and Campylobacter fetus and N-glycans as targets for ...

254
N-linked glycosylation in Campylobacter jejuni and Campylobacter fetus and N-glycans as targets for antibody-based detection A thesis submitted to the University of Manchester for the degree of Doctor of Philosophy in the Faculty of Biology, Medicine and Health 2017 Danielle Weaver School of Biological Sciences / Division of Infection, Immunity and Respiratory Medicine

Transcript of and Campylobacter fetus and N-glycans as targets for ...

N-linked glycosylation in Campylobacter jejuni

and Campylobacter fetus and N-glycans as

targets for antibody-based detection

A thesis submitted to the University of Manchester for the degree of Doctor

of Philosophy in the Faculty of Biology, Medicine and Health

2017

Danielle Weaver

School of Biological Sciences / Division of Infection,

Immunity and Respiratory Medicine

2

Table of Contents

List of Figures ....................................................................................................................... 8

List of Tables ...................................................................................................................... 11

Abbreviations ..................................................................................................................... 12

Abstract ............................................................................................................................... 15

Declaration ......................................................................................................................... 16

Copyright Statement ........................................................................................................... 16

Acknowledgements ............................................................................................................ 17

Chapter 1. Introduction ....................................................................................................... 18

1.1. The Genus Campylobacter .......................................................................................... 19

1.1.1 C. jejuni and C. coli are the most common Campylobacter pathogens ................ 19

1.1.2. Campylobacter fetus: a veterinary and human pathogen. .................................... 20

1.1.3. Emerging Campylobacter pathogens. .................................................................. 21

1.1.4. Overview of Campylobacter in the farming industry. .......................................... 22

1.2. Features of Protein glycosylation. ............................................................................... 23

1.2.1. N-linked protein glycosylation in Eukaryotes and Archaea ................................. 24

1.3. O-linked protein glycosylation in Bacteria .................................................................. 25

1.4. Overview of N-glycosylation in Bacteria .................................................................... 26

1.5. N-linked glycosylation in Campylobacter ................................................................... 27

1.5.1. Campylobacter jejuni N-glycosylation. ................................................................ 27

1.5.1.1. The C. jejuni N-linked glycosylation pathway .................................................. 28

1.5.1.2 The Campylobacter oligosaccharyltransferase, PglB ......................................... 30

1.5.2. The role of N-glycosylation in C. jejuni ............................................................... 31

1.5.3. N-glycosylation in other Campylobacter species. ................................................ 32

1.5.3.1. N-linked glycosylation in C. fetus ................................................................. 35

1.5.4. N-linked glycosylation in closely related genera, Helicobacter and Wolinella ... 36

1.5.5. N-linked glycosylation in hydrothermal vent Bacteria ........................................ 36

1.6. Campylobacter detection/identification ...................................................................... 37

3

1.6.1. Culture and phenotypic testing ............................................................................. 37

1.6.2. Molecular methods for detecting Campylobacter in poultry ............................... 38

1.6.3. Methods of detecting Campylobacter fetus in cattle. ........................................... 43

1.6.4. Emerging methods for Campylobacter detection ................................................. 44

1.6.5. N-linked glycans as targets for antibody-based detection. ................................... 45

1.7 Aims ............................................................................................................................. 47

Chapter 2. Materials and Methods. ..................................................................................... 48

2.1 Bacterial strains and plasmids ................................................................................. 49

2.2 Isolation of Campylobacter from retail chicken ...................................................... 49

2.2.1 Culture and isolation of Campylobacter chicken meat isolates ....................... 49

2.2.2 Verification and Molecular typing of isolates .................................................. 49

2.3 Whole cell lysate preparation for SDS-PAGE ........................................................ 50

2.4 SDS-PAGE and western blotting ............................................................................ 50

2.5 Dot blot assay .......................................................................................................... 51

2.6 Flow cytometry ........................................................................................................ 51

2.7. Polymerase Chain Reaction ..................................................................................... 52

2.8. DNA agarose gel electrophoresis ............................................................................ 54

2.9. DNA restriction digestion and ligation .................................................................... 55

2.10. E. coli competent cell preparation and heat shock transformation ...................... 55

2.11. C. fetus conjugation ............................................................................................. 55

2.12. Protein expression in E. coli BL21 (DE3) ........................................................... 56

2.13. Nickel affinity chromatography ........................................................................... 56

2.14. Ion exchange chromatography ............................................................................. 57

2.15. Post-purification protein treatment ...................................................................... 57

2.16. Mass Spectrometry .............................................................................................. 57

2.16.1. Sample preparation and MALDI-MS analysis ............................................. 57

2.16.2. LC-MS/MS ................................................................................................... 58

2.17. Bioinformatics ..................................................................................................... 59

4

Chapter 3. Characterisation of an antiserum raised against a Campylobacter jejuni N-

linked glycoprotein ............................................................................................................. 60

3.1. Introduction ............................................................................................................. 61

3.1.1. Overview of C. jejuni NGRP antiserum production process ........................... 61

3.1.2. Project aims ...................................................................................................... 61

3.2. Characterisation of CjNgp reactivity against Campylobacter whole cell lysate

extracts. ............................................................................................................................... 63

3.2.1. The reactivity of CjNgp antiserum against C. jejuni N-linked glycan. ............ 63

3.2.2. CjNgp reactivity with a variety of C. jejuni strains ......................................... 65

3.2.3. CjNgp reactivity with recent Campylobacter chicken meat isolates ............... 67

3.2.4. Campylobacter species specificity of CjNgp. .................................................. 69

3.3. CjNgp binding untreated cells of Campylobacter. .................................................. 71

3.3.1. CjNgp binding to untreated C. jejuni cells is not dependent on N-linked

glycosylation. .................................................................................................................. 71

3.3.2. Reactivity against the NGRP protein influences the ability of CjNgp to bind

untreated C. jejuni cells. ................................................................................................. 73

3.3.3. CjNgp binding to untreated cells of Campylobacter species ........................... 75

3.3.4. CjNgp labels whole cells of C. jejuni 11168H ................................................. 77

3.4. Discussion ................................................................................................................ 79

Chapter 4. Development and characterisation of an antiserum raised against a

Campylobacter fetus N-linked glycoprotein ...................................................................... 83

4.1 Introduction ............................................................................................................. 84

4.2 Production of a recombinant Campylobacter fetus N-linked glycoprotein for use as

immunogen. ........................................................................................................................ 86

4.2.1 Using a C. fetus chromosomal integration vector to integrate an N-glycoprotein

was not a suitable approach. ........................................................................................... 86

4.2.2 Production of a C. fetus conjugative vector encoding an N-glycoprotein. ...... 88

4.2.3 Production of N-linked glycoprotein immunogen in C. fetus. ......................... 90

4.2.4 Purification of the N-linked glycoprotein immunogen from C. fetus. ............. 92

5

4.3 Characterisation of C. fetus NGRP antiserum (CfNgp) reactivity with

Campylobacter whole cell lysate extracts. ......................................................................... 95

4.3.1 CfNgp reactivity with C. fetus fetus NCTC 10842. ......................................... 95

4.3.2 CfNgp reactivity with numerous C. fetus strains. ............................................ 97

4.3.3 Campylobacter species specificity of CfNgp. .................................................. 99

4.3.4 C. jejuni N-linked glycan structural specificity of CfNgp. ............................ 101

4.4 CfNgp binding to Campylobacter cells. ................................................................ 103

4.4.1 CfNgp binding untreated cells of C. fetus. ..................................................... 103

4.4.2 The influence of pglB on CfNgp C. fetus cell binding. .................................. 104

4.4.3 CfNgp binding to untreated cells of Campylobacter species. ........................ 106

4.5 Discussion .............................................................................................................. 108

Chapter 5. Towards engineering the C. fetus N-linked glycan using glycocompetent E.

coli containing a hybrid C. jejuni-C. fetus system ............................................................ 115

5.1 Introduction ................................................................................................................ 116

5.1.1 Comparison of the C. fetus and C. jejuni pgl loci ............................................... 116

5.2 Strategy to investigate the role of C. fetus glycosyltransferases in N-glycan assembly

by integrating them into an E. coli system. ...................................................................... 118

5.2.1 An uncharacterised predicted glycosyltransferase gene lies within C. fetus pgl

locus .............................................................................................................................. 118

5.2.2 Model of a hybrid C. jejuni/C. fetus glycosylation machinery. ........................... 120

5.3 Identification of initial C. fetus glycosyltransferase to act on C. jejuni pglH::kn

trisaccharide. ..................................................................................................................... 122

5.3.1 Cloning strategy ................................................................................................... 122

5.3.1.1 Cloning pglH1 into pETNGRP ............................................................... 122

5.3.1.2 Cloning pglH2 into pETNGRP ............................................................... 125

5.3.1.3 Cloning Cf1389 into pETNGRP ............................................................. 127

5.3.1.4 Cloning CjpglH into pETNGRP ............................................................. 129

5.3.2 Only CfpglH1 decreased the mobility of NGRP glycoforms in the ppglpglH::kn

background. .................................................................................................................. 131

6

5.4 Identification of second C. fetus glycosyltransferase to act on C. jejuni pglH::kn

trisaccharide. ..................................................................................................................... 133

5.4.1 Cloning strategy ................................................................................................... 133

5.4.1.1 Construction of pNH1H2 ............................................................................. 133

5.4.1.2 Construction of pNH1-1389 ......................................................................... 133

5.4.2 C. fetus pglH2 decreased NGRP glycoform mobility in the ppglpglH::kn CfpglH1

background. .................................................................................................................. 135

5.5 Structural analysis of N-linked glycoproteins produced from hybrid systems. ......... 137

5.5.1 Purification of NGRP from C. jejuni/C. fetus hybrid N-linked glycosylation

systems in E. coli .......................................................................................................... 137

5.5.2. MALDI-TOF MS analysis of purified NGRP glycoforms ................................. 139

5.5.2 Production and purification of modified NGRP in hybrid systems. ................... 143

5.5.3 MS analysis of purified NGRP3. ......................................................................... 146

5.5.3.1 The sequon-containing tryptic peptide of unglycosylated NGRP3 was

detected by MALDI TOF Mass spectrometry. ......................................................... 146

5.5.3.2 MALDI-TOF MS analysis of glycosylated NGRP3 using CHCA matrix. .. 149

5.5.3.3 MALDI-TOF MS analysis of glycosylated NGRP3 using DHB matrix. ..... 151

5.5.3.4 LC-MS/MS analysis of glycosylated NGRP3 .............................................. 153

5.6 Construction and analysis of hybrid pgl system containing C. fetus pglH genes and

predicted glycosyltransferase cf1389. .............................................................................. 159

5.6.1 Cloning strategy ................................................................................................... 159

5.6.2 Addition of cf1389 to a ppglpglH::kn CfpglH1H2 background resulted in three

apparent forms of NGRP3 ............................................................................................ 160

5.6.3 MALDI-TOF MS analysis of glycosylated NGRP3 produced in the presence of C.

fetus pglH genes and cf1389. ........................................................................................ 162

5.7 Discussion ................................................................................................................... 164

Chapter 6. C. fetus N-glycoproteome prediction and the conservation of N-glycoproteins

between Campylobacter species ....................................................................................... 168

6.1 Introduction ................................................................................................................ 169

6.2 C. fetus N-glycoproteome prediction and validation .................................................. 172

7

6.2.1 The predicted C. fetus fetus 82-40 N-glycoproteome .......................................... 172

6.2.2.2 Validation of a newly predicted N-glycoprotein, Cf0445 ............................ 180

6.3 Analysis of the C. fetus predicted N-glycoproteome .................................................. 182

6.3.1 Subcellular locations of predicted C. fetus N-glycoproteins. .............................. 182

6.3.2 Sequon distribution in C. fetus predicted N-glycoproteins. ................................ 185

6.3.3 Amino acid composition of sequons in C. fetus predicted N-glycoproteins. ...... 187

6.4 Predicted conservation of N-glycoproteins within the Campylobacter genus ........... 189

6.4.1 Predicted conservation of N-glycoproteins between C. jejuni and C. fetus ........ 189

6.4.2.1 N-glycosylation of components of the CmeABC multidrug efflux system is

predicted to be conserved in several Campylobacter species .................................. 195

6.5 Discussion ................................................................................................................... 197

Chapter 7. Conclusion and future work ............................................................................ 202

7.1 N-linked glycoprotein glycans as targets for antibody-based detection/identification

.......................................................................................................................................... 203

7.2 Developing glycocompetent E. coli producing the C. fetus N-linked glycan. ........... 206

7.3 The C. fetus predicted N-glycoproteome and putative conservation of N-glycoproteins

amongst Campylobacter species. ..................................................................................... 208

7.4 Final conclusion .......................................................................................................... 211

References ........................................................................................................................ 213

Appendix .......................................................................................................................... 252

Final word count: 50,869

8

List of Figures

Figure 1.1. Campylobacter jejuni Pgl pathway. 29

Figure 1.2. N-linked glycan structures produced by Campylobacter species,

Helicobacter pullorum and Wolinella succinogenes. 34

Figure 3.1. N-linked glycan structural specificity of CjNgp antiserum. 64

Figure 3.2. Reactivity of CjNgp antiserum with diverse C. jejuni reference strains. 66

Figure 3.3. Reactivity of CjNgp antiserum with Campylobacter chicken meat

isolates. 68

Figure 3.4. Campylobacter species specificity of CjNgp antiserum. 70

Figure 3.5. CjNgp C. jejuni cell binding in dot blot assay. 72

Figure 3.6. CjNgp cell binding upon pre-incubation with BL21 ± NGRP lysates. 74

Figure 3.7. CjNgp binding to cells of Campylobacter species. 76

Figure 3.8. Flow cytometric analysis of CjNgp antiserum labelling of fluorescent

C. jejuni 11168gfp4. 78

Figure 4.1 Chromosomal integration of NGRP into C. fetus did not produce

sufficient protein yield. 87

Figure 4.2. Construction of pG1-N and conjugation into C. fetus. 89

Figure 4.3. Production of glycosylated NGRP in C. fetus 91

Figure 4.4. Nickel affinity chromatography purification of glycosylated NGRP

from C. fetus. 94

Figure 4.5. Initial characteriseration of CfNgp C. fetus reactivity. 96

Figure 4.6. CfNgp antiserum C.fetus strain coverage. 98

Figure 4.7. Campylobacter species specificity of CfNgp antiserum 100

Figure 4.8. C. jejuni N-linked glycan structural specificity of CfNgp antiserum. 102

Figure 4.9. CfNgp C. fetus fetus and C. fetus venerealis cell binding in dot blot

assay. 103

Figure 4.10. The pglB-independency of CfNgp C. fetus cell binding. 105

Figure 4.11. CfNgp binding to cells of Campylobacter species. 107

9

Figure 5.1. Comparison of the pgl loci and N-glycan structures of C. jejuni and

C. fetus 119

Figure 5.2. Plan to investigate predicted glycosyltransferases in C. fetus N-linked

glycosylation by creating hybrid C. jejuni/C fetus pgl system in E. coli. 121

Figure 5.3. Construction of pNH1. 124

Figure 5.4. Construction of pNH2. 126

Figure 5.5. Construction of pN1389. 128

Figure 5.6. Construction of pNCjH. 130

Figure 5.7. Identification of C. fetus PglH1 glycosyltransferase activity using a

C. jejuni pglH::kn N-glycan structure. 132

Figure 5.8. Construction of pNH1H2 and pNH1-1389. 134

Figure 5.9. Further construction of a hybrid C. jejuni/C. fetus N-glycan through

C. fetus PglH2 activity. 136

Figure 5.10. Purification of NGRP from hybrid C. jejuni/C. fetus pgl systems

expressed in E. coli BL21. 138

Figure 5.11. MALDI-TOF mass spectrometry analysis of NGRP expressed in

hybrid C. jejuni/C. fetus pgl systems. 142

Figure 5.12. Amino acid sequence of NGRP3, a modified version of NGRP

optimised for glycopeptide analysis by mass spectrometry. 143

Figure 5.13. Purification of NGRP3 from hybrid C. jejuni/C. fetus pgl systems

expressed in E. coli BL21. 145

Figure 5.14. MALDI-TOF Mass spectrometry analysis of unglycosylated NGRP3. 148

Figure 5.15. MALDI-TOF Mass spectrometry analysis of glycosylated NGRP3

using CHCA matrix. 150

Figure 5.16. MALDI-TOF Mass spectrometry analysis of glycosylated NGRP3

using DHB matrix. 152

Figure 5.17. LC-MS/MS analysis of unglycosylated NGRP3 and NGRP3 modified

with C. jejuni pglH::kn N-glycan. 154

Figure 5.18. LC-MS/MS analysis of NGRP3. 156

Figure 5.19. Two possible mechanisms of N-glycan assembly in the hybrid

C. jejuni/C. fetus systems. 158

Figure 5.20. Production and purification of NGRP3 from E. coli BL21 161

10

ppglpglH::kn pX12-N3.

Figure 5.21. MALDI-TOF Mass spectrometry analysis NGRP3 produced in

C. jejuni/C.fetus hybrid system containing three predicted transferases from

C. fetus.

163

Figure 6.1 Manual curation Bioinformatics workflow for prediction of

N-glycoproteomes in Campylobacter. 171

Figure 6.2 Decision tree for subcellular localisation assignment of proteins during

manual curation of results from bioinformatics N-glycoprotein prediction pipeline. 173

Figure 6.3. Validation of PglB-dependent modification of Cf0781 in C. fetus. 179

Figure 6.4. Validation of PglB-dependent modification of Cf0445 in C. fetus. 181

Figure 6.5. Subcellular localisation predictions of C. fetus proteins. 184

Figure 6.6. Distribution of N-glycosylation sequons in C. jejuni and C. fetus

experimentally identified and predicted N-glycoproteins. 186

Figure 6.7. Logo representations of amino acid abundance in N-glycosylation

sequons of C. jejuni and C. fetus known and predicted N-glycoproteins. 188

Figure 6.8 Schematic of N-glycoprotein conservation analysis results. 190

11

List of Tables

Table 2.1. Primer sequences. 52

Table 5.1. Predicted tryptic digest of NGRP when expressed from pET22(+)-

derived plasmids. 139

Table 5.2. Predicted m/z of potential tryptic glycopeptides from NGRP. 141

Table 5.3. Predicted tryptic digest of NGRP3 when expressed from pET22(+)-

derived plasmids. 147

Table 5.4. Predicted m/z of potential tryptic glycopeptides from NGRP3. 149

Table 6.1. Complete predicted C. fetus N-glycoproteome. 174

Table 6.2. Identification of N-glycosylation sequon-containing C.fetus

homologues of known C. jejuni N-glycoproteins. 191

Table 6.3 Putative conservation of experimentally identified C. jejuni N-

glycoproteins in Campylobacter species. 196

Appendix Table 1. Bacterial strains used. 252

Appendix Table 2. Plasmids used. 254

12

Abbreviations

Anti-WC 11168 Anti-whole cell C. jejuni 11168H antiserum

Asn/N Asparagine

Asp/D Aspartate

ATCC American Type Culture Collection

Bis-tris Bis-(2-hydroxy-ethyl)-amino-tris(hydroxymethyl)-methane

BLAST Basic Local Alignment Search Tool®

BLASTP Protein Basic Local Alignment Search Tool®

BSA Bovine serum albumin

BVC Bovine Venereal Campylobacteriosis

CCDA Charcoal cefoperazone deoxycholate agar

CDP Cytidine diphosphate

CDT Cytolethal distending toxin

CELLO SubCELlular LOcalisation predictor

CFU Colony forming units

CHCA α-cyano-4-hydroxycinnamic acid

cPCR Colony PCR

CV Column volume

C. f fetus C. fetus subspecies fetus

C. f ven C. fetus subspecies venerealis

C. hyo hyo C. hyointestinalis subspecies hyointestinalis

DHB 2,5-dihydroxybenzoic acid

DiNAcBac Di-N-acetylbacillosamine

DNA Deoxyribonucleic acid

EDTA Ethylenediaminetetraacetic acid

EFSA European Food Safety Authority

ER Endoplasmic reticulum

FA Formic acid

fOS Free oligosaccharide

Gal Galactose

GalNAc N-acetylgalactosamine

13

GBS Guillain Barré Syndrome

gDNA Genomic DNA

GDP Guanosine diphosphate

GFP Green Fluorescent Protein

Glc Glucose

GlcNAc N-acetylglucosamine

Glu/E Glutamic acid

GT Glycosyltransferase

Hex Hexose

HexNAc N-acetylhexosamine

His Histidine

IPTG Isopropyl β-D-1-thiogalactopyranoside

LB Lysogeny broth

LC-MS/MS Liquid chromatography-tandem mass spectrometry

LDS Lithium dodecyl sulfate

LLO Lipid-linked oligosaccharide

LPS Lipopolysaccharide

MALDI LIFT-

TOF/TOF MS

Matrix-assisted laser desorption/ionisation LIFT-time-of-

flight/time-of-flight mass spectrometry

Man Mannose

MOPS 3-(N-morpholino)propanesulfonic acid

NCBI National Centre for Biotechnology Information

NCTC National Collection of Type Cultures

OD600 Optical density at 600 nm

OIE World Organisation for Animal Health

OPG Osmoregulated periplasmic glucan

OST Oligosaccharyltransferase

P Proline

PBS Phosphate buffered saline

PBS-T Phosphate-buffered saline - Tween

PCR Polymerase Chain Reaction

RBS Ribosome binding site

R-PE R-Phycoerythrin

14

SDS-PAGE Sodium dodecyl sulfate-polyacrylamide gel electrophoresis

Ser/S Serine

SNP Single nucleotide polymorphism

SOC Super optimal broth

SpI Signal peptide I

SpII Signal peptide II (lipoprotein)

TAE Tris base, acetic acid and EDTA

TFA Trifluoroacetic acid

Thr/T Threonine

TM Transmembrane

Tris Tris(hydroxymethyl)aminomethane

T4SS Type IV secretion system

UDP Uridine diphosphate

UndP Undecaprenyl phosphate

VBNC Viable but non-culturable

15

Abstract

N-linked glycosylation in Campylobacter jejuni and Campylobacter fetus and N-

glycans as targets for antibody-based detection

Danielle Weaver, Doctor of Philosophy at the University of Manchester, 2017.

Campylobacter spp., especially C. jejuni and C. coli, are the leading cause of bacterial

gastroenteritis in Europe. There is a recognised need to develop detection tools which can

be performed on farms to facilitate reducing the presence of Campylobacter in poultry. A

similar application could be beneficial for detection of C. fetus, a veterinary pathogen

which causes significant economic loss in the cattle industry. Campylobacter species

perform protein N-linked glycosylation and in C. jejuni at least 150 proteins, many of

which are surface-exposed, may be modified. Therefore, the first portion of this thesis

investigated the feasibility of using N-linked glycans as targets for antibody-based

detection of Campylobacter species. To do this, a His-tagged N-glycoprotein was

expressed and purified from C. fetus and used as immunogen to raise an antiserum termed

CfNgp. The Campylobacter N-glycan reactivity of this antiserum was characterised and it

was shown to react with N-glycoproteins and cells of C. fetus and other emerging

Campylobacter species such as C. concisus. Immunoblotting techniques and flow

cytometry were used to characterise an antiserum (CjNgp) raised against a C. jejuni N-

linked glycoprotein and demonstrated that it can specifically detect cells of C. jejuni,

C. coli and other emerging Campylobacter species found in poulty. This thesis also

describes the investigation of the relatively uncharacterised C. fetus N-linked

glycosylation system. Functional analysis of C. fetus predicted glycosyltransferases was

acheived by developing glycocompetent E. coli containing a hybrid C. jejuni/C. fetus pgl

system. The N-glycan structures biosynthesised were analysed using mass spectrometry

and this novel approach discovered the activity of two C. fetus glycosyltransferase

enzymes. Finally, this work used a bioinformatics pipeline to produce a C. fetus predicted

N-linked glycoproteome and experimentally verified a newly identified N-linked

glycoprotein. This pipeline was also applied to investigate the putative conservation of N-

linked glycoproteins throughout the Campylobacter genus and highlighted ‘core’ N-linked

glycoproteins which are key targets for experimental investigation. Overall, this work

demonstrates that Campylobacter N-linked glycans are attractive targets for antibody-

based detection, expands our knowledge of C. fetus N-linked glycosylation and contributes

to the broader understanding of this intriguing aspect of Campylobacter biology.

16

Declaration

No portion of the work referred to in this thesis has been submitted in support of an

application for another degree or qualification of this or any other university or other

institute of learning

Copyright Statement

I. The author of this thesis (including any appendices and/or schedules to this thesis)

owns certain copyright or related rights in it (the “Copyright”) and she has given

The University of Manchester certain rights to use such Copyright, including for

administrative purposes.

II. Copies of this thesis, either in full or in extracts and whether in hard or electronic

copy, may be made only in accordance with the Copyright, Designs and Patents

Act 1988 (as amended) and regulations issued under it or, where appropriate, in

accordance Presentation of Theses Policy You are required to submit your thesis

electronically Page 11 of 25 with licensing agreements which the University has

from time to time. This page must form part of any such copies made.

III. The ownership of certain Copyright, patents, designs, trademarks and other

intellectual property (the “Intellectual Property”) and any reproductions of

copyright works in the thesis, for example graphs and tables (“Reproductions”),

which may be described in this thesis, may not be owned by the author and may be

owned by third parties. Such Intellectual Property and Reproductions cannot and

must not be made available for use without the prior written permission of the

owner(s) of the relevant Intellectual Property and/or Reproductions.

IV. Further information on the conditions under which disclosure, publication and

commercialisation of this thesis, the Copyright and any Intellectual Property and/or

Reproductions described in it may take place is available in the University IP

Policy (http://documents.manchester.ac.uk/DocuInfo.aspx?DocID=24420), in any

relevant Thesis restriction declarations deposited in the University Library, The

University Library’s regulations

(http://www.library.manchester.ac.uk/about/regulations/) and in The University’s

policy on Presentation of Theses.

17

Acknowledgements

I would like to thank my supervisor, Dr. Dennis Linton, for providing his time and

guidance during my PhD. I am also very grateful to Dr. Adrian Jervis for passing on his

wisdom at the beginning and for continued help and advice upon leaving the group. I

would like to extend my gratitude to the Society for Applied Microbiology for funding this

PhD.

I would like to acknowledge all those in the Microbiology lab for the technical assistance

and conversation along the way. In particular, thanks to Ange for all her help and advice. I

would also like to thank Nader for all the laughs and out-of-hours discussions.

Finally, I would like to thank my friends and family for all the good times. I am

particularly thankful to Elena for supporting me and providing positivity when mine had

run dry. Finally, a special thanks to Mam and Dad for all the support and encouragement

throughout.

18

Chapter 1. Introduction

19

1.1. The Genus Campylobacter

1.1.1 C. jejuni and C. coli are the most common Campylobacter pathogens

Campylobacter is a genus of Gram-negative Epsilon-proteobacteria with a low GC content

and a curved or helical rod morphology (Lastovica et al., 2014). All but two species,

C. gracilis and C. hominis, are motile (Vandamme et al., 1995; Lawson et al., 2001). The

majority are microaerobic, but some prefer anaerobic conditions and require hydrogen or

formate for growth (Kaakoush et al., 2015). The genus contains a variety of pathogens

able to cause gastroenteritis in humans, and the occurrence of campylobacteriosis appears

to have risen in the recent decade (Kaakoush et al., 2015). In particular, C. jejuni and

C. coli are the most common agents of campylobacteriosis. Together with C. upsaliensis

and C. lari these form the so-called ‘thermotolerant’ Campylobacter group. All

Campylobacter species grow between 30 and 42 °C, but this group grow best at 42° C and

are commonly found in birds (Lastovica et al., 2014). The presence of these organisms in

poultry is the main reservoir for human infection (EFSA, 2010). In the UK over 60% of

retail chicken meat is positive for Campylobacter, and over 10% is heavily contaminated

(> 1000 colony forming units per gram) (PHE, 2017). In the European Union

Campylobacter is the most common bacterial cause of gastroenteritis, with more than

200,000 cases in 2015 (EFSA, 2016).

As the major human pathogen of the genus, C. jejuni is the best characterised species.

Infrequently, Campylobacter infections can also lead to Guillain Barré Syndrome (GBS)

which causes temporary paralysis that can last many months, and even be fatal (van Doorn

et al., 2008). This is a result of molecular mimicry, as certain variants of C. jejuni lipo-

oligosaccharide (LOS) found on its surface can induce an autoimmune response against

human neuronal gangliosides (Young et al., 2007). C. jejuni LOS and its capsular

polysaccharide are structurally variable which is thought to help evade the host immune

response. In addition, C. jejuni has a variety of features which aid its adherence to and

invasion of host cells. These include adhesins such as CadF and JlpA, the flagellar type III

secretion system and the mitosis-blocking cytotoxin, cytolethal distending toxin or CDT

(Jin et al., 2001; Konkel et al., 2004, 1997; Pickett and Whitehouse, 1999).

There have been growing concerns of drug resistance within Campylobacter species, in

particular against the cornerstone treatments, fluoroquinolone and macrolide antibiotics.

Indeed, the World Health Organisation has recently designated fluoroquinolone-resistant

Campylobacter spp. as high priority in regards to the need for novel antibiotics (WHO,

20

2017). The expansion of macrolide resistance is particularly concerning in C. coli, where

rates of resistant human isolates were over three times higher in 2014 compared to 2011

(NARMS, 2016). Of particular note is CmeABC, an efflux pump that provides protection

against bile salts but also contributes to drug resistance (Lin et al., 2002). In addition to

developing intrinsic mutations which can provide resistance, it can also work in

conjunction with other mutations such as in the DNA gyrase gene, gyrA, to mediate

enhanced antibiotic resistance (Luangtongkum et al., 2009; Luo et al., 2003). This system

has been found to play a role in drug resistance of several Campylobacter species (Guo et

al., 2010).

1.1.2. Campylobacter fetus: a veterinary and human pathogen.

Campylobacter fetus consists of three subspecies, C. fetus subsp. fetus, C. fetus subsp.

venerealis and C. fetus subsp. testudinum (Patrick et al., 2013; Veron and Chatelain,

1973). C. fetus subsp. fetus causes disease in sheep, cattle and, less frequently, humans.

Veterinary infections often result in abortion (Campero et al., 2005), whereas human

infections can involve a variety of outcomes ranging from diarrhoea to systemic disease

(Wagenaar et al., 2014). Most C. fetus subsp. fetus human infections involve

immunocompromised, pregnant or elderly individuals. In contrast, C. fetus subsp.

venerealis is an extremely rare cause of human vaginosis (Holst et al., 1987), but has a

specific host niche within the reproductive tract of cattle, where it can cause abortion and

temporary infertility (Mshelia et al., 2010). This disease, named Bovine Venereal

Campylobacteriosis (BVC), is a considerable burden on the cattle industry as it can cause

infertility for up to 8 months (Mshelia et al., 2007). The most recently described member

of the species, C. fetus testudinum, is primarily associated with reptiles but can also infect

humans (Choi et al., 2016; Fitzgerald et al., 2014; Patrick et al., 2013). Although, like C.

fetus fetus, C. fetus testudinum can cause systemic infection in the immunocompromised

or elderly, these instances have been described much less frequently in the literature.

C. fetus harbour many virulence factors also found in C. jejuni. For example, it contains

CDT, the invasion protein CiaB and adhesins such and CadF and PEB1 (Ali et al., 2012).

In contrast to C. jejuni, C. fetus does not have a capsular polysaccharide. However, a

striking and distinct aspect of C. fetus biology is the crystalline array of S-layer proteins

on its surface. S layers are a common feature in Archaea but are less common in Bacteria,

making C. fetus relatively unusual. In C. fetus, the S layer is subject to antigenic variation

due to alternating expression of two S layer protein homologs, SapA and SapB (Garcia et

21

al., 1995), contributing to immune evasion and serum resistance (Blaser et al., 1988;

Thompson, 2002).

Comparative genomics recently identified considerable conservation of virulence factors

between C. fetus subspecies, including the S layer proteins, CadF, CiaB and CdtABC

(Gilbert et al., 2016). However, there are differences proposed to explain the distinct host

niches of each subspecies (Gilbert et al., 2016). For example, the most defining feature of

C. fetus venerealis which separates it from the other subspecies is the existence of a

pathogenicity island (PAI) encoding a type IV secretion system (T4SS) (Gorkiewicz et al.,

2010).

1.1.3. Emerging Campylobacter pathogens.

Several Campylobacter species are said to be ‘emerging’ pathogens as their incidence in

disease is becoming more appreciated due to recent advances in detection and

identification techniques (Man, 2011). These species include C. lari, C. upsaliensis, C.

concisus and C. hyointestinalis and likely remain underrecognised as causes of human

disease (Kaakoush et al., 2015). For example, a study of paediatric diarrhoeic stools

reported the isolation of Campylobacter species other than C. jejuni and C. coli from 48%

of samples (Allos and Lastovica, 2008). In addition, many of these emerging

Campylobacter species harbour cytolethal distending toxin (CDT) genes (Man, 2011).

C. lari has been isolated from a variety of organisms including mammals, birds and even

molluscs (Man, 2011). In humans, C. lari can cause enteritis (Simor and Wilcox, 1987)

and more infrequently, bacteraemia (Martinot et al., 2001; Werno et al., 2002). Similarly,

C. upsaliensis can cause gastroenteritis (Goossens et al., 1990) and rare cases of

bacteraemia and abortion have been documented (Gurgan and Diker, 1994; Patton et al.,

1989). Unlike other thermotolerant Campylobacter species which commonly reside in

birds, this species is primarily found in dogs and cats (Baker et al., 1999).

C. ureolyticus, recently reclassified from the genus Bacteroides (Vandamme et al., 2010),

is another emerging Campylobacter pathogen. Its incidence in human infection appears to

be increasing and O’Donovan et al. (2014) argue that it may now represent the second

most common Campylobacter species causing gastroenteritis, surpassing C. coli.

Interestingly, cattle appear to be the reservoir for this species, with transmission via

contaminated milk (Koziel et al., 2012).

C. concisus was first identified from the human oral cavity (Tanner et al., 1981) and has

since been isolated from dogs and cats (Chaban et al., 2010; Petersen et al., 2007). This

22

species can cause gastroenteritis often with milder symptoms that last longer than

infections caused by C. jejuni/C. coli (Kaakoush et al., 2015; Nielsen et al., 2012). It has

also been linked with inflammatory bowel disease and periodontitis; however, its role in

these diseases is still contested (Kaakoush and Mitchell, 2012).

C. hyointestinalis consists of two subspecies, C. hyointestinalis subsp. hyointestinalis and

C. hyointestinalis subsp. lawsonii, which are found in the gastrointestinal tracts of cattle

and swine, respectively (Gebhart et al., 1985; Miller et al., 2016; On et al., 1995). There

are rare cases of C. hyointestinalis enteritis in humans (Edmonds et al., 1987) thought to

be transmitted from infected swine, where it also causes enteritis (Gorkiewicz et al., 2002).

Furthermore, C. hyointestinalis subsp. hyointestinalis can cause potentially fatal enteritis

in calves (Diker et al., 1990).

1.1.4. Overview of Campylobacter in the farming industry.

Campylobacter are usually transmitted to humans through consumption of contaminated

meat. Therefore, to reduce human infection, the problem needs to be addressed at the farm

level. Campylobacter spp. are widely distributed in the environment, having been

identified in wild birds, domestic animals, freshwater and sewage (Jones, 2001). As a

result, biosecurity measures are often implemented on farms to reduce livestock exposure.

In broiler chickens, maternal antibodies seem to provide some protection as colonisation

does not usually occur until chicks are at least 2 weeks old (Sahin et al., 2003). However,

once acquired, the colonisation often persists throughout its lifetime (Lee and Newell,

2006). It is traditionally thought that this colonisation is asymptomatic and thus

Campylobacter is often viewed as a chicken gut commensal (Hermans et al., 2012).

However, recent data suggest that C. jejuni colonisation induces a host inflammatory

response and, in certain chicken breeds, can cause chronic inflammation and diarrhoea

(Humphrey et al., 2014). Further studies have also identified that colonisation can impair

nutrient uptake and growth in commercial chickens (Awad et al., 2015a, 2015b).

The majority of Campylobacter transmission through a broiler flock occurs horizontally,

although there is some evidence suggesting vertical transfer can occur (Newell and

Fearnley, 2003). Transmission is mostly via the faecal-oral route, particularly due to

contaminated water. Once colonisation is established in one individual, Campylobacter

can spread rapidly throughout an entire flock within a few days (Lee and Newell, 2006).

However, the often asymptomatic nature of Campylobacter colonisation in poultry means

contamination of a flock can go unnoticed without active monitoring. Therefore, regular

23

surveillance of Campylobacter infection should be undertaken. Methods implemented for

identifying these organisms are outlined in section 1.6.

At the farm level, preventative measures include enhanced biosecurity, chicken

vaccination, chlorinating drinking water and competitive exclusion (e.g. probiotics)

(Umaraw et al., 2017). Antibiotics have traditionally been used as both preventative and

treatment measures. However, as discussed, the emergence of antibiotic resistance within

Campylobacter is concerning and hence alternative measures are being sought (Johnson et

al., 2017). For example, both bacteriocin and bacteriophage therapy can successfully

reduce the load of Campylobacter in broilers (Johnson et al., 2017; Umaraw et al., 2017).

Another option to prevent Campylobacter infection in humans is to eliminate the organism

during carcass processing, where contamination of meat with cecal contents often occurs.

These methods of treatment include multi-stage scalding of carcasses, chilling, irradiation

and freezing the meat (Umaraw et al., 2017).

C. fetus infection of cattle causes significant economic losses in agriculture, with reports

of up to a 66% reduction in gross profit in the initial year of infection (McMillen et al.,

2006). As this organism can also cause infections in humans, an infected herd also

presents a human health risk. Bulls can act as asymptomatic carriers (Samuelson and

Winter, 1966), meaning infected animals are difficult to identify. Thus, herds should be

routinely surveyed for infection. Tools used to detect and identify C. fetus from samples

will be discussed in section 1.6. Prevention and control measures for C. fetus include

increased biosecurity, artificial insemination, isolation of infected individuals and

vaccination of bulls (Mshelia et al., 2007). All members of contaminated herds are often

vaccinated to reduce the duration of infection, however, cows can still remain infected for

more than one mating season (OIE, 2008). Although antibiotic treatment is successful for

bulls, it usually does not fully eliminate C. fetus from the reproductive tract of cows

(Mshelia et al., 2007). Due to this, C. fetus infection often results in culling to prevent

further infection spread and hence incurs further cost (Truyers et al., 2014).

1.2. Features of Protein glycosylation.

A key aspect of Campylobacter biology is the extensive protein glycosylation systems.

The following sections outline the general features of protein glycosylation within all three

Domains of life, with a particular focus on N-linked glycosylation in Bacteria. Protein

24

glycosylation describes the enzymatic covalent linkage of carbohydrates onto proteins. It

is predicted that over 70% of eukaryotic proteins are glycosylated (Dell et al., 2010). The

diversity of glycan structures identified in eukaryotes is astounding, and a single

glycosylation site can harbour one of a variety of complex structures resulting in vast

microheterogeneity (Moremen et al., 2012). As a result, protein glycosylation serves many

functions, such as protein folding and cell adhesion, which can in turn have implications in

immunity, inflammation and cancer (Moremen et al., 2012). The two main types of protein

glycosylation are N-linked and O-linked. The former involves transfer of a glycan onto a

nitrogen atom within an asparagine (Asn or N) residue. The latter corresponds to addition

of a glycan onto an oxygen atom of the amino acid residues, serine (Ser) and threonine

(Thr) (Dell et al., 2010). In animals, well characterised O-glycosylated proteins include

mucins and collagen, each modified with distinct glycan structures. However, N-linked

glycosylation is the most predominant type of glycosylation, as around 90% of eukaryotic

glycoproteins are modified with N-glycans (Apweiler et al., 1999).

1.2.1. N-linked protein glycosylation in Eukaryotes and Archaea

In eukaryotes, N-glycan structures are first assembled at the cytosolic face of the

endoplasmic reticulum (ER) membrane onto a lipid carrier (dolichol phosphate in

eukaryotes), forming the lipid-linked oligosaccharide (LLO) precursor (Burda and Aebi,

1999). Several glycosyltransferase enzymes encoded by the asparagine-linked

glycosylation (ALG) pathway genes act sequentially to construct this LLO. At the

cytoplasmic face, a glycan structure consisting of two N-acetylglucosamine (GlcNAc)

residues and five mannose (Man) residues is formed by glycosyltransferase enzymes

which utilise soluble nucleotide-activated sugar donors (Breitling and Aebi, 2013). This

intermediate is then flipped across the membrane and additional ALG glycosyltransferases

elaborate this structure within the lumen of the ER to give GlcNAc2 Man9Glc3. Finally, the

oligosaccharyltransferase (OST) complex transfers the constructed glycan en bloc from

the precursor to target polypeptides containing the acceptor tripeptide sequon, N-X-S/T

(where X ≠ P) (Moremen et al., 2012). The OST is a hetero-oligomeric enzyme consisting

of up to nine subunits in higher eukaryotes, although it is the STT3 protein which is

catalytically active (Kelleher et al., 2003; Kelleher and Gilmore, 2005). There are two

isoforms of this subunit, STT3A and STT3B. The former is involved in co-translational N-

glycosylation, where the OST complex associates with the SEC61 translocon and modifies

polypeptides as they are entering the rough ER (Shrimal et al., 2015). An OST complex

containing STT3B then acts on any remaining unmodified N-glycosylation sites in a co-

25

translational or post-translocational manner (Ruiz-Canada et al., 2009; Shrimal et al.,

2015). Although all N-glycoproteins are first modified with an identical N-glycan

structure, this is then extensively remodelled in the Golgi apparatus (Moremen et al.,

2012). The N-glycans can be trimmed and further elaborated, often with residues such as

fucose and sialic acid, resulting in a variety of structures that can be complex and highly

branched.

N-glycosylation was considered a unique feature of eukaryotes until the S layer protein of

an archaea, Halobacterium salinarum, was identified to contain an N-linked glycan

(Mescher et al., 1974). The most characterised archaeal N-glycosylation pathway is now

that of Haloferax volcanii, which is encoded by Agl (archaeal glycosylation) genes (Jarrell

et al., 2014). This N-glycosylation process begins at the cytoplasmic face of the cell

membrane, where sugar residues are transferred to a dolichol-phosphate or dolichol-

pyrophosphate lipid carrier to form LLO (Schwarz and Aebi, 2011). An unusual feature of

this process is that two distinct types of LLO are formed, one containing the first four

residues of the N-glycan and the other carrying a mannose residue which will be the

terminal sugar (Eichler, 2013). Once these LLOs have been flipped across the membrane,

in a similar manner as the eukaryotic pathway, the OST transfers the tetrasaccharide to

asparagine residues within N-X-S/T (X ≠ P) motifs. The archaeal OSTs are a single

enzyme rather than a complex, and are homologous to the eukaryotic STT3 catalytic

subunit (Magidovich and Eichler, 2009). The final residue of the N-glycan is transferred

by the AglS enzyme (Cohen-Rosenzweig et al., 2012). Overall, archaeal N-glycan

pathways and the structures they construct are more diverse than analogous eukaryotic

systems (Schwarz and Aebi, 2011). For example, in Archaea at least twenty types of sugar

residues have been identified in N-linked glycans (Eichler, 2013), with uronic acid

residues appearing as a distinct feature. Furthermore, considering the limited number of

organisms studied so far, this diversity will likely expand considerably in the future

(Eichler, 2013).

1.3. O-linked protein glycosylation in Bacteria

O-linked glycosylation was first identified in Bacteria when a trisaccharide was identified

on the Neisseria meningitidis pilin protein (Stimson et al., 1995). Pilin O-glycosylation

also occurs in Neisseria gonorrhoeae and numerous pilin glycosylation (pgl) genes

encoding glycosyltransferases and sugar biosynthesis enzymes are now known (Aas et al.,

26

2007; Power et al., 2003). It was later demonstrated that Neisseria pilin modification was

performed by a general O-linked glycosylation pathway and at least twelve O-

glycoproteins has been identified (Ku et al., 2009; Vik et al., 2009). Analogous to

eukaryotic N-glycosylation, these pathways consist of OSTs, PglL and PglO in N.

meningitidis and N. gonorrhoeae, respectively, which mediate the block transfer of glycan

onto Ser or Thr residues (Aas et al., 2007; Dell et al., 2010; Faridmoayer et al., 2007). The

O-glycan structure of N. meningitidis consists of two galactose (Gal) residues and either

2,4-diacetamido-2,4,6-trideoxyhexose (DATDH) or 2-glyceramido-4-acetamido-2,4,6-

trideoxyhexose (GATDH) as the reducing end sugar (Chamot-Rooke et al., 2007).

Distinct O-linked glycosylation systems where sugar residues are sequentially added to

flagellar proteins exist in many Bacteria, including Campylobacter (Logan, 2006). C.

jejuni modifies the flagellin protein, FlaA, with as many as 19 O-glycans (Thibault et al.,

2001). These O-glycans are composed mainly of pseudaminic acid, legionaminic acid and

their derivatives and such modification is required for flagellar assembly and motility

(Goon et al., 2003; Logan et al., 2009; Thibault et al., 2001). Genetic loci encoding the O-

glycosylation system display considerable interstrain variation and can contain upto 50

genes (Mcnally et al., 2006). Many of these genes contain homopolymeric tracts which

likely alter glycan structure due to phase variation, as demonstrated for Cj1295 (Hitchen et

al., 2010). However, this complex system and its numerous glycosyltransferases remain

relatively uncharacterised.

1.4. Overview of N-glycosylation in Bacteria

In 2002, the discovery of an N-linked glycosylation system in C. jejuni (Wacker et al.,

2002; Young et al., 2002) showed that this process exists in all three domains of life.

General similarities can be drawn between the N-glycosylation systems of Bacteria and

Eukaryotes. In Bacteria, N-glycan structures are constructed as LLO precursors at the

cytoplasmic face of the inner membrane, flipped into the periplasm and a homologue of

the eukaryotic STT3 subunit then catalyses transfer of N-glycan structures en bloc onto

target sequences. In contrast to the eukaryotic system, Bacterial N-glycan structures

transferred by an OST can differ in a species specific manner. However, the extensive

downstream processing applied to eukaryotic N-glycans is not a feature observed in

Bacteria (Dell et al., 2010).

27

A striking aspect of Bacterial N-glycosylation is that, unlike in eukaryotes and archaea

where the process is universal (Dell et al., 2010), it only occurs in some species. Thus far,

N-glycosylation has been identified mainly in Gram-negative Bacteria, and only within

some species of the epsilonproteobacteria (Campylobacter, Helicobacter, Wolinella),

deltaproteobacteria (Desulfovibrio) and gammaproteobacteria (Haemophilus influenzae)

(Dell et al., 2010). However, the Gammaproteobacterial pathway is unique as it occurs in

the cytoplasm and does not utilise an LLO and hence likely evolved separately (Nothaft

and Szymanski, 2013). This pathway was identified in Haemophilus influenzae where an

adhesin, HMW1, was found to be modified with mono or di-hexose N-glycans (thought to

contain glucose and galactose residues) at up to 31 sites (Grass et al., 2010; Gross et al.,

2008).

1.5. N-linked glycosylation in Campylobacter

1.5.1. Campylobacter jejuni N-glycosylation.

The N-glycosylation locus of C. jejuni was first described as the wla locus and thought to

be involved in LPS biosynthesis (Fry et al., 1998). However, it was soon recognised to

encode a protein glycosylation system and thus renamed the pgl locus (Szymanski et al.,

1999). The original LPS biosynthesis association was likely due to the presence of an

upstream gene, waaC, which can indeed modify LOS (Klena et al., 1998). In addition, the

pgl locus contains a bifunctional UDP-GlcNAc/ UDP-glucose 4-epimerase, gne (also

known as galE), which provides sugar donor substrates for the LOS, capsular

polysaccharide and protein glycosylation pathways (Bernatchez et al., 2005; Fry et al.,

2000).

Soybean Agglutinin (SBA) lectin affinity purification was used to identify the first

putative N-glycoproteins of C. jejuni, PEB3 and CgpA, and as SBA is an N-

acetylgalactosamine (GalNAc) -specific lectin this suggested the N-glycan structure

contained this sugar (Linton et al., 2002). The 16 kb pgl locus can be functionally

transferred into E. coli allowing recombinant N-glycoprotein production (Wacker et al.,

2002). Using such ‘glycocompetent E. coli’ and tandem mass spectrometry (MS), the N-

glycan structure produced by the C. jejuni pgl pathway was identified as a heptasaccharide

consisting of five N-acetyl-hexosamine (HexNAc) residues, a 2,4-diacetamido-2,4,6-

trideoxyhexose (DATDH) and a branching hexose (Hex) (Wacker et al., 2002). Nuclear

28

magnetic resonance (NMR) analysis of SBA-purified C. jejuni N-glycoproteins confirmed

the N-linked glycan structure as GalNAc-α1,4-GalNAc-α1,4-[Glcβ1,3-]GalNAc- α1,4-

GalNAc-α1,4-GalNAc- β1,3-Bacillosamine-β1 (Young et al., 2002). Bacillosamine refers

to 2,4-diacetamido-2,4,6-trideoxy-β-D-glucopyranose and is correctly known as Di-N-

acetylbacillosamine (DiNAcBac). This heptasaccharide structure is conserved among

members of this species and over 60 C. jejuni N-glycoproteins have been experimentally

identified (Scott et al., 2014, 2011; Young et al., 2002). However, bioformatics analyses

suggest that approximately 150 C. jejuni proteins may be N-glycosylated (Frost, 2015;

Scott et al., 2011).

1.5.1.1. The C. jejuni N-linked glycosylation pathway

The C. jejuni N-glycosylation process has been studied extensively and, using data

acquired from in vitro assays and pgl gene mutagenesis both in C. jejuni and the

glycocompetent E. coli system, the role of Pgl enzymes is now well understood (Fig. 1.1).

Biosynthesis of the C. jejuni N-linked glycan first occurs in the cytoplasm, where the

structure is assembled onto undecaprenyl pyrophosphate (UndP) to create the LLO

precursor. Pgl glycosyltransferases mediate this process and although soluble, they are

relatively hydrophobic and likely interact with the cell membrane (Glover et al., 2005).

All Pgl glycosyltransferases utilise nucleotide sugars as donors and thus the pgl locus

encodes enzymes responsible for their biosynthesis. A three stage reaction catalysed by

PglDEF creates UDP-DiNAcBac (Olivier et al., 2006), providing substrate for PglC to

transfer the first residue of the N-glycan, DiNAcBac, to the undecaprenyl-phosphate

(UndP) precursor (Glover et al., 2006; Linton et al., 2005). PglAJH then act sequentially

to transfer the five GalNAc residues, respectively, with PglH adding the final three

residues (Glover et al., 2005; Troutman and Imperiali, 2009). PglAJH all utilise UDP-

GalNAc as a donor substrate which is provided by the bifunctional epimerase, Gne

(Bernatchez et al., 2005). Finally, transfer of the β 1,3-linked branching glucose residue is

catalysed by the glucosyltransferase PglI (Glover et al., 2005; Kelly et al., 2006; Linton et

al., 2005). In vitro assays using purified C. jejuni PglH found it was a polymerase with a

single active site and that its affinity for the N-glycan increases with increasing GalNAc

residues (Troutman and Imperiali, 2009). It is thought that upon PglH transferring three

GalNAc residues the reaction is inhibited until PglI competes for the product and transfers

the final branching residue, blocking further PglH binding (Troutman and Imperiali,

2009). The complete N-glycan precursor is then flipped across the inner membrane into

the periplasm by an ABC transporter, PglK (Kelly et al., 2006). The inner membrane

29

protein OST, PglB, uses this LLO as substrate and transfers the heptasaccharide structure

onto target asparagine residues in protein acceptor sequences (Wacker et al., 2002). The

acceptor sequon in C. jejuni is D/E-X-N-X-S/T (where X ≠ P), which is an extended

version of the eukaryotic equivalent (Kowarik et al., 2006b).

C. jejuni also produces free oligosaccharide (fOS) structurally identical to its N-glycan in

a PglB-dependent manner (Liu et al., 2006; Nothaft et al., 2009). This hydrolytic activity

of PglB requires the WWDYG motif known to be essential for N-linked protein

glycosylation (Nothaft et al., 2009). A reduction in free oligosaccharide release due to

altered PglB activity was observed in response to high salt concentrations or sucrose

(Nothaft et al., 2009). In contrast, the extent of N-linked glycosylation appeared

unchanged under these conditions. This indicated that fOS may be similar to

osmoregulated periplasmic glucans (OPGs) known in other proteobacterial subsets

(Nothaft et al., 2009). OPGs are found in alpha, beta and gamma proteobacteria and can be

30

cyclic or linear glucose polymers (Bohin, 2000). An absence of OPGs can cause

pleiotropic effects and are associated with the virulence of many pathogens (Bontemps-

Gallo and Lacroix, 2015). OPGs have been suggested to be involved in osmoadaptation,

cell signalling, cell envelope integrity and cell division (Bontemps-Gallo et al., 2017).

