Design and Synthesis of Sialic Acid Conjugates as Inhibitors of EKC-causing Adenoviruses

Susanne Johansson

Department of Chemistry

Umeå University 2008

DOCTORAL THESIS

2

COPYRIGHT © SUSANNE JOHANSSON

ISBN: 978-91-7264-545-5 PRINTED IN SWEDEN BY VMC-KBC

UMEÅ UNIVERSITY, UMEÅ 2008 Cover: The adenovirus electron microscopy image (2nd from top) is printed with permission from the originator: Linda M. Stannard, Univesity of Cape Town, http:/www.uct.ac.za/depts./mmi/.

3

Title: Design and synthesis of sialic acid conjugates as inhibitors of EKC-causing adeno-viruses

Author: Susanne Johansson, Department of Chemistry, Umeå University, SE-901 87 Umeå, Sweden

Abstract The combat against viral diseases has been, and still is, a major challenge in the field of drug development. Viruses are intracellular parasites that use the host cell ma-chinery for their replication and release. Therefore it is difficult to target and destroy the viral particle without disturbing the essential functions of the host cell. This thesis describes studies towards antiviral agents targeting adenovirus type 37 (Ad37), which causes the severe ocular infection epidemic keratoconjunctivitis (EKC). Cell surface oligosaccharides serve as cellular receptors for many pathogens, including viruses and bacteria. For EKC-causing adenoviruses, cell surface oligo-saccharides with terminal sialic acid have recently been shown to be critical for their attachment to and infection of host cells. The work in this thesis support these re-sults and identifies the minimal binding epitope for viral recognition. As carbo-hydrate–protein interactions in general, the sialic acid–Ad37 interaction is very weak. Nature overcomes this problem and vastly improves the binding affinity by presenting the carbohydrates in a multivalent fashion. Adenoviruses interact with their cellular receptors via multiple fiber proteins, whereby it is likely that the ideal inhibitor of adenoviral infections should be multivalent. This thesis includes design and synthesis of multivalent sialic acid glycoconjugates that mimic the structure of the cellular receptor in order to inhibit adenoviral attachment to and infection of human corneal epithelial (HCE) cells. Synthetic routes to three different classes of sialic acid conjugates, i.e. derivatives of sialic acid, 3’-sialyllactose and N-acyl modified sialic acids, and their multivalent counterparts on human serum albumine (HSA) have been developed. Evaluation of these conjugates in cell binding and cell infectivity assays revealed that they are effective as inhibitors. Moreover the results verify the hypothesis of the multivalency effect and clearly shows that the power of inhibition is significantly increased with higher orders of valency. Potential inhibi-tors could easily be transferred to the eye using a salve or eye drops, and thereby they would escape the metabolic processes of the body, a major drawback of using carbohydrates as drugs. The results herein could therefore be useful in efforts to develop an antiviral drug for treatment of EKC.

Keywords Sialic acid, sialylmimetics, sialylation, multivalency, glycoconjugates, adenovirus, antiviral agents, epidemic keratoconjunctivitis

ISBN: 978-91-7264-545-5

4

5

Contents

List of Papers ...................................................................................................... 7

List of Abbreviations.......................................................................................... 9

1. Introduction .................................................................................................. 11 1.1 Sialic acid conjugates in biology and chemistry ................................. 12 1.2 Multivalency in glycobiology............................................................... 16 1.3 Adenoviruses, causative agents of EKC .............................................. 18

2. Aims of the thesis......................................................................................... 20

3. Sialic acid glycoconjugates as inhibitors of EKC-causing adenoviruses . 21 3.1 Synthesis of sialylated oligosaccharides .............................................. 22 3.2 Conjugation of sialylated oligosaccharides to HSA............................ 25 3.3 Biological evaluation of multivalent 3’-sialyllactose as inhibitor of EKC-causing adenoviruses ......................................................................... 27 3.4 Structural simplification of the inhibitor ............................................. 28 3.5 Mode of action....................................................................................... 32 3.6 Summary ................................................................................................ 34

4. Design, synthesis and evaluation of nonnatural sialic acids and sialylmimetics ................................................................................................... 36

4.1 Structure-based design and docking of nonnatural analogues of sialic acid ............................................................................................................... 36 4.2 Synthesis and biological evaluation of N-acyl modified sialic acids . 41 4.3 Synthesis and biological evaluation of multivalent N-acyl modified sialic acids .................................................................................................... 48 4.4 Evaluation of benzoic acids as sialylmimetics .................................... 49 4.5 Summary ................................................................................................ 53

5. Structure-based methods to investigate interactions of ligands with Ad37........................................................................................................................... 54

5.1 Relaxation-edited 1H-NMR spectroscopy for identification of potential inhibitors of Ad37 ........................................................................ 54 5.2 X-ray crystallography studies of nonnatural sialic acids in complex with the Ad37 fiber knob ............................................................................ 58 5.3 Summary ................................................................................................ 61

6

6. Concluding remarks and future prospectives ............................................. 62

7. Populärvetenskaplig sammanfattning ......................................................... 64

8. Acknowledgements ...................................................................................... 66

9. Appendices ................................................................................................... 69 Appendix I.................................................................................................... 69 Appendix II .................................................................................................. 71 Appendix III................................................................................................. 72

10. References .................................................................................................. 74

7

List of Papers

I S. Johansson, N. Arnberg, M. Elofsson, G. Wadell, and J. Kihlberg; Multivalent HSA conjugates of 3’-sialyllactose are potent inhibitors of adenoviral cell attachment and infec-tion. ChemBioChem, 2005, 6, 358-364. DOI: 10.1002/cbic.200400227

II S. Johansson, E. Nilsson, M. Elofsson, N. Ahlskog, J. Kihlberg, and N. Arnberg; Multivalent sialic acid conju-gates inhibit adenovirus type 37 from binding to and infecting human corneal epithelial cells. Antiviral Research, 2007, 73, 92-10. DOI: 10.1016/j.antiviral.2006.08004

III S. Johansson, E. Nilsson, W. Qian, N. Arnberg, J. Kihlberg, and M. Elofsson; Design, synthesis and evaluation of N-acyl modified sialic acids as inhibitors of EKC-causing adeno-viruses. Manuscript.

Paper I and II have been reprinted with permission from the publishers: Elsevier and Wiley-VCH Verlag GmbH & Co.

8

9

List of Abbreviations

3’SLA 3’-sialyllactose Ac acetyl AcOH acetic acid Ad adenovirus AgOTf silver trifluoromethanesulfonate, silver triflate AIDS aquired immunodeficiency syndrome aq. aqueous Bn benzyl Boc tert-butoxycarbonyl BSA bovine serum albumine Bu butyl CAR coxsackie adenovirus receptor Cbz benzyloxycarbonyl COSY correlation spectroscopy CPM counts per minute Da dalton Dec decyl DIC N,N’-diisopropylcarbodiimide DIPEA diisopropylethylamine DMAP N,N-dimethylaminopyridine DMF N,N’-Dimethylformamide DMTST dimethyl(methylthio)sulfoniumtrifluoromethanesufonate EKC epidemic keratoconjunctivitis EM electron microscopy eq. equivalents Et ethyl et al. and others (et alii) FAB fast atom bombardment FFU fluorescent focus units FITC fluorescein thioisocyanate Fmoc 9-fluorenylmethoxycarbonyl Gal galactose GalNAc N-acetylgalactosamine Glc glucose HATU O-(7-azabenzotriazole-1-yl)-N,N,N’,N’-

tetramethyluronium hexafluorophosphate

10

HCE human corneal epithelial HIV human immunodeficiency virus HPLC high performance liquid chromatography HRMS high resolution mass spectroscopy HSA human serum albumine IC50 50 % inhibitory concentration iPr isopropyl KDN 2-keto-3-deoxynononic acid L leaving group LC liquid chromatography Lys lysine MALDI matrix-assisted laser desorption ionization Me methyl MeSBr methylsulfenylbromide MeSOTf methylsulfenyl trifluoromethanesulfonate MS mass spectroscopy Mw molecular weight Neu5Ac N-acetylneuraminic acid Neu5Gc N-glycolylneuraminic acid NIS N-iodosuccinimide NMR nuclear magnetic resonance PBS phosphate buffered saline Ph phenyl Pr propyl Pro proline Py pyridine quant. quantitative r.t. room temperature rmsd root mean square deviation ROMP ring opening methatesis polymerisation RSA rabbit serum albumine SA sialic acid TfO trifluoromethanesulfonate, triflate TFA trifluoroacetic acid TfOH trifluoromethanesulfonic acid, triflic acid THF tetrahydrofuran Thr threonine TLC thin layer chromatography TMSOTf trimethyl sulfonium trifluoromethanesulfonate TOF time of flight Troc trichloroethoxycarbonyl Tyr tyrosine Val valine

11

1. Introduction

For a long time, viral diseases were considered untreatable by chemo-terapheutic means since viruses were known to be dependent on the machin-ery of their host cells for replication. Targeting viruses selectively without disturbing any of the essential functions of the cell was thought to be an in-superable problem. Preparates that improve the immune defence to a particu-lar virus (i.e. vaccines) have indeed been, and still are, very successful but they provide modest or no therapeutic effect in individuals already infected. Unfortunately, for some major viral diseases it has not been possible to pro-duce efficient vaccines. To improve the defense against viral infections, an additional approach based on antiviral compounds is necessary. Attempts to combat viral infections in such ways can be traced to the early 1950s and the first drug was introduced in the beginning of the 1960s for prophylaxis of smallpox.1,2 However, it was not until 1977 the breakthrough of antiviral chemotherapy came by the launch of the nucleoside analogue Acyclovir, the first truly selective agent demonstrating that selectivity for viral enzymes over human host enzymes was a realistic goal. In addition, in the middle of 1980s, the human immunodeficiency virus (HIV) was identified as the causative agent of aquired immunodeficiency syndrome (AIDS). An inten-sive search for inhibitors of this retrovirus started and a general increase in research on antivirals followed. Fortunately, at the same time, important tools such as genome sequencing, structure-based design and high through-put screening required to carry out effective research appeared and facili-tated the progress. During the past decade, the development of new antiviral agents has been growing rapidly. In 1990 there were only five licensed drugs targeting viruses and today there are more than 40 on the market and at least as many in preclinical trials.1,3 However, the progress of antiviral drugs is still far behind the antibacterial agents. There are several reasons for this but the primary one is still the fact that viruses are more difficult to combat in a safe way. The need for an improved defense against viruses is growing rap-idly as more viral diseases are recognized and already existing treatments need to be modified due to resistance and drug related side effects. Today the majority of the drugs on the market are nucleoside analogues that inhibit viral polymerases. However, during the recent years nonenzymatic processes of the viral life cycle such as adsorption to host cells, fusion and penetration through the cell membrane have became focus for the research of new anti-viral agents.2,4

12

1.1 Sialic acid conjugates in biology and chemistry Carbohydrate–protein interactions are crucial in several physiological proc-esses.5,6 Cell surface oligosaccharides often serve as cellular receptors for many different pathogens, including viruses and bacteria. They are also of great importance for recognition and signaling when the immune system combats such infections. An important family of carbohydrates involved in many biological processes and crucial for the development and life of higher animals are the sialic acids.7,8 These are commonly found in terminal posi-tions of cell surface glycoconjugates such as glycoproteins and glycolipids. Thereby they are ideally located to participate in carbohydrate–protein inter-actions that mediate recognition phenomena. Sialic acids have a very spe-cific glycosidic linkage pattern and are found either α(2,3)- or α(2,6)-linked to hexoses or α(2,8)-linked to other sialic acids. There are about 50 natural members of the sialic acid family; all of them comprise the neuraminic acid (Neu) scaffold and differ only in the substitution pattern at the amino group and at the hydroxyl groups (Figure 1). The amino group of neuraminic acid can either be acetylated or glycolylated, and the nonglycosidic hydroxyl groups can be acetylated. Usually, there is only one O-acetyl group, often at O-9, but di- and tri-O-acetylated derivatives are also known. The most abun-dant sialic acid is N-acetylneuraminic acid (Neu5Ac, 5-acetamido-3,5-dideoxy-D-glycero-D-galacto-non-2-ulopyranosonic acid). Other important derivatives are the N-glycolylneuraminic acid (Neu5Gc) and the non-aminated form of sialic acid (3-deoxy-D-glycero-D-galacto-non-2-ulopyranosonic acid, KDN).

O

COOH

R2

R4

R9R8

R7

1

2

3

45

67

8

9

R2 = ! OH if free but in nature mostly linked "(2,3) or "(2,6) to hexoses, or "(2,8)

to other sialic acids

R4 = OH or OAcR7 = OH or OAc

R5 = NH2, NHAc,

NH-glycolyl or OH

R9 = OH, OAc,

O-lactyl or O-phosphate

R5

R8 = OH, OAc,

O-sulfate or O-methyl

Figure 1: A general overview of the sialic acid family. The sialic acids are mostly found in terminal positions of cell surface glycoconjugates (Figure inspired by R. Schauer7).

13

In cellular interactions sialic acids function in a dual fashion either by mask-ing recognition sites9 or contrarily by constituting a recognition site them-selves.10 In their protective role they can mask specific biological epitopes and by reversible sialylation (sialidase/sialyltransferase) they may regulate cellular interactions. Two essential recognition sites masked by sialic acids are galactose residues, involved in many physiological and pathological events, and immunological recognition sites where loss of sialic acid could lead to autoimmune diseases. In contrast to masking, sialic acids are the most frequent ligands for pathogenic and nonpathogenic microorganisms.11,12 The best known and longest studied example is influenza viruses that attach to cellular sialic acid by the viral lectin hemagglutinin prior to entry and infection of the host cell.13 A well studied example of pathogenic bacteria that colonize tissue via binding to sialic acid is Helicobacter pylori that is the causative agent of gastric inflammation and possibly cancer.14,15 This mani-fold of possibilities makes understanding of the biological role of sialic acid glycoconjugates difficult. The availability of complex sialic acid conjugates is of key importance to unravel the mechanisms of sialic acid recognition and masking. Isolation of pure and structurally defined glycoconjugates is cumbersome, whereby effective routes for synthesis of these conjugates are crucial. Both chemical and enzymatic routes are used to prepare sialic acid containing glycoconjugates but only chemical methods will be discussed in this thesis.

Formation of a glycosidic bond is generally performed through the conden-sation of an electrophilic glycosyl donor, bearing a potential leaving group at its anomeric center, with a nucleophilic glycosyl acceptor. The glycosidic reaction can occur at any unprotected hydroxyl group of the acceptor. In addition, the glycosylation can result in formation of either the α- or the β-anomer, making control of the regio- and stereoselectivity of this reaction important. This can be accomplished if the hydroxyl groups and the ano-meric center of the acceptor are protected with appropriate protecting groups. A glycosylation normally refers to an O-glycosylation if nothing else is specified, but N-, S-, and C- glycosylations are also possible. The field of glycosylations has been extensively reviewed.16-18 Therefore, the focus of this text is restricted to sialic acid chemistry and strategies to selectively synthesize α-O-glycoconjugates of sialic acid, the naturally occurring form of sialic acid conjugates involved in the biological recognition events dis-cussed above.

14

Although glycosylations with sialic acids (sialylations) can be performed with similar approaches as for less complicated hexoses, e.g. galactose and glucose, the formation of α-linked sialic acids is far more complex and pro-vides a great challenge for carbohydrate chemists (Figure 2).19 The specific difficulties arise from structural features of the sialic acid scaffold.10,20 Firstly, the electron withdrawing carboxylic acid at the anomeric center (C2) disfavors oxonium ion formation and it also restricts formation of the glyco-sidic bond due to steric hindrance. Secondly, lack of a substituent at the C3 position precludes neighboring group participation to assist and/or direct the sterical outcome of the reaction. Thirdly, this lack of a C3 substituent, in combination with the electron deficient anomeric center, makes sialic acids prone to 2,3-elimination.

O

COOR

L

OAc

AcOOAc

AcO

AcHN

promotor O

OAc

AcOOAc

AcO

AcHN

COORO

COOR

OR

OAc

AcOOAc

AcO

AcHN

O

OR

COOR

OAc

AcOOAc

AcO

AcHN

O

OAc

AcOOAc

AcO

AcHN

COOR

!-glycoside of sialic acid

"-glycoside of sialic acid

Neu2en5Ac

X

donor

acceptor (ROH)

+ HL

+ HL

+ HL, ROH

Figure 2: Direct chemical synthesis of α-glycosides of sialic acid. An appropriate leaving group (L) at the donor and a corresponding promotor for activation, directs the stereoselectivity at the anomeric center. Common byproducts are the β-isomer and the 2,3-elimination product of the donor.

To minimize the side reactions and increase the yield of sialylations, other donor/promotor strategies compared to those used in classical glycosylations are required.10,21 The oldest sialylation approach is the one using anomeric halides, chorides in particular, as donors and silver or mercury salts as pro-motors. One classical example is the Koenig-Knorr reaction,22 using Ag2CO3 as promotor. These reactions often result in α/β mixures and elimination products. Nowadays, the halogen derivatives are only used in sialylations of simpler alcohols. When more complex and hindered alcohols like sugars are to be sialylated other approaches are preferred. The major breakthrough in sialic acid chemistry was the development of new highly efficient and stereoselctive donor/promotor systems in the early 1990s.

15

A widely applied approach uses 2-thiosialosides that can be activated in the presence of soft electrophiles. The most important 2-thioderivatives includes S-alkyl/aryl (e.g. methyl, ethyl, phenyl)23 and xanthate sialosides24 and the most commonly used activation reagents are methylsulfenyl trifluoro-methanesulfonate (MeSOTf), dimethyl(methylthio)sulfonium trifluoro-methanesulfonate (DMTST) and N-iodosuccinimide (NIS)/trifluoro-methanesulfonic acid (TfOH). The excellence of the thioderivatives lies in their chemical stability making them compatible with a manifold of reaction conditions used in carbohydrate chemistry. Another important class of do-nors are the 2-phosphites25 that are very reactive and only require a catalytic amount of TMSOTf for their activation. The phosphite donors have been reported to be better than the thioderivatives for sialylation of more sterically hindered alcohols. Not mentioned above, but of great importance to ano-meric selectivity, is the participating solvent acetonitrile (MeCN).21 The nitrilium ion replaces the leaving group at the donor and adopts a preferred axial configuration. Subsequent nucleophilic substitution with an unpro-tected hydroxyl group of the acceptor gives mainly the α-isomeric sialyl-ation product.

O

COOR

L

OAc

AcOOAc

AcO

AcHN XROH

promotor O

OAc

AcOOAc

AcO

AcHN

COOR

X

O R

H

O

COOR

OR

OAc

AcOOAc

AcO

AcHN X

O

COOR

L

OAc

AcOOAc

AcO

AcHNROH

promotor O

OAc

AcOOAc

AcO

AcHN

COOR

O

H

R

O

OR

COOR

OAc

AcOOAc

AcO

AcHN

X X X

HL

HL

Figure 3: Indirect synthesis of α- and β-glycosides of sialic acids. An auxiliary is incorporated at the C3 position to direct the stereoselectivity (Above: equatorial auxiliary/α-glycoside, Below: axial auxiliary/β-glycoside).

The direct chemical methods discussed above are most common but indirect methods where auxiliaries are introduced at the C3 position to control the anomeric selectivity are also used.21 The auxiliaries (e.g. 3-O, 3-bromo, 3-thio or 3-seleno auxiliaries) direct the reaction towards 2,3-trans glycosides of sialic acids, whereby α-glycosides are formed by the use of an equatorial auxiliary and β-glycosides are formed if an axial auxiliary is used (Figure 3). However, these methods require several additional reaction steps in order to introduce and later on, remove the auxiliary. In addition, stereoselective introduction of auxiliaries at the C3 position is not easily performed.

16

During the recent years, a number of new direct and indirect methods have emerged20,26-30 but unfortunately there is still no general method that allows sialylations of a wide range of acceptors in high yield and stereoselectivity. Each specific sialylation needs to be carefully examined to obtain a success-ful result.

