Journal of Applied Bacteriology 1988,65,69-78 270911 2/87

Numerical analysis of electrophoretic protein patterns of Campylobacter laridis and allied thermophilic campylobacters from the natural environment

R.J. OWEN*, M . COSTAS, LESLEY SLOSS & F.J. BoLToNt National Collection of Type Cultures, Central Public Health Laboratory, London N W 9 5HT and ?Public Health Laboratory, Royal Injirmary, Preston, Lancashire PRI 6PS, U K

Received 23 December 1987 revised 14 March 1988 and accepted 19 March 198%

OWEN, R.J., COSTAS, M., SLOSS, L. & BOLTON, F.J. 1988. Numerical analysis of electrophoretic protein patterns of Campylobacter laridis and allied thermophilic campylobacters from the natural environment. Journal of Applied Bacteriology 65, 69-78.

Twenty-one strains comprising Campylobacter laridis (nine), nalidixic acid sensitive campylobacters (NASC) (four), and urease-positive thermophilic campylobacters (UPTC) (eight) were characterized by one-dimensional SDS-PAGE of cellular pro- teins. The UPTC and NASC strains included six from river water, two from mussels and four from sea water. The type strains of three other Campylobacter species were included for reference. The protein patterns, which contained 45-50 discrete bands, were highly reproducible and were used as the basis for two numerical analyses. In the first, which included all the protein bands, the 21 strains formed nine clusters at the 80% similarity (S) level. The typical C. laridis strains were restricted to two phenons (2 and 5 ) ; the atypical strains being distributed among the remaining phenons. In the second analysis, which excluded the principal protein bands (W 48.5 kD range), the 21 strains formed five clusters at the 80% S level. The typical C. laridis strains were relatively homogeneous and fell into a single phenon (2) within which two subgroups were discernable. The atypical strains were more heter- ogeneous with respect to background protein pattern, with representatives appear- ing in all five phenons. An electropherotyping scheme comprising six electropherotypes, and based on both analyses is proposed. The high within-group S level and separation from reference strains of Campylobacter in the second analysis, suggested that UPTC and NASC strains belonged within C. laridis poss- ibly as biovars.

The species Campylobacter laridis (Benjamin et al., 1983), was described for nalidixic acid- resistant thermophilic campylobacters (NARTC), first isolated from the cloaca1 swabs of wild seagulls of the genus Larus (Skirrow & Benjamin 1980). The organism was isolated infrequently from man and other avian and mammalian species (Benjamin et al., 1983; Karmali & Skirrow 1984), but several cases of illness linked to C. laridis have been reported

* Corresponding author.

indicating that the species may be an enteric pathogen for the normal and immunosuppres- sed human host (Nachamkin et al. 1984; Tauxe et al. 1985; Simor & Wilcox 1987; Borczyk et al. 1987).

The isolation of C. laridis and atypical bio- types has also been reported from river water in north-west England (Bolton et al. 1985, 1987) as well as from marine sources including sea water, cockles, and mussels (Bolton et al. 1985). The atypical organisms differ from C. laridis either in producing urease, in which case they are

70 R . J. Owen et al. Table 1. Sources of Campylobacter strains studied

Reference no. Cumpylohacter in phenogram specieslgroup Reference no(s) Source of isolation

1 2 3 4 5

6 7 8 9

10 I I 12 13

14 15 16 17

18 19 20 21 22 23 24 25 26 27

NASC UPTC NASC UPTC U PTC

U PTC C. jejuni C. coli c. fetus

UPTC UPTC UPTC U PTC

C. laridis NASC C. faridis C. laridis

U PTC NASC C. laridis C. laridis C. laridis C. laridis C. laridis C. luridis U PTC UPTC

subsp. feius

A697185; 2991/F3/82 A689/85; 10231/6/83 A696j85; 1064/83 NCTC 11845; 12830/1/82 A691/85 (same strain

as no. 4 above A692/85; 3980/1/83 NCTC 11351' NCTC 1 1366T NCTC 10842T

A694/85; 1547/1/85 NCTC 11937; A1/9/82 A688185; 2991/F1/82 A687/85 (same strain

as no. 11 above) NCTC 11457 A695/85; 3979183 A2518 1 A2718 1

A693/85; 1295185 NCTC 11844; 3980/SW/82 A21/81 A24181 NCTC 11352' NCTC 11458 A22181 A79182 NCTC 11928; A1/1/82 A686185 (same strain

as no. 26 above)

