The design, synthesis and biological evaluation of neuraminic acid-based probes of Vibrio cholerae sialidase

Article abstract of DOI:10.1016/S0957-4166(99)00552-2

A molecular modelling study using the program GRID has been used to investigate the structural requirements of a potential inhibitor binding to Vibrio cholerae sialidase. A number of favourable interactions were predicted between the sialidase and Neu2en derivatives containing hydroxyl- or halogen-substituted acyl groups on the C-5 amine. As a result of this study, a detailed analysis of the interactions of C-5-substituted Neu2en derivatives with the active site of V. cholerae sialidase was undertaken using a conformational searching routine based on molecular dynamics. Based on the results of these molecular design studies several N-acyl-Neu2en-based probes were prepared and evaluated for sialidase inhibition. As envisaged, and pleasingly, the designed compounds were found to be accommodated by the enzyme's active site architecture, and to be strong inhibitors of V. cholerae sialidase. Copyright (C) 2000 Elsevier Science Ltd.

Full text of DOI:10.1016/S0957-4166(99)00552-2

TETRAHEDRON:
ASYMMETRY
Pergamon
Tetrahedron: Asymmetry 11 (2000) 53–73
The design, synthesis and biological evaluation of neuraminic
acid-based probes of Vibrio cholerae sialidase
Jennifer C. Wilson, Robin J. Thomson, Jeffrey C. Dyason, Pas Florio, Kaylene J. Quelch,
Samia Abo and Mark von Itzstein ∗
Department of Medicinal Chemistry, Monash University (Parkville Campus), 381 Royal Parade, Parkville, Victoria 3052,
Australia
Received 9 November 1999; accepted 2 December 1999
Abstract
A molecular modelling study using the program GRID has been used to investigate the structural requirements
of a potential inhibitor binding to Vibrio cholerae sialidase. A number of favourable interactions were predicted
between the sialidase and Neu2en derivatives containing hydroxyl- or halogen-substituted acyl groups on the C-5
amine. As a result of this study, a detailed analysis of the interactions of C-5-substituted Neu2en derivatives
with the active site of V. cholerae sialidase was undertaken using a conformational searching routine based on
molecular dynamics. Based on the results of these molecular design studies several N-acyl-Neu2en-based probes
were prepared and evaluated for sialidase inhibition. As envisaged, and pleasingly, the designed compounds were
found to be accommodated by the enzyme’s active site architecture, and to be strong inhibitors of V. cholerae
sialidase. © 2000 Elsevier Science Ltd. All rights reserved.
1. Introduction
The significant biological roles played by sialic acids (e.g. N-acetylneuraminic acid, Neu5Ac 1)
in normal physiological processes, and also in disease states and microbial infection, has led to the
widespread study of the proteins that recognize this class of compounds.1–4 Knowledge of the three-
dimensional structure of a sialic acid-recognizing protein through an X-ray crystal structure can lead
to a greater understanding of protein–substrate/inhibitor interactions and aid in the design of potential
inhibitors. The X-ray crystal structures of a number of sialic acid-recognizing proteins, both metabolizing
enzymes, e.g. N-acetylneuraminate lyase5 and several sialidases,6–21 as well as adhesion proteins such
as influenza virus haemagglutinin22 and E-selectin,23 have been solved. In the case of sialidases from
strains of influenza virus, analyses of X-ray crystal structures have shown that a number of amino
∗
Corresponding author. Fax: +61-3-9903-9672; e-mail: mark.vonitzstein@vcp.monash.edu.au
0957-4166/00/$ - see front matter © 2000 Elsevier Science Ltd. All rights reserved.
PII: S0957-4166(99)00552-2
tetasy 3187