1.2 Multivalency in glycobiology Despite the central role of sialic acids and carbohydrates in general, there are relatively few carbohydrate based therapeutics. One of the reasons for this is their poor oral bioavailability and high sensitivity towards metabolism lead-ing to rapid clearance. Another major drawback with carbohydrates in drug development is their weak (millimolar) affinity to proteins making it difficult to design high affinity drugs to target specific carbohydrate binding proteins. Nature solves this problem elegantly by using carbohydrates in a multivalent fashion and thereby makes strong attachments from weak binders. Multiple copies of the carbohydrate are linked closely together to produce multivalent binding epitopes on one biological molecule or organism (cell surface) that could simultaneously associate with multiple binding sites on another entity (bacterium, cell, protein, or virus). This results in an increase of the binding affinity by several orders of magnitude beyond the multiplication factor of the valency.31-33 During the latest decades there has been a considerable in-terest in mimicking the natural multivalent saccharide display in order to design high affinity carbohydrate ligands with potential in drug develop-ment.34-36 Some notable studies where multivalent carbohydrates have been used to inhibit viral and bacterial attachment to cell surface receptors clearly demonstrate the promise of multivalency.37-41 To succeed with such conju-gates it is important to understand the underlying mechanisms through which a multivalent ligand can enhance affinity compared to its monovalent coun-terpart (Figure 4).32,42-44 The most commonly discussed mechanism, simulta-neous binding of multivalent ligands to multiple protein binding entities, is often referred to as the chelate effect and it could be accomplished in several ways. Multivalent ligands can bind to oligomeric binding entities (4a), to one binding entity with multiple carbohydrate binding sites (4b) or by subsite binding if the binding epitope possess an additional binding site close to the primary site (4c). In all these cases multivalent carbohydrate ligands could receive a tighter binding than their monovalent counterparts due to a smaller entropy cost. This primarily results from the fact that most of the entropy cost of binding the multivalent ligand is paid with the first carbohydrate–protein interaction and that the following interactions will occur without additional entropy costs (except for conformational costs) presupposed that the ligands perfectly match their binding sites.

17

Figure 4: Mechanisms of multivalent ligand binding to cell surface receptors. (a) Multivalent ligands can occupy several binding sites simultaneously by interacting with oligomeric receptors. (b) Multivalent ligands can bind to more than one binding site at one receptor. (c) Multivalent ligands can bind to both primary and secondary binding sites on one receptor. (d) Multivalent ligands possess a higher local concen-tration of the ligands (Picture inspired by L. Kiessling et al.32,42).

A further contribution to the affinity enhancement is that multivalent carbo-hydrate ligands display a higher local concentration of carbohydrates that can result in higher functional affinity even if only one binding entity is en-gaged (4d). Another unique effect of multivalent carbohydrate ligands that is of great relevance for their potential as inhibitors of adhesion is steric stabi-lization. Due to the size and hydration shell of multivalent ligands they could, if bound to a pathogen, sterically prevent proximity to the cellular surface and thereby hinder the pathogen from attachment. A less discussed but still very important mechanism by which multivalent ligands could pre-vent pathogens from infecting cells is the possibility to crosslink pathogens and thereby cause aggregation.44 While probing multivalent cell surface binding systems, it is important to point out that the strength of the multiva-lent carbohydrate–protein interaction not only depends on the valency of the two molecules but also on their structural arrangement. A compatible orga-nization of the carbohydrates and the binding sites of the protein is required to get a successful interaction.32,45

Naturally occurring multivalent carbohydrate structures are widespread but in general they are only found in limited amounts and are often too structur-ally complex and heterogeneous to be useful when targeting a specific bio-logical system. Using synthetic multivalent glycoconjugates makes it possi-ble to vary the presentation of the carbohydrates in terms of valency, scaf-fold structure, size, shape and flexibility. There are several different methods to generate multivalent saccharide displays. Natural scaffolds such as pro-teins like serum albumin, shorter peptides or lipids provide easily accessable neoglycoconjugates.46,47

18

Such conjugates have been, and still are, used when the purpose is to quali-tatively show that multivalent binding is involved in a specific system. How-ever, when the purpose is to explore the specific requirements for binding, the scaffold needs to be more properly designed and controlled. Nonnatural scaffolds can be constructed to span ranges from flexible to relatively rigid structures and display the carbohydrates in a very controlled way.42,44,48 Line-ar scaffolds have been obtained from polymerization of modified acryl-amides37,39,49 or from ring opening metathesis (ROMP)50-52. In order to gen-erate spherical or partly spherical scaffolds dendrimers, highly branched oligomeres, have been extensively used. 45,53-55 A noncovalent alternative for multivalent presentation of carbohydrates is to use liposomes as scaf-folds.56,57 Liposomes represent an ideal mimic of the cell surface membranes where the carbohydrate moieties could move freely in the two-dimensional membrane space. All these different methods will of course have their indi-vidual advantages and disadvantages and the preferred method for an opti-mal multivalent carbohydrate display will vary with the target biological system.

1.3 Adenoviruses, causative agents of EKC Adenoviruses constitute the Adenoviridae family of viruses and cause infect-ions in humans and animals all over the world.58-61 Three adenovirus sero-types, (Ad8, Ad19 and Ad37) have frequently been reported as the major cause of the severe ocular infection epidemic keratoconjunctivities (EKC).62,63 The virus uses its protruding fibers to interact with cellular recep-tors. Each viral particle has twelve fiber proteins and the receptor binding domain, the “knob”, consists of a homotrimer that presents three separate binding sites (Figure 5).64 Although adenoviruses in general use the cox-sackie adenovirus receptor (CAR), it has recently been demonstrated that the EKC-causing serotypes, use sialic acid as a functional cellular receptor. For Ad37 it has been determined that the receptors consist of glycoproteins car-rying at least one terminal sialic acid residue linked by an α(2-3)-glycosidic bond to a neighboring galactose saccharide.65,66 Since the fiber knob protein of the other two EKC-causing adenoviruses are highly homologous to that of Ad37, it is likely that they also use receptors containing terminal α(2-3) sialic acids.67

EKC-causing adenoviruses are usually transferred by direct contact such as eye-to-hand-to-eye within an already infected individual or as eye-to-hand-to-hand-to-eye between infected and healthy individuals. As a result of that, EKC is predominant in densely populated areas and has been well docu-mented in East Asia68-70 but also in North America and Europe.62,71 For ex-ample, half a million individuals fall ill in EKC every year in Japan alone.72

19

As a result of the infection, the patient is handicapped for several weeks due to irritation, tearing, photophobia and in the most severe cases, various levels of reduced sight. In some cases the infection leads to permanent reduction of eyesight. A major consequence of the longlasting infection in addition to the substantial suffering of the patients are economic losses.62 So far there are no licensed antiviral drugs for treatment of EKC. It is important to point out that the capacity of adenoviruses to cause EKC correlates with their ability to use glycoconjugates with terminal sialic acids as cellular receptors.66 In view of this it appears possible that sialylated saccharides could act as inhibitors of binding of the virus to host cells, i.e. as antiviral agents. Since the eye is easily accessable for topical administration by the use of a salve or eye drops, the antiviral drug would not have to be administred orally and thereby it would be affected by metabolic processes to a less extent. As a conse-quence, most of the above discussed drawbacks with carbohydrate based drugs could be circumvented when using carbohydrate derivatives to treat EKC.

Figure 5: The adenoviral particle. The virus interacts with a cellular receptor via the fiber proteins protruding from the viral particle. The receptor binding domain, con-sists of a homotrimeric “knob” that presents three separate binding sites. The figure is adopted with permission from Prof. W.C Russell, University of St Andrews, St Andrews, UK, and reprinted with permission from the publisher, The society for general microbiology.73

20

2. Aims of the thesis

The overall aim of this thesis was to develop potential antiviral agents that target adenoviruses of type Ad8, Ad19 and Ad37, the causative agents of the severe ocular infection EKC. The more specific objectives were to

i) verify that EKC-causing adenoviruses use sialic acid as cellular receptor and that their attachment to the receptor can be pre-vented by synthetic glycoconjugates of sialic acid,

ii) identify the minimal binding epitope required for viral recogni-tion, and

iii) on the basis that adenoviruses, like viruses in general, attach to host cells in a multiple fashion, study the effect of multivalent antiviral agents for potential treatment of EKC.

To achieve these objectives, the work in this thesis focuses on the design and synthesis of sialic acid glycoconjugates. When obtained, these derivatives were to be transformed into their corresponding multivalent conjugates and evaluated for their ability to inhibit EKC-causing adenoviruses from binding to and infecting human corneal epithelial (HCE) cells. Development of effi-cient synthetic routes to the inhibitors and reliable biological assays were expected to be critical for the success of this work.

21

3. Sialic acid glycoconjugates as inhibitors of EKC-causing adenoviruses

As described in chapter one, EKC-causing adenoviruses use glycoconjugates with terminal α(2-3)-linked sialic acids on HCE cells as cellular receptors. In order to develop an antiviral agent that blocks the sialic acid binding sites on the viral fiber protein and thereby prevent attachment and subsequent infec-tion, information of the carbohydrate receptor is of significant value. Since all α(2-3) sialyltransferases characterized so far only transfer sialic acid to galactose,74 it appears likely that the neighboring saccharide is galactose. In addition, unpublished studies of the binding of the viral fiber protein to gly-colipids using a thin layer chromatography (TLC) overlay assay suggest that the carbohydrate receptor consists of the pentasaccharide [NeuAcα2-8NeuAcα2-3(GalNAcβ1-4)Galβ1-4Glc] (1, Figure 6) or smaller parts thereof (2-4, Figure 6).75 Since the virus interacts with the cellular receptor in a multivalent fashion it is likely that an antiviral agent should be multiva-lent in order to be more efficient, e.g. a glycoconjugate prepared by attach-ment of several oligosaccharides to a suitable carrier. On the basis of this, the initial strategy towards an inhibitor was to synthesize the pentasaccharide 1, the smaller derivatives 2-4 and their corresponding multivalent conju-gates.

O

COOH

OH

HO

O

HO

AcHNOO

HOOH

R

OH

O

OH

OH

O

O

3

O

NHAcHO

HO OH

O

COOH

HO

HOOH

HO

AcHN

O

COOH

OH

HO

O

HO

AcHNOO

HOOH

R

OH

O

OH

OH

O

HO

O

COOH

HO

HOOH

HO

AcHN

O

COOH

OH

HO

OH

HO

AcHNOO

HOOH

R

OH

O

OH

OH

O

OO

NHAcHO

HO OH

O

COOH

OH

HO

OH

HO

AcHNOO

HOOH

R

OH

O

OH

OH

O

HO

1 2

4 Figure 6: The suggested cell surface receptor structure for EKC-causing adenovi-ruses (1) and smaller parts thereof (2-4).

22

3.1 Synthesis of sialylated oligosaccharides A retrosynthetic analysis of the pentasaccharide 1 resulted in the four build-ing blocks 5-8 (Figure 7). Three of these (5-7) may be readily synthesized from commercial saccharides,76-78 whereas the disialic acid derivative 8 can be prepared by controlled hydrolysis of colomic acid,79 which is a commer-cially available linear polymer of α(2-8)-linked sialic acid or by 8-O-sialylation of sialic acid.80 However, the oligosaccharides 1 and 2 constitute significant synthetic challenges due to the disialic acid moiety.80-82 There-fore, the primary focus of this project was to develop a synthetic route to the smaller oligosaccharides 3 and 4 (Figure 6) and their multivalent counter-parts.

O

COOH

OH

HO

O

HO

AcHNOO

HOOH

R

OH

O

OH

OH

O

OO

NHAcHO

HO OH

O

COOH

OH

HOOH

HO

AcHN

OOBnO

OBn

R

OBn

O

OBn

OBn

HO

HO

SCSOEtO

COOMe

OAc

AcOOAc

AcO

AcHN

SMeO

NHTrocAcO

AcO OAc

O

COOMe

OAc

AcO

O

AcO

AcHN OAc

O

COOMe

OAc

AcOOAc

AcO

AcHN

OOHO

OH

OH

OH

O

OH

OH

HO

HO

OHO

COOH

OH

HOOH

HO

AcHN

OHO

NH2HO

HO OH

O

COOH

OH

HO

O

HO

AcHN O

O

COOH

OH

HOO

HO

AcHN

Sialic acid Galactosamine Lactose Colomic acid

1

5 6 7 8

Figure 7: Retrosynthetic analysis of the cell surface receptor structure 1.

The 2-azidoethyl lactoside 12 (Scheme 1) was selected as core structure. It allows glycosylation with sialic acid and N-acetylgalactosamine (GalNAc) and subsequent conjugation to a carrier via the azide. From lactose, the 3’,4’-O-isopropylidene-2-bromoethyl lactoside 9 was synthesized by traditional carbohydrate chemistry.78 This compound was then successfully converted into the 2-azidoethyl lactoside 10 in 97 % yield by treatment with sodium azide and crown ether in dimethylformamide (DMF).83

23

Benzylation of the unprotected hydroxyl groups gave 11, and subsequent hydrolysis of the 3’4’-isopropylidene acetal resulted in the target lactoside core 12, with free hydroxyl groups in the 3’ and 4’ positions that could be glycosylated, and an azide functionality that could be used for conjugation. From this lactoside it would be possible to synthesize the tetrasaccharide 3 via a protected precursor of trisaccharide 4 (Figure 6).

O

COOMe

OAc

AcOOAc

AcO

AcHNOO

BnOOBn

O

OBn

O

OBn

OBn

O

OO

NHTrocAcO

AcO OAc13

OO

HOOH

OO

OH

OHO

O

N3

9 10

OOBnO

OBn

O

OBn

O

OBn

OBnO

O

N3

11

OOBnO

OBn

O

OBn

O

OBn

OBnHO

HO

N3

12

OH

O

OBn

O

HO OBn

O O

OBn

OBnBnO

O

COOMe

OAc

AcOOAc

AcO

AcHN ON3

14

N3

OO

HOOH

OO

OH

OHO

O

Br

OH

a) b)

c) d)

e)

Scheme 1: Synthesis of the protected cell surface receptor analogues 13 and 14. Reagents and conditions: a) NaN3, 15-crown-5, DMF, 97 %; b) BnBr, NaH, DMF, 68 %; c) TFA/H2O (9:1), CH2Cl2, quant.; d) 5 (Figure 7), AgOTf, MeSBr, –60 °C, CH3CN/CH2Cl2 (9:4), 56 %; e) 6, NIS/TfOH, CH2Cl2, –20 ºC, 44 %.

Due to its well accepted efficiency in stereoselective α-O-sialyations,78,84,21,85 the sialic acid donor 576 (Figure 7) was used to sialylate the 3’ position of the lactoside 12 (Scheme 1). The reaction was promoted by MeSOTf,86,87 gener-ated in situ by reaction of methylsulfenylbromide (MeSBr) with silver trifluoromethanesulfonate (AgOTf),88 and performed in a mixture of di-chloromethane and acetonitrile at low temperature. The α-isomer of the pro-tected trisaccharide 13 was obtained in 56 % yield. The regioselectivity of this sialylation is due to the greater reactivity of the equatorial hydroxyl group compared to the axial C4 hydroxyl group. Further glycosylation at the 4’ hydroxyl group of 13 with GalNAc to achieve the tetrasaccharide 14 was accomplished by the thiomethyl donor 677,89 (Figure 7) and with NIS/TfOH as promotor system.77 The protected tetrasaccharide 14 was obtained in 44 % yield. In order to conjugate this compound to a carrier all protecting groups had to be removed and the azide had to be reduced to a primary amine.

24

The initial strategy was removal of the trichloroethoxycarbonyl (Troc) group, deacetylation, hydrolysis of the methyl ester and cleavage of the ben-zyl groups with simultaneous reduction of the azide in the given order. Un-fortunately it was more difficult than expected and problems occurred at the very first step (Scheme 2, unpublished results). Selective deprotection of the Troc protected amine of GalNAc using Zn/acetic acid77,90 was tried but dis-carded since it also reduced the azide. The second strategy, selective reduc-tion of the azide with SnCl2, PhSH, Et3N91 followed by protection of the generated amine with an orthogonal protecting group (benzyloxycarbonyl, Cbz)92, was also unsuccessful and resulted in an uncharacterized mixture of products. Despite the fact that other research groups have succeeded in the synthesis of similar compounds93,94 the decision was taken not to complete the synthesis of the tetrasaccharide at this stage and instead direct the focus towards the less complex trisaccharide derivative 18 (Scheme 3).

O

COOMe

OAc

AcOOAc

AcO

AcHNOO

BnOOBn

O

OBn

O

OBn

OBn

O

OO

NHTrocAcO

AcO OAc14

N3

O

COOMe

OAc

AcOOAc

AcO

AcHNOO

BnOOBn

O

OBn

O

OBn

OBn

O

OO

NH2AcO

AcO OAc15

N3

Zn, HOAc, r.t.

O

COOMe

OAc

AcOOAc

AcO

AcHNOO

BnOOBn

O

OBn

O

OBn

OBn

O

OO

NHTrocAcO

AcO OAc16

NH2

SnCl2, PhSH, Et3N, CH3CN, r.t.

Scheme 2: Two initial strategies for deprotection of tetrasaccharide 14. Deprotection of the GalNAc moiety of 14 to get compound 15 was not possible under these condi-tions without simultaneous azide reduction. Reduction of the azide of 14 to get the primary amine 16 resulted in a mixture of products.

Hypothetically, the GalNAc residue is of less importance for the adenoviral recognition compared to the essential sialic acid residue and thereby an in-hibitor without the GalNAc residue would most likely still be able to bind to the fiberprotein and prevent the virus from attaching to the host cell. The trisacccharide 13 was completely deprotected in three steps including deace-tylation, hydrolysis of the methyl ester and cleavage of the benzyl groups with simultaneous reduction of the azide (Scheme 3).95 The unprotected aminoethylglycoside of sialyllactose, 18, was obtained in 85 % yield from the protected trisaccharide 13 and in an overall yield of 11 % from lactose.

25

13

O

OBn

O

HO OBn

O O

OBn

OBnBnO

O

COOMe

OAc

AcOOAc

AcO

AcHN ON3

17

O

OBn

O

HO OBn

O O

OBn

OBnBnO

O

COOH

OH

HOOH

HO

AcHN ON3

18

O

OH

O

HO OH

O O

OH

OHHO

O

COOH

OH

HOOH

HO

AcHN ONH2

a)

b)

Scheme 3: Deprotection of the trisaccharide 13. Reagents and conditions: a) 1. NaOMe, MeOH, r.t., 2. H2O, r.t., 87 %, 2 steps; b) H2(g), Pd/C, EtOH/HCl, r.t., 98 %.

3.2 Conjugation of sialylated oligosaccharides to HSA As discussed in section 1.2, a multivalent display of oligosaccharides can be obtained by the use of several different carriers. Proteins have been widely used as carriers since they provide desirable physicochemical properties such as solubility and hydrodynamic size. In addition, the availability of amino acid side chains provides suitable functionalities for conjugation. The most commonly used carriers for the preparation of neoglycoproteins are the bo-vine (BSA), rabbit (RSA) and human (HSA) serum albumins since they are easily accessable. Generally, most of the conjugation procedures use lysine amino groups of the protein for formation of the covalent bond between the carbohydrate moiety and the protein. Common methods includes formation of a -C(=X)NH-linkage (X = O, S, or NH), or formation of a secondary amine by reductive amination.47 During the latest years an alternative method for conjugation has been introduced, where an amine on the carbo-hydrate residue is covalently linked to an amine of the protein by a squaric acid diester.96-99 This method relies on the pH dependent reactivity of squaric acid diesters compared to squaric acid amide esters (Scheme 4). A dialkyl squarate reacts readily with amines at pH 7 affording a squaric acid amide ester, for further conversion to the diamide a pH of 9 is required. Advantages of this method are that it is compatible with both primary and secondary amines and that it is performed under mildly basic conditions making it suit-able for acid sensitive molecules such as carbohydrates.99 Another major advantage of this method is that unreacted squaric acid glycoside derivates can be recovered from the conjugation mixture.100

26

OO

OHHO

OO

OR1R1O

Squaric acid Squaric acid diester

R2NH2, pH 7

R3NH2,pH 9

OO

OR1R2HN

OO

NHR3R2HN

Squaric acid amide ester Squaric acid diamide

R1OH

Scheme 4: Conjugation of saccharides to proteins by squaric acid. (R1 = Me, Et, Dec usually, R2= saccharide-spacer, R3= protein).