River water River water Sea water River water

Sea water Bovine, faeces Pig, faeces Sheep, fetus, brain

Mussels River water River water

Dog, faeces Sea water Duck, faeces Foal with diarrhoea,

faeces Mussels Sea water Simian, healthy Cow, faeces Gull, cloaca1 swab Child, faeces Water, park lake Child, faeces River water

NCTC, National Collection of Type Cultures, Central Public Health Laboratory, London, U K ; all other strains prefixed by the letter A are field isolates, with alternative numbers. T, Type strain.

referred to as the urease-positive thermophilic campylobacter (UPTC) group (Bolton et al. 1985) or in being sensitive to nalidixic acid (Bolton et al. 1987) (herein designated the NASC group).

The purpose of the present study was to investigate the use of high-resolution polyacry- lamide gel electrophoresis (PAGE) of proteins combined with computerized analysis of the profiles as a means of characterizing both typical and atypical C. laridis strains because the relationships of the latter to C. laridis are unclear. As shown previously (Costas & Owen 1983; Costas et al., 1987a), protein pattern com- puterized analysis is a valuable approach for establishing similarities within and between Campylobacter species and avoids the subjectiv- ity of conventional visual comparisons of

PAGE patterns which have been used in a number of studies of other Campylobacter species (Morris & Park 1973; Hanna et al., 1983; Ferguson & Lambe 1984; Pearson et al. 1984; Megraud et al. 1985).

Materials and Methods

B A C T E R I A L S T R A I N S

The 21 strains of Campylobacter laridis and allied (UPTC and NASC) campylobacters (including the type strain NCTC 11352) and three reference strains of other Campylobacter species used are listed in Table 1 with their alternative strain numbers and sources. Conven- tional test results on these strains were described previously (Benjamin et al. 1983;

Protein patterns of C . laridis 71 Bolton et al. 1985). Details on the isolation of UPTC and nalidixic acid-sensitive C . laridis strains were described by Bolton et ul. (1987).

M E D I U M A N D C U L T U R E M E T H O D S

All strains were grown on 5% (v/v) defibrinated horse blood agar. Cultures were incubated for 48 h at 37°C under microaerophilic conditions (about 5% oxygen) attained by incubating in anaerobic jars (without catalyst) from which 75% of the air had been withdrawn (580 mmHg below atmospheric pressure) and replaced with an anaerobic gas mixture containing 10% hydrogen, 10% carbon dioxide and 80% nitro. gen.

P R E P A R A T I O N OF P R O T E I N S A M P L E S

For each protein sample, approximately 0.05- 0.1 g wet weight of the bacteria were harvested by centrifugation at 13 OOO rev/min (11 600 g) for 10 min. The pellet was washed twice in sterile distilled water and finally resuspended in about 60 p1 of double strength lysis buffer [20% v/v glycerol, 2% v/v 2-mercaptoethanol, 4% w/v sodium dodecyl sulphate (SDS) and 70% v/v stacking gel buffer]. The protein samples were then extracted as described by Costas et al. (1987 a,b).

E L E C T R O P H O R E S I S

Samples were run on discontinuous SDS- polyacrylamide gels which were cast to allow for a 10 mm stacking gel. The final polyacrylamide concentrations were 10% w/v for the separation gel and 5%) w/v for the stacking gel. Full details of the methods used in gel preparation and elec- trophoresis are described by Costas et al. (1987 ah).

S C A N N I N G OF GELS

The stained protein patterns in the dried gels were scanned (LKB Ultroscan 2202 laser densitometer) at a scan speed of 20 cm/min and an absorbance range 0 to 1. The absorbance ( Y ) values were recorded on magnetic disk (Apple Duodisk) as integers between 0 and 1000 using an interface program SCAN (Heyden Datasy- stems Ltd, Hendon, London) which was amended to incorporate an XVIA real time

clock (Xcalibur Computers Ltd, Northampton) enabling a fixed number of data points to be collected by an Apple IIe microcomputer (500 data points were collected for each 10 cm trace). The retention times and total areas of each protein band were calculated using an LKB 2220 recording integrator.