However, the function of fOS in C. jejuni remains unclear. Nevertheless, it appears to be a

prominent feature in C. jejuni as it accounts for around 2.5 % of its dry cell weight

(Dwivedi et al., 2013).

1.5.1.2 The Campylobacter oligosaccharyltransferase, PglB

Campylobacter PglB enzymes are encoded on the genomes of all but one Campylobacter

species, Campylobacter canadensis (Nothaft et al., 2012). They are members of the

glycosyltransferase superfamily GT-C, which include glycosyltransferases that harbour

between 8 and 13 transmembranes helixes and contain a modified DxD catalytic motif

(Liu and Mushegian, 2003). The X-ray structure of C.lari PglB with an acceptor peptide

was solved and demonstrated that its periplasmic and transmembrane domains contribute

to acceptor sequon binding and catalysis (Lizak et al., 2011). This work also identified

amino acid residues key for catalysis. Although the full X-ray structure of C. jejuni PglB

is unavailable, mutagenesis studies have similarly revealed the importance of the

corresponding amino acids for binding and catalysis (Nothaft and Szymanski, 2013). The

substrate specificity of C. jejuni PglB is relatively relaxed and it is able to transfer

truncated C. jejuni N-glycan structures to protein (Linton et al., 2005). Furthermore, PglB-

mediated transfer of diverse E. coli- and Pseudomonas aeruginosa-derived O-antigen

structures onto protein was observed in E. coli (Feldman et al., 2005).

The N-glycosylation sequon that PglB modifies is D/E-X-N-X-S/T (where X ≠ P)

(Kowarik et al., 2006b) and in vitro assays suggest that DQNAT results in optimal N-

glycosylation (Chen et al., 2007). The requirement for the extended sequon in comparison

to the eukaryotic counterpart, N-X-S/T, was structurally explained when Lizak et al.

(2011) identified that residue R331 of C. lari PglB forms a stabilising salt bridge with the

-2 Asp. This residue is conserved in Bacteria (Lizak et al., 2011) and the equivalent

residue (R328) in C. jejuni PglB has since been implicated in restricting the sequon

specificity also (Ollis et al., 2014). However, using glycopeptide enrichment and tandem

MS, rare instances of atypical sequons have been identified in N-glycoproteins from C.

jejuni 11168O (the ‘original’ 11168 isolate) (Scott et al., 2014). Three examples of non-

canonical sequons were found and had leucine or glutamine at the -2 position or alanine at

the +2 position.

31

It was first thought that C. jejuni PglB modified folded proteins and could only modify N-

glycosylation sequons present at exposed flexible loops (Kowarik et al. 2006). However,

more recent crystal structures of C. jejuni N-glycoproteins demonstrates that N-

glycosylation sequons can lie within alpha-helices or structured turns (Kawai et al., 2012;

Rangarajan et al., 2007; Silverman and Imperiali, 2016). Furthermore, in vivo analyses by

Silverman & Imperiali (2016) suggest that N-linked glycosylation is coupled with Sec

pathway-mediated protein translocation and hence N-glycoproteins can be modified prior

to complete folding. This is reminiscent of the eukaryotic system, where the OST complex

can couple with the SEC61 translocon resulting in co-translational N-glycosylation of

nascent polypeptides as they enter the ER (Shrimal et al., 2015). C. lari PglB itself is an

N-glycoprotein, harbouring two glycosylation sites which are modified in glycocompetent

E. coli (Lizak et al., 2011). The PglB of C. jejuni also contains an experimentally

identified N-glycosylation sequon (Scott et al., 2011).

1.5.2. The role of N-glycosylation in C. jejuni

In eukaryotes, N-glycosylation has numerous functions and is intricately involved in

protein folding, quality control, targeting and secretion (Helenius and Aebi, 2001;

Moremen et al., 2012). In C. jejuni, the specific function(s) of N-linked glycosylation are

unknown, however, loss of this modification has pleiotropic effects and impairs virulence

(Nothaft and Szymanski, 2013). For example, pglH and pglB mutants have reduced ability

to adhere to and invade human epithelial cells in vitro and colonise the gastrointestinal

tracts of chickens in vivo (Jones et al., 2004; Karlyshev et al., 2004; Szymanski et al.,

2002). Furthermore, mutation of pglE or pglF, which are involved in the biosynthesis of

DiNABac residue, was also shown to reduce colonisation in chickens (Hendrixson and

DiRita, 2004). However, the mechanism through which N-linked glycosylation impairs C.

jejuni virulence is unclear.

There are limited studies regarding the function conferred by N-linked glycosylation of

individual C. jejuni N-glycoproteins and most studies have failed to identify a requirement

for this modification. Thus far, influence of N-glycosylation on protein function has only

been demonstrated for VirB10 (Larsen et al., 2004), a protein component of the plasmid-

encoded type IV Secretion System (T4SS) of C. jejuni 81-176. Larsen et al. (2004)

demonstrated that N-glycosylation of VirB10 significantly contributes to natural

competence as loss of VirB10 or its N-glycosylation conferred similar reductions in

natural transformation efficiency. In contrast, abolishing N-glycosylation of a zinc

transporter system component, ZnuA, did not affect its function (Davis et al., 2009). In

32

addition, N-glycosylation did not influence the antigenicity or abundance of a surface-

exposed lipoprotein, JlpA (Scott et al., 2009). Futhermore, N-glycan modification was not

vital for the activity of two mechanosensitive channel N-glycoproteins, Cj1025 and

Cj0263 (Kakuda et al., 2012). Finally, C. jejuni expressing a non-glycosylated version of

Cj1496c, a periplasmic N-glycoprotein which contributes to adherence and invasion in

vitro and chick colonisation, did not give a similar phenotype to the gene knockout strain

(Kakuda and DiRita, 2006).

As discussed, current experimental evidence does not suggest that N-glycosylation is

required for the function of specific proteins. Instead, N-linked glycosylation may have a

broader function. For example, Alemka et al. (2013) demonstrated that N-glycosylation

may protect proteins from degradation by chicken gut proteases. In this study, chicken

cecal contents reduced the fitness or growth of a C. jejuni pglB mutant compared to the

wildtype and this phenotype could be partially complemented by addition of protease

inhibitors (Alemka et al., 2013). Another study has found that an Fc fusion protein

containing the extracellular region of the human Macrophage Galactose-type lectin (MGL)

receptor can bind the terminal GalNAc residues of C. jejuni N-glycoproteins (van Sorge et

al., 2009). Furthermore, the authors found that, in comparison to wildtype cells, pglA,

pglH or pglJ mutants induced higher production of interleukin 6 by human dendritic cells

in vitro, suggesting that Campylobacter N-glycans may dampen the host immune response

(van Sorge et al., 2009).

1.5.3. N-glycosylation in other Campylobacter species.

N-linked glycosylation is found in all Campylobacter species excluding C. canadensis and

the general features of the C. jejuni pgl locus are usually conserved in similarly organised

loci (Jervis et al., 2012; Nothaft et al., 2012). However, in many instances, distinct

predicted glycosyltransferase and sugar biosynthesis genes are also present. In line with

this, the N-glycans produced by Campylobacter species are structurally diverse (Jervis et

al., 2012; Nothaft et al., 2012)(Fig. 1.2). However, there is some conservation of N-glycan

structure among certain species. In addition, Nothaft et al. (2012) found that eighteen

Campylobacter species produce fOS that is structurally identical to their corresponding N-

glycans. Interestingly, DiNAcBac has been identified as the reducing end sugar in every

Campylobacter N-glycan analysed thus far. Furthermore, DiNAcBac is unique to a subset

of bacterial pathogens including Neisseria gonorrhoeae and Acinetobacter baumannii,

prompting suggestion of a putative role in virulence (Morrison and Imperiali, 2014). In

33

line with this, Schutter et al. (2017) have recently described inhibitors active against the

PglD enzyme.

A C. jejuni N-glycan reactive antiserum, hR6 (Amber and Aebi, unpublished), reacts with

all the thermotolerant Campylobacter species, C. jejuni, C. coli, C. upsaliensis and C. lari

(Jervis et al., 2012; Nothaft et al., 2012). Furthermore, members of this group contain pgl

loci most closely resembling that of C. jejuni. Unsurprisingly, C. coli, C. upsaliensis and

C. helveticus are now known to produce an identical heptasaccharide N-glycan as C.

jejuni (Jervis et al., 2012; Nothaft et al., 2012). However, C. lari lacks a pglI homologue

and thus it produces a hexsaccharide N-glycan without the branching glucose found in the

N-glycan of the other thermotolerant species.

The hR6 antiserum is not reactive with Campylobacter species more distantly related to C.

jejuni, such as C. fetus and C. concisus, and the pgl loci of these species are more variable

(Jervis et al., 2012). For example, some Campylobacter species including C. concisus

have two pglB homologues, referred to as pglB1 and pglB2 (Jervis et al., 2012). In C.

concisus, insertional mutagenesis of these homologues has indicated that pglB1 is

responsible for the majority of N-glycosylation activity but the role of pglB2 remains

unclear (Frost, 2015). The variation in Campylobacter pgl pathways means that the N-

glycan structures identified thus far exhibit much diversity, with some species known to

produce more than one N-glycan structure. Notably, four distinct N-glycan structures have

been found in the human gut commensal bacterium, C. hominis (Nothaft et al., 2012).

There have been conflicting reports regarding the glycan structure produced by the pgl

pathway of C. concisus. Nothaft et al. (2012) reported a fOS structure including a 234 Da

residue from C. concisus 13826. In contrast, Jervis et al. (2012) identified in vitro–

generated glycopeptides harbouring a 217 Da residue-containing N-glycan using

membrane preparations from C. concisus NCTC 11485. Nothaft et al. (2012) proposed

that these differences may be due to intraspecies diversity, which is broadly known as a

feature of C. concisus (Deshpande et al., 2013). C. ureolyticus, an emerging human

pathogen, has been demonstrated to produce an identical N-linked glycan to C. concisus

13826 (Nothaft et al., 2012). However, the pgl locus of C. concisus contains more

predicted glycosyltransferase genes than in C. ureolyticus which suggests that more than

one N-glycan structure might be produced by C. concisus.

34

35

1.5.3.1. N-linked glycosylation in C. fetus

C. fetus contains homologues of all C. jejuni pgl genes excluding pglI. In addition, there

are two additional open reading frames that putatively encode a glycosyltransferase and a

sugar biosynthesis enzyme (Jervis et al., 2012). The pgl locus of C. hyointestinalis is

similar to that of C. fetus and identical N-glycan structures have been identified in both

species (Jervis et al., 2012; Nothaft et al., 2012). Using an in vitro N-glycosylation assay

and MALDI LIFT- TOF/TOF MS, Jervis et al (2012) observed the production of a

hexasaccharide N-glycan which appeared similar to that of C. jejuni but lacked a terminal

HexNAc residue. However, Nothaft et al. (2012) combined LC MS/MS and NMR to

analyse C. fetus fOS and identified that, in contrast to C. jejuni, C. fetus produce two

hexasaccharide fOS structures which contain GlcNAc sugars and differ at the non-

reducing end. These structures, referred to here as type 1 and type 2, were α-GlcNAc-6-[

β-Glc-3]-α-GlcNAc-4-α-GlcNAc-4-α-GalNAc-3-α,β-diNAcBac and α-GlcNAc-6-[β-

GlcNAc-3] -α-GlcNAc-4-α-GlcNAc-4-α-GalNAc-3-α,β-diNAcBac, respectively. MS

analysis of C. fetus N-glycopeptides confirmed that two types of N-glycan are transferred

to protein: HexNAc-[HexNAc]-HexNAc3-diNAcBac and HexNAc-[Hex]-HexNAc3-

diNAcBac. These structures most likely contain identical sugar residues as the type 1 and

type 2 fOS structures, respectively. Furthermore the glycoforms were found at a ratio of

~4:1, with the type 2 GlcNAc branch-containing structure predominant both in fOS and N-

glycopeptides (Nothaft et al., 2012). The authors suggested that having two N-glycan

structures may provide protection from the host immune system (Nothaft et al., 2012),

which is thought to be the case for O-linked glycosylation in Neisseria (Borud et al.,

2014).

Excluding N-glycan structural analysis, there are limited data regarding the C. fetus pgl

pathway. However, analysis of C. fetus glycopeptides identified 32 unique N-

glycosylations sites consisting of the C. jejuni-type sequon, D/E-X-N-X-S/T (where X ≠

P) (Nothaft et al., 2012). These N-glycopeptides correspond to 25 N-glycoproteins.

Furthermore, Nothaft et al. (2012) used C. fetus fOS to generate polyclonal antisera,

termed GRPII-1 and GRPII-2, reactive against the type 1 and type 2 structures,

respectively. Immunoblotting demonstrated these antisera to react with numerous

presumed N-glycoproteins in C. fetus cell lysates, suggesting a similar extent of protein N-

glycosylation in C. fetus as in C. jejuni.

36

1.5.4. N-linked glycosylation in closely related genera, Helicobacter and Wolinella

N-glycosylation systems similar to those of Campylobacter are now known in other

epsilonproteobacteria, such as Helicobacter and Wolinella succinogenes. PglB othologues

have been identified in three Helicobacter species, H. pullorum, H. canadensis and H.

winghamensis, of which H. pullorum is most characterised (Jervis et al., 2010). H.

pullorum was first identified from poultry and humans with gastroenteritis (Stanley et al.,

1994). This species is considered an emerging enterohepatic pathogen (Javed et al., 2017)

and has also been associated with Crohn’s disease, chronic gallbladder inflammation and

bactaeremia (Apostolov et al., 2005; Bohr et al., 2004; Tee et al., 2001). However, its

pathogenicity is debated as Ceelen et al. (2005) found that H. pullorum prevalence in

human stool did not vary between healthy individuals and those with gastrointestinal

disease. Interestingly, H. pullorum contains two pglB orthologues, pglB1 and pglB2,

which are not found within an ordered locus (Jervis et al., 2010). H. pullorum PglB1 could

N-glycosylate two of four sequons present in the C. jejuni N-glycoprotein, Cj0114, in

glycocompetent E. coli (Jervis et al., 2010). Furthermore, using an in vitro assay, the N-

glycan structure transferred by PglB1 was identified as a pentasaccharide containing

unknown 216 and 217 Da sugar residues and lacking the common reducing end sugar,

DiNAcBac (Jervis et al., 2010). However, the role of PglB2 remains unclear and, unlike

pglB1, attempts to inactivate pglB2 in H. pullorum failed, suggesting that it may be

essential (Jervis et al., 2010).

W. succinogenes is a non-pathogenic species originally isolated from a bovine rumen

(Tanner et al., 1981; Wolin et al., 1961). This species contains a pgl cluster similarly

ordered as in C. jejuni but with additional predicted glycosyltransferases present (Baar et

al., 2003). As with C. jejuni, a single pglB gene is present and the reducing end sugar of its

N-glycan is DiNAcBac (Dell et al., 2010). However, the N-glycan is a hexasaccharide

which, similarly to H. pullorum, contains a 216 Da sugar and an uncharacterised 232 Da

sugar (Dell et al., 2010).

1.5.5. N-linked glycosylation in hydrothermal vent Bacteria

Most recently, pglB homologues were identified in proteobacteria which dwell in

hydrothermal vents and limited research has begun to gain insight into the activity of these

enzymes (Mills et al., 2016). Deferribacter desulfuricans, Nitratiruptor tergarcus and

Sulfurovum lithotrophicum contain functional PglB homologues which can, to some

extent, complement C. jejuni PglB activity in glycocompetent E. coli (Mills et al., 2016).

However, the PglB enzymes of these thermophilic Bacteria displayed stricter specificity

37

for target proteins compared to C. jejuni PglB. The OSTases of N. tergarcus and S.

lithotrophicum had a similar requirement for a negatively charged residue at the -2

position of an acceptor sequon. The D. desulfuricans PglB displayed low activity in E. coli

and its requirements could not be studied further, however, it lacks the R331 residue

implicated in C. jejuni PglB sequon specificity and thus likely does not require the

extended sequon (Mills et al., 2016). Such an observation was made for the PglB enzyme

encoded by Desulfovibrio desulfuricans which, like Deferribacter, is a sulfate-reducing

member of the delta-proteobacteria (Ielmini and Feldman, 2011). The N-glycan structures

synthesised by the Pgl systems of these deltaproteobacterial species are unknown.

Nevertheless, the discovery of C. jejuni PglB homologues in these species further

highlights the question of why N-glycosylation exists in such an ecologically diverse

subset of Bacteria. Mills et al. (2016) suggested that in these deep sea vent Bacteria N-

glycosylation may provide thermostability and protection against high osmolarity.

1.6. Campylobacter detection/identification

1.6.1. Culture and phenotypic testing

Isolation and identification of Campylobacter from samples is a particularly time-

consuming process which involves enrichment, solid microbiological culture isolation of

the organism and then further testing of Campylobacter-like colonies (Oyarzabal and

Fernández, 2016). Samples with relatively low numbers of Campylobacter cells, such as

food or environmental samples, are first enriched by incubating in specialised broth such

as Bolton or Preston for at least 24 hours (Kim et al., 2009). Enriched samples are then

plated onto solid culture media. For the identification of Campylobacter in a clinical

sample, Public Health England guidelines suggest microaerobic incubation on

Campylobacter selective media such as charcoal cefoperazone deoxycholate agar (CCDA)

for 48 hours at 42 °C (PHE, 2014). Cells from a Campylobacter-like colony are then

examined microscopically to confirm a curved or ‘S’ shaped rod. This is followed by

Gram staining and an oxidase test to confirm the identity as Campylobacter. Therefore,

confirming a Campylobacter species is present using this method takes at least a few days.

In order to speciate the identified organism, PHE recommend further biochemical testing

(either manually or using commercial kits), MALDI-TOF MS and/or PCR.

38

The well-established protocol described above is optimal for the isolation of C. jejuni and

C. coli from samples. However, other Campylobacter species can have different growth

requirements and isolation is sub-optimal using such procedures. For example, a German

hospital identified a “pseudo-outbreak” of C. concisus upon introducing hydrogen into the

culture environment (Casanova et al., 2015). Furthermore, C. fetus and C. upsaliensis are

susceptible to the concentrations of cefoperazone present in CCDA selective media

(Kulkarni et al., 2002; Schulze et al., 2006). Therefore, modified methods are required to

isolate all possible Campylobacter species from a sample.

1.6.2. Molecular methods for detecting Campylobacter in poultry

As traditional culture methods for the detection and identification of Campylobacter spp.

have limitations, molecular methods of identification are often adopted. Such techniques

have been applied for detecting Campylobacter spp. from a variety of samples, including

human stool, poultry faeces and meat. Molecular methods either target DNA or proteins.

DNA-based technologies adapted for Campylobacter detection include polymerase chain

reaction (PCR) and loop-mediated isothermal amplification (LAMP) (Dong et al., 2014;

Gharst et al., 2013). Techniques targeting proteins utilise antibodies and include latex

agglutination tests (LATs) and enzyme immunoassays (EIAs) (Gharst et al., 2013). A

recent project undertaken by the Department for Environment, Food and Rural Affairs

(DEFRA) (code: OZ0621) concluded that there is a significant need for a rapid on-farm

test to assist Campylobacter control strategies. The available methods for detection of

Campylobacter spp. will be discussed in this context.

Several distinct PCR-based methods have been implemented for Campylobacter detection

and identification, including single reaction PCR, multiplex PCR and quantitative PCR.

Commonly used target sequences of Campylobacter include ribosomal DNA and the

flagellin genes, flaA and flaB (Kirk and Rowe, 1994; Linton et al., 1997; Oyofo et al.,

1992). Muliplex PCR (mPCR) uses more than one primer pair per reaction which can, for

example, allow for simultaneous identification of different Campylobacter species. Kamei

et al. (2016) described an mPCR which targets the CDT and can identify and differentiate

between C. jejuni, C. coli, C. upsaliensis, C. lari, C. fetus and C. hyointestinalis in pure

culture. Quantitative PCR (qPCR) (also known as Real Time PCR) involves coupling

fluorescent detection techniques with conventional PCR to allow amplification to be

monitored in ‘real time’ (Heid et al., 1996). This method is faster than traditional PCR as

it negates the need for further processing following the reaction, such as gel

electrophoresis (Jensen et al., 2005). De Boer et al. (2015) combined three qPCR assays to

39

detect Campylobacter spp. in 926 broiler faeces samples and found that 83% of samples

were positive in comparison to the 65% detected by traditional culture methods. PCR

based assays can offer such higher sensitivity due to detecting viable but non-culturable

(VBNC) or dead Campylobacter cells (Gharst et al., 2013). However, the sensitivity can

decrease significantly when using complex sample matrices, thus it is often advisable to

enrich samples and extract DNA before analysis (Gharst et al., 2013). For example, a 24

hour enrichment step followed by DNA extraction is recommended when using the

commercially available qPCR assay, iQ-Check® Campylobacter (BIO-RAD). Another

limitation of PCR methods is that they are often less effective for the detection of the so-

called ‘emerging’ Campylobacter species as they are usually optimised for the

identification of C. jejuni and C. coli (E.g. The iQ-Check® assay cannot detect C.

upsaliensis). Thus, these techniques are not promising tools for on-site detection of

Campylobacter species in poultry.

Loop-mediated isothermal amplification (LAMP) is a highly sensitive method in which a

gene can be amplified using a single 60-65 °C step (Notomi et al., 2000). Furthermore, a

positive result can be seen by the naked eye due to the production of a turbid solution

(Hara-Kudo et al., 2005). Dong et al. (2014) applied a LAMP assay for the detection of C.

jejuni from enriched cattle faeces and found that the assay was less sensitive to inhibitors

found in the enrichment broth in comparison to PCR. However, the reported sensitivity

was 84 %. Therefore, although such methods are relatively rapid, they still require

significant sample preparation and in practice do not offer high sensitivity. Thus, these

methods are not highly suited to point-of-care testing in a farm environment.

In general, DNA-based methods tend to have higher sensitivity than antibody-based tools

(Platts-Mills et al., 2014). However, the former usually requires specialised equipment and

hence samples need transporting to a laboratory for analysis. As a result, this incurs further

cost and time. In contrast, antibody-based methods or immunological methods are simpler

and produce rapid results. Therefore, antibody-based methods are appropriate for

analysing samples on-site.

Several types of immunological assays for the detection of Campylobacter spp. have been

described in the literature. Such techniques can utilise polyclonal or monoclonal

antibodies. Polyclonal antibodies describe a set of antibodies which although all reactive

against one antigen, have varying specificities and affinities (Boenisch, 2009). In contrast,

monoclonal antibodies are those produced from a single B-cell clone and all bind identical

epitopes. Polyclonal antiserum is relatively simple to produce by immunizing animals,

40

such as rabbits, with an antigen and later harvesting the sera from its blood. The

production of monoclonal antibodies is more complex and time consuming, as it requires

isolation of an animals B cells, fusion with a myeloma cell line (to confer immortality) and

subsequent screening to identify the desired antibody-producing cell line (Boenisch,

2009).

Latex agglutination tests (LATs) are well established antibody-based tests that have long

been used to confirm pathogen identification from clinical samples. The principle involves

coupling antibodies (commonly polyclonal) to latex so that the formations of antigen-

antibody complexes are made visible to the naked eye. The application of this method for

Campylobacter identification was first reported by Hodinka and Gilligan (1988). More

recently, Miller et al. (2008) reviewed three such LAT assays marketed for Campylobacter

detection: Microgen M46 Campylobacter (Microgen Bioproducts Ltd.), CAMPY (jcl)

(Scimedx Corporation) and Dryspot Campylobacter (Oxoid). However, the Microgen and

CAMPY (jcl) assays gave false positives when analysing Acinetobacter baumanii, an

organism commonly co-isolated with Campylobacter (Oyarzabal et al., 2005). In addition,

none of the assays were able to identify all eight C. upsaliensis strains tested. Furthermore,

LATs are prone to non-specific agglutination. For example, Hazeleger et al. (1992) found

commercial LAT assays to give unacceptable amounts of non-specific agglutination when

testing against enriched stool samples. The authors also reported that the limit of detection

could vary considerably in a strain- and species-dependent manner, varying from 7 x 105 –

5 x 108 CFU/ml. Hence, it has been suggested that these assays are useful for confirming

culture-based identification but not for independent use (Gharst et al., 2013).

More robust antibody-based detection methods are enzyme immunoassays (EIAs), which

involve coupling antibodies to enzymes to afford a colour change as a result of the

antibody-antigen interaction. These include enzyme-linked immunosorbent assays

(ELISAs) in microplate formats and lateral flow immunoassays. The latter are modified

EIAs contained within a plastic casing commonly referred to as a ‘dipstick’ (Oyarzabal

and Battie, 2012). Lateral flow assays (LFAs) often contain antibodies labelled with

colloidal gold nanoparticles, which upon aggregating on the test line produce a visible

colour change (Sajid et al., 2015). Although ELISA formats allow for quantification,

LFAs produce faster results and are simpler to use (Granato et al., 2010).

There are commercial automated ELISA systems, VIDAS® and miniVIDAS®

(bioMérieux), marketed for the identification of Campylobacter from food samples.

However, the sensitivity of these devices relies on 48 hour enrichment of samples (Liu et

41

al., 2009). Furthermore, they are not portable and hence not applicable to on-farm testing.

Non-automated commercial ELISA formats also exist for the detection of Campylobacter,

including RIDASCREEN® Campylobacter (R-BioPharm AG), Premier® CAMPY

(Meridian Bioscience) and ProSpecT™ Campylobacter (Remel). However, these assays

can suffer from poor sensitivity and considerable false positive results are often reported

(Oyarzabal and Battie, 2012; Tissari and Rautelin, 2007). For example, studies have

reported ProSpecT to have a sensitivity and specificity as low as 62% and 84%,

respectively, when analysing stool samples (Hindiyeh et al., 2000; Tribble et al., 2008).

Furthermore, Giltner et al. (2013) found that Premier® CAMPY (Meridian Biosciences)

gave considerable false positives, and warned that although it was a fast alternative to

routine identification procedures, all positive results should be subsequently confirmed

with culture. In addition, these EIAs are in microtiter plate format which would not be

ideal for on-site detection at farms.

Lateral flow assays are the most applicable type of immunoassay for point-of-care testing,

providing rapid results alongside being extremely user-friendly. Commercially available

LFAs for the detection of Campylobacter include RIDA®QUICK, ImmunoCard STAT!,

Xpect, SinglePath and NH Immunochromato. These assays are marketed for analysis of

clinical stool samples or food samples. Although most suggest pre-enrichment of samples

some assays, such as ImmunoCard STAT and RIDAQUICK, are intended to analyse

patient stool samples directly, requiring only dilution of sample with reagents provided.

However, a recent study conducted by the CDC ruled that available immunoassays were

not appropriate for standalone Campylobacter spp. identification from stool due to

unacceptably low sensitivities and positive predictive values (PPVs; the probability that a

positive result will be a true positive) (Fitzgerald et al., 2016). Specifically, this work

analysed two microplate format assays, Premier and ProSpecT, and two LFAs,

ImmunoCard STAT and Xpect. Nevertheless, Oyarzabal and Battie (2012) argue that

optimisation of such assays may allow for considerable improvement. Indeed, many of the

commercial assays utilise anti-Campylobacter polyclonal antisera for which no further

details are available. As the targets of these sera are not publicly known, it can be argued

that more focused research onto appropriate target antigens may be necessary to create

more efficacious assays. Furthermore, the use of culture methods as a reference from

which to calculate sensitivity and specificity of other assays has been raised into question.

For example, VBNC cells of Campylobacter may be detected by non-culture-based tests

and due to comparison with a reference culture method, this result may be construed as a

false positive. In evidence of this, when analysing available EIAs, Granato et al., (2010)

42

found that upon using RT-PCR to further analyse contradictory results between EIAs and

culture, the PPV of the EIAs improved (i.e. less false positives were apparent).

Furthermore, Gómez-Camarasa et al. (2014) found that the PPV of RIDAQUICK changed

from 77% to 94% when including PCR in the reference method. This assay performed

well when analysing stool sample with minimal preparation, further suggesting that LFAs

are promising for on-site analysis at farms. However, analysis of the species specificity of

this assay has not been published.

Although there are numerous assays described for the detection and identification of

Campylobacter species, manufacturers usually recommend an enrichment step be included

to concentrate samples prior to analysis (Gharst et al., 2013). The only available assay

targeted for Campylobacter detection directly from poultry samples is SinglePath Direct

Campylobacter poultry kit (Merck). Merck claims this test can be performed in less than 2

hours and provides a sensitivity and specificity of 96% and 100%, respectively, when

analysing chicken caecal droppings (n = 120). The limit of detection is reported at 107

CFU/ml, which allows for differentiation between ‘Campylobacter high and low risk

flocks’. Targeting flocks with a high burden may be a feasible option for reducing chicken

meat-associated human Campylobacteriosis, as a risk assessment study found that

incidence could be 30 times lower if the level of Campylobacter on chicken carcasses was

reduced by 2 log (Rosenquist et al., 2003). However, this product is no longer available to

purchase. Wadl et al. (2009) reported the development of an LFA capable of detecting 7.2

Log CFU/g C. jejuni cells in chicken faeces. As chickens are often reported to shed up to

107-10

8 CFU/g of

C. jejuni cells in their faeces, this test seemed feasible. However, in

reality the assay did not perform well, only identifying 3 (17%) of the 18 infected flocks

tested.

The Food Standards Agency (FSA) recently published results of a project which aimed to

identify an appropriate method for detecting Campylobacter-infected flocks on-site

(Madden et al., 2016). The project analysed commercially available applications including

two lateral flow devices (SinglePath and ImmunoCard STAT! CAMPY), a LAMP

(Loopamp) kit and a qPCR (Mericon Campylobacter) kit. It was established that the LFDs

did not have sufficient sensitivity and specificity whereas the LAMP kit produced false

positives with Arcobacter spp., bacteria commonly found in chickens. This study

concluded that the qPCR kit was most suitable overall for rapidly detecting

Campylobacter infected-flocks. However, it was not able to be performed on-site and

hence farmers were not informed of the results until after at least 36 hours. Therefore,

there remains a need for rapid detection methods which can be performed on-site for the

43

detection of Campylobacter. Furthermore, ease-of-use would be a significant advantage,

meaning farmers can complete the test unaided. Hence, the development of more sensitive

and specific LFAs would be ideal.

Overall, EIAs for the identification of Campylobacter tend to suffer from poor sensitivity

and often give a high rate of false positives. Furthermore, the majority available are

marketed for the identification of C. jejuni and C. coli from clinical specimens and/or

enriched food samples. Therefore, there is still a requirement for tools which are optimised

to provide rapid results for Campylobacter monitoring on-site and also able to detect the

more ‘emerging’ pathogens within the genus.

1.6.3. Methods of detecting Campylobacter fetus in cattle.

Traditionally, C. fetus infection was identified solely using culture methods which,

contrary to those used for other Campylobacter species, includes incubation at 37 °C and

5% hydrogen if possible (Vandamme, 2000). The defining biochemical tests to

differentiate between C. fetus fetus and C. fetus venerealis are the glycine tolerance and

hydrogen sulphide production tests, for which only the former is positive. However, the

validity of this has been called into question as a variant of C. fetus venerealis termed

biovar intermedius (Veron and Chatelain, 1973) can be positive in these assays (van der

Graaf-van Bloois et al., 2014). Furthermore, such features often do not parallel virulence

capabilities, for example, strains typed as C. fetus fetus based on phenotypic and

genotyping methods can display virulence features on the accessory genome (van der

Graaf-van Bloois et al., 2014). Therefore, the authors argued that C. fetus diagnostics,

particularly the methods used for subspecies differentiation, need considerable reviewing.

As described in section 1.6.2., there are several commercially available Campylobacter

detection tests. However, they are mostly optimised for the detection of the so-called

thermotolerant species. For example, the manufacturer of the DrySpot kit (Thermo

Scientific) advices that results can be variable with C. fetus. More recently, C. fetus

identification time has been reduced by screening samples using an ELISA and

subsequently confirming positives using culture (OIE, 2017). This assay utilises

monoclonal antibodies targeting C. fetus LPS (Brooks et al., 2002) and has a reported

100% sensitivity (Brooks et al., 2004; Devenish et al., 2005). However, samples require

incubation for at least four days prior to analysis and the procedure itself takes a further

two days (Devenish et al., 2005; OIE, 2017). An immunofluorescence test (Mellick et al.,

1965) using these monoclonal antibodies can be applied directly to genital fluid, however,

it requires several steps and an immunofluorescence microscope (OIE, 2017).

44

Several DNA-based methods have been reported for the identification of C. fetus. Efforts

have mainly focused on developing methods which can differentiate between C. fetus fetus

and C. fetus venerealis, as the latter is more concerning to the farming industry. For

example, the identification of a C. fetus venerealis-infected bull at an artificial

insemination facility can result in temporary suspension of activities and recent semen

samples to be discarded (van der Graaf-van Bloois et al., 2013). However, subspecies

differentiation of C. fetus is notoriously difficult, and often requires more complex

methods such as amplified fragment length polymorphism (Wagenaar et al., 2001) and

multilocus sequence typing (van Bergen et al., 2005). Multiple authors have reported PCR

assays for such purpose, although only those which target nahE of C. fetus venerealis can

do so reliably (OIE, 2017). However, as discussed above, the clinical relevance of such

subspecies differentiation has been challenged (van der Graaf-van Bloois et al., 2014).

Available DNA-based assays for C. fetus identification were reviewed by van der Graaf-

Bloois et al. (2013). Although some assays gave 100% sensitivity and specificity, all were

applied to pure cultures or extracted DNA. For example, these authors reported the PCR

method described by Hum et al. (1997) to give 100% sensitivity, however, Schmidt et al.

(2010) found the sensitivity to be 85.7% when analysing spiked preputial scrapings.

Furthermore, although there have been reports of using qPCR assays to detect C. fetus in

preputial samples (Chaban et al., 2012), these procedures still require sample

transportation to a laboratory and some sample preparation. Thus, such assays are most

appropriate for typing samples following culture identification in a laboratory, not direct

detection. Therefore, no methods available for detecting C. fetus are practical for on-site

detection. Development of such assays could provide farmers with a simple method to

monitor C. fetus infection, and thus rapidly identify an infected herd. In addition, this

could improve C. fetus detection in developing countries which is of particular difficulty

due to a lack of resources for technical molecular diagnostic methods (Mshelia et al.,

2010).

1.6.4. Emerging methods for Campylobacter detection

In the past decade, more novel methods such as biosensors and microarrays have been

developed for the detection of Campylobacter spp. Such methodologies allow for

detection of multiple pathogens simultaneously. For example, Suo et al. (2010) described a

DNA-based microarray capable of detecting C. jejuni, E. coli O157:H7, Salmonella

enterica and Listeria monocytogenes from fresh meat samples. This methodology

essentially allows the results of numerous qPCR assays to be visualised on a single glass

45

slide, thus bypassing the need for agarose gel electrophoresis. There are also similar

platforms which automate qPCR assays and are available commercially, such as

Verigene® Enteric Pathogens test and BioFire FilmArray™. These methods can provide

extensive analysis of a sample with high sensitivity (Huang et al., 2016), however, they

are relatively complex and hence are most suited to diagnostic laboratories and not field

analysis.

Biosensors generally utilise DNA or antibodies as detecting agents. The interaction of the

sensing biomolecule with an analyte is detected by a transducer which converts this signal

into an electrical output (Borisov and Wolfbeis, 2008). The method by which the

transducer detects an interaction defines the type of biosensor as optical, electrochemical

or mass-sensitive (Yang et al., 2013). Surface plasmon resonance devices are a particularly

popular form of optical biosensor which offer fast, highly sensitive and label-free analysis

without the need for numerous reagents (Skottrup et al., 2008; Yang et al., 2013).

Furthermore, commercial platforms are increasingly accessible. For example,

Gnanaprasaka et al. (2011) developed a DNA-based biosensor able to detect C. jejuni

using the Spreeta™ SPR platform. A comprehensive review of biosensors described for

the detection of Campylobacter spp. by Yang et al. (2013) found that C. jejuni was often

the sole target. Furthermore, although these biosensors have high sensitivity for C. jejuni,

it often reduces when analysing other species. For example, an immunosensor described

by Sapsford et al. (2004) could detect as little as 9.7 x 102

cells of C. jejuni but only 7.8 x

105

or more C. coli cells. In addition, there are no reports in the literature of biosensors

optimised for the detection of C. fetus. The development of antibodies capable of detecting

the common and emerging Campylobacter species with equal sensitivity could be a merit

to these methodologies. Furthermore, as the technology expands and devices become

increasingly portable these approaches will likely be readily applicable to on-farm testing.

1.6.5. N-linked glycans as targets for antibody-based detection.

As discussed, there is a need for fast and simple detection methods for pathogenic

Campylobacter species found in poultry and also the veterinary pathogen, C. fetus. In

particular, detection tests which could be performed on-site at farms are desirable.

Antibody-based detection methods such as lateral flow devices or immunosensors are

suitable technologies. For detection of Campylobacter species, a surface-exposed target

antigen which is a dominant and constant feature would be ideal.

46

The C. jejuni cell surface displays an abundance of carbohydrates, including the LOS,

capsular polysaccharide and O-glycosylated flagella (Karlyshev et al., 2005). These

features display interstrain diversity and phase-variable expression and are therefore

limited as targets for detection/identification of Campylobacter species. However, the C.

jejuni N-linked glycosylation pathway is not phase variable and produces only a single N-

glycan structure (Linton et al., 2005). At least 150 proteins may be N-glycosylated in C.

jejuni, including outer membrane proteins which could be surface exposed. The C. jejuni

N-glycan structure is conserved in closely related Campylobacter species such as C. coli

but not in more distantly related species such as C. fetus (Jervis et al., 2012; Nothaft et al.,

2012). Therefore, C. jejuni N-linked glycans might be considered promising targets for

antibody-based detection.

The N-glycan-reactive antiserum, hR6, is often used in Campylobacter N-glycosylation

research as it reacts with the two or three terminal GalNAc residues of the C. jejuni

heptasaccharide (Amber and Aebi, unpublished). As a result, hR6 is reactive with N-

glycoproteins of all Campylobacter species producing the C. jejuni type N-glycan and also

C. lari (Jervis et al., 2012; Nothaft et al., 2012). This antiserum does not react with other

Campylobacter species that produce N-glycans which vary at the non-reducing end in

comparison to C. jejuni, such as C. fetus and C. concisus. Furthermore, fluorescent

microscopy has demonstrated that hR6 can label cells of C. jejuni, suggesting that many

N-glycoproteins are surface exposed (Alemka et al., 2013). Therefore, C. jejuni N-glycan-

reactive antiserum could be used for specific, antibody-based detection of C. jejuni, C. coli

and other emerging species.

Detection assays for the human and veterinary pathogen, C. fetus, are limited. The cell

surface of C. fetus is unusual amongst Campylobacter species in that it contains a

paracrystalline layer of antigenically-variable surface layer proteins (SLPs) (Garcia et al.,

1995; Thompson, 2002). C. fetus do not produce capsular polysaccharide but do harbour

putative loci for synthesis of variable LOS and O-glycosylated flagellin (Gilbert et al.,

2016). Although C. fetus produce two N-linked glycan structures, they do not appear to be

subject to variable expression. Furthermore, antisera raised against each of the C. fetus

fOS structures are reactive with numerous C. fetus N-glycoproteins but not those of other

species such as C. jejuni, C. coli and C. concisus (Nothaft et al., 2012). Furthermore,

fluorescent microscopy revealed that these antisera can label cells of C. fetus fetus and C.

fetus venerealis (Nothaft et al., 2012). Therefore, similar to in C. jejuni, C. fetus N-linked

47

glycans are favourable targets for antibody-based detection as they appear a conserved

feature of C. fetus cells.

1.7 Aims

There are two overall aims for this research. The first is the investigation into

Campylobacter N-linked glycans as targets for antibody-based detection/identification of

Campylobacter species (Chapters 3 and 4). The second describes functional

characterisation of N-linked glycosylation in C. fetus (Chapters 5 and 6).

As discussed, it would useful to develop assays for the on-site detection of Campylobacter

human pathogens in poultry and the veterinary pathogen, C. fetus, in cattle. Antibody-

based detection methods are simple and rapid and therefore applicable for this purpose.

Furthermore, Campylobacter N-linked glycans could be attractive targets for antibody-

based detection/identification. To this end, this project aimed to characterise two antisera

raised against the N-glycans of C. jejuni and C. fetus (Chapters 3 and 4, respectively), and

evaluate the potential of these antisera to be used to detect Campylobacter species. An

antiserum, CjNgp, against a C. jejuni N-glycoprotein was already in production (Jervis &

Linton, unpublished) and this antiserum was characterised here. However, a C. fetus N-

glycan-reactive antiserum was unavailable and was newly developed as part of the work

described in this thesis.

Secondly, as the C. fetus N-linked glycosylation system is relatively uncharacterised, this

project aimed to analyse this further. Glycocompetent E. coli containing the C. jejuni pgl

locus is a powerful tool used to study C. jejuni N-glycosylation. Such a system expressing

the C. fetus N-glycan would be similarly useful for understanding N-glycosylation system

of this species. Therefore, this project aimed to develop glycocompetent E. coli producing

the C. fetus N-glycan structures (Chapter 5). In addition, using a bioinformatics tool, it

aimed to predict the C. fetus N-glycoproteome and assess conservation of predicted N-

glycoproteins more broadly amongst Campylobacter species (Chapter 6).

48

Chapter 2. Materials and Methods.

49

2.1 Bacterial strains and plasmids

Bacterial strains used are listed in Appendix Table 1. All Campylobacter strains were

grown on blood agar (BA) plates containing Columbia blood agar base (Oxoid Ltd.)

supplemented with 5% defibrinated horse blood (TCS Biosciences) and containing

selective antibiotics as required. Most Campylobacter strains were grown in a VA500

workstation (Don Whitley Ltd., UK) in a microaerobic atmosphere (85% N2, 10% CO2,

and 5% O2). However, C. concisus, C. lanienae, C. sputorum and C. hyointestinalis

require a microaerobic hydrogen-containing environment generated in a 2.5 litre gas

chamber (Oxoid Ltd.) using a CampyGen™ atmosphere generation system sachet and 10

ml of a 1% sodium borohydride solution. Most Campylobacter species were incubated at

42°C, with C. fetus and C. concisus incubated at 37 °C. E. coli strains were grown in

lysogeny broth (LB) or LB agar (Oxoid Ltd.) containing antibiotics as required and

incubated at 37 °C. Antibiotics were purchased from Sigma and used at the following

concentrations: 50 µg/ml kanamycin, 100 µg/ml ampicillin and 34 µg/ml chloramphenicol.

For Campylobacter, chloramphenicol was used at 17 µg/ml.

2.2 Isolation of Campylobacter from retail chicken

2.2.1 Culture and isolation of Campylobacter chicken meat isolates

Chicken meat was purchased from a variety of supermarkets during 2014 and 2015.

Packaging and tools used to handle meat were sterilised using 100% ethanol, and samples

were obtained aseptically. Sample-inoculated swabs or ~ 1 inch2

skin pieces were

incubated in Bolton broth (Oxoid Ltd.) with Bolton selective supplement (Fluka

Analytical) for 48 hours at 42°C in a VA500 workstation-generated microaerobic

atmosphere. Samples were plated onto Blood Free Campylobacter selectivity agar base

(Oxoid Ltd.) with CCDA selective supplement (Fluka Analytical) and incubated for a

further 48 hours, before single Campylobacter-like colonies were picked and streaked onto

Columbia blood agar base (Oxoid Ltd.) supplemented with 5% defibrinated horse blood

(TCS Biosciences).

2.2.2 Verification and Molecular typing of isolates

Isolates were confirmed as Campylobacter using PCR with genus-specific 16S rDNA

primers, C412F and C1288 (892 and 893 in Table 2.1), as described by Linton et al.

(1996). Species-specific PCR was performed using primer pairs HIP400F/HIP1134R for

C. jejuni and CC18F and CC519R for C. coli (Linton et al., 1997). To ensure strains were

distinct, isolates were typed by sequencing the hypervariable region of flaB using primers

50

1246 and 1247 (Table 2.1). Amplicons were sequenced on an Applied Biosystems 3730

DNA analyzer (Life Technologies). Isolates (DW1-10) are described in Appendix Table 1.

2.3 Whole cell lysate preparation for SDS-PAGE

Whole cell lysates were prepared using BugBuster™ Protein Extraction Reagent

(Novagen) or incubating cells suspended in sample buffer at 95 °C for 10 minutes. Cells

were harvested from agar plates and suspended in 0.6 ml PBS. Absorbance at 600 nm

(OD600) was measured using an Ultrospec 2100 pro UV/visible spectrophotometer

(Amersham Biosciences). The remaining 0.5 ml cell suspension was centrifuged at 16000

g for 3 minutes and the supernatant removed. Whole cell lysates were produced by

resuspending this pellet in 50-100 µl BugBuster™ Protein Extraction Reagent (Novagen)

with 0.5 µl DNaseI (Qiagen) and incubating for 20 minutes at room temperature. Cell

lysate solutions were centrifuged at 16000 g for 3 minutes and the supernatant transferred

to a clean Eppendorf. NuPAGE™ LDS sample buffer (4X) (Thermofisher Scientific) and

water were added to the supernatants to give a final volume of 10 µl per optical density at

OD600 of 0.1. The samples were then incubated at 95°C for 5 minutes. Alternatively, for

samples created without the use of BugBuster™ Protein Extraction Reagent, a cell

suspension was made to a final volume of 10 µl per optical density at OD600 of 0.1 using

NuPAGE™ LDS sample buffer (4X), incubated at 95°C for 10 minutes and centrifuged at

16000 g at 16162 xg for 20 minutes. The supernatant was transferred to a fresh Eppendorf

and stored at 4 °C.

2.4 SDS-PAGE and western blotting

Samples were loaded onto NuPage Bis-tris precast 12% gels (Life Technologies) and

electrophoresed at 200 V for 50 minutes in NuPage MOPS buffer using an XCell

SureLock™ Mini-Cell Electrophoresis System (Life Technologies) and BIORAD

powerpac300. Alternatively, samples were run on hand-cast gels containing 12% ProtoGel

Acrylamide (National Diagnostics) in SDS-PAGE running buffer (containing 0.3% (w/v)

Tris, 1.4% (w/v) glycine and 0.1% (w/v) SDS). Hand-cast gels were electrophoresed at

160 V for 80 minutes using the apparatus described above. Precision plus All Blue

Standards (Bio-Rad) was used as a molecular weight marker. Gels were stained with

InstantBlue (Expedeon) coomassie stain for at least 20 minutes at room temperature and

washed in water prior to imaging.

For western blotting, SDS-PAGE gels were transferred to a nitrocellulose membrane in an

XCell II™ Blot Module (Life Technologies) for 1 hour at 30V in Transfer buffer

51

containing 0.24 % (w/v) Tris and 1.1 % (w/v) glycine. Membranes were blocked with 5 %

(w/v) powdered milk (Marvel) or 5 % BSA in PBS with 0.1 % (w/v) Tween (PBS-T) at

4°C overnight or 30 minutes at room temperature on a rocker. Membranes were probed

with primary antibodies in PBS-T with either 3 % (w/v) milk or BSA at room temperature

for one hour. Antibody dilutions were as follows: 1/2000 (CjNgp), 1/1000 (CfNgp) or

1/1000 (Anti-HexaHis). Membranes were washed three times in PBS-T. IRDye® 800CW

Goat anti-Rabbit IgG secondary antibodies (LI-COR Biosciences) or IRDye® 680RD

Donkey anti-Mouse IgG secondary antibodies (LI-COR Biosciences) were added at a

dilution of 1/10000 in 3 % milk or 3 % BSA-PBS-T and incubated at room temperature

for one hour. Membranes were washed three times with PBS-T. Membranes were then

stored in PBS until visualised using an ODYSSEY Infrared Imaging System (LI-COR

Biosciences) and associated Image Studio™ software.

2.5 Dot blot assay

The dot blot assay protocol was modified from Qian et al. (2008). Cells were grown for 24

hours on blood agar plates, harvested, washed in 600 µl PBS and resuspended at various

OD600 values. Cells numbers were estimated based on an OD600 of 1 being equal to around

1 x 109

C. jejuni cells per ml (Qian et al., 2008). Cell suspensions (1 µl) were pipetted onto

a nitrocellulose membrane, left to air-dry for 15 minutes and the membrane blocked in

PBS with 0.1% Tween and 5% BSA for 30 minutes. Following a brief wash in PBS-T,

primary and secondary antibodies were incubated in 3% BSA-PBS-T for 1 hour and the

membrane washed with PBS-T three times following incubations. Dot blots were stored in

PBS and visualised using an ODYSSEY (LI-COR Biosciences) Infrared scanner and

associated software.

2.6 Flow cytometry

Cells were harvested from overnight blood agar plates, suspended into PBS and

centrifuged at 12470 g for 3 minutes. Cells were washed again in PBS, resuspended at an

OD600 of 0.5 in PBS with 0.05 % (w/v) Tween and 1 % (w/v) BSA and incubated at room

temperature for 30 minutes on an end-over-end rocker. Pimary antibody was added at a

dilution of 1/250 and incubated for an hour at room temperature. Cells were washed once

in PBS-T-BSA, R-phycoerythrin Goat Anti-Rabbit IgG secondary antibodies (Life

Technologies) added at 1/250 in PBS-T-BSA and incubated for 45 minutes at room

temperature in the dark. Cells were washed with PBS-T-BSA and resuspended in PBS.

Data was aquired on a Canto Flow Cytometer using BDFACSDIVA software. Ungated

52

data analysis was performed using Flowjo version 9 (Tree Star) and event counts were set

to 10,000.

2.7. Polymerase Chain Reaction

Genomic DNA was extracted using an Archive Pure DNA cell/tissue kit according to

manufacturer’s instructions (5PRIME) and PCR performed in a Techne thermal cycler

(Bibby Scientific Ltd.). Primers (Table 2.1) were purchased from Eurofins MWG Operon

and used at a concentration of 1 pmol µl-1

. For colony PCR, MyTaq Red Mix (Bioline)

was used as the PCR master mix. Target DNA was included at a concentration of 1 ng µl-

1. Reactions were 30 cycles of: 94°C for 30 seconds, annealing temperature at 5 degrees

below the melting temperature for 30 seconds and elongation at 72°C for 30 seconds kb-1

.

Upon completion, samples were held at 4°C. For PCR for use in molecular cloning,

CloneAmp HiFi PCR premix (Takara Bio USA) was used and reaction parameters were as

suggested by the manufacturer. PCR products were purified using QIAquick PCR

purification kit (Qiagen) and amplicons sequenced on an Applied Biosystems 3730 DNA

analyzer (Life Technologies) using between 300 and 500 ng of DNA and 4 pmol of primer

in 10 µl of milliQ water.

Table 2.1. Primer sequences.