Due to the above discussed properties of proteins and available knowledge within our group, a protein was chosen as carrier for oligosaccharide 18 syn-thesized in section 3.1. When this project was about to start, the cattle dis-ease bovine spongiform encephalopathy (BSE, “mad cow disease”) was frequently reported in the media, whereby the use of bovine material seemed inappropriate in development of a potential antiviral drug. Therefore, HSA became the choice of protein carrier and the newly reported and successful squaric acid method was used for conjugation (Scheme 5).99 By varying the number of equivalents of the sialyllactose derivative, the degree of multiva-lency could be varied with reproducible results. Five neoglycoproteins 20a-e ranging from 3-19 incorporated saccharides per HSA were synthesized.* The average degree of incorporation was determined by MALDI-TOF MS by using the centre of the single-charged neoglycoprotein peak.101

*O

OH

O

HO OH

O O

OH

OHHO

O

COOH

OH

HOOH

HO

AcHN ONH

20a-e, n=3,4,12,17 and 19

OO

NH

O

OH

O

HO OH

O O

OH

OHHO

O

COOH

OH

HOOH

HO

AcHN ONH

OO

OC10H21

19

n

HSA

18

O

OH

O

HO OH

O O

OH

OHHO

O

COOH

OH

HOOH

HO

AcHN ONH2

a)

b)

Scheme 5: Conjugation of 3’-sialyllactose to HSA. Reagents and conditions: a) Squaric acid didecylester, DMF, Et3N, r.t., 40 %; b) NaHCO3 (pH 9.0), HSA, r.t. * The experimental conditions for 20b (4-valent) and 20c (17-valent) were used to synthesize 20a (3-valent) and 20d (19-valent) respectively, indicating reproducible conjugation to HSA.

27

3.3 Biological evaluation of multivalent 3’-sialyllactose as inhibitor of EKC-causing adenoviruses In order to investigate whether the multivalent 3’-sialyllactose derivatives could work as inhibitors of EKC-causing adenoviruses, two different assays were performed.† Firstly, their ability to inhibit adhesion to HCE cells was investigated by a binding assay based on 35S-labeled Ad37 virions.102,103 Secondly, to investigate if they also prevented Ad37 from infecting HCE cells, an infectivity assay in which adenovirus-infected cells are visualized as fluorescent focus units (FFU) was used.104 65

Multivalent 3’-sialyllactose inhibits Ad37 from binding to HCE cells 35S-labeled Ad37 virions were preincubated with a dilution series of 3’-sialyllactose or the 4-, 12- or 17-valent 3’-sialyllactose HSA conjugates 20b-d before being added to HCE cells. After incubation, unbound virions were washed away, and cell-associated radioactivity was measured in a scintilla-tion counter. It was found that both commercial 3’-sialyllactose and the syn-thesized multivalent sialyllactose derivatives were able to inhibit adhesion of Ad37 virions to HCE cells. Thereby the hypothesis of the minor importance of the GalNAc moiety of the inhibitor proved to be justified. In addition the data clearly showed that the inhibition was significantly increased with higher orders of multivalency (Figure 8). The multivalent compound with the highest degree of multivalency, 20d, shows an IC50 value that is more than 100-fold better than 3’-sialyllactose.

Figure 8: Dose-dependent inhibition of Ad37 attachment to HCE cells with 3’-sialyllactose (3’SLA) and multivalent HSA conjugates of 3’SLA (20b–d). The re-sults are presented as a mean of two independent experiments. The experiments were repeated three times with reproducible results (for details see paper I). Printed with permission from the publisher; Wiley-VCH Verlag GmbH & Co. KGaA.

† The biological evaluation was performed in collaboration with Dr. N. Arnberg and E. Nils-son, Division of Virology, Department of Microbiology, Umeå University, Sweden.

28

Multivalent 3’-sialyllactose inhibits Ad37 from infecting HCE cells Nonlabeled Ad37 virions were preincubated with a dilution series of 3’-sialyllactose or the 19-valent 3’-sialyllactose HSA conjugate 20e before being incubated with HCE cells. Unbound virions were washed away and cell-associated virions were allowed to infect the cells. The cells were finally stained with rabbit polyclonal anti-Ad37 serum followed by FITC (fluo-rescein thioisocyanate) labeled swine anti-rabbit IgG antibodies and exam-ined in an immunofluorescence microscope. Once again the results demon-strated the enormous improvement in inhibitory power that can be obtained with multivalent structures. The 19-valent derivative clearly inhibited infec-tion at a nanomolar level whereas 3’-sialyllactose was not able to inhibit the infection at a concentration of 1 mM (Figure 9).

19-valent 3’SLA (mM)

0 0.01 0.1 13’SLA (mM)

0 0.01 0.1 1 Figure 9: Dose-dependent inhibition of Ad37 infection of HCE cells with 19-valent 3’-sialyllactose (20e) and 3’SLA (for details se table 2 paper I). 19-valent 3’SLA was 1000-fold more efficient as inhibitor than 3’SLA.

3.4 Structural simplification of the inhibitor From the results above it was clear that the less complex 3’-sialyllactose derivative was sufficient to prevent an EKC-causing adenovirus from bind-ing to and infecting HCE cells. Thus the synthesis of the more complex sac-charides (1-3, Figure 6) and their corresponding multivalent conjugates be-came less relevant and was abandoned. Could it be possible to simplify the structure further and still maintain the inhibitory power?

From the crystal structure of the sialic acid interacting domain of the Ad37 fiber knob protein in complex with sialyllactose it was found that only the sialic acid residue contributed to the carbohydrate–protein interaction.105 No part of the lactose residue was involved in complex formation in the crystal.

29

Hypothetically, multivalent sialic acid would therefore be able to prevent EKC-causing adenoviruses from binding to and infecting HCE cells to a similar extent as multivalent 3’-sialyllactose. An initial investigation of this hypothesis was made by comparing the inhibitory effect of different sialic acid containing saccharides. Sialic acid, disialic acid, trisialic acid, and 3’-sialyllactose were compared in the binding assay described in section 3.3. The results demonstrated that all these saccharides had similar efficiency in inhibiting Ad37 virions from binding to HCE cells (Figure 10), with IC50 values within the 1-5 mM range. Therefore it was decided to base the second generation of synthetic multivalent adenovirus inhibitors on sialic acid alone. A major advantage of a sialic acid conjugate is that it requires a less compli-cated synthesis than a conjugate based on sialyllactose. However, sialic acid chemistry is always challenging and an efficient synthetic route to a sialic acid derivative that could be conjugated to HSA had to be developed.

Figure 10: Dose-dependent inhibition of Ad37 attachment to HCE cells with sialic acid (SA), disialic acid (di-SA), trisialic acid (tri-SA) and 3’SLA. All saccharides showed IC50 values within 1-5 mM. Printed with permission from the publisher; Elsevier.

Synthesis of multivalent sialic acid HSA conjugates Since sialic acid itself is smaller than 3’-sialyllactose it seemed likely that the aglycon in this case should be longer than the ethyl aglycon used previ-ously. In addition, the optimal spacer derivative would preferably contain both the hydroxyl functionality for sialylation and an amino functionality for conjugation to HSA in order to save synthetic steps. Therefore, commer-cially available Fmoc-aminoalcohols were selected as aglycons. The longest available aminoalcohol at the time when this project was initiated was Fmoc-aminohexanol. Different donor/promotor system had to be evaluated in order to obtain the sialylated derivative 22 in reasonable yield and α/β-selectivity (Table 1).

30

Table 1: Sialylation of Fmoc-aminohexanol with different donor/promotor systems.

O

COOMe

R

OAc

AcOOAc

AcO

AcHN O

COOMe

O

OAc

AcOOAc

AcO

AcHN

NHFmoc

Fmoc-aminohexanol, promotor, CH3CN/CH2Cl2

21 22

R promotor eq promotor

CH2Cl2: MeCN

T (°C)

yield (%)

α:β ratio

21a Cl AgOTf 1.5 9:4 –10 8 not estimated

21b SC(S=)OEt MeSBr/ AgOTf

1 9:4 –60 14-30 3:1

21c SPh IBr/ AgOTf

1.3 3:2 –72 40-70 3:1-4:1

Since 2-chloro sialic acid is suggested to be the general donor of choice to sialylate simple aliphatic alcohols,21,106 21a was the first donor to be used in efforts to get the target sialoside (Table 1). Unfortunately, this method re-sulted in a very low yield of the sialoside 22 (8 %). Moreover, when the simple alcohol 1-penten-5-ol was sialylated under the same conditions only traces of the sialylated product was obtained (data not shown). Based on these results, the 2-chloro donor of sialic acid was discarded. Instead, the xanthate donor 21b and its corresponding promotor system used previously for the sialylation of the lactoside 12 (Scheme 1) was investigated. The yield was still not satisfactory, only 14-30 % and with an α/β ratio of 3:1 at the best. Therefore the, at that time, new method relying on the 2-thiophenyl donor 21c promoted by IBr/AgOTf26 was evaluated for this reaction. The yield was significantly improved (40-70 %) but the α/β ratio was still mod-erate (3:1-4:1). In order to investigate if the spacer length of the acceptor could have an influence on the reaction, the donor/promotor system 21c/IBr/AgOTf was used to sialylate the one carbon shorter Fmoc-aminopentanol (Scheme 6). Illustrating the complexity of the sialic acid chemistry, the yield of the sialylation was now raised. The sialoside 23 was obtained in 60-90 % yield and, more importantly, the α/β ratio increased to 7:1. However, in order to isolate the pure α-isomer, extensive purification via repeated SiO2 column chromatography and preparative HPLC was nec-essary. Therefore the pure α-isomer was only obtained in 26 % yield. This low yield was correlated with the problems of separating the two isomers, the more β-isomer of the product, the more α-isomer was lost in the purifica-tion. Therefore, efforts were made in order to improve the α/β ratio of the reaction. Statistical experimental design107 was used to investigate if the α-selectivity could be increased by variation of the reaction parameters; i) tem-perature, ii) promotor equivalents, iii) MeCN/CH2Cl2 ratio, and iv) acceptor structure. Unfortunately, no significant improvement was achieved.

31

Conditions with potential to increase the α-yield slowed down the reaction rate and gave a lower total yield of the product (unpublished results, for de-tails see Appendix I). It was decided to keep the conditions as for the synthe-sis of 23 and instead direct the focus on improving the method of purifica-tion. By using chiral preparative HPLC,108 it was possible to completely separate the two isomers and thereby the isolated yield of the α-isomer of 23 was significantly increased.

O

COOMe

SPh

OAc

AcOOAc

AcO

AcHN O

COOMe

O

OAc

AcOOAc

AcO

AcHNNHFmoc

O

COOH

O

OH

HOOH

HO

AcHNNH

OO

NH

*HSA

n26a-c, n=3, 8 and 13

O

COOH

O

OH

HOOH

HO

AcHNNH2

21c 23

24

O

COOH

O

OH

HOOH

HO

AcHNNH

25

OO

OC10H21

a) b)

c)

d)

Scheme 6: Synthesis of multivalent HSA conjugates of sialic acid. Reagents and conditions: a) Fmoc-aminopentanol, IBr, AgOTf, CH3CN/CH2Cl2 (3:2), –72 ºC, 60-90 %, α:β 7:1; b) 1. NaOMe, MeOH, 2. LiOH (aq), r.t.; c) Didecylsquarate, DMF, Et3N, r.t., 62 % from 24; d) NaHCO3 (pH 9.0), HSA, DMF, r.t.

The sialic acid derivative 23 was fully deprotected by deacetylation and si-multaneous Fmoc removal followed by hydrolysis of the methyl ester to give 24. It was then conjugated to HSA according to the methodology described in section 3.2. Multivalent sialic acid HSA conjugates 26a-c were obtained with 3, 8 and 13 incorporated saccharides per HSA respectively (number of incorporated saccharides according to MALDI-TOF measurements).

Biological evaluation of the sialic acid conjugates To evaluate the multivalent sialic acid conjugates, the cell binding assay and the cell infectivity assay described in section 3.3 were used. Once again the multivalency effect was clear and even more interestingly was that the sialic acid conjugates were as powerful inhibitors as the 3’-sialyllactose conju-gates. In both the binding assay and the infectivity assay the multivalent sialic acid conjugates were 1000-fold more efficient than sialic acid (Figure 11).

32

0

13- valent SA (mM)

0.005 0.05

0 1 5

SA (mM)

Figure 11: Multivalent sialic acid inhibits Ad37 from binding to and infecting HCE cells 1000-fold better than sialic acid. Top: Dose-dependent inhibition of Ad37 attachment to HCE cells with sialic acid (SA) and multivalent sialic acid HSA con-jugates 26a-c (3-, 8-, and 13-valent). Bottom: Dose-dependent inhibition of Ad37 infection of HCE cells with sialic acid or the 13-valent sialic acid HSA conjugate 26c. Each yellow-green dot corresponds to one infected cell. The data represent an average of at least three independent experiments for each assay (for details see Paper II). Printed with permission from the publisher; Elsevier.

3.5 Mode of action In section 1.2, different mechanisms by which a multivalent ligand could enhance the affinity of a carbohydrate–protein interaction are outlined. In the specific example of the multivalency effect obtained above, the binding af-finity enhancement is most likely due to the mechanisms of aggregation and

33

steric stabilization. This assumption is based on the structures of the HSA conjugates and the viral fiber knob protein. HSA is an average sized protein with a molecular weight of 70 kDa. The lysine residues to which the saccha-rides can be conjugated (the most solvent-exposed) are evenly spread at the protein surface.109 The theoretical spacing between the saccharide moieties is at a minimum 10-20 Å and at a maximum 100-110 Å.‡110 Compared to the viral particle (Figure 5, section 1.3), the size of the HSA conjugates are equal to the size of the fiber knob proteins. Due to the large distance between the fiber knobs, it would not be possible to connect two fiber knobs situated at the same viral particle through one sialic acid HSA conjugate. In addition, since the distance between the three binding sites on one fiber knob is quite small, i.e. around 9 Å,§111 it is not likely that there is enough space for two or three saccharide-spacer-lysine units to bind simultaneously. Thereby the mechanisms of the chelate effect and the higher local concentration are un-likely. More likely one HSA conjugate is capable of binding to two or more fiber proteins from different viral particles simultaneously. This would result in a three-dimensional network of viruses and HSA conjugates as illustrated in Figure 12. This network would aggregate the viral particles and thereby prevent attachment and infection. In the eye, the aggregates could possibly be rinsed out by the tear flow.

Figure 12: A schematic illustration (not drawn to scale) of the multivalency effect when multivalent sialic acid neutralizes adenoviruses by aggregation. A) In the case of a monomeric inhibitor (sialic acid), all sialic acid interacting domains need to be occupied in order to prevent each virion from binding to and infecting target cells. B) In the case of a multivalent inhibitor (sialic acid conjugated to HSA), the virions are aggregated and thereby prevented from binding to and infecting the target cells.

‡ Representative distances between the most solvent-exposed lysins were measured from the crystal structure of HSA (1BMO.pdb) using the MOE software program.110 §Distance measured from the crystal structure of the Ad37–sialyllactose complex (1UXA.pdb) using sybyl7.2 software.111

34

In order to verify this mechanism of aggregation a centrifugation based ex-periment was set up.** 35S-labeled Ad37 virions were incubated with or with-out 13-valent sialic acid on HSA. The mixtures were centrifuged at increas-ing centrifugal force and the resulting supernatant and pellet were separated. Scintillation analysis of the radioactivity in the two fractions showed that virions treated with multivalent sialic acid were pelleted more efficiently than the virions that were not (Figure 13). At the highest centrifugal force most of the untreated virions were still found in the supernatant whereas those incubated with 13-valent sialic acid were pelleted at lower centrifugal force. These results clearly support the hypothesis that multivalent sialic acid neutralizes the virions by aggregation.

Figure 13: 13-valent sialic acid (26c, SA-HSA) neutralizes Ad37 virions by aggre-gation (S = supernatant, P = pellet). Printed with permission from the publisher; Elsevier.

3.6 Summary Synthetic multivalent sialylated glycoproteins that mimic the cellular recep-tor of EKC-causing adenoviruses have been synthesized and evaluated in two different assays as inhibitors of adenoviral cell attachment and infection. It was shown that the 3’-sialyllactose derivatives 20a-e and the less complex sialic acid derivatives 26a-c, efficiently prevented Ad37 from binding to and infecting HCE cells. Moreover, the results clearly demonstrate that the power of inhibition is significantly increased with higher orders of valency. The most potent conjugates inhibit Ad37 from binding to and infecting the cells with a 1000-fold lower IC50 value as compared to 3’-sialyllactose and

** Performed in collaboration with Dr. Niklas Arnberg and Emma Nilsson, Division of Virol-ogy, Department of Microbiology, Umeå University, Umeå, Sweden.

35

sialic acid. The two classes of conjugates are equally efficient as inhibitors, but sialic acid conjugates provide a much more efficient route to an inhibitor than the more complex 3’-sialyllactose conjugates. Synthesis of 26a-c were performed in only four steps while the 3’-sialyllactose conjugates 20a-e re-quired 11 steps and gave a lower overall yield. These differences are signifi-cant considering the time and cost of producing an antiviral drug. In view of the size and structure of the multivalent sialic acid conjugates and the virus particles, it was proposed that neutralization of the Ad37 virions are due to aggregation. This mechanism of the multivalent effect was successfully veri-fied by quantifying aggregation of Ad37 in presence of multivalent sialic acid by simple centrifugation. This fast and simple assay could be applied for initial screening of new multivalent inhibitor candidates and the overall results presented herein could be useful in efforts to develop an antiviral drug for treatment of EKC.

36

4. Design, synthesis and evaluation of nonnatural sialic acids and sialylmimetics

The study in Chapter 3 demonstrated that multivalent sialic acid conjugates are effective inhibitors of EKC-causing adenoviruses. However, in order to investigate the scope and limitations of a potential antiviral drug it was de-cided to optimize these structures further. The weak affinity of carbo-hydrate–protein interactions originates from features of the carbohydrate structure, i.e: i) the lack of hydrophobic groups, ii) the low contribution of hydrogen-bond interactions from the hydroxyl groups to the overall binding affinity due to competing solvatisation and iii) the significant entropy cost when several, in general flexible, hydrogen-bonding groups are con-strained.112 In order to improve the affinity of carbohydrates to target pro-teins, as well as other properties of significant importance for their use as therapeutic agents (i.e. metabolic stability and bioavailability), modified carbohydrates or compounds that mimic carbohydrates have been designed. This can be accomplished by two main strategies. In the first approach the parent carbohydrate core is maintained but the structure is modified to achieve appropriate properties by removal, addition or change of functional groups. In the second approach, the entire carbohydrate core is replaced with a noncarbohydrate framework presenting only those functional groups in-volved in critical receptor interactions.20,112 Regardless of which method that is used, the success is highly dependent on proper knowledge of the inter-actions between the carbohydrate and the target protein. In this study, struc-ture-based design and docking were used to increase the understanding of the recognition of sialic acid by EKC-causing adenoviruses. Based on these results, a library of N-acyl modified sialic acids and a sialylmimetic were designed, synthesized and evaluated.

4.1 Structure-based design and docking of nonnatural analogues of sialic acid Structure-based design has been used extensively in drug development and is proven to be a valuable tool in the design of new chemical entities.113 One key methodology used in both hit identification and lead optimization is docking of small molecules into the protein binding site. The docking pro-

37

cess consists of two parts, accurate prediction of the bioactive conformation and orientation (pose) of a ligand in the binding site, and prediction of the binding affinity of the protein–ligand complex. Docking has been applied in pharmaceutical research since the early 1980s and a large number of docking programs have been reported in the literature such as DOCK, FlexX, Glide and GOLD. In general, all docking programs consist of three main parts, i) a function to place the ligand in the binding site, ii) a search algoritm to ex-plore different binding modes, and iii) a scoring function to rank different binding poses. The key characteristic of a good and reliable docking pro-gram is its ability to reproduce the experimental binding mode of ligands and the ability of its scoring function to score and rank ligands according to their experimental binding affinity.114,115 However, before these methods can be used to study the interaction of interest, a high qualitative three-dimensional model of the protein structure is required. The primary method for determin-ing protein structures is by X-ray crystallography, discussed in more details in section 5.2. Two other methods commonly used for solving protein-structures, not discussed in this thesis, are solution NMR experiments116 or homology modeling.117,118 In the Protein Data Bank (PDB) a large number of structures of proteins and other biological macromolecules are available for free.

Figure 14: Left; The molecular surface of the sialic acid interacting domain of the fiber protein of Ad37 in complex with 3’-sialyllactose. The fiber protein consists of a homotrimer presenting three identical binding sites for sialic acid. Right; Four interactions assumed to be essential for recognition and function of sialic acid as cellular receptor for Ad37: A salt bridge from the axial carboxyl group to Lys345, two hydrogen bonds between 4’OH and Tyr312, and 5’NH and Pro317, and a hy-drophobic interaction between the N-acetyl group and the hydrophobic bottom of the binding site. The surface is colored blue for polar parts, brown for nonpolar parts and green for intermediate polarity.