C O M P U T A T I O N

The raw data were aligned by computer on the initial (stacking gel/separation gel) interface. This band and the final (bromophenol blue marker) bands were then deleted using Vidi- chart (Heyden Datasystems Ltd). The number of data points in each trace was reduced to a standard 460 point trace after subtraction of the initial and final bands. A general background trend in each trace was removed to increase dis- crimination between patterns, The background cut-off was set at 0.6. Similarity between all pairs of standard traces was calculated using the Pearson product moment correlation coefficient, and expressed as a percentage for convenience. Strains were then clustered by the method of unweighted pair group average linkage (UPGMA). A second analysis was performed to exclude major protein bands in the 4M8.5 kD range. The analysis was based on 393 data points in this case. Background trend removal, calculation of similarity and clustering were all carried out on an Apple IIe microcomputer using programs T MATRIX and T CLUSTER, adapted from PET Basic to APPLESOFT Basic. Full details of these programs are given elsewhere (Jackman et al. 1983).

Results

G E N E R A L F E A T U R E S OF P A G E - P R O T E I N P A T T E R N S

One-dimensional SDS-PAGE of whole cell protein extracts of 21 typical and atypical C. laridis strains produced patterns containing 40 to 45 discrete bands with molecular weights of 18-90 kD (Fig. 1). Differences among strains in the total pattern were evident principally in the major protein bands corresponding to molecu- lar sizes in the range 4 M 8 . 5 kD; these accounted for up to 20% of the total protein content. There were also small but discernable differences among profiles in the other protein

72 R . J. Owen et al.

13 12 1 1 109 8 7 X 6 5 4 3 2 I 27 26 25 24 23 22 21 20 19 18 17 X 16 15 14

Fig. 1. Electrophoretic protein patterns of Campylobacter laridis, UPTC and NASC groups, and reference strains of other Campylobacter species. The numbers refer to those used in Table 1, Fig. 2 and Fig. 3. Molecular weight markers (track labelled X) are (from top to bottom): ovotransferrin 7 6 7 8 kD; albumin, 66.25 kD; ovalbumin, 45 kD; carbonic anhydrase, 30 kD; myoglobin, 17.2 kD. The brackets on the vertical axis indicate the principal protein band region ( U 8 . 5 kD range) omitted in cluster analysis 2.

bands, which constituted the background pattern.

R E P R O D U C I B I L I T Y

The protein patterns of the Campylobacter luridis strains were highly reproducible, both within and between gels. Duplicate protein samples from three strains (NCTC 11845, NCTC 11928 and NCTC 11937; see Figs 2 and 3) electrophoresed on two different gels were compared, and their computed average simi- larities based on the whole patterns were in excellent agreement with a mean and S.D. of 96.3 f 0.9%. Average similarity between these samples based on background pattern analysis was 94.0 0.0%. Molecular weight protein standards were also included in each gel, and in all cases estimates of their similarity were 96% although they provided a less objective measure of reproducibility as they were based on just four bands.

N U M E R I C A L A N A L Y S I S

Numerical analysis of PAGE total protein pro- files (analysis 1) based on the determination of the Pearson correlation coeflicient and UPGMA clustering revealed nine distinct clus- ters comprising the 21 strains of C. laridis

(typical and atypical strains) at the 80% simi- larity (S) level, as shown in Fig. 2. The molecu- lar weights of the principal protein bands found in each phenon were as follows: phenon 1 = 41.1 f 0.04 kD; 2 = 41.6 & 0.1 kD; 3 = 41.5 f 0.07 kD; 4 = 41.1 and 42.6 kD; 5 = 42.6 f 0.0 kD; 6 = 42.6 f 0.0 kD; 7 = 44.7 f 0.1 kD; 8 = 43.4 f 0.0 and 44.7 0.0 kD; 9 = 46.5 kD. The average between- and within-group similarities of the clusters are listed in Table 2. Phenons 2 and 5 contained biochemically typical strains of C. laridis, whereas the other phenons contained the atypical strains. There was no clear separation between the UPTC and NASC strains as several were in the same phenon (see phenons 1 and 7). The main anomoly in this analysis was the fact that the C. jejuni type strain, which has low DNA relatedness to C. laridis (Benjamin et al. 1983), was linked at 73% S to C. laridis phenons 5 and 6, although the reference strains of C. coli and C.fetus were well separated from C. laridis. The molecular weight of the principal protein band in the C. jejuni profile correspond closely to that of the C. laridis phenon 3 , 4 and 5 strains, which resulted in a high computed simi- larity between the two species.