Name (number) Sequence Reference

(if applicable)

T7 universal F (80)

()promoter

GGATGACACTTTTCGGAGC N/A

T7 universal R (82) CATTGTAGCACGTGTGTC N/A

Cm200Rout (546) CTTGGAAAGGAACACCGCCG

N/A

Cj0114NpET22-R1 (790) TCTTTGCGTTCACAAGCAGC Jervis & Linton,

unpublished C412F (892) ACCTGATTGCGCAAGGATG Linton et al., 1996

C1288 (893) GCGACTGGATCCATGTAATTAGTTT

CTTAATCG

Linton et al., 1996

NGRP-F (1184) GTAACTAGATCTGCGAAATTAAGA

CAAAGTG

Jervis & Linton,

unpublished

FlaB_Cjc-L (1426) GCTAAGAGATCTTCAATTTGCTGCT

TCTTTTAT

Korczak et al., 2009

CjflaBR (1247) TACATTGGATCCGCTATTTTCATGC

AAAATTTCC

This work

rrlfetusF (1268) TAAGTGCAAGTGGTAGGAG This work

53

rrlfetusR (1269) GCTCTTATCTCATCCTAGC This work

porAcatBamHIF (1296) CGTTCAGGATCCTCGAGCTTAAAATTA

CACGCC

This work

porAcatBamHIR (1297) ACTGACGGATCCGCTAGCCATGGTTAT

TTATTCAGC

This work

pCfporAcatseqF (1302)

GGTAGCGAAATTCCTTGTC

This work

NGRPwithssNdeIF (1320) GGCGGCATATGAAAAAAATATTCACA

GTAGCTC

This work

NGRPwithssNdeIF (1322) GCCGGCATATGATCATCTTGGGTATTT

TGAG

This work

HIP400F (1366) TCAATTTGCTGCTTCTTTTAT Linton et al., 1997

HIP1134R (1367) CTTTTTCATTGTTCGCACTC Linton et al., 1997

CC18F (1368) TCACTGTGCTAAATCTCCTC Linton et al., 1997

CC519R (1369) TTGCCTAGATCTGTGAAGCAAAAAT

CATGGATG

Linton et al., 1997

CfpglH1BamHIF (1378) GTAAGCAGATCTTTAGGCATTTTTA

ACCTCGAC

This work

CfpglH2BclIR (1379) GTAACTTGATCAGCGAAATTAAGA

CAAAGTG

This work

CfpglH2BglIIF (1385) AATACGGGATCCTCACTGTGCTAAA

TCTCCTC

This work

CfpglH1BglIIR (1386) CGCCCTATAGTGAGTCG This work

Cf1389BamHIF (1387) GTTATATCTCCCATGGCCATCGCC This work

CfpglH1checkR (1388) CCATGATTACGCCAAGCGCGC This work

CfpglH2checkR (1389) TACCCGGGTTAGTGATGGTGATGGT

GATGGTGATGATCATCTTGGGTATT

TTGAG

This work

Cf1389checkR (1390) GAAGAGGGTTTGGGTGGTG This work

CjpglHBamHIF (1397) AGCTAGCTTCGCATAATAACTTG This work

CjpglHBamHIR (1398) GGTATGATTTCTACAAAGCGAG This work

CfpglH2BclIF (1399) ATAAAAGACTATCGTCGCGTG This work

Cf1389BamHIR (1406) GCTAGTTATTGCTCAGCGG This work

pGEMT7checkR (1407) TGATGTCGGCGATATAGG This work

pETNGRPupstream (1411) TAATACGACTCACTATAGG This work

54

CjpglHcheckR (1424) GATAGCTGATCACTTTTTCATTGTT

CGCACTC

This work

pETNpglcheck R (1425) TTAGGCATTTTTAACCTCGGC This work

M13 universal F (1432) GCGGCTGTGATAGTTTAGTCG N/A

Cf1389BclIF (1439) AGAACGGTTAATCCACTTGC This work

pRYGG1/2_seq_F (1445) GCTCGCTGATCACTATTTTCATGCA

AAATTTCC This work

CfH2midcheck_R (1450) CTCAGACGTAAAATTAAAGCCG This work

NGRPSmaI_H8_R (1484) TTGTTTCTCGAGATCATCTTGGGTAT

TTTGAG This work

CfpglH1_F_out (1485) ATGGCCATGGGAGATATAACAAGC

AATTC This work

Cf1389_mid_F (1486) TTCAGGCTTACTTGCGTCT This work

CfpglH1midR (1558) CGTCACGGATCCATGAATATAGGC

AAGGGTATAGC This work

Cf0445BamHI_F (1566) CGGCGCGGATCCATGAAAAAAATA

CTTAGCCTTATGTTC This work

Cf0781BamHI_F (1567)

TACCCGGGTTAGTGATGGTGATGGT

GATGGTGATGTTCACTTACGTCAATCC

CAA

This work

Cf0781SmaI_R (1572)

TACCCGGGTTAGTGATGGTGATGGT

GATGGTGATGCTTAAATAAAGAAAT

TTCTTC

This work

Cf0445SmaI_R (1573) GTAAAACGACGGCCAGT This work

2.8. DNA agarose gel electrophoresis

DNA electrophoresis was performed in a Mini-Sub Cell GT cell (Bio-Rad) in 1% agarose

gels containing 0.1% ethidium bromide in TAE (40 mM Tris-acetate, 1 mM EDTA, pH 8)

buffer. Gels were exposed to 110 V for one hour using a Consort EV265 electrophoresis

power supply. Samples were prepared using 6X loading buffer (30% (w/v) glycerol,

0.25% (w/v) bromophenol blue, 0.25% (w/v) xylene cyanol FF). Hyperladder 1 kb

(Bioline) was used as the DNA molecular weight marker. Gels were visualised using a

Uvipro Silver geldoc (UVITEC).

55

2.9. DNA restriction digestion and ligation

Plasmids were extracted using QIAprep Spin miniprep kit (Qiagen) and PCR products

were purified using QIAquick PCR purification kit (Qiagen). DNA concentrations were

measured using a NanoDrop 1000 spectrophotometer (Thermo Scientific). Restriction

enzymes used were either purchased from Roche or NEB. Most restriction digests were

performed for one hour at 37°C, whereas SmaI enzyme was incubated at 25 °C. T4 DNA

Ligase (NEB) and 10X Ligation buffer were purchased from Roche. Ligation reactions

were performed at 4°C overnight.

2.10. E. coli competent cell preparation and heat shock

transformation

All reagents, pipette tips and falcon tubes were pre-chilled and centrifugation was

performed at 4 °C. One ml of an overnight culture was used to inoculate 100 ml pre-

warmed LB broth. Cells were grown at 37°C with shaking until an OD600 of 0.6-0.7,

placed on ice for 15 minutes, transferred to 50 ml falcon tubes, centrifuged at 3500 rpm

for 10 minutes in a Heraeus™ Megafuge 40R centrifuge (Thermo Scientific) and the

supernatant discarded. Cells were washed with 25 ml cold 0.1M CaCl2, centrifuged as

above and the supernatant removed. Cells were resuspended in 2 ml cold 0.1M CaCl2 and

incubated on ice for at least two hours. Following this, glycerol was added to give a final

concentration of 15% and the suspension aliquoted into 100 µl fractions. Cells were snap-

frozen in liquid nitrogen and stored at -80°C.

For purchased competent E. coli XL-1 Blue cells (Agilent), transformations were carried

out according to manufacturer’s instructions. For laboratory-prepared cells, 50 µl of cells

were aliquoted into pre-chilled tubes, 1.7 µl of β-mercaptoethanol added and incubated for

ten minutes on ice. One µl of DNA was added, cells incubated for 30 minutes on ice, 42°C

for 45 seconds and then incubated on ice for two minutes. 900 µl of preheated SOC

medium (20 g tryptone, 5 g yeast extract, 0.5 g NaCl per 1L) was added and tubes

incubated at 37 °C for one hour with shaking. 200 µl of cells were plated onto four LB

plates containing appropriate antibiotics and incubated overnight at 37 °C.

2.11. C. fetus conjugation

C. fetus conjugation was carried out according to the protocol of Kienesberger et al.

(2007), with slight modification based on personal communication (Kienesberger, 2015).

All C. fetus incubation steps and the mating step were carried out under microaerobic

56

conditions. C. fetus cells were incubated on blood agar plates supplemented with 5%

defibrinated horse blood (with selection if necessary) for 24 hours at 37 °C. The donor

strain, E. coli S17-1 λpir pG1-N, was in cultured in 5 ml LB with chloramphenicol

overnight at 37 °C and 0.5 ml of this overnight culture was subsequently used to inoculate

3 ml of LB with chloramphenicol and incubated for 3 hours at 37 °C. The OD600 of this

culture was measured and ~107

cells (based on OD600 of 0.1 equating to 5 x 107

cells/ml)

were resuspended in 20 µl of sterile PBS. C. fetus cells were harvested from the 24 hour

incubation plate and suspended in sterile PBS. Based on an OD600 of 0.1 equating to 5 x

108

cells, 109 C. fetus cells were centrifuged and the 20 µl of prepared E. coli cells were

used to resuspend this pellet. This mating mixture was spotted onto a nitrocellulose filter

(25 mm, 0.22 µm pore size)(Millipore), placed in the centre of a BA plate and incubated

microaerobically for 3 hours at 37 °C to allow plasmid mobilisation. Cells were harvested

from the filter in a sterile 20 ml universal tube using 1 ml sterile PBS (prewarmed to 37

°C), vortexed and plated onto four BA plates containing Nalidixic acid (75 µg/ml)

(ACROS Organics) and chloramphenicol (30 µg/ml) to select for C. fetus and the

transferred plasmid, respectively. When using C. fetus pglB::kn, these plates also included

kanamycin. Plates were incubated for 8-14 days at 37 °C.

2.12. Protein expression in E. coli BL21 (DE3)

Protein expression from E. coli was achieved using plasmids derived from pET22 b (+)

(Novagen). E. coli BL21 containing the appropriate expression vector was incubated in 5

ml LB broth (with selective antibiotics) overnight (16 hours) at 37 °C. This culture was

then used to inoculate 250 ml LB broth (with relevant antibiotics) and incubated at 37 °C

until the culture reached an OD600 of 0.6. At this point, 1 mM IPTG (Thermo Fisher

Scientific) was added and the culture incubated for either 2 hours at 37 °C or overnight at

16 °C to allow for protein expression. To confirm the desired protein had been produced, a

whole cell lysate sample was prepared and analysed using SDS-PAGE and western

blotting as described above.

2.13. Nickel affinity chromatography

Following protein induction in E. coli or 24 hour incubation of C. fetus, cells were

centrifuged at 8000 rpm in a SORVALL® RC5B Plus centrifuge for 20 minutes at 4 °C.

Cell pellets were weighed and cells resuspended to a final volume of 5 ml/g in Nickel

binding buffer: 20 mM sodium phosphate (pH 7.4), 0.5 M NaCl, 20 mM imidazole

containing Complete™ mini EDTA-free protease inhibitor cocktail and DNase (1/1000).

Cells were sonicated on ice at 30% setting for 5 second pulses with 10 second intervals.

57

Sonication was carried out for at least 4 minutes. To remove cell debris, the sonicated

solution was centrifuged at 16,000 rpm in a SORVALL® RC5B Plus centrifuge for 30

minutes at 4 °C. The supernatant was removed and filtered using a 0.35 µm pore size

syringe filter (Millex).

All buffers were filtered and degassed prior to use. Proteins were purified using a 5 ml

HisTrap FF column (GE Healthcare). The column was first washed with at least 25 ml of

water and equilibrated with 5 ml of 20 mM imidazole binding buffer (solution as above).

Sample was applied and the flow-through collected. The protein-bound column was

washed with 75 ml of 50 mM imidazole wash buffer and sample eluted with 10 ml of 500

mM imidazole elution buffer. Eluate was collected in 1 ml fractions (E1-10).

2.14. Ion exchange chromatography

Nickel affinity chromatography eluate was dialysed against 20 mM Tris pH 7.6 0.01 M

NaCl. Briefly, sample was applied to a HiTrapQ HP 1 ml column (GE healthcare) on an

ÄKTA HPLC system and a gradient from 20 mM Tris pH 7.6 0.01 M NaCl to 100% 20

mM Tris pH 7.6 1 M NaCl over 20 ml applied.

2.15. Post-purification protein treatment

The appropriate nickel affinity chromatography eluate fractions (as determined by SDS-

PAGE and coomassie-staining) were pooled and dialysed against 30 mM Tris pH 8.0, 30

mM NaCl overnight at 4 °C. Protein solutions were concentrated using Vivaspin

ultrafiltration columns (Sartorius) and centrifuged at 7000 rpm until sufficiently

concentrated.

For preparation of the glycosylated NGRP purified from C. fetus for use as an

immunogen, the dialysed purification eluate was concentrated to ~ 2 ml, a sample ran

alongside BSA standards and coomassie-stained (Fig. 4.4C). The total amount of NGRP

glycoforms present was estimated as around 1 mg using densitometry within Image Studio

software (LiCor biosciences). Aliquots of this preparation were lyophilised and sent to

Antibody Production Services Ltd (Life Science Group Ltd.) for production of polyclonal

rabbit antisera using their standard 77 day immunisation protocol.

2.16. Mass Spectrometry

2.16.1. Sample preparation and MALDI-MS analysis

Samples were separated using SDS-PAGE and proteins visualised with coomassie stain.

Following destaining in water overnight, protein bands of interest were excised using a

58

sterile scalpel, sliced into ~ 1 mm3

pieces and transferred to a sterile eppendorf. Gel pieces

were incubated in 20 µl of 50 mM ammonium bicarbonate (AMBIC) pH 8.4 at room

temperature for 5 minutes and the supernatant removed. This step was repeated until the

coomassie stain was no longer visible. Destained gel pieces were dried for 15 minutes in a

vacuum centrifuge. Dried gel pieces were rehydrated in 20 µl of a working solution of

trypsin (25 ng/µl in 50 mM AMBIC pH 8.4) (Promega) and incubated at room

temperature for 15 minutes. If the gel pieces swelled above the liquid level, 20-30 µl of 50

mM AMBIC pH 8.4 was added to submerge them. The in-gel tryptic digests were then

incubated overnight at 37 °C. Following incubation, the supernatant was transferred to a

sterile Eppendorf. To stop the reaction, the gel pieces were incubated in 50 µl of

Trifluoroacetic acid (TFA) optima™ LC-MS grade (Fisher Scientific) for 10 minutes at 37

°C. To elute peptides, the pieces were incubated in acetonitrile (ACN) for 15 minutes at 37

°C. The tryptic peptide-containing supernatant was removed and combined with the

hydrophilic peptides initially extracted. This TFA and ACN step was repeated once. The

tryptic peptide solution was then dried in a vacuum centrifuge for upto two hours.

Finally, peptides were concentrated and purified using C18 ZipTip® pipette tips (Merck).

Samples were acidified using 10 µl of 0.1% trifluoroacetic acid (TFA). ZipTips were

equilibrated using two cycles of 50% ACN followed by five cycles of 0.1% TFA. Samples

were bound to the tip using five cycles. Samples were washed with two cycles of 0.1%

TFA and eluted in 4-10 µl in 50% ACN, 0.1 % FA. Samples were either analysed using α-

cyano-4-hydroxycinnamic acid (CHCA) (Sigma) or 2,5-Dihydroxybenzoic acid (DHB)

(Sigma) as the matrix. CHCA was prepared as a 1 mg/ml solution in 80% ACN, mixed 1:1

with sample solution. Two microlitres of this solution was applied to a ground steel

MALDI target plate and left to air dry. DHB was prepared at 20 mg/ml in 0.1 % TFA and

mixed 1:1 with the sample solution. Half a microlitre of this solution was pipetted onto a

ground steel MALDI target plate and left to air dry. MALDI-TOF MS was performed on a

Bruker Ultraflex II mass spectrometer in positive-ion relfection mode and at 40-50% laser

power. Data were analysed using FlexAnalysis version 3.0 software (Bruker Daltonics).

2.16.2. LC-MS/MS

Digested samples were produced as previous and analysed by LC-MS/MS using an

UltiMate® 3000 Rapid Separation LC (RSLC, Dionex Corporation) coupled to a Q

Exactive™ HF Hybrid Quadrupole-Orbitrap™ Mass Spectrometer (Thermo Fisher

Scientific) with Higher-energy Collisional Dissociation (HCD) fragmentation.

59

Samples were separated using a multistep gradient from 95% A (0.1% formic acid in

water) and 5% B (0.1% formic acid in ACN) to 7% B at 1 min, 18% B at 35 min, 27% B

in 43 min and 60% B at 44 min at 300 nL min-1

, using a 75 mm x 250 μm i.d. 1.7 M

CSH C18, analytical column (Waters).

Raw data files were converted to .mgf files using Mascot Daemon software (Matrix

Science) and data analysed using mMass v 5.5 (Niedermeyer and Strohalm, 2012). Files

were scanned for precursor ions at predicted m/z ratios. Fragmentation spectra were

manually analysed for the presence of glycan-derived species such as m/z 204.09, which

corresponds to a HexNAc residue and is characteristic of glycopeptide fragmentation

spectra obtained using HCD (Zhao et al., 2011). In addition, spectra were scanned for

disaccharide oxonium ions such as m/z 366 (HexHexNAc), as they can help elucidate

glycan structure (Toghi Eshghi et al., 2016).

2.17. Bioinformatics

The molecular weight of proteins was predicted using the compute pI/Mw tool of the

ExPASy bioinformatics resource portal (Swiss Institute of Bioinformatics). The following

programs were downloaded locally on a computer running a Linux operating system to

facilitate use of the N-glycoprotein prediction Perl script described by Frost (2015): NCBI

BLAST version (v) 2.3.0+ (Camacho et al., 2009), Pftools v 2.3 (ExPASy), PSORTb v

3.0.4 (Yu et al., 2006), LipoP v 1.0 (Juncker et al., 2003) and TMHMM v 2.0b (Krogh et

al., 2001). Proteomes, excluding protein sequences derived from pseudogenes, were

extracted in FASTA format from genomes using Artemis Genome Browser (Rutherford et

al., 2000). These proteomes were then fed to the perlscript. The tab delimited text files

produced by the perlscript were copied to Microsoft Excel for manual curation. Manual

analysis utilised CELLO v 2.5 (Yu et al., 2006) and SignalP 4.1 Server (Petersen et al.,

2011). Sequence logos were created using WebLogo 3 (Crooks et al., 2004). Predicted N-

glycoprotein comparisons utilised the online standard protein BLAST® (BLASTP)

program and query sequences were searched against the revelant organism within the non-

redundant protein sequence database. The total percent identity between two proteins was

determined using the BLAST® global alignment tool.

60

Chapter 3. Characterisation of an

antiserum raised against a

Campylobacter jejuni N-linked

glycoprotein

61

3.1. Introduction

3.1.1. Overview of C. jejuni NGRP antiserum production process

At the start of this project, an antiserum raised against a C. jejuni N-linked glycoprotein

antiserum was in production (Jervis and Linton, unpublished). This antiserum was raised

against a truncated form of a C. jejuni glycoprotein, Cj0114 that has four N-linked

glycosylation sites, two in an N-terminal coiled-coil domain and two in a flexible linker

region. Its C-terminal tetratricopeptide repeat (TPR) domain is not N-glycosylated and

hence was removed to give the truncated form, termed NGRP (N-Glycosylation Reporter

Protein). Due to being small and heavily N-glycosylated this recombinant protein made an

ideal candidate to use for the production of a C. jejuni N-linked glycoprotein antiserum.

Production of NGRP glycosylated with the C.jejuni heptasaccharide was achieved using

glycocompetent E. coli (Wacker et al., 2002). NGRP was cloned into pET22b(+) to give

pNGRP which allowed expression of His6-tagged NGRP directed to the periplasm. This

plasmid was transformed into E. coli BL21 (DE3) cells containing the plasmid ppgl

(encoding the functional pgl locus of C. jejuni) (Wacker et al., 2002). Glycosylated His6-

tagged NGRP was purified by Nickel-affinity chromatography and used as an immunogen

for production of rabbit polyclonal antiserum (Antibody Production Services Ltd.). The

final antiserum produced was named CjNgp (C. jejuni N-glycoprotein).

3.1.2. Project aims

There is a considerable need for detection methods that could be used on-site to detect

Campylobacter in poultry. Antibody-based detection/identification methods are

particularly suitable, as they are generally rapid and simple to use. As discussed in chapter

1, N-linked glycans are promising targets for antibody-based detection or identification of

Campylobacters. More specifically, an identical N-glycan structure is produced by C.

jejuni, C. coli and C. upsaliensis, pathogenic species which are found on chicken meat.

Therefore, antibodies targeting this N-glycan structure have the potential to be developed

into an assay able to detect several important Campylobacter pathogens. Indeed,

previously described antisera reactive against the C. jejuni heptasaccharide have been

shown to react similarly well with cell lysates of C. coli and C upsaliensis (Jervis et al.,

2012; Nothaft et al., 2012). In addition, these antisera are able to react with cell lysates of

62

the hexasaccharide N-glycan-producing C. lari, another Campylobacter pathogen that

contaminates chicken meat. In conclusion, it appears that antibodies against the C. jejuni

heptasaccharide N-glycan could be used to create a test able to detect at least four species

of Campylobacter which are a human infection risk in association with chicken meat. This

project aimed to to assess the feasibility of CjNgp to be used in an antibody-based method

for identification/detection of important Campylobacter pathogens found in poultry. To do

this, the following aspects of antiserum CjNgp were characterised:

C. jejuni N-linked glycan reactivity and specific N-glycan structural specificity.

C. jejuni strain coverage and reactivity against recent chicken meat isolates.

Campylobacter species specificity.

Binding untreated cells of Campylobacter species.

63

3.2. Characterisation of CjNgp reactivity against Campylobacter

whole cell lysate extracts.

3.2.1. The reactivity of CjNgp antiserum against C. jejuni N-linked glycan.

In order to assess if CjNgp was reactive towards the C. jejuni N-linked heptasaccharide, it

was used to probe western blots containing C. jejuni wildtype and a range of pgl mutant

whole cell lysates (Fig. 3.1B). Whole cell lysates were separated by SDS-PAGE and then

either stained with coomassie (Fig. 3.1A), or western blotted for probing with antiserum

(Fig. 3.1B). With a C. jejuni lysate, CjNgp reacted with many proteins (Fig. 3.1B). In

particular, a strong reactive band at around 42 kDa was visible. When probed against a C.

jejuni pglB mutant only a single reactive band at around 35 kDa was visible. The

oligosaccharyltransferase PglB is responsible for transferring the N-linked glycan structure

onto target proteins, thus a mutant lacks N-linked glycoproteins (Wacker et al., 2002).

Therefore, all but a single CjNgp-reactive band in a C. jejuni lysate were pglB-dependent.

Thus, CjNgp is reactive towards the N-linked glycan produced by C. jejuni.

Deletion of C. jejuni pgl glycosyltransferase genes results in transfer of truncated glycans

onto target proteins (Linton et al., 2005) and N-linked glycan structures produced by each

strain are presented (Fig. 3.1C). As with a wildtype lysate, CjNgp exhibited reactivity with

numerous bands within a pglI mutant lysate, but with reduced intensity (Fig. 3.1B). In

contrast to its particularly strong reactivity with an ~ 42 kDa band in a wildtype lysate,

with a pglI mutant it reacted with two distinct bands at around 42 kDa and 41 kDa in size.

Reactivity with lysates from pglHJA mutants was reduced to a similar level to that

observed with the pglB mutant. Importantly, it appeared that the single reactive band

remaining in the pglB and pglHJA mutants decreased in size as glycan structure was

truncated, suggesting it corresponds to a glycoprotein. This is likely Cj0114, the native N-

glycoprotein from which NGRP was constructed, as its apparent unglycosylated size (35

kDa) in C. jejuni pglB::kn correlates with what is predicted (35.4 kDa). Therefore, CjNgp

is reactive with this native protein, but its N-glycan reactivity was likely abolished upon

reduction to tri-, di- and monosaccharides. These data confirm that CjNgp has significant

reactivity against the C. jejuni N-linked glycan, the majority of which is directed towards

the terminal four sugars (GalNAc3Glc) of the heptasaccharide structure. However, the

antiserum also appears to react with the N-glycoprotein Cj0114 regardless of N-glycan

modification.

64

65

3.2.2. CjNgp reactivity with a variety of C. jejuni strains

In order to assess if CjNgp has consistently high reactivity with N-glycoproteins of a range

of C. jejuni strains, it was used to probe western blots containing whole cell lysates of an

additional 11 strains. This strain set included Penner reference strains and clinical strains

from patients with Miller-Fischer syndrome (F1-F3) and Guillian Barré Syndrome (G2)

(Appendix Table 1). Whole cell lysates were separated by SDS-PAGE and then either

stained with coomassie (Fig. 3.2A), or western blotted for probing with antiserum (Fig.

3.2B). As previously, CjNgp reacted with a multitude of bands in a C. jejuni 11168H

whole cell lysate (Fig. 3.2B). A C. jejuni 11168H pglB mutant lysate served as a negative

control. In contrast to the result seen with the previous pglB::kn sample, CjNgp did not

react with a 35 kDa band in this instance. This could be due to a difference in extraction

method (BugBuster® Protein Extraction Reagent) that fails to solubilise Cj0114 and/or

possibly due to reduced stability of this protein when not N-glycosylated.

CjNgp strongly reacted with whole cell lysates of each of the strains tested and although

glycoprotein profiles were very similar, subtle differences were observed. Again, a strong

band at around 42 kDa in size is seen with all the strains tested, though the mobility and

intensity of the band does vary between strains. Therefore, the CjNgp antiserum reacted

with numerous glycoproteins in a variety of C. jejuni strains. Hence, CjNgp antiserum has

the potential to detect a broad range of C. jejuni strains if integrated into an antibody-

based test.

66

67

3.2.3. CjNgp reactivity with recent Campylobacter chicken meat isolates

To verify that the CjNgp antiserum reacts with currently circulating strains of

Campylobacter, strains were isolated from retail chicken meat. Isolates were identified

using Campylobacter-specific PCR (Linton et al., 1996) followed by C. jejuni and C. coli

species-specific PCR assays (Linton et al., 1997). Strains were typed by sequencing of the

flaB variable region (See section 2.2). Isolation experiments carried out in 2014 and 2015

from two supermarkets resulted in 10 distinct strains, 6 C. jejuni and 4 C. coli (Appendix

Table 1).

Whole cell lysates were separated by SDS-PAGE and then either stained with coomassie

(Fig. 3.3A), or western blotted for probing with antiserum CjNgp (Fig. 3.3B). Many

immunoreactive bands were evident with each isolate. The pattern of reactivity with each

C. jejuni isolate was very similar to that observed with C. jejuni 11168H. As for the C.

coli isolates, there were a few differences in the reactive band profile however the overall

level of reactivity was just as marked. Overall, CjNgp reacted very strongly towards whole

cell lysates of Campylobacter recently found to be infecting retail chicken meat, whether

they were C. jejuni or C. coli. These data suggest CjNgp is appropriate for detecting

Campylobacter strains recently contaminating chicken flocks.

68

69

3.2.4. Campylobacter species specificity of CjNgp.

To investigate the species specificity of the CjNgp antiserum, it was used to probe whole

cell lysates of Campylobacter species which encompass five N-linked glycan structural

groups described (Jervis et al., 2012). Whole cell lysates were separated by SDS-PAGE

and then either stained with coomassie (Fig. 3.4B), or western blotted for probing with

antiserum CjNgp (Fig. 3.4A).

As seen previously, CjNgp reacted with C. jejuni 11168H but not the corresponding pglB

knockout (Fig. 3.4A). As expected given the results of the chicken isolates reactivity,

CjNgp reacted to a similarly strong degree with C. coli. Alongside these, CjNgp also gave

numerous reactive bands with C. upsaliensis and C. helveticus, species which produce an

identical N-linked heptasaccharide structure. With C. lari, CjNgp reacted with several

bands but at a lower intensity. Compared with the C. jejuni glycan structure, C. lari

produces a linear hexasaccharide lacking the glucose sugar (Schwarz et al., 2011),

suggesting that CjNgp binds this residue within C. jejuni N-glycan. CjNgp did not react

with whole cell lysates of any of the four remaining species tested (C. concisus, C. fetus,

C. hyointestinalis and C lanienae) (Fig. 3.4A). These four species all produce distinct N-

glycans that vary more considerably in structure compared to the C. jejuni group and C.

lari. These results confirm that CjNgp is highly reactive towards the N-linked glycan

structure produced by C. jejuni, C. coli, C. upsaliensis and C. helveticus and, to a lesser

extent, the linear hexasaccharide structure produced by C. lari. However, CjNgp does not

bind to C. concisus, C. fetus, C. hyointestinalis or C. lanienae N-linked glycans. These

data indicate CjNgp could specfically detect both the common and so-called ‘emerging’

Campylobacter pathogens found in poultry, but not cross react with other Campylobacter

species.

70

71

3.3. CjNgp binding untreated cells of Campylobacter.

3.3.1. CjNgp binding to untreated C. jejuni cells is not dependent on N-linked

glycosylation.

Many assays for detecting/identifying Campylobacter require samples to be pre-treated,

for example cells are lysed or DNA extracted, before analysis. A reagent that can directly

bind untreated cells would allow preparation time to be reduced. Hence, for CjNgp to be a

successful reagent in an antibody-based Campylobacter detection system, it would be

useful if it could bind untreated Campylobacter cells. To investigate this, CjNgp was used

to probe dot blots containing untreated cells of C. jejuni. Furthermore, to assess if

antibody reactivity was against N-glycan, C. jejuni pgl mutants were also included in the

experiment. Although a pglB mutant does not produce N-linked glycoproteins, the glycan

structure is still in the cell in its precursor form, lipid-linked oligosaccharide (LLO) (Jervis

et al., 2012). This lipid-linked precursor migrates very rapidly in SDS-PAGE of cell

lysates so that it does not interfere with antibody-protein interactions, making the strain a

suitable control in 3.2. However, it is possible that dot blots containing untreated cells may

present LLO in this strain. Therefore, it was thought that a pglB mutant may not be the

most appropriate negative control. To circumvent this potential issue, the pglA mutant,

which produces a severely truncated N-glycan, was also included. As shown in 3.2.1.,

CjNgp does not react with the N-glycan structure present in this mutant. Dot blots

performed with no primary antibody present or using pre-immune serum gave no

reactivity (not shown). As a positive control, anti- whole cell C. jejuni 11168 (anti-WC

11168) antiserum was included (Fig. 3.5B) (Linton, unpublished). Dot blots were

performed similarly to the protocol described by Qian et al. (2008). Briefly, the dot blot

procedure involved harvesting cells from agar plates following 24 hours growth, washing

in 1ml PBS, and resuspending at various OD600 values. Cell numbers are estimated based

on OD600 of 1 equating to around 1 x 109 C. jejuni cells (Qian et al., 2008). One µl of each

cell suspension was pipetted onto nitrocellulose membranes and air dried for 15 minutes.

Membranes were then blocked, probed and washed as in a western blotting procedure (see

methods).

When used in a dot blot assay, anti-WC 11168 detected as little as 104

cells of C. jejuni

wildtype, pglB::kn and pglA::kn strains (Fig. 3.5B). Although this limit of detection was

consistent with each strain, the observed reactivity was lower with the pglB mutant in

comparison to the wildtype and pglA mutant. When CjNgp was used to probe a dot blot, it

72

could detect 105

cells and above (Fig. 3.5A). Using CjNgp, slightly more reactivity was

seen with the wildtype strain in comparison to each knockout. These data show that CjNgp

is able to bind to untreated cells of C. jejuni in a dot blot assay format. However, the

antiserum does not perform as well as anti-WC 11168 antiserum. Furthermore, cells

lacking the pglB or pglA genes were still bound by CjNgp to an extent not far below that

observed with the wildtype. This suggests that the cell binding observed is not due to N-

linked glycan reactivity within the antiserum. As CjNgp displays reactivity against Cj0114

(the native progenitor of NGRP) (Section 3.2.1.), reactivity against this protein may be

responsible for C. jejuni cell binding.

73

3.3.2. Reactivity against the NGRP protein influences the ability of CjNgp to bind

untreated C. jejuni cells.

Section 3.3.1 demonstrated that the ability of CjNgp to detect untreated cells of C. jejuni

was not dependent on N-linked glycosylation. Using western blotting, the antiserum was

inferred to react with the Cj0114 protein, the basis for production of the N-glycan carrier

immunogen, NGRP (See section 3.2.1.). Therefore, the NGRP reactivity within the

antiserum may be influencing its ability to bind untreated C. jejuni cells. To investigate

this, dot blots were performed using CjNgp that had been pre-treated to adsorb out

reactivity against unglycosylated NGRP. To achieve this, CjNgp was pre-incubated with a

membrane containing lysates of E. coli BL21 expressing NGRP prior to use in a dot blot

assay. To account for the loss of antibodies due to non-specific binding, pre-incubation

with a blank membrane was performed as a control. Also, to ensure no loss of antibody

binding was due to adsorption of E.coli-reactive antibodies, CjNgp pre-treated with a

membrane containing BL21 lysates was also included. Following each pre-incubation,

CjNgp was used to probe dot blots containing untreated cells of C. jejuni wildtype, C.

jejuni pglB::kn and C. jejuni pglA::kn, as previously (Fig. 3.6). In addition a spot of

purified NGRP was included to assess NGRP reactivity present.

When pre-incubated with a blank membrane, CjNgp weakly reacted with as few as 105

cells of all strains tested (Fig. 3.6A). As seen previously, CjNgp reacted similarly with C.

jejuni cells with and without functional pglB or pglA genes, giving a strong signal with

each of the strains when 107

cells were present. CjNgp gave a very strong reaction with

purified NGRP. Upon pre-incubation with BL21 lysates, no change in cell binding was

observed (Fig. 3.6B). In addition, a similarly strong signal was seen with purified NGRP.

When CjNgp was pre-incubated with BL21 NGRP lysates, a marked change in cell

binding was observed (Fig. 3.6C). This pre-adsorbed CjNgp could only reasonably detect

5 x 105

C. jejuni wildtype cells and the overall intensity of reactivity was considerably

lower. When probing C. jejuni pglB::kn cells, a signal was only identifiable with 107

cells.

With C. jejuni pglA::kn, there was no signal at all using CjNgp. However, even following

this BL21 NGRP pre-incubation, some slight NGRP reactivity still remained.

In summary, removing most of the NGRP-reactive antibodies from CjNgp resulted in a

reduction in its ability to bind C. jejuni cells. In this assay, with wildtype C. jejuni cells, its

limit of detection appeared to shift from 105 cells to 5 x 10

5 cells upon removing the

majority of NGRP-reactive antibodies. Moreover it could no longer detect C. jejuni cells

lacking pglA and only gave a sufficient signal with a pglB mutant when 107

cells were

74

present. Therefore, these data suggest that CjNgp contains a large proportion of antibodies

reactive with the NGRP protein rather than the heptasaccharide N-glycan. Furthermore,

these data suggest that these NGRP-reactive antibodies considerably contribute to, but are

not fully responsible for, the ability of CjNgp to bind untreated cells.

75

3.3.3. CjNgp binding to untreated cells of Campylobacter species

As described in previous sections, CjNgp can bind to untreated cells of C. jejuni,

suggesting it would be feasible to integrate this antiserum into a C. jejuni detection tool.

Moreover, there are several other Campylobacter species found alongside C. jejuni in

poultry that are also known to cause human disease. A tool able to detect such

Campylobacter species would be useful. This work has previously demonstrated that

CjNgp is highly reactive with lysates of disease-causing Campylobacter species found on

poultry, but not with lysates of other Campylobacter pathogens, including the veterinary

pathogen, C. fetus venerealis (section 3.2.4). To investigate if this pattern is maintained

with regards to untreated cell binding, CjNgp was used to probe dot blot assays containing

five Campylobacter species. Again, anti-WC 11168 was used for comparison.

In this dot blot assay anti-WC 11168 could detect 5 x 105

cells of each of the five species

used (Fig. 3.7B). However, this antiserum reacted with C. fetus fetus considerably less in

comparison to the remaining four species (C. jejuni, C. coli, C, upsaliensis and C. lari).

When CjNgp was used, it could detect 5 x 105

cells and above of C. jejuni, C. coli, C,

upsaliensis and C. lari (Fig. 3.7A). However, it gave the strongest reaction with C. jejuni

and C. coli, then C. upsaliensis then C. lari. CjNgp was not able to bind upto 8 x 105

cells

of C. fetus.

In conclusion, when testing against untreated cells, the Campylobacter species specificity

of CjNgp was similar to the observation made when using cell lysates. This observation is

that CjNgp is most reactive towards Campylobacter species producing the N-linked

heptasaccharide structure, gives lower reactivity with C. lari and harbours no reactivity for

C. fetus. Here it was also shown that, as previously found with C. jejuni cells in section

3.3., CjNgp has a lower affinity for the Campylobacter species tested here in comparison

to the anti-WC 11168 serum. However, in contrast to the anti-WC 11168 serum, CjNgp

does not cross react with untreated cells of C. fetus.

76

77

3.3.4. CjNgp labels whole cells of C. jejuni 11168H

Previous analysis using dot blot assays demonstrated that CjNgp can bind untreated cells.

However, cells may have lysed using this method as it involves drying the cells onto a

membrane. As N-linked glycoproteins are either periplasmic or membrane proteins, a

considerable percentage could be surface-exposed and accessible for antibody binding

when cells are whole. Therefore, to assess if CjNgp is able to label whole cells of C. jejuni

flow cytometric analysis was used. To allow for cell detection, the green fluorescent

protein (GFP)-producing strain C. jejuni 11168Hgfp4 developed by Jervis et al. (2015)

was utilised. The FITC-channel was used to detect GFP (Fig. 3.8 x-axes). Antibody

labelling of cells was detected using R-phycoerthrin (R-PE) goat anti-rabbit IgG secondary

antibody (Fig. 3.8, y-axes). Cells not incubated with antiserum were included to rule out

any staining resulting from the secondary antibody (Fig. 3.8A). As a positive control, anti-

whole cell C. jejuni 11168 (anti-WC 11168) antiserum was included (Fig. 3.8B) (Linton,

unpublished). Finally, pre-immune rabbit serum served to investigate if non-specific

binding occurred with normal rabbit serum alone (Fig. 3.8D).

GFP production was unexpectedly low in all samples (Fig.3.8 x-axes), however, the

analysis still gave a useful indication of antibody labelling. As seen in figure 3.8A, C.

jejuni cells not treated with serum displayed very minimal R-PE labelling. When

incubated with anti-WC 11168, the cells exhibited an obvious shift in R-PE staining, with

the majority of the population giving a high signal (Fig. 3.8B). In contrast, pre-immune

rabbit antiserum did not confer significant R-PE staining of C. jejuni (Fig. 3.8D). Finally,

a large majority of the population exhibited R-PE staining upon treatment with CjNgp

antiserum (Fig. 3.8C). However, this shift was seen with fewer of the population and the

signal achieved was not as intense in comparison to with the anti-WC 11168 antiserum.

Therefore, CjNgp appears capable of labelling whole cells of C. jejuni 11168, but not to

the same as extent as an anti-whole cell C. jejuni 11168 antiserum.

78

79

3.4. Discussion

This Chapter described the characterisation of an antiserum, CjNgp, raised against a C.

jejuni N-glycoprotein, NGRP, expressed in glycocompetent E. coli. CjNgp antiserum was

assessed with regards to its Campylobacter reactivity; its N-linked glycan structural

specificity, strain coverage, species specificity and the ability to bind Campylobacter cells.

The overall aim was to preliminarily assess the feasibility of integrating the antiserum into

a Campylobacter detection/identification test.

The CjNgp antiserum was very reactive with a C. jejuni 11168H lysate and gave only a

single band of reactivity upon deletion of pglB, demonstrating its reactivity against the N-

linked glycan structure. The ~ 35 kDa CjNgp-reactive band observed with a pglB mutant

likely corresponds to Cj0114, the native protein from which NGRP was developed. A

band putatively corresponding to this protein was observed within each C. jejuni pgl

mutant lysate tested – with its mobility changing dependent on N-glycan length.

Therefore, CjNgp harbours reactivity against this protein with and without N-linked

glycan modification. CjNgp N-glycan reactivity was influenced by loss of the branching

glucose sugar upon deletion of pglI. Further truncation of the N-glycan suggested the

antiserum is most reactive with the terminal four residues (GalNAc3Glc). An antiserum,

termed hR6, specific for the C. jejuni N-glycan already exists in the literature (Amber and

Aebi, unpublished). In contrast to these data, Nothaft et al. (2009) demonstrated that hR6

reacts equally well with C. jejuni wildtype and pglI::kn lysates, suggesting its reactivity is

independent of the branching glucose sugar. Instead its reactivity is mostly dependent on

the terminal three GalNAc residues, having very minimal reactivity against a pglH::kn

lysate and no reactivity if the N-glycan is further truncated. This difference is not

surprising as these are polyclonal antisera and is likely due to varying antibody responses

in the immunized rabbits.

The pgl locus is conserved within C. jejuni, however, prediction and/or experimental

identification of C. jejuni N-glycoproteins has only been performed in a few strains (Ding

et al., 2009; Scott et al., 2011). Therefore the extent of variation in the size of the C. jejuni

N-linked glycoproteome among different strains is undetermined. The amount of N-linked

glycosylation that takes place within C. coli is even less understood. CjNgp reacted with a

multitude of bands within the lysates of 11 C. jejuni reference strains and clinical isolates,

demonstrating that an equally large N-glycoproteome is present in several members of this

species. Furthermore, CjNgp displayed consistently high reactivity within C. jejuni and C.

coli strains recently isolated from retail chicken meat. These observations provide further

80

evidence that the N-linked glycan structure produced by C. jejuni and C. coli is well

conserved and that N-glycosylation remains a predominant protein modification among

strains of these species. Furthermore, it suggests this N-glycan-reactive antiserum could

create a detection tool suitable for a variety of C. jejuni and C. coli strains, including those

recently contaminating poultry.

The Campylobacter species specificity of CjNgp was assessed by probing western blots of

whole cell lysates of a variety of species. CjNgp was very reactive towards the three

Campylobacter species (C. coli, C. upsaliensis and C. helveticus) known to produce the C.

jejuni-type heptasaccharide N-glycan structure. This is in agreement with the findings of

Jervis et al. (2012) and Nothaft et al. (2012) when using hR6 antiserum, and provides

further evidence that C. jejuni N-glycan-reactive antisera are promising tools for detecting

these Campylobacter pathogens. However, in contrast to the characterised reactivity of

hR6, CjNgp was considerably less reactive with C. lari. As C. lari produces a linear N-

glycan identical to that of a C. jejuni pglI mutant, this observation further proves the

importance of this branching glucose residue for CjNgp reactivity. These data demonstrate

a drawback of using polyclonal sera for detection tools, as variability in rabbit responses

may result in such differences in reactivity. However, these data suggest the CjNgp

antiserum could be promising for the broad detection of Campylobacter species found in

chickens, including the common C. jejuni and C. coli and those considered to be emerging

pathogens (C. upsaliensis and C. lari) (Man, 2011). Such broad reactivity against

pathogenic Campylobacter species found in poultry is favourable as methods for

Campylobacter detection are often restricted to C. jejuni and C. coli. For example, Miller

et al. (2008) found that three commercially available Campylobacter LAT assays were

unable to identify any of the eight C. upsaliensis strains tested. Furthermore, these data

suggest that a detection tool using CjNgp could be similarly sensitive when detecting C.

jejuni, C. coli and C. upsaliensis. This would be another advantage over currently

available methods of detection because, for example as seen with an immunosensor

developed by Sapsford et al. (2004), they can have lower sensitivity for Campylobacter

species other than C. jejuni.

CjNgp did not react with cell lysates of any other Campylobacter species tested, each of

which are known to produce N-glycan structures distinct from that of C. jejuni (Jervis et

al., 2012; Nothaft et al. 2012). The hR6 antiserum described elsewhere also displays an

absence of reactivity with these species (Jervis et al., 2012; Nothaft et al., 2012). Therefore

CjNgp has the potential to be used to identify common Campylobacter human

81

gastrointestinal pathogens without cross reacting with other species such as C. fetus, C.

hyointestinalis, C. concisus or C. lanienae.

To integrate CjNgp into a rapid detection/identification tool it would be advantageous for

it bind to Campylobacter cells without pre-treatment steps. Dot blot assays showed that

CjNgp could bind untreated cells of C. jejuni wildtype, pglB::kn and pglA::kn strains to a

similar extent, suggesting N-glycosylation was not important for cell binding. Adsorbing

NGRP-reactive antibodies from the antiserum demonstrated that NGRP reactivity

considerably influenced the ability of CjNgp to bind C. jejuni cells. However, the

remaining antibodies within CjNgp were able to bind C. jejuni wildtype cells but not C.

jejuni pglA::kn cells, suggesting that these antibodies were N-glycan-reactive. These

antibodies were also able to bind C. jejuni pglB::kn cells, but to a reduced extent. This

reduction in binding as opposed to complete loss is likely due to the presence of LLO

within these pglB::kn cells. Therefore, NGRP reactivity considerably influences the ability

of CjNgp to bind untreated cells of C. jejuni, even though it is a periplasmic protein. The

reason this protein is readily accessible in a dot blot assay may be due to surface damage

of the cells during preparation. Nevertheless, it appears that the N-glycan reactive

antibodies within CjNgp are able to bind untreated C. jejuni cells, and thus C. jejuni N-

glycan remains a potential target for developing a rapid antibody-based detection assay.

Dot blot assays also demonstrated CjNgp can bind cells of C. coli, C. upsaliensis and C.

lari but not C. fetus. In contrast to the similar reactivity against whole cell lysates of C.

jejuni, C. coli and C. upsaliensis, CjNgp was more reactive with C. jejuni cells than those

of the other species. It should be noted that NGRP-reactive antibodies may have bound to

these cells as they contain Cj0114 homologues but that such antibodies would likely be

more reactive with Cj0114 and thus have higher affinity for C. jejuni. However, as N-

glycan-reactive antibodies present within CjNgp can bind untreated cells of C. jejuni, it is

envisaged that purification of the N-glycan-reactive antibodies from this antiserum would

still provide a reagent able to similarly bind C. jejuni, C. coli, C. upsaliensis and C. lari

cells. Furthermore, in contrast to anti-WC 11168, CjNgp does not cross react with C. fetus

cells. Therefore, an advantage of CjNgp over an anti-whole cell serum is that it is specific

for the common gastrointestinal Campylobacter pathogens found in poultry.

Flow cytometric analysis demonstrated that CjNgp could bind whole cells of C. jejuni,

suggesting that N-glycans may be significantly surface-exposed. Surface localisation of N-

glycan in C. jejuni has previously been shown with hR6 antiserum using fluorescent

microscopy (Alemka et al., 2013). However, flow cytometric analysis here indicated that

82

an N-glycan reactive antiserum may not label C. jejuni cells as efficiently as an anti-whole

cell 11168 antiserum. Nevertheless, CjNgp still has the advantage of specificity for the

common Campylobacter human pathogens found on chicken meat. Furthermore, future

work to purify and concentrate N-glycan reactive antibodies from the antiserum could

allow enhanced Campylobacter sensitivity (See Chapter 7).

This chapter has demonstrated that CjNgp is an antiserum reactive against the C. jejuni N-

linked glycan structure, more specifically the four terminal sugar residues (GalNAc3Glc).

This reactivity results in the ability to react with a multitude of C. jejuni and C. coli N-

linked glycoproteins from numerous strains, including reference strains, clinical isolates

and recent retail chicken meat contaminants. In addition CjNgp reacts significantly with C.

upsaliensis, C. helveticus and, to a lesser extent, C. lari N-glycoproteins. Importantly, it

does not cross react with the N-linked glycoproteins produced by other species of

Campylobacter such as C. fetus and C. consisus. Furthermore, this antiserum is able to

bind untreated cells of C. jejuni, C. coli, C.upsaliensis and C. lari. A portion of this

binding activity is due to presence of protein-reactive antibodies present in the serum.

However, upon removing these antibodies, CjNgp still exhibits the ability to bind to

untreated cells of C. jejuni. Furthermore, CjNgp was demonstrated to label whole cells of

C. jejuni. Therefore, CjNgp has been shown to be a promising candidate for development

of a tool capable of detecting harmful poultry meat contaminants such as C. jejuni, C. coli,

C.upsaliensis and C. lari.

83

Chapter 4. Development and

characterisation of an antiserum raised

against a Campylobacter fetus N-linked

glycoprotein

84

4.1 Introduction

C. fetus consists of three subspecies, C. fetus fetus (C. f fetus), C. fetus venerealis (C. f

ven) (Veron and Chatelain, 1973) and the more recently proposed C. fetus testudinum

(Patrick et al., 2013). C. f fetus causes sporadic abortion in sheep and cattle (van Bergen et

al., 2005). However, it is also an opportunistic human pathogen capable of systemic

infections (Wagenaar et al., 2014). C.f ven is the causative agent of Bovine Genital

Campylobacteriosis (BCG), which is a significant burden to the cattle industry (Mshelia et

al., 2010). C. fetus testudinum is associated with reptiles but has also proven to be an

opportunistic human pathogen (Patrick et al., 2013).

Traditional culture and isolation of C. fetus is known to be difficult and solely relying on

this method can allow infected herds to remain undiagnosed (Truyers et al., 2014).

Although numerous molecular assays have been developed, most are better suited to

research than routine diagnostics (van der Graaf-van Bloois et al., 2013). Furthermore,

these methods require laboratories with sufficient equipment, a factor which is limiting for

developing countries (Mshelia et al., 2010). Instead, simple detection methods that can be

used in the field are desirable. Antibody-based detection is suitable for such on-farm

detection as it can be quick, easy and require very little sample preparation. Current

immunological methods utilised for diagnosis of bovine genital campylobacteriosis tend to

have poor sensitivity and can be misleading due to reliance on the varying humoral

responses of cattle (Truyers et al., 2014). For this reason it has been suggested that several

methods should be used in parallel to obtain accurate results (Truyers et al., 2014).

Therefore, improved tests are needed. In addition, such assays could be applicable to

detecting C. fetus in human samples as there are currently none available.

As described in 3.1, Campylobacter N-linked glycans are promising target candidates for

antibody-based detection tools. C. fetus produce N-glycan structures distinct from that of

all other Campylobacter species except for C. hyointestinalis (Jervis et al., 2012; Nothaft

et al., 2012). C. hyointestinalis consists of two subspecies, C. hyointestinalis lawsonii and

C. hyointestinalis hyoinestinalis (C. hyo hyo) (On et al., 1995). The former is associated

with swine whereas the latter is isolated from bovine intestinal tracts (Miller et al., 2016).

Therefore, as C. hyo hyo is not specifically found in the reproductive tract of cattle it

should not pose a serious cross contamination risk when screening for C. fetus. However,

C. hyo hyo can lead to fatal enteritis in calves (Diker et al., 1990) and detection of this

species in significantly contaminated preputial/vaginal samples could also be favourable.

Additionally, C. hyointestinalis has been associated with human gastrointestinal illness

85

and even bacteraemia (Gorkiewicz et al., 2002; Pasternak et al., 1984). Hence, C. fetus N-

glycans could be utilised to develop a test suitable for detecting this species, and possibly

C. hyointestinalis, in agricultural animals and/or human samples. Antisera, termed GRPII-

1 and GRPII-2, raised against each of the two reported C. fetus free oligosaccharide

structures display considerable reactivity with numerous proteins in C. fetus fetus and C.

fetus venerealis .(Nothaft et al., 2012). Although promising, each antiserum displayed

varying characteristics and further research into the potential for C. fetus N-glycans to be

used as detection targets was warranted. The study described here aimed to develop a C.

fetus N-linked glycan-reactive polyclonal antiserum to allow for further research.

Firstly, due to the ease and success of using NGRP as the glycoprotein immunogen when

raising the CjNgp antiserum (see Chapter 3), it was chosen again here. Previous

production of glycosylated NGRP was undertaken in glycocompetent E. coli expressing

the C. jejuni pgl locus (Wacker et al., 2002). However, no such E. coli system exists for

producing N-glycoproteins with the C. fetus N-glycan structure. To develop an equivalent

system for C. fetus N-glycan production in E. coli would have meant laborious cloning

and screening with no guarantee of a correctly functioning end-product. Instead, this

project aimed to integrate NGRP into C. fetus to allow for glycosylation via the native pgl

system.

Following successful development of a C. fetus N-glycan-reactive antiserum, this project

aimed to characterise the serum to assess its feasibility to be used for C. fetus

identification or detection purposes. Similarly as in Chapter 2, the following features of

CfNgp were characterised:

C. fetus N-glycan reactivity.

C. fetus subspecies and strain coverage.

Campylobacter species specificity.

C. jejuni N-linked glycan structural specificity.

Ability to bind cells of Campylobacter species.

86

4.2 Production of a recombinant Campylobacter fetus N-linked

glycoprotein for use as immunogen.

4.2.1 Using a C. fetus chromosomal integration vector to integrate an N-

glycoprotein was not a suitable approach.

Genetic manipulation of C. fetus is difficult and available tools are relatively limited. Gene

integration within the 23S rRNA gene, rrl, on the C. jejuni genome has been demonstrated

(Karlyshev and Wren, 2005). These vectors exploit the fact that C. jejuni harbours three

identical copies of ribosomal RNA genes. As C. fetus also has multiple, identical copies of

ribosomal rRNA genes, this study aimed to similarly develop a suicide vector which could

allow integration of an N-glycoprotein gene into the 23S rRNA gene of C. fetus.

Briefly, a 1.6 kb region of the C. fetus rrl gene was amplified from C. fetus fetus 10482

genomic DNA (gDNA) using primers 1268 and 1269 (Table 2.1) and ligated to pGEMT-

easy vector (Promega) to give pGEM23SCf. Jervis et al. (2015) previously developed a

suicide vector, pCJC4, for high-level protein expression in C. jejuni which included a

construct containing the C. jejuni porA promoter and a chloramphenicol resistance cassette

(cat). This construct was utilised here and was amplified from pCJC4 using primers 1296

and 1297 (Table 2.1) which add BamHI sites to each end. This construct was inserted

within the centre of the rrl region of pGEM23SCf to give plasmid pCfporAcat. In chapter

3, using glycocompetent E. coli, NGRP was directed to the periplasm using the pelB

signal sequence of pET22b(+) to allow modification by the C. jejuni N-linked

glycosylation machinery (Jervis and Linton, unpublished). To produce glycosylated

NGRP in C. fetus a different approach was required. NGRP was amplified along with the

native Cj0114 signal peptide (Fig. 4.1A) using primers 1320 and 1322 (Table 2.1) which

add a C-terminal Hexa-Histag and an NdeI site to each end. This PCR product was cloned

between the porA promoter and the cat cassette within pCfporAcat to give plasmid pCfN

(Fig. 4.1B).