38

From the X-ray structure of the complex of 3’-sialyllactose bound to the fiber protein of Ad37, four main interactions between sialic acid and the protein were identified (Figure 14).105 The axial carboxylic acid forms a salt bridge with the amino group of a lysine close to the surface of the binding site, the 4’-hydroxyl group and the 5’-amide are involved in hydrogen bonds deeper inside the binding site, and the N-acetyl group of sialic acid is di-rected into a deep hydrophobic pocket in the bottom of the binding site. Most likely all of these interactions are essential for recognition and function of sialic acid as a cellular receptor. The glycerol moiety on the other hand, does not seem to be involved in any important interactions. Moreover, the hydrophobic pocket of the binding site continues into a bigger and even more hydrophobic cavity. With this information in hand, the work towards a high affinity inhibitor started based on the parent sialic acid structure. It was decided to keep all the polar functionalities involved in the four main inter-actions and to focus the work on the possibilities to increase the affinity through hydrophobic interactions. It was hypothesized that a large and hy-drophobic substituent at the sialic acid amine would be directed into the deeper cavity in the bottom of the binding site, and thereby capturing addi-tional hydrophobic interactions. Removal of the glycerol chain would in-crease the hydrophobicity and have the possibility to increase the affinity due to reduced entropy cost for desolvation of polar functional groups. In order to predict the effect of hydrophobic substituents at the amine and re-moval of the glycerol chain, a set of modified sialic acids were selected to be docked into the X-ray structure of the Ad37 fiber protein (Figure 15).

HOOH

HO

OHO

COOH

OH

RHN

A, R = COEtB, R = COPrC, R = COiPrD, R = COBuE, R = COPh

OHO

COOH

OH

AcHN

F

Figure 15: A set of six modified sialic acids (A-F) were docked into the sialic acid binding site of Ad37 and investigated for their potential to increase the binding affinity

39

At the time for this project, GOLD119 had been reviewed as one of the most reliable programs in predicting accurate binding poses.115,120 In order to evaluate the reliability for GOLD in this specific case, the ability to repro-duce the experimental binding mode of sialic acid was studied. The 3’-sialyllactose ligand was taken out of the X-ray structure and a new confor-mation of sialic acid was docked back into the binding site. All docked poses of the ligand were clustered within a 1.5 Å root mean square deviation (rmsd) from the experimental binding mode of sialic acid, indicating reliabil-ity of the docking (Figure 16).120

Figure 16: Overlay of the calculated binding poses of sialic acid (displayed as lines) and the experimental binding pose of the sialic acid residue of 3’-sialyllactose (dis-played as capped sticks) in the complex with the Ad37 fiber protein. An average rmsd value of 1.5 Å between the X-ray structure and the calculated structures veri-fies a successful prediction of the binding mode by the docking program.

Based on these results the same docking method was used on the modified sialic acids A-F shown in Figure 15. For all sialic acids, 20 docking poses were generated and evaluated manually by visual inspection. Compounds with a majority of the docked poses clustered close to the experimental bind-ing pose of sialic acid, and in a plausible conformation finding the hydrogen bonds and the salt bridge were predicted as good binders with potential to increase the binding affinity. Compounds with docked poses spread out in the binding site and not clustered at all were assumed to have unspecific biding modes and predicted to be poor binders. The docked poses of ligand F with the glycerol chain removed, were tightly clustered with a similar rmsd to the experimental binding mode of sialic acid as docked sialic acid (data not shown). This result support the hypothesis that the glycerol chain is of minor importance for the binding of sialic acid to the Ad37 fiber protein.

40

The results of the N-acyl modified derivatives A-E revealed that it was not as easy as expected to access the deep cavity that extends the binding site (Figure 17). Sialic acids with larger substituents failed to adopt conform-ations with the substituent directed into the cavity and instead they were forced to a binding mode not likely to pick up any of the important interac-tions discussed above. Only the propaneamide was predicted as a good binder (Figure 17a). A significant majority of the docked poses were clus-tered close to the experimental binding pose of sialic acid in a conformation with possibility of finding all four important interactions. The poses of the butane- and the pentaneamides were not clustered at all and none of them were close to the experimental pose of sialic acid (Figure 17b). The isobu-tane- and the benzamide were expected to give nonclustered poses as seen for the butane- and the pentaneamide. Surprisingly, they were closely clus-tered in one single binding pose for the benzamide and two binding poses for the isobutaneamide (Figure 17c and d). These results indicate a plausible second binding possibility inside the binding site even though the confor-mations fails to pick up the three hydrogen bonds.

B

C D

A

Figure 17: Overlays of the calculated binding poses for N-acyl modified sialic acids (displayed as lines) and the experimental binding pose of the sialic acid residue of 3’-sialyllactose (displayed as capped sticks) in the complex with the Ad37 fiber protein. A) The docked structures of the propaneamide of sialic acid are clustered close to the experimental binding pose. B) The poses of the pentaneamide are not clustered at all. The poses of the C) isobutaneamide and the D) benzamide are clus-tered in new binding modes with major differences compared to the experimental pose of sialic acid.

41

The propaneamide was the only modified sialic acid predicted to be a good binder based on the criteria discussed above. It is, however, important to point out that this prediction is based on a computer simulated protein–ligand interaction. It can not be excluded that any of the other compounds could be active, or that the propaneamide could be inactive in a biological system. As for most docking approaches, the docking algorithm used by the GOLD program only allows the ligand to be flexible and considers the pro-tein to be rigid.114,120 Thus, features such as induced fit or other conforma-tional changes that could occur upon binding are not taken into account for example, a structure estimated to be too big to fit the binding site could actu-ally be a good binder after conformational changes of the protein.113 There-fore it is relevant to evaluate both good and poor binders experimentally to explore the possibilities and limitations of the binding site. On the basis of this it was decided to synthesize and evaluate both the promising pro-paneamide and the butane- and pentaneamides. Since the docking of the benzamide and the isobutaneamide indicates other binding possibilities in-side the binding site than the experimental pose of 3’-sialyllactose, these compounds were also decided to synthesize. In order to further expand the variation within the library a couple of compounds with small and hydro-philic substituents were included. The final library selected for synthesis is visualized in Figure 18.

O

O

F

FF

HO

HN

R =O

COOH

OH

OH

HOOH

HO

NHR

O

J

E

G

A

B

C

D

F

H

I

Figure 18: Structure-based design and docking was used to design a library of ten sialic acids with modified N-acyl chains A-J.

4.2 Synthesis and biological evaluation of N-acyl modified sialic acids In the first part of this study, the nonnatural N-acyl modified sialic acids designed in section 4.1 were to be synthesized, biologically evaluated and compared to sialic acid in order get information about how modifications of the N-acyl chain affect the affinity to the Ad37 fiber protein. Secondly, the multivalent counterparts of potential inhibitors were to be synthesized through the same squaric acid/HSA methodology used in the previous study with multivalent sialic acid and 3’-sialyllactose conjugates. According to

42

this, it would be valuable if the synthetic strategy to generate the library dis-played in Figure 18 also provides an efficient route to the multivalent pre-cursors of the library, i.e. their corresponding aminopentyl sialosides. The sialic acid derivatives 27 (Figure 19) fulfill these requirements and were thereby chosen as the key building blocks in this study. The anomeric thio-phenyl group of 27 offers efficient protection of the anomeric center during the synthesis of the modified sialic acids21 and can be converted to an ano-meric hydroxyl group to afford the target structures 28. In addition, sialyl-ation of Fmoc-aminopentanol with 27 according to the procedure described in section 3.4 would provide a route to the aminopentyl derivatives 29.

27

O

COOH

R1

OH

HOOH

HO

NHR

O

28, R1 = OH29, R1 = O(CH2)5NH2

O

COOMe

SPh

OAc

AcOOAc

AcO

NHR

O

Figure 19: The target structures 27 can be converted to both the N-acyl modified sialic acids 28 and to their multivalent precursors 29.

Synthesis of N-acyl modified sialic acids The first challenge in the synthesis of the target sialosides 28 was the deacet-ylation of the acetamide of sialic acid. Direct hydrolysis of the acetamide under either acidic or basic conditions has been reported to cause decomp-osition of the sialic acid structure.121 Therefore a milder procedure was used. The acetamide of sialic acid derivative 30 (Scheme 7) was converted into the t-butoxycarbonyl (Boc) acetamide 31, followed by deacetylation and O-reacetylation to give 32. Subsequent Boc removal gave the amine 33.122,123 Acylation of 33 to the modified amides 27a-e was performed with the corre-sponding anhydride in the presence of DMAP. These amide couplings were slower than expected and it was observed that when using the larger anhyd-rides, O to N acetyl migration or possibly direct O-acylation occurred. In order to minimize the formation of such products, the reactions were stopped when the consumption of the amine seized according to LCMS. To afford the more functionalized derivatives 27f-h, benzoic acid activated by HATU, methylchloroformate, and methylisocyanate were used respectively in the presence of DIPEA. The crude products were purified by column chroma-tography, and additional preparative HPLC when necessary. This gave com-pounds 27a-h in 52-89 % yield from 32.

43

O

COOMe

SPh

OAc

AcOOAc

AcO

R1R2NO

COOMe

SPh

OAc

AcOOAc

AcO

RCOHN

27e, R = CF327f, R = Ph27g, R = OCH327h, R = NHCH3

O

COOMe

OH

OAc

AcOOAc

AcO

RCOHNO

COOH

OH

OH

HOOH

HO

RCOHNO

COOH

O

NH

HOHO

HO R

O

27a, R = CH2CH327b, R = CH2CH2CH327c, R = CH(CH3)227d, R = CH2CH2CH2CH3

(e)

O

COOH

SPh

OH

HOOH

HO

RCOHN

38f, R = Ph38g, R = CH3O38h, R = CH3NH38i, R = OC(CH3)3

38a, R = CH2CH338b, R = CH2CH2CH338c, R = CH(CH3)238d, R = CH2CH2CH2CH338e, R = CF3

30 R1 = Ac, R2 = H

31 R1 = Ac, R2 = Boc

32 R1 = H, R2 = Boc

a)

b)

c)

O

COOMe

SPh

OAc

AcOOAc

AcO

H3NCF3COO

33

d)

f)

27a-h32

e)

f)

34a, R = CH2CH334b, R = CH2CH2CH334c, R = CH(CH3)234d, R = CH2CH2CH2CH334e, R = CF3

35, R = OC(CH3)3

36a, R = CH2CH336b, R = CH2CH2CH3

37a, R = CH2CH337b, R = CH2CH2CH3

Scheme 7: Synthesis of amide functionalized sialic acids. Reagents and conditions: a) Boc2O, DMAP, THF, reflux, 2 h; b) 1. NaOMe, MeOH, r.t., 3 h, 2. Ac2O, Py, r.t., O.N.; c) 90% TFA (aq), CH2Cl2, r.t., 1-2 h; d) 27a-e: (RCO)2O, DMAP, CH2Cl2, r.t., 5-11 h, 27f: PhCOOH, HATU, DIPEA, CH2Cl2, r.t., 3 h, 27g: CH3OCOCl, DIPEA, CH2Cl2, r.t., 1 h, 27h: CH3NCO, DIPEA, CH2Cl2, r.t., 4 h, (52-89 % from 32); e) NIS, MeCN:H2O 10:1, r.t., 3 h, (42-60 %); f) 1. NaOMe, MeOH, 2. LiOH (aq), MeOH, (44-94 %).

Unfortunately, deprotection of compounds 27a-h turned out to be an unfore-seen challenge. First, hydrolysis of the thiophenyl derivatives 27a-e and 32 by NIS, H2O and MeCN124 afforded only moderate yields of 34a-e and 35 (hydrolysis of 27f-h not investigated). Second, and more decisive, the fully deprotected sialic acids could not be isolated after the final steps, i.e. deace-tylation and methyl ester hydrolysis. Although the methyl ester hydrolysis was performed under relatively mild conditions (6 eq LiOH (aq) at room temperature) structural studies show that the sialic acid scaffold was de-graded. Deprotection of 34a and b by these conditions resulted in a mixture of sialic acids 36a-b and the degraded compounds 37a-b. This is probably due to the involvement of the keto α-hydrogens in the open-chain form ac-cording to the mechanism outlined in Scheme 8, as reported by others.125,126 Despite several attempts to avoid this transformation it was not possible to prepare and isolate fully deprotected N-acyl modified sialic acids. However, the corresponding deprotected thiophenyl derivatives 38a-i were synthesized in 44-94 % yields from 27a-h and 32 by convenient deacetylation and hydrolysis.

44

OCOOH

O

NH

HO

HO

HO

34a, b-H2O O

CO2Li

O

NH

HO

HO

HO

HOH

HOOH

HO

CO2Li

O

39 40 37a, b

HNR

O

R

O

R

O

Scheme 8: Suggested mechanism for the alkaline degradation of sialic acid126. R = CH2CH3 or CH2CH2CH3.

Since the anomeric hydroxyl group of sialic acid plays a critical role in the degradation mechanism above it was decided to replace that group so that it was not possible to form the open chain of sialic acid. A methyl group was considered to be a good replacement for the hydroxyl group in terms of bind-ing properties. The crystal structure of 3’-sialyllactose in complex with the fiber knob of Ad37 showed that the lactose residue is directed out of the binding site and therefore a small and neutral substituent in the anomeric position of sialic acid is suggested not to affect binding. This was verified by comparing sialic acid and α-O-methyl sialic acid as inhibitors of Ad37 at-tachment to HCE cells using the binding assay described in section 3.3. No significant difference between the two compounds could be detected (data not shown). In addition, there are published procedures for straightforward synthesis of the α-methylglycoside, compatible with the chemistry required for the N-acyl substitution.127 Therefore, it was decided to synthesize an equivalent library to the one in Figure 18 based on the corresponding α-methylsialosides (Scheme 9). The functionalization of 41 was performed in a similar way as described for the thiosialosides (Scheme 7). However, instead of the anhydrides, the corresponding HATU-activated carboxylic acids were used to afford 46a-e. This procedure shortened the reaction time and mini-mized the acetyl migration. Derivatives 46f-h were synthesized by use of trifluoroacetic anhydride, methylchloroformate and methylisocyanate, re-spectively. Pure products were afforded by repeated column chromatography and all derivatives were obtained in reasonable yields (26-92 % from 44). The target compounds 47a-i were obtained after O-deacetylation and methyl ester hydrolysis of 46a-h and 43 (74-100 % yield). However, the trifluoroacetamide of sialic acid, 46f, was partly hydrolyzed during the de-protection. Since it was not possible to separate the amine from the amide, it was excluded from the library.

45

O

COOMe

OMe

OR3

R3OOR3

R3O

R1R2NO

COOMe

OMe

OAc

AcOOAc

AcO

RCOHN

46e, R = Ph46f, R =CF346g, R = OCH346h, R = NHCH3

O

COOH

OMe

OH

HOOH

HO

RCOHN

46a, R = CH2CH346b, R = CH2CH2CH346c, R = CH(CH3)246d, R = CH2CH2CH2CH3

dO

COOMe

OMe

OAc

AcOOAc

AcO

H2NCF3COO

45

e)

43

47f, R =CF347g, R = OCH347h, R = NHCH347i, R = OC(CH3)3

47a, R = CH2CH347b, R = CH2CH2CH347c, R = CH(CH3)247d, R = CH2CH2CH2CH347e, R = Ph

g)

f, g)

41 R1 = Ac, R2 = H, R3 = Ac

42 R1 = Ac, R2 = Boc, R3 = Ac

43 R1 = H, R2 = Boc, R3 = H

44 R1 = H, R2 = Boc, R3 = Ac

a)

b)

c)

Scheme 9: Synthesis of N-acyl modified α-O-methylsialosides. Reagents and condi-tions: a) Boc2O, DMAP, THF, reflux, 2 h; b) NaOMe, MeOH, r.t., 3 h, 65 % from 41; c) Ac2O, Py, r.t., O.N., 92 %; d) 90% TFA (aq), CH2Cl2, r.t., 1-2 h; e) 46a-e: RCOOH, HATU, DMAP, CH2Cl2, r.t., 3-5 h, 46f: (CF3CO)2O, DMAP, CH2Cl2, r.t., 0.5 h, 46g: CH3OCOCl, DIPEA, CH2Cl2, r.t., 1.5 h, 46h: CH3NCO, DIPEA, CH2Cl2, r.t., 4 h (26-92 % from 44); f) NaOMe, MeOH, r.t., 2 h; g) 2M LiOH (aq), MeOH, r.t., O.N. (74-100 %, 2 steps).

Biological evaluation of N-acyl modified sialic acids The N-acyl modified sialic acids 47a-i (f excluded) and commercial Neu5Gc (N-glycolylneuraminic acid) were evaluated and compared with sialic acid for their ability to inhibit Ad37 attchment to HCE cells, using the binding assay described in section 3.3. The experimental settings were adjusted in order to minimize the consumption of the compounds, but otherwise it was performed as before (Table 2, for details see Paper III). The results varied between the experiments and the standard deviations were large, whereby only rough conclusions about the inhibitory potency of the compounds could be drawn. Compounds that prevented Ad37 from attaching to HCE cells to a similar extent as sialic acid were defined as potential inhibitors and are de-noted + + in Table 2, compounds that were slightly active but not as good as sialic acid are denoted +, and compounds with no effect are denoted –. It was found that except for sialic acid, only the propaneamide 47a was effective as inhibitor of Ad37 attachment to HCE cells. The benzamide 47e and the methylcarbamate 47g were as active as sialic acid in one experiment respect-ively but in the case of 47g the data were very uncertain. All of the other compounds were concluded to have no effect as inhibitors.

46

Table 2: Inhibition of Ad37 attachment to HCE cells with the N-acyl modified sialic acids 47a-i (f excluded) and commercial NeuGc. The experiments consist of a dilu-tion series of each compound (2, 4, 8 and 16 mM) and the data presented in this table represents an average of two duplicates. Compounds are denoted + + if they prevented Ad37 from attaching to HCE cells to the same extent as sialic acid or more, + if the were slightly effective, and – if they had no effect at all.

Compound experiment 1 experiment 2 experiment 3 SA + + + + + + 47a + + no data + + 47b - + - 47c - - no data 47d - + - 47e + + + + 47g + + + - 47h + - + 47i - - -

Neu5Gc - - -

In order to obtain more information it was decided to further investigate compounds that were somewhat active in the binding assay. Derivatives 47a-c, 47e and 47g were therefore evaluated for their ability to inhibit Ad37 in-fection of HCE cells, using the infectivity assay described in section 3.3. The experiments were repeated twice and produced reproducible data. The re-sults confirmed the activity of 47a and 47g (Figure 20). The benzamide 47e on the contrary, was found to have no effect in the infectivity assay even though it seemed quite promising in the binding assay. Compounds 47b and 47c were included as negative controls and as expected they had no effect as inhibitors. The results from these biological experiments and the docking studies in section 4.1 are in good agreement, only the smallest N-acyl sub-stituents fits into the binding site and compounds with larger substituents on the amine fail to bind. It is fascinating that propaneamide 47a is a potent inhibitor of Ad37 HCE cell attachment and infection, while the butaneamide 47b that only differs from 47a by a single methylene group has no effect at all. These results indicate that there is little flexibility in the sialic acid inter-acting site of the Ad37 fiber protein. Although it was a disappointment that the sialic acids with larger N-acyl groups failed to occupy the hydrophobic cavity of the binding site, it was encouraging to find that different N-acyl groups clearly influenced the binding.

47

Ad37 infection of HCE cells in presence of N-acyl modified sialic acids

Figure 20: Dose-dependent inhibition of Ad37 infection of HCE cells with N-acyl modified sialic acids 47a, 47b, 47c, 47e and 47g (black lines), compared to sialic acid (SA, dashed lines). The derivatives 47a and 47g were effective as inhibitors in the same range as sialic acid while 47b, 47c and 47e showed no or only modest activity. The dilution of virions was adjusted in order to obtain approximately 100 infected cells per view field in the control wells (untreated cells). The data repre-sents an average of duplicates within one experiment and the experiment was re-peated twice with reproducible results.

48

4.3 Synthesis and biological evaluation of multivalent N-acyl modified sialic acids In order to further investigate the differences in inhibition obtained in section 4.2, the multivalent counterparts of compounds 47a-c were synthesized and evaluated. It was assumed that these conjugates would display a similar mul-tivalency effect as observed earlier and thereby increase the differences be-tween the compounds. The multivalent conjugates 52a-c were syntesized in seven steps starting from the thiophenyl donor 32 (Scheme 10). Fmoc-aminopentanol was sialylated with 32 using the procedure described in sec-tion 3.4 and 48 was obtained in 68 % yield. Boc removal and subsequent acylation afforded amides 49a-c in 18-64 % yield. Once again, larger sub-stituents resulted in lower yields. In addition, during the Boc removal step small extents of methyl ester hydrolysis and partial O-deacetylation were observed, which further explains the modest yield of the amides. The amides were deprotected to give 50a-c and the crude products were converted to the squarate amides 51a-c in 60-80 % yield from 49a-c. Finally 51a-c were con-jugated to HSA to afford the multivalent sialosides 52a-c, with 14-17 incor-porated saccharides per HSA respectively. The degree of incorporation was determined by MALDI-TOF MS, using the center of the single-charged neo-glycoprotein peak. The results were confirmed by gel electrophoresis.