54
J. C. Wilson et al. / Tetrahedron: Asymmetry 11 (2000) 53–73
acids that line and surround the walls of the active site pocket are invariant in all strains of the virus
characterized so far.8 This finding led to the rational design,24,25 using molecular modelling techniques
such as described herein, and development of the potent and selective inhibitor of the sialidase, 5-
acetamido-2,6-anhydro-3,4,5-trideoxy-4-guanidino-D-glycero-D-galacto-non-2-enoic acid 2.25–31 The 4-
deoxy-4-guanidino-Neu5Ac2en derivative 2 has subsequently been developed as the anti-influenza drug
Zanamivir (Relenza®).29–31
Sialic acid-recognizing proteins such as the sialidases from pathogenic organisms are obvious targets
for the design of potential therapeutic agents.6,8,29 The sialidase associated with the bacterium Vibrio
cholerae is believed to have two roles in cholera disease. Firstly, V. cholerae sialidase is part of a multi-
enzyme mucinase complex,32 which may function in the pathogenesis of the bacterium by reducing
the viscosity of the gastrointestinal mucus, thereby facilitating access of the bacterium to the target
epithelial cells. Secondly, as depicted in Fig. 1, V. cholerae sialidase cleaves sialic acids from higher
order gangliosides to give GM1, the putative cholera toxin receptor.33,34 It therefore appears that the
sialidase may increase the virulence of V. cholerae by increasing the number of receptor sites available
for binding of cholera toxin.
A number of structure–activity studies have been carried out on V. cholerae sialidase.27,35–45 These
studies have included the use of modified sialic acid glycosides as potential substrates,35–37,41,45 and
2,3-unsaturated sialic acid derivatives27,37–39,43,44 based on 5-acetamido-2,6-anhydro-3,5-dideoxy-D-
glycero-D-galacto-non-2-enoic acid (N-acetyl-Neu2en, Neu5Ac2en, 3), as well as non-carbohydrate
compounds,36,42 as potential inhibitors. Neu5Ac2en 3 itself is a µM inhibitor of the enzyme.27 It is
postulated to be a transition-state analogue,14,46 based on its half-chair conformation,47 mimicking the
putative sialosyl cation intermediate14,48,49 4 in the proposed mechanism of sialidase catalysis.
In 1994, the X-ray structure of V. cholerae sialidase solved to 2.3 Å was published17 allowing a
closer examination of the enzyme active site and comparison17 with the sialidases from influenza virus8
and Salmonella typhimurium.15 As shown in Fig. 2 the sialidase from V. cholerae consists of a central
catalytic domain that is associated with two lectin domains.
Described herein is an investigation of the active site of V. cholerae sialidase using molecular modelling
techniques. In particular, we have examined the area of the active site that accommodates the C-5
acetamido group of Neu5Ac. Based on the results of this study, a number of C-5-modified derivatives of
Neu5Ac2en 3 have been synthesized and evaluated as inhibitors of the sialidase. Furthermore, we have
taken the opportunity to interpret previously reported inhibition data38,39,43,44 in light of our conclusions
drawn from the present molecular modelling study.

J. C. Wilson et al. / Tetrahedron: Asymmetry 11 (2000) 53–73
55
Fig. 1. Cleavage of a higher-order ganglioside by Vibrio cholerae sialidase to give GM1, the putative receptor for cholera toxin
2. Results and discussion
2.1. Molecular modelling
The overall topology of the V. cholerae active site cavity, in which Neu5Ac2en17 or Neu5Ac bind with
the β-face of the pyranose ring towards the floor of the cavity, can be described as a pocket with a deep
cavity directly below the C-4-epi position which extends underneath, around to the C-5 position as shown
in Fig. 3. The upper surface of the pocket is formed by Asp250 that sits directly above the C-4 position,
Asn318 which sits above the C-5 acetamido group, and hydrophobic residues Phe638 and Leu580 which
encase the C-8/C-9 position.
The V. cholerae sialidase active site contains an array of residues that are conserved among the bacterial
sialidases from S. typhimurium15 and Micromonospora viridifaciens,18 the sialidases from different
strains of influenza virus,8–14 and the intramolecular ‘trans-sialidase’ from the leech (Macrobdella
decora),19 for which X-ray crystal structures have been solved. The conserved array of residues in V.
cholerae sialidase, which have been detailed by Crennell et al. (summarised in Table 1),17 are those that
are potentially involved in the enzyme catalytic mechanism. However, there are some differences between
the sialidases in other areas of the active site, in particular in the vicinity of C-4 and of the C-6 glycerol
side chain of Neu5Ac.17–19 In each of the sialidases crystallized to date,8,15,17–19 there is a hydrophobic
pocket which accommodates the C-5 acetamido substituent of Neu5Ac. In V. cholerae sialidase, this