Successful transformation of pCfN into C. fetus was challenging to attain. Transformants

were checked by PCR which amplified both the porA promoter and the sequence of NGRP

using primers 546 and 1302 (Table 2.1) (not shown). Western blotting whole cell lysates

of positive clones demonstrated two weak anti-His(6)-reactive bands around 28 kDa and

31 kDa that were not present in wildtype C. fetus (Fig. 4.1C). Transformation of pCfN into

C. fetus pglB::kn was unsuccessful and therefore it could not be confirmed that these anti-

87

his(6)-reactive proteins were N-glycosylated versions of NGRP. Furthermore, the poor

yield of protein was not suitable for producing sufficient immunogen to raise an

antiserum. Therefore, an alternative approach was used to integrate an N-glycoprotein in

C. fetus.

88

4.2.2 Production of a C. fetus conjugative vector encoding an N-glycoprotein.

As discussed above, integrating an N-glycoprotein-encoding gene onto the chromosome of

C. fetus was challenging and unsuccessful for producing a modified N-glycoprotein in C.

fetus in this case. However, Kienesberger et al. (2007) have described a conjugative

plasmid transmission method for introducing exogenous DNA into C. fetus. This was

achieved using E. coli – C. fetus shuttle vectors which utilise the C. fetus sapA promoter to

drive expression of downstream genes (Kienesberger et al., 2007). One such vector,

pRYGG1 (Appendix Table 2), was used to express the target glycoprotein, NGRP, in C.

fetus.

Similarly to in section 4.2.1, NGRP was amplified along with the native Cj0114 signal

peptide from C. jejuni 11168H gDNA using primers 1427 and 1484 (Table 2.1). This

primer pair introduced a 5’ BamHI site and a 3’ polyhistidine (His8) tag-encoding

sequence followed by a SmaI site. The resulting 600 bp product (not shown) was digested

with BamHI/SmaI and ligated into SmaI and BamHI-digested pRYGG1 to give pG1-N

(Fig. 4.2D). Following transformation, colonies were checked by PCR using primer pair

1445 and 1484 (Fig. 4.2A) and NGRP insert sequenced using primers 1445 and 1432

(indicated in Fig. 4.2D). To allow for conjugative transfer, pG1-N was transformed into

electro-competent E.coli S17-1 λpir cells. The predicted size of NGRP with a His8-tag is

21.8 kDa. Production of an ~ 22 kDa His-tagged protein in an E.coli S17-1 λpir lysate was

confirmed by western blotting (Fig. 4.2B). These donor cells were subsequently used to

introduce pG1-N into both wildtype C. fetus fetus and the corresponding pglB::kn mutant

(Jervis et al., 2012) via conjugation and strains were verified by colony PCR using primers

1445 and 1484 (Fig. 4.2C).

89

90

4.2.3 Production of N-linked glycoprotein immunogen in C. fetus.

Whole cell lysates of C. fetus pG1-N and corresponding pglB::kn strain were subjected to

SDS-PAGE and coomassie staining to visualise NGRP production (Fig. 4.3A). E. coli

S17-1 λpir pG1-N was included as a reference. In addition, plasmid-free versions of each

strain were included as negative controls.

Consistent with Figure 4.1C, an ~ 22 kDa band was visible in the coomassie-stained cell

lysate of E. coli S17-1 λpir only when it harboured pG1-N (Fig. 4.3A). Up to four

glycoforms of NGRP (predicted ~ 23.2, 24.4, 25.6 and 26.8 kDa) were expected in C.

fetus pG1-N, but were not evident. Neither was unglycosylated NGRP (~ 22 kDa)

observed in C. fetus pglB::kn pG1-N.

To detect NGRP expression in these C. fetus strains, whole cell lysates were western

blotted and probed with anti-His6 antibodies (Fig. 4.3B). Similar to the observation in

4.2.2, an anti-His6 reactive band at around 22 kDa was observed with E. coli S17-1 λpir

pG1-N (Fig. 4.3B, lane 2). However, at least three anti-His6-reactive bands between 27

and 31 kDa in size were visible in a C. fetus pG1-N lysate (Fig. 4.3B, lane 3). Moreover,

in the N-glycosylation-deficient pglB mutant, a single anti-His6-reactive band was

observed at around 25 kDa (Fig. 4.3B, lane 4). Importantly, no anti-His6-reactive bands

were evident in C. fetus strains lacking pG1-N (Fig. 4.3B, lanes 5 and 6).

Although the presence of multiple anti-His6-reactive forms reducing to a single higher

mobility band upon pglB deletion indicates N-linked glycan modification, the mobility of

this form (Fig. 4.3B, lane 4) was less than predicted and of that observed in E. coli (Fig.

4.3B, lane 2). To ensure the observed proteins were NGRP-derivatives, the NGRP-reactive

CjNgp antiserum was also used to probe C. fetus lysates (Fig. 4.3C). CjNgp antiserum

reacts with NGRP and C. jejuni N-linked glycan but not with C. fetus N-linked glycan

(See 3.2.4). Hence, this antiserum would react with NGRP produced in C. fetus but would

not bind any C. fetus N-glycan. Each band reactive with anti-His6 antibodies was also

bound by CjNgp, suggesting these His-reactive bands are forms of NGRP. In addition, an

unexpected band reactive with CjNgp alone was observed at around 42 kDa. Therefore,

these data suggest that, although the mobility of unglycosylated NGRP was less than

observed in E. coli, NGRP was expressed and N-glycosylated in C. fetus. However, a

larger CjNgp-reactive protein was also observed. One explanation for this band is an

NGRP-derivative without an accessible His-tag. Nevertheless, as this unexplained band

was not anti-His6-reactive, it was not deemed problematic as glycosylated NGRP was to

be purified by its His-tag using Nickel affinity chromatography.

91

92

4.2.4 Purification of the N-linked glycoprotein immunogen from C. fetus.

Although some discrepancies were observed in the apparent size of NGRP produced in C.

fetus pG1-N, glycoforms of His-tagged NGRP were evident and Nickel-affinity

chromatography was employed to purify this glycosylated NGRP (gNGRP). Filtered cell

lysate was prepared from material harvested from 200 plates of C. fetus pG1-N. Lysate

was applied to a 5 ml HisTrap FF column, washed with 75 ml of 50 mM imidazole wash

buffer and eluted with 10 ml 500 mM imidazole elution buffer, with samples collected in 1

ml fractions (E1-10). Elution fractions 5 and 6 were separated by SDS-PAGE and

coomassie stained (Fig. 4.4A). Lysate, lysate flow through, wash flow through and elution

fraction 5 were separated by SDS-PAGE, western blotted and probed with anti-His6 and

CjNgp (Fig. 4.4B).

Elution fractions 5 and 6 contained bands between 27 and 31 kDa in size (Fig, 4.4A) that

were reactive with anti-His6 and CjNgp (Fig. 4.4B), confirming the presence of His-

tagged NGRP forms. These protein bands were analysed by mass spectrometry and the

results searched against C. jejuni and C. fetus protein sequences within Swiss-Prot and

TrEMBL databases. C. jejuni Cj0114 (YbgF), the native progenitor of NGRP, was the first

protein hit, confirming these proteins are NGRP. However, glycosylated NGRP had only

been partially purified, as other proteins were evident in the eluate (Fig. 4.4A). Three

distinct bands at 42, 44, and 46 kDa were CjNgp-reactive but not anti-His6 reactive (Fig.

4.4B), although they had been enriched by Nickel-affinity purification. Furthermore, the

pattern of these CjNgp-reactive bands was not identical to that observed with coomassie-

staining, as a stronger band was evident at around 46 kDa when coomassie-stained. To

identify this ~ 46 kDa protein the band was excised, subjected to mass spectrometry and

the results searched against C. jejuni and C. fetus protein sequences within Swiss-Prot and

TrEMBL databases. The protein with the most peptide matches (56) was a C. fetus

glutamate dehydrogenase (CFF8240-1557) with a theoretical size of 49.8 kDa and the

second hit was C. jejuni Cj0114 (9 peptide matches), the native protein from which NGRP

was constructed. This suggested that the significant band at ~ 46 kDa corresponded to

CFF8240-1557 but the three CjNgp-reactive proteins were derived from NGRP. As these

forms were not anti-His6-reactive, the possibility of these proteins being translational

read-through products was explored by analysing the codons downstream of NGRP in

pG1-N. However, this is unlikely as it would only give a ~ 25 kDa protein (Maximum size

being ~ 29.8 kDa if four N-glycan sites modified.).

93

Attempts to further purify gNGRP using ion exchange chromatography were unsuccessful

(not shown). An absorbance peak at the end of the salt concentration gradient

corresponded to fractions containing similar protein bands as prior to the chromatography

and thus no separation was observed. Due to poor yield and time constraints, further

rounds of purification were not undertaken. Instead, nickel affinity chromatography

elution fractions were pooled and dialysed to remove imidazole and reduce salt levels. The

preparation was concentrated to ~ 2 ml and a sample ran alongside BSA standards and

coomassie-stained (Fig. 4.4C). The total amount of NGRP glycoforms present was

estimated to be around 1 mg using densitometry within Image Studio software (LiCor

biosciences). Aliquots of the preparation were lyophilised and sent to Antibody Production

Services Ltd (Life Science Group Ltd.) for production of polyclonal rabbit antisera using

their standard 77 day immunisation protocol.

94

95

4.3 Characterisation of C. fetus NGRP antiserum (CfNgp)

reactivity with Campylobacter whole cell lysate extracts.

4.3.1 CfNgp reactivity with C. fetus fetus NCTC 10842.

As described in section 4.2, a recombinant N-glycoprotein was produced and partially

purified from C. fetus NCTC 10842 and used as an immunogen to develop a C. fetus N-

glycan-reactive polyclonal antiserum. The resulting antiserum was named CfNgp (C. fetus

N-glycoprotein). To initially identify if this antiserum contained any C. fetus N-glycan

reativity a test bleed sample (day 49) was used to probe a western blot containing whole

cell lysates of C. fetus 10842 wildtype and pglB::kn strains (Fig. 4.6C). The observed

reactivity was compared to that of the pre-immune serum (Day 0) (Fig. 4.5B).

Pre-immune serum gave numerous faint reactive bands identical between wildtype and

pglB knockout (Fig. 4.5B). Day 49 CfNgp serum displayed considerably higher reactivity

with the C. fetus cell lysates (Fig. 4.5C). When probing C. fetus wildtype lysate, CfNgp

resulted in many reactive bands. In contrast, with C. fetus pglB::kn lysate, CfNgp gave

only one strong reactive band at around 49 kDa. In summary, immunization with

glycosylated NGRP resulted in an antiserum reactive against numerous C.fetus proteins in

a pglB-dependent manner, indicating they were N-linked glycoproteins. The antiserum

also contained reactivity against a single pglB-independent band, likely the contaminating

C. fetus glutamate dehydrogenase (predicted 49.8 kDa) identified in the partially purified

NGRP preparation used for immunization. These data indicate the potential of CfNgp for

detecting C. fetus N-linked glycoproteins. In the following sections the antiserum was

used in its final form (day 77 terminal bleed) and employed at the most appropriate

dilution (1/1000), as decided by probing blots similar to Figure 4.5C with CfNgp at a

variety of dilutions (not shown).

96

97

4.3.2 CfNgp reactivity with numerous C. fetus strains.

Previous N-glycan structural characterisation work in C. fetus has identified two distinct

structures in both C. fetus fetus 82-40 and C. fetus venerealis, α-GlcNAc-6-[ β-Glc-3]-α-

GlcNAc-4-α-GalNAc-3-α,β-diNAcBac and α-GlcNAc-6-[β-GlcNAc-3]-α-GlcNAc-4-α-

GalNAc-3-α,β-diNAcBac, with the latter being most predominant (Nothaft et al., 2012).

However, only a HexNAc-[Hex]-HexNAc3-diNAcBac N-glycan species was identified in

the strain used in this thesis, C. fetus fetus 10842 (Jervis et al., 2012). To ensure CfNgp

antiserum is equally reactive with strains of C. fetus other than in which the immunogen

was produced, it was used to probe western blots of cell lysates from a variety of C. fetus

strains (Fig. 4.6B). This included C. fetus 10842 wildtype and pglB::kn, C. fetus veneralis

10354 as well as 18 C. fetus human isolates obtained from Public Health England (PHE).

PHE isolates were collected from a variety of human samples (including blood, tissue and

faeces) between 2004 and 2015 (Table 2.1). Although only two have been subtyped as C.

fetus fetus, it is likely that most or all are also this subspecies as C. fetus venerealis is very

rarely found in humans (Wagenaar et al., 2014). All samples were seperated by SDS-

PAGE and coomassie stained to confirm protein level consistency (Fig. 4.6A, isolates 9-

18 are not shown).

As in the previous section, CfNgp gave numerous immunoreactive bands with wildtype

C.ff 10842, all except one of which were pglB-dependent (Fig. 4.6B). With the C. f ven

lysate, CfNgp exhibited similar reactivity, with a notable band at less than 15 kDa. With

all the C. fetus human isolates, CfNgp gave a near identical pattern to that observed with

C. f fetus 10842 (Fig. 4.6B, isolates 9-18 not shown). Therefore, CfNgp reacts similarly

well with cell lysates of numerous C. fetus strains, encompassing both subspecies.

98

99

4.3.3 Campylobacter species specificity of CfNgp.

Previously a C. jejuni N-linked glycan-reactive antiserum was shown to specifically react

with species that produce structurally related N-linked glycan (section 3.2.4). This

antiserum does not react with species of Campylobacter such as C. fetus and C. concisus

which produce a variety of less structurally related N-glycan structures. This feature

makes the antiserum a promising reagent for creating a detection tool specific for an

important set of Campylobacter pathogens. To a similar end, the species specificity of

CfNgp was determined by probing western blots of whole cell lysates of a variety of

Campylobacter species (Fig. 4.7B).

CfNgp produced numerous immunoreactive bands with wildtype C. fetus lysate and only a

single ~ 49 kDa band with a pglB mutant (Fig. 4.7B). CfNgp also reacted with all the

remaining Campylobacter species tested, which could be divided into a higher and a lower

reactivity group. With the higher reactivity group (consisting of C. hyointestinalis, C.

concisus, C. lanienae and C. sputorum) the immunoreactivity observed was similar to that

observed with C. fetus. This is unsurprising for C. hyointestinalis as it produces identical

N-glycan structures to C. fetus (Jervis et al., 2012; Nothaft et al., 2012). However, this

level of reactivity was not expected with C. concisus, C. lanienae and C. sputorum as they

produce related but structurally distinct N-glycan structures.

The lower reactivity group (C. jejuni, C. coli, C. upsaliensis, C. helveticus and C. lari)

produced fewer immunoreactive bands with reduced intensity (Fig. 4.7B). These five

species are reactive with the C.jejuni N-glycan antiserum, CjNgp (Fig. 3.4, section 3.2.4),

and produce either the canonical C. jejuni-type N-glycan heptasaccharide or the C. lari

hexasaccharide structure lacking a branching glucose (Jervis et al., 2012; Nothaft et al.,

2012). Therefore, although CfNgp reacts with numerous N-glycoproteins produced by C.

fetus and C. hyointestinalis, it is not specific for their N-glycan structures as it also reacts

similarly with cell lysates of C. consisus, C. sputorum and C. lanienae. CfNgp also reacts

with whole cell lysates of C. jejuni, C. coli, C. upsaliensis, C. helveticus and C. lari,

however, the observed reactivity was of lower intensity. This suggests that the antiserum

contains antibodies against diverse N-glycan structures and this was investigated further

using C. jejuni pgl mutants that produce a variety of N-glycan structures.

100

101

4.3.4 C. jejuni N-linked glycan structural specificity of CfNgp.

Section 4.3.3 demonstrated the broad reactivity of CfNgp against Campylobacter species,

with a higher and a lower reactivity group identified. Species within the higher reactivity

group produce N-glycans more structurally related to C. fetus than those in the lower

reactivity group. To prove the immunoreactive bands observed with species in the lower

reactivity group were N-glycoproteins, CfNgp was used to probe a western blot of C.

jejuni wildtype and pglB::kn lysates. To further elucidate the C. jejuni N-glycan structural

specificity of CfNgp, lysates of C. jejuni pgl mutants producing truncated N-glycans were

also tested (Fig. 4.8B). C. fetus fetus wildtype and pglB::kn lysates were included as a

positive and negative control, respectively.

As seen previously, CfNgp reacted with a number of proteins in a C. jejuni cell lysate (Fig.

4.8B). Upon deletion of pglB, only one reactive band remained, with the rest being pglB-

dependent and hence a result of N-glycan reactivity. This remaining reactive protein (~ 35

kDa) is likely Cj0114, the protein from which NGRP was constructed, as glycosylated

NGRP was used as immunogen and CfNgp undoubtedly contains antibodies against the

protein. When the branching glucose residue was absent from the C. jejuni N-glycan (pglI

mutant), CfNgp produced a similar number of reactive bands. Interestingly, in a few cases

this truncation appeared to increase the intensity of the reactivity. Upon further loss of the

three terminal GalNAc residues (pglH mutant), CfNgp was reactive with fewer proteins.

Further truncation in pglJ and pglA mutants resulted in only a single reactive protein, at

around 37 kDa and 36 kDa, respectively. These remaining reactive bands correspond to

Cj0114 harbouring the respective truncated N-glycan structures, hence antibodies within

CfNgp are likely reactive against the protein and not the truncated N-glycan structures.

In conclusion, the majority of CfNgp reactivity with a C. jejuni cell lysate is pglB-

dependent, with the exception of the reactivity against Cj0114. Furthermore, CfNgp C.

jejuni N-glycan reactivity is independent of the branching glucose, the presence of which

appears to reduce the level of reactivity in some cases. Instead, the majority of the CfNgp

C. jejuni N-glycan reactivity is directed towards the four terminal GalNAc residues.

102

103

4.4 CfNgp binding to Campylobacter cells.

4.4.1 CfNgp binding untreated cells of C. fetus.

Bacterial detection tools often require pre-treatment of cells, for example, a cell lysis or

DNA extraction step. However, if a reagent can effectively bind untreated cells the

preparation time can be reduced, allowing faster detection. Therefore it would be useful if

CfNgp could react with untreated cells of C. fetus. To test this, CfNgp was used to probe

dot blots containing untreated cells of C. fetus fetus and C. fetus venerealis (Fig. 4.9A).

Polyclonal serum raised against cells of C. jejuni 11168H, anti-WC 11168, was also used

for comparison as it is known to react with C. fetus cells (See section 3.3.4) (Fig. 4.9B).

Similar blots performed without primary antibody or probed with pre-immune serum gave

no reactivity (not shown). Dot blots were performed similarly to the protocol described by

Qian et al. (2008), as explained in section 3.3.1.

In this dot blot assay, CfNgp could detect 103

or more cells of C. fetus fetus (Fig. 4.9A).

CfNgp had a lower affinity for cells of C. fetus venerealis, requiring a least 104

cells for

detection. In contrast, anti-WC 11168 was only able to detect as few as 10

5 cells of either

subspecies (Fig. 4.9B). Therefore, CfNgp can bind untreated cells of both C. fetus fetus

and C. fetus venerealis.

104

4.4.2 The influence of pglB on CfNgp C. fetus cell binding.

Section 4.4.1. demonstrated that CfNgp is able to bind untreated cells of C. fetus. To

investigate if this immunoreactivity is N-glycoprotein-dependent, the antiserum was used

to probe dot blots containing untreated cells of C. fetus wildtype and pglB::kn (Fig.

4.10B). The respective pre-immune serum was also tested (Fig. 4.10A). As discussed in

3.3.1., LLO is found within Campylobacter pglB mutants and hence may be bound by

anti-N-glycan antibodies if presented in a dot blot assay. In 3.3.1, this issue was taken into

account and a C. jejuni pglA mutant also included in the experiment. However, it should

be noted that a C. fetus pglA mutant was not available here. Hence, this experiment served

only to assess if cell binding was due to N-glycoprotein reactivity, not overall N-glycan

reactivity.

Due to the immunogen being only partially purified, there is some reactivity against C.

fetus proteins within the antiserum, as evidenced by mild reactivity with a pglB::kn lysate

upon western blotting (see section 4.3.1). Therefore, the influence of such reactivity on the

ability of CfNgp to bind cells was also investigated. To achieve this, some of this protein

reactivity was adsorbed from the antiserum by pre-incubating CfNgp with a membrane

containing C. fetus pglB::kn cell lysates before use in a dot blot (Fig. 4.10D). To account

for potential loss of reactivity due to non-specific binding to a membrane, CfNgp was also

pre-incubated with an empty membrane prior to dot blotting (Fig. 4.10C).

When probed with the rabbit pre-immune serum, no cells were detected (Fig. 4.10A).

Upon probing with CfNgp, as little as 103

cells of C fetus wt and pglB::kn were detectable,

with no difference in reactivity between each strain (Fig. 4.10B). Upon pre-incubation

with a blank membrane, the binding ability of CfNgp was unchanged (Fig. 4.10C).

However upon pre-incubation with a membrane containing C. fetus pglB::kn lysates, the

cell binding ability of CfNgp was reduced (Fig. 4.10D). Instead, the antiserum only gave a

significant signal when at least 104

cells of either strain were present. Furthermore,

although the limit of detection did not differ, this treated CfNgp appeared to bind wildtype

cells with a slightly stronger affinity than cells of the pglB mutant.

In conclusion, absorption of some C. fetus protein cross-reactivity reduced the ability of

CfNgp to bind untreated C. fetus cells. This suggests that unglycosylated protein reactivity

within CfNgp influences its ability to detect cells of C. fetus. Furthermore, although a

marginally higher affinity for wildtype cells indicated some cell binding was due to N-

glycoprotein-reactivity, it appeared the majority of cell binding activity within CfNgp was

not influenced by PglB activity. However, in contrast to a western blot, where the LLO

105

(likely abundant in a pglB::kn mutant) is not present, the antiserum may have had access

to this component in the untreated cells displayed by this dot blot method. Therefore, it is

likely that some proportion of the binding observed with C. fetus pglB::kn is due to the

availability of LLO and not C. fetus proteins. Nevertheless, these data suggest that both N-

linked glycan reactivity and protein reactivity contribute to the ability of CfNgp to bind C.

fetus cells.

106

4.4.3 CfNgp binding to untreated cells of Campylobacter species.

Previous experiments found CfNgp to be reactive against cell lysates of several

Campylobacter species, even those which produce N-glycan structures different from the

C. fetus N-glycan. This reactivity could be divided into two groups, one of high reactivity

and one of low reactivity. The high reactivity group contained C. fetus, C. hyointestinalis,

C. consisus, and C. lanienae. The low reactivity group consisted of C. jejuni, C. coli, C.

upsaliensis, C. helveticus and C. lari. Section 4.4.1. showed that CfNgp can bind to

untreated cells of C. fetus, which belongs to the high reactivity group. Here it was

investigated if CfNgp is able to bind untreated cells of species from the low reactivity

group in a similar dot blot assay. Cells of Campylobacter species from this ‘low reactivity’

group (excluding C. helveticus) were spotted onto nitrocellulose and probed with CfNgp

(Fig. 4.11A). The cell number spotted ranged from 5 x 103

to 104

cells, as this includes the

lower range at which CfNgp was previously observed to detect C. fetus cells. Again, anti-

WC 11168 was used for comparison purposes (Fig. 4.11B).

As expected, CfNgp detected between 5 x 103

and 104

cells of C. fetus fetus (Fig. 4.11A).

CfNgp was also able to detect C. jejuni, C. coli and C. lari at these cell numbers.

However, the observed signal was at a lower intensity. Furthermore, C. upsaliensis was

very weakly detected by CfNgp at these cell numbers. In contrast, anti-WC 11168 was

unable to detect C. fetus at any cell number used here (Fig. 4.11B). The remaining four

species were detectable by anti-WC 11168 at all cell numbers, with C. jejuni giving the

strongest signal. In conclusion, although CfNgp is able to easily detect between 5 x 103

and 104

C. fetus cells, the antiserum can also detect species from the ‘low reactivity’ group

within this cell number range. However, the reaction intensity is discriminatory between

C. fetus and those species from the ‘low reactivity’ group, as observed when analysing the

antiserum using western blotting.

107

108

4.5 Discussion

Rapid antibody-based tools such as lateral flow devices would be advantageous for

C. fetus detection in cattle and possibly identification from human samples also. C. fetus

N-linked glycan has the potential to be a useful target for detecting this species. Indeed,

antisera raised against BSA conjugates of each of the two C. fetus fOS structures can label

C. fetus fetus cells using fluorescence microscopy (Nothaft et al., 2012). This study aimed

to produce a C. fetus N-glycan-reactive antiserum and characterise its potential to be used

for C. fetus detection purposes. A recombinant glycoprotein (NGRP) was expressed and

glycosylated in the native C. fetus and purified. This purified N-glycoprotein was used to

raise polyclonal antiserum termed CfNgp. The Campylobacter reactivity of CfNgp was

assessed including its C. fetus strain coverage, Campylobacter species specificity, and

ability to bind untreated cells of Campylobacter.

This study described the first documented integration of a recombinant N-glycoprotein

into C. fetus. A His-tagged N-glycoprotein, NGRP, was first introduced using a newly

developed chromosomal integration vector which utilised a strong C. jejuni promoter.

However, this approach proved to be unsuitable for producing N-glycoprotein immunogen

for two reasons. Firstly, it was difficult to successfully gain C. fetus transformants and

therefore was achieved in wildtype C. fetus but not the N-glycosylation-negative control

strain C. fetus pglB::kn. Secondly, the wildtype C. fetus transformants did not produce

sufficient protein yield as anti-His(6)-reactive bands were barely visible when western

blotting whole cell lysates. In agreement with this, Kienesberger et al. (2007) have shown

that C. fetus require endogenous promoter elements to drive significant gene expression

and thus developed suitable conjugative vectors for this species. Therefore His-tagged

NGRP was integrated into one such vector and successfully conjugated into C. fetus

wildtype and pglB::kn strains. This approach resulted in the successful production of

readily-detectable forms of His-tagged NGRP in C. fetus.

Interestingly, the apparent size of NGRP was larger in C. fetus compared to what was

expected and that observed in the E. coli donor strain. As the conjugated plasmids were

not re-isolated from C. fetus and sequenced it is possible that a mutation in the NGRP

sequence created a larger sized protein. Alternatively, it could have been due to additional

post-translational modification in C. fetus, but this was beyond the scope of this study.

Nevertheless, NGRP appeared to be glycosylated by PglB at several sites, as expected.

Although detectable by western blotting, glycoprotein expression was not as high as

anticipated. Hence, purifying sufficient quantities of glycoprotein immunogen was

109

relatively laborious. NGRP was successfully extracted from C. fetus cell lysates using

Nickel affinity chromatography; however, several additional proteins were evident in the

partially purified preparation. Three proteins at around 42, 44, and 46 kDa in size were

reactive with CjNgp antiserum, indicating they could be NGRP-derivatives. A similar

phenomenon of a CjNgp-reactive band at around 40 kDa has been observed when

expressing NGRP using a pET22(+)b vector in E. coli BL21 (not shown). The closest

characterised homologue of Cj0114, the native protein from which NGRP was

constructed, is E. coli YbgF (21% identity). YbgF contains an N-terminal coiled coil

domain (Krachler et al., 2010). As NGRP includes the N-terminal long α-helical region of

Cj0114, it could be speculated that these observed bands may relate to NGRP forming a

dimeric coiled-coil. Indeed, bioinformatics analysis on NGRP using COILS/PCOILS tool

gives a probability of 1 for a predicted coiled coil at its N-terminus. Therefore, these lower

mobility forms could have been dimers of NGRP glycoforms. However, there was an

abundance of protein at around 46 kDa when coomassie staining in comparison to western

blotting, suggesting that another CjNgp-unreactive protein was present. This was

identified as CFF8240-1557, a C. fetus predicted glutamate dehydrogenase, using mass

spectrometry. This protein is not rich in Histidine residues (1.98%), but a SWISS-MODEL

prediction suggests the protein forms a homo-hexamer with each subunit having 3 His

residues in a surface exposed alpha helix. This may explain why the protein was extracted

using Nickel affinity chromatography. A brief attempt to further purify gNGRP using ion-

exchange chromatography was unsuccessful and, due to time constraints, a partially

purified NGRP preparation was used as antigen for raising rabbit polyclonal antiserum,

termed CfNgp.

Initial characterisation of CfNgp at day 49 found it reacted with numerous N-

glycoproteins in C. fetus 10842, the strain from which the antigen was derived. In C. fetus

pglB::kn CfNgp reacted with a few proteins, in particular a strong reaction was observed

with an ~ 49 kDa protein. This is likely the glutamate dehydrogenase which contaminated

the NGRP purification preparation. Upon completion of the antiserum and when CfNgp

was used at its most appropriate dilution, reactivity against this glutamate dehydrogenase

remained. However, other proteins within C. fetus pglB::kn no longer gave a significant

reaction with the antiserum, suggesting that the majority of reactivity was now against C.

fetus N-glycoproteins. Therefore, CfNgp reacted with numerous C. fetus fetus N-

glycoproteins and also reacted similarly well with N-glycoproteins of the causative agent

of BVC, C. fetus venerealis.

110

The only other reported sera reactive against C. fetus N-glycans are GRPII-1 and GRPII-2

developed by Nothaft et al. (2012). These polyclonal sera were raised against free

oligosaccharide (fOS)-BSA conjugates of each distinct fOS structure extracted from C.

fetus fetus 82-40. GRPII-1 was raised against HexNAc-[Hex]-HexNAc3-diNAcBac-BSA

(type 1) and GRPII-2 against HexNAc-[HexNAc]-HexNAc3-diNAcBac-BSA (type 2)

(Nothaft et al., 2012). In similarity with CfNgp, GRPII-1 reacts equally with cell lysates of

C. fetus fetus and C.fetus venerealis. In contrast, GRPII-2 is more reactive with a C. fetus

fetus cell lysate. Nothaft et al. (2012) reported that type 2 glycopeptides are around 3.5

times more abundant than type 1 glycopeptides in C. fetus fetus 82-40. However, Jervis et

al. (2012) only identified the type 1 glycan structure when analysing C. fetus fetus 10842

N-glycosylation in vitro. It may be that, in contrast, this strain produces more type 1 N-

glycan than type 2 N-glycan. This could explain why the reactivity pattern of CfNgp bears

more similarity to GRPII-1 than GRPII-2, if NGRP was modified with more type 1 than

type 2 N-glycan. Alternatively, as the exact enzymes responsible for creating these

strucures are unknown and the complete genome sequence of this strain is not available, it

is possible that C. fetus 10842 may only produce the type 1 N-glycan.

Furthermore, CfNgp reacted similarly with 18 C. fetus clinical isolates as it did with C.

fetus fetus 10842. These data suggest that C. fetus strains recently infecting humans have

an unchanged N-glycan structure and similar N-glycoproteome size in comparison to type

strains isolated decades ago and since maintained in laboratories. This finding is

encouraging in regard to the potential for using C. fetus N-glycans as detection targets.

The Campylobacter species specificity of CfNgp was investigated and, as expected based

on their identical reported glycan structures (Jervis et al., 2012; Nothaft et al., 2012),

CfNgp reacted with C. hyointestinalis hyointestinalis to a similar extent as with C. fetus.

This is in agreement with the behaviour of GRPII-1, but not GRPII-2 as it reacts

minimally with C. hyo hyo (as with C. f ven) (Nothaft et al., 2012). Further species tested

all produce N-glycan structures distinct from that of C. fetus, however, CfNgp reacted

similarly well with C. concisus, C. lanienae and C. sputorum sputorum also. As this

reactivity was not proven to be due to N-glycan reactivity, it cannot be ruled out that

CfNgp harbours some reactivity against native proteins of these species, possibly due to

homology with C. fetus proteins which contaminated the antigen preparation. However,

the broad extent and pattern of reactivity suggests the majority is N-glycoprotein

reactivity.

111

The observations made here are somewhat similar to that of Nothaft et al. (2012) who

reported that their GRPII sera reacted weakly with C. lanienae and C. sputorum. Notably

GRPII-2 demonstrated considerably less reactivity with C. lanienae and C. sputorum

sputorum than GRPII-1. However, with CfNgp we observed equivalent reactivity with C.

fetus and these species. C. sputorum sputorum produces a heptasaccharide structure which

is similar to the C. fetus type 2 N-glycan except that it has an additional terminal residue

which is a glucose (Jervis et al., 2012; Nothaft et al., 2012). Furthermore, an identical N-

glycan structure was found in C. lanienae. Nothaft et al. (2012) proposed that the

branching GlcNAc of these species is similar configured as the GlcNAc at the

nonreducing end of the C. fetus N-glycan and that this causes the weak reactivity of GRPII

sera with these species. The apparent higher reactivity CfNgp has for these species in

comparison to GRPII sera may be due to antibodies with such reactivity being more

dominant in CfNgp serum. Another explanation may be that CfNgp bears reactivity

against the reducing end of the N-glycan, which is identical in these Campylobacter

species. In addition, these epitopes could encompass the amino acid sequon of N-

glycoproteins, explaining why this observation is more pronounced with CfNgp than

GRPII-1, as GRPII-1 was raised against a fOS-BSA conjugate and not a traditional N-

linked glycoprotein. The considerable cross-reactivity that CfNgp displays against C.

lanienae and C. sputorum could pose an issue if it was used to detect C. fetus in cattle. For

example, C. sputorum is found in cattle but does not cause disease (On et al., 1998).

Therefore this cross-reactivity could cause an inappropriate positive result.

In contrast to the considerable reactivity CfNgp displayed with C. concisus 13826, GRPII

sera do not react with this strain (Nothaft et al., 2012). Interestingly, N-glycan structural

data produced by Nothaft et al. (2012) regarding C. concisus 13826 conflicts findings from

this laboratory (Frost & Linton, unpublished). The striking difference observed between

the C. concisus reactivity of GRPII sera and CfNgp here may be due to differences evident

between the version of the strain used by Nothaft et al. (2012) and what was utilised in this

study. Alternatively, C. concisus 13826 may produce more than one N-glycan structure, a

phenomenon known in C. hominis and C. fetus (Nothaft et al., 2012). However, to result in

this conflicting reactivity between sera, the expression of these N-glycans would need to

vary in some manner. Lastly, the theory discussed above regarding reactivity towards the

reducing end of the N-glycan structure could also explain some of the reactivity of CfNgp

against the C. concisus N-glycan. This is unlikely to be the sole factor because if this was

true, CfNgp would presumably be equally reactive with Campylobacter species such as C.

jejuni as they produce N-glycans which are identical at the reducing end. However, this is

112

not the case (see below). Nevertheless, as C. ureolyticus produces an identical N-glycan to

C. concisus (Nothaft et al., 2012), this suggests that CfNgp may also react with this

emerging Campylobacter species.

Furthermore, CfNgp displayed weak reactivity with Campylobacter species which

produce the C. jejuni N-glycan structure (C. jejuni/coli/upsaliensis/helveticus) and also C.

lari. In contrast, the GRP-II sera harbour no reactivity for these species (Nothaft et al.,

2012). Again, this could be due to CfNgp being reactive with the identical reducing end

portion of these N-glycan structures. Further elucidation of the specific nature of this

reactivity was achieved by analysing C. jejuni pgl mutants. This experiment demonstrated

that CfNgp does not react with the branching glucose residue; in fact its absence appeared

to increase reactivity with some proteins. Instead, CfNgp reactivity with C. jejuni

glycoproteins depended on the presence of the four terminal GalNAc residues. At least

two GalNAc residues appeared to be required for reactivity with C. jejuni glycoproteins to

be evident. These data appear to conflict the theory that CfNgp could harbour reactivity

against residues at the reducing end of the N-glycan. However, Kelly et al. (2006) reported

that no N-glycan is detectable via NMR and lectin blotting in C. jejuni pglHJA knockout

mutants. This suggests that the reduction in reactivity seen with these mutants could be

influenced by N-glycoprotein depletion and does not prove that the serum is unreactive

with the reducing end. Together, these data suggest that CfNgp contains antibodies

reactive against several N-glycan structural epitopes which in turn confers broad reactivity

against Campylobacter species.

The dot method was used as a preliminary investigation into the ability of CfNgp to bind

untreated cells of Campylobacter. This assay demonstrated that CfNgp was able to detect

as few as 103 cells of C. fetus fetus and 10

4 of C. fetus ven. Previously Nothaft et al. (2012)

reported GRPII sera were able to label C. fetus cells using fluorescence microscopy.

However, no difference in the ability to label each C. fetus subspecies was obvious in their

study. The variation in the limit of detection between each subspecies seen here may be

apparent due to a dot blot assay allowing for more sensitive assessment of cell binding

ability. The variation itself may be explained by differing amounts of surface accessible

N-glycans between each subspecies. In addition, some C. fetus fetus-specific antibodies

may exist within CfNgp due to C. fetus fetus proteins contaminating the antigen

preparation. Although minimal reactivity is seen with a C. fetus pglB::kn cell lysate,

components of the cell which are not available when probing western blots cannot be

accounted for (e.g. LLO/insoluble proteins) and may be exposed during dot blotting.

Furthermore, surface accessible proteins that are mildly reactive during western blotting

113

conditions may impact cell surface binding more significantly than expected due to correct

folding providing additional conformational epitopes.

Although Nothaft et al. (2012) reported cell labelling of C. fetus with GRPII sera, the

authors did not verify that this result was solely due to N-glycoprotein reactivity. Here,

CfNgp was able to detect cells of C. fetus pglB::kn to the same extent as the wildtype

strain. However, if some C. fetus protein-reactive antibodies were adsorbed from the

serum its cell detection limit increased to 104

cells and the displayed reactivity was of a

higher intensity with the wildtype strain. This suggests that protein reactivity within the

antiserum is responsible for some portion of the cell binding observed. However, this

study did not ensure that all non-glycan reactivity within the antiserum was removed so it

does not elucidate to what extent the influence is. Nevertheless, as removing some of this

reactivity allowed for a difference in the ability to bind the wildtype and pglB::kn strain to

be visible, it proves that the N-glycan reactivity does contribute to the ability to bind

untreated cells. Notably, C. fetus pglB::kn contains N-linked glycan precursor, LLO,

which could result in N-glycan-reactive antibodies binding to this strain. LLO may have

been exposed if cell surface damage occurred during preparation and/or cells did not

remain intact upon membrane drying prior to the dot blot assay. Therefore, anti-N-glycan

antibodies within CfNgp may contribute to cell binding to a higher extent than is

suggested by the results reported here. Hence, further investigation into the ability of N-

glycan reactive antibodies to bind to C. fetus cells is warranted. A C. fetus pglA::kn mutant

would be a more appropriate control for investigating this, however, such a mutant is

currently unavailable.

CfNgp was also able to bind cells of C. jejuni-type N-glycan producing species and C.

lari, albeit at a lower intensity, in a dot blot assay. This is in contrast to the GRPII antisera

produced by Nothaft et al. (2012) which were unable to label C. jejuni cells under

fluorescence microscopy. Therefore it seems that the cross reactivity CfNgp displayed

with cell lysates of these Campylobacter species still persists when probing untreated

cells. These results suggest that if CfNgp were integrated into a detection test, it could

potentially give positive results when faced with the so-called ‘thermophilic’

Campylobacter species such as C. jejuni and C. lari. Although there have been cases of C.

jejuni causing abortion in cattle and sheep (Larson et al., 1992; Mannering et al., 2006),

these species do not usually reside in the reproductive tracts of ruminants. However, they

are often found in the intestinal tracts of healthy animals (Stanley and Jones, 2003) and

this could pose a slight cross-contamination risk.

114

In conclusion, a recombinant N-glycoprotein produced in C. fetus fetus was successfully

used to raise an antiserum, CfNgp, reactive with N-linked glycoproteins produced by a

variety of C. fetus fetus strains and C. fetus venerealis. In contrast to previously reported

sera reactive against C. fetus N-linked glycan, CfNgp also reacts similarly with the N-

linked glycans of other Campylobacter species, such as C. concisus. This antiserum can

also bind to untreated cells of many Campylobacter species. Although protein-reactive

antibodies contributed to this ability to bind cells, adsorbing some of this reactivity

demonstrated that N-glycan-reactive antibodies were also able to bind C. fetus cells.

Therefore, these data suggest that CfNgp could be used to develop a rapid test which can

detect several Campylobacter species, with particular emphasis on C. fetus and rarer,

‘emerging’ pathogenic Campylobacter species such as C. concisus and possibly C.

ureolyticus.

115

Chapter 5. Towards engineering the

C. fetus N-linked glycan using

glycocompetent E. coli containing a

hybrid C. jejuni-C. fetus system

116

5.1 Introduction

Chapter 4 described the development of an antiserum reactive against N-glycans of several

Campylobacter species including C. fetus and C. concisus. Such an antiserum could be

used in an immunoassay for detecting a variety of Campylobacter species. However,

production and purification of the N-linked glycoprotein immunogen from C. fetus was

time consuming and laborious with a low yield. This method would not be suitable for

larger scale production of C. fetus glycoprotein if further antisera were made. In Chapter 3,

to obtain immunogen for raising antiserum CjNgp, high-yield production of C. jejuni N-

linked glycoprotein was achieved with relative ease using the glycocompetent E. coli

system developed by Wacker et al. (2002). The development of an analogous C. fetus N-

linked glycosylation system functioning in E. coli would similarly be useful for larger

scale production of C. fetus N-glycoprotein immunogen for production of CfNgp antisera.

It would also be useful for further research into the C. fetus N-linked glycosylation process

itself.

5.1.1 Comparison of the C. fetus and C. jejuni pgl loci

To reconstitute the C. fetus N-glycosylation system in E. coli, the appropriate genes need

identifying. Although the C. fetus pgl locus has not been fully characterised, the function

of some genes are inferred from homology with the C. jejuni pgl locus (Fig. 5.1A).

Similarities between the C. jejuni and C. fetus pgl loci reflect the similarities in their N-

glycan structures, with identical sugar residues (DiNAcBacGalNAc2) at the reducing end

(Fig, 5.1B) (Jervis et al., 2012; Nothaft et al., 2012). Both species contain pglK and pglB,

the former encoding the flippase mediating the transit of the LLO precursor across the

inner membrane (Kelly et al., 2006) and the latter the oligosaccharyltransferase which

transfers the assembled N-glycan onto target proteins (Wacker et al., 2002). The C. fetus

pgl locus contains homologues of C. jejuni genes encoding sugar biosynthesis enzymes

responsible for production of UDP-DiNAcBac (pglDEF) (Olivier et al., 2006) and UDP-

GalNAc (gne also known as galE) (Bernatchez et al., 2005), sugar-nucleotide donors used

in the Pgl systems of both species. C. fetus also contains orthologues of C. jejuni genes

encoding the glycosyltransferases, pglCAJ, which mediate the transfer of DiNAcBac and

the two subsequent GalNAc residues (Fig, 5.1B) (Glover et al., 2006, 2005; Linton et al.,

2005) found in each N-glycan structure.

117

The non-reducing end of the C. jejuni and C. fetus N-glycans vary in structure and hence

their pgl loci begin to differ with regards to the genes responsible for constructing this

portion. C. jejuni pglH encodes the glycosyltransferase which acts after PglJ, adding a

further three GalNAc residues sequentially (Troutman and Imperiali, 2009). However, the

C. fetus N-glycan does not contain these additional GalNAc residues, but rather two

GlcNAc residues (Nothaft et al., 2012). The C. fetus pgl locus harbours two homologues

of C. jejuni pglH, namely cf1385 and cf1386, and referred to as pglH2 and pglH1,

respectively. The products of pglH2 and pglH1 share 38% and 39% identity, respectively,

with C. jejuni PglH and 48% identity with one another. As with C. jejuni pglH, C. fetus

pglH1 and pglH2 are members of Glycosyltransferase Family 4 of the CAZy

(carbohydrate active enzymes) database. PglH1, PglH2 and C. jejuni PglH contain a GT1

amsD-like domain (with a putative nucleotide diphosphate-binding pocket) and a

Glycosyltransferase group 1 domain, placing them within the Glycosyltransferase GTB-

type superfamily. It is likely that one or both of PglH1H2 are responsible for transfer of at

least two GlcNAc residues onto the DiNAcBacGalNAc2-UndP precursor.

In C. jejuni, the final sugar transferred to the N-glycan precursor is the branching glucose

residue added by PglI (Glover et al., 2005; Linton et al., 2005). Although C. fetus

produces N-glycan with a branching glucose residue, it does not have a pglI homologue.

Furthermore, there is no obvious candidate within the C. fetus pgl locus for the addition of

the alternative branching residue, GlcNAc. However, it is possible that either C. fetus

PglH1 or PglH2 could be responsible for transfer of this residue. Therefore, based on

homology with C. jejuni pgl genes alone, there are no genes predicted to encode the

glycosyltransferases responsible for adding these branching sugars (Glc or GlcNAc).

118

5.2 Strategy to investigate the role of C. fetus glycosyltransferases

in N-glycan assembly by integrating them into an E. coli system.

5.2.1 An uncharacterised predicted glycosyltransferase gene lies within C. fetus pgl

locus

Three uncharacterised genes, cf1389-cf1391, are situated between pglK and galE within

the C. fetus 82-40 pgl locus. Of these, cf1389 is annotated as a predicted

glycosyltransferase matching to Pfam protein family PF00535, a diverse family which

transfer sugar from UDP-Glc, UDP-GalNAc, GDP-Mannose or CDP-abequose to a

variety of substrates. The gene product is, like C. jejuni PglI, characterised as a

Glycosyltransferase 2 family protein within the CAZy (carbohydrate active enzymes)

database. C. fetus cf1389 is not a homologue of C. jejuni pglI, and alignment using

BLASTP gives 31% identity over only 17% of PglI. However, in C. fetus the glucose

residue is linked to a GlcNAc, not a GalNAc as in C. jejuni, and hence a considerably

different enzyme may be responsible for its addition. Therefore, it is possible that cf1389

encodes a glycosyltransferase responsible for adding either of the branching sugars (Glc or

GlcNAc) to the C. fetus N-linked glycan.

119

120

5.2.2 Model of a hybrid C. jejuni/C. fetus glycosylation machinery.

As discussed, the first three residues of the C. fetus N-glycan (DiNAcBacGalNAc2) are

identical to those in C. jejuni. Therefore, using a truncated N-glycan consisting of these

residues could provide a base onto which the remaining residues of C. fetus N-glycan

could be transferred. Such a system would provide an alternative to cloning the entire pgl

locus of C. fetus into E. coli to create C. fetus N-linked glycoprotein. Furthermore,

glycosyltransferases which build the remainder of the C. fetus N-glycan could be

investigated by introducing gene candidates to the system individually and in various

combinations. Such a truncated trisaccharide can be produced in E. coli using

ppglpglH::kn (Linton et al., 2005). C. jejuni PglB does not have strict substrate specificity,

as it can transfer truncated C. jejuni N-glycan structures (Linton et al., 2005) and even

diverse O-antigen structures produced by E. coli and Pseudomonas aeruginosa onto

protein (Feldman et al., 2005). Hence it would likely be able to transfer the C. fetus N-

glycan to a target protein. Therefore, a plan to engineer the full C. fetus N-glycan in E. coli

by introducing C. fetus predicted transferases to the available C. jejuni pglH::kn system

was devised (Fig. 5.2).

121

122

5.3 Identification of initial C. fetus glycosyltransferase to act on C.

jejuni pglH::kn trisaccharide.

5.3.1 Cloning strategy

The strategy devised included using the available ppglpglH::kn plasmid which, when

expressed in E. coli BL21 with a target protein, results in modification with a truncated

trisaccharide version of the C. jejuni N-linked glycan structure (Linton et al., 2005).

Previously this laboratory successfully used a C. jejuni glycoprotein, Cj0114, as a

glycosylation reporter protein in H. pullorum (Jervis et al., 2010). Since then, a truncated

version, termed NGRP (N-glycosylation reporter protein) (See 3.1.1), was constructed

(Jervis & Linton, unpublished). His-tagged NGRP was introduced using pETNGRP (see

Appendix Table 2) and used as a target protein for N-glycosylation.

To avoid requiring a third plasmid, each C. fetus glycosyltransferase gene candidate was

introduced onto the backbone of pETNGRP. A BglII restriction site upstream of the T7

promoter (Fig. 5.3A) was used for insertion of each gene of interest to avoid the predicted

glycosyltransferases being over-expressed upon IPTG induction. The ability to combine

BglII-digested DNA with compatible ends created by BamHI and BclI was also utilised.

The open reading frame of each gene and 80-100 bp upstream of the start codon was

cloned to incorporate the ribosomal start site and any potential native promoter elements.

Notably, the complete genome sequence of the C. fetus strain used here, NCTC 10842,

was not available hence the genome of C. fetus 82-40 was used as a template. To identify

if any of the C. fetus predicted glycosyltransferases could transfer sugar onto the C. jejuni

pglH::kn N-glycan, each gene was individually introduced into pETNGRP. To provide a

positive control, C. jejuni PglH activity was complemented in this system by re-

introducing a functional copy of pglH at the BglII site of pETNGRP. Construction of these

vectors is described in the following sections.

5.3.1.1 Cloning pglH1 into pETNGRP

pglH1 was amplified from C. fetus fetus 10842 genomic DNA (gDNA) using primers

1378 and 1386 (Table 2.1), introducing a 5’ BamHI and a 3’ BglII site. Agarose gel

electrophoresis confirmed production of the predicted 1.2 kb product (Fig. 5.3B). This was

digested with BamHI and BglII and ligated to BglII-digested pETNGRP. Following

transformation, ligation of pglH1 was demonstrated by colony PCR (cPCR) using primers

82 and 1378 (indicated in Fig. 5.3C) to amplify pglH1 along with ngrp (results not

123

shown). Purified plasmids from positive clones were subjected to further PCR to validate

the size of pglH1 using primers 1388 and 1411 (Fig. 5.3C). PCR products around 1.2 kb in

size were produced from each plasmid, as expected (Fig. 5.3D). Plasmids were sequenced

using primers 1411 and 1425. This revealed two basepair mismatches, -51G>T and

(769C>T) in each of the three plasmids. The former did not appear to interrupt any

obvious promoter elements and the latter was a silent mutation. As these mismatches were

identical in all three clones and did not appear likely to interrupt gene expression or

function they were assumed to be single nucleotide polymorphisms (SNPs) in the C. fetus

strain being used. Furthermore, the sequence of NGRP was confirmed as correct via

sequencing using a universal T7 promoter primer (Table 2.1, number 80). This plasmid

was designated pNH1.

124

125

5.3.1.2 Cloning pglH2 into pETNGRP

pglH2 was amplified from C. fetus fetus 10842 gDNA using primer pair 1379/1399 (Table

2.1), which introduced a BclI site at each end (Fig. 5.4A). This product was digested with

BclI and ligated to BglII-digested pETNGRP. Following transformation, colonies were

screened by PCR using primers 1389 and 1411 (Fig. 5.4B). Colonies 5 and 8 gave PCR

products around 1.2 kb in size, as expected (Fig. 5.4C). Plasmids extracted from these

clones were sequenced with primers 1411 and 1425. Each plasmid had a 655A>C

mutation within pglH2, translating to a K219Q substitution in the protein product.

Repeating the cloning process created another plasmid (clone 15) with this mutation.

Hence it was presumed a C. fetus fetus 10842 SNP. A further PCR check was included

using primers 82 and 1411, in order to also cover the NGRP region of each plasmid (Fig.

5.4B). A predicted product of 1.97 kb was expected from this reaction. Although each

clone produced a PCR product ~ 2 kb in size, clones 5 and 8 each produced an additional

unexpected product, either larger or smaller in size, respectively (Fig. 5.4D). Therefore,

plasmid extracted from clone 15 was chosen and termed pNH2.

126

127

5.3.1.3 Cloning Cf1389 into pETNGRP

Cf1389 was amplified from C. fetus fetus 10842 gDNA using primers 1387 and 1406

(Table 2.1), adding a BamHI site to each end of the PCR product. An ~ 1.1 kb PCR

product, as predicted, was observed (Fig. 5.5A). Following BamHI digestion, the product

was ligated with BglII-digested pETNGRP. Following transformation, positive clones

were identified by PCR using primers 1390 and 1411 (Fig. 5.5B). Clones 1 and 3 gave

PCR products at ~ 1.2 kb (Fig. 5.5C). Plasmids were purified and sequenced using primers

1411 and 1425. One base mismatch, 783A>G, was found in each clone. However, this

mutation was silent as the codon continued to encode a glycine residue (GGA>GGG). In

addition, the NGRP sequence within each plasmid was confirmed by sequencing using

primer 80. The plasmid, termed pN1389, purified from clone 1 was used in subsequent

work.

128

129

5.3.1.4 Cloning CjpglH into pETNGRP

To provide a positive control, the C. jejuni glycosyltransferase gene, pglH, was similarly

cloned into pETNGRP. This is predicted to genetically complement the pglH insertional

knockout and restore heptasaccharide N-glycan production. pglH was amplified from C.

jejuni 11168H gDNA using primer pair 1397/1398 (Table 2.1), which add a BamHI site to

each end of the PCR product. Production of an ~1.2 kb PCR product was demonstrated via

agarose gel electrophoresis (predicted product 1.24 kb)(Fig. 5.6A) and following BamHI

digestion, this product was ligated with BglII-digested pETNGRP. Transformatants were

were identified by PCR using primers 1411 and 1424 (Fig. 5.6B). PCR products of ~ 1.3

kb were confirmed from clones 3 and 7 (Fig. 5.6C). Plasmid from clone 3 was sequenced

using primers 1411 and 1425 with no mismatches found. Furthermore the plasmid was

sequenced using primer 80 to check the sequence of NGRP and this plasmid was termed

pNCjH.