49a, R = CH2CH349b, R = CH2CH2CH349c, R = CH(CH3)2

O

COOMe

O(CH2)5NHFmoc

OAc

AcOOAc

AcO

BocHNO

COOMe

O(CH2)5NHFmoc

OAc

AcOOAc

AcO

RCOHN

O

COOH

O(CH2)5NH2

OH

HOOH

HO

RCOHN

O

COOH

O(CH2)5

OH

HOOH

HO

RCOHN NH

OO

HSA

NH

O

COOH

O(CH2)5

OH

HOOH

HO

RCOHN OC10H21

OO

NH

b, c)

d, e) f)

g)

48

50a-c 51a-c

52a-c, n = 14-17

n

O

COOMe

SPh

OAc

AcOOAc

AcO

BocHN

32

a)

Scheme 10: Synthesis of multivalent HSA conjugates of N-acyl modified sialic acids. Reagents and conditions: a) Fmoc-aminopentanol, molecular sieves (3Å), AgOTf, 1M IBr in CH2Cl2, CH3CN:CH2Cl2 (3:2), −72 °C, 4 h, 68 %; b) 90 % TFA (aq), CH2Cl2, r.t., 1 h; c) RCOOH, HATU, DIPEA, CH2Cl2, r.t., 9-24 h, 18-64 % from 48; d) NaOMe, MeOH, r.t., 1 h; e) 1M LiOH (aq), MeOH, O.N.; f) didecyl squarate, DMF, Et3N, r.t., 11 h, 60-80 % from 49a-c; g) NaHCO3, HSA, r.t., 27 h.

49

Biological evaluation of multivalent N-acyl modified sialic acids The multivalent sialosides 52a-c were evaluated and compared to multiva-lent sialic acid (17-valent) for their ability to prevent Ad37 from infecting HCE cells (Figure 21), using the infectivity assay described in section 3.3. As expected, the conjugates with larger N-acyl substituents, i.e. 52b and c, showed no effect as inhibitors. The results obtained with the multivalent propaneamide 52a were more surprising. The nonconjugated propaneamide 47a was equally efficient as sialic acid in preventing Ad37 from HCE cell attachment and infection (Figure 20), but its multivalent counterpart 52a had much lower effect than multivalent sialic acid. It was only at the highest concentration (0.05 mM) that an effect was observed. However, 52a was more efficient than 47a that had no effect at concentrations below 4 mM.

Figure 21: Dose-dependent inhibition of Ad37 infection of HCE cells with 17-valent sialic acid (SA) and the multivalent N-acyl modified sialic acid conjugates 52a-c (14-17-valent). 17-valent sialic acid are the most potent inhibitor.

4.4 Evaluation of benzoic acids as sialylmimetics Although the results from the structure-based design and the docking study, indicates that it could be possible to increase the affinity by modifying the sialic acid structure, the biological evaluation of the N-acyl modified sialic acids indicates no improvements in affinity compared to the parent sialic acid structure. In addition, the synthesis of modified sialic acid is very com-plicated and time consuming, and not suitable for drug development. In or-der to find high affinity compounds and to decrease the synthetic complex-ity, it was desirable to replace the entire sialic acid structure with a non-

50

carbohydrate sialylmimetic that orients the required functional groups as in the parent sialic acid structure. During the recent years a number of sia-lylmimetics have been reported.20,112 A highly studied area is mimetics of the cell surface oligosaccharide sialyl LewisX (sLex, Figure 22) that interacts with the cell surface glycoprotein E-selectin. This interaction represents an initial step in the recruitment of white blood cells to damaged tissues. The sLex mimetic in Figure 22, where the entire sialic acid residue has been re-placed with a carboxylate group, is one of several mimetics that shows ex-cellent binding affinity for E-selectin.112,128,129 The success of sLex mimetics has encouraged development of sialylmimetics targeting other sialic acid recognizing proteins. One remarkable example is the anti-influenza drug Relenza, where a modified structure of the dehydrated form of sialic acid, Neu2en5Ac, mimics the transition state of terminal sialic acids in complex with sialidase during release of new viruses (Figure 22).130 During the recent years, a large number of mimetics of this transition-state analogue have been reported.131-134 Inspired by these studies, we decided to investigate the poten-tial of sialylmimetics as inhibitors of EKC-causing adenoviruses. A study with sialylmimetics based on a benzoic acid scaffold was therefore per-formed.

O

NHAc

OH

O OHO

OOH

HOOH

HOOH

OHOO

COOH

OH

HOOH

HO

AcHN

OMe

O OH

R

HOOC

O COOH

HOOH

HO

AcHN

HN

H2N

NH

AcHN

N

OEt

EtCOOH

sialyl LewisX (sLex)

O COOH

HOOH

HO

AcHN

OH

Neu2en5Ac

mimetic of sLex

COOH

H2N

NHAc

mimetics of Neu2en5Ac

Relenza

Figure 22: Successful mimetics of sialyl LewisX and Neu2en5Ac.20,131,132

Design and synthesis of benzoic acids mimicking sialic acid By using the same procedure as for the modified sialic acids in section 4.1, a set of substituted benzoic acids (Figure 23, 53a-i), presenting the important functionalities of sialic acid (COOH, 4’OH and NHAc) and with N-acyl modifications, were docked in the binding site of the fiber protein of Ad37 and evaluated manually by visual inspection. The docked poses of structures 53a, 53b and 53g almost perfectly overlapped their 1-COOH, 3-OH and 4-NAc substituents with the corresponding functionalites of sialic acid in its

51

experimental binding pose (Figure 23). The structures 53c-f and 53h failed to find these interaction points. As for the N-acyl modified sialic acids in section 4.1, structures with large substituents, were spread out in the binding site and theoretically to big to fit inside the binding site. However, the results of 53a, 53b, and 53g indicates that substituted benzoic acids could function as silylmimetics and be potential inhibitors of EKC-causing adenoviruses. To investigate this, compound 53a was synthesized by published proce-dures131,135, evaluated and compared to sialic acid.

Figure 23: Left: Substituted benzoic acids 53a-h investigated by docking as sia-lylmimetics. Right: Overlay of a calculated binding pose for sialylmimetic 53a (dis-played as lines) and the experimental binding pose of the sialic acid residue of 3’-sialyllactose (displayed as capped sticks) in complex with the Ad37 fiber protein. The docked structures of the sialylmimetic were close to the experimental binding pose of sialic acid, with the substituents on the aromatic ring oriented in the same directions as the corresponding functionalities of sialic acid (marked with dashed squares).

COOH

R2

NHAc

R1

COOH

OH

R3

HO

53a R1 = H, R2 =OH53b R1 = OH, R2 = OH53c R1 =CH2OH, R2 = OH53d R1 = OH, R2 = NH253e R1 =CH2OH, R2 = NH2

53f R3 = NHCOPh

53g R3 = NHCOEt53h R3 = NHCOiPr

52

Biological evaluation of a sialylmimetic In order to investigate whether the sialylmimetic 53a could inhibit Ad37 from binding to and infecting HCE cells, it was evaluated in the binding and infectivity assays described in section 3.3 (Figure 24). The results were con-tradictory. The sialylmimetic was potent as inhibitor of Ad37 attachment to HCE cells, but it had no effect as inhibitor of Ad37 infection of HCE cells unless very high concentrations were used. These results indicate that 53a binds to the fiber protein, but since it does not prevent the virus from infect-ing the cells this is most likely due to unspecific hydrophobic binding at the surface of the fiber protein and not at the specific binding site.

Figure 24: Biological evaluation of sialylmimetic 53a shows that it prevents Ad37 attachment to HCE cells to the same extent as sialic acid (top) while it has less effect as inhibitor of Ad37 infection (bottom) compared to sialic acid. The data represents an average of duplicates in one experiment, respectively. The experiments were repeated three times for the binding assay and twice for the infectivity assay, with reproducible results.

53

4.5 Summary In this study, structure-based design was used to study the possibility to en-hance the binding affinity of sialic acid to the fiber protein of Ad37 by N-acyl modifications. A docking study of N-acyl modified sialic acids was performed, and the results indicated that changes of the N-acyl group would affect binding. Ten nonnatural sialic acids were synthesized, biologically evaluated and compared to sialic acid using the HCE cell binding and infec-tivity assays described in section 3.3. During the synthesis of these com-pounds it was found that the sialic acid hemiketal structure was degraded under alkaline conditions. Therefore the N-acyl modified sialic acids were synthesized as their α-O-methylglycosides. It was found that only sialic ac-ids with small N-acyl groups, i.e. CH3CH2- (47a) and CH3O- (47g) instead of the methyl group of sialic acid, were efficient as inhibitors. Sialic acids with larger substituents showed no activity at all. However, as for carbohydrate–protein interactions in general, the sialic acid–Ad37 interaction is weak. In order to obtain more efficient inhibitors, the multivalent HSA conjugates 52a-c were synthesized and evaluated. Unfortunately the propaneamide 52a, that was as potent as sialic acid in the monomeric case, was found to be a much less potent inhibitor as compared to multivalent sialic acid. These re-sults indicate that the sialic acid–Ad37 interaction is more complex than expected. Although N-acyl group extensions seem promising from the crys-tal structure, the biological results reveal the opposite. The binding site seems to be very rigid and sialic acid is the most potent structure. In order to find an inhibitor with higher affinity than sialic acid, further structure-activity investigations are required. Since sialic acid synthesis is complicated and time consuming, it would be desirable to use a noncarbohydrate frame-work to mimic sialic acid. This would be advantageous in terms of less com-plex synthesis and an increased possibility of structural modifications. In this study a benzoic acid scaffold was investigated for its potential to mimic sialic acid. From docking studies it was found that benzoic acids substituted with a hydroxyl group in position 3 and an acetamide in position 4, could be positioned in the sialic acid binding site of Ad37. The functional groups were oriented as for the corresponding groups of sialic acid and all of the interactions believed to be important were found. The benzoic acid 53a, was synthesized, biologically evaluated and compared to sialic acid. Interest-ingly, this sialylmimetic prevented Ad37 attachment to HCE cells to a simi-lar extent as sialic acid, but did not affect infection of the cells. However, this benzoic acid structure was proven to mimic sialic acid to some extent and it is well motivated to examine its structure-activity relationships by synthesis of further analogues.

54

5. Structure-based methods to investigate interactions of ligands with Ad37

The two biological assays used to identify inhibitors that prevent Ad37 from bindning to and infecting HCE cells are very reliable and of high quality. However, these assays require large amounts of the compounds to be evalu-ated (milligrams). Often these compounds are obtained in limited amounts from complicated and expensive synthetic work. It would be preferred to use a less consumptive assay for initial screening and identification of potential inhibitors and then use the more sophisticated assays to rank identified hits. Initially surface plasmon resonance (SPR) using a Biacore 3000 instrument was extensively investigated for this purpose but with limited success.††136,137 In addition, neither SPR nor the binding assay or the infectivity assay pro-vide any information about how the inhibitors bind in the active site of the fiber knob protein. The sialic acid based inhibitors are assumed to bind in a similar fashion as sialic acid by using the same interaction points, but in the case of sialylmimetics one could assume alternative binding poses or even alternative binding sites. Therefore, in order to find a fast and less consump-tive assay for identification of potential inhibitors, 1H-NMR spectroscopy was evaluated for screening of ligands binding to the fiber knob protein of Ad37. X-ray crystallography was then used to obtain the binding poses of active inhibitors.

5.1 Relaxation-edited 1H-NMR spectroscopy for identification of potential inhibitors of Ad37 Relaxation-edited one dimensional 1H-NMR experiments make it possible to study the binding of small molecules to proteins. The methodology relies on the difference in T2 relaxation rate of macromolecules (proteins) compared to small molecules (ligands).

†† The sialyllactose–Ad37 interaction was used as model system for evaluating the binding of sialic acid derivatives to the Ad37 fiber knob protein. The fiber knob was immobilized to the sensorchip and sialyllactose was used as analyte. Various coupling techniques where investi-gated but difficulties in preparing a stable and reproducible sensorchip surface made the binding measurements unreliable.

55

Proteins have a very rapid relaxation and the experimental settings can be set so as to allow the signals from the protein to disappear, while signals from ligands in solution still can be observed, although with somewhat reduced intensity. Ligands bound to a protein adopt the relaxation behaviour of the protein during the time they are part of the protein–ligand complex. The observed T2 relaxation rate of the ligands will be an average of the rate of the free and bound forms. As a consequence, the relaxation time and thereby the signal intensity of a ligand will decrease as a function of binding strength. By measuring the intensity of ligand signals in absence and presence of pro-tein it is possible to identify ligands that bind to the protein (Figure 25). If the investigated ligands have comparable T2 relaxation rates, series of com-pounds can be investigated and ranked in terms of binding affinity.138,139

In this study, relaxation-edited 1H-NMR spectroscopy was used to identify and rank binding of the N-modified sialic acids 47a-i (f excluded), NeuGc and the sialylmimetic 53a to the fiber knob protein of Ad37 (Table 3, unpub-lished results, for details see Appendix II). Galactose was included as a non-binding control to allow determination of the observed reduction of signal intensities upon binding. The H3ax signal in the sialic acid core was used to determine binding of the sialic acids to the fiber knob. For the sialylmimetic 53a, the aromatic signals were used. Signal reduction was related to sialic acid that was used as a positive control.

Figure 25: Binding of sialic acid to the fiber knob protein of Ad37, studied by re-laxation-edited 1H-NMR spectroscopy. Shown here is the reduction of the H3ax signal intensity of sialic acid upon binding. Galactose (not visible within the spectral range of this figure) was added as a nonbinding control to allow determination of the reduced signal intensity.

56

For a majority of the compounds, the results from this study were in close agreement with the results from the binding and infectivity assays in Chapter 4. In presence of Ad37 fiber knobs, the 1H-NMR signals of compounds 47a, 47e and 53a were reduced to a similar extent as sialic acid or more, indi-cating binding. The signals of 47b, 47d, 47h, 47i, and NeuGc were not re-duced at all, indicating no bindning. For compound 47c and 47g lower signal reductions than for sialic acid were observed.

For the three compounds 47c, 47e and 47g the results obtained in this assay differed compared to the results obtained previously in the following way; i) compound 47c that had no effect on viral cell binding and infection, was here found to bind to the Ad37 fiber knob, ii) compound 47e that had some effect in the binding assay but no effect as inhibitor of infection was found to bind equally well as sialic acid and, iii) compound 47g that had varied re-sults in the binding assay but was effective as inhibitor of infection, was here found to bind only weakly to the fiber knob. The false positive results of 47c and 47e might be explained by unspecific binding to other sites of the pro-tein due to their hydrophobic N-acyl groups, but the weak binding of 47g is more puzzling. It is important to point out that this experiment was only performed once and needs to be reproduced. Unfortunately, it was not possi-ble to reproduce these results within the timelimits of this thesis due to diffi-culties in preparing fiber knob proteins with sufficient quality for NMR stud-ies. However, based on these preliminary results and with the assumption that they can be reproduced, this method is reliable for screening for poten-tial inhibitors of EKC-causing adenoviruses. The primary advantage of this method compared to the binding and infectivity assays used before is the small amounts of ligands (micrograms) required. An additional advantage of great importance for fast evaluation of many compounds, is that this method directly identifies potential inhibitors from mixtures of compounds.138

57

Table 3: Relative binding to the fiber knob protein of Ad37a (for experimental de-tails see Appendix II).

O

COOH

OMe

OH

HOOH

HO

RCOHN

47e, R = Ph47g, R = OCH347h, R = NHCH347i, R = OC(CH3)3

47a, R = CH2CH347b, R = CH2CH2CH347c, R = CH(CH3)247d, R = CH2CH2CH2CH3

COOH

OH

NHAc

53a

O

COOH

OH

OH

HOOH

HO

HNHO

O

NeuGc

Compound 53a 47a 47e NeuAc 47c 47g 47b 47d 47h 47i NeuGc

Relative affinityb

1.7 1.5 1.2 1 0.8 0.4 0 0 0 0 0

aAll spectra were scaled to have the same signal intensity for the nonbinding control, galac-tose. The reduction of the signal intensity upon binding to Ad37 was measured on H3ax for the sialic acids and on the aromatic signals for 53a. The binding was related to sialic acid. b >15% RSD from baseline variations. Signals reduced less than 15% were set to 0 and as-sumed to have no affinity.

Since 53a was shown to bind to the Ad37 fiber knob in this study and in the binding assay, but not in the infectivity assay, it was decided to investigate its mode of action further. In order to find out if this sialylmimetic binds to the same binding site as sialic acid, a competition study was performed. Thus, relaxation-edited 1H-NMR experiments were used to study the binding of 53a to the Ad37 fiber knob in prescence of sialic acid. As before galactose was used as nonbinding control and the samples were prepared as a dilution series of 53a with constant concentrations of sialic acid, galactose and Ad37 fiber knobs (for experimental details see Appendix II). As expected, at low concentrations, both the signals from sialic acid and the signals from 53a were reduced. Higher concentrations of the sialylmimetic did not result in recovered signals from sialic acid and the aromatic sialylmimetic signals were still reduced. This indicates that the sialylmimetc binds to the fiber knob at a different binding site than sialic acid, or that it binds by unspecific binding.

58

5.2 X-ray crystallography studies of nonnatural sialic acids in complex with the Ad37 fiber knob X-ray crystallography is the primary method to obtain structural information with atomic resolution for proteins and other biological macromolecules such as DNA and RNA. The method interprets the diffraction pattern that emerge from a crystal of the investigated protein when targeted by an X-ray beam (Figure 26). The directions and intensities of the X-rays diffracted from the crystal are received and measured by a detector and converted to an image of the crystal contents by computer simulations. Since the X-ray beam is diffracted by the clouds of electrons belonging to the molecules in the crystal, the resulting image is a three-dimensional electron density map. Conversion of the electron density map to a structure of the crystallized pro-tein is performed by building a molecular model of the protein that fits real-istically to the map. In general, a known amino acid sequence of the protein is fitted piecemeal to the electron density map followed by a series of re-finements. The final output is a molecular model of the protein structure presented as x, y, z Cartesian coordinates for each atom in the protein (not hydrogens). This model can be used in studies of molecular action and inter-action, protein foldings or protein function. It is thus important to point out that the determined structure is a static model describing a dynamic structure with vibrations, torsions and movements. Understanding of the shortcomings of the crystallographic model such as the uncertainty of the precise positions of the atoms, disordered regions, unexplained density and undetected parts of the sequence is essential for proper use of the model.140

Figure 26: Schematic illustration of an X-ray diffraction experiment for biological macromolecules.

59

In this study, X-ray crystallography was used to investigate the binding of 47a, 47e, 47g and 53a to the fiber protein of Ad37.‡‡ From the results of the docking study and the biological evaluation of these compounds presented in chapter 4 and 5.1 it is suggested that, 47a and 47g bind to the Ad37 fiber knob in a similar fashion as sialic acid, whereas the benzamide 47e and the sialylmimetic 53a are suspected to bind via other interactions. The aim with this study was to determine the active conformation of the potential inhibi-tors 43a and 43g and to investigate how (if) 47e and 53a bind to the fiber knob, in order to explain their action as inhibitors of attachment but not of infection.

Co-crystals of the Ad37 fiber knob in complex with each ligand were grown by the hanging drop method.105,140 Data were collected using synchrotron radiation and the crystals diffracted to 1.8 Å resolution. The structures were solved by molecular replacement using the previous structure of the complex of Ad37–sialyllactose105 (unpublished results, for experimental data see Ap-pendix III). As control, crystals with sialyllactose were reproduced. Unfortu-nately, the ligands 47e and 53a were not visible in the structure of the known binding site of sialic acid or anywhere at the fiber protein, indicating no binding. Most likely the results obtained for these compounds as potential inhibitors of Ad37 attachment, result from unspecific binding to the fiber protein. More successful results were obtained for the structures of com-pound 47a and 47g. They were both clearly observed within the sialic acid binding site and with well defined electron density (Figure 27). Overlays with the Ad37–sialyllactose complex showed that both 47a and 47g were positioned in the same manner as sialic acid. All of the important interac-tions were fulfilled and the extra atoms on the N-acetyl group were directed into the hydrophobic part of the binding site. In addition, it was found that the compounds but not the protein residues were slightly shifted with respect to bound sialic acid. These results, in combination with the biological evaluations in Chapter 4, indicate that the protein is rigid. However, it is surprising that the binding site does not tolerates large N-acyl substituents on the sialic acid since it from the crystal structures seems to be space enough for N-acyl extensions.