To avoid possible distortion caused by weight- ing of the principal protein bands and improve

Phenon

I [

c. coli a C. 9 - fetus 9

2

I I l I I I I I I I l 1 1 l 1 1 l 1 I I 1 1 1 I 1 1 1 1 1 1

6::

3[

4 -

Protein patterns of C. laridis * 73

C jejuni

3-

4 -

5 -

I / -

12.

13-

the discrimination between species, a second numerical analysis was carried out based on the background pattern only (analysis 2), which

excluded bands in the 40-48.5 kD region. The analysis produced just five phenons at the 80% S level with lower (< 75% S) average similarities

74 Phenon

2 5 -

24

19

2.

4 .

1. 5 -

1 1 .

.i 0 ::: cieiuni c COll 8

C fetus 9 -

R. J . Owen et al.

i : q -

I 5 : t

7::

J 1 I . i I _ - L 1 - l _ l L I I 1 1 1 1 I I I I I I I I I I I I I I I I00 90 80 70 60 50 40

to the reference strains of C. coli, C. jejuni and Phenon 1 contained five strains; NCTC C. fetus subsp. fetus (Fig. 3). The average 11928 (in duplicate samples) and two other between- and within-group similarities of the UPTC field strains, and two NASC field strains. clusters are listed in Table 3. The phenons com- Sources of isolates were river water and marine. prised the following strains (from the top of the Phenon 2 contained 10 strains which were all phenogram). typical members of C. [aridis except for one

Tabl

e 2.

Mea

n in

tra-

and

inte

r-ph

enon

per

cen

t sim

ilarit

ies

for a

naly

sis

1 as

det

erm

ined

by

the

Pear

son

prod

uct m

omen

t cor

rela

tion

coef

ficie

nt (r

) and

UPG

MA

clu

ster

ing.

All

phen

ons

are

wrf

orm

ed a

t the

79%

S le

vel

Cam

pylo

bact

er

Cam

pylo

bact

er

Cam

pylo

bact

er

Phen

on 1

Ph

enon

2

Phen

on 3

Ph

enon

4

Phen

on 5

Ph

enon

6

Phen

on 7

Ph

enon

8

Phen

on 9

je

juni

’ co

li’

fetu

sT

(n =

3)

(n =

5)

(n =

2)

(n =

1)

(n =

4)

(n =

2)

(n =

3)

(n =

3)

(n =

1)

(n =

1)

(n =

1)

(n =

1)

Phen

on 1

86

.0 f 1

.4*

Phen

on 2

77

.6 f 5

.1

84.7

f 3.

7 Ph

enon

3

77.5

f 1

.2

78.5

f 3

.3

89

Phen

on 4

67

.7 f 0.5

68.4

k 3.

4 65

.5 f 1

.5

(100

) Ph

enon

5

61.5

f 5

.4

75.8

f 5

.7

70.8

f 3.

6 67

.3 &

2.6

88

.8 k 2.

2 Ph

enon

6

61.0

f 5

.7

64.2

f 7.

0 66

.3 f 4.

7 61

.0

0.0

77.1

5.

6 91

0

Phen

on 8

39

.2 f 6.

7 49

.7 f 4.

0 44

.3 f 2

.9

61.3

k 3

.3

61.9

k 4

.4

56.5 f 1

.5

19.3

k 3

.4

95.3

2 0

.5

“c

Cam

pylo

bact

er je

juni

T 46

.6 _+

2.3

62

.6 f 5.0

58.5

k 4

.5

58

74.7

k 1

.6

65.5

k 1

.5

64.3

f 3

.4

69.0

k 1

.6

51

(100

) C

ampy

loba

cter

col

iT

33.6

f 4

-0

45.0

f 3.

4 41

.5

4.5

44

48.5

5 6

.2

30.0

f 1

.0

72.3

f 0.

9 63

.3 -t

1.2

77

53

(100)

Cam

pylo

bact

erfe

tusT

18

.0 f 7.

1 32

.6 k

5.3

22

.0 f 1

.0

34

44.5

k 3.

3 26

.5 k

1.5

59

.6 f 4.

8 44.0 k

0.8

70

55

64

(100

)

2 Ph

enon

9

46.3

f 4.

8 52

.0 f 1

.8

41.5

f 1

.5

53

51.5

k 5

.4

41.5

f 0.

5 75

.6 k 2.