56
J. C. Wilson et al. / Tetrahedron: Asymmetry 11 (2000) 53–73
Fig. 2. Structure of Vibrio cholerae sialidase. The central domain is the sialidase function and is flanked by lectin domains
pocket is relatively large, with quite extensive cavities adjacent to and below the C-5 acetamido group. In
a general inspection of the C-5 binding domain we concluded that an interaction between the active site
residue Asn318 and the C-5 acylamino carbonyl is possible. This observation supports the view35,41,44,45
that this carbonyl group may be required for recognition by V. cholerae sialidase.
Fig. 4 shows some of the interactions between the active site residues of V. cholerae sialidase and
Neu5Ac2en 3. Similar active site interactions are observed for Neu5Ac 1.
2.1.1. GRID calculations
The program GRID50 was used to investigate the active site of V. cholerae sialidase. GRID is a
molecular design tool that can be used to investigate the steric and electronic interactions between
functional group probes and biomolecules. Contour maps are calculated to identify regions of favourable
interaction energy between the functional group probes and the biomolecule. In these calculations only
contributions within a 35 Å cube centred on the catalytically important residues were considered. A
range of functional group probes were investigated for this study and, as detailed below, several of these
have provided useful information. The areas of favourable interaction energy noted were examined with
α-N-acetylneuraminic acid (2-α-Neu5Ac) modelled in the active site. Contour maps showing the most

J. C. Wilson et al. / Tetrahedron: Asymmetry 11 (2000) 53–73
57
Fig. 3. A view of 2-α-Neu5Ac modelled in the active site of Vibrio cholerae sialidase
favourable interactions for carboxyl, hydroxyl, chloro, amino, and methyl probes, with 2-α-Neu5Ac
positioned in the active site, are presented in Figs. 5–9, respectively.
2.1.1.1. Carboxyl group probe. The carboxyl group probe (Fig. 5) showed a region of favourable
interaction energy (in magenta) at −7 kcal mol−1 that accurately predicted the positioning of the Neu5Ac
C-2 carboxylate located in a cluster of three arginine residues (224, 635 and 712) within the active site.
These arginine residues form strong interactions with the acid group on Neu5Ac, as determined from
binding energy calculations, and therefore have a role in holding the Neu5Ac residue in place during the
catalytic process.49 GRID analysis revealed no other positions of favourable interaction for a carboxylate
group within the active site. Other similar probes such as phosphate, sulfone, and nitro groups were also
predicted to bind well within the triarginyl cluster. However, from the point of view of inhibitor design,

58
J. C. Wilson et al. / Tetrahedron: Asymmetry 11 (2000) 53–73
Table 1
The important conserved residues in the active site of Vibrio cholerae sialidase17
Fig. 4. Important interactions between the active site residues of Vibrio cholerae sialidase and Neu5Ac2en 3
substitution of these groups on the sialic acid framework at C-2, in preference to a carboxylate group, is
not expected to be advantageous. Indeed, the phosphonic acid analogue of Neu5Ac2en,51 showed a slight
reduction in inhibition of V. cholerae sialidase compared to Neu5Ac2en.52 Crystal structures determined
with 2-deoxy-phosphonic acid analogues of Neu5Ac bound to S. typhimurium16 and influenza virus21
sialidases showed that interaction between the phosphonyl group and the triarginyl cluster was a major
contributing factor in the determination of the inhibitor binding mode.16,21
2.1.1.2. Hydroxyl group probe. The hydroxyl group probe (Fig. 6) showed at −11 kcal mol−1 regions
of favourable interaction energy (in blue) corresponding to the location of the carboxylate group on
Neu5Ac. At the same contour level ‘hot-spots’ were also seen located near the methyl of the C-5