‡‡ This study was performed in collaboration with Drs. Delphine Gulligay, Thibaut Crepin and Stephen Cusack, EMBL, Grenoble, France.

60

Figure 27: The crystal structures of the N-acyl modified sialic acids 47a (Left) and 47g (Right), in complex with Ad37 fiber knob protein has been solved to 1.8 Å resolution by molecular replacement. Both ligands are observed with well defined electron density (Top) within the the binding site of sialic acid (Middle). Overlays with the sialyllactose–Ad37 structure (Bottom) show that the N-acyl modified sialic acids binds to the binding site in a similar manner as sialic acid, with the extra atoms on the acetyl group directed down to the hydrophobic part of the binding site. Black electron density map: unbiased difference density (2.50 σ), Blue: final refined map (0.95 σ); White residues: chain C of the homotrimer of the Ad37 fiber protein, yel-low residues = chain B; Orange structure: N-acyl modified sialic acid, Pink: sialic acid.

61

5.3 Summary Relaxation-edited 1H-NMR spectroscopy was evaluated and confirmed to be a fast and efficient method for identification of potential inhibitors of EKC-causing adenoviruses. Binding of the N-acyl modified sialic acids 47a-i (f excluded) and the sialylmimetic 53a to the Ad37 fiber knob was evaluated by this method and the results were comparable to those from the cell based assays used in Chapter 4. Biological evaluation of ligands by 1H-NMR spec-troscopy is advantageous since a much lower amount of the ligands is re-quired and since it is possible to evaluate mixtures of compounds. This method could be suitable for initial screening of new potential inhibitors assuming that the results observed in this study are reproducible. Com-pounds 47a, 47e, 47g, and 53a, were crystallized with the Ad37 fiber knob protein and studied by X-ray crystallography. The results show that 47a and 47g bind to the fiber knob in a similar binding mode as sialic acid with the extra atoms on the acetyl group directed towards the hydrophobic bottom of the binding site. The sialyl mimetic 53a and the sialic acid benzamide 47e were not found to bind to the fiber knob, indicating that their efficiency in preventing Ad37 from attaching to HCE cells is more likely a result of un-specific binding than alternative binding poses or binding sites. The results obtained in this study could be useful in development and evaluation of fu-ture inhibitors of EKC-causing adenoviruses. The rate of success will in-crease by use of efficient methods for fast and reliable hit identification and detailed structural information of their interaction with the fiber knob.

62

6. Concluding remarks and future prospectives

Prior to this work, studies were published suggesting that EKC-causing ade-noviruses use oligosaccharides with terminal sialic acids as cellular receptor. It was also hypothesized that sialic acid could be used as a component in an antiviral drug targeting viral adhesion. At present there are no licenced drugs for treatment of EKC. This thesis describes the synthesis and biological evaluation of sialic acid derivatives as inhibitors of EKC-causing adenovi-ruses. The main achievements and conclusions from this work can be sum-marized as follows:

• Efficient synthetic methods to achieve derivatives of sialic acid and 3’-sialyllactose have been developed, and a general route for conjugating these derivatives to a carrier, for example proteins, have been described.

• Multivalent HSA conjugates of 3’-sialyllactose and sialic acid prevent

EKC-causing adenoviruses from attaching and infecting HCE cells. This verifies that EKC-causing adenoviruses use sialic acid as a cellular recep-tor and support the hypothesis of using sialic acid in an antiviral drug.

• The cellular receptor of EKC-causing adenoviruses most likely consists of

a complex oligosaccharide with terminal sialic acid. The work in this the-sis elucidates that sialic acid alone is sufficient for viral recognition, whereby the synthesis of potent inhibitors is significantly reduced.

• Carbohydrate–protein interactions are generally weak. Herein the low

millimolar affinity of sialic acid to the Ad37 fiber protein has been effi-ciently enhanced by preparation of multivalent sialic acid HSA conju-gates. The results clearly show that the inhibitory power is significantly improved with an increasing order of multivalency.

• The possibilities to improve the binding affinity by structural modifica-

tions of the sialic acid core were investigated by structure-based design using the crystal structure of the Ad37 fiber protein in complex with sia-lyllactose. Synthesis and biological evaluation of N-acyl modified sialic acids revealed that only relatively small modifications were tolerated. Un-fortunately no high affinity structures were found, but valuable know-ledge for future efforts to find potential inhibitors was gained.

63

• Efficient methods for biological evaluation are crucial for identification of potential inhibitors. In this thesis two efficient assays for rapid screening of new compounds are presented. Binding of small molecules to the Ad37 fiber knob protein was studied by relaxation-edited 1H-NMR spectros-copy. It was also found that the potency of multivalent structures to neu-tralize EKC-causing adenoviruses can be studied by an aggregation based assay.

Based on these conclusions there are several openings for development of the next generation of inhibitors. The primary focus for future research would be efficient assays for screening for potential inhibitors. The prelimi-nary results of the NMR study presented in this thesis were very promising and should be further investigated. If Ad37 fiber knob protein could be iso-lated in sufficient quality for relaxation-edited 1H-NMR studies, this method could be ideal for screening of large sets of compounds. An alternative method for hit identification, already under investigation, is the use of car-bohydrate microarrays to study the interaction of ligands with Ad37. A car-bohydrate array with sialic acid conjugated to microtiterplate wells or glass plates could be easily constructed, using the same method of conjugation as for the synthesis of HSA conjugates presented in this thesis. By a competi-tion study, where the sialic acid–Ad37 interaction is studied in presence of potential inhibitors, inhibitors with higher affinity to Ad37 than sialic acid could be identified. In the search of successful inhibitors it would be interest-ing to further investigate the potential of sialylmimetics based on benzoic acid. Although the sialylmimetic synthesized in this work had modest effi-ciency in preventing Ad37 from infecting HCE cells, it prevented attachment and the docking studies of related structures were promising which moti-vates a more extensive study. Noncarbohydrate sialylmimetics are advanta-geous in terms of less complex synthesis and the possibilities to perform modifications. They are also more suitably for drug development. Since the results in this thesis indicate that the sialic acid interacting site of Ad37 is rigid, it seems likely that the multivalent display needs to be more properly investigated to find an optimal arrangement of inhibitors. It would for in-stance, be very interesting to investigate the possibility to link three inhibi-tors together for simultaneous binding to the three binding sites of the Ad37 fiber knob.

To the best of our knowledge, the present study is the first one to reveal that synthetic multivalent conjugates can be used to inhibit EKC-causing adeno-viruses. Efficient inhibitors could easily be transferred to the eye through a salve or eye drops, and thereby they could escape metabolism. The results presented herein provides valuable information and tools for future efforts to develop an antiviral drug and the chances to find a treatment for EKC in short time seems promising.

64

7. Populärvetenskaplig sammanfattning

Virus och bakterier har i alla tider orsakat svåra sjukdomar hos människor och djur. Människan har försökt ligga steget före och komma på effektiva behandlingsmetoder. Bakteriella sjukdomar har sedan 40-talet kunnat be-handlas med framförallt antibiotika. Virus är däremot mycket svårare att bekämpa eftersom de lever som parasiter på våra egna celler och utnyttjar deras maskineri för att föröka sig. Därför är det svårt att oskadliggöra ett virus utan att samtidigt skada våra egna celler. Under flera år ansågs det helt omöjligt att lyckas med detta och ytterst lite forskning publicerades inom området. Det stora genombrottet kom 1977 då man framställde den kemiska substansen Acyclovir och fann att den selektivt oskadliggjorde herpesvirus utan att skada värdcellernas funktioner. Detta bevisade att det faktiskt var möjligt att på ett säkert sätt bekämpa virus med kemiska substanser. I mitten av 80-talet fann man att HIV (human immunodeficiency virus) orsakar AIDS (aquired immunodeficiency syndrome) och då tog forskningen kring virusbekämpning fart på allvar. På 90-talet fanns endast fem licensierade läkemedel mot virussjukdomar på marknaden. Idag är det närmare 50 varav de flesta är bromsmediciner mot HIV/AIDS. Virusbekämpningen ligger dock fortfarande långt efter kampen mot bakterier samtidigt som behovet hela tiden ökar. Fler virusrelaterade sjukdomar upptäcks och redan exister-ande läkemedel måste kompletteras eller ersättas på grund av biverkningar eller resistenta virus. Nya substanser och nya bekämpningsmetoder måste således tas fram kontinuerligt.

Den svåra ögonsjukdomen EKC (epidemisk keratokonjunctivit) orsakas av virus tillhörande släktet adenovirus och är vanlig i tätbefolkade områden, framförallt i Asien. Som ett exempel kan nämnas att enbart i Japan drabbas ca en halv miljon individer av denna sjukdom årligen. Infektionen gör pa-tienten handikappad under flera veckor vilket innebär mycket lidande och stora ekonomiska förluster. Viruspartikeln tar sig in i värdcellen genom att först binda till cellytan med hjälp av fiberprotein som sticker ut från viruset. Nyligen har man funnit att benägenheten hos EKC-orsakande adenovirus att infektera värdceller är helt beroende av att de kan känna igen och binda till cellreceptorer innehållande kolhydraten sialinsyra. Det huvudsakliga målet i det här arbetet har varit att ta fram föreningar som binder till virusets fiber-protein och därmed inhiberar (blockerar) denna inbindning och förhindrar viruset från att infektera cellerna. Varje virus har tolv fiberproteiner som kan

65

fästa till cellreceptorer och det är därför troligt att en inhibitor som binder till flera fiberproteiner samtidigt är mer effektiv än en som bara binder till ett. Detta kan åstadkommas med så kallade multivalenta strukturer, d.v.s. struk-turer där flera inhibitorer har kopplats samman. För att efterlikna den natur-liga cellreceptorn användes sialinsyra (1) som utgångpunkt för inhibitorerna. De sialinsyrabaserade föreningarna 2 och 3 syntetiserades och användes för att framställa de multivalenta konjugaten 5 och 6 genom att koppla upp dem på proteinet humant serum albumin (HSA). De multivalenta HSA-konjugaten utvärderades biologiskt med avseende på förmåga att förhindra adenovirus från att binda till och infektera hornhinneceller. Båda konjugaten var effektiva som inhibitorer och en tydlig multivalenseffekt erhölls, d.v.s. ett ökat antal inhibitorer på HSA-molekylerna gav en kraftigt ökad inhiber-ingseffekt. I båda fallen erhölls en tusenfaldig ökning av inhiberingen jäm-fört med de enskilda kolhydraterna. I ett försök att utveckla ännu effektivare inhibitorer studerades en kristallstruktur av sialinsyra bunden till fiberpro-teinet. Tio nya, icke naturliga, sialinsyraderivat (4) designades, syntetise-rades och kopplades upp på HSA enligt den tidigare strategin. Ingen av dessa strukturer var bättre som inhibitor än sialinsyra men nya kristallstrukturer visar att ett par av dem binder in på rätt ställe i fiberproteinet och att de med sina nya funktionaliteter hittar nya intressanta interaktionsmöjligheter. Då kemi med sialinsyra är mycket komplicerad och ofta leder till låga utbyten och många biprodukter var det önskvärt att försöka hitta ett nytt strukturele-ment som sialinsyra kunde bytas ut mot, ett så kallat mimetikum. Förening 7 syntetiserades och jämfördes med sialinsyra. Det visade sig att denna fören-ing aktivt inhiberar inbindning av EKC-orsakande adenovirus till hornhinne-celler och till och med något bättre än sialinsyra. Dessa studier med icke-naturliga sialinsyror och ett sialylmimetikum visar tydligt att det finns poten-tial att hitta nya strukturer för inhibering och ger vägledning mot nästa gen-erations inhibitorer. I nuläget finns inga licensierade läkemedel för behandling av EKC. Eftersom strukturer som inhiberar EKC-orsakande ade-novirus skulle kunna appliceras direkt i ögat, t.ex. i form av en salva, och därmed inte behöver ta sig igenom kroppens nedbrytningsprocesser skulle de i förlängningen kunna användas i en behandlingsmetod mot EKC.

O

COOH

O

OH

HOOH

HO

NH

R1NH

OO

NH

n, n=3-19

HSAO

COOH

OH

HOOH

HO

NH R1R2

O

COOH

OH

NH

1 R1 = OH, R2 = CH3, sialinsyra

2 R1 = O(CH2)5NH2, R2 = CH3

3 R1 = O-laktos-O(CH2)2NH2, R2 = CH3

4 R1 = OCH3, R2 =Et, Pr, iPr, Bu, Ph, OMe, NHMe

R2

O O

5 R1 = (CH2)5, R2 = CH3

6 R1 = laktos-O(CH2)2, R2 = CH3

7

Figur 28: Föreningar som syntetiserats och utvärderats som adenovirusinhibitorer i detta arbete.

66

8. Acknowledgements

Det är med blandade känslor jag tar mig an detta kapitel och sätter punkt för fem års arbete. Jag har vetat om den här dagen ända sedan starten men ändå känns det som om den smugit sig på mig. Jag har läst andra doktoranders Acknowledgements, glatts med dem och peppat dem inför den stundande disputationen, fullt medveten (fast ändå inte) om att det en dag skulle bli min tur. Nu är det min tur och det är både skönt och skrämmande på samma gång. Skönt att bli klar, javisst det är ju det här jag har jobbat så hårt för men också vemodigt att lämna det bakom sig. Vad händer nu? På dom föregående sidorna har jag summerat mitt arbete. Många stunder har det varit det roligaste jag gjort och andra stunder har det varit tyngre. Jag skulle aldrig ha tagit mig igenom denna utmaning om det inte hade varit för alla er kollegor, vän-ner och familj som under dessa år stöttat mig i både med- och motgångar, på jobbet såväl som i livet. Ni är ovärdeliga. Mina varmaste tack går till,

Mina två huvudhandledare, Jan och Mikael. Det har varit otroligt roligt och lärorikt att få jobba med er. Tack Jan för att du gav mig den här chansen och för att du med stor trygghet har guidat mig igenom stort som smått. Tack Mikael för att du med ett fantastiskt engagemang hoppade på tåget i halvtid och för att du alltid lyckats vända motgångarna till något positivt. Tack till er båda för stöd och uppmuntran när allt inte riktigt gått som planerat både inom projekten och privat.

Min högstadielärare, Rolf “Fosse” Fosselius. Jag hade aldrig gett mig in på den naturvetenskapliga banan om det inte hade varit för dig. Du gjorde kemi- och fysik-timmarna inspirerande och annorlunda och fick mig att tro på att jag kunde klara av den utmaning naturvetenskapligt program på gymnasiet och fortsatta högskolestudi-er innebar. Du hade rätt!

Mina samarbetspartners, Niklas, min biträdande handledare, tack för att du med stor entusiasm introducerat mig i adenovirus och virologi i allmänhet och för att du alltid har tagit dig tid att svara på alla mina frågor. Emma, tack för all hjälp med de biologiska utvärderingarna och alla “omtag” med bilderna nu på slutet. Göran, tack för att du bidragit med stor kunskap och trevliga anekdoter inom virologi. Weixing, thank you for help with the synthesis of the multivalent N-acyl modified sialic acids. Stephen Cusack, Delphine Gulligay and Thibaut Crepin, thank you for your fast and impressive work with the crystal structures of the N-acyl modified sialic acids. Det har varit ett sant nöje att samarbeta med er!

67

Mina kollegor på avdelningen för Organisk kemi (numera Kemiska Institutionen), ni har gjort det kul att gå till jobbet varje dag. Tack för ni tillsammans gjort vår av-delning till en fantastisk arbetsplats, med många intressanta diskussioner både på labbet och i fikarummet, med mycket skratt, och med många roliga fester. Ett extra tack vill jag ge till alla seniorer på avdelningen för en välorganiserad forskningsmil-jö och för den enorma kunskapsbas ni så generöst delat med er av. Tack för alla frågor ni tagit er tid att svara på och alla goda råd ni givit.

Frippe, för ditt intresse, många roliga diskussioner och din fantastiska förmåga att inspirera.

Anna och David, för all hjälp och goda råd rörande dockningar och beräkningskemi i allmänhet. Tack för noggrann genomläsning och framförallt, tack för att ni alltid tar er tid. Anna, du är världens bästa på att få en att känna sig stolt över sitt arbete.

Mattias, för att du funnits till hands med din expertis när det krisat nere vid 400:an. Tack också för ett trevligt samarbete med sialinsyra-Ad37 studien och för att du så tålmodigt introducerat mig i låssignaler, vattenundertryckning, relaxationstider, spin-lock filter mm.

David, Tomas och Andreas för att ni så generöst delat med er av era enorma kun-skaper i organisk kemi. Det har varit en ovärdelig tillgång. Ett extra tack till dig Andreas som agerat kemi-mentor åt mig ända sedan A-kursen.

Sara, Stina och Veronica, tack för noggrann genomläsning och många bra kom-mentarer. Ni är fulla med klokheter och helt fantastiska vänner. Jag är så glad att jag fått dela min doktorand- och fritid med er. Sara det kommer att bli fruktansvärt tomt utan ditt dagliga sällskap och tack för att du är en ”top notch” barnvakt åt Anna. Tack Veronica för en underbar energiladdar helg precis när det som bäst behövdes.

Alla som jag har haft det stora nöjet att dela kontor och/eller lab med, Anna L, Anna K, David, Fredrik, Ida, Lotta, Maciej, Mattias, Mogge, Sara, Tengel, Weixing, och alla projektarbetare och ex-jobbare som passerat (ingen nämnd ingen glömd). Alla har ni bidragit till en trivsam och omväxlande arbetsmiljö.

Carina, Ann-Helen, Pelle, Kenneth, alla på Kemiförrådet och all övrig TA-personal som gör att allt runt omkring fungerar. Utan er hjälp skulle många saker varit väldigt mycket jobbigare.

Sara, Ida, Lotta och Mirja för att ni fick mig att känna mig varmt välkommen tillbaka efter föräldraledigheten.

Malin och Ante-Larsson, vilken partyelit ni är! Tack för många roliga fester ihop, det är alltid 100 % kul med er.

68

Mina vänner, ni erbjuder en oas långt bort från organisk kemi. Tack för många roli-ga stunder vid sidan av jobbet, det är sånt som bygger upp den nödvändiga energin för att orka prestera. Bea, Carro, Karin, Maria R och Maria O, järngänget från TNK som fortfarande hänger kvar. Det är alltid roligt att umgås med er, tack för att ni finns!

Collryd/Olofssons, TWAN, det är lika trevligt och avslappnat att hänga med er på söndagar i kolbäcksparken såväl som på goda middagar och storhelger. Roliga dis-kussioner över en kopp kaffe i ert kök suddar bort all stress direkt.

David & Stina, Kicki & Björn, Pelle & Hanna, Maria & Anders och Markus & Sara. Tack för otaliga middagsbjudningar, utflykter, fjällresor, fikakvällar, fester och annat skoj som ni bjudit in till under de här åren.

Lisa, Marre, Kicki och Maria Thorin, stockholmsbrudarna som aldrig låter mig glömma mitt ursprung. Det är alltid lika kul att komma “hem” till er.

Mina nära och kära, tjocka släkten är kanske inte rätt ordval då vi faktiskt är en väldigt liten släktskara. Däremot är vi desto tajtare. Frida, Helena och Sissi, mina “systrar”, för även om jag är syskonlös så är det precis det ni är. Tack för att ni ger mig perspektiv på saker och ting. Ingun och Arne tack för att dörren alltid står öp-pen i Umeå och i Nyböle även om jag utnyttjat det allt för sällan den sista tiden.

Familjen Jonsson, Margareta, Sven-Göran och Johan, det är alltid trevligt och avkopplande att komma ner till Kristinehamn och träffa er. Ni är min extra familj.

Mamma och Totte, vad jag än har gett mig in på så har ni stöttat och trott på mig. Det är tryggt att veta att ni finns där vad som än händer.

Slutligen ett STORT TACK till min familj, Pär och Anna. Ni är det finaste jag har och ni får mig att komma ihåg vad som är viktigt här i livet. Pär du är min allra bästa vän, tack för att du funnits vid min sida och peppat mig under slutspurten. Anna, att få vara mamma till dig är min största glädje i livet. Ni har fått ta den största smällen den här våren. Nu är det snart över, nu är jag snart hemma hos er igen.

Jag har säkert glömt någon eller några som förtjänat att tackas, i så fall vill jag att ni ska veta att det bara är just här på pappret. I verkligheten kommer jag ihåg och upp-skattar er.