9 56

.3 k

2.0

(1

00)

$. x 3 5 % 9

5. g

Phen

on 7

48

.5 f 7.

2 55

.5 f 2

.1

48.0

k 4.

0 62

.0 f 4.

5 57

.5 k

5.1

44

.3 f 4.

1 89

.0 f 5

.7

c.+

* Mea

n an

d st

anda

rd d

evia

tion

of si

mila

rity

estim

ates

. N

umbe

r in

par

enth

eses

indi

cate

onl

y on

e st

rain

incl

uded

in th

e ph

enon

; n, n

umbe

r of

str

ains

in e

ach

phen

on; T

, typ

e st

rain

.

- Ta

ble

3. M

ean

intr

a- a

nd i

nter

-phe

non

per

cent

sim

ilarit

ies

for

anal

ysis

2 a

s de

term

ined

by

the

Pear

son

prod

uct

mom

ent

corr

elat

ion

coef

ficie

nt (

r) an

d U

PGM

A c

lust

erin

g. A

ll ph

enon

s ar

e fo

rmed

at t

he 8

1% S

leve

l

Cam

pylo

bact

er

Cam

pylo

bact

er

Cam

pylo

bact

er

Phen

on 1

Ph

enon

2

Phen

on 3

Ph

enon

4

Phen

on 5

je

juni

T co

iiT

fetu

sT

(n =

6)

(n =

10)

(n

= 2

) (n

= 3

) (n =

3)

(n =

1)

(n =

1)

(n =

1)

Phen

on 1

84

.3 f 4

.7*

Phen

on 2

74

.5 i- 6

.6

83.1

f 3

.3

Phen

on 3

68

.0 &

4.8

73

.1

3.6

85

Phen

on 4

77

.8 i-

5.1

15

.7 k 3.

1 70

.2

1.6

92.3

f 1

.2

Phen

on 5

67

.1 f 5

.7

64.9

f 4

.3

64.2

k 4.

0 79

.6 f 3

.5

94.0

f 0

.8

Cam

pylo

bact

er je

juni

T 60

.3 &

4.4

64

.3 _

+ 2.

5 55

.0 i-

0.0

66

.3 k 2.

0 56

.7 _+

2.6

(1

00)

Cnm

pylo

bact

er co

liT

64.3

k 6.

5 69

.8 f 3

.9

61.5

f 3.

5 61

-6

2.0

62.3

f 1

.2

73

(100

) C

ampy

loba

cter

fetu

sT

42.0

& 3

-0

46.1

i- 4

.6

36.5

f 1

-5

49.0

f 2

.2

36.6

f 2

.5

61

50

(100

)

* As

Tab

le 2.

4

ul

76 R. J . Owen et al. Table 4. Protein electropherotypes (PE-types) within Campylobacter luridis

Phenon designation No. of Name of C. laridis Reference

PE-type NA-1* NA-2t strains subgroup strain Ia 5 2b 4 biovar typical NCTC 11352T Ib 1 2b 1 biovar UPTC A693185 I1 2 2a 5 biovar typical NCTC 11457 I11 3 3 2 biovar NASC NCTC 11844 IV 8 5 2 biovar UPTC NCTC 11937 V 2, 7 4 2 biovar UPTC NCTC 11845 VI 1, 4, 6, 7, 9 1 5 biovars NASC and UPTC NCTC 11928

* Numerical analysis based on complete protein pattern. t Numerical analysis based on background pattern.

UPTC field strain from a mussel. The phenog- ram showed there were two subclusters within the strains at the 82% S level. All five strains in subcluster 2a which included NCTC 11457 were from bird or mammalian sources. Subcluster 2b contained NCTC 11352T, NCTC 11458, three other typical C . laridis isolates and the UPTC strain.

Phenon 3 contained NCTC 11844 and one field strain, both of which were members of the NASC group and isolated from sea water.

Phenon 4 contained two UPTC strains; these were NCTC 11845 (in duplicate samples) and a field strain; both were from river water.

Phenon 5 contained two UPTC strains; NCTC 11937 (in duplicate samples) and a field strain; both were from river water.