130

131

5.3.2 Only CfpglH1 decreased the mobility of NGRP glycoforms in the

ppglpglH::kn background.

To investigate if C. fetus glycosyltransferases are able to transfer sugar (s) to the C. jejuni

pglH::kn truncated N-glycan structure, pNH1, pNH2 and pN1389 were individually

transformed into competent cells of E. coli BL21 (λDE3) ppglpglH::kn. For reference,

NGRP glycosylated with the full C. jejuni heptasaccharide N-linked glycan was produced

by transforming pETNGRP and ppgl into E. coli BL21 (λDE3). To create the truncated N-

glycan structure, pETNGRP was introduced into E. coli BL21 (λDE3) ppglpglH::kn. As a

positive control, pETNCjH was introduced into E. coli BL21 (λDE3) ppglpglH::kn to

complement PglH activity and reconstitute the C. jejuni heptasaccharide N-glycan

structure. Presence of the appropriate plasmids was confirmed by PCR (not shown).

NGRP production was induced by addition of 1 mM IPTG and following incubation at

25°C overnight whole cell lysates were prepared. A western blot of whole cell lysates was

probed with anti-His6 antibodies (Fig. 5.7). E. coli ppgl pETNGRP produced three

detectable forms of His-tagged NGRP. The major forms migrated at around 23 and 26

kDa, corresponding to unglycosylated NGRP (21.6 kDa predicted size) and NGRP with a

single glycosylation site modified (referred to as g1, labelled with asterisk), respectively.

The least dominant form migrated at 25 kDa and likely corresponds to unmodified NGRP

with it its signal peptide present. Upon inactivation of the C. jejuni pglH gene on ppgl, the

g1 form now migrated at around 24 kDa. This g1 form represents modification with a

truncated trisaccharide N-glycan, as C. jejuni PglH catalyses transfer of the fourth, fifth

and sixth sugar residues of the C. jejuni heptasaccharide (Troutman and Imperiali, 2009).

Introduction of C. fetus pglH2 or cf1389 along with ppglpglH::kn did not alter the

migration of g1 NGRP forms. However, adding C. fetus pglH1 into the system caused the

electrophoretic mobility of the g1 form to decrease, increasing in apparent size by ~ 1

kDa. Another His-tagged glycoform of NGRP was visible at around 27 kDa, which

correlates with two N-glycosylation site being modified (g2). Finally, upon re-

introduction of C. jejuni pglH into the system (on pNCjH), the g1 form returned to its

wildtype apparent size of around 26 kDa. Furthermore, the g2 form was also visible in this

strain.

In summary, introducing the C. fetus predicted glycosyltransferase gene, pglH1, decreased

the electrophoretic mobility of NGRP glycoforms produced by E. coli containing

ppglpglH::kn. Such a shift is consistent with addition of glycan residue (s). In contrast,

addition of genes pglH2 and cf1389 did not influence the apparent electrophoretic mobility

132

of NGRP glycoforms in this system. These data suggest that the predicted

glycosyltransferase, PglH1, is able to utilise the truncated C. jejuni pglH::kn N-glycan

structure as an acceptor substrate. This would indicate that PglH1 is a glycosyltransferase

responsible for addition of the fourth sugar residue of the C. fetus hexasaccharide

structure.

133

5.4 Identification of second C. fetus glycosyltransferase to act on C.

jejuni pglH::kn trisaccharide.

5.4.1 Cloning strategy

The reduced electrophoretic mobility of NGRP glycoforms due to C. fetus pglH1 allowed

investigation into the identity of the glycosyltransferase able to transfer sugar(s) to this

new N-glycan structure. Thus, plasmids containing pglH1 and also one of the two

glycosyltransferase genes, pglH2 and cf1389, were constructed.

5.4.1.1 Construction of pNH1H2

As described in 5.2.3, C. fetus pglH1 and pglH2 are overlapping genes. Therefore, they

were amplified together with primers 1378 and 1379 (Table 2.1), adding a 5’ BamHI and a

3’ BclI site. An ~ 2.2 kb PCR product was confirmed via agarose gel electrophoresis (Fig.

5.8A), digested with BamHI and BclI, ligated with BglII-digested pETNGRP and the

ligation electroporated into E. coli XL-1 Blue. Transformants were identified by

production of an ~ 2.3 kb cPCR product from primers 1411 and 1389 (not shown).

Plasmid was extracted from clone 11 and sequenced using primers 1411, 1425 and 1450.

Sequences were identical to those found previously when individually cloning pglH1 and

pglH2 (See 5.3.1), confirming that these are strain-specific SNPs. The plasmid was termed

pNH1H2 (Fig. 5.8B).

5.4.1.2 Construction of pNH1-1389

To create pNH1-1389, cf1389 was cloned into pNH1 at the BglII site downstream of

pglH1. Primers 1439 and 1406 (Table 2.1) were used to amplify cf1389 with a 5’ BclI and

a 3’ BamHI site. An ~ 1.1 kb PCR product was confirmed by gel electrophoresis (1.13 kb

predicted) (Fig. 5.8C) and, following digestion, was ligated into BglII-digested pNH1 and

the ligation electroporated into E. coli XL-1 Blue. Transformants were identified by the

production of an ~ 2.3 kb cPCR product from primers 1411 and 1390 (not shown).

Plasmid DNA was extracted from clones 15 and 28 and sequenced using primers 1425 and

1485. The basepair mismatch (783A>G), a silent mutation also found when cloning

cf1389 into pETNGRP, was detected, confirming a strain-specific SNP. The plasmid was

termed pNH1-1389 (Fig. 5.8D).

134

135

5.4.2 C. fetus pglH2 decreased NGRP glycoform mobility in the ppglpglH::kn

CfpglH1 background.

To investigate if either of the C. fetus predicted glycosyltransferase genes, pglH2 or

cf1389, could transfer sugar (s) to the N-glycan created by E. coli BL21 (λDE3)

ppglpglH::kn pNH1, the plasmids described in 5.4.1 were each transformed into E. coli

BL21 (λDE3) ppglpglH::kn. Clones positive for pNH1H2 and pNH1-1389 were identified

by PCR using primer pairs 1411/1450 and 1411/1390, respectively. NGRP production was

induced in each strain as previously (section 5.3.2.), and a western blot of whole cell

lysates of these induced strains along with induced E. coli BL21 (λDE3) ppglpglH::kn

containing pNH1 or pNCjH was probed with anti-His6 antibodies (Fig. 5.9).

As previously, all strains produced unglycosylated NGRP around 23 kDa (Fig. 5.9) and E.

coli BL21 (λDE3) ppglpglH::kn pNH1 produced g1 and g2 forms at ~ 25 kDa and ~27

kDa, respectively. Notably, g1 and g2 forms with pNH1H2 had reduced electrophoretic

mobility, whilst glycosylated forms produced with pNH1-1389 did not. With pNCjH,

bands corresponding to g1 and g2 forms modified with full size C. jejuni N-glycan were

evident ( ~26.5 kDa and ~ 29 kDa). Therefore, adding cf1389 into E. coli ppglpglH::kn

pNH1 system had no detectable effect on NGRP glycoforms, but pglH2 caused a

reduction in electrophoretic mobility consistent with addition of sugar residue (s). These

data suggest that the pglH2-encoded predicted glycosyltransferase is able to transfer sugar

residue (s) to the N-glycan structure following the action of PglH1. Therefore, this would

indicate that PglH2 is a glycosyltransferase which transfers the fifth sugar residue onto the

C. fetus N-glycan structure.

136

137

5.5 Structural analysis of N-linked glycoproteins produced from

hybrid systems.

5.5.1 Purification of NGRP from C. jejuni/C. fetus hybrid N-linked glycosylation

systems in E. coli

His-tagged NGRP was purified from E. coli BL21 (λDE3) containing ppglpglH::kn and

pETNGRP, pNH1 or pNH1H2 using nickel affinity chromatography. Cells were induced

with 1 mM IPTG overnight at 16 °C, sonicated and applied to a 5 ml HisTrap FF column.

Cell lysate, lysate flow-through and eluate were separated by SDS-PAGE and stained with

coomassie (Fig. 5.10). The unglycosylated and g1 forms of NGRP were apparent in the

induced cell lysates of each strain (Fig. 5.10, indicated by asterisks), and absent from the

column flow-through. Consistent with western blotting in 5.3.2 and 5.4.2, the mobility of

the g1 form decreased upon addition of C. fetus pglH1 then pglH2 into the system (pNH1

then pNH1H2). The unglycosylated and g1 forms of NGRP purified from E. coli BL21

(λDE3) ppglpglH::kn pETNGRP were not distinct bands due to their similar mobility and

the high concentration of each protein in the eluate (Fig. 5.10, left). Glycosylated forms

were apparent in eluate from E. coli containing pNH1 and pNH1H2, with g1 and g2 forms

evident at around 25 and 27 kDa with pNH1 and around 25.5 and 27.5 kDa with pNH1H2

(Fig. 5.10, pNH1 – middle, pNH1H2 – right). Therefore, NGRP was purified from the

hybrid systems created in 5.3 and the observation of glycoforms with reduced

electrophoretic mobility was further evidenced. These preparations were analysed by mass

spectrometry to structurally characterise the attached N-glycans.

138

139

5.5.2. MALDI-TOF MS analysis of purified NGRP glycoforms

To structurally characterise the N-glycans produced by the hybrid C. jejuni/C. fetus pgl

systems, the dialysed eluates described in 5.5.1 were separated by SDS-PAGE and bands

were excised, digested with trypsin and subjected to MALDI-TOF MS using α-cyano-4-

hydroxycinnamic acid (CHCA) matrix. The g2 forms of NGRP purified from E. coli BL21

(λDE3) containing ppglpglH::kn and pNH1 or pNH1H2 were excised for analysed. As

only the g1 form of NGRP purified from E. coli BL21 (λDE3) ppglpglH::kn pETNGRP

was available, and was not fully distinguishable from the unglycosylated band, the upper

limit of this band was excised for analysis.

Each sample gave peaks corresponding to at least seven of the ten predicted tryptic

peptides of NGRP (Table 5.1 peptides 2-4, 6-9), confirming the identity of this protein. A

further peak corresponds to a predicted tryptic peptide with m/z 1885.9662 arising due to

one missed cleavage site (residues 21-36).

140

Three predicted tryptic peptides from NGRP (Table 5.1 peptides 1, 2 and 7) contain N-

glycosylation sequons, and the predicted m/z values of these peptides with possible N-

glycan structures attached are displayed in Table 5.2. The largest sequon-containing

peptide of predicted m/z 4059.8411 (Table 5.2, peptide 1) contains two sequons and, as

glycan modification can significantly reduce ionisation efficiency (Song et al., 2013),

detecting such a large glycopeptide seemed the least likely. As expected, no peaks

corresponded to this sequon-containing peptide modified with any of the predicted N-

glycan structures in the spectra. The two remaining potential glycopeptides (Table 5.2,

peptides 2 and 7) each have a single sequon and smaller predicted m/z values of 2478.1743

and 1216.6168, and hence were more likely to have higher detectability. Unfortunately, no

species with such mass to charge ratios were found in any NGRP preparation. Instead,

peaks with m/z values corresponding to these peptides without glycan were evident (Fig.

5.11, indicated with arrows). One explanation for these findings could be that the two

sequons within the largest tryptic peptide (4059.8411) were preferentially modified,

creating large glycopeptides which could not be detected (smallest predicted m/z;

5329.1587 – modified with truncated N-glycan produced by ppglpglH::kn). Therefore,

MALDI MS analysis of NGRP, with its four N-glycan sites, did not yield structural N-

glycan data. In light of this issue, further MALDI-TOF MS analysis was undertaken using

a modified form of NGRP.

141

142

143

5.5.3 Production and purification of modified NGRP in hybrid systems.

Due to difficulties in observing glycopeptide ions when analysing trypsin-digested NGRP

using MALDI-TOF MS, a different protein was integrated into the hybrid C. jejuni/C.

fetus pgl systems to allow N-glycan structural analysis. A modified version of NGRP,

termed NGRP3, is a glycosylation reporter protein created to be more amenable for mass

spectrometry (MS) analysis (Jervis & Linton, unpublished). To achieve this, three of the

four N-glycan sequons were inactivated by site directed mutagenesis and the single

remaining sequon is located in the flexible linker region that appears to be more readily

modified (Jervis et al. 2010 and section 5.5.2). In addition, residues surrounding this

remaining sequon were altered to create a smaller tryptic peptide which has a higher

positive charge, to enhance its ionisation efficiency during mass spectrometry (Fig. 5.12).

To replace NGRP in the hybrid systems developed in 5.3 and 5.4, the original NGRP

sequence was first digested out of pNH1/pNH1H2 using XhoI and NcoI. NGRP3 was

amplified from pNGRP3 using primers 790 and 1184 (Table 2.1), adding a 5’ NcoI and 3’

XhoI. Following digestion with XhoI and NcoI, the product was ligated with digested

pNH1 and pNH1H2 to give pH1-N3 and pH1H2-N3, respectively. Following

transformation, the presence of NGRP3 was identified by PCR using primers 790 and

1184. Extracted plasmids were sequenced using primer 82 to check the NGRP3 sequence.

Primers 1411 and 1425 were used to sequence the pglH1 region of pH1-N3. The pglH1-

pglH2 region of pH1H2-N3 was sequenced using primers 1388, 1389, 1411, 1425 and

1559.

144

Plasmids pNGRP3, pH1-N3 and pH1H2-N3 were transformed into E. coli BL21 (λDE3)

ppglpglH::kn. Positive clones were identified by PCR using primers 790 and 1184 for

pNGRP3 and primers 1388 and 1411 for pH1-N3 or pH1H2-N3. To produce

unglycosylated protein, pNGRP3 was also transformed into E. coli BL21 (λDE3). NGRP3

production was induced in each strain using 1 mM IPTG at 16°C overnight. Whole cell

lysates were separated by SDS-PAGE and coomassie-stained, with unglycosylated

NGRP3 present as an ~ 21 kDa protein (predicted 21.78 kDa) (Fig. 5.13A). A g1 form of

NGRP3 was evident in each strain carrying C.fetus pgl genes, at around 23 and 23.5 kDa

with pglH1 and pglH1H2, respectively.

NGRP3 was purified from lysates of each IPTG-induced strain using Nickel affinity

chromatography. Eluates were dialysed against 30 mM Tris 0.15 M NaCl pH 7.5. Bands

corresponding to unglycosylated NGRP at ~ 20 kDa were evident in each eluate when

diluted (1/10) and separated by SDS-PAGE (Fig. 5.13B). In the presence of ppglpglH::kn,

a g1 form of NGRP at around 21 kDa was also purified. Upon addition of pH1-N3 or

pH1H3-N3 the g1 forms purified were around 22 kDa or 22.5 kDa, respectively.

Therefore, forms of NGRP3 glycosylated with varying N-glycan structures similar to

those observed with the hybrid C. jejuni/C. fetus pgl systems in sections 5.3 and 5.4 were

purified. The structures of these N-glycans were analysed by mass spectrometry.

145

146

5.5.4 MS analysis of purified NGRP3.

5.5.4.1 The sequon-containing tryptic peptide of unglycosylated NGRP3 was detected by

MALDI TOF Mass spectrometry.

NGRP3-containing eluates (as described in 5.5.3) were separated by SDS-PAGE, protein

bands excised and tryptic peptides prepared for MALDI-TOF MS analysis. To confirm

that the sequon-containing tryptic peptide of NGRP3 (Table 5.3 peptide 8) could be

detected when unmodified, unglycosylated NGRP3 produced in E. coli BL21

ppglpglH::kn pH1-N3 was excised and analysed. Two matrices, CHCA and 2, 5-

dihydroxybenzoic acid (DHB), were used to assess which was most appropriate for

detection of this peptide. The spectra using each matrix are compared in Figure 5.14.

Better coverage of NGRP3 was achieved using DHB (Fig. 5.14B), with 7 peaks matching

possible tryptic peptides (Table 5.3, peptides 1-4, 6, 8 and 9) compared to 3 matches with

CHCA (peptides 2,4 and 8). A further three peaks (m/z 1885, 2718 and 3217) were

identified in each spectrum that correspond to masses of possible tryptic peptides arising

due to a missed cleavage site. The sequon-containing tryptic peptide (predicted m/z

1167.55) was identified at m/z 1167.32 and 1167.38 using CHCA and DHB, respectively.

However, the peak identified using DHB was of very low intensity (Fig 5.14B, indicated

by grey arrow). Therefore, the sequon-containing tryptic peptide from NGRP3 was readily

detectable by MALDI-TOF MS when using CHCA as the sample matrix, but did not give

a significant peak with DHB matrix.

147

148

149

5.5.4.2 MALDI-TOF MS analysis of glycosylated NGRP3 using CHCA matrix.

As CHCA allowed improved identification of the sequon-containing tryptic peptide from

unglycosylated NGRP in comparison to DHB, this matrix was chosen for analysing

NGRP3 glycoforms using MALDI-TOF mass spectrometry. NGRP3-containing eluates

were separated by SDS-PAGE and coomassie stained (as described in 5.5.3), and the g1

forms of NGRP3 excised for mass spectrometry analysis.

At least 7 peaks corresponded to predicted NGRP3 tryptic peptides in each sample

analysed (Fig. 5.15, Peptides 1-4, 6, 9 and 10 in Table 5.3), confirming the identity of

NGRP3. A peak with a m/z value corresponding to tryptic peptide 7 (Table 5.3) was

produced by NGRP3 from pH1H2-N3. In addition, peaks corresponding to the masses of

tryptic peptides arising from missed cleavages were identified in all samples (m/z 1886

from all samples and m/z 2134 from pN3 and pH1-N3). Peaks corresponding to the

sequon-containing tryptic peptide, HHTDSNSTIR (Table 5.3 peptide 8), were not

observed.

The m/z values of remaining peaks were cross referenced against the predicted m/z values

of tryptic peptide 8 (HHTDSNSTIR) modified with the expected N-glycan structures

(Table 5.4). No peaks from pN3 or pN3-H1 had m/z values similar to any of the predicted

glycopeptides. Therefore, no structural data for the attached N-glycans were obtained

using MALDI-TOF MS with CHCA matrix.

150

151

5.5.4.3 MALDI-TOF MS analysis of glycosylated NGRP3 using DHB matrix.

CHCA matrix allowed identification of the sequon-containing tryptic peptide from

unglycosylated NGRP3 (Section 5.5.4.1) but did not when this peptide was N-

glycosylated (Section 5.5.4.2). Although with DHB the peak intensity for this unmodified

peptide was low, this matrix is commonly used for glycan analysis (Morelle et al., 2006)

and Mills et al. (2016) successfully identified glycopeptides from a similar NGRP variant

using this matrix. Hence, the glycoforms of NGRP3 analysed above were reanalysed using

DHB matrix (Fig. 5.16). The mass range of the detector was selected between m/z 1700

and 2500.

In the m/z range tested, only unmodified tryptic peptides 1-5 were detectable. Peaks

corresponding to peptides 1-4 were each found from pH1-N3 and pH1H2-N3 and those

corresponding to 1,3 and 4 from pN3 (Table 5.3). Additionally, each NGRP glycoform

gave two peaks corresponding to tryptic peptides with one miscleavage (m/z 1885 and

2134). Of the remaining peaks, one at m/z 2207.906 evident in pH1H2-N3 (Fig. 5.16,

bottom panel) could correspond to the sequon-containing tryptic peptide with a

DiNAcBac-HexNAc4 modification (Table 5.4, m/z 2207.87). Unfortunately, this peak was

of low intensity and tandem MS fragmentation was unsuccessful.

In summary, MALDI-TOF mass spectrometry analysis of NGRP3 glycoforms mainly

resulted in identification of peaks corresponding to unmodified tryptic peptides. However,

using DHB matrix, an ion with at m/z 2207 was identified in glycosylated NGRP3

produced from pH1H2-N3. This species could be a glycopeptide containing a DiNAcBac-

HexNAc4 modification. This would suggest that the addition of both C. fetus pglH

homologs into a C. jejuni pglH::kn system results in the transfer of a further two HexNAc

residues onto the N-linked glycan structure. However, tandem MS data could not be

obtained from this species and hence the structure was not confirmed. To elucidate if this

species is such a glycopeptide, more sensitive mass spectrometry analysis was undertaken.

In addition, further analysis was warranted to confirm the structure of N-glycan present in

glycosylated NGRP3 expressed from pN3 and pH1-N3.

152

153

5.5.4.4 LC-MS/MS analysis of glycosylated NGRP3

To investigate the N-glycan structures created by the E. coli hybrid pgl systems described

in this chapter, purified NGRP3 preparations were analysed by LC-MS/MS. Analysis was

performed on the Q Exactive™ HF Hybrid Quadrupole-Orbitrap™ Mass Spectrometer

with Higher-energy Collisional Dissociation (HCD) fragmentation. MS/MS scan events

with precursor ions at the predicted m/z values were identified using mMass v5.5

(Niedermeyer and Strohalm, 2012) and scanned for a HexNAc oxonium ion (m/z 204.09),

which is characteristic of glycopeptide fragmentation spectra obtained using HCD (Zhao

et al., 2011). Marker ions at m/z 126, 138, 144, 168 and 186 can also be found,

representing further fragmented HexNAc residues (Zhao et al., 2011). Disaccharide

oxonium ions such as m/z 366 (HexHexNAc) can also be identified, which can help

elucidate glycan structure (Toghi Eshghi et al., 2016). Identification of these ions was used

to confirm scans corresponded to glycopeptides and are labelled on each spectrum (Fig.

5.17).

The presence of the unmodified sequon-containing tryptic peptide, 145

HHTDSNSTIR154

,

as a doubly charged species at m/z 584.21 was confirmed in NGRP3 produced in the

absence of Campylobacter pgl genes. Fragmentation of this parent ion resolved peaks

corresponding to 7 y-ions and 1 b-ion (Fig 5.17A). From glycosylated NGRP3 produced

in the presence of ppglpglH::kn, a precursor ion with m/z 901.41 corresponding to

145HHTDSNSTIR

154 with DiNAcBac-HexNAc2 in a doubly charged state was confirmed.

Fragmentation resolved peaks representing the peptide with partial N-glycans attached,

including DiNAcBac-HexNAc and DiNAcBac (Fig 5.17B).

154

155

A precursor ion of m/z 722.99 was identified from glycosylated NGRP3 produced in the

presence of ppglpglH::kn and pH1-N3 (5.18A). This m/z corresponds to

145HHTDSNSTIR

154 with DiNAcBac-HexNAc3-Hex in a triply charged state. This

confirms previous data suggesting that PglH1 modifies the trisaccharide produced by the

C. jejuni pglH::kn locus and identifies the transferred residues as a HexNAc and a Hex.

Fragmentation yielded ions with m/z values 1395 and 1598 corresponding to the peptide

with truncated glycan structures of DiNAcBac and DiNAcBac-HexNAc, respectively (Fig

5.18A). Further ions corresponded to y-ions with DiNAcBac, DiNAcBac-HexNAc or

DiNAcBac-HexNAc2 modification. Although no ions containing DiNAcBacHexNAc3

were identified, the marker ion of m/z 366.14 (HexNAcHex) was detected. Therefore, the

lack of a species containing DiNAcBacHexNAc3 is likely due to the first fragmentation

event removing the terminal HexNAc and Hex residues in conjunction.

Analysis of glycosylated NGRP3 produced in the presence of ppglpglH::kn and pH1H2-

N3 also yielded a precursor ion of m/z 722.99 (5.18B, upper spectrum). Fragmentation of

this ion yielded a peak, m/z 1778, corresponding to the m/z of a b7 ion with a DiNAcBac-

HexNAc3-Hex modification. A peak at an m/z of a b2 peptide ion was also evident.

However, another precursor ion of m/z 736.66 representing 145

HHTDSNSTIR154

with

DiNAcBac-HexNAc4 in a triply charged state (Fig 5.18B, lower spectrum) was identified

from glycosylated NGRP3 produced in this system. Notably, precursor ions at this m/z

were over three times more abundant than m/z 722.99 ions (112 and 29 scan events,

respectively). Fragmentation of a m/z 736.66 parent ion produced species corresponding to

the peptide with partial N-glycan structures of DiNAcBac-HexNAc3, DiNAcBac-

HexNAc2, DiNAcBac-HexNAc and DiNAcBac attached (Fig.5.18B, lower spectrum). In

addition, several species with m/z values corresponding to y-ions with either DiNAcBac or

DiNAcBac-HexNAc were identified. Ions corresponding to the unmodified peptide and its

b2, b3 and b4 ions were also observed. Lastly, an ion of m/z 407.17 was identified which

likely corresponds to two HexNAc residues. These data suggest that two pentasaccharide

N-glycan structures were present, the DiNAcBac-HexNAc3-Hex structure first produced

by introduction of PglH1 alone and another structure, DiNAcBac-HexNAc4, produced

only when PglH2 is also present.

156

157

Therefore, LC-MS/MS analysis confirmed the production of NGRP3 modified with a

DiNAcBac-HexNAc2 glycan from E. coli BL21 ppglpglH::kn pETNGRP. These data

provided evidence that upon addition of C. fetus pglH1 into the system, a further HexNAc

and Hex residue were transferred to the N-glycan structure giving DiNAcBac-HexNAc3-

Hex. Upon introduction of C. fetus pglH2 in conjunction with pglH1, an N-glycan

structure consisting of DiNAcBac-HexNAc4 was also present. Therefore, these data

suggest that C. fetus PglH1 may be responsible for adding the first GlcNAc residue and

the branching glucose within the C. fetus N-linked glycan. Secondly, these data suggest

that PglH2 may transfer the alternative branching GlcNAc residue (See Fig. 5.19, pathway

to the left). The latter structure is the more dominant structure in C. fetus (Nothaft et al.,

2012), which may explain the abundance of this ion in comparison to the other. However,

it should be noted that as the C. jejuni PglI enzyme was present in the system, it is possible

that C. fetus PglH1 attached a GlcNAc residue which was then acted upon by C. jejuni

PglI, transferring a branching glucose residue (Fig. 5.19, pathway to the right). Such

interference would be similar to its native role of transferring a branching glucose residue

to the third GalNAc residue of the C. jejuni N-glycan (Kelly et al., 2006). Nevertheless,

these data suggest that pglH1H2 can mediate the construction of a partial C. fetus N-linked

glycan using a C. jejuni pglH::kn trisaccharide N-glycan as a substrate. To engineer the

full N-glycan structure, the predicted transferase gene cf1389 was introduced into this

pglH1H2-containing system.

158

159

5.6 Construction and analysis of hybrid pgl system containing C.

fetus pglH genes and predicted glycosyltransferase cf1389.

5.6.1 Cloning strategy

This chapter has demonstrated a role for C. fetus pglH homologues in building the C. fetus

N-glycan, adding two of the three GlcNAc residues and possibly the branching Glc. To

investigate if the predicted glycosyltransferase gene cf1389 is involved in creating the

remainder of the C. fetus N-glycan structure, a plan to introduce this gene along with

pglH1 and pglH2 was devised. Briefly, this plan included cloning the three genes together

into pGEM®-T-easy vector, amplifying the construct with BclI ends, then digesting and

ligating it with BglII-digested pETNGRP. Genes pglH1 and pglH2 were amplified as in

5.4.1.1. The product was A-tailed using MyTaq™ Red Mix (Bioline) and ligated into

pGEM®-T-Easy, creating pGEMH1H2. Positive clones were identified by cPCR using

primers 80 and 81. The sequence of pglH1H2 was checked by sequencing using primers

80, 81 and 1450. Cf1389 was amplified as in 5.3.1.3, digested with BamHI and ligated

with BamH1-digested pGEMH1H2 to give pGEMX12. Following transformation, positive

clones were identified by cPCR using primer 80 and 1388. Further cPCR to confirm that

pglH1H2 and cf1389 were in the same orientation was performed using primers 80 and

1390 (1188 bp predicted). The presence of all three correctly sized genes was confirmed

by PCR using primers 1379 and 1439. Cf1389 was sequenced using primers 80 and 1486.

The pglH1H2-cf1389 region was amplified from pGEMX12 using primers 1379 and 1439

(Table 2.1), adding BclI sites to each end. This ~3.3 kb product was digested with BclI

and ligated into BglII-digested pETNGRP. Positive transformants were identified by

cPCR using primers 1390 and 1411 (Table 2.1). Plasmid was purified and the sequence of

the C. fetus genes confirmed by sequencing using primers 1388, 1411, 1425, 1485 and

1486. To offer improved performance during mass spectrometry analysis NGRP was

replaced with NGRP3 as with vectors in 5.5.2, creating pX12-N3. C. fetus genes were

sequenced using primers 1388, 1389, 1411, 1425, 1485 and 1558. Following

transformation of pX12-N3 into E. coli BL21 (λDE3) ppglpglH::kn, positive clones were

identified by PCR using primers 1390 and 1411 (not shown).

160

5.6.2 Addition of cf1389 to a ppglpglH::kn CfpglH1H2 background resulted in

three apparent forms of NGRP3

NGRP3 production was induced in E. coli BL21 (λDE3) ppglpglH::kn pX12-N3 using 1

mM IPTG at 16°C overnight. Notably, the yield of biomass was lower compared with

previous strains used in this chapter (~ 6.7 g pellet from 1 L induced culture, in contrast to

between 10 and 16g pellets obtained per 1 L culture). Although NGRP3 production was

not obvious in the coomassie-stained lysate of this induced strain (not shown), his6-tagged

NGRP3 was purified using Nickel affinity chromatography. The sample was eluted in 10

mL elution buffer and 1 mL fractions collected. Fractions 4-8 were combined, dialysed

against 30 mM Tris 0.15 M NaCl pH 7.5 and separated by SDS-PAGE (Fig. 5.18A).

Proteins were observed at around 19.5 kDa and 21.5 kDa, likely corresponding to

unglycosylated NGRP3 and NGRP3- DiNAcBac-HexNAc4, respectively. In addition a

considerable amount of a protein at around 23 kDa was purified, suggesting the possible

presence of another form of NGRP3.

A western blot of this eluate and that from E. coli BL21 (λDE3) ppglpglH::kn pH1H2-N3

was probed with anti-his6 antibodies (Fig. 5.20B). Due to the low biomass yield from

pX12-N3, approximately 20 times more eluate sample was loaded onto the gel in

comparison with the eluate from pH1H2-N3. In both strains unglycosylated NGRP3 was

confirmed at around 21 kDa (predicted 21.78 kDa) and a g1 form of NGRP3 at around

23.5 kDa. However, the 23 kDa protein observed by coomassie-staining was not reactive,

suggesting this protein was not a form of NGRP3 and rather another protein contaminating

the preparation.

In summary, the addition of cf1389 into the hybrid glycosylation system resulted in lower

biomass and protein yield upon induction. Nevertheless, production of a g1 form of

NGRP3 similarly sized to that seen previously was identified. To further study the forms

of NGRP3 produced in this system, they were analysed by mass spectrometry.

161

162

5.6.3 MALDI-TOF MS analysis of glycosylated NGRP3 produced in the presence

of C. fetus pglH genes and cf1389.

To investigate the N-glycan structure of glycosylated NGRP3 purified from E. coli BL21

(λDE3) ppglpglH::kn pX12-N3, the g1 band was excised and analysed by MALDI-TOF

mass spectrometry using CHCA. Peaks with m/z ratios corresponding to 9 tryptic peptides

of NGRP3 (one with 1 missed cleavage site) were identified (Fig. 5.21). This included a

relatively small peak at m/z 1167.793, corresponding to the unmodified sequon-containing

tryptic peptide, indicating the presence of unglycosylated NGRP3. However, another small

peak with an m/z of 2208.393 was produced, suggesting the presence of the tryptic peptide

modified with DiNAcBac-HexNAc4 (as found with pH1H2-N3).

Therefore, these data suggest that both unglycosylated NGRP3 and NGRP3 modified with

DiNAcBac-HexNAc4 were produced. The peak corresponding to the unmodified sequon-

containing tryptic peptide may be due to cross contamination from the ~ 21.5 kDa

unglycosylated form of NGRP3 evident in the preparation (Fig. 5.21). Alternatively, it

could be due to unglycosylated NGRP3 with its signal peptide attached which, if present,

may have been indistinguishable from glycosylated NGRP3 by coomassie staining and

western blotting. If this is true, another tryptic peptide with a predicted m/z of 3735.9077

would be possible, however, this was above the detectable mass range during analysis so

would not have been identified. Nevertheless, these data suggest that introduction of

cf1389 into the hybrid pgl system did not change the N-glycan structure created and hence

this gene does not appear to play a role in C. fetus N-linked glycosylation.

163

164

5.7 Discussion

This Chapter investigated the role of predicted C. fetus transferase genes, cf1389, pglH1

and pglH2, in assembling residues at the non-reducing end of the N-linked glycan

structure. These genes were integrated into an E. coli system which produces an N-

glycoprotein reporter protein, NGRP, modified with a truncated C. jejuni N-glycan to

identify if they were able to add sugar residues onto this structure.

Each candidate gene was singly introduced into the E. coli ppglpglH::kn system to identify

the first-acting glycosyltransferase gene. Western blotting revealed that addition of

CfpglH1 reduced the electrophoretic mobility of NGRP, indicating an increase in its

molecular weight, likely due to transfer of a larger N-glycan structure. Upon introducing

cfpglH1 in combination with either cfpglH2 or cf1389, a further shift in electrophoretic

mobility was observed with cfpglH2 only. Therefore, it appeared that the gene products of

cfpglH1 and cfpglH2 were able to act sequentially on a C. jejuni pglH::kn N-glycan

structure.

The various N-glycosylated NGRP products were purified using nickel affinity

chromatography and the doubly N-glycosylated (g2) form was analysed using MALDI-

TOF MS. However, none of the three possible tryptic glycopeptides were resolved. Instead

two of these possible tryptic peptides were observed in an unmodified state. The

remaining possible tryptic glycopeptide contains two N-glycosylation sequons, and if fully

modified, even with the truncated C. jejuni pglH::kn glycan, would give a peptide with an

m/z of 5329. Hence, this peptide was likely not resolved due to creating an ion with such a

large m/z. Therefore, it may be that the g2 form of NGRP is predominantly modified at the

two sites, N150 and N156, which reside within a single tryptic peptide. In support of this,

site-directed mutagenesis of the native precursor of NGRP, Cj0114, has previously

identified that these sites are preferentially modified in H. pullorum, as these are predicted

to reside in a flexible linker region within the protein (Jervis et al., 2010).

To alleviate the issues regarding MS analysis of NGRP, an alternative version of this

protein, NGRP3, was integrated into the systems developed. This version has been altered

to create a novel sequon-containing tryptic peptide that is small and positively charged

(Jervis & Linton, unpublished). NGRP3 was similarly modified by the hybrid N-

glycosylation systems and analysed by MALDI MS. Two different sample matrices were

tested, CHCA and DHB, the latter often applied for glycopeptide analysis. No possible

glycopeptides were identified from the systems containing CfpglH1 alone with either

165

matrix. However, using DHB, a possible glycopeptide was identified from gNGRP3

produced in the presence of cfpglH1H2, although attempts to further fragment this ion

failed. Nevertheless, the size of the peptide indicated an N-glycan structure consisting of

DiNAcBacHexNAc4. Therefore, these data suggested that cfpglH1H2 work in sequence to

transfer the first two GalNAc residues of the C. fetus N-glycan structure.

To gain further insight into the N-glycan structures produced by the hybrid systems,

gNGRP3 samples were analysed using LC-MS/MS. Confirming the MALDI MS findings,

glycopeptides containing an N-glycan structure of DiNAcBacHexNAc4 were identified

from NGRP3 produced in the presence of cfpglH1H2. However, when cfpglH1 alone was

present, glycopeptides containing a DiNAcBacHexNAc3Hex N-glycan were identified. In

addition, this structure was also found when pglH1H2 were present, albeit at a lower

abundance in comparison to the DiNAcBacHexNAc4-containing glycopeptides. Two

possible hypotheses were formed (Fig. 5.19) regarding the mechanism behind these

observed N-glycan structures. In each case, C. fetus PglH1 adds a GlcNAc residue onto

the C. jejuni pglH::kn trisaccharide. Following this, either C. fetus PglH1 or C. jejuni PglI

transfers a branching glucose to this intermediate tetrasaccharide to form the

DiNAcBacHexNAc3Hex N-glycan identified in the pglH1-containing system. The

DiNAcBacHexNAc4 N-glycan identified upon introducing C. fetus pglH2 into this system

is then a result of PglH2 transferring either the branching or linear GlcNAc residue onto

the tetrasaccharide intermediate (created by the PglH1-mediated addition of a GlcNAc).

C. jejuni PglI transfers the branching Glc onto GalNAc within the C. jejuni N-glycan

structure (Linton et al., 2005). Although C. fetus PglH1 is hypothesised to add a GlcNAc

in this system, and hence should not provide a substrate for C. jejuni PglI, this was not

verified. Therefore, it is plausible that interference from C. jejuni PglI occurred. As this

reaction would not be optimal for this enzyme, it can be expected that the Hex-containing

structure would be less abundant than that created by PglH2 activity. Indeed, the Hex-

containing structure appeared over three times less abundant than the DiNAcBacHexNAc4

N-glycan. However, in C. fetus the GlcNAc branch-containing structure is 4 times more

abundant than the Glc-branching structure (Nothaft et al., 2012). Therefore, the relative

abundancies of the N-glycan structures observed do not favour either hypothesis. If

CfPglH1 does catalyse the addition a GlcNAc and a Glc residue onto two different

acceptor substrates, GalNAc and GlcNAc, this would be a striking finding. Although

bacterial glycosyltransferases are broadly known to have more relaxed acceptor

specificities than their eukaryotic counterparts (Johnson, 1999), such relaxed specificity is

166

not typical for the GTs of the ε-proteobacteria Pgl systems. It has been demonstrated that,

if a GlcNAc is transferred in place of DiNAbac (by an E. coli GT), this can be utilised as

an acceptor substrate by C. jejuni PglA. However, this happens at a lower efficiency and is

only observed in the E. coli host system not in vivo (Linton et al., 2005). Furthermore,

although there are examples of bacterial GTs which can utilise more than one sugar donor

(Naegeli et al., 2014), this is not a representative feature of Pgl systems. For example,

although C. jejuni PglH transfers multiple sugar residues onto the N-linked glycan

structure, each residue is a GalNAc and hence the enzyme has specificity for only a single

sugar donor. Therefore, the suggested role of C. fetus PglH1 would be an unusual feature

of N-linked glycosylation in ε-proteobacteria.

As introducing cfpglH1 and H2 did not build the full C. fetus N-glycan hexasaccharide

structure, it was hypothesised that the predicted glycosyltansferase, cf1389, may be

responsible for transfer of the final GlcNAc residue. However, introducing this gene into

the hybrid C. jejuni/C. fetus pgl system containing cfH1H2 caused a reduction in both

biomass yield and N-glycosylation efficiency. Furthermore, the N-glycosylation that was

identified appeared to include the DiNAcBacHexNAc4, as found in the system without

cf1389 present. Therefore, it seemed that the product of cf1389 did not contribute to N-

glycosylation in this system, and even inhibited the process. This prompted further

bioinformatics analysis into this gene and the two uncharacterised genes downstream, all

of which are located within the C. fetus pgl locus in C. fetus fetus 82-40.

As these genes are not found in the C. fetus 04/554 pgl locus, it was hypothesised that this

region may result from integration of bacteriophage DNA into the genome. Indeed,

PHASTER (PHAge Search Tool Enhanced Release) web server, a tool for identifying

prophage DNA (Arndt et al., 2016), found this region to be probable prophage DNA

within the C. fetus 82-40 genome. Genes cf1389 and cf1390 matched sequences from

Chronobacter phage ENT47670 (Genbank accession HQ201308), identified from

Chronobacter sakazakii. The former matched with a bactoprenol glucosyl transferase,

which could explain the negative impact introduction of this gene had on the C. fetus

engineered system created in this chapter. Bactoprenol glucosyl transferases catalyse the

addition of glucose to undecaprenyl-pyrophosphate (also known as UndPP/Bactoprenol)

(Allison and Verma, 2000). Hence, if Cf1389 transfers a glucose/GlcNAc residue onto

UndP, it could interfere with the N-linked glycosylation system by utilising the UndP

substrate needed to create LLO precursor. Furthermore, this could have a far-reaching

effect in E. coli, inhibiting other systems which require an UndP substrate, such as

167

peptidoglycan and lipopolysaccharide biosynthesis (Heijenoort, 2001; Wang and Quinn,

2010). Such interference may only be significant in the E. coli system due to a potentially

higher copy number of Cf1389 than when expressed in vivo.

Cf1390 matched an uncharacterised hypothetical protein of Chronobacter phage

ENT47670. Performing protein BLASTP searches of Cf1390 identified a possible

Glucos_trans_II domain, placing it within the GtrII superfamily. The canonical protein of

this superfamily is glucosyl transferase II (gtrII) from Shigella phage SfII, which mediates

transfer of a glucose residue onto a rhamnose present within the O-antigen repeat unit

(Mavris et al., 1997). Furthermore, performing a BLASTP search with the predicted

product of cf1391 found the protein to have a GtrA-like domain. GtrA is a Shigella phage

SfX protein thought to mediate transport of UndP-Glc across the inner membrane.

Therefore, it appears that cf1389-91 constitute a horizontally-acquired LPS O-antigen

modification locus termed GtrABC, as documented in Shigella and Salmonella (Allison &

Verma, 2000; Davies et al., 2013). In line with this, the more recent availability of

complete genomes of C. hyointestinalis (Miller, Yee, & Chapman, 2016) reveal there are

no homologues of cf1389-1391 in these species. Therefore, as these C. hyointestinalis

produce identical N-glycan structures to C. fetus, this suggests that none of these genes

play a role in the N-glycosylation pathway.

In conclusion, glycosyltransferase activity of C. fetus PglH1 and PglH2 was demonstrated

using the C. jejuni pglH::kn trisaccharide as a substrate. These data revealed that CfPglH1

likely mediates transfer of one GlcNAc residue and is followed by the activity of CfPglH2,

which may add a further GlcNAc residue. Furthermore, CfPglH1 may also transfer a

branching glucose residue. However, this transfer could have been due to interference

from the C. jejuni PglI enzyme. Nevertheless, these enzymes were not sufficient to

construct the entire non-reducing end of the C. fetus N-glycan, as the N-glycan structures

produced by the pglH1H2-containing system each lacked a GlcNAc residue. Investigation

into another potential glycosyltransferase within the C. fetus pgl locus found this protein to

not be involved in the N-glycosylation process, instead likely forming part of an O-antigen

serotype conversion prophage locus. Therefore, the glycosyltransferase responsible for

adding the final GlcNAc residue of the C. fetus N-glycan structure remains unknown.

Furthermore, the identity of the enzyme responsible for transferring the branching glucose

residue to the C. fetus N-glycan remains unclear.

168

Chapter 6. C. fetus N-glycoproteome

prediction and the conservation of N-

glycoproteins between Campylobacter

species

169

6.1 Introduction

The C. fetus Pgl system has not been studied extensively and the extent of N-glycan

protein modification in comparison to C. jejuni is not known. In C. jejuni there are over 60

experimentally identified N-glycoproteins (Scott et al., 2014, 2011; Young et al., 2002). In

C. fetus, Nothaft et al. (2012) identified 65 N-glycopeptides using LC-MS/MS,

corresponding to 32 distinct N-glycosylation sites from 25 proteins. Numerous reactive

bands were evident upon probing C. fetus whole cell lysates with antiserum raised against

C. fetus free oligosaccharides (Nothaft et al., 2012), which suggests there are many more

N-glycoproteins. This is also evident from the extent of reactivity observed with an

antiserum raised against a C. fetus N-linked glycoprotein (See Chapter 4). Although

glycopeptide enrichment coupled with LC-MS/MS is a powerful tool for discovering

glycopeptides, there are limitations as many N-glycoproteins may be insoluble, or

glycosylation sites may reside within large tryptic peptides which cannot be resolved

easily. In addition, glycoproteins which are less abundant can be missed. In line with this,

an N-glycoproteome prediction pipeline suggests there are at least 148 potential N-linked

glycoproteins in C. jejuni NCTC 11168 (Frost, 2015). Therefore, it is likely that C. fetus

N-glycosylates more proteins than have been experimentally identified thus far, and that

the N-glycoproteome prediction pipeline could reveal more potential N-glycoproteins.

As C. jejuni is known to N-glycosylate numerous proteins, many studies have tried to

elucidate the function of this modification. For example, Alemka et al. (2013)

demonstrated that N-glycosylation influences the growth of C. jejuni in the presence of

chicken caecal contents ex vivo, likely due to N-glycan modification providing protection

from the action of chicken proteases. Other evidence also suggests that Campylobacter N-

glycosylation is broadly important for bacterial fitness, cell attachment, colonisation and

survival in the host (Hendrixson and DiRita, 2004; Szymanski et al., 2002; van Sorge et

al., 2009). However, studies on the influence of N-glycosylation on individual proteins are

limited and largely do not suggest that modification alters protein activity. An influence of

N-glycosylation on the function of virB10, a type IV secretion system component, has

been demonstrated (Larsen et al., 2004). However, studies on proteins such as ZnuA and

JlpA suggest that N-glycan modification does not significantly affect protein function

(Davis et al., 2009; Kakuda and DiRita, 2006; Scott et al., 2009). Therefore, it can be

argued that N-glycosylation does not afford advantages to specific target proteins, but

rather the abundance of the modification itself provides an overall advantage, such as

170

increased resistance to proteases, sufficient for continued selection. If this is correct, there

is likely low level conservation of the identity of N-glycosylation target proteins between

different Campylobacter species, provided that each species has a significant number of

modified proteins overall. As experimental data regarding the N-glycoproteins of

Campylobacter species other than C. jejuni are relatively sparse, bioinformatics analysis

could also be utilised to putatively identify the extent of N-glycoprotein conservation

among C. jejuni and these species.

This chapter describes the prediction of the C. fetus fetus 82-40 N-glycoproteome using

the previously developed bioinformatics pipeline (Frost, 2015). This prediction pipeline is

a perl script which inputs protein sequences into PSORTb (Yu et al., 2006), LipoP

(Juncker et al., 2003) and TMHMM (Krogh et al., 2001) to predict subcellular localisation

and then identifies any D/E-X-N-S/T sequons in the sequences, generating an output file

which can be manually curated to produce a list of predicted N-glycoproteins (Fig. 6.1).

PglB-mediated modification of a newly predicted N-glycoprotein was confirmed

experimentally in C. fetus. Lastly, the conservation of N-glycoproteins among

Campylobacter species including C. fetus was investigated.

171

172

6.2 C. fetus N-glycoproteome prediction and validation

6.2.1 The predicted C. fetus fetus 82-40 N-glycoproteome

Protein sequences, extracted from the C. fetus fetus 82-40 genome using Artemis

(Rutherford et al., 2000), were fed into the N-glycoprotein prediction pipeline (Fig. 6.1)

and the output file manually curated as described in Figure 6.2. Proteins not confidently (<

9) assigned a subcellular location by PSORTb but with predicted transmembrane (TM)

regions or signal peptide cleavage sites by the other programs were assigned as ‘unknown

extracytoplasmic’ and still considered a potential target for PglB. Although the predicted

N-glycoproteome of C. jejuni 11168 NCTC was previously established by Frost (2015)

using this pipeline, the process was repeated to allow for accurate comparison with the

newly generated C. fetus data. Furthermore, the protein sequences extracted from the C.

jejuni 11168 genome by Frost (2015) included the products of pseudogenes; such

sequences were excluded from this analysis. Four experimentally identified C. jejuni N-

glycoproteins (Cj0494, Cj1345c, Cj1496c and Cj1032) had been missed by the prediction

pipeline as they were predicted as cytoplasmic. These proteins were added to give the total

predicted C. jejuni N-glycoproteome consisting of 158 proteins.

Of the total 1719 C. fetus proteins, 308 (18%) contained N-glycosylation sequon

sequences. Of these 308, 129 were identified as possible N-glycoproteins based on

subcellular localisation predictions (Table 6.1). This included 20 of the 25 N-glycoproteins

previously experimentally identified by Nothaft et al. (2012). The remaining five proteins

(CFF8240_0198, 0217, 0365, 0372 and 1782) were not identified as they were classified

as ‘unknown’ by PSORTb and contained no predicted transmembrane domains or signal

peptide cleavage sites according to TMHMM and LipoP, respectively (Table 6.1, shown

without colour). Subsequent analysis of these proteins using an alternative localisation

tool, CELLO (Yu et al., 2006), predicted that all but Cf0372 were extracytoplasmic.

However, SignalP 4.1 (Petersen et al., 2011), a server which predicts signal peptide (SP)

cleavage sites, suggested that Cf0372 does harbour a SP cleavage site. Therefore, although

LipoP can identify SpI or SpII (lipoprotein) signal peptide cleavage sites, it appears that

SignalP may be more accurate for identifying SpI cleavage sites. This is likely due to the

tool being optimised for specific identification of SpI cleavage sites and possibly because

it is recently updated (version 4.1 is currently available) whereas there is only a single

version of LipoP released in 2003. These findings indicate that the prediction pipeline

could be improved if more accurate protein localisation prediction tools were included.

173

Upon combining the experimental data set of Nothaft et al. (2012) with this prediction

pipeline result, the predicted total C. fetus N-glycoproteome consists of 134 proteins,

making up 7.8% of its total proteome and containing 219 sequons in total. These figures

are similar to that found in C. jejuni, which has a predicted total N-glycoproteome of 158

proteins, constitutes 9.7% of its total proteome and contains 185 sequons. Therefore, it

appears that the extent of N-glycosylation is as significant in C. fetus as it is in C. jejuni.

174

Table 6.1. Complete predicted C. fetus N-glycoproteome. Colour indicates predicted

location assignment based on prediction pipeline described in Figure 6.1: Blue; inner

membrane. Green; periplasmic. Red; Outer membrane. Purple; extracellular. Grey; unknown

but extracytoplasmic. White; predicted as cytoplasmic. Bold indicates experimental

identification by Nothaft et al. (2012).