Sussi

69

9. Appendices

Appendix I Statistical experimental design was used to investigate the possibility to im-prove the α-selectivity of the sialylation of Fmoc-aminoalcohols with 21c (Scheme 6). Four factors; temperature, MeCN/CH2Cl2 ratio, equivalents IBr, and length of the acceptor, were varied in two levels (high/low) as listed below. A 24-1 fractional factorial design107 was generated using the MODDE141 software resulting in eight experiments plus three center points that were to be conducted (Table 4).

Variables: low high Temperature; -20 °C -78 °C Eq. IBr: 1 eq 2 eq % MeCN in CH2Cl2; 50 % 80 % Acceptor; Fmoc-aminoethanol (A) Fmoc-aminopentanol (B)

General experimental settings: 21c (Scheme 6, 50 mg, 0.06 mmol), Fmoc-aminoalcohol (4 eq) and pow-dered molecular sieves (3 Å, 40 mg) were stirred in a mixture of CH3CN and CH2Cl2 (3 mL) for 1 h under nitrogen. AgOTf (2 eq relative IBr) was added and the reaction mixture was cooled. IBr in CH2Cl2 (1 M) was added drop wise during 5 min and the mixture was then stirred for 3 h, DIPEA (8 eq) was added and stirring was continued for 0.5 h. The mixture was allowed to attain room temperature and was then filtered and concentrated at reduced pressure. The residue was partly purified by automated parallel SiO2 column chromatography142 (toluene:Acetone; 3:1-3:2). The total α/β-yield and the α/β ratio was used as responses (Table 4). The α/β ratio was determined by COSY NMR experiments of the α/β mixture. The ratio was calculated by integrating the cross peaks of the H3eq/H3ax J-coupling for the α- and β-residue, respectively.

70

Table 4: The 24-1 fractional factorial design and the afforded responses for sialyl-ation of Fmoc-aminoalcohols with 21c.

Entry Run order IBr (eq)

MeCN (%)

Acceptor Temp (°C)

Yield (%)

α/β ratio

1 11 1 50 A -78 81 8:1 2 7 2 50 A -20 95 4:1 3 8 1 80 A -20 76 3:1 4 3 2 80 A -78 64 6:1 5 12 1 50 B -20 64 9:1 6 9 2 50 B -78 90 7:1 7 6 1 80 B -78 58 18:1 8 5 2 80 B -20 56 12:1 9 2 1.5 65 B -49 58 20:1 10 10 1,5 65 B -49 58 15:1 11 11 1,5 65 B -49 70 26:1

Results: No valid model could be calculated for the data due to large variation in the center points. The design needed to be completed with additional experi-ments but at that point there were no time for more synthesis. However, since the α/β ratio are as high as 20:1 or more for some entrys in the table above, it was decided to investigate some of those further. The settings of the center points (i.e. entry 9-11) was used in two independent experiments to synthesize 23 (Scheme 6) in the same scale as before (200 mg of 21c). These conditions resulted in a much lower yield and a large amount of unreacted starting material. In addition it turned out that the value of the α/β ratio var-ied with the method used to determine it. If measured by COSY experiments as in the design above, the ratio for these reactions was 15:1 and 18:1 respec-tively, but if calculated from 1D 1H-NMR the ratio was only 7:1 and 8:1. This is due to baseline variations in the 1D spectra. Since the α/β ratio of 23 originally was determined to 7:1 from 1D 1H-NMR it is likely that the origi-nal ratio is in the range of the best entrys in the experimental design above. Therefore it was concluded that the original settings were the best to use and that the isolated α-yield should instead be increased by improved procedures for purification.

71

Appendix II Expression and purificaion of Ad37 fiber knob protein were performed by published procedures143 in collaboration with Emma Nilsson, Division of Virology, Department of Microbiology, Umeå University, Umeå, Sweden. The prepared stock solution of the protein contained large amounts of impu-rities (mainly glycerol) that obscured important spectral peaks. Purification by buffer exchange was performed by centrifugal filtration. Glycerol was still present in the samples during recording of spectra but the glycerol 1H-NMR signals did not overlap with the signals used for measurements. In attempts to repeat the experiment, the fiber knob was purified more exten-sively to remove all glycerol residues. However these purifications inacti-vated the protein and it could not be used to study binding. Due to difficul-ties with preparing fiber knobs with sufficient quality for NMR studies, these experiments were only possible to perform once within the timelimits of this thesis. Experimental procedure for relaxation-edited 1H-NMR experiments Each sample contained fiber knob protein of Ad37 (50 µM), ligand (N-acyl modified sialic acids 47, sialylmimic 53a, NeuGc or sialic acid, 50 µM), and galactose (included as a nonbinding control, 50 µM) in phosphate buffer with 10 % D2O. The samples were prepared from a solution of fiber knob protein in phospate buffer (16 mg/ml) and stock solutions of the ligands and galactose (1 mM in phosphate buffer). Additional phosphate buffer was added to obtain a total volume of 550 µl. A 200 ms cmpg spin-lock filter was used to efficiently remove the signals from the fiber knob protein and bound ligands.139 Suppression of the water signal was accomplished with a WA-TERGATE sequence.144 Spectra were recorded at 25 °C on a Bruker DRX 600 MHz spectrometer. Reduction in signal intensity was related to sialic acid measured on the H3ax proton for sialic acids and on the aromatic signals for 53a.

Experimental procedure for competition study of 53a and sialic acid binding to the fiber knob protein of Ad37 Each sample contained fiber knob protein of Ad37 (50 µM), sialic acid (50 µM), and galactose (included as a nonbinding control, 50 µM) in phosphate buffer with 10 % D2O. To three individual samples were added 25, 50 and 100 µM of 53a respectively. The samples were prepared and specra were recorded as above.

72

Appendix III Ad37 fiber knob protein purification, crystallisation and structure de-termination Performed by Drs: Delphine Gulligay, Thibaut Crepin and Stephen Cusack, EMBL Grenoble, France.

Protein expression and purification. Ad37 fiber knob (residues 177-365 of the full length fiber) was cloned into the pPROEX HTb plasmid. The protein was expressed in BL21 Star (DE3) bacterial strain in LB medium at 37 °C for 4 h after induction with 1 mM IPTG. The pellet was resuspended in buffer (150 mM NaCl, 30 mM Tris pH 7.5) and incubated with lysozyme (1 mg/ml) on ice for 30 min and sonicated 12 times for 10 s each time. After centrifugation at 45,000 g for 30 min at 4 °C, the cleared lysate was loaded on 10 mL of Ni-sepharose resin in column in presence of 10 mM imidazole, washed first with 50 mL of 150 mM NaCl, 30 mM Tris pH 7.5, 10 mM imidazole, then with 40 mL of 150 mM NaCl, 30 mM Tris pH 7.5, 20 mM imidazole and then eluted in several 5 mL frac-tions with buffer (150 mM NaCl, 30 mM Tris pH 7.5, 0.5 M imidazole). The His tag was cleaved off by an overnight incubation with 1/100 His-tagged TEV protease at 10 °C. Imidazole was eliminated by dialysis against buffer 150 mM NaCl, 30 mM Tris pH 7.5, 20 mM imidazole. Cleaved protein was then incubated with 5 mL of Ni-sepharose for 1h at 4 °C, and recovered in the flow through and washing fractions (5-6 x 5 mL). These fractions were pooled, concentrated 7-8 mg/mL, purified over a Superdex 200 column and the peak collected concentrated to 15 mg/mL for crystallization.

Crystallization and structure determination. Co-crystals of Ad37 fiber knobs (at 13 mg/mL) with each ligand were grown by the hanging drop method105,140 using a reservoir solution of 25-27 % polyethylene glycol 8000, 50 mM zinc acetate, and 100 mM HEPES (pH 7.15, 7.3 or 7.5). Crystals have been fast frozen in nitrogen prior to data col-lection without cryoprotection. Data were collected on beamline ID14-2, and ID14-4 at the European Synchroton Radiation Facility (Grenoble, France). As a control, crystals with 20 mM sialyllactose were reproduced.105 The best diffraction was obtained with crystals grown in condition 25 % PEG 8000, 50 mM Zinc acetate, 100 mM HEPES pH 7.5. Data were collected on ID14.4 beamline at 1.8 Å resolution.

73

For 47a, 47e, and 47g the best diffracting crystals were grown in presence of 20 mM of ligand in the same crystallization condition as for sialyllactose. Data were collected on ID14-eh2 beamline at 1.8 Å resolution. Structure was solved by molecular replacement using the Ad37–sialic acid structure. Data collection for 47e at 1.8 Å resolution and structure resolution were per-formed in the same way but the ligand was not visible in the structure in the known site for sialic acid. For 53a, crystals were obtained in presence of 5-10 mM ligand instead of 20 mM. The best diffracting crystals were grown in presence of 10 mM ligand in crystallization solution 27 % PEG 8000, 50 mM Zinc acetate, 100 mM HEPES pH 7.5. Data collection at 1.8 Å resolu-tion and structure resolution were performed in the same way as before but the ligand was not visible in the structure in the known site for sialic acid.

Table 5: Data collection and refinement statistics for structures of Ad37 fiber pro-tein in complex with the N-acyl modified sialic acids 47a and 47g.

47a 47g Space-group P21 P21 Unit-cell dimensions: a (Å) 61.27 61.14 b (Å) 69.99 69.93 c (Å) 75.00 74.45 β (°) 94.42 94.35 Range of resolution (Å) 30-1.8 30-1.8 R-merge 4.5 (46.7) 5.3 (60.8) Completeness 99.8 (99.8) 99.3 (98.2) Refinement: No. of unique reflections 58537 57656 R-factor 0.183 0.183 Rfree 0.226 0.223

74

10. References

(1) Jones, P. S., Strategies for antiviral drug discovery. Antiviral Chemistry & Chemotherapy, 1998, 9, 283-302.

(2) Uhnoo, I., Antivirala läkemedels möjligheter och begränsnin-gar. Tidskriften Medikament, 2002, 2, 24-29.

(3) De Clercq, E., Three decades of antiviral drugs. Nature Rev. Drug Disc., 2007, 6, 941-941.

(4) De Clercq, E., Strategies in the design of antiviral drugs. Na-ture Rev. Drug Disc., 2002, 1, 13-25.

(5) Dwek, R. A., Glycobiology: Toward understanding the func-tion of sugars. Chem. Rev., 1996, 96, 683-720.

(6) Varki, A., Biological roles of oligosaccharides - all of the theories are correct. Glycobiology, 1993, 3, 97-130.

(7) Schauer, R., Achievements and challenges of sialic acid re-search. Glycoconj. J., 2000, 17, 485-499.

(8) Schauer, R., Sialic acids: Fascinating sugars in higher animals and man. Zoology, 2004, 107, 49-64.

(9) Schauer R, Sialic acids and their role as biological masks. Trends Biochem. Sci., 1985, 10, 357-360.

(10) Schauer, R.; Kammerling, J. P., Chemistry, biochemistry and biology of sialic acids. Vol. 29b, Amsterdam, Elsevier Sci-ence, 1997.

(11) Varki, A., Sialic acids as ligands in recognition phenomena. FASEB J., 1997, 11, 248-255.

(12) Kelm, S.; Schauer R., Sialic acids in molecular and cellular interactions. Vol. 175, San Diego, USA, Academic Press, 1997.

(13) Paulson, J. C., The receptors. Vol. 2, New York, Academic press, 1985.

(14) Mahdavi, J.; Sonden, B.; Hurtig, M.; Olfat, F. O.; Forsberg, L. et al, Helicobacter pylori sabA adhesin in persistent infection and chronic inflammation. Science, 2002, 297, 573-578.

(15) Hirmo, S.; Kelm, S.; Iwersen, M.; Hotta, K.; Goso, Y. et al, Inhibition of Helicobacter pylori sialic acid-specific haemag-glutination by human gastrointestinal mucins and milk glyco-proteins. FEMS Immun. Med. Microbiol., 1998, 20, 275-281.

(16) Schmidt, R. R.; Castro-Palomino, J. C.; Retz, O., New aspects of glycoside bond formation. Pure Appl. Chem., 1999, 71, 729-744.

75

(17) Sinay, P., Recent advances in glycosylation reactions. Pure Appl. Chem., 1991, 63, 519-528.

(18) Boons, G. J., Strategies in oligosaccharide synthesis. Tetrahe-dron, 1996, 52, 1095-1121.

(19) Kanie, O.; Hindsgaul, O., Synthesis of oligosaccharides, gly-colipids and glycopeptides. Curr. Opin. Struct. Biol., 1992, 2, 674-681.

(20) Kiefel, M. J.; von Itzstein, M., Recent advances in the synthe-sis of sialic acid derivatives and sialylmimetics as biological probes. Chem. Rev., 2002, 102, 471-490.

(21) Boons, G. J.; Demchenko, A. V., Recent advances in O-sialylation. Chem. Rev., 2000, 100, 4539-4566.

(22) Schmidt, R. R., Recent developments in the synthesis of gly-coconjugates. Pure Appl. Chem., 1989, 61, 1257-1270.

(23) Kanie, O.; Kiso, M.; Hasegawa, A., Methylthioglycosides of N-acetylneuraminic acid and dimethyl(methylthio)sulfonium triflate. J. Carbohydr. Chem., 1988, 7, 501-506.

(24) Martichonok, V.; Whitesides, G. M., Stereoselective alpha-sialylation with sialyl xanthate and phenylsulfenyl triflate as a promotor. J. Carbohydr. Chem., 1996, 61, 1702-1706.

(25) Kondo, H.; Ichikawa, Y.; Wong, C. H., Beta-sialyl phosphite and phosphoramidite - synthesis and application to the chemoenzymatic synthesis of CMP-sialic acid and sialyl oli-gosaccharides. J. Am. Chem. Soc., 1992, 114, 8748-8750.

(26) Meijer, A.; Ellervik, U., Study of interhalogens/silver trifluoromethanesulfonate as promoter systems for high-yielding sialylations. J. Org. Chem., 2002, 67, 7407-7412.

(27) Ikeda, K.; Aizawa, M.; Sato, K.; Sato, M., Novel glycosyla-tion reactions using glycosyl thioimidates of N-acetyl-neuraminic acid as sialyl donors. Bioorg. Med. Chem. Lett., 2006, 16, 2618-2620.

(28) De Meo, C.; Parker, O., Thiomidoyl approach to the synthesis of alpha-sialosides. Tetrahedron-Asym., 2005, 16, 303-307.

(29) Ress, D. K.; Linhardt, R. J., Sialic acid donors: Chemical synthesis and glycosylation. Curr. Org. Synth., 2004, 1, 31-46.

(30) De Meo, C.; Demchenko, A. V.; Boons, G. J., A stereoselec-tive approach for the synthesis of alpha-sialosides. J. Org. Chem., 2001, 66, 5490-5497.

(31) Mammen, M.; Choi, S. K.; Whitesides, G. M., Polyvalent interactions in biological systems: Implications for design and use of multivalent ligands and inhibitors. Angew. Chem. Int. Edit., 1998, 37, 2755-2794.

(32) Kiessling, L. L.; Pohl, N. L., Strength in numbers: Non-natural polyvalent carbohydrate derivatives. Chem. Biol., 1996, 3, 71-77.

76

(33) Lee, Y. C.; Lee, R. T., Carbohydrate-protein interactions - basis of glycobiology. Acc. Chem. Res., 1995, 28, 321-327.

(34) Borman, S., Multivalency: Strength in numbers. Chem. Eng. News, 2000, 78, 48-53.

(35) Ashwell G., M. A., The role of surface carbohydrates in the hepatic recognition and transport of circulating glycoproteins. Advances in enzymology and related subjects, 1974, 41, 99-128.

(36) Lee, Y. C.; Lee, R. T.; Hardy, M. R.; Lonngren, J.; Arnarp J.; et al, Binding of synthetic oligosaccharides to the hepatic Gal/GalNac lectin. Dependence on fine structural features., J. Biol. Chem., 1983, 258, 199-202.

(37) Matrosovich, M. N.; Mochalova, L. V.; Marinina, V. P.; Byramova, N. E.; Bovin, N. V., Synthetic polymeric sialoside inhibitors of influenza-virus receptor-binding activity. FEBS Lett., 1990, 272, 209-212.

(38) Gamian, A.; Chomik, M.; Laferriere, C. A.; Roy, R., Inhibi-tion of influenza-A virus hemagglutinin and induction of in-terferon by synthetic sialylated glycoconjugates. Can. J. Microbiol., 1991, 37, 233-237.

(39) Spaltenstein, A.; Whitesides, G. M., Polyacrylamides bearing pendant alpha-sialoside groups strongly inhibit agglutination of erythrocytes by influenza-virus. J. Am. Chem. Soc., 1991, 113, 686-687.

(40) Fan, E.; Zhang, Z.; Minke, W. E.; Hou, Z.; Verlinde, C.et al, High-affinity pentavalent ligands of Escherichia coli heat-labile enterotoxin by modular structure-based design. J. Am. Chem. Soc., 2000, 122, 2663.

(41) Kitov, P. I.; Sadowska, J. M.; Mulvey, G.; Armstrong, G. D.; Ling, H. et al, Shiga-like toxins are neutralized by tailored multivalent carbohydrate ligands. Nature, 2000, 403, 669-672.

(42) Kiessling, L. L.; Gestwicki, J. E.; Strong, L. E., Synthetic multivalent ligands in the exploration of cell-surface interac-tions. Curr. Opin. Chem. Biol., 2000, 4, 696-703.

(43) Dam, T. K.; Roy, R.; Das, S. K.; Oscarson, S.; Brewer, C. F., Binding of multivalent carbohydrates to concanavalin a and dioclea grandiflora lectin - thermodynamic analysis of the "multivalency effect". J. Biol. Chem., 2000, 275, 14223-14230.

(44) Lundquist, J. J.; Toone, E. J., The cluster glycoside effect. Chem. Rev., 2002, 102, 555-578.

(45) Reuter, J. D.; Myc, A.; Hayes, M. M.; Gan, Z. H.; Roy, R. et al, Inhibition of viral adhesion and infection by sialic-acid-conjugated dendritic polymers. Bioconj. Chem., 1999, 10, 271-278.

77

(46) Wong, S. Y. C., Neoglycoconjugates and their applications in glycobiology. Curr. Opin. Struct. Biol., 1995, 5, 599-604.

(47) Lee, Y. C.; Lee, L. R., Neoglycoconjugates, preparation and applications, San Diego, California, Academic Press Inc, 1994.

(48) Fan, E.; Merrit, E. A., Combating infectious diseases through multivalent design. Curr. Drug Targets: Infectious Disorders, 2002, 2, 161-167.

(49) Choi, S. K.; Mammen, M.; Whitesides, G. M., Generation and in situ evaluation of libraries of poly(acrylic acid) presenting sialosides as side chains as polyvalent inhibitors of influenza-mediated hemagglutination. J. Am. Chem. Soc., 1997, 119, 4103-4111.

(50) Manning, D. D.; Hu, X.; Beck, P.; Kiessling, L. L., Synthesis of sulfated neoglycopolymers: Selective P-selectin inhibitors. J. Am. Chem. Soc., 1997, 119, 3161-3162.

(51) Kanai, M.; Mortell, K. H.; Kiessling, L. L., Varying the size of multivalent ligands: The dependence of concanavalin A binding on neoglycopolymer length. J. Am. Chem. Soc., 1997, 119, 9931-9932.

(52) Mortell, K. H.; Gingras, M.; Kiessling, L. L., Synthesis of cell agglutination inhibitors by aqueous ring-opening metathesis polymerization. J. Am. Chem. Soc., 1994, 116, 12053-12054.

(53) Vrasidas, I.; de Mol, N. J.; Liskamp, R. M. J.; Pieters, R. J., Synthesis of lactose dendrimers and multivalency effects in binding to the cholera toxin B subunit. Eur. J. Org. Chem., 2001, 4685-4692.

(54) Boas, U.; Heegaard, P. M. H., Dendrimers in drug research. Chem. Soc. Rev., 2004, 33, 43-63.

(55) Thompson, J. P.; Schengrund, C. L., Inhibition of the adher-ence of cholera toxin and the heat-labile enterotoxin of Es-cherichia coli to cell-surface gm1 by oligosaccharide-derivatized dendrimers. Biochem. Pharmacol., 1998, 56, 591-597.

(56) Spevak, W.; Nagy, J. O.; Charych, D. H.; Schaefer, M. E.; Gilbert, J. H.et al, Polymerized liposomes containing C-glycosides of sialic-acid - potent inhibitors of influenza-virus invitro infectivity. J. Am. Chem. Soc., 1993, 115, 1146-1147.