Discussion

Total protein patterns provide a reliable means of characterizing bacterial strains of clinical and environmental interest, especially those for which there are no other widely available typing methods. Some method of typing strains of C. luridis is required to clarify the relationship between illness in man and the presence of this organism in natural aquatic environments. Lior (1984) recognized two possible biotypes of C. laridis based on ability to hydrolyse DNA, and more recently utilized serotyping to characterize isolates in a waterborne gastroenteritis outbreak in Canada (Borczyk et al. 1987). Our study has shown that, on the basis of protein patterns, the atypical (urease-positive or nalidixic acid- sensitive) strains of C . laridis, originally described by Bolton and colleagues (Bolton el al. 1985, 1987) and isolated from fresh water and

marine sources, overall have a high level of average similarity (2 71% S) to typical members of C. laridis. The indication that these atypical strains are members of C. laridis is supported by similar DNA base compositions of 31 mol % and positive dot-blot hybridization tests (Owen & Ahmed, unpublished results). The protein patterns of the C. laridis strains were quite dis- tinct from those of both C. fetus subsp. fetus and C . coEi but their distinction from those of C. jejuni was less clear cut and depended on the type of data used for the numerical analysis.

On the basis of both the protein pattern data analyses, several distinct protein electro- pherotypes (PE-types) within the C . laridis cluster were identified (Table 4). The four strains designated PE-type. Ia were biochemically typical members of C. laridis sensu strict0 (Benjamin et al. 1983) and formed a homoge- neous cluster in numerical analyses 1 and 2. Carnpylobacter laridis NCTC 11352T and the two strains isolated from children were electro- pherotype type Ia and previous DNA homology data showed they were closely related (72% DNA similarity) (Benjamin et al. 1983). In analysis 2, which was based on background pat- terns, a UPTC mussel isolate (no. 18 in the phenograms) clustered with these C. laridis strains and is designated PE-type Ib. However, it clustered differently in analysis 1 (see Phenon 1). The five strains that were designated PE-type 11, were also biochemically typical members of C. laridis, had high levels of DNA homology to the type strain (Benjamin et al. 1983) and formed discrete clusters in both numerical analyses. The reference strain for this PE-type is NCTC 11457.

The biochemically atypical C. laridis strains were more heterogeneous in their protein pat-

Protein patterns of C . laridis 77 terns and clustered in several phenons. Two NASC strains (nos 15 and 19 in the phenograms) clustered together in both numeri- cal analyses and were designated PE-type 111; NCTC 11844 is the reference strain. Two UPTC strains also clustered together in both analyses and were designated PE-type IV, with NCTC 11937 as the reference strain. The remaining seven strains clustered differently depending on the analysis used but for the purposes of the present study were designated PE-types V and VI based on the background pattern analysis. When the total pattern data was used, however, they proved to be extremely heterogeneous and fell in six different clusters linked at 72% S or above. The apparent anomolous grouping of C. jejuni in Fig. 2, and the above discrepancies between phenograms derived from total and background patterns in the case of some strains, suggest further investigation is needed into the development of alternative coefficients for expressing protein pattern similarity. The prob- lems highlighted in this study using the Pearson correlation coefficient for comparing protein patterns, that contain a high proportion of the total protein in a limited number of bands, were also encountered in a previous study of Proui- dencia species (Costas et al. 1987b).

The results indicate that the urease-positive and nalidixic acid-sensitive thermophilic campy- lobacters from the natural environment are either distinct species closely related to C. laridis, or are subspecies or biovars of C. laridis. In previous studies of catalase-negative campy- lobacters, we found that PAGE-protein group- ings show excellent agreement with species based on other data (Costas et al. 1987a). The strains of C. laridis sensu stricto formed discrete clusters in the present study, but neither of the numerical analyses show unequivocally that the NASC and UPTC strains constituted two separate definable entities and we suggest that they are considered biovars of C. laridis. The correlations between the conventional bio- chemical properties, PE-typing and DNA char- acteristics of these strains, and proposals for a revised description to accommodate them in C. laridis will be discussed elsewhere.

This research was carried out in the framework of contract no. BP. 0133 UK of the Biotech- nology Action Programme of the Commission of the European Communities.

References BENJAMIN, J., LEAPER, S., OWEN, R.J. & SKIRROW,

M.B. 1983 Description of Campylobacter laridis, a new species comprising the nalidixic acid resistant thermophilic Campylobacter (NARTC) group. Current Microbiology 8, 231-238.

BOLTON, F.J., HOLT, A.V. & HUTCHINSON, D.N. 1985 Urease-positive thermophilic campylobacters. Lancet i, 1217-1218.