Locus

Tag

No. of

sequons Sequon sequence (s) Predicted location

Cf0010 2 EENSS, DQNGS Extracellular

Cf0014 1 DTNSS Extracytoplasmic

Cf0018 2 DINLS, DINSS Inner Membrane

Cf0026 1 DFNRS Extracellular

Cf0065 1 EINAT Inner Membrane

Cf0067 3 DKNST, DDNGT, DDNKT Extracytoplasmic

Cf0092 1 EINAS Extracellular

Cf0107 3 DANAT, DTNDT, DNNIT Periplasmic

Cf0117 1 EQNAS Outer Membrane

Cf0118 1 EANST Inner Membrane

Cf0119 2 DYNGT, ENNST Inner Membrane

Cf0121

(ccoG) 1 DVNDT Inner Membrane

Cf0134 1 DDNFT Extracytoplasmic

Cf0136 1 DDNKS Extracytoplasmic

Cf0138 1 DINSS Inner Membrane

Cf0151 1 DKNQS Inner Membrane

Cf0191 1 ENNTT Extracytoplasmic

Cf0198 2 DINST, DINGS Cytoplasmic

Cf0217 2 DANYS, DLNDT Cytoplasmic

Cf0243 1 DENST Inner Membrane

Cf0272 1 ELNAT Inner Membrane

Cf0290 7 DKNNS, ESNMT, DENVT, DSNLS, DENES,

DINLS, DKNTT Extracytoplasmic

Cf0301 2 DLNLT, DYNTT Extracytoplasmic

Cf0312 11

DANLS, EANIT, EFNFS, DTNSS, DANAS,

DVNYS, DNNIS, DINIT, DINGT, DTNIT,

ETNES

Outer Membrane

Cf0358 2 EGNLT, DLNNS Extracytoplasmic

Cf0365 1 DNNET Cytoplasmic

Cf0367 2 DINKT, DKNGT Inner Membrane

Cf0368 2 EDNIS, DNNFT Inner Membrane

Cf0372 2 DVNRT, DNNSS Cytoplasmic

Cf0384 2 DTNAT, DLNLT Extracytoplasmic

Cf0398 1 DENVS Extracytoplasmic

Cf0401 1 DGNIS Inner Membrane

Cf0404 1 DVNNT Inner Membrane

Cf0431 4 EDNIT, DHNIT, DSNFS, DKNIS Inner Membrane

Cf0440 1 ERNLS Periplasmic

Cf0443 1 DHNTT Extracytoplasmic

Cf0444 3 DKNET, EQNAS, DNNET Extracytoplasmic

Cf0445 1 DENLT Periplasmic

Cf0449 1 DVNTT Extracytoplasmic

Cf0459 1 DLNLT Inner Membrane

175

Cf0461 1 DNNVT Extracellular

Cf0490 3 DINET,EINFT,DGNDT Extracellular

Cf0502 1 DANYS Extracytoplasmic

Cf0505 2 ENNES, DINGT Extracytoplasmic

Cf0511 1 EINIS Inner Membrane

Cf0517 2 DINST, DLNST Outer Membrane

Cf0529 1 DINFS Inner Membrane

Cf0536 1 ELNLT Inner Membrane

Cf0537 1 DWNSS Inner Membrane

Cf0553 2 DINQT, EQNSS Inner Membrane

Cf0557 1 EKNSS Inner Membrane

Cf0569 2 DVNDT, ESNAT Inner Membrane

Cf0572 3 EQNAS, DINGT, DDNSS Extracytoplasmic

Cf0583 1 ESNTS Inner Membrane

Cf0607 4 DGNIT, DQNET, ESNVS, EYNST Outer Membrane

Cf0625 2 DTNGT, DINGS Inner Membrane

Cf0636 2 EQNRT, DKNIT Extracytoplasmic

Cf0682 1 DFNLS Inner Membrane

Cf0684 1 DDNNT Inner Membrane

Cf0699 6 ELNLT, DLNYT, DFNIS, DANAS, DVNVT,

DENKT Outer Membrane

Cf0705

(flgC) 1 ELNKS Periplasmic

Cf0708 6 DSNVT, DTNET, DENIT, DANAS, DLNKT,

DNNIT Extracytoplasmic

Cf0723 1 DINFS Inner Membrane

Cf0725

(int) 2 EKNAS, ENNMS Inner Membrane

Cf0728

(secD) 1 DSNLT Inner Membrane

Cf0735 2 DINGS, DNNTT Extracytoplasmic

Cf0738 2 EKNVT, EDNNS Extracellular

Cf0745 1 ENNNT Outer Membrane

Cf0748 1 DNNDS Inner Membrane

Cf0758 1 EQNAT Extracytoplasmic

Cf0760 1 DFNQT Outer Membrane

Cf0781 1 DINGT Periplasmic

Cf0786

(pyrD) 1 DKNAT Inner Membrane

Cf0787 1 EKNES Inner Membrane

Cf0799 1 DFNIT Inner Membrane

Cf0800 2 DRNLS, DLNTT Periplasmic

Cf0802

(cutF) 3 DSNGS, DINQT, DDNET Outer Membrane

Cf0825 1 DDNNS Inner Membrane

Cf0845

(murG) 1 EHNIS Inner Membrane

Cf0855 1 DINSS Inner Membrane

Cf0872 2 EHNDT, DRNIT Inner Membrane

Cf0875

(ppk) 1 EKNSS Inner Membrane

Cf0898 1 DNNSS Inner Membrane

Cf0944 2 DQNFT, DGNST Inner Membrane

176

Cf0948 2 DLNLT, EANST Outer Membrane

Cf0954 1 EKNRT Extracytoplasmic

Cf0955 1 ETNFT Extracytoplasmic

Cf1003

(era) 1 ESNKT Inner Membrane

Cf1004 1 DSNIT Extracytoplasmic

Cf1005 2 DVNVS, DINVS Extracytoplasmic

Cf1009 2 ETNQS, EKNRS Extracytoplasmic

Cf1026 1 DENRS Extracytoplasmic

Cf1086 1 EGNET Inner Membrane

Cf1126 6 DKNIS, DKNIS, DSNIS, DKNIT, DQNIT,

DENSS Extracytoplasmic

Cf1144

(cvpA) 2 DLNRT, DSNET Inner Membrane

Cf1147 1 DSNNT Extracytoplasmic

Cf1158

(napB) 2 DKNKS, DSNES Periplasmic

Cf1165 1 DINLT Inner Membrane

Cf1182 1 DLNVT Inner Membrane

Cf1217 1 DINRS Inner Membrane

Cf1221 1 DLNDS Extracytoplasmic

Cf1227 1 EVNQS Inner Membrane

Cf1242

(putP) 1 DKNSS Inner Membrane

Cf1268 2 DANRS, DRNIS Inner Membrane

Cf1277 1 DLNLT Inner Membrane

Cf1298

(feoB) 1 ESNIT Inner Membrane

Cf1304 1 DLNRT Extracytoplasmic

Cf1343 1 DKNST Outer Membrane

Cf1353

(pglG) 2 ENNIS, DINIS Inner Membrane

Cf1363 2 DANLS, DTNET Inner Membrane

Cf1372 1 ELNVT Extracytoplasmic

Cf1383

(pglB2) 2 DYNAS, DQNFT Inner Membrane

Cf1417 1 EDNIS Outer Membrane

Cf1439 1 EYNAT Outer Membrane

Cf1479 2 DANQT, DDNKT Extracytoplasmic

Cf1499 1 EFNDS Extracytoplasmic

Cf1518 2 DLNLT, EKNSS Outer Membrane

Cf1522 1 ELNIT Inner Membrane

Cf1523 1 EANAT Inner Membrane

Cf1530 1 DRNRS Extracytoplasmic

Cf1543 1 EYNES Extracytoplasmic

Cf1568 1 EMNDT Inner Membrane

Cf1620 1 DNNET Inner Membrane

Cf1648 1 EYNFS Outer Membrane

Cf1673 1 DKNSS Inner Membrane

Cf1723 1 DENLS Inner Membrane

Cf1729 1 DINAT Inner Membrane

Cf1730 1 EHNGT Inner Membrane

Cf1735 1 EFNAT Inner Membrane

177

Cf1737 1 DKNAS Inner Membrane

Cf1750 1 DFNDS Extracytoplasmic

Cf1756

(nhaA) 1 DFNLS Inner Membrane

Cf1782 3 EYNQT, DSNST, DVNIS Cytoplasmic

Cf1804 1 ETNKT Inner Membrane

178

6.2.2. Validation of predicted C.fetus N-glycoproteins

Numerous N-glycoproteins have been predicted in C. fetus using the bioinformatics

pipeline described above with 20 being previously identified experimentally by Nothaft et

al. (2012). However, a predicted extracytoplasmic protein containing an N-glycosylation is

not necessarily an N-glycoprotein in vivo. For example, the sequon could reside within a

buried structural feature and hence not be accessible (Silverman and Imperiali, 2016). To

demonstrate that N-glycoproteins predicted using the bioinformatics pipeline were

modified in vivo, His-tagged versions of predicted proteins were introduced into C. fetus

wildtype and pglB::kn strains. Although the prediction pipeline utilised the genome of C.

fetus fetus 82-40, this strain was not available and thus C. fetus fetus 10842 was utilised

instead.

6.2.2.1 Proof of principle validation of experimentally identified C. fetus N-glycoprotein,

Cf0781

As a proof of principle, a His-tagged version of a known C. fetus N-glycoprotein, Cf0781

(Gene ID: CFF8240_0781), was introduced into C. fetus wildtype and pglB::kn strains.

This protein was chosen as it was predicted to be periplasmic and contained only a single,

experimentally confirmed sequon (Nothaft et al., 2012). cf0781 was amplified from C.

fetus 10842 gDNA using primer pair 1567/1572 (Table 2.1), adding a 5’ BamHI site and

3’ Octa-His tag and SmaI site. An ~ 0.8 kb product was confirmed using agarose gel

electrophoresis (not shown). The digested product was ligated to BamHI- and SmaI-

digested pRYGG1 to give p0781 and transformed into E. coli XL-1 Blue. The sequence of

the octa-his-tagged version of Cf0781 in p0781 is shown in Figure 6.3A. Positive clones

were confirmed by cPCR using primers 1445 and 1572. The sequence of cf0781 within

p0781 was checked using primers 1445 and 1432. Plasmid p0781 was then transformed

into E. coli S17 λpir cells in preparation for conjugation into C. fetus wildtype and

pglB::kn strains. Primers 1432 and 1445 were also used to identify positive E. coli S17

clones and then C. fetus transformants. To confirm production of this protein, a western

blot containing whole cell lysates of C. fetus wildtype and pglB::kn strains with and

without p0781 was probed with anti-His(6) antibodies (Fig. 6.3B).

No His-reactive proteins were detected in the absence of p0781 (Fig. 6.3B). Upon

introduction of the plasmid into wildtype C. fetus, an anti-His6 reactive protein was

detected at around 27 kDa. This is consistent with the predicted size, ~ 27.4 kDa, of

(8)His-tagged Cf0781 with a single N-glycan modification. In the presence of p0781 but

without pglB, an anti-His6 reactive protein was observed at around 26 kDa, which

179

correlates with the predicted size of 26.4 kDa for unglycosylated Cf0781. These data

confirm that Cf0781 is modified in a C. fetus PglB dependent manner.

180

6.2.2.2 Validation of a newly predicted N-glycoprotein, Cf0445

A His(8)-tagged version of a protein, Cf0445, predicted to be an N-glycoprotein using the

bioinformatics approach described in section 6.2 was introduced into C. fetus. Cf0445 is a

relatively small predicted periplasmic protein containing a cytochrome c553 domain and a

single predicted N-glycosylation sequon.

cf0445 (Gene ID: CFF8240_0445) was amplified from C. fetus 10842 gDNA using primer

pair 1566/1573 (Table 2.1), adding a 5’ BamHI site and 3’ Octa-His tag and SmaI site. An

~ 0.5 kb product was confirmed using agarose gel electrophoresis (not shown). The

digested product was ligated to BamHI- and SmaI-digested pRYGG1 to give p0445 and

transformed into E. coli XL-1 Blue. The sequence of the octa-his-tagged version of

Cf0445 in p0445 is shown in Figure 6.4A. Positive clones were confirmed by PCR using

primers 1445 and 1573. The sequence of cf0445 was confirmed using primers 1445 and

1432. Verified plasmid was transformed into E. coli S17 λpir cells in preparation for

conjugation into C. fetus wildtype and pglB::kn. All subsequent cPCR used to confirm

positive clones utilised primers 1432 and 1445. To confirm production of the protein, a

western blot containing whole cell lysates of C. fetus wildtype and pglB::kn strains with

and without p0445 was probed with anti-His(6) antibodies (Fig. 6.4B).

No reactivity was identified in C. fetus cells in the absence of p0445 (Fig. 6.4B). In the

presence of p0445, an anti-His(6)-reactive protein was produced, with apparent sizes of

around 19 and 17.5 kDa in wildtype and pglB::kn C. fetus cells, respectively. These likely

correspond to unmodified (predicted 17.1 kDa) and g1 form (predicted 18.3 kDa) of His-

tagged Cf0445. Therefore, these data demonstrate that Cf0445 is indeed modified in a C.

fetus PglB-dependent manner.

181

182

6.3 Analysis of the C. fetus predicted N-glycoproteome

6.3.1 Subcellular locations of predicted C. fetus N-glycoproteins.

Following the methodology outlined in Figure 6.2, N-glycoproteins were assigned a

predicted subcellular localisation. This bioinformatics pipeline previously found that the

majority of C. jejuni experimentally identified and predicted N-glycoproteins were

designated as inner membrane proteins (53.2%), followed by the outer membrane (9.1%),

periplasmic (6.5%) and extracellular (0.6%) (Frost, 2015). Similar analyses were

performed on C. fetus proteins to identify the predicted subcellular localisations of N-

glycoproteins experimentally identified by Nothaft et al. (2012), the predicted total N-

glycoproteome (described in 6.2.1) and the total set of C. fetus proteins predicted to

extracytoplasmic (Fig. 6.5).

Analysis of the 25 experimentally identified C. fetus N-glycoproteins revealed that the

largest proportion, consisting of nine proteins, was those predicted to be extracytoplasmic

but not confidently assigned a specific location (Fig. 6.5A, ‘unknown extracytoplasmic’).

In addition, five proteins had no indication of residing outside the cytoplasm and thus

were predicted as cytoplasmic (Fig. 6.5A). The remaining eleven proteins consisted of

four inner membrane, four outer membrane and three periplasmic proteins. No proteins

were predicted to be extracellular.

In comparison, analysis of the C. fetus predicted total N-glycoproteome (those predicted

and identified experimentally) (Fig. 6.5B) found that the largest proportion was predicted

to be inner membrane proteins (49.3%) and the smallest fraction was predicted to be

extracellular (4.5%). There were a considerable number (26.9%) of proteins assigned as

‘unknown extracytoplasmic’ due to contradictions between the three prediction tools

included in the pipeline. For example, 7 were categorised as cytoplasmic by PSORTb but

had a signal peptide and/or transmembrane (TM) region(s) predicted by LipoP and

TMMHM, respectively. The remainder of the proteins assigned as ‘unknown

extracytoplasmic’ were unclassifiable using PSORTb, but also displayed either a predicted

signal peptide or TM region. Furthermore, as above, there were five proteins (accounting

for 3.7% in this instance) assigned as cytoplasmic due to having no indication of being

extractyoplasmic.

These data reveal that outer membrane N-glycoproteins may be twice as abundant as

periplasmic N-glycoproteins in C. fetus, accounting for 10.4% and 5.2% of the predicted

183

total N-glycoproteome, respectively (Fig. 6.5B). This finding is not simply a result of their

relative abundance within the extracytoplasmic pool of proteins (Fig. 6.5C: 6.4% and

5.1% for OM and periplasmic, respectively). These observations are similar to what has

been observed in C. jejuni. Although Scott et al. (2011) found that, of the experimentally

identified N-glycoproteins, more were predicted to be periplasmic than outer membrane

proteins, combining these N-glycoproteins with those predicted by the bioinformatics

pipeline suggested there are potentially more outer membrane (8.9%) than periplasmic

(6.3%). The lower abundance of membrane proteins found experimentally may be because

membrane proteins are more challenging to extract and solubilise prior to mass

spectrometry analysis (Scott et al., 2011). In summary, the localisation of N-glycoproteins

in C. fetus appears to be similar to that of C. jejuni with predominantly inner membrane

proteins followed by outer membrane then periplasmic.

184

185

6.3.2 Sequon distribution in C. fetus predicted N-glycoproteins.

Previous analysis of sequon frequency within individual N-glycoproteins found that 85%

of C. jejuni 11168 predicted N-glycoproteins contain a single sequon (Frost, 2015) (Fig.

6.6A). Similarly to this, using the predicted total C. jejuni N-glycoproteome established

here, it was found that 72% of proteins harboured only a single sequon and 17% contained

two sequons (Fig. 6.6A). The most extensively glycosylated predicted N-glycoprotein

identified, Cj0633, contained eight sequons and Scott et al. (2014) have experimentally

identified two of these sequons in C. jejuni. Furthermore, eight glycoforms are observed

by western blotting when Cj0633 is expressed in glycocompetent E. coli (Frost, 2015). To

determine the distribution of sequons within experimentally identified and predicted C.

fetus N-glycoproteins, the frequency of sequons per protein was analysed.

Analysis of the 33 C. fetus N-glycosylation sequons experimentally identified by Nothaft

et al. (2012) demonstrated that 76% of N-glycoproteins contained a single modified

sequon and 20% contained two sequons. The highest number of modified sequons per

protein was four, which was evident in only a single protein. Similarly, examination of the

C. fetus predicted total N-glycoproteome found the majority of N-glycoproteins contained

one (64%) or two (25%) sequons (Fig. 6.6B). However, the remaining proteins had 3, 4, 6,

7 or 11 sequons. The protein with eleven potential N-glycosylation sequons,

CFF8240_0312, was experimentally determined as an N-glycoprotein by Nothaft et al.

(2012); however, the glycopeptides corresponded to only one of the possible eleven

sequons. This protein is predicted to reside in the outer membrane, and hence it could be

that the possible sites are within TM helices and not accessible for modification. However,

the TMHMM server predicts only a single TM helix encompassing residues 7 and 26, with

the remainder of the protein following this residing outside the cell. This suggests that as

no sequons lie within the TM region they still have the potential to be modified (Table

6.1).

In conclusion, the predicted N-glyoproteome of C. fetus has revealed that its N-

glycoproteins may be more extensively N-glycosylated than indicated by experimental

data. The pattern of sequon distribution was similar to that of C. jejuni. Strikingly, one

predicted N-glycoprotein, Cf0312, contained 11 possible sequons and is potentially the

most highly N-glycosylated protein identified in Campylobacter thus far.

186

187

6.3.3 Amino acid composition of sequons in C. fetus predicted N-glycoproteins.

The extended sequon, D/E-X-N-X-S/T (X ≠ P), is required for C. jejuni N-glycosylation

(Kowarik et al. 2006) and other Campylobacter species including C. fetus also appear to

follow this rule (Nothaft et al., 2012). Chen et al. (2007) proposed that DQNAT is the

optimum acceptor sequence for C. jejuni PglB using an in vitro assay. In agreement with

this, in vitro analysis of C. lari PglB found Thr to be favoured over Ser at the +2 position,

giving a higher rate of glycosylation (Gerber et al., 2013). Furthermore, analysis of N-

glycosylation sequons in all predicted and verified C. jejuni N-glycoproteins demonstrated

that Asp (D) is most common in the -2 position. However, Ser (S) is more abundant at the

+2 position (Frost, 2015). The amino acid composition of sequons within the C. jejuni

predicted total N-glycoproteome is shown in Figure 6.7A. To identify the most abundant

amino acids in C. fetus N-glycosylation sequons, the distribution of residues in both

predicted and known C. fetus N-glycoproteins was analysed.

Analysing sequon sequences from C. fetus N-glycopeptides identified by Nothaft et al.

(2012) revealed overrepresentation of Asp (94%) and Thr (78%) at the -2 and +2

positions, respectively (not shown). This trend is also observed when analysing sequons of

the predicted total N-glycoproteome, although the relative abundancies are not as marked,

with 70% Asp and 55% Thr at the -2 and +2 positions, respectively (Fig. 6.7B). Therefore,

these data suggest that, in agreement with in vitro Campylobacter PglB substrate

specificity studies, C. fetus N-glycosylation sequons preferentially consist of D-X-N-X-T

sequences.

188

189

6.4 Predicted conservation of N-glycoproteins within the

Campylobacter genus

6.4.1 Predicted conservation of N-glycoproteins between C. jejuni and C. fetus

Although many N-glycoproteins from C. jejuni and C. fetus have been experimentally

identified, the extent of conservation of N-glycosylation targets among these species is

unknown. As discussed in section 6.1, some evidence suggests that N-glycosylation is not

required for the function of specific target proteins and rather the modification may afford

a general advantage, such as improved resistance against proteolytic enzymes (Alemka et

al., 2013). Understanding the conservation of N-glycoproteins among Campylobacter

species may provide suggestive evidence of proteins which require N-glycosylation. To

this end, a sample of the core N-glycoproteins found in C. jejuni and C. fetus were

identified using a combination of the bioinformatics pipeline described above and the

protein BLAST® search tool (NCBI) (BLASTP).

Previous work identified the core N-glycoproteome of C. jejuni using experimentally

validated and predicted N-glycoproteins from four C. jejuni strains (Frost, 2015). This list

included 109 N-glycoproteins, 65 of which have been experimentally identified as N-

glycoproteins (Frost, 2015; Scott et al., 2014, 2011; Young et al., 2002). To identify if

these experimentally identified N-glycoproteins were conserved in C. fetus, the protein

sequences were queried against C. fetus fetus 82-40 using BLASTP to identify

homologues. Matches with an E value of 0.01 or higher were not designated as

homologues. Results from the N-glycoprotein prediction pipeline were then used to

identify if the homologues contained N-glycosylation sequons (Table 6.2). Seventeen of

the C. jejuni experimentally identified N-glycoproteins had no homologues in C. fetus

(Fig. 6.8). Of the 48 C. fetus proteins with homologues among the C. jejuni N-

glycoproteins, 34 contained potential N-glycosylation sequons, 10 of which have been

experimentally identified as sites of N-glycosylation in C. fetus (Nothaft et al. 2012, Table

6.2 experimentally identified in bold). Notably, every C. fetus protein homologue of C.

jejuni outer membrane proteins contained sequons. In contrast only two of the five C. fetus

homologues of C. jejuni periplasmic N-glycoproteins contained sequons. Therefore,

overall, 31 of the C. jejuni known N-glycoproteins did not have N-glycosylation sequon-

containing homologues in C. fetus.

190

Four of the C. fetus proteins identified in this comparison, Cf0214, Cf0318, Cf0928 and

Cf1178, were not within the predicted total C. fetus N-glycoproteome established in

section 6.2.1 as there was no indication of extracytoplasmic localisation. However, further

analysis of these proteins using CELLO predicted all except Cf0928 as outer membrane.

Cf0928 was predicted as either extracellular or outer membrane with similar probabilities.

Therefore, based on homology with known C. jejuni N-glycoproteins, this comparison

revealed four sequon-containing C. fetus proteins which may be N-glycosylated and hence

could be included in its predicted N-glycoproteome. This would increase the C. fetus

predicted total N-glycoproteome to 138 proteins.

In summary, of the 65 experimentally identified C. jejuni N-glycoproteins, 34 had

homologues which also contained N-glycosylation sequons in C. fetus, suggesting a

moderate level of conservation between the N-glycosylation protein targets of these

species. If excluding C. jejuni N-glycoproteins which had no significant protein

homologue in C. fetus, as many as 71% of N-glycoproteins were predicted to be modified

in both species. Furthermore, more C. jejuni outer membrane N-glycoproteins than

periplasmic N-glycoproteins were predicted to be conserved as N-glycoproteins in C.

fetus.

191

Table 6.2. Identification of N-glycosylation sequon-containing C.fetus homologues

of known C. jejuni N-glycoproteins. Colour represents subcellular location

assignment; Blue; inner membrane. Green; periplasmic. Red; Outer membrane. Purple;

extracellular. Grey; unknown extracytoplasmic. Light grey; assigned as cytoplasmic.

‘No homol’ indicates no homologous protein was identified in C. fetus. Bold indicates

the protein has been experimentally identified as an N-glycoprotein.

Locus Tag

(Name) # Sequon(s)

Sequon

position(s)

Identity

(coverage)

Locus

Tag

(Name)

# Sequon(s)

Sequon

position

(s)

Cj0011c 1 EANFT 49-53 50 (93) Cf0510 0 - -

Cj0017c

(dsbI) 2

EINKT DNNFS

3-7

290-294 45 (99) Cf1646 0 - -

Cj0081

(cydA) 2

DNNES

ENNDT

283-287

351-355 no homol

Cj0089 1 DFNKS 73-77 no homol

Cj0114 4

ENNFT

DANLS

DSNST

ENNNT

99-103

153-157

171-175

177-181

41 (99) Cf1518 2 DLNLT

EKNSS

104-108

153-157

Cj0131 1 DDNTS 73-77 52(93) Cf1499 1 EFNDS 83-87

Cj0143c

(znuA) 1 EQNTS 26-30 44 (98) Cf1731 0 - -

Cj0152c 6

EQNNT

ETNRT DKNIS

ENNIS

ENNTT

DFNIS

126-130

163-167

182-186

188-192

193-197

250-254

32 (91) Cf0290 7

DKNNS

ESNMT

DENVT

DSNLS

DENES

DINLS

DKNTT

101-105

124-128

136-140

141-145

146-150

220-224

242-246

Cj0158 2 EKNIS

DKNHS

24-28

119-123 no homol

Cj0168c 1 DVNQT 26-30 no homol

Cj0176c 1 DLNKT 29-33 no homol

Cj0182 3 DSNST

ENNAT

EKNIS

58-62

70-74

394-398

61(84) Cf0147 0 - -

Cj0200c 1 DNNKT 33-37 no homol

Cj0235c

(secG) 2

ENNNT

DVNSS

87-91

118-122 63 (71) Cf1783 0 - -

Cj0238 2 DANIS

DENSS

24-28

56-60 42 (94) Cf1737 1 DKNAS 45-49

Cj0256

(eptC) 1 ENNHT 213-217 48 (96) Cf1723 1 DENLS 474-478

Cj0277

(mreC) 1 DQNST 91-95 44 (92) Cf0318 1 DKNST 75-79

Cj0289c

(peb3) 1 DFNVS 88-92 no homol

Cj0313 2 DLNLS

DGNIT

173-177

196-200 47 (100) Cf0272 1 ELNAT 207-211

Cj0365c

(cmeC) 2

EANYS

ENNSS

30-34

47-51 36 (96) Cf0117 1 EQNAS 354-358

Cj0366c

(cmeB) 2

DRNVS DRNAS

634-638

653-657 61 (99) Cf0118 1 EANST 638-642

Cj0367c

(cmeA) 2

DFNRS DNNNS

121-125

271-275 45 (95) Cf0119 2

DYNGT

ENNST

247-251

273-277

Cj0371 1 DLNGT 75-79 53 (93) Cf1659 0 - -

192

Cj0397c 1 DFNNT 105-109 37 (98) Cf1147 1 DSNNT 104-108

Cj0399 1 DLNNT 179-183 45 (100) Cf1144

(cvpA) 2

DLNRT

DSNET

160-164

180-184

Cj0454c 1 ENNKS 91-95 32 (90) Cf1372 1 ELNVT 121-125

Cj0494 1 DNNIT 26-30 no homol

Cj0508 1 DANLS 312-316 60 (97)

Cf1032 0 - -

Cj0511 1 DQNIS 67-71 66 (99) Cf0944 2 DQNFT

DGNST

64-68

145-149

Cj0515 3 DFNIS ELNAT

DFNAS

103-107

207-211

234-238

38 (90) Cf0948 2 DLNLT

EANST

96-100

130-134

Cj0530 5

DGNLS

DFNAS DFNIT

DFNAS

DSNKT

156-160

180-184

389-393

519-523

617-621

28 (77) Cf0928 8

DYNIS

DSNLT

DINDT

DLNIT

DANSS

DLNRT

DGNFT

DINMT

2-6

164-168

260-264

272-276

391-395

416-420

478-482

578-582

Cj0587 1 DNNLS 282-286 no homol

Cj0592c 4

DINQS

ENNES

ENNQS

DVNMT

97-101

103-107

127-131

137-141

no homol

Cj0599 3 EANIT

DLNST

DNNIT

97-101

109-113

168-172

50 (99) Cf0517 2 DINST

DLNST

111-115

157-161

Cj0608 2 DLNLT DANNT

35-39

412-416 36 (97) Cf0760 1 DFNQT 38-42

Cj0610c 5

DENLS DENTS

DANIS

ENNRS

EENAS

82-86

98-102

113-117

296-300

331-335

no homol

Cj0633 8

DNNKS

DLNTS

EQNLS

EQNLS

EQNQS

DTNLT

DQNLT

DKNHS

73-77

90-94

97-101

104-108

110-114

123-127

129-133

180-184

39 (94) Cf1126 6

DKNIS

DKNIS

DSNIS

DKNIT

DQNIT

DENSS

69-73

82-86

97-101

104-108

133-137

141-145

Cj0648 3 ESNTS

DFNLS

EGNVT

49-53

82-86

103-107

31 (95) Cf1187 2 ELNST

DDNIT

41-45

90-94

Cj0652

(pbpC) 3

DRNGT

DLNAS

ENNNT

55-59

99-103

467-471

54 (99) Cf1182 1 DLNVT 96-100

Cj0694 3 DFNKT

DQNIS DQNSS

132-136

306-310

426-430

42 (99) Cf0569 2 DVNDT

ESNAT

239-243

297-301

Cj0734c

(hisJ) 1 ESNAS 27-31 48 (89) Cf1232 0 - -

Cj0776c 3

DENQS

ENNQS

DTNTS

87-91

103-107

111-115

no homol

193

Cj0783

(napB) 1 EANFT 48-52 44 (97)

Cf1158

(napB) 2

DKNKS

DSNES

24-28

162-166

Cj0843c 4

DANLT

DANAS

DYNKT

DLNKS

97-101

327-331

374-378

537-541

47 (95) Cf0107 3 DANAT

DTNDT

DNNIT

81-85

136-140

159-163

Cj0846 2 DFNST

DLNTS

208-212

280-284 39 (98) Cf1153 0 - -

Cj906c 2 DKNIS

DINIS

53-57

273-277 47 (99) Cf1582 0 - -

Cj0944c 2 ENNAS

DENST

219-223

238-242 44 (47) Cf0574 0 - -

Cj0958c

(YidC) 2

EQNIT

DENGS

40-44

154-158 58 (97) Cf0553 2

DINQT

EQNSS

36-40

45-49

Cj0982c

(cjaA) 1 DSNIT 137-141 30 (66) Cf1233 0 - -

Cj0983

(jlpA) 1 DINAS 144-148 no homol

Cj1007c 1 DVNRT 17-21 51 (95) Cf0855 1 DINSS 172-176

Cj1013c 3 ENNNS

DENLT

DLNST

178-182

230-234

530-534

46 (98) Cf0404 1 DVNNT 636-640

Cj1032

(cmeE) 1 DQNGT 199-203 no homol

Cj1053c 2 DINVS DNNQS

75-79

96-100 no homol

Cj1126c

(pglB) 1 DYNQS 532-536 46 (97)

Cf1383

(pglB2) 2

DYNAS

DQNFT

580-584

595-599

Cj1219c 1 DVNIT 47-51 34 (94)

Cf0699

6

ELNLT

DLNYT

DFNIS

DANAS

DVNVT

DENKT

68-72

85-89

206-210

431-435

764-768

831-835

Cj1345c 4

DDNNT

DYNIT

EINAS

DGNET

31-35

59-63

155-159

348-352

50 (97) Cf0214

5

DDNNT

DYNIT

DQNVS

EYNAS

DGNET

22-26

50-54

106-110

290-294

338-342

Cj1373 2 DINRT DQNAS

134-138

497-501 48 (98) Cf0367 2

DINKT

DKNGT

131-135

185-189

Cj1444c

(kpsD) 2

DQNLS

ENNLT

37-41

50-54 no homol

Cj1496c 2 EVNAT

DNNAS

71-75

167-171 41 (97) Cf0372 2

DVNRT

DNNSS

71-75

171-175

Cj1565c

(pflA) 3

DKNFS

DNNAS

EGNFS

250-254

456-460

495-499

29 (98) Cf0217 2 DANYS

DLNDT

262-266

498-502

Cj1621 1 DLNKT 197-201 34 (100) Cf1768 0 - -

Cj1661 1 ENNQS 188-192 52 (99) Cf0527 0 - -

Cj1670c

(cgpA) 4

DQNIT

DVNKS

EKNSS ESNST

26-30

71-75

104-108

111-115

40 (99) Cf0198 2 DINST

DINGS

26-30

78-82

194

6.4.2 Predicted conservation of N-glycoproteins among further Campylobacter species

Section 6.4.1 identified 34 experimentally identified C. jejuni N-glycoproteins which also

have predicted N-glycoprotein homologues in C. fetus. To identify if these putatively

conserved N-glycoproteins were also found in other Campylobacter species, the same

methodology was applied to species closely related to C. jejuni and C. fetus: C. coli

RM4661, C. lari RM2100, C. hyointestinalis LMG 9260 and C. concisus 13826. The

findings are displayed in Table 6.3 and the percent identity between each C. jejuni known

N-glycoprotein and its homologue is shown. Precent identity was found using BLAST®

global alignment tool.

Nineteen of the thirty-four C. jejuni N-glycoproteins had homologues containing N-

sequons in all Campylobacter species analysed (Table 6.3). As previously, this conserved

set contained mostly predicted inner membrane proteins. Three were predicted as outer

membrane proteins and none were predicted to reside in the periplasm. One N-

glycoprotein (Cj0843c) did not have a homologue in C. coli but in all other species there

were N-sequon-containing homologues. The remaining fourteen proteins had homologues

in all species tested, but N-sequons were not identified in every homologue. Eleven

proteins lacked an N-sequon-containing homologue in a single species whereas three

proteins lacked N-sequon-containing homologues in two species.

No overall pattern in the presence/absence of N-sequons was immediately apparent and

the conservation of sequons did not appear to correlate with percentage identity. For

example, homologues of Cj0530 had low percentage identity (as low as 23%) but

contained sequons whereas the C. coli homologue of NapB has high (84%) identity but

lacks a sequon. Therefore, it appeared that the observed conservation of N-glycosylation

sequon occurrence was independent of amino acid sequence identity.

In conclusion, there appears to be moderate conservation of proteins predicted to be N-

glycosylated between different Campylobacter species. Nineteen C. jejuni N-

glycoproteins had putative N-glycoprotein homologues in the five Campylobacter species

tested. This suggests that these proteins may be ‘core’ Campylobacter N-glycoproteins

and make promising candidates for experimentally addressing if N-glycosylation can

influence protein function.

195

6.4.2.1 N-glycosylation of components of the CmeABC multidrug efflux system is

predicted to be conserved in several Campylobacter species

The C. jejuni CmeABC multidrug efflux system (Lin et al., 2002) is a known colonisation

factor in chickens (Hermans et al., 2011). Mutation of the CmeABC locus or its regulator,

CmeR, can increase the susceptibility of C. jejuni to bile salts and a variety of

antimicrobials (Lin et al., 2002). Moreover, a CmeB mutant is unable to colonise chickens

(Lin et al., 2003). All three components have been experimentally identified as N-

glycoproteins in C. jejuni (Scott et al., 2011; Wacker et al., 2002). CmeA (also known as

AcrA) is the most extensively studied and is often used as a ‘model’ N-glycoprotein

(Nothaft et al., 2010). It has been reported that abolishing CmeA N-glycan modification

confers reductions in antimicrobial resistance and chicken colonisation similar to a cmeA

knockout, suggesting modification significantly influences its function (Nothaft and

Szymanski, 2013). In addition, Alemka et al. (2013) described data which suggests N-

glycosylation of CmeA provides resistance from proteolysis. However, both these data

sets remain unpublished.

N-glycosylation sequon-containing homologues of CmeA and CmeC were identified in C.

fetus, C. coli, C. lari, C. concisus and C. hyointestinalis. However, N-sequon-containing

homologues of CmeB were only identified in C. coli and C. lari, not C. concisus or C.

hyointestinalis. Further analysis of the predicted sequons present in the identified

homologues revealed that in other Campylobacter species, components were often

predicted to contain only a single sequon. However, although each C. jejuni component

has two predicted N-glycosylation motifs, only one sequon has been experimentally

identified from each component. This may explain why only one motif is conserved per

protein in other species.

In conclusion, N-glycan modification is a putatively conserved feature of the CmeABC

system in Campylobacter species, with CmeA and CmeC being predicted N-glycoproteins

in the further five species studied. CmeB is conserved as a sequon-containing predicted N-

glycoprotein in three of the five species studied. Therefore, this bioinformatics analysis

further suggests that N-glycosylation may be important for the CmeABC multidrug efflux

system.

196

Table 6.3 Putative conservation of experimentally identified C. jejuni N-

glycoproteins in Campylobacter species. Background colour of C. jejuni N-

glycoproteins represents subcellular location assignment; Blue; inner membrane.

Green; periplasmic. Red; Outer membrane. Purple; extracellular. Grey; unknown

extracytoplasmic. Light grey; assigned as cytoplasmic. No colour indicates that

the homologue is also a predicted N-glycoprotein. Dark grey background

indicates that the homologue does not contain a sequon. The percentage identity is

shown for each homologue (identified using BLAST® global alignment). A cross

indicates there is no significant homologue. Bold and underlined indicates the

homologue has been experimentally identified as an N-glycoprotein in C. fetus

(Nothaft et al., 2012).

Predicted N-glycoprotein homologue in Campylobacter

species

C. fetus C. coli C. lari C. hyo C. concisus

C. je

jun

i p

rote

in n

ame

CmeB 60 98 75 59 59

NapB 44 84 65 45 49

Cj0131 50 99 60 50 50

Cj0114 40 75 52 40 36

EptC 47 75 51 46 22

MreC 41 75 54 44 44

Cj0313 47 78 54 46 47

Cj0399 45 81 59 44 41

Cj0454c 29 59 31 27 26

Cj0515 35 64 45 35 34

Cj0843c 45 54 43 41

Cj1013c 46 79 63 46 47

PglB 45 81 56 45 47

Cj1219c 33 72 43 32 31

Cj1373 47 83 60 45 46

Cj0152c 29 62 38 31 30

Cj0238 41 84 53 42 39

CmeC 35 88 54 37 33

CmeA 43 88 56 43 43

Cj0397c 37 65 43 40 39

Cj0511 65 89 67 64 64

Cj0530 24 69 39 30 31

Cj0599 49 91 63 49 42

Cj0608 35 79 49 33 33

Cj0633 36 62 45 38 30

Cj0648 30 75 42 29 32

PbpC 54 82 60 54 54

Cj0694 42 73 48 42 40

Cj0958c 56 81 68 56 56

Cj1007c 49 83 56 50 45

Cj1345c 48 91 66 49 46

Cj1496c 36 88 59 38 38

PflA 28 80 47 31 31

CgpA 39 78 43 37 28

197

6.5 Discussion

This chapter described the in silico prediction of the total C. fetus N-glycoproteome and

investigated the conservation of N-linked glycosylation sequon occurence among proteins

of several Campylobacter species. By combining an available experimental data set

(Nothaft et al., 2012) and that of a bioinformatics pipeline, a total C. fetus predicted N-

glycoproteome containing 134 proteins was identified. However, the bioinformatics

pipeline only predicted 80% of the experimentally identified C. fetus N-glycoproteins,

highlighting flaws in its ability to correctly predict the subcellular location of proteins.

Further analysis of these N-glycoproteins not identified by the pipeline using online

prediction servers, CELLO and SignalP, identified them as extracytoplasmic. Therefore, it

would improve sensitivity of the pipeline if these tools were integrated in the future.

Nevertheless, these data suggest that N-glycan modification is a significant feature of C.

fetus, occurring to a similar extent as in C. jejuni, which has 148 predicted N-

glycoproteins (Frost, 2015). These figures correspond to 7.8 % and 9.5 % of total proteins

predicted in C. fetus and C. jejuni, respectively. These results correlate well with the

results of probing C. fetus lysates with C. fetus N-glycan-reactive antibodies, which

previously indicated the presence of a multitude of N-glycoproteins (Nothaft et al. 2012,

Chapter 4). Therefore, this bioinformatics analysis further suggests that Campylobacter

species N-glycosylate a considerable number of proteins. However, despite significant

advances in understanding the mechanism of N-linked glycosylation in Campylobacter,

the question as to why Campylobacter proteins are N-glycosylated remains unanswered.

Furthermore, whether N-glycan modification is required for the function of specific

proteins or serves a more general role remains unclear. Only limited experimental data

exists surrounding these questions and further investigation to gain a more holistic view of

Campylobacter N-glycosylation may provide insight into these questions. Towards this

end, data gained using the N-glycoprotein prediction pipeline was analysed in further

detail.

Further analysis of the C. fetus predicted N-glycoproteome found that most were predicted

to reside in the inner membrane, which is also true for predicted C. jejuni N-glycoproteins

(Frost, 2015). Predicted outer membrane N-glycoproteins were twice as abundant as

predicted periplasmic N-glycoproteins, although total numbers of predicted outer

membrane and periplasmic proteins are more similar. Although predicted outer membrane

N-glycoproteins were also more (1.4X) abundant than periplasmic in C. jejuni (Frost,

198

2015), the difference is more marked in C. fetus. As N-glycosylation has been reported to

confer resistance against chicken gut protease activity (Alemka et al., 2013), this relatively

higher abundance of predicted outer membrane N-glycoproteins may reflect a preferential

requirement for surface-exposed proteins to be modified with N-glycan to provide such

protection.

The distribution of sequons among N-glycoproteins was also investigated. Using the

experimentally identified N-glycosylation sequons identified from C. fetus, sequon

frequency per protein was between 1 and 4, with 76% of the identified N-glycoproteins

harbouring a single sequon. Analysis of N-glycosylation motifs identified in the C. fetus

predicted N-glycoproteome found that although most (64%) proteins were still predicted

to contain a single site, a larger proportion (25%) was predicted to be modified at two

sites. Furthermore, proteins with between 1 and 11 potential modification sites were

identified. Structural prediction of the N-glycoprotein with the most predicted sequons

(Cj0312) suggested the 11 sequons reside outside the cell. Importantly, one of these

sequons was experimentally identified in a glycopeptide by Nothaft et al. (2012),

providing evidence that it is indeed an N-glycoprotein. Extensive N-glycosylation has

been documented in a C. jejuni protein, Cj0633, which has 8 modification sites that can be

modified in the available E. coli Pgl system (Frost, 2015). In the future, it would be

interesting to confirm if Cf0312 is indeed N-glycosylated to such a high extent, and if so,

would make it the most extensively N-glycosylated protein reported in ε-proteobacteria

thus far. It would also be the N-glycoprotein containing the second-most modification

sites in Bacteria overall, with HMW1 of Haemophilus influenzae having 31 sites modified

with mono or di-hexoses (Gross et al., 2008). Furthermore, it may be that such high level

N-glycosylation protects this C. fetus protein from degradation by host gut proteases, as

suggested in C. jejuni (Alemka et al., 2013).

The amino acid composition of sequons from experimentally identified C. fetus N-

glycoproteins was analysed, finding that the most common motif was D-X-N-X-T. This

motif was also most abundant in the sequons of the total C. fetus predicted N-

glycoproteome. This is in agreement with in vitro assays investigating the activity of C.

jejuni and C. lari PglB enzymes, which found that D and T at the -2 and +2 positions,

respectively, confer the highest efficiency of glycosylation (Gerber et al., 2013; Kowarik

et al., 2006b). However, this finding is in contrast to the composition of sequons in C.

jejuni, where Ser is more prominent at the +2 position when analysing experimentally

identified N-glycoproteins or the total predicted set (Frost, 2015). These findings suggest

that D-X-N-X-T may be the preferred N-glycosylation sequon in C. fetus. However, as

199

these data rely mostly on predicted and not proven N-glycosylation sequons, they must be

taken with caution and further experimental validation is required. Futhermore, it may be

that the most efficient N-glycosylation sequons are not the most abundant.

Although the bioinformatics pipeline predicted numerous N-glycoproteins in C. fetus, the

presence of an extended D/E-X-N-X-S/T sequon is not sufficient for glycosylation.

Although it was originally thought that N-glycosylation sites had to reside within flexible

loops in order to be modified in C. jejuni (Kowarik et al. 2006), more recent evidence

suggests that this is not the case. The crystal structures of C. jejuni N-glycoproteins JlpA

and PEB3 revealed N-glycosylation sites present in an alpha-helix and structured turn,

respectively (Kawai et al., 2012; Rangarajan et al., 2007). Silverman & Imperiali (2016)

have since demonstrated that N-linked glycosylation is coupled with protein translocation

mediated by the Sec pathway and hence N-glycoproteins can be modified prior to

complete folding. Nevertheless, although it appears that PglB can modify regions within a

variety of protein structures, if sequons reside within buried structural features they may

not be modified (Silverman and Imperiali, 2016). Therefore, some of the predicted N-

glycosylation sequons identified in this chapter may not be accessible. Nevertheless, His-

tagged versions of two predicted N-glycoproteins identified by the workflow were shown

to be modified by PglB in C. fetus. Glycopeptides from the first protein, Cf0718, had been

previously identified by Nothaft et al. (2012) and hence this served as a proof of principle

to demonstrate that N-glycosylation could be successfully validated using this approach.

Secondly, PglB-mediated modification of a previously unknown predicted N-glycoprotein,

Cf0445, was demonstrated. This provides some evidence towards the accuracy of the N-

glycoprotein predictions of the bioinformatics pipeline; however, more extensive

experimental validation/identification is required.

Limited experimental evidence has demonstrated that N-glycosylation is not required for

the function of individual proteins in C. jejuni. Some evidence suggests that N-

glycosylation instead serves a more general role and provides protection from chicken gut

proteases (Alemka et al., 2013). However, it could be that particular proteins have a

stronger requirement for N-glycosylation to provide such protection than others. For

example, N-glycosylation may be particularly advantageous for surface-exposed proteins.

Nevertheless, it can be argued that if a protein requires N-glycosylation, whether it is for

protein function or protection, there would be an evolutionary pressure to maintain this

modification. Therefore, understanding the conservation of N-glycoproteins among

Campylobacter species may provide suggestive evidence of proteins which require N-

glycosylation to function and contribute to uncovering the role of Campylobacter N-linked

200

glycosylation. Towards this end, the conservation of sequon occurrence for experimentally

validated C. jejuni N-glycoproteins which appear to be core N-glycoproteins for this

species was analysed. Thirty-four of sixty-five experimentally identified C. jejuni N-

glycoproteins were predicted to be N-glycosylated in C. fetus, and ten of these were

previously experimentally identified as C. fetus N-glycoproteins (Nothaft et al., 2012).

Interestingly, more outer membrane proteins were conserved as predicted N-glycosylation

targets than periplasmic proteins. This finding may reflect the need for certain surface-

exposed proteins to be N-glycosylated to provide protection from protease activity. Some

of the C. fetus homologues identified were not characterised as predicted N-glycoproteins

by the bioinformatics workflow but did contain N-glycosylation sequons. This issue was

alleviated by manual analysis using CELLO which predicted the proteins were

extracytoplasmic, further highlighting that improvements to the localisation predictive

ability of the bioinformatics workflow could be made. Therefore, this analysis also

identified another four C. fetus predicted N-glycoproteins, increasing the total predicted N-

glycoproteome to contain 138 proteins. The conserved C. jejuni proteins cover a range of

functions, including motility (PflA) (Yao et al., 1994), drug efflux (CmeABC)(Lin et al.,

2002), post-translational modification (EptC) (Scott et al., 2012), nitrate reduction (NapB)

(Pittman et al., 2007) and cell shape determination (MreC) (Divakaruni et al., 2007).

These data suggested that the targets of N-glycan modification may be moderately

conserved in Campylobacter and warranted investigation into conservation among other

species.

Further analysis identified that of the 34 C. jejuni N-glycoproteins with homologues

predicted to be N-glycosylated in C. fetus, 19 of these were also predicted to be N-

glycoproteins in another four Campylobacter species (C. coli, C. lari, C. hyointestinalis

and C. concisus). The conservation of N-glycosylation sequons did not appear to strictly

correlate with high percentage identity homologues which suggested N-glycan

modification may be required for the function of certain proteins. However, it may be that

N-glycosylation serves a general role, such as protection from proteases (Alemka et al.,

2013), and that certain proteins have a stronger requirement for this. For example, more

surface exposed proteins may benefit from such protection more than proteins present

inside the cell. In agreement with this, there were generally more predicted outer

membrane N-glycoproteins than periplasmic and more outer membrane than periplasmic

N-glycoproteins were putatively conserved among species. However, an argument against

this theory is that the majority of N-glycoproteins are not surface exposed as most are

201

predicted to be inner membrane proteins. Therefore, it is clear that further experimental

study into the role of N-glycosylation in protein function is warranted.

Interestingly, this analysis found that ZnuA and JlpA are not conserved as N-glycoproteins

throughout these Campylobacter species. This is consistent with experimental evidence

suggesting that N-glycosylation is not required for their function in C. jejuni (Davis et al.,

2009; Scott et al., 2009). Therefore, it can be speculated that these proteins are not

representative in such studies, and that further analysis of specific conserved N-

glycoproteins needs to be undertaken. For example, the conserved proteins included

CmeA and CmeC from the tripartite drug efflux system, CmeABC (Lin et al., 2002), and

PflA (Yao et al., 1994), a protein associated with motility in Campylobacter. It would be

interesting to identify if N-glycosylation is essential for the function of these

Campylobacter proteins in particular.

In summary, the complete C. fetus predicted N-glycoproteome containing 138 candidate

targets was identified. The general features of this N-glycoproteome bear similarity with

the predicted N-glycoproteome of C. jejuni, containing largely inner membrane proteins

followed by outer membrane and periplasmic proteins. The analysis revealed an N-

glycoprotein with 11 potential N-glycosylation sites which, if fully modified, would make

it the most extensively N-glycosylated protein in ε-proteobacteria to date. Two of the

predicted N-glycoproteins were demonstrated to be modified by C. fetus PglB in vivo,

providing limited evidence of the robustness of this predictive approach. Lastly,

conservation of predicted N-glycoproteins across Campylobacter species was investigated.

Of the 65 experimentally confirmed C. jejuni N-glycoproteins thought to be part of the

‘core’set for the species, 34 had homologues predicted to be N-glycosylated in C. fetus.

Further analysis revealed that 19 are also predicted to be N-glycoproteins in C. coli, C.

lari, C. hyointestinalis and C. concisus, suggesting some conservation of N-glycoproteins

between several Campylobacter pathogens covering broad host niches. Interestingly, this

set did not include N-glycoproteins for which N-glycan modification is not thought to be

responsible for function. Instead, it may be that the putatively conserved N-glycoproteins

identified here have a requirement for N-glycosylation. These findings inform the wider

question of what is/are the function(s) of the N-glycosylation of so many Campylobacter

proteins and identify conserved N-glycoproteins for which further study is attractive.

202

Chapter 7. Conclusion and future work

203

7.1 N-linked glycoprotein glycans as targets for antibody-based

detection/identification

One focus of this thesis was assessing Campylobacter N-linked glycans as targets for

antibody-based detection of Campylobacter species. Chapter 3 described the

characterisation of an antiserum, CjNgp, raised against a recombinant C. jejuni N-

glycoprotein, NGRP. This antiserum reacted with numerous N-glycoproteins in many C.

jejuni strains, including strains recently isolated from chicken meat. CjNgp also reacted

with the N-glycoproteins of C. coli, C. upsaliensis, C. helveticus and, to a lesser extent, C.

lari. CjNgp could also detect untreated cells of these Campylobacter species. The

antiserum contained antibodies reactive with both the carrier protein, NGRP, and the

associated N-linked glycan, but it was demonstrated that both types of antibodies

interacted with C. jejuni cells. Furthermore, flow cytometry demonstrated that CjNgp

labelled whole cells of C. jejuni, although to a lesser extent that an anti-whole cell C.

jejuni antiserum. One advantage of CjNgp over the whole cell antiserum was that it does

not cross react with C. fetus. Therefore, CjNgp shows promise as a tool for detecting both

the common human Campylobacter pathogens (C. jejuni and C. coli) and also emerging

pathogens found in poultry (C. upsaliensis and C. lari) without cross-reacting with other

Campylobacter species such as C. fetus and C. hyointestinalis. However, these species are

not found in chicken and would not usually pose a cross-reactivity risk. Instead, these

species can be found in cattle and swine and therefore this suggests that a detection test

using CjNgp could also be used to identify Campylobacter species such as C. jejuni and C.

coli from cattle and pigs without producing false-positives with species such as C. fetus

and C. hyointestinalis, respectively. For example, although poultry is the main reservoir

for human campylobacteriosis (EFSA, 2010), one study found that over 70% of beef liver

products are infected with Campylobacter (Noormohamed and Fakhr, 2013). Therefore,

such a tool for detecting C. jejuni, C. coli and emerging Campylobacter pathogens from

other farmed animals could also be useful.

Chapter 4 described the development and characterisation of a novel antiserum, CfNgp,

raised against a recominant N-glycoprotein containing C. fetus N-linked glycan. This

involved expression of the recombinant N-glycoprotein to be used as immunogen, NGRP,

in C. fetus and subsequent purification. The resulting polyclonal antiserum was reactive

with N-linked glycoproteins of C. fetus fetus, C. fetus venerealis and a variety of C. fetus

clinical isolates. CfNgp displayed broad reactivity with Campylobacter species,

204

particularly with C. hyointestinalis, C. consisus, C. sputorum and C. lanienae. This

antiserum was also able to bind Campylobacter cells. Protein-reactive antibodies within

the antiserum partly contributed to this ability. Overall, these data suggest that CfNgp

could be utilised to detect C. fetus and also less common Campylobacter pathogens, such

as C. hyointestinalis and C. concisus. Notably, CfNgp was similarly reactive with C. fetus

and C. sputorum. This cross-reactivity poses a significant disadvantage to using this

antiserum for detecting C. fetus in cattle as C. sputorum is found in cattle but does not

cause disease in these animals. However, species such as C. sputorum and C. concisus can

cause gastroenteritis in humans (Lindblom et al., 1995) and therefore it appears this

antiserum could be better suited for detection of Campylobacter species from human

samples. Furthermore, Campylobacter species including C. concisus and C. ureolyticus

are particularly difficult to isolate and culture in the laboratory and such a detection test

could improve detection of these species. However, as the previously described C. fetus N-

glycan-reactive GRP-II antisera display very little or no reactivity against species such as

C. concisus and C. sputorum (Nothaft et al., 2012), it can be argued that, overall, C. fetus

N-linked glycan is still a potential target for antibody-based detection of C. fetus in cattle.

This research instead highlights that N-glycan-reactive antibodies with stricter specificity

would be favourable for such a purpose.

One significant feature of both antisera is the presence of protein-reactive antibodies,

particularly those reactive with NGRP and its native progenitor, Cj0114. These antibodies

influence the ability of CjNgp and CfNgp to bind cells of C. jejuni and C. fetus,

respectively. There are homologues of Cj0114 in C. coli, C. upsaliensis, C. lari and

C. fetus. In light of this, the CjNgp cell binding reactivity observed with these species

could be markedly influenced by NGRP reactivity within the antiserum. Similarly, such

reactivity within CfNgp may influence the observed cell binding of C. jejuni and others

tested here. Therefore, such protein reactivity is a potential issue which needs to be

addressed further. It would be interesting to purify the N-glycan-reactive antibodies from

these antisera and analyse if their Campylobacter specificity is altered. Furthermore,

purification and concentration of N-glycan-reactive antibodies may make the reagents

more reactive towards target Campylobacter cells. For example, carrying this out with

CjNgp could produce a reagent which is more effective at labelling C. jejuni cells than the

anti-whole cell C. jejuni 11168 antiserum. Furthermore, the sensitivity of this reagent may

be more consistent when detecting cells of C. jejuni, C coli and C. upsaliensis. This would

provide an advantage over currently available detection methods, as they can have reduced

sensitivity for species other than C. jejuni (Sapsford et al., 2004). Antibody purification

205

could be achieved using an affinity chromatography column containing an N-glycoprotein

other than NGRP/Cj0114. Alternatively a column containing the fOS produced by each

species could be developed. Purification of fOS could be undertaken according to the

protocol of Nothaft et al. (2012), which was successfully used to create fOS-BSA

conjugate immunogens for producing the C. fetus N-glycan-reactive GRPII sera. Such

affinity chromatography columns would retain N-glycan-reactive antibodies, which could

be eluted from the column upon removing the non-N-glycan-reactive antibodies and other

serum components. These approaches would generate high-titre reagents specific for the

respective N-linked glycan structures and may allow more sensitive detection of the

corresponding Campylobacter species.