(57) Bruehl, R. E.; Dasgupta, F.; Katsumoto, T. R.; Tan, J. H.; Bertozzi, C. R. et al, Polymerized liposome assemblies: Bi-functional macromolecular selectin inhibitors mimicking physiological selectin ligands. Biochemistry, 2001, 40, 5964-5974.

(58) Benkö, M.; Harrach, B.; Russel, W. C., Family adenoviridae. New York, Academic Press, 2000, 227-238.

78

(59) Shenk, E. T., Adenoviridae: The viruses and their replication. 4th edition, volume 2, Philadelphia, Lippincott Williams & Williams Publishers, 2001, 2265-2300.

(60) Horwitz, M. S., Adenoviruses. 4th edition, volume 2, Phila-delphia, Lippincott Williams & Williams Publishers, 2001, 2301-2326.

(61) Wadell, G.; Hammarskjöld, M-L.; Winberg, G.; Varsanyl, M.; Sundell, G., Genetic variability of adenoviruses. Ann. N.Y. Acad. Sci, 1980, 354, 16-42.

(62) Ford, E.; Nelson, K. E.; Warren, D., Epidemiology of epi-demic keratoconjunctivitis. Epidemiol. Rev., 1987, 9, 244-261.

(63) Gordon, Y. J.; Aoki, K.; Kinchington, Adenovirus keratocon-junctivitis. Pepose, J. S., Holland, G. N., and Wilhelmus, K. R. (ed), Ocular infection and immunity, C.V. Mosby, St. Louis, Mo, 1996.

(64) Henry, L. J.; Xia, D.; Wilke, M. E.; Deisenhofer, J.; Gerhard, R. D., Characterisation of the knob domain of the adenovi-rustype 5 fiber protein expressed in escherichia coli. J. Virol., 1994, 68, 5239-5246.

(65) Arnberg, N.; Edlund, K.; Kidd, A. H.; Wadell, G., Adenovirus type 37 uses sialic acid as a cellular receptor. J. Virol., 2000, 74, 42-48.

(66) Arnberg, N.; Kidd, A. H.; Edlund, K.; Olfat, F.; Wadell, G., Initial interactions of subgenus D adenoviruses with A549 cellular receptors: Sialic acid versus αv integrins. J. Virol., 2000, 74, 7691-7693.

(67) Arnberg, N.; Mei, Y. F.; Wadell, G., Fiber genes of adenovi-ruses with tropism for the eye and the genital tract. Virology, 1997, 227, 239-244.

(68) Aoki, K.; Kawana, R.; Matsumoto, I.; Wadell, G.; Dejong, J. C., Viral conjunctivitis with special reference to adenovirus type-37 and enterovirus-70 infection. Jpn. J. Ophthalmol., 1986, 30, 158-164.

(69) Paparello, S. F.; Rickman, L. S.; Mesbahi, H. N.; Ward, J. B.; Siojo, L. G.et al, Epidemic keratoconjunctivitis at a united-states military base - republic of the philippines. Mil. Med., 1991, 156, 256-259.

(70) Yamadera, S.; Yamashita, K.; Akatsuka, M.; Kato, N.; Inouye, S.et al, Adenovirus surveillance, 1982-1993, japan. A report of the national epidemiological surveillance of agents in japan. Jpn. J. Med. Sci. Biol., 1995, 48, 199-210.

(71) Kemp, M. C.; Hierholzer, J. C.; Cabradilla, C. P.; Obijeski, J. F., The changing etiology of epidemic keratokonjunctivitis: Antigenic and restriction enzyme analyses of adenovirus types 19 and 37 isolated over a 10-year period. J. Infect. Dis., 1983, 148, 24-33.

79

(72) Aoki, K.; Yoshitugu, T., A twenty-one year surveillance of adenoviral conjunctivitis in sapporo, japan. Int. Ophthamol. clin., 2002, 42, 49-54.

(73) Russell, W. C., Update an adenovirus and its vectors. J. Gen-eral Virol., 2000, 81, 2573-2604.

(74) Dall'Olio, F.; Chiricolo, M., Sialyltransferases in cancer. Gly-coconj. J., 2001, 18, 841-850.

(75) unpublished results in collaboration with Dr. Susanne Tene-berg, Goteborg university, Goteborg, Sweden

(76) Marra, A.; Sinay, P., Stereoselective synthesis of 2-thioglycosides of n-acetylneuraminic acid. Carbohydr. Res., 1989, 187, 35-42.

(77) Bhattacharyya, P.; Slawin, A. M. Z.; Woollins, J. D., Hetero-cycle formation using [PhP(Se)(mu-Se)](2). The crystal struc-tures of [PhP(Se)(mu-Se)](2), PhP(Se)Se-2(C7H10) and PhP(Se)(mu-Se)(mu-NPh)P(Se)Ph. J. Chem. Soc.-Dalton Trans., 2001, 300-303.

(78) Bengtsson, M.; Broddefalk, J.; Dahmen, J.; Henriksson, K.; Kihlberg, J. et al, Convergent synthesis of neoglycopeptides by coupling of 2-bromoethyl glycosides to cysteine and ho-mocysteine residues in T cell stimulating peptides. Glycoconj. J., 1998, 15, 223-231.

(79) Roy, R.; Pon, R. A., Efficient synthesis of alpha(2-8)-linked N-acetyl and N-glycolylneuraminic acid disaccharides from colominic acid. Glycoconj. J., 1990, 7, 3-12.

(80) Castro-Palomino, J. C.; Tsvetkov, Y. E.; Schmidt, R. R., 8-O-sialylation of neuraminic acid. J. Am. Chem. Soc., 1998, 120, 5434-5440.

(81) Hasegawa, A.; Ishida, H.; Isogai, Y.; Ishida, H.; Kiso, M., Synthetic studies on sialoglycoconjugates. 50. Total synthesis of ganglioside GD2. J. Carbohydr. Chem., 1993, 12, 1217-1222.

(82) Castro-Palomino, J. C.; Simon, B.; Speer, O.; Leist, M.; Schmidt, R. R., Synthesis of ganglioside GD3 and its com-parison with bovine GD3 with regard to oligodendrocyte apoptosis mitochondrial damage. Chemistry-A European Journal, 2001, 7, 2178-2184.

(83) Bergh, A., A hydrophobic homobifunctional amine-reactive reagent for neoglycoconjugate synthesis. Deparment of Bioorganic Chemistry, Center for Chemistry and Chemical Engineering, Lund Institute of Technology, Lund University, 2001.

(84) Elofsson, M.; Kihlberg, J., Synthesis of Tn and sialyl Tn building-blocks for solid-phase glycopeptide synthesis. Tetra-hedron Lett., 1995, 36, 7499-7502.

80

(85) Marra, A.; Sinay, P., A novel stereoselective synthesis of N-acetyl-alpha-neuraminosyl-galactose disaccharide derivatives, using anomeric S-glycosyl xanthates. Carbohydr. Res., 1990, 195, 303-308.

(86) Dasgupta, F.; Garegg, P. J., Use of sulfenyl halides in carbo-hydrate reactions.1. Alkyl sulfenyl triflate as activator in the thioglycoside-mediated formation of beta-glycosidic linkages during oligosaccharide synthesis. Carbohydr. Res., 1988, 177, C13-C17.

(87) Lonn, H.; Stenvall, K., Exceptionally high-yield in glycosyla-tion with sialic-acid - synthesis of a GM3 glycoside. Tetrahe-dron Lett., 1992, 33, 115-116.

(88) Silvestri, M. G.; Wong, C. H., Opening of thiiranes: Prepara-tion of orthogonal protected 2-thioglyceraldehyde. J. Org. Chem., 2001, 66, 910-914.

(89) Higashi, K.; Suzaki, H., A novel glycosidation promoted by the cobination of trimethylsilyl halide and zinc triflate. Chem. Pharm. Bull., 1992, 40, 2119-2022.

(90) Windholz, T.; Johnston, D., Trichloroethoxycarbonyl: A generally applicable protecting group. Tetrahedron Lett., 1967, 27, 2555-2557.

(91) Bartra, M.; Romea, P.; Urpí, F.; Vilarrasa, J., A fast procedure for the reduction of azides and nitro compounds based on the reducing ability of Sn(SR)3-species. Tetrahedron, 1990, 46, 587-594.

(92) Hansen, M. M.; Harkness, A. R.; Coffey, D. S.; Bordwell, F. G.; Zhao, Y. Y., Substrate acidities and conversion times for reactions of amides with di-tert-butyl dicarbonate. Tetrahe-dron Lett., 1995, 36, 8949-8952.

(93) Sugimoto, M.; Numata, M.; Koike, K.; Nakahara, Y.; Ogawa, T., Synthetic studies on cell-surface glycans. 50. Total synthe-sis of gangliosides GM1 and GM2. Carbohydr. Res., 1986, 156, C1-C5.

(94) CastroPalomino, J. C.; Ritter, G.; Fortunato, S. R.; Reinhardt, S.; Old, L. J. et al, Efficient synthesis of ganglioside GM2 for use in cancer vaccines. Angew. Chem. Int. Ed., 1997, 36, 1998-2001.

(95) Sorme, P.; Qian, Y. N.; Nyholm, P. G.; Leffler, H.; Nilsson, U. J., Low micromolar inhibitors of galectin-3 based on 3'-derivatization of N-acetyllactosamine. Chembiochem, 2002, 3, 183-189.

(96) Kamath, V. P.; Diedrich, P.; Hindsgaul, O., Use of diethyl squarate for the coupling of oligosaccharide amines to carrier proteins and characterization of the resulting neoglycoproteins by maldi-tof mass spectrometry. Glycoconj. J., 1996, 13, 315-319.

81

(97) Tietze, L. F.; Schroter, C.; Gabius, S.; Brinck, U.; Goerlach-graw, A. et al, Conjugation of para-aminophenyl glycosides with squaric acid diester to a carrier protein and the use of neoglycoprotein in the histochemical detection of lectins. Bio-conj. Chem., 1991, 2, 148-153.

(98) Tietze, L. F.; Arlt, M.; Beller, M.; Glusenkamp, K. H.; Jahde, E. et al, Anticancer agents.15. Squaric acid diethyl ester - a new coupling reagent for the formation of drug biopolymer conjugates - synthesis of squaric acid ester amides and dia-mides. Chem. Ber., 1991, 124, 1215-1221.

(99) Bergh, A.; Magnusson, B. G.; Ohlsson, J.; Wellmar, U.; Nils-son, U. J., Didecyl squarate - a practical amino-reactive cross-linking reagent for neoglycoconjugate synthesis. Glycoconj. J., 2001, 18, 615-621.

(100) Kovác, P. Studies towards a rationally designed conjugate vaccine for cholera using synthetic carbohydrate antigens, Chapter 10, The Royal Society of Chemistry, Cambridge, UK, 2006.

(101) Harvey, D. J.; Kuster, B.; Naven, T. J. P., Perspectives in the glycosciences - matrix-assisted laser desorption/ionization (MALDI) mass spectrometry of carbohydrates. Glycoconj. J., 1998, 15, 333-338.

(102) Mei, Y. F.; Lindman, K.; Wadell, G., Two closely related adenovirus genome types with kidney or respiratory tropism differ in their binding to epithelial cells of various origins. Vi-rology, 1998, 240, 254-266.

(103) Segerman, A.; Arnberg, N.; Eriksson, A.; Lindman, K.; Wadell, G., There are two different species B adenovirus re-ceptors: sBAR, common to species BB and B2 adenoviruses, and SB2AR, excusively used by species B2 adenoviruses. J. Virol., 2003, 77, 1157-1162.

(104) Wadell, G.; Allard, A.; Hierholzer, J. C., Mannual of clinical microbiology. 7th edition, Washington Publisher, ASM Press, 1999, 970-982.

(105) Burmeister, W. P.; Guilligay, D.; Cusack, S.; Wadell, G.; Arnberg, N., Crystal structure of species D adenovirus fiber knobs and their sialic acid binding sites. J. Virol., 2004, 78, 7727-7736.

(106) Dekany, G.; Wright, K.; Ward, P.; Toth, I., Synthesis of sialyl Lewis X analogues. J. Carbohydr. Chem., 1996, 15, 383-398.

(107) Lundstedt, T.; Seifert, E.; Abramo, L.; Thelin, B.; Nystrom, A. et al, Experimental design and optimization. Chemom. In-tell. Lab. Syst., 1998, 42, 3-40.

(108) Daicel-chiralcel OD column (250 x 20 mm, 10um, Chiral Technologies, Illkirch, France

82

(109) Lapolla, A.; Fedele, D.; Reitano, R.; Arico, N. C.; Seraglia, R.et al, Enzymatic digestion and mass spectrometry in the study of advanced glycation end products/peptides. J. Am. Soc. Mass Spectrom., 2004, 15, 496-509.

(110) MOE-Molecular operating environment, version 2004.03; Chemical Computing Group Inc., Montreal, C., 2004.

(111) SYBYL, version 7.2; Tripos Inc.: St Louis, MO, 2004. (112) Sears, P.; Wong, C. H., Carbohydrate mimetics: A new strat-

egy for tackling the problem of carbohydrate-mediated bio-logical recognition. Angew. Chem. Int. Ed., 1999, 38, 2301-2324.

(113) Gohlke, H.; Klebe, G., Approaches to the description and prediction of the binding affinity of small-molecule ligands to macromolecular receptors. Angew. Chem. Int. Ed., 2002, 41, 2645-2676.

(114) Kitchen, D. B.; Decornez, H.; Furr, J. R.; Bajorath, J., Dock-ing and scoring in virtual screening for drug discovery: Meth-ods and applications. Nature Rev. Drug Disc., 2004, 3, 935-949.

(115) Taylor, R. D.; Jewsbury, P. J.; Essex, J. W., A review of pro-tein-small molecule docking methods. J. Comput.-Aided. Mol. Des., 2002, 16, 151-166.

(116) Guntert, P., Structure calculation of biological macromole-cules from NMR data. Quart. Rev. Biophys., 1998, 31, 145-237.

(117) Bajorath, J.; Stenkamp, R.; Aruffo, A., Knowledge-based model-building of proteins - concepts and examples. Protein Science, 1993, 2, 1798-1810.

(118) Hillisch, A.; Pineda, L. F.; Hilgenfeld, R., Utility of homol-ogy models in the drug discovery process. Drug Disc. Today, 2004, 9, 659-669.

(119) GOLD- Genetic Optimisation for Ligand Docking, version 3.1.1; The Cambridge Crystallographic Datacenter, Cam-bridge, U. K., 2004.

(120) Verdonk, M. L.; Cole, J. C.; Hartshorn, M. J.; Murray, C. W.; Taylor, R. D., Improved protein-ligand docking using GOLD. Proteins: Struct. Funct. Genetics, 2003, 52, 609-623.

(121) Pan, Y. B.; Ayani, T.; Nadas, J.; Wen, S. M.; Guo, Z. W., Accessibility of N-acyl-D-mannosamines to N-acetyl-D-neuraminic acid aldolase. Carbohydr. Res., 2004, 339, 2091-2100.

(122) Gervay, J.; Flaherty, T. M.; Nguyen, C., Solution phase syn-thesis of (1->5)-amide linked sugar amino acid dimers derived from sialic acids. Tetrahedron Lett., 1997, 38, 1493-1496.

83

(123) Sherman, A. A.; Yudina, O. N.; Shashkov, A. S.; Menshov, V. M.; Nifantiev, N. E., Preparative route to N-glycolylneuraminic acid phenyl 2-thioglycoside donor and synthesis of Neu5Gc-alpha-(2 -> 3 ')-lactosamine 3-aminopropyl glycoside. Carbohydr. Res., 2002, 337, 451-457.

(124) Mogemark, M.; Gustafsson, L.; Bengtsson, C.; Elofsson, M.; Kihlberg, J., A fluorinated selenide linker for solid-phase syn-thesis of n-pentenyl glycosides. Org. Lett., 2004, 6, 4885-4888.

(125) Saito, K.; Sugai, K.; Fujikura, K.; Yamada, N.; Goto, M.et al, Thermal-degradation of sodium N-acetylneuraminate. Carbo-hydr. Res., 1989, 185, 307-314.

(126) Sugiyama, N.; Saito, K. I.; Fujikura, K.; Sugai, K.; Yamada, N. et al, Thermal and photochemical degradation of sodium N-acetylneuraminate. Carbohydr. Res., 1991, 212, 25-36.

(127) Byramova, N. E.; Tuzikov, A. B.; Bovin, N. V., A simple procedure for the synthesis of the methyl and benzyl gly-cosides of Neu5ac and 4-epi-Neu5ac - conversion of the ben-zyl and methyl glycosides of Neu5Ac into N-trifluoroacetylneuraminic acid benzyl glycosides. Carbohydr. Res., 1992, 237, 161-175.

(128) Simanek, E. E.; McGarvey, G. J.; Jablonowski, J. A.; Wong, C. H., Selectin-carbohydrate interactions: From natural ligands to designed mimics. Chem. Rev., 1998, 98, 833-862.

(129) Kuribayashi, T.; Ohkawa, N.; Satoh, S., Aryl C-glycosides: Physiologically stable glycomimetics of sialyl Lewis X. Bioorg. Med. Chem. Lett., 1998, 8, 3307-3310.

(130) Vonitzstein, M.; Wu, W. Y.; Kok, G. B.; Pegg, M. S.; Dya-son, J. C. et al, Rational design of potent sialidase-based in-hibitors of influenza-virus replication. Nature, 1993, 363, 418-423.

(131) Singh, S.; Jedrzejas, M. J.; Air, G. M.; Luo, M.; Laver, W. G. et al, Structure-based inhibitors of influenza-virus sialidase - a benzoic-acid lead with novel interaction. J. Med. Chem., 1995, 38, 3217-3225.

(132) Lew, W.; Wu, H. W.; Chen, X. W.; Graves, B. J.; Escarpe, P. A. et al, Carbocyclic influenza neuraminidase inhibitors pos-sessing a C-3-cyclic amine side chain: Synthesis and inhibi-tory activity. Bioorg. Med. Chem. Lett., 2000, 10, 1257-1260.

(133) Bianco, A.; Brufani, M.; Manna, F.; Melchioni, C., Synthesis of a carbocyclic sialic acid analogue for the inhibition of in-fluenza virus neuraminidase. Carbohydr. Res., 2001, 332, 23-31.

84

(134) Babu, Y. S.; Chand, P.; Bantia, S.; Kotian, P.; Dehghani, A. et al, Discovery of a novel, highly potent, orally active, and se-lective influenza neuraminidase inhibitor through structure-based drug design. J. Med. Chem., 2000, 43, 3482-3486.

(135) Jedrzejas, M. J.; Singh, S.; Brouillette, W. J.; Laver, W. G.; Air, G. M. et al, Structures of aromatic inhibitors of influ-enza-virus neuraminidase. Biochemistry, 1995, 34, 3144-3151.

(136) Fagerstam, L. G.; Frostellkarlsson, A.; Karlsson, R.; Persson, B.; Ronnberg, I., Biospecific interaction analysis using sur-face-plasmon resonance detection applied to kinetic, binding-site and concentration analysis. J. Chromatography, 1992, 597, 397-410.

(137) Jonsson, U.; Fagerstam, L.; Ivarsson, B.; Johnsson, B.; Karls-son, R. et al, Real-time biospecific interaction analysis using surface-plasmon resonance and a sensor chip technology. Bio-techniques, 1991, 11, 620-&.

(138) Hajduk, P. J.; Olejniczak, E. T.; Fesik, S. W., One-dimensional relaxation- and diffusion-edited NMR methods for screening compounds that bind to macromolecules. J. Am. Chem. Soc., 1997, 119, 12257-12261.

(139) Åberg, V.; Hedenström, M.; Pinkner, J. S.; Hultgren, S. J.; Almqvist, F., C-terminal properties are important for ring-fused 2-pyridones that interfere with the chaperone function in uropathogenic E-coli. Org. Biomol. Chem., 2005, 3, 3886-3892.

(140) Rhodes, G. Crystallography made crystal clear. 2nd ed., San Diego, California, USA, Academic Press, 2000.

(141) MODDE 6.0, Umetrics, Box 7960, S-90719 Umeå, Sweden, available from www.umetrics.com.

(142) FlashPackTM FlashMaster, Jones Chromatography, Hengoed, UK

(143) Arnberg, N.; Kidd, A. H.; Edlund, K.; Nilsson, J.; Pring-Akerblom, P. et al, Adenovirus type 37 binds to cell surface sialic acid through a charge-dependent interaction. Virology, 2002, 302, 33-43.

(144) Piotto, M.; Saudek, V.; Sklenar, V., Gradient-tailored excita-tion for single-quantum NMR-spectroscopy of aqueous-solutions. J. Biomol. NMR, 1992, 2, 661-665.