BOLTON, F.J., COATES, D., HUTCHINSON, D.N. & GODFREE, A.F. 1987 A study of thermophilic campylobacters in a river system. Journal of Applied Bacteriology 62, 167-176.

BORCZYK, A,, THOMPSON, J.S., SMITH, D., LIOR, H. & DEVITT, B. 1987 Water-borne outbreak of Campylo- bacter laridis-associated gastroenteritis in Ontario, Canada. In Proceedings of the IVth International Workshop on Campylobacter Infections, Abstract No. 209. Goteborg, Sweden.

COSTAS, M. & OWEN, R.J. 1983 The classification and identification of campylobacters using total protein profiles. In Campylobacter 11, Proceedings of the Second International Workshop on Campylobacter Infections ed. Pearson, A.D., Skirrow, M.B., Rowe, B., Davies, J.R. & Jones, D.M. pp. 40. London: Public Health Laboratory Service.

COSTAS, M., OWEN, R.J. & JACKMAN, P.J.H. 1987a Classification of Campylobacter sputorum and allied campylobacters based on numerical analysis of elec- trophoretic protein patterns. Systematic and Applied Microbiology 9, 125-131.

COSTAS, M., HOLMES, B. & SLOSS, L.L. 1987b Numeri- cal analysis of electrophoretic protein patterns of Prooidencia rustigianii strains from human diar- rhoea and other sources. Journal of Applied Bacte- riology 63, 319-328.

FERCUWN JR., D.A. & LAMBE JR., D.W. 1984 Differen- tiation of Campylobacter species by protein banding patterns in polyacrylamide slab gels. Journal of Clinical Microbiology 20,453460.

HANNA, J., NEILL, S.D., O’BRIEN, J.J. & ELLIS, W.A. 1983 Comparison of aerotolerant and reference strains of Campylobacter species by polyacrylamide gel electrophoresis. International Journal of Sys- temic Bacteriology 33, 143-146.

JACKMAN, P.J.H., FELTHAM, R.K.A. & SNEATH, P.H.A. 1983 A program in BASIC for numerical taxonomy of microorganisms based on electrophoretic protein patterns. Microbios Letters 23, 87-93.

KARMALI, M.A. & SKIRROW, M.B. 1984 Taxonomy of the genus Campylobacter. In Campylobacter Infec- tion in Man and Animals ed. Butzler, J.P. pp. 1-20. Boca Raton, Fla: CRC Press.

LIOR, H. 1984 New, extended biotyping scheme for Campylobacter jejuni, Campylobacter coli and ‘Campylobacter laridis’. Journal of Clinical Micro- biology 20,636-640.

MEGRAUD, F., BONNET, F., GARNIER, M. & LABOU- LIATTE, H. 1985 Characterization of ‘Campylobacter pyloridis’ by culture, enzymatic profile, and protein content. Journal of Clinical Microbiology 22, 1007- 1010.

MORRIS, J.A. & PARK, R.W. 1973 A comparison using gel electrophoresis of cell proteins of campylobac-

78 R . J . Owen et al. ters (vibrios) associated with infertility, abortion and swine dysentry. Journal of General Micro- biology 78, 165-178.

A.M., R o o p 11, R.M. & SMIBERT, R.M. 1984 Campy- lobacter laridis causing bacteremia in an immuno- compromised patient. Annals of Internal Medicine 101.55-57.

PEARSON, A.D., BAMFORTH, J., BOOTH, L., HOLDSTOCK, G., IRELAND, A,, WALKER, C., HAWTIN, P. &

electrophoresis of spiral bacteria from the gastric antrum. Lancet i, 1349-1350.

NACHAMKIN, I., STOWELL, c., SKALINA, D., JONES,

MILLWARD-SADLER, H. 1984 Polyacrylamide gel

SIMOR, A.E. & WILCOX, L. 1987 Enteriw associated with Campylobacter laridis. Journal of Clinical Microbiology 2, 10-12.

SKIRROW, M.B. & BENJAMIN, J . 1980 ‘1001’ campylo- bacters: cultural characteristics of intestinal campy- lobacters from man and animals. Journal of Hygiene (Cambridge) 85,427442.

TAUXE, R.V., PATTON, C.M., EDMONDS, P., BARRETT, T.J., BRENNER, D.J. & BLAKE, P.A. 1985 Illness associated with Campylobacter laridis, a newly re- cognized Campylobacter species. Journal of Clinical Microbiology 21, 222-225.