Such purified anti-N-glycan antibodies could subsequently be integrated into detection

platforms such as LFDs or biosensors and their performance assessed in a more ‘real

world’ applications. Such analysis could include determining the sensitivity and

specificity of the platform. In particular, the occurrence of cross-reactivity with bacterial

species often resident within poultry caeca and human faeces would be necessary for

CjNgp and CfNgp, respectively. For example, Clostridium, Bacteroides, and

Gallibacterium species are often found in poultry caeca (Pauwels et al., 2015; Stanley et

al., 2015). In line with this, the compatibility of such assays when directly testing

poultry/human samples could be investigated.

As NGRP-reactivity was a significant feature of the antisera described here, it may be

more appropriate to use an alternative N-glycoprotein antigen when producing such

antisera in the future. For example, Nothaft et al. (2016) have developed a small

‘GlycoTag’ from a C. jejuni N-glycoprotein which is N-glycosylated at nine sites and was

readily expressed and purified from E. coli. Such a small protein may result in less

protein-reactive antibodies in the antiserum. Alternatively, an N-glycoprotein unique to

the respective target Campylobacter species could be used so that any resulting protein

reactivity should not cause cross-reactivity with other species. This could be particularly

useful for further development of C. fetus N-glycan-reactive antisera so that the cross-

reactivity with C. jejuni and other species is reduced. Furthermore, the development of

glycocompetent E. coli producing the C. fetus N-linked glycan would provide a more

effective method of producing and purifying large quantities of C. fetus N-glycoprotein

(See below) and an alternative carrier protein, for example GlycoTag, could be introduced

into such a system. Another option would be to use fOS-BSA conjugates as immunogen,

which was successfully utilised by Nothaft et al. (2012) to develop the GRP-II sera. As

described in section 4, these antisera displayed stricter Campylobacter species specificity

206

than CfNgp as they did not react with C. concisus. However, it must be noted that

polyclonal antisera can display batch-to-batch variation and such specificity may not be

reproducible. In conclusion, this study has highlighted the limitations of using polyclonal

antisera and it can be argued that monoclonal antibodies would be more suitable in the

future. However, the production of monoclonal antibodies is more time-consuming and

expensive than raising polyclonal sera and these findings demonstrate that the latter are a

feasible tool to preliminarily assess potential target antigens.

Nevertheless, in their current form, these antisera will be useful for N-glycosylation

research. CjNgp can be used similarly as the previously available hR6 antiserum, and

assist investigation into the N-linked glycosyation systems of C. coli, C. lari and others.

Due to its broad reactivity against Campylobacter species, CfNgp will prove a valuable

tool for Campylobacter N-glycan research into the less established systems in species such

as C. fetus and C. concisus. For example, C. concisus harbours two pglB enzymes and this

antiserum could aid the study of their activities. In addition, CfNgp has the potential to be

used for N-glycoprotein pulldown experiments to identify N-glycoproteins in multiple

Campylobacter species.

7.2 Developing glycocompetent E. coli producing the C. fetus N-

linked glycan.

Chapter 5 described a strategy toward developing glycocompetent E. coli able to

synthesise the C. fetus N-glycan structures. By exploiting the known structural features of

the C. jejuni and C. fetus N-linked glycan, this approach used part of the C. jejuni Pgl

system to produce a trisaccharide N-glycan which could be used as a basal substrate by

predicted C. fetus glycosyltransferases. Introducing C. fetus pglH1 and pglH2 into this

system, respectively, demonstrated that these genes encoded enzymes able to sequentially

modify this truncated N-glycan. Structural analysis revealed that PglH1 and PglH2 could

each add a HexNAc sugar, likely GlcNAc residues given the structure of C. fetus N-

glycan. Although this approach of combining two N-linked glycosylation pathways was

partially successful, it was limited by unexpected interactions with the components of the

two systems. Upon introduction of PglH1, a Hex sugar was also transferred along with the

presumed GlcNAc, however, the enzyme responsible for this transfer is to be determined

as C. jejuni PglI may have interfered with the system. Another predicted

207

glycosyltransferase, encoded by Cf1389, did not contribute to further N-glycan

biosynthesis and rather appeared to cause deleterious effects in the glycocompetent E. coli.

Subsequent genome analysis suggests that this glycosyltransferase is instead part of a

prophage O-antigen modification locus. Nevertheless, this work successfully developed a

hybrid C. jejuni/C. fetus N-glycosylation system in E. coli and provided strong evidence

that C.fetus PglH1 and PglH2 are active glycosyltransferases that act sequentially to add

GlcNAc residues to the N-linked glycan structure.

The main limitation of this work is the likely interference by the C. jejuni PglI enzyme.

Therefore, future work would require C. jejuni PglI to be removed from the system and

could be achieved by insertional mutagenesis of pglI within ppglpglH::kn. Re-analysis of

the N-glycan structures produced in the absence of C. jejuni pglI would identify if the Hex

residue in question was still transferred upon introduction of C. fetus pglH1. If so, it would

seem that C. fetus PglH1 was responsible for the transfer of this residue. However, if the

N-glycan structure did not contain a Hex residue in the absence of C. jejuni pglI it would

suggest that the presence of this residue was due to interference from the C. jejuni

pathway. In this case, it could be concluded that C. fetus PglH1 is responsible for the

transfer of the HexNAc residue, likely a GlcNAc, only. Such activity would better reflect

its homology with C. jejuni PglH enzyme which transfers linear GalNAc residues to the

C. jejuni N-glycan. NMR could then be used to characterise the specific sugars and

glycosidic linkages present to confirm that the HexNAc residues transferred by C. fetus

glycosyltransferases are GlcNAc residues, as predicted.

Another limitation of this approach was the difficulty in obtaining N-glycan structural data

using MALDI-TOF-MS. Although using an improved version of NGRP, NGRP3,

somewhat improved the outcome of the MALDI-TOF-MS analysis, it was not sufficient.

Therefore, LC-MS/MS anaylsis was undertaken to alleviate these issues. However, LC-

MS/MS is a more costly technology and produces large quantities of spectra which require

time-consuming data analysis. In contrast, MALDI-TOF-MS is a cheaper and simple

alternative. Another variant of NGRP, modified to produce a further positively charged N-

glycan sequon-containing tryptic peptide, was successfully used by Mills et al. (2016) to

gain N-glycan structural data using MALDI-TOF-MS. Introducing this NGRP variant into

the experimental strategy described here could therefore improve the applicability of this

approach.

Finally, to identify the glycosyltransferase(s) responsible for transferring the remaining

C. fetus N-linked glycan residues, predicted glycosyltransferase genes common to C. fetus

208

and C. hyointestinalis could be introduced into the system. Such investigation is now

possible with the recent availability of complete C. hyointestinalis genome sequences

(Miller et al., 2016). Development of the full C. fetus N-glycan structures in E. coli would

allow more rapid and high-yield production of N-glycoprotein, which could readily

provide antigen for antibody purification (see above). Furthermore, as the majority of

Campylobacter N-linked glycan structures are structurally identical at the reducing end,

this work provides an experimental framework which could be adapted for investigating

glycosyltransferase enzymes of other Campylobacter species.

7.3 The C. fetus predicted N-glycoproteome and putative

conservation of N-glycoproteins amongst Campylobacter species.

Chapter 6 described prediction of the C. fetus N-glycoproteome using a bioinformatics

pipeline which analyses the predicted subcellular locations of proteins and the occurrence

of N-glycosylation sequons. Combining experimental data and the results of the N-

glycoprotein prediction pipeline, it was found that there are at least 134 possible N-linked

glycoproteins in C. fetus. PglB modification of two predicted N-glycoproteins, Cf0718 and

Cf0445, the latter of which was previously unknown as an N-glycoprotein, was validated

in C. fetus. However, the bioinformatics pipeline did not predict some previously

experimentally identified N-glycoproteins due to incorrect assignment of their subcellular

location, whereas manual analysis using another localisation prediction tool, CELLO,

identified these proteins as extracytoplasmic. Therefore, it appears the subcellular location

predictive ability of the pipeline could be improved if this tool was integrated.

Analysis of the C. fetus predicted N-glycoproteome found that it generally had similar

features to that previously found with C. jejuni (Frost, 2015; Scott et al., 2014). For

example, investigating the predicted subcellular localisations of C. fetus predicted N-

glycoproteins found that most N-glycoproteins are likely inner membrane proteins, as

found in C. jejuni (Frost, 2015; Scott et al., 2014). Similar to what was previously found

with the C. jejuni predicted N-glycoproteome (Frost, 2015), there were more predicted

outer membrane N-glycoproteins than periplasmic, further suggesting that N-glycosylation

may be particularly important for surface exposed proteins. This could be due to N-glycan

modification increasing resistance to proteolytic degradation by extracellular proteases, as

suggested for C. jejuni by Alemka et al. (2013). Like in C. jejuni, most C. fetus predicted

209

N-glycoproteins contain only one or two sequons. However, there were instances where

more sequons were evident, with one predicted N-glycoprotein identified containing 11

sequons making this potentially the most N-glycosylated Campylobacter protein.

Although C. fetus N-glycoproteins experimentally identified by Nothaft et al. (2013)

contain the extended sequon, D/E-X-N-X-S/T (X ≠ P), known to be a requirement for N-

linked glycosylation in C. jejuni, the acceptor sequon specificity of C. fetus PglB has not

been studied in detail. This prediction pipeline found that, in agreement with

Campylobacter PglB in vitro assays (Chen et al., 2007; Gerber et al., 2013), the amino

acid composition of sequons in C. fetus predicted N-glycoproteins most commonly

contained aspartate and threonine at the -2 and +2 positions, respectively. However, the

abundance of these amino acid residues does not mean that D-X-N-X-T are the most

efficiently modified sequons in vivo. The acceptor sequon specificity of C. fetus PglB

could be investigated further by validating several of the predicted N-glycoproteins

identified by the pipeline in vivo, as was done with Cf0718 and Cf0445 here. Site-directed

mutagenesis of the asparagine residues within predicted sequons would confirm the sites

of N-glycosylation and identify if any are preferentially modified. Together, such analyses

may begin to reveal the structural preferences and acceptor sequon specificity of the

C. fetus PglB enzyme, which is relatively uncharacterised. Furthermore, this methodology

could be used to investigate if Cf0312 is modified at the eleven N-glycan sequons

predicted and therefore demonstrate if it is the most N-glycosylated Campylobacter

protein identified thus far.

One flaw of this prediction pipeline is that it gives no indication of where the identified

sequons reside within protein structures. Experimental evidence suggests that sequons

which lie within buried structural features may not be modified (Silverman and Imperiali,

2016) and thus presence of a sequon is not sufficient for N-glycosylation. Therefore, it

would be useful to integrate structural predictions into the pipeline to identify N-sequons

which are found in buried structural regions and thus not likely to be modified in vivo.

Detailed protein structure modelling of all predicted N-glycoproteins would be a large and

complex task bioinformatically, however, PredictProtein (Yachdav et al., 2014) is an

online tool which gives a simple output of protein structural prediction and solvent

accessibility. Such a tool may be appropriate for integration into the N-glycoprotein

prediction pipeline. Including such analysis would undoubtedly reduce the size and alter

the overall features of the predicted N-glycoproteome, although the extent of this is

unknown. A subset of these predicted N-glycoproteins could then be validated in vivo to

provide evidence of the accuracy of the pipeline.

210

The second part of Chapter 6 investigated the conservation of a set of experimentally

identified C. jejuni N-glycoproteins thought to be core N-glycoproteins of this species.

The role of N-linked glycosylation in Campylobacter is unclear and limited experimental

evidence has demonstrated that N-glycosylation is not required for the function of

individual proteins in C. jejuni (Davis et al., 2009; Kakuda et al., 2012; Kakuda and

DiRita, 2006; Scott et al., 2009). Some evidence suggests that N-glycosylation instead

serves a more general role and provides protection from chicken gut proteases (Alemka et

al., 2013). Understanding the conservation of N-glycoproteins among Campylobacter

species may help identify proteins which require N-glycosylation for their function and/or

protection, as they will likely be conserved as N-glycosylation targets. Such N-

glycoproteins would provide promising candidates for experimentally studying the role of

N-glycosylation and this investigation served to identify a portion of these N-

glycoproteins. Fourty-eight of the sixty-five C. jejuni N-glycoproteins studied had C. fetus

homologues. Thirty-four of these contained sequons and thus are predicted to be N-

glycosylated in C. fetus. Notably, ten of these predicted C. fetus N-glycoproteins were

experimentally identified by Nothaft et al. (2012). More outer membrane proteins than

periplasmic proteins were predicted to be N-glycoproteins in both species, suggesting that

surface exposure may be a stronger drive for the conservation of N-glycosylation. It can be

speculated that such conservation is due the proposed protection of N-glycan against

protease-mediated degradation, as suggested by Alemka et al. (2013). Analysis among a

further four Campylobacter species, C. coli, C. lari, C.hyointestinalis and C. concisus,

revealed that nineteen of the thirty-four C. jejuni N-glycoproteins also had predicted N-

glycoprotein homologues in these species. Therefore, there appeared to be some

conservation of the proteins targeted for N-glycan modification throughout the

Campylobacter genus.

Interestingly, N-glycoproteins for which experimental research identified N-glycosylation

was dispensible for protein function are those which are not predicted to be conserved as

N-glycoproteins among Campylobacter species. It may be that there is no evolutionary

pressure driving conservation of N-glycan modification of these proteins because it does

not confer a significant advantage. Therefore, in contrast, it can be argued that N-

glycoproteins which are putatively conserved as N-glycoproteins among Campylobacter

species may be so because N-glycan modification does contribute to protein function

and/or protection. Therefore, this work identified promising protein candidates for

experimentally researching the role of N-glycans. In particular, a few N-glycoproteins

identified have known phenotypes (eg. Antibiotic resistance and motility for CmeABC

211

and PflA, respectively), and thus it would be experimentally tractable to abolish N-

glycosylation of these proteins and identify if it is required for function.

An issue with this method is that when analysing the conservation of N-glycoproteins

among Campylobacter species, the number and location of conserved N-sequons was not

analysed. If N-sequons in an N-glycoprotein are similarly located in its homologue, this

may further suggest that N-glycosylation is important for the function of those proteins.

To investigate this, protein sequence alignments and protein structure predictions for the

conserved set of N-glycoproteins could be created. Such analysis may reveal the specific

N-glycosylation sites that are conserved in several species and therefore pinpoint which

sequons to investigate experimentally.

Overall, this bioinformatics prediction approach revealed a large number of possible

C. fetus N-glycoproteins and identified a subset of the possible ‘core’ N-glycoproteins

among Campylobacter species. Although this approach has its drawbacks, predicting N-

glycoproteins complements experimental methods of identification which, at present,

remain limiting. Furthermore, this analysis highlighted areas of improvement to the

pipeline which would provide an increasingly powerful tool for investigating N-linked

glycosylation in Campylobacter species. In the future, this valuable tool could be further

adapted to contribute to the understanding of N-linked glycosylation in other Bacterial

species.

7.4 Final conclusion

This thesis provided evidence that Campylobacter N-linked glycans are attractive targets

for antibody-based detection of Campylobacter species. A C. jejuni N-glycan-reactive

antiserum, CjNgp, showed promise for specific detection of Campylobacter pathogens

found in poultry, including C. jejuni, C. coli and C. lari. Furthermore, this antiserum could

be useful for detecting these organisms in other farmed animals without cross-reacting

with Campylobacter species that may be resident. An antiserum, CfNgp, was raised

against a recombinant C. fetus N-glycoprotein and shown to be able to detect the

veterinary and human pathogen, C. fetus and emerging human pathogens such as

C. concisus and C. ureolyticus. As its broad reactivity against Campylobacter N-linked

glycans could potentially cause cross-reactivity when sampling cattle, it was concluded

that this antiserum may be better suited for detection of such Campylobacter species from

212

human samples. Limitations of these antisera, in their neat form, include protein-cross

reactivity and low sensitivity. Nevertheless, this work demonstrated that, with further

improvement, N-glycan-reactive antibodies are suitable for developing Campylobacter

detection assays that are rapid and simple to perform. Such assays would be valuable for

detecting Campylobacter pathogens in agricultural and/or human samples and could be an

asset to Campylobacter surveillance programs. In addition, the N-glycan-reactive antisera

described here will also be a useful tool for experimental research into Campylobacter N-

linked glycosylation systems.

Secondly, this thesis developed glycocompetent E. coli containing a hybrid C. jejuni/C.

fetus Pgl system. This novel approach successfully demonstrated the activity of two

C. fetus glycosyltransferases, PglH1 and PglH2. Although this experimental system

proved useful and adaptable, it was hindered by possible interference from the C. jejuni

pathway. However, removal of a single C. jejuni enzyme would alleviate this issue and

provide a robust system for continued investigation into C. fetus glycosyltransferases. This

approach would also be a valid method to investigate glycosyltransferases in other

Campylobacter species. In particular, it provides a favourable alternative to newly cloning

entire pgl loci into E. coli or constructing insertional pgl gene knockouts in

Campylobacter, both of which are relatively difficult and laborious tasks.

In conclusion, this thesis has identified the C. fetus predicted N-glycoproteome and

revealed similarities with that of C. jejuni. Furthermore, investigation into the putative

conservation of N-glycosylation protein targets revealed considerable conservation among

six Campylobacter species and identified key candidates for experimental study into the

function of Campylobacter N-linked glycosylation. This study identified limitations within

the bioinformatics pipeline which could inform subsequent work and move towards

developing a more robust N-glycoprotein prediction tool. In the future, this bioinformatics

analysis could further advance the understanding of N-glycosylation throughout the

Campylobacter genus. Futhermore, it has the potential to be adapted for predicting N-

glycoproteins in other bacterial species, for example, the deep-sea vent Bacteria which

were recently identified to contain N-linked glycosylation machinery (Mills et al., 2016).

In conclusion, this work has demonstrated that prediction of Campylobacter N-

glycoproteins is informative and complements experimental N-linked glycosylation

research.

213

References

Aas, F.E., Vik, A., Vedde, J., Koomey, M., Egge-Jacobsen, W., 2007. Neisseria

gonorrhoeae O-linked pilin glycosylation: functional analyses define both the

biosynthetic pathway and glycan structure. Mol. Microbiol. 65, 607–24.

doi:10.1111/j.1365-2958.2007.05806.x

Alemka, A., Nothaft, H., Zheng, J., Szymanski, C.M., 2013. N-glycosylation of

Campylobacter jejuni surface proteins promotes bacterial fitness. Infect. Immun. 81,

1674–1682. doi:10.1128/IAI.01370-12

Ali, A., Soares, S.C., Santos, A.R., Guimarães, L.C., Barbosa, E., Almeida, S.S., Abreu,

V. a C., Carneiro, A.R., Ramos, R.T.J., Bakhtiar, S.M., Hassan, S.S., Ussery, D.W.,

On, S., Silva, A., Schneider, M.P., Lage, A.P., Miyoshi, A., Azevedo, V., 2012.

Campylobacter fetus subspecies: Comparative genomics and prediction of potential

virulence targets. Gene 508, 145–156. doi:10.1016/j.gene.2012.07.070

Allison, G.E., Verma, N.K., 2000. Serotype-converting bacteriophages and O-antigen

modification in Shigella flexneri. Trends Microbiol. 8, 17–23. doi:10.1016/S0966-

842X(99)01646-7

Allos, B.M., Lastovica, A.J., 2008. Clinical Significance of Campylobacter and Related

species other than Campylobacter jejuni and Campylobacter coli, in: Nachamkin, I.,

Szymanski, C.M., Blaser, M.J. (Eds.), Campylobacter, Third Edition. American

Society of Microbiology, Washington, pp. 123–149.

doi:10.1128/9781555815554.ch7

Apostolov, E., Al-Soud, W.A., Nilsson, I., Kornilovska, I., Usenko, V., Lyzogubov, V.,

Gaydar, Y., Wadström, T., Ljungh, A., 2005. Helicobacter pylori and other

Helicobacter species in gallbladder and liver of patients with chronic cholecystitis

detected by immunological and molecular methods. Scand. J. Gastroenterol. 40, 96–

214

102.

Apweiler, R., Hermjakob, H., Sharon, N., 1999. On the frequency of protein

glycosylation, as deduced from analysis of the SWISS-PROT database. Biochim.

Biophys. Acta 1473, 4–8.

Arndt, D., Grant, J.R., Marcu, A., Sajed, T., Pon, A., Liang, Y., Wishart, D.S., 2016.

PHASTER: a better, faster version of the PHAST phage search tool. Nucleic Acids

Res. 44, 1–6. doi:10.1093/nar/gkw387

Awad, W.A., Molnár, A., Aschenbach, J.R., Ghareeb, K., Khayal, B., Hess, C., Liebhart,

D., Dublecz, K., Hess, M., 2015a. Campylobacter infection in chickens modulates the

intestinal epithelial barrier function. Innate Immun. 21, 151–160.

doi:10.1177/1753425914521648

Awad, W.A., Smorodchenko, A., Hess, C., Aschenbach, J.R., Molnár, A., Dublecz, K.,

Khayal, B., Pohl, E.E., Hess, M., 2015b. Increased intracellular calcium level and

impaired nutrient absorption are important pathogenicity traits in the chicken

intestinal epithelium during Campylobacter jejuni colonization. Appl. Microbiol.

Biotechnol. 99, 6431–41. doi:10.1007/s00253-015-6543-z

Baar, C., Eppinger, M., Raddatz, G., Simon, J., Lanz, C., Klimmek, O., Nandakumar, R.,

Gross, R., Rosinus, A., Keller, H., Jagtap, P., Linke, B., Meyer, F., Lederer, H.,

Schuster, S.C., 2003. Complete genome sequence and analysis of Wolinella

succinogenes. Proc. Natl. Acad. Sci. U. S. A. 100, 11690.

doi:10.1073/pnas.1932838100

Baker, J., Barton, M.D., Lanser, J., 1999. Campylobacter species in cats and dogs in South

Australia. Aust. Vet. J. 77, 662–6.

215

Benjamin, J., Leaper, S., Owen, R.J., Skirrow, M.B., 1983. Description of Campylobacter

laridis, a new species comprising the nalidixic acid resistant

thermophilicCampylobacter (NARTC) group. Curr. Microbiol. 8, 231–238.

doi:10.1007/BF01579552

Bernatchez, S., Szymanski, C.M., Ishiyama, N., Li, J.J., Jarrell, H.C., Lau, P.C., Berghuis,

A.M., Young, N.M., Wakarchuk, W.W., 2005. Single bifunctional UDP-GlcNAc/Glc

4-epimerase supports the synthesis of three cell surface glycoconjugates in

Campylobacter jejuni. J. Biol. Chem. 280, 4792–4802.

Blaser, M.J., Smith, P.F., Repine, J.E., Joiner, K.A., 1988. Pathogenesis of Campylobacter

fetus infections. Failure of encapsulated Campylobacter fetus to bind C3b explains

serum and phagocytosis resistance. J. Clin. Invest. 81, 1434–1444.

doi:10.1172/JCI113474

Bohin, J.-P., 2000. Osmoregulated periplasmic glucans in Proteobacteria. FEMS

Microbiol. Lett. 186, 11–19. doi:10.1111/j.1574-6968.2000.tb09075.x

Bohr, U.R.M., Glasbrenner, B., Primus, A., Zagoura, A., Wex, T., Malfertheiner, P., 2004.

Identification of enterohepatic Helicobacter species in patients suffering from

inflammatory bowel disease. J. Clin. Microbiol. 42, 2766–8.

doi:10.1128/JCM.42.6.2766-2768.2004

Bontemps-Gallo, S., Bohin, J.-P., Lacroix, J.-M., 2017. Osmoregulated Periplasmic

Glucans. EcoSal Plus 7. doi:10.1128/ecosalplus.ESP-0001-2017

Bontemps-Gallo, S., Lacroix, J.-M., 2015. New insights into the biological role of the

osmoregulated periplasmic glucans in pathogenic and symbiotic bacteria. Environ.

Microbiol. Rep. 7, 690–697. doi:10.1111/1758-2229.12325

216

Borisov, S.M., Wolfbeis, O.S., 2008. Optical biosensors. Chem. Rev. 108, 423–461.

doi:10.1021/cr068105t

Borud, B., Anonsen, J.H., Viburiene, R., Cohen, E.H., Samuelsen, A.B.C., Koomey, M.,

2014. Extended glycan diversity in a bacterial protein glycosylation system linked to

allelic polymorphisms and minimal genetic alterations in a glycosyltransferase gene.

Mol. Microbiol. 94, 688–699. doi:10.1111/mmi.12789

Breitling, J., Aebi, M., 2013. N-linked protein glycosylation in the endoplasmic reticulum.

Cold Spring Harb. Perspect. Biol. 5, a013359. doi:10.1101/cshperspect.a013359

Brooks, B.W., Devenish, J., Lutze-Wallace, C.L., Milnes, D., Robertson, R.H., Berlie-

Surujballi, G., 2004. Evaluation of a monoclonal antibody-based enzyme-linked

immunosorbent assay for detection of Campylobacter fetus in bovine preputial

washing and vaginal mucus samples. Vet. Microbiol. 103, 77–84.

doi:10.1016/j.vetmic.2004.07.008

Brooks, B.W., Robertson, R.H., Lutze-Wallace, C.L., Pfahler, W., 2002. Monoclonal

antibodies specific for Campylobacter fetus lipopolysaccharides. Vet. Microbiol. 87,

37–49. doi:10.1016/s0378-1135(02)00026-3

Burda, P., Aebi, M., 1999. The dolichol pathway of N-linked glycosylation. Biochim.

Biophys. Acta - Gen. Subj. 1426, 239–257. doi:10.1016/S0304-4165(98)00127-5

Camacho, C., Coulouris, G., Avagyan, V., Ma, N., Papadopoulos, J., Bealer, K., Madden,

T.L., 2009. BLAST+: architecture and applications. BMC Bioinformatics 10, 421.

doi:10.1186/1471-2105-10-421

Campero, C.M., Anderson, M.L., Walker, R.L., Blanchard, P.C., Barbano, L., Chiu, P.,

Martínez, A., Combessies, G., Bardon, J.C., Cordeviola, J., 2005.

217

Immunohistochemical identification of Campylobacter fetus in natural cases of

bovine and ovine abortions. J. Vet. Med. Ser. B Infect. Dis. Vet. Public Heal. 52,

138–141. doi:10.1111/j.1439-0450.2005.00834.x

Casanova, C., Schweiger, A., Von Steiger, N., Droz, S., Marschall, J., 2015.

Campylobacter concisus pseudo-outbreak caused by improved culture conditions. J.

Clin. Microbiol. 53, 660–662. doi:10.1128/JCM.02608-14

Ceelen, L., Decostere, A., Verschraegen, G., Ducatelle, R., Haesebrouck, F., 2005.

Prevalence of Helicobacter pullorum among patients with gastrointestinal disease and

clinically healthy persons. J. Clin. Microbiol. 43, 2984–6.

doi:10.1128/JCM.43.6.2984-2986.2005

Chaban, B., Chu, S., Hendrick, S., Waldner, C., Hill, J.E., 2012. Evaluation of a

Campylobacter fetus subspecies venerealis real-time quantitative polymerase chain

reaction for direct analysis of bovine preputial samples. Can. J. Vet. Res. 76, 166–

173.

Chaban, B., Ngeleka, M., Hill, J.E., 2010. Detection and quantification of 14

Campylobacter species in pet dogs reveals an increase in species richness in feces of

diarrheic animals. BMC Microbiol. 10, 73. doi:10.1186/1471-2180-10-73

Chamot-Rooke, J., Rousseau, B., Lanternier, F., Mikaty, G., Mairey, E., Malosse, C.,

Bouchoux, G., Pelicic, V., Camoin, L., Nassif, X., Dumenil, G., 2007. Alternative

Neisseria spp. type IV pilin glycosylation with a glyceramido acetamido

trideoxyhexose residue. Proc. Natl. Acad. Sci. 104, 14783–14788.

doi:10.1073/pnas.0705335104

Chen, M.M., Glover, K.J., Imperiali, B., 2007. From peptide to protein: comparative

analysis of the substrate specificity of N-linked glycosylation in C. jejuni.

Biochemistry 46, 5579–5585. doi:10.1021/bi602633n

218

Choi, H.S., Shin, S.U., Bae, E.H., Ma, S.K., Kim, S.W., 2016. Infectious Spondylitis in a

Patient with Chronic Kidney Disease: Identification of Campylobacter fetus subsp.

testudinum by 16S Ribosomal RNA Sequencing. Jpn. J. Infect. Dis. 69, 517–519.

doi:10.7883/yoken.JJID.2015.461

Cohen-Rosenzweig, C., Yurist-Doutsch, S., Eichler, J., 2012. AglS, a novel component of

the Haloferax volcanii N-glycosylation pathway, is a dolichol phosphate-mannose

mannosyltransferase. J. Bacteriol. 194, 6909–16. doi:10.1128/JB.01716-12

Crooks, G.E., Hon, G., Chandonia, J.-M., Brenner, S.E., 2004. WebLogo: A Sequence

Logo Generator. Genome Res. 14, 1188–1190. doi:10.1101/gr.849004

Davies, M.R., Broadbent, S.E., Harris, S.R., Thomson, N.R., van der Woude, M.W., 2013.

Horizontally acquired glycosyltransferase operons drive Salmonellae

lipopolysaccharide diversity. PLoS Genet. 9. doi:10.1371/journal.pgen.1003568

Davis, L.M., Kakuda, T., DiRita, V.J., 2009. A Campylobacter jejuni znuA orthologue is

essential for growth in low-zinc environments and chick colonization. J. Bacteriol.

191, 1631–1640. doi:10.1128/JB.01394-08

De Boer, P., Rahaoui, H., Leer, R.J., Montijn, R.C., Van Der Vossen, J.M.B.M., 2015.

Real-time PCR detection of Campylobacter spp.: A comparison to classic culturing

and enrichment. doi:10.1016/j.fm.2015.05.006

de Lorenzo, V., Timmis, K.N., 1994. Analysis and construction of stable phenotypes in

gram-negative bacteria with Tn5- and Tn10-derived minitransposons. Methods

Enzymol. 235, 386–405.

De Schutter, J.W., Morrison, J.P., Morrison, M.J., Ciulli, A., Imperiali, B., 2017.

Targeting Bacillosamine Biosynthesis in Bacterial Pathogens: Development of

219

Inhibitors to a Bacterial Amino-Sugar Acetyltransferase from Campylobacter jejuni.

J. Med. Chem. 60, 2099–2118. doi:10.1021/acs.jmedchem.6b01869

Dell, A., Galadari, A., Sastre, F., Hitchen, P., 2010. Similarities and differences in the

glycosylation mechanisms in prokaryotes and eukaryotes. Int. J. Microbiol. 2010,

148178. doi:10.1155/2010/148178

Deshpande, N.P., Kaakoush, N.O., Wilkins, M.R., Mitchell, H.M., 2013. Comparative

genomics of Campylobacter concisus isolates reveals genetic diversity and provides

insights into disease association. BMC Genomics 14, 585. doi:10.1186/1471-2164-

14-585

Devenish, J., Brooks, B., Perry, K., Milnes, D., Burke, T., McCabe, D., Duff, S., Lutze-

Wallace, C.L., 2005. Validation of a monoclonal antibody-based capture enzyme-

linked immunosorbent assay for detection of Campylobacter fetus. Clin. Diagn. Lab.

Immunol. 12, 1261–1268. doi:10.1128/cdli.12.11.1261-1268.2005

Diker, K.S., Diker, S., Ozlem, M.B., 1990. Bovine diarrhea associated with

Campylobacter hyointestinalis. Zentralbl. Veterinarmed. B 37, 158–60.

Ding, W., Nothaft, H., Szymanski, C.M., Kelly, J., 2009. Identification and Quantification

of Glycoproteins Using Ion-Pairing Normal-phase Liquid Chromatography and Mass

Spectrometry. Mol. Cell. Proteomics 8, 2170–2185. doi:10.1074/mcp.M900088-

MCP200

Divakaruni, A. V., Baida, C., White, C.L., Gober, J.W., 2007. The cell shape proteins

MreB and MreC control cell morphogenesis by positioning cell wall synthetic

complexes. Mol. Microbiol. 66, 174–188. doi:10.1111/j.1365-2958.2007.05910.x

Dong, H.-J., Cho, A.-R., Hahn, T.-W., Cho, S., 2014. Development of a loop-mediated

220

isothermal amplification assay for rapid, sensitive detection of Campylobacter jejuni

in cattle farm samples. J. Food Prot. 77, 1593–1598. doi:10.4315/0362-028X.JFP-14-

056

Dwivedi, R., Nothaft, H., Reiz, B., Whittal, R.M., Szymanski, C.M., 2013. Generation of

free oligosaccharides from bacterial protein N-linked glycosylation systems.

Biopolymers 99, 772–783. doi:10.1002/bip.22296

Edmonds, P., Patton, C.M., Griffin, P.M., Barrett, T.J., Schmid, G.P., Baker, C.N.,

Lambert, M. a, Brenner, D.J., 1987. Campylobacter hyointestinalis associated with

human gastrointestinal disease in the United States. J Clin Microbiol 25, 685–691.

EFSA, 2016. The European Union summary report on trends and sources of zoonoses,

zoonotic agents and food-borne outbreaks in 2015. EFSA J. 14.

doi:10.2903/j.efsa.2016.4634

EFSA, 2010. Scientific Opinion on Quantification of the risk posed by broiler meat to

human campylobacteriosis in the EU. EFSA J. 8, 1–89.

doi:10.2903/j.efsa.2010.1437.Available

Eichler, J., 2013. Extreme sweetness: protein glycosylation in archaea. Nat. Rev.

Microbiol. 11, 151–156. doi:10.1038/nrmicro2957

Faridmoayer, A., Fentabil, M.A., Mills, D.C., Klassen, J.S., Feldman, M.F., 2007.

Functional characterization of bacterial oligosaccharyltransferases involved in O-

linked protein glycosylation. J. Bacteriol. 189, 8088–98. doi:10.1128/JB.01318-07

Feldman, M.F., Wacker, M., Hernandez, M., Hitchen, P.G., Marolda, C.L., Kowarik, M.,

Morris, H.R., Dell, A., Valvano, M.A., Aebi, M., 2005. Engineering N-linked protein

glycosylation with diverse O antigen lipopolysaccharide structures in Escherichia

221

coli. Proc. Natl. Acad. Sci. U. S. A. 102, 3016–3021. doi:10.1073/pnas.0500044102

Fitzgerald, C., Patrick, M., Gonzalez, A., Akin, J., Polage, C.R., Wymore, K., Gillim-

Ross, L., Xavier, K., Sadlowski, J., Monahan, J., Hurd, S., Dahlberg, S., Jerris, R.,

Watson, R., Santovenia, M., Mitchell, D., Harrison, C., Tobin-D’Angelo, M.,

DeMartino, M., Pentella, M., Razeq, J., Leonard, C., Jung, C., Achong-Bowe, R.,

Evans, Y., Jain, D., Juni, B., Leano, F., Robinson, T., Smith, K., Gittelman, R.M.,

Garrigan, C., Nachamkin, I., 2016. Multicenter Evaluation of Clinical Diagnostic

Methods for Detection and Isolation of Campylobacter spp. from Stool. J. Clin.

Microbiol. 54, 1209–1215. doi:10.1128/JCM.01925-15

Fitzgerald, C., Tu, Z. c., Patrick, M., Stiles, T., Lawson, A.J., Santovenia, M., Gilbert,

M.J., van Bergen, M., Joyce, K., Pruckler, J., Stroika, S., Duim, B., Miller, W.G.,

Loparev, V., Sinnige, J.C., Fields, P.I., Tauxe, R. V., Blaser, M.J., Wagenaar, J.A.,

2014. Campylobacter fetus subsp. testudinum subsp. nov., isolated from humans and

reptiles. Int. J. Syst. Evol. Microbiol. 64, 2944–2948. doi:10.1099/ijs.0.057778-0

Fouts, D.E., Mongodin, E.F., Mandrell, R.E., Miller, W.G., Rasko, D.A., Ravel, J.,

Brinkac, L.M., DeBoy, R.T., Parker, C.T., Daugherty, S.C., Dodson, R.J., Durkin,

A.S., Madupu, R., Sullivan, S.A., Shetty, J.U., Ayodeji, M.A., Shvartsbeyn, A.,

Schatz, M.C., Badger, J.H., Fraser, C.M., Nelson, K.E., 2005. Major structural

differences and novel potential virulence mechanisms from the genomes of multiple

Campylobacter species. Plos Biol. 3, 72–85. doi:10.1371/journal.pbio.0030015

Frost, H.M., 2015. N-linked protein glycosylation in Campylobacter and Helicobacter

species. Ph.D. Univ. Manchester.

Fry, B.N., Feng, S., Chen, Y.Y., Newell, D.G., Coloe, P.J., Korolik, V., 2000. The galE

gene of Campylobacter jejuni is involved in lipopolysaccharide synthesis and

virulence. Infect. Immun. 68, 2594–2601. doi:10.1128/IAI.68.5.2594-2601.2000

222

Fry, B.N., Korolik, V., Brinke, J.A., Pennings, M.T.T., Zalm, R., Teunis, B.J.J., Coloe,

P.J., Zeijst, B.A.M. Van Der, 1998. The lipopolysaccharide biosynthesis locus of

Campylobacter jejuni 81116.

Garcia, M.M., Lutze-Wallace, C.L., Denes, A.S., Eaglesome, M.D., Holst, E., Blaser,

M.J., 1995. Protein shift and antigenic variation in the S-layer of Campylobacter fetus

subsp. venerealis during bovine infection accompanied by genomic rearrangement of

sapA homologs. J. Bacteriol. 177, 1976–80.

Gebhart, C.J., Edmonds, P., Ward, G.E., Kurtz, H.J., Brenner, D.J., 1985. “Campylobacter

hyointestinalis” sp. nov.: a new species of Campylobacter found in the intestines of

pigs and other animals. J. Clin. Microbiol. 21, 715–720.

Gerber, S., Lizak, C., Michaud, G., Bucher, M., Darbre, T., Aebi, M., Reymond, J.L.,

Locher, K.P., 2013. Mechanism of bacterial oligosaccharyltransferase: In vitro

quantification of sequon binding and catalysis. J. Biol. Chem. 288, 8849–8861.

doi:10.1074/jbc.M112.445940

Gharst, G., Oyarzabal, O. a, Hussain, S.K., 2013. Review of current methodologies to

isolate and identify Campylobacter spp. from foods. J. Microbiol. Methods 95, 84–

92. doi:10.1016/j.mimet.2013.07.014

Gilbert, M.J., Miller, W.G., Yee, E., Zomer, A.L., van der Graaf-van Bloois, L.,

Fitzgerald, C., Forbes, K.J., Méric, G., Sheppard, S.K., Wagenaar, J. a, Duim, B.,

2016. Comparative genomics of Campylobacter fetus from reptiles and mammals

reveals divergent evolution in host-associated lineages. Genome Biol. Evol. 8, 2006–

19. doi:10.1093/gbe/evw146

Giltner, C.L., Saeki, S., Bobenchik, A.M., Humphries, R.M., 2013. Rapid detection of

Campylobacter antigen by enzyme immunoassay leads to increased positivity rates. J.

Clin. Microbiol. 51, 618–620. doi:10.1128/JCM.02565-12

223

Glover, K.J., Weerapana, E., Chen, M.M., Imperiali, B., 2006. Direct biochemical

evidence for the utilization of UDP-bacillosamine by PglC, an essential glycosyl-1-

phosphate transferase in the Campylobacter jejuni N-linked glycosylation pathway.

Biochemistry 45, 5343–5350. doi:10.1021/bi0602056

Glover, K.J., Weerapana, E., Imperiali, B., 2005. In vitro assembly of the

undecaprenylpyrophosphate-linked, heptasaccharide for prokaryotic N-linked

glycosylation. Proc. Natl. Acad. Sci. U. S. A. 102, 14255–14259.

doi:10.1073/pnas.0507311102

Gnanaprakasa, T.J., Oyarzabal, O.A., Olsen, E. V, Pedrosa, V.A., Simonian, A.L., 2011.

Tethered DNA scaffolds on optical sensor platforms for detection of hipO gene from

Campylobacter jejuni. Sensors and Actuators B-Chemical 156, 304–311.

doi:10.1016/j.snb.2011.04.037

Gómez-Camarasa, C., Gutiérrez-Fernández, J., Rodríguez-Granger, J.M., Sampedro-

Martínez, A., Sorlózano-Puerto, A., Navarro-Marí, J.M., 2014. Evaluation of the

rapid RIDAQUICK Campylobacter® test in a general hospital. Diagn. Microbiol.

Infect. Dis. 78, 101–104. doi:10.1016/j.diagmicrobio.2013.11.009

Goon, S., Kelly, J.F., Logan, S.M., Ewing, C.P., Guerry, P., 2003. Pseudaminic acid, the

major modification on Campylobacter flagellin, is synthesized via the Cj1293 gene.

Mol. Microbiol. 50, 659–71.

Goossens, H., Vlaes, L., De Boeck, M., Pot, B., Kersters, K., Levy, J., De Mol, P., Butzler,

J.P., Vandamme, P., 1990. Is Campylobacter upsaliensis an unrecognised cause of

human diarrhoea? Lancet. 335, 584–6.

Gorkiewicz, G., Feierl, G., Zechner, R., Zechner, E.L., 2002. Transmission of

Campylobacter hyointestinalis from a pig to a human. J. Clin. Microbiol. 40, 2601–5.

doi:10.1128/JCM.40.7.2601

224

Gorkiewicz, G., Kienesberger, S., Schober, C., Scheicher, S.R., Guelly, C., Zechner, R.,

Zechner, E.L., 2010. A genomic island defines subspecies-specific virulence features

of the host-adapted pathogen Campylobacter fetus subsp venerealis. J. Bacteriol. 192,

502–517. doi:10.1128/jb.00803-09

Granato, P. a, Chen, L., Holiday, I., Rawling, R. a, Novak-Weekley, S.M., Quinlan, T.,

Musser, K. a, 2010. Comparison of premier CAMPY enzyme immunoassay (EIA),

ProSpecT Campylobacter EIA, and ImmunoCard STAT! CAMPY tests with culture

for laboratory diagnosis of Campylobacter enteric infections. J. Clin. Microbiol. 48,

4022–7. doi:10.1128/JCM.00486-10

Grass, S., Lichti, C.F., Townsend, R.R., Gross, J., St. Geme, J.W., 2010. The Haemophilus

influenzae HMW1C protein is a glycosyltransferase that transfers hexose residues to

asparagine sites in the HMW1 adhesin. PLoS Pathog. 6, e1000919.

doi:10.1371/journal.ppat.1000919

Gross, J., Grass, S., Davis, A.E., Gilmore-Erdmann, P., Townsend, R.R., St. Geme, J.W.,

2008. The Haemophilus influenzae HMW1 adhesin is a glycoprotein with an unusual

N-linked carbohydrate modification. J. Biol. Chem. 283, 26010–26015.

doi:10.1074/jbc.M801819200

Guo, B., Lin, J., Reynolds, D.L., Zhang, Q., 2010. Contribution of the multidrug efflux

transporter CmeABC to antibiotic resistance in different Campylobacter species.

Foodborne Pathog. Dis. 7, 77–83. doi:10.1089/fpd.2009.0354

Gurgan, T., Diker, K.S., 1994. Abortion associated with Campylobacter upsaliensis. J.

Clin. Microbiol. 32, 3093–4.

Hara-Kudo, Y., Yoshino, M., Kojima, T., Ikedo, M., 2005. Loop-mediated isothermal

amplification for the rapid detection of Salmonella. FEMS Microbiol. Lett. 253, 155–

161. doi:10.1016/j.femsle.2005.09.032

225

Hazeleger, W.C., Beumer, R.R., Rombouts, F.M., 1992. The use of latex agglutination

tests for determining Campylobacter species. Lett. Appl. Microbiol. 14, 181–184.

doi:10.1111/j.1472-765X.1992.tb00679.x

Heid, C.A., Stevens, J., Livak, K.J., Williams, P.M., 1996. Real time quantitative PCR.

Genome Res. 6, 986–94. doi:10.1101/GR.6.10.986

Helenius, A., Aebi, M., 2001. Intracellular functions of N-linked glycans. Science 291,

2364–9. doi:10.1126/science.291.5512.2364

Hendrixson, D.R., DiRita, V.J., 2004. Identification of Campylobacter jejuni genes

involved in commensal colonization of the chick gastrointestinal tract. Mol.

Microbiol. 52, 471–484. doi:10.1111/j.1365-2958.2004.03988.x

Hermans, D., Pasmans, F., Heyndrickx, M., Van Immerseel, F., Martel, A., Van Deun, K.,

Haesebrouck, F., 2012. A tolerogenic mucosal immune response leads to persistent

Campylobacter jejuni colonization in the chicken gut. Crit. Rev. Microbiol. 38, 17–

29. doi:10.3109/1040841X.2011.615298

Hermans, D., Van Deun, K., Martel, A., Van Immerseel, F., Messens, W., Heyndrickx,

M., Haesebrouck, F., Pasmans, F., 2011. Colonization factors of Campylobacter

jejuni in the chicken gut. Vet. Res. 42, 82. doi:10.1186/1297-9716-42-82

Hindiyeh, M., Jense, S., Hohmann, S., Benett, H., Edwards, C., Aldeen, W., Croft, A.,

Daly, J., Mottice, S., Carroll, K.C., 2000. Rapid detection of Campylobacter jejuni in

stool specimens by an enzyme immunoassay and surveillance for Campylobacter

upsaliensis in the greater Salt Lake City area. J. Clin. Microbiol. 38, 3076–3079.

Hitchen, P., Brzostek, J., Panico, M., Butler, J.A., Morris, H.R., Dell, A., Linton, D., 2010.

Modification of the Campylobacter jejuni flagellin glycan by the product of the

226

Cj1295 homopolymeric-tract-containing gene. Microbiology 156, 1953–1962.

doi:10.1099/mic.0.038091-0

Hodinka, R.L., Gilligan, P.H., 1988. Evaluation of the campyslide agglutination test for

confirmatory identification of selected Campylobacter species. J. Clin. Microbiol. 26,

47–9.

Holst, E., Wathne, B., Hovelius, B., Mårdh, P.A., 1987. Bacterial vaginosis:

microbiological and clinical findings. Eur. J. Clin. Microbiol. 6, 536–41.

Huang, R.S.P., Johnson, C.L., Pritchard, L., Hepler, R., Ton, T.T., Dunn, J.J., 2016.

Performance of the Verigene® enteric pathogens test, Biofire FilmArrayTM

gastrointestinal panel and Luminex xTAG® gastrointestinal pathogen panel for

detection of common enteric pathogens. Diagn. Microbiol. Infect. Dis. 86, 336–339.

doi:10.1016/j.diagmicrobio.2016.09.013

Hum, S., Quinn, K., Brunner, J., On, S., 1997. Evaluation of a PCR assay for identification

and differentiation of Campylobacter fetus subspecies. Aust. Vet. J. 75, 827–831.

doi:10.1111/j.1751-0813.1997.tb15665.x

Humphrey, S., Chaloner, G., Kemmett, K., Davidson, N., Williams, N., Kipar, A.,

Humphrey, T., Wigley, P., 2014. Campylobacter jejuni is not merely a commensal in

commercial broiler chickens and affects bird welfare. MBio 5, e01364-14.

doi:10.1128/mBio.01364-14

Ielmini, M. V., Feldman, M.F., 2011. Desulfovibrio desulfuricans PglB homolog

possesses oligosaccharyltransferase activity with relaxed glycan specificity and

distinct protein acceptor sequence requirements. Glycobiology 21, 734–742.

doi:10.1093/glycob/cwq192

227

Jarrell, K.F., Ding, Y., Meyer, B.H., Albers, S.-V., Kaminski, L., Eichler, J., 2014. N-

linked glycosylation in Archaea: a structural, functional, and genetic analysis.

Microbiol. Mol. Biol. Rev. 78, 304–41. doi:10.1128/MMBR.00052-13

Javed, S., Gul, F., Javed, K., Bokhari, H., 2017. Helicobacter pullorum: An emerging

zoonotic pathogen. Front. Microbiol. 8, 604. doi:10.3389/fmicb.2017.00604

Jensen, a N., Andersen, M.T., Dalsgaard, a, Baggesen, D.L., Nielsen, E.M., 2005.

Development of real-time PCR and hybridization methods for detection and

identification of thermophilic Campylobacter spp. in pig faecal samples. J. Appl.

Microbiol. 99, 292–300. doi:10.1111/j.1365-2672.2005.02616.x

Jervis, A.J., Butler, J.A., Wren, B.W., Linton, D., 2015. Chromosomal integration vectors

allowing flexible expression of foreign genes in Campylobacter jejuni. BMC

Microbiol. 15, 230. doi:10.1186/s12866-015-0559-5

Jervis, A.J., Butler, J. a., Lawson, A.J., Langdon, R., Wren, B.W., Linton, D., 2012.

Characterization of the structurally diverse N-linked glycans of Campylobacter

species. J. Bacteriol. 194, 2355–2362. doi:10.1128/JB.00042-12

Jervis, A.J., Langdon, R., Hitchen, P., Lawson, A.J., Wood, A., Fothergill, J.L., Morris,

H.R., Dell, A., Wren, B., Linton, D., 2010. Characterization of N-Linked Protein

Glycosylation in Helicobacter pullorum. J. Bacteriol. 192, 5228–5236.

doi:10.1128/jb.00211-10

Jin, S., Joe, A., Lynett, J., Hani, E.K., Sherman, P., Chan, V.L., 2001. JlpA, a novel

surface-exposed lipoprotein specific to Campylobacter jejuni, mediates adherence to

host epithelial cells. Mol. Microbiol. 39, 1225–36.

Johnson, K.F., 1999. Synthesis of oligosaccharides by bacterial enzymes. Glycoconj. J. 16,

228

141–146. doi:10.1023/A:1026440509859

Johnson, T.J., Shank, J.M., Johnson, J.G., 2017. Current and potential treatments for

reducing Campylobacter colonization in animal hosts and disease in humans. Front.

Microbiol. 8, 1–14. doi:10.3389/fmicb.2017.00487

Jones, K., 2001. Campylobacters in water, sewage and the environment. J. Appl.

Microbiol. 90, 68S–79S. doi:10.1046/j.1365-2672.2001.01355.x

Jones, M.A., Marston, K.L., Woodall, C.A., Maskell, D.J., Linton, D., Karlyshev, A. V,

Dorrell, N., Wren, B.W., Barrow, P.A., 2004. Adaptation of Campylobacter jejuni

NCTC11168 to high-level colonization of the avian gastrointestinal tract. Infect.

Immun. 72, 3769–76. doi:10.1128/IAI.72.7.3769-3776.2004

Juncker, A.S., Willenbrock, H., von Heijne, G., Brunak, S., Nielsen, H., Krogh, A., 2003.

Prediction of lipoprotein signal peptides in Gram-negative bacteria. Protein Sci. 12,

1652–1662. doi:10.1110/ps.0303703

Kaakoush, N.O., Castano-Rodriguez, N., Mitchell, H.M., Man, S.M., 2015. Global

epidemiology of Campylobacter infection. Clin. Microbiol. Rev. 28, 687–720.

doi:10.1128/CMR.00006-15

Kaakoush, N.O., Mitchell, H.M., 2012. Campylobacter concisus – A new player in

intestinal disease. Front. Cell. Infect. Microbiol. 2, 1–15.

doi:10.3389/fcimb.2012.00004

Kakuda, T., DiRita, V.J., 2006. Cj1496c encodes a Campylobacter jejuni glycoprotein that

influences invasion of human epithelial cells and colonization of the chick

gastrointestinal tract. Infect. Immun. 74, 4715–4723. doi:10.1128/IAI.00033-06

229

Kakuda, T., Koide, Y., Sakamoto, A., Takai, S., 2012. Characterization of two putative

mechanosensitive channel proteins of Campylobacter jejuni involved in protection

against osmotic downshock. Vet. Microbiol. 160, 53–60.

doi:10.1016/j.vetmic.2012.04.044

Kamei, K., Kawabata, H., Asakura, M., Samosornsuk, W., Hinenoya, A., Nakagawa, S.,

Yamasaki, S., 2016. A cytolethal distending toxin gene-based multiplex PCR assay

for Campylobacter jejuni, C. fetus, C. coli, C. upsaliensis, C. hyointestinalis, and C.

lari. Jpn. J. Infect. Dis. 69, 256–258. doi:10.7883/yoken.JJID.2015.182

Karlyshev, A. V., Ketley, J.M., Wren, B.W., 2005. The Campylobacter jejuni glycome.

FEMS Microbiol. Rev. 29, 377–390. doi:10.1016/j.femsre.2005.01.003

Karlyshev, A. V., Linton, D., Gregson, N.A., Wren, B.W., 2002. A novel paralogous gene

family involved in phase-variable flagella-mediated motility in Campylobacter jejuni.

Microbiology 148, 473–480. doi:10.1099/00221287-148-2-473

Karlyshev, a V, Everest, P., Linton, D., Cawthraw, S., Newell, D.G., Wren, B.W., 2004.

The Campylobacter jejuni general glycosylation system is important for attachment

to human epithelial cells and in the colonization of chicks. Microbiology 150, 1957–

1964. doi:10.1099/mic.0.26721-0

Karlyshev, a V, Wren, B.W., 2005. Development and application of an insertional system

for gene delivery and expression in Campylobacter jejuni. Appl. Environ. Microbiol.

71, 4004–4013. doi:10.1128/AEM.71.7.4004

Kawai, F., Paek, S., Choi, K.-J., Prouty, M., Kanipes, M.I., Guerry, P., Yeo, H.-J., 2012.

Crystal structure of JlpA, a surface-exposed lipoprotein adhesin of Campylobacter

jejuni. J. Struct. Biol. 177, 583–588. doi:10.1016/j.jsb.2012.01.001

230

Kelleher, D.J., Gilmore, R., 2005. An evolving view of the eukaryotic

oligosaccharyltransferase. Glycobiology 16, 47R–62R. doi:10.1093/glycob/cwj066

Kelleher, D.J., Karaoglu, D., Mandon, E.C., Gilmore, R., 2003. Oligosaccharyltransferase

isoforms that contain different catalytic stt3 subunits have distinct enzymatic

properties. Mol. Cell 12, 101–111. doi:10.1016/S1097-2765(03)00243-0

Kelly, J., Jarrell, H., Millar, L., Tessier, L., Fiori, L.M., Lau, P.C., Allan, B., Szymanski,

C.M., 2006. Biosynthesis of the N-linked glycan in Campylobacter jejuni and

addition onto protein through block transfer. J. Bacteriol. 188, 2427–2434.

doi:10.1128/JB.188.7.2427-2434.2006

Kienesberger, S., Gorkiewicz, G., Joainig, M.M., Scheicher, S.R., Leitner, E., Zechner,

E.L., 2007. Development of experimental genetic tools for Campylobacter fetus.

Appl. Environ. Microbiol. 73, 4619–4630. doi:10.1128/AEM.02407-06

Kim, S.A., Lee, Y.M., Hwang, I.G., Kang, D.H., Woo, G.J., Rhee, M.S., 2009. Eight

enrichment broths for the isolation of Campylobacter jejuni from inoculated

suspensions and ground pork. Lett. Appl. Microbiol. 49, 620–626.

doi:10.1111/j.1472-765X.2009.02714.x

Kirk, R., Rowe, M.T., 1994. A PCR assay for the detection of Campylobacter jejuni and

Campylobacter coli in water. Lett. Appl. Microbiol. 19, 301–3.

Konkel, M.E., Garvis, S.G., Tipton, S.L., Anderson, D.E., Cieplak, W., 1997.

Identification and molecular cloning of a gene encoding a fibronectin-binding protein

(CadF) from Campylobacter jejuni. Mol. Microbiol. 24, 953–63.

Konkel, M.E., Klena, J.D., Rivera-Amill, V., Monteville, M.R., Biswas, D., Raphael, B.,

Mickelson, J., 2004. Secretion of virulence proteins from Campylobacter jejuni is

231

dependent on a functional flagellar export apparatus. J. Bacteriol. 186, 3296–3303.

doi:10.1128/JB.186.11.3296-3303.2004

Korlath, J.A., Osterholm, M.T., Judy, L.A., Forfang, J.C., Robinson, R.A., 1985. A point-

source outbreak of campylobacteriosis associated with consumption of raw milk. J.

Infect. Dis. 152, 592–6.

Kowarik, M., Numao, S., Feldman, M.F., Schulz, B.L., Callewaert, N., Kiermaier, E.,

Catrein, I., Aebi, M., 2006a. N-linked glycosylation of folded proteins by the

bacterial oligosaccharyltransferase. Science (80-. ). 314, 1148–1150.

doi:10.1126/science.1134351

Kowarik, M., Young, N.M., Numao, S., Schulz, B.L., Hug, I., Callewaert, N., Mills, D.C.,

Watson, D.C., Hernandez, M., Kelly, J.F., Wacker, M., Aebi, M., 2006b. Definition

of the bacterial N-glycosylation site consensus sequence. Embo J. 25, 1957–1966.

doi:10.1038/sj.emboj.7601087

Koziel, M., Lucey, B., Bullman, S., Corcoran, G.D., Sleator, R.D., 2012. Molecular-based

detection of the gastrointestinal pathogen Campylobacter ureolyticus in

unpasteurized milk samples from two cattle farms in Ireland. Gut Pathog. 4, 14.

doi:10.1186/1757-4749-4-14

Krachler, A.M., Sharma, A., Cauldwell, A., Papadakos, G., Kleanthous, C., 2010. TolA

modulates the oligomeric status of YbgF in the bacterial periplasm. J. Mol. Biol. 403,

270–285. doi:10.1016/j.jmb.2010.08.050

Krogh, A., Larsson, B., von Heijne, G., Sonnhammer, E.L.., 2001. Predicting

transmembrane protein topology with a hidden markov model: application to

complete genomes. J. Mol. Biol. 305, 567–580. doi:10.1006/jmbi.2000.4315

232

Ku, S.C., Schulz, B.L., Power, P.M., Jennings, M.P., 2009. The pilin O-glycosylation

pathway of pathogenic Neisseria is a general system that glycosylates AniA, an outer

membrane nitrite reductase. Biochem. Biophys. Res. Commun. 378, 84–89.

doi:10.1016/j.bbrc.2008.11.025

Kulkarni, S.P., Lever, S., Logan, J.M.J., Lawson, A.J., Stanley, J., Shafi, M.S., 2002.

Detection of Campylobacter species: a comparison of culture and polymerase chain

reaction based methods. J. Clin. Pathol. 55, 749–53.

Larsen, J.C., Szymanski, C., Guerry, P., 2004. N-linked protein glycosylation is required

for full competence in Campylobacter jejuni 81-176. J Bacteriol.186, 6508–6514.

doi:10.1128/JB.186.19.6508

Larson, D.J., Wesley, I. V., Hoffman, L.J., 1992. Use of oligodeoxynucleotide probes to

verify Campylobacter jejuni as a cause of bovine abortion. J. Vet. Diagnostic

Investig. 4, 348–351. doi:10.1177/104063879200400324

Lastovica, A.J., On, S.L.W., Zhang, L., 2014. The Family Campylobacteraceae, in: The

Prokaryotes. Springer Berlin Heidelberg, Berlin, Heidelberg, pp. 307–335.

doi:10.1007/978-3-642-39044-9_274

Lawson, A.J., Logan, J.M., On, S.L., Stanley, J., 2001. Campylobacter hominis sp. nov.,

from the human gastrointestinal tract. Int. J. Syst. Evol. Microbiol. 51, 651–660.

doi:10.1099/00207713-51-2-651

Lee, M.D., Newell, D.G., 2006. Campylobacter in Poultry: Filling an Ecological Niche.

Avian Dis. 50, 1–9. doi:10.1637/7474-111605R.1

Lin, J., Michel, L.O., Zhang, Q., 2002. CmeABC functions as a multidrug efflux system in

Campylobacter jejuni. Antimicrob. Agents Chem. 46, 2124–2131.

233

doi:10.1128/AAC.46.7.2124

Lin, J., Sahin, O., Michel, L.O., Zhang, Q., 2003. Critical role of multidrug efflux pump

CmeABC in bile resistance and in vivo colonization of Campylobacter jejuni. Infect.

Immun. 71, 4250–4259. doi:10.1128/IAI.71.8.4250

Lindblom, G.B., Sjögren, E., Hansson-Westerberg, J., Kaijser, B., 1995. Campylobacter

upsaliensis, C. sputorum sputorum and C. concisus as common causes of diarrhoea in

Swedish children. Scand. J. Infect. Dis. 27, 187–8.

Linton, D., Allan, E., Karlyshev, A. V, Cronshaw, A.D., Wren, B.W., 2002. Identification

of N-acetylgalactosamine-containing glycoproteins PEB3 and CgpA in

Campylobacter jejuni. Mol. Microbiol. 43, 497–508.

Linton, D., Dorrell, N., Hitchen, P.G., Amber, S., Karlyshev, A. V., Morris, H.R., Dell, A.,

Valvano, M. a., Aebi, M., Wren, B.W., 2005. Functional analysis of the

Campylobacter jejuni N-linked protein glycosylation pathway. Mol. Microbiol. 55,

1695–1703. doi:10.1111/j.1365-2958.2005.04519.x

Linton, D., Karlyshev, A. V, Hitchen, P.G., Morris, H.R., Dell, A., Gregson, N.A., Wren,

B.W., 2000. Multiple N-acetyl neuraminic acid synthetase (neuB) genes in

Campylobacter jejuni: identification and characterization of the gene involved in

sialylation of lipo-oligosaccharide. Mol. Microbiol. 35, 1120–1134.

doi:10.1046/j.1365-2958.2000.01780.x

Linton, D., Lawson, A.J., Owen, R.J., Stanley, J., 1997. PCR detection, identification to

species level, and fingerprinting of Campylobacter jejuni and Campylobacter coli

direct from diarrheic samples. J. Clin. Microbiol. 35, 2568–2572.

Linton, D., Owen, R.., Stanley, J., 1996. Rapid identification by PCR of the genus

234

Campylobacter and of five Campylobacter species enteropathogenic for man and

animals. Res. Microbiol. 147, 707–718. doi:10.1016/S0923-2508(97)85118-2

Liu, J., Mushegian, A., 2003. Three monophyletic superfamilies account for the majority

of the known glycosyltransferases. Protein Sci. 12, 1418–31. doi:10.1110/ps.0302103

Liu, L., Hussain, S.K., Miller, R.S., Oyarzabal, O.A., 2009. Efficacy of Mini VIDAS for

the detection of Campylobacter spp. from retail broiler meat enriched in bolton broth,

with or without the supplementation of blood. J. Food Prot. 72, 2428–2432.

Liu, X., McNally, D.J., Nothaft, H., Szymanski, C.M., Brisson, J.-R., Li, J., 2006. Mass

spectrometry-based glycomics strategy for exploring N-linked glycosylation in

Eukaryotes and Bacteria. Anal. Chem. 78, 6081–6087. doi:10.1021/ac060516m

Lizak, C., Gerber, S., Numao, S., Aebi, M., Locher, K.P., 2011. X-ray structure of a

bacterial oligosaccharyltransferase. Nature 474, 350–355. doi:10.1038/nature10151

Logan, J.M., Burnens, a, Linton, D., Lawson, a J., Stanley, J., 2000. Campylobacter

lanienae sp. nov., a new species isolated from workers in an abattoir. Int. J. Syst.

Evol. Microbiol. 50 Pt 2, 865–72.

Logan, S.M., 2006. Flagellar glycosylation - a new component of the motility repertoire?

Microbiology 152, 1249–1262. doi:10.1099/mic.0.28735-0

Logan, S.M., Hui, J.P.M., Vinogradov, E., Aubry, A.J., Melanson, J.E., Kelly, J.F.,

Nothaft, H., Soo, E.C., 2009. Identification of novel carbohydrate modifications on

Campylobacter jejuni 11168 flagellin using metabolomics-based approaches. FEBS

J. 276, 1014–1023. doi:10.1111/j.1742-4658.2008.06840.x

235

Luangtongkum, T., Jeon, B., Han, J., Plummer, P., Logue, C.M., Zhang, Q., 2009.

Antibiotic resistance in Campylobacter: emergence, transmission and persistence.

Future Microbiol. 4, 189–200. doi:10.2217/17460913.4.2.189

Luo, N., Sahin, O., Lin, J., Michel, L.O., Zhang, Q., 2003. In vivo selection of

Campylobacter isolates with high levels of fluoroquinolone resistance associated

with gyrA mutations and the function of the CmeABC efflux pump. Antimicrob.

Agents Chem. 47, 390–4. doi:10.1128/aac.47.1.390-394.2003

Madden, R.H., Hutchison, M., Young, F., Taylor, M., 2016. Development of a rapid on-

farm test for the detection of Campylobacter. FSA Project FS241049.

Magidovich, H., Eichler, J., 2009. Glycosyltransferases and oligosaccharyltransferases in

Archaea: putative components of the N -glycosylation pathway in the third domain of

life. FEMS Microbiol. Lett. 300, 122–130. doi:10.1111/j.1574-6968.2009.01775.x

Man, S.M., 2011. The clinical importance of emerging Campylobacter species. Nat. Rev.

Gastroenterol. Hepatol. 8, 669–685. doi:10.1038/nrgastro.2011.191

Mannering, S. a., West, D.M., Fenwick, S.G., Marchant, R.M., O’Connell, K., 2006.

Pulsed-field gel electrophoresis of Campylobacter jejuni sheep abortion isolates. Vet.

Microbiol. 115, 237–242. doi:10.1016/j.vetmic.2006.01.005

Martinot, M., Jaulhac, B., Moog, R., De Martino, S., Kehrli, P., Monteil, H., Piemont, Y.,

2001. Campylobacter lari bacteremia. Clin. Microbiol. Infect. 7, 96–97.

doi:10.1046/j.1469-0691.2001.00212.x

Mavris, M., Manning, P.A., Morona, R., 1997. Mechanism of bacteriophage SfII-mediated

serotype conversion in Shigella flexneri. Mol. Microbiol. 26, 939–50.

doi:10.1046/j.1365-2958.1997.6301997.x

236

McMillen, L., Fordyce, G., Doogan, V.J., Lew, A.E., 2006. Comparison of culture and a

novel 5’ Taq nuclease assay for direct detection of Campylobacter fetus subsp.

venerealis in clinical specimens from cattle. J. Clin. Microbiol. 44, 938–45.

doi:10.1128/JCM.44.3.938-945.2006

Mcnally, D.J., Hui, J.P.M., Aubry, A.J., Mui, K.K.K., Guerry, P., Brisson, J.-R., Logan,

S.M., Soo, E.C., 2006. Functional characterization of the flagellar glycosylation locus

in Campylobacter jejuni 81–176 using a focused metabolomics approach. J. Biol.

Chem. 281, 18489-18498. doi:10.1074/jbc.M603777200

Mellick, P.W., Winter, A.J., Mcentee, K., 1965. Diagnosis of vibriosis in the bull by use

of the fluorescent antibody technique. Cornell Vet. 55, 280–94.

Mescher, M.F., Stroming.Jl, Watson, S.W., 1974. Protein and carbohydrate composition

of cell envelope of Halobacterium salinarium. J. Bacteriol. 120, 945–954.

Miller, R.S., Speegle, L., Oyarzabal, O. a., Lastovica, A.J., 2008. Evaluation of three

commercial latex agglutination tests for identification of Campylobacter spp. J. Clin.

Microbiol. 46, 3546–3547. doi:10.1128/JCM.01027-08

Miller, W.G., Yee, E., Chapman, M.H., 2016. Complete genome sequences of

Campylobacter hyointestinalis subsp. hyointestinalis strain LMG 9260 and C.

hyointestinalis subsp. lawsonii strain LMG 15993. Genome Announc. 4, e00665-16.

doi:10.1128/genomeA.00665-16

Mills, D.C., Jervis, A.J., Abouelhadid, S., Yates, L.E., Cuccui, J., Linton, D., Wren, B.W.,

2016. Functional analysis of N-linking oligosaccharyl transferase enzymes encoded

by deep-sea vent proteobacteria. Glycobiology 26, 398–409.

doi:10.1093/glycob/cwv111

237

Morelle, W., Canis, K., Chirat, F., Faid, V., Michalski, J.C., 2006. The use of mass

spectrometry for the proteomic analysis of glycosylation. Proteomics 6, 3993–4015.

doi:10.1002/pmic.200600129

Moremen, K.W., Tiemeyer, M., Nairn, A. V., 2012. Vertebrate protein glycosylation:

diversity, synthesis and function. Nat. Rev. Mol. Cell Biol. 13, 448–462.

doi:10.1038/nrm3383

Morrison, M.J., Imperiali, B., 2014. The renaissance of bacillosamine and its derivatives:

Pathway characterization and implications in pathogenicity. Biochemistry 53, 624–

638. doi:10.1021/bi401546r

Mshelia, G.D., Amin, J.D., Woldehiwet, Z., Murray, R.D., Egwu, G.O., 2010.

Epidemiology of bovine venereal campylobacteriosis: Geographic distribution and

recent advances in molecular diagnostic techniques. Reprod. Domest. Anim. 45.

doi:10.1111/j.1439-0531.2009.01546.x

Mshelia, G.D., Singh, J., Amin, J.D., Woldehiwet, Z., Egwu, G.O., Murray, R.D., 2007.

Bovine venereal campylobacteriosis: an overview. CAB Rev. Perspect. Agric. Vet.

Sci. Nutr. Nat. Resour. 2, 14 pp. doi:10.1079/PAVSNNR20072080

Naegeli, A., Michaud, G., Schubert, M., Lin, C.W., Lizak, C., Darbre, T., Reymond, J.L.,

Aebi, M., 2014. Substrate specificity of cytoplasmic N-glycosyltransferase. J. Biol.

Chem. 289, 24521–24532. doi:10.1074/jbc.M114.579326

Newell, D.G., Fearnley, C., 2003. Sources of Campylobacter colonization in broiler

chickens. Appl. Environ. Microbiol. 69, 4343–4351. doi:10.1128/AEM.69.8.4343

Niedermeyer, T.H.J., Strohalm, M., 2012. mMass as a software tool for the annotation of

cyclic peptide tandem mass spectra. PLoS One 7, e44913.

238

doi:10.1371/journal.pone.0044913

Nielsen, H.L., Engberg, J., Ejlertsen, T., B?cker, R., Nielsen, H., 2012. Short-term and

medium-term clinical outcomes of Campylobacter concisus infection. Clin.

Microbiol. Infect. 18, E459–E465. doi:10.1111/j.1469-0691.2012.03990.x

Noormohamed, A., Fakhr, M.K., 2013. A higher prevalence rate of Campylobacter in

retail beef livers compared to other beef and pork meat cuts. Int. J. Environ. Res.

Public Health 10, 2058–68. doi:10.3390/ijerph10052058

Nothaft, H., Davis, B., Lock, Y.Y., Perez-Munoz, M.E., Vinogradov, E., Walter, J., Coros,

C., Szymanski, C.M., 2016. Engineering the Campylobacter jejuni N-glycan to create

an effective chicken vaccine. Sci. Rep. 6, 26511. doi:10.1038/srep26511

Nothaft, H., Liu, X., McNally, D.J., Li, J., Szymanski, C.M., 2009. Study of free

oligosaccharides derived from the bacterial N-glycosylation pathway. Proc. Natl.

Acad. Sci. U. S. A. 106, 15019–15024. doi:10.1073/pnas.0903078106

Nothaft, H., Liu, X., McNally, D.J., Szymanski, C.M., 2010. N-linked protein

glycosylation in a bacterial system, in: Methods in Molecular Biology (Clifton, N.J.).

pp. 227–243. doi:10.1007/978-1-60761-454-8_16

Nothaft, H., Scott, N.E., Vinogradov, E., Liu, X., Hu, R., Beadle, B., Fodor, C., Miller,

W.G., Li, J., Cordwell, S.J., Szymanski, C.M., 2012. Diversity in the protein n-

glycosylation pathways within the Campylobacter Genus. Mol. Cell. Proteomics 11,

1203–1219. doi:10.1074/mcp.M112.021519

Nothaft, H., Szymanski, C.M., 2013. Bacterial protein N-glycosylation: New perspectives

and applications. J. Biol. Chem. 288, 6912–6920. doi:10.1074/jbc.R112.417857

239

Notomi, T., Okayama, H., Masubuchi, H., Yonekawa, T., Watanabe, K., Amino, N., Hase,

T., 2000. Loop-mediated isothermal amplification of DNA. Nucleic Acids Res. 28,

E63.

O’Donovan, D., Corcoran, G.D., Lucey, B., Sleator, R.D., 2014. Campylobacter

ureolyticus: A portrait of the pathogen. Virulence 1–9. doi:10.4161/viru.28776

Olivier, N.B., Chen, M.M., Behr, J.R., Imperiali, B., 2006. In vitro biosynthesis of UDP-

N,N ’-diacetylbacillosamine by enzymes of the Campylobacter jejuni general protein

glycosylation system. Biochemistry 45, 13659–13669. doi:10.1021/bi061456h

Ollis, A. a, Zhang, S., Fisher, A.C., DeLisa, M.P., 2014. Engineered

oligosaccharyltransferases with greatly relaxed acceptor-site specificity. Nat. Chem.

Biol. 1–9. doi:10.1038/nchembio.1609

On, S.L., Bloch, B., Holmes, B., Hoste, B., Vandamme, P., 1995. Campylobacter

hyointestinalis subsp. lawsonii subsp. nov., isolated from the porcine stomach, and an

emended description of Campylobacter hyointestinalis. Int. J. Syst. Bacteriol. 45,

767–74. doi:10.1099/00207713-45-4-767

On, S.L.W., Atabay, H.I., Corry, J.E.L., Harrington, C.S., Vandamme, P., 1998. Emended

description of Campylobacter sputorum and revision of its infrasubspecific (biovar)

divisions, including C. sputorum biovar paraureolyticus, a urease-producing variant

from cattle and humans. Int. J. Syst. Bacteriol. 48, 195–206. doi:10.1099/00207713-

48-1-195

Oyarzabal, O.A., Fernández, H., 2016. Isolation and identification of Campylobacter spp.

in poultry, in: Campylobacter spp. and related organisms in poultry. Springer

International Publishing, Cham, pp. 19–35. doi:10.1007/978-3-319-29907-5_2

240

Oyarzabal, O.A., Macklin, K.S., Barbaree, J.M., Miller, R.S., 2005. Evaluation of agar

plates for direct enumeration of Campylobacter spp. from poultry carcass rinses.

Appl. Environ. Microbiol. 71, 3351–3354. doi:10.1128/aem.71.6.3351-2254.2005

Oyarzabal, O. a, Battie, C., 2012. Immunological methods for the detection of

Campylobacter spp. - Current applications and potential use in biosensors, in: Trends

in Immunolabelled and Related Techniques. InTech, pp 203-225.

Oyofo, B.A., Thornton, S.A., Burr, D.H., Trust, T.J., Pavlovskis, O.R., Guerryl, P., 1992.

Specific detection of Campylobacter jejuni and Campylobacter coli by using

polymerase chain reaction. J. Clin. Microbiol. 30, 2613–2619.

Palmer, S.R., Gully, P.R., White, J.M., Pearson, A.D., Suckling, W.G., Jones, D.M.,

Rawes, J.C.L., Penner, J.L., 1983. Water-borne outbreak of Campylobacter

gastroenteritis. Lancet 321, 287–290. doi:10.1016/S0140-6736(83)91698-7

Pasternak, J., Bolivar, R., Hopfer, R.L., Fainstein, V., Mills, K., Rios, A., Bodey, G.P.,

Fennell, C.L., Totten, P.A., Stamm, W.E., 1984. Bacteremia caused by

Campylobacter-like organisms in two male homosexuals. Ann. Intern. Med. 101,

339–41.

Patrick, M.E., Gilbert, M.J., Blaser, M.J., Tauxe, R. V., Wagenaar, J. a., Fitzgerald, C.,

2013. Human infections with new subspecies of Campylobacter fetus. Emerg. Infect.

Dis. 19, 1678–1680. doi:10.3201/eid1910.130883

Patton, C.M., Shaffer, N., Edmonds, P., Barrett, T.J., Lambert, M.A., Baker, C., Perlman,

D.M., And, S., Brenner3, D.J., 1989. Human disease associated with Campylobacter

upsaliensis (Catalase-negative or weakly positive Campylobacter species) in the

United States. J. Clin. Microbiol. 66–73.

241

Pauwels, J., Taminiau, B., Janssens, G.P.J., De Beenhouwer, M., Delhalle, L., Daube, G.,

Coopman, F., 2015. Cecal drop reflects the chickens’ cecal microbiome, fecal drop

does not. J. Microbiol. Methods 117, 164–170. doi:10.1016/j.mimet.2015.08.006

Penner, J.L., Hennessy, J.N., Congi, R. V., 1983. Serotyping of Campylobacter jejuni and

Campylobacter coli on the basis of thermostable antigens. Eur. J. Clin. Microbiol. 2,

378–383. doi:10.1007/BF02019474

Petersen, R.F., Harrington, C.S., Kortegaard, H.E., On, S.L.W., 2007. A PCR-DGGE

method for detection and identification of Campylobacter, Helicobacter, Arcobacter

and related Epsilobacteria and its application to saliva samples from humans and

domestic pets. J. Appl. Microbiol. 103, 2601–2615. doi:10.1111/j.1365-

2672.2007.03515.x

Petersen, T.N., Brunak, S., von Heijne, G., Nielsen, H., 2011. SignalP 4.0: discriminating

signal peptides from transmembrane regions. Nat. Methods 8, 785–6.

doi:10.1038/nmeth.1701

Pickett, C.L., Whitehouse, C.A., 1999. The cytolethal distending toxin family. Trends

Microbiol. 7, 292–297. doi:10.1016/S0966-842X(99)01537-1

Pittman, M.S., Elvers, K.T., Lee, L., Jones, M.A., Poole, R.K., Park, S.F., Kelly, D.J.,

2007. Growth of Campylobacter jejuni on nitrate and nitrite: electron transport to

NapA and NrfA via NrfH and distinct roles for NrfA and the globin Cgb in protection

against nitrosative stress. Mol. Microbiol. 63, 575–590. doi:10.1111/j.1365-

2958.2006.05532.x

Platts-Mills, J.A., Liu, J., Gratz, J., Mduma, E., Amour, C., Swai, N., Taniuchi, M.,

Begum, S., Peñataro Yori, P., Tilley, D.H., Lee, G., Shen, Z., Whary, M.T., Fox, J.G.,

McGrath, M., Kosek, M., Haque, R., Houpt, E.R., 2014. Detection of Campylobacter

in stool and determination of significance by culture, enzyme immunoassay, and PCR

242

in developing countries. J. Clin. Microbiol. 52, 1074–80. doi:10.1128/JCM.02935-13

Power, P.M., Roddam, L.F., Rutter, K., Fitzpatrick, S.Z., Srikhanta, Y.N., Jennings, M.P.,

2003. Genetic characterization of pilin glycosylation and phase variation in Neisseria

meningitidis. Mol. Microbiol. 49, 833–47.

Public Health England, 2017. A microbiological survey of Campylobacter contamination

in fresh whole UK-produced chilled chickens at retail sale. FSA Project FS102121.

Qian, H., Pang, E., Du, Q., Chang, J., Dong, J., Say, L.T., Fook, K.N., Ai, L.T., Kwang, J.,

2008. Production of a monoclonal antibody specific for the major outer membrane

protein of Campylobacter jejuni and characterization of the epitope. Appl. Environ.

Microbiol. 74, 833–839. doi:10.1128/AEM.01559-07

Rangarajan, E.S., Bhatia, S., Watson, D.C., Munger, C., Cygler, M., Matte, A., Young,

N.M., 2007. Structural context for protein N-glycosylation in bacteria: The structure

of PEB3, an adhesin from Campylobacter jejuni. Protein Sci. 16, 990–995.

doi:10.1110/ps.062737507

Rosenquist, H., Nielsen, N.L., Sommer, H.M., Nørrung, B., Christensen, B.B., 2003.

Quantitative risk assessment of human campylobacteriosis associated with

thermophilic Campylobacter species in chickens. Int. J. Food Microbiol. 83, 87–103.

doi:10.1016/S0168-1605(02)00317-3

Ruiz-Canada, C., Kelleher, D.J., Gilmore, R., 2009. Cotranslational and posttranslational

N-glycosylation of polypeptides by distinct mammalian OST isoforms. Cell 136,

272–83. doi:10.1016/j.cell.2008.11.047

Rutherford, K., Parkhill, J., Crook, J., Horsnell, T., Rice, P., Rajandream, M.A., Barrell,

B., 2000. Artemis: sequence visualization and annotation. Bioinformatics 16, 944–5.

doi:10.1093/BIOINFORMATICS/16.10.944

243

Sahin, O., Luo, N., Huang, S., Zhang, Q., 2003. Effect of Campylobacter-specific

maternal antibodies on Campylobacter jejuni colonization in young chickens. Appl.

Environ. Microbiol. 69, 5372-5379. doi:10.1128/AEM.69.9.5372

Sajid, M., Kawde, A.-N., Daud, M., 2015. Designs, formats and applications of lateral

flow assay: A literature review. J. Saudi Chem. Soc. 19, 689–705.

doi:10.1016/j.jscs.2014.09.001

Samuelson, J.D., Winter, A.J., 1966. Bovine vibriosis: the nature of the carrier state in the

bull. J. Infect. Dis. 116, 581–92.

Sandstedt, K., Ursing, J., 1991. Description of Campylobacter upsaliensis sp. nov.

previously known as the CNW Group. Syst. Appl. Microbiol. 14, 39–45.

doi:10.1016/S0723-2020(11)80359-0

Sapsford, K.E., Rasooly, A., Taitt, C.R., Ligler, F.S., 2004. Detection of Campylobacter

and Shigella species in food samples using an array biosensor. Anal. Chem. 76, 433–

440. doi:10.1021/ac035122z

Schmidt, T., Venter, E.H., Picard, J. a, 2010. Evaluation of PCR assays for the detection of

Campylobacter fetus in bovine preputial scrapings and the identification of

subspecies in South African field isolates. J. S. Afr. Vet. Assoc. 81, 87–92.

doi:10.4102/jsava.v81i2.111

Schulze, F., Bagon, A., Müller, W., Hotzel, H., 2006. Identification of Campylobacter

fetus subspecies by phenotypic differentiation and PCR. J. Clin. Microbiol. 44, 2019–

24. doi:10.1128/JCM.02566-05

Schwarz, F., Aebi, M., 2011. Mechanisms and principles of N-linked protein

glycosylation. Curr. Opin. Struct. Biol. 21, 576–582. doi:10.1016/j.sbi.2011.08.005

244

Schwarz, F., Lizak, C., Fan, Y.Y., Fleurkens, S., Kowarik, M., Aebi, M., 2011. Relaxed

acceptor site specificity of bacterial oligosaccharyltransferase in vivo. Glycobiology

21, 45–54. doi:10.1093/glycob/cwq130

Scott, N.E., Bogema, D.R., Connolly, A.M., Falconer, L., Djordjevic, S.P., Cordwell, S.J.,

2009. Mass spectrometric characterization of the surface-associated 42 kDa

lipoprotein JlpA as a glycosylated antigen in strains of Campylobacter jejuni. J.

Proteome Res. 8, 4654–4664. doi:10.1021/pr900544x

Scott, N.E., Marzook, N.B., Cain, J. a, Solis, N., Thaysen-Andersen, M., Djordjevic, S.P.,

Packer, N.H., Larsen, M.R., Cordwell, S.J., 2014. Comparative proteomics and

glycoproteomics reveal increased N-linked glycosylation and relaxed sequon

specificity in Campylobacter jejuni NCTC11168 O. J. Proteome Res.

doi:10.1021/pr5005554

Scott, N.E., Nothaft, H., Edwards, A.V.G., Labbate, M., Djordjevic, S.P., Larsen, M.R.,

Szymanski, C.M., Cordwell, S.J., 2012. Modification of the Campylobacter jejuni N-

linked glycan by EptC protein-mediated addition of phosphoethanolamine. J. Biol.

Chem. 287, 29384–29396. doi:10.1074/jbc.M112.380212

Scott, N.E., Parker, B.L., Connolly, A.M., Paulech, J., Edwards, A.V.G., Crossett, B.,

Falconer, L., Kolarich, D., Djordjevic, S.P., Højrup, P., Packer, N.H., Larsen, M.R.,

Cordwell, S.J., 2011. Simultaneous glycan-peptide characterization using hydrophilic

interaction chromatography and parallel fragmentation by CID, higher energy

collisional dissociation, and electron transfer dissociation MS applied to the N-linked

glycoproteome of Campylobacter jejuni. Mol. Cell. Proteomics 10, M000031–

MCP201. doi:10.1074/mcp.M000031-MCP201

Shrimal, S., Cherepanova, N.A., Gilmore, R., 2015. Cotranslational and

posttranslocational N-glycosylation of proteins in the endoplasmic reticulum. Semin.

Cell Dev. Biol. 41, 71–78. doi:10.1016/j.semcdb.2014.11.005

245

Silverman, J.M., Imperiali, B., 2016. Bacterial N-glycosylation efficiency is dependent on

the structural context of target sequons. J. Biol. Chem. 291, 22001–22010.

doi:10.1074/jbc.M116.747121

Simor, A.E., Wilcox, L., 1987. Enteritis associated with Campylobacter laridis. J. Clin.

Microbiol. 25, 10–2.

Skottrup, P.D., Nicolaisen, M., Justesen, A.F., 2008. Towards on-site pathogen detection

using antibody-based sensors. Biosens. Bioelectron. 24, 339–348.

doi:10.1016/j.bios.2008.06.045

Song, E., Pyreddy, S., Mechref, Y., 2013. Quantification of glycopeptides by multiple

reaction monitoring LC-MS/MS. Rapid Commun. Mass Spectrom. 26, 1941–1954.

doi:10.1002/rcm.6290.Quantification

Stanley, D., Geier, M.S., Chen, H., Hughes, R.J., Moore, R.J., 2015. Comparison of fecal

and cecal microbiotas reveals qualitative similarities but quantitative differences.

BMC Microbiol. 15, 51. doi:10.1186/s12866-015-0388-6

Stanley, J., Burnens, A.P., Linton, D., On, S.L.W., Costas, M., Owen, R.J., 1992.

Campylobacter helveticus sp. nov., a new thermophilic species from domestic

animals: characterization, and cloning of a species-specific DNA probe. J. Gen.

Microbiol. 138, 2293–2303. doi:10.1099/00221287-138-11-2293

Stanley, J., Linton, D., Burnens, A.P., Dewhirst, F.E., On, S.L., Porter, A., Owen, R.J.,

Costas, M., 1994. Helicobacter pullorum sp. nov.-genotype and phenotype of a new

species isolated from poultry and from human patients with gastroenteritis.

Microbiology 140 ( Pt 1, 3441–9. doi:10.1099/13500872-140-12-3441

Stimson, E., Virji, M., Makepeace, K., Dell, A., Morris, H.R., Payne, G., Saunders, J.R.,

246

Jennings, M.P., Barker, S., Panico, M., 1995. Meningococcal pilin: a glycoprotein

substituted with digalactosyl 2,4-diacetamido-2,4,6-trideoxyhexose. Mol. Microbiol.

17, 1201–14.

Studier, F.W., Moffatt, B.A., 1986. Use of bacteriophage T7 RNA polymerase to direct

selective high-level expression of cloned genes. J. Mol. Biol. 189, 113–130.

doi:10.1016/0022-2836(86)90385-2

Suo, B., He, Y., Paoli, G., Gehring, A., Tu, S.-I., Shi, X., 2010. Development of an

oligonucleotide-based microarray to detect multiple foodborne pathogens. Mol. Cell.

Probes 24, 77–86. doi:10.1016/j.mcp.2009.10.005

Szymanski, C.M., Burr, D.H., Guerry, P., 2002. Campylobacter protein glycosylation

affects host cell interactions. Infect. Immun. 70, 2242–2244.

doi:10.1128/iai.70.4.2242-2244.2002

Szymanski, C.M., Yao, R.J., Ewing, C.P., Trust, T.J., Guerry, P., 1999. Evidence for a

system of general protein glycosylation in Campylobacter jejuni. Mol. Microbiol. 32,

1022–1030. doi:10.1046/j.1365-2958.1999.01415.x

Tanner, A.C.R., Badger, S., Lai, C.-H., Listgarten, M.A., Visconti, R.A., Socransky, S.S.,

1981. Wolinella gen. nov., Wolinella succinogenes (Vibrio succinogenes Wolin et al.)

comb. nov., and description of Bacteroides gracilis sp. nov., Wolinella recta sp. nov.,

Campylobacter concisus sp. nov., and Eikenella corrodens from humans with

periodontal disease. Int. J. Syst. Bacteriol. 31, 432–445. doi:10.1099/00207713-31-4-

432

Tee, W., Montgomery, J., Dyall‐ Smith, M., 2001. Bacteremia caused by a Helicobacter

pullorum– like organism. Clin. Infect. Dis. 33, 1789–1791. doi:10.1086/323983

247

Thibault, P., Logan, S.M., Kelly, J.F., Brisson, J.-R., Ewing, C.P., Trust, T.J., Guerry, P.,

2001. Identification of the carbohydrate moieties and glycosylation motifs in

Campylobacter jejuni flagellin. J. Biol. Chem. 276, 34862–34870.

doi:10.1074/jbc.M104529200

Thompson, S. a, 2002. Campylobacter surface-layers (S-layers) and immune evasion.

Ann. Periodontol. 7, 43–53. doi:10.1902/annals.2002.7.1.43

Tissari, P., Rautelin, H., 2007. Evaluation of an enzyme immunoassay-based stool antigen

test to detect Campylobacter jejuni and Campylobacter coli. Diagn. Microbiol. Infect.

Dis. 58, 171–175. doi:10.1016/j.diagmicrobio.2006.12.023

Toghi Eshghi, S., Yang, W., Hu, Y., Shah, P., Sun, S., Li, X., Zhang, H., 2016.

Classification of tandem mass spectra for identification of N- and O-linked

glycopeptides. Sci. Rep. 6, 37189. doi:10.1038/srep37189

Tribble, D.R., Baqar, S., Pang, L.W., Mason, C., Houng, H.-S.H., Pitarangsi, C., Lebron,

C., Armstrong, A., Sethabutr, O., Sanders, J.W., 2008. Diagnostic approach to acute

diarrheal illness in a military population on training exercises in Thailand, a region of

Campylobacter hyperendemicity. J. Clin. Microbiol. 46, 1418–25.

doi:10.1128/JCM.02168-07

Troutman, J.M., Imperiali, B., 2009. Campylobacter jejuni PglH is a single active site

processive polymerase that utilizes product inhibition to limit sequential glycosyl

transfer reactions. Biochemistry 48, 2807–2816. doi:10.1021/bi802284d

Truyers, I., Luke, T., Wilson, D., Sargison, N., 2014. Diagnosis and management of

venereal campylobacteriosis in beef cattle. BMC Vet. Res. 10, 280.

doi:10.1186/s12917-014-0280-x

248

Umaraw, P., Prajapati, A., Verma, A.K., Pathak, V., Singh, V.P., 2017. Control of

Campylobacter in poultry industry from farm to poultry processing unit: A review.

Crit. Rev. Food Sci. Nutr. 57, 659–665. doi:10.1080/10408398.2014.935847

van Bergen, M. a P., Dingle, K.E., Maiden, M.J., Newell, D.G., van Der Graaf-Van

Bloois, L., van Putten, J.P.M., Wagenaar, J. a, 2005. Clonal nature of Campylobacter

fetus as defined by multilocus sequence typing. J. Clin. Microbiol. 43, 5888–5898.

doi:10.1128/JCM.43.12.5888

van der Graaf-van Bloois, L., Miller, W.G., Yee, E., Rijnsburger, M., Wagenaar, J.A.,

Duim, B., 2014. Inconsistency of phenotypic and genomic characteristics of

Campylobacter fetus subspecies requires reevaluation of current diagnostics. J. Clin.

Microbiol. 52, 4183–8. doi:10.1128/JCM.01837-14

van der Graaf-van Bloois, L., van Bergen, M. a P., van der Wal, F.J., de Boer, A.G., Duim,

B., Schmidt, T., Wagenaar, J. a, 2013. Evaluation of molecular assays for

identification Campylobacter fetus species and subspecies and development of a C.

fetus specific real-time PCR assay. J. Microbiol. Methods 95, 93–7.

doi:10.1016/j.mimet.2013.06.005

van Doorn, P.A., Ruts, L., Jacobs, B.C., 2008. Clinical features, pathogenesis, and

treatment of Guillain-Barré syndrome. Lancet Neurol. 7, 939–950.

doi:10.1016/S1474-4422(08)70215-1

van Sorge, N.M., Bleumink, N.M.C., van Vliet, S.J., Saeland, E., van der Pol, W.-L., van

Kooyk, Y., van Putten, J.P.M., 2009. N-glycosylated proteins and distinct

lipooligosaccharide glycoforms of Campylobacter jejuni target the human C-type

lectin receptor MGL. Cell. Microbiol. 11, 1768–81. doi:10.1111/j.1462-

5822.2009.01370.x

Vandamme, P., Daneshvar, M.I., Dewhirst, F.E., Paster, B.J., Kersters, K., Goossens, H.,

249

Moss, C.W., 1995. Chemotaxonomic analyses of Bacteroides gracilis and

Bacteroides ureolyticus and reclassification of B. gracilis as Campylobacter gracilis

comb. nov. Int. J. Syst. Bacteriol. 45, 145–52. doi:10.1099/00207713-45-1-145

Vandamme, P., Debruyne, L., De Brandt, E., Falsen, E., 2010. Reclassification of

Bacteroides ureolyticus as Campylobacter ureolyticus comb. nov., and emended

description of the genus Campylobacter. Int. J. Syst. Evol. Microbiol. 60, 2016–2022.

doi:10.1099/ijs.0.017152-0

Veron, M., Chatelain, R., 1973. Taxonomic study of the Genus Campylobacter Sebald and

Veron and designation of the neotype strain for the type species, Campylobacter fetus

(Smith and Taylor) Sebald and Veron. Int. J. Syst. Bacteriol. 23, 122–134.

doi:10.1099/00207713-23-2-122

Vik, A., Aas, F.E., Anonsen, J.H., Bilsborough, S., Schneider, A., Egge-Jacobsen, W.,

Koomey, M., 2009. Broad spectrum O-linked protein glycosylation in the human

pathogen Neisseria gonorrhoeae. Proc. Natl. Acad. Sci. 106, 4447–4452.

doi:10.1073/pnas.0809504106

Wacker, M., Linton, D., Hitchen, P.G., Nita-Lazar, M., Haslam, S.M., North, S.J., Panico,

M., Morris, H.R., Dell, A., Wren, B.W., Aebi, M., 2002. N-linked glycosylation in

Campylobacter jejuni and its functional transfer into E. coli. Science 298, 1790–

1793. doi:10.1126/science.298.5599.1790

Wadl, M., Pölzler, T., Flekna, G., Thompson, L., Slaghuis, J., Köfer, J., Hein, I., Wagner,

M., 2009. Easy-to-use rapid test for direct detection of Campylobacter spp. in

chicken feces. J. Food Prot. 72, 2483–8.

Wagenaar, J.A., van Bergen, M.A.P., Newell, D.G., Grogono-Thomas, R., Duim, B.,

2001. Comparative study using amplified fragment length polymorphism

fingerprinting, PCR genotyping, and phenotyping to differentiate Campylobacter

250

fetus strains isolated from animals. J. Clin. Microbiol. 39, 2283–2286.

doi:10.1128/jcm.39.6.2283-2286.2001

Wagenaar, J. a., Van Bergen, M. a P., Blaser, M.J., Tauxe, R. V., Newell, D.G., Van

Putten, J.P.M., 2014. Campylobacter fetus infections in humans: Exposure and

disease. Clin. Infect. Dis. 58, 1579–1586. doi:10.1093/cid/ciu085

Werno, A.M., Klena, J.D., Shaw, G.M., Murdoch, D.R., 2002. Fatal Case of

Campylobacter lari prosthetic joint infection and bacteremia in an immunocompetent

patient. J. Clin. Microbiol. 40, 1053–1055. doi:10.1128/JCM.40.3.1053–1055.2002

Wolin, M.J., Wolin, E.A., Jacobs, N.J., 1961. Cytochrome-producing anaerobic Vibrio

succinogenes, sp. n. J. Bacteriol. 81, 911–7.

Yachdav, G., Kloppmann, E., Kajan, L., Hecht, M., Goldberg, T., Hamp, T.,

Honigschmid, P., Schafferhans, A., Roos, M., Bernhofer, M., Richter, L., Ashkenazy,

H., Punta, M., Schlessinger, A., Bromberg, Y., Schneider, R., Vriend, G., Sander, C.,

Ben-Tal, N., Rost, B., 2014. PredictProtein--an open resource for online prediction of

protein structural and functional features. Nucleic Acids Res. 42, W337–W343.

doi:10.1093/nar/gku366

Yang, X., Kirsch, J., Simonian, A., 2013. Campylobacter spp. detection in the 21st

century: a review of the recent achievements in biosensor development. J. Microbiol.

Methods 95, 48–56. doi:10.1016/j.mimet.2013.06.023

Yao, R., Burr, D.H., Doig, P., Trust, T.J., Niu, H., Guerry, P., 1994. Isolation of motile

and non-motile insertional mutants of Campylobacter jejuni: The role of motility in

adherence and invasion of eukaryotic cells. Mol. Microbiol. 14, 883–893.

doi:10.1111/j.1365-2958.1994.tb01324.x

251

Young, K.T., Davis, L.M., Dirita, V.J., 2007. Campylobacter jejuni: molecular biology

and pathogenesis. Nat. Rev. Microbiol. 5, 665–679. doi:10.1038/nrmicro1718

Young, N.M., Brisson, J.R., Kelly, J., Watson, D.C., Tessier, L., Lanthier, P.H., Jarrell,

H.C., Cadotte, N., Michael, F.S., Aberg, E., Szymanski, C.M., 2002. Structure of the

N-linked glycan present on multiple glycoproteins in the gram-negative bacterium,

Campylobacter jejuni. J. Biol. Chem. 277, 42530–42539.

doi:10.1074/jbc.M206114200

Yu, C.-S., Chen, Y.-C., Lu, C.-H., Hwang, J.-K., 2006. Prediction of protein subcellular

localization. Proteins Struct. Funct. Bioinforma. 64, 643–651.

doi:10.1002/prot.21018

Zhao, P., Viner, R., Teo, C.F., Boons, G.-J., Horn, D., Wells, L., 2011. Combining high-

energy C-trap dissociation and electron transfer dissociation for protein O-GlcNAc

modification site assignment. J. Proteome Res. 10, 4088–4104.

doi:10.1021/pr2002726

252

Appendix

Table 2.1. Bacterial strains used.

Bacterial strain Reference/Source

Campylobacter jejuni 11168H Karlyshev et al. 2002

C. jejuni 11168gfp4 Jervis et al. 2015

C. jejuni 11168H pgI::kan Linton et al. 2002

C. jejuni 11168H pglA::kan Linton et al. 2002

C. jejuni 11168H pglB::kan Wacker et al. 2002

C. jejuni 11168H pglD::kan Linton et al. 2002

C. jejuni 11168H pglE::kan Linton et al. 2002

C. jejuni 11168H pglF::kan Linton et al. 2002

C. jejuni 11168H pglH::kan Linton et al. 2002

C. jejuni 11168H pglJ::kan Linton et al. 2002

C. jejuni 81116 Palmer et al. 1983

C. jejuni 81-176 Korlath et al. 1985

C. jejuni DW10 This study

C. jejuni DW2 This study

C. jejuni DW3 This study

C. jejuni DW7 This study

C. jejuni DW8 This study

C. jejuni DW9 This study

C. jejuni Guillain-Barré strain G2 (Linton et al., 2000)

C. jejuni Miller Fisher strain F1 (Linton et al., 2000)

C. jejuni Miller Fisher strain F2 (Linton et al., 2000)

C. jejuni Miller Fisher strain F3 (Linton et al., 2000)

C. jejuni Penner serotype reference strain P1 Penner et al. 1983

C. jejuni Penner serotype reference strain P10 Penner et al. 1983

C. jejuni Penner serotype reference strain P19 Penner et al. 1983

C. jejuni Penner serotype reference strain P3 Penner et al. 1983

C. jejuni Penner serotype reference strain P4 Penner et al. 1983

Campylobacter coli DW1 This study

C. coli DW4 This study

C. coli DW5 This study

C. coli DW6 This study

C. coli RM2228 Veron and Chatelain 1973

Campylobacter concisus 13826 Tanner et al. 1981

Campylobacter fetus clinical isolate 1* PHE

Campylobacter fetus clinical isolate 10# PHE

Campylobacter fetus clinical isolate 11~ PHE

Campylobacter fetus clinical isolate 12+ PHE

Campylobacter fetus clinical isolate 13# PHE

Campylobacter fetus clinical isolate 14# PHE

253

Campylobacter fetus clinical isolate 15+ PHE

Campylobacter fetus clinical isolate 16# PHE

Campylobacter fetus clinical isolate 17~

PHE

Campylobacter fetus clinical isolate 19 PHE

Campylobacter fetus clinical isolate 2+ PHE

Campylobacter fetus clinical isolate 3+ PHE

Campylobacter fetus clinical isolate 4+ PHE

Campylobacter fetus clinical isolate 5 PHE

Campylobacter fetus clinical isolate 7+

PHE

Campylobacter fetus clinical isolate 8#

PHE

Campylobacter fetus clinical isolate 9+

PHE

Campylobacter fetus subsp. fetus NCTC 10842 Veron and Chatelain 1973

Campylobacter fetus subsp. fetus NCTC 10842

pglB::kan

Jervis et al. 2012

Campylobacter fetus subsp. fetus clinical isolate 18! PHE

Campylobacter fetus subsp. fetus clinical isolate 6+ PHE

Campylobacter fetus subsp. venerealis NCTC 10354

Veron and Chatelain 1973

Campylobacter helveticus NCTC 12470 Stanley et al. 1992

Campylobacter hyointestinalis subsp. hyointestinalis

NCTC 11608

Gebhart et al. 1985

Campylobacter lanienae NCTC 13004 Logan et al. 2000

Campylobacter lari RM2100 Benjamin et al. 1983

Campylobacter sputorum subsp. sputorum NCTC 11528 Veron and Chatelain 1973

Campylobacter upsaliensis NCTC 11541 Sandstedt and Ursing 1991

Campylobacter upsaliensis RM3195 Fouts et al. 2005

E. coli BL21 (λDE3) Studier and Moffatt 1986

E. coli S17-1 λpir de Lorenzo and Timmis

1994 E. coli XL-1 Blue Agilent

PHE; Public Health England. For clinical isolates, symbol indicates site of origin:

*Tissue

+Blood

#Faeces

~Hip aspirate

!Shoulder.

254

Appendix Table 2. Plasmids used.

Plasmid Description Resistance Reference

pRYGG1 sapA108 promoter in multiple cloning

region of pRY111 Chlor

Kienesberger et al.

2007

pG1-N

pRYGG1 containing bases 1 to 561 of

cj0114 plus eight 3’ histidine codons

cloned at the BamHI and SmaI sites.

Chlor This study

pET22 (+)b

pET system vector carrying an N-

terminal pelB signal sequence and a C-

terminal His-Tag®.

Amp Novagen

pNGRP

pET22(+)b with bases 73 to 561 of

cj0114 cloned between NcoI and XhoI

sites.

Amp Jervis & Linton,

unpublished

pNH1 pNGRP with bases -105 to 1053 of C.

fetus pglH1 cloned at the BglII site Amp This study

pNH2 pNGRP with bases -77 to 1046 of C.

fetus pglH2 cloned at the BglII site Amp This study

pN1389 pNGRP with bases -98 to 1010 of C.

fetus cf1389 cloned at the BglII site Amp This study

pNCjH pNGRP with bases -136 to 1080 of C.

jejuni pglH cloned at the BglII site Amp This study

pNH1H2

pNGRP with bases -105 to 1053 and 1

to 1046 of C. fetus pglH1 and pglH2,

respectively, cloned at the BglII site

Amp This study

pNH1-1389 pNH1 with bases -98 to 1010 of C. fetus

cf1389 cloned at the BglII site Amp This study

pNX12

pNGRP with bases -105 to 1053 and 1

to 1046 of C. fetus pglH1 and pglH2,

respectively, and bases -98 to 1010 of C.

fetus cf1389 cloned at the BglII site

Amp This study

ppgl pACYC containing the 16 kb C. jejuni

pgl locus. Chlor Wacker et al. 2002

ppglpglH::kn ppgl with a kanamycin cassette inserted

within pglH.

Chlor

Kan Linton et al. 2005

p0445

pRYGG1 with the ORF of C. fetus

cf0445 plus eight 3’ histidine codons

cloned at the BamHI and SmaI sites.

Chlor This study

p0781

pRYGG1 with the ORF of C. fetus

cf0781 plus eight 3’ histidine codons

cloned at the BamHI and SmaI sites.

Chlor This study

Chlor; chloramphenicol. Amp; ampicillin. Kan; kanamycin.