J. C. Wilson et al. / Tetrahedron: Asymmetry 11 (2000) 53–73
59
Fig. 5. Contour map showing the most favourable interactions (in magenta) at −7 kcal mol−1 for the carboxyl group probe with
2-α-Neu5Ac modelled in the active site
acetamido group that as previously mentioned resides in quite a deep pocket of the active site. This
led us to the conclusion that the N-glycolyl [NHC(O)CH2OH] derivative of neuraminic acid should be
accommodated well in that region of the active site.
Additionally, at −7 kcal mol−1, regions of contours (not shown) were seen extending from the C-3
position on Neu5Ac to the position of the C-4 hydroxyl group. In this region of the active site are situated
Arg245 (2.8/3.3 Å) and Asp292 (2.4/3.6 Å) which would be expected to interact with the C-4 hydroxyl
group, contributing to these favourable contour energies. Similar ‘pairs’ of arginine and aspartate residues
are observed to interact with the C-4 hydroxyl group in the sialidases from S. typhimurium,15 M.
viridifaciens,18 and the intramolecular trans-sialidase from leech.19 In influenza virus sialidase, however,
there do not appear to be extensive interactions with the C-4 hydroxyl group,16 which is accommodated
in a large cavity.8
Examination of the interactions of the hydroxyl group probe in the glycerol side chain binding region

60
J. C. Wilson et al. / Tetrahedron: Asymmetry 11 (2000) 53–73
Fig. 6. Contour map showing the most favourable interactions (in blue) at −11 kcal mol−1 for the hydroxyl group probe with
2-α-Neu5Ac modelled in the active site
revealed contours (not shown), at −6 kcal mol−1, located near the C-7 hydroxyl and C-8 hydroxyl groups
of Neu5Ac. In this region of the active site, Asn318 could interact with the C-7 hydroxyl group (3.1 Å)
and C-9 hydroxyl group (2.5 Å). In addition, Asp637 form a hydrogen bond with the C-8 hydroxyl group
(2.9 Å). The C-7 hydroxyl group could also participate in a hydrogen bonding interaction with solvent
exposed Asp250 (2.5 Å).
2.1.1.3. Halogen probes. The chloro (Fig. 7) and fluoro probes showed several areas of favourable
interaction. The chloro probe had contours (in yellow) at −4 kcal mol−1 extending from the C-3 position
on Neu5Ac to, and below, the C-4 position. Favourable contours were also located in the vicinity of the
methyl of the C-5 acetamido group.

J. C. Wilson et al. / Tetrahedron: Asymmetry 11 (2000) 53–73
61
Fig. 7. Contour map showing the most favourable interactions (in yellow) at −4 kcal mol−1 for the chloro probe with
2-α-Neu5Ac modelled in the active site
Interestingly, no favourable contours were seen for the bromo or iodo probes in the C-5 pocket.
The most likely explanation for this result is that both bromine and iodine atoms are too bulky to be
accommodated in this region of the active site.
2.1.1.4. Nitrogen-based probes. In general the nitrogen-based probes (e.g. amino group probe, Fig. 8)
gave the best interaction energies with contours (in blue) being distinguishable at −18 kcal mol−1. Close
examination of these contours revealed that they were in regions below the epi-position at the C-4 position
of Neu5Ac, in the vicinity of the acidic residues Glu243, Glu619 and Asp292. At higher interaction
energies (−9 kcal mol−1, not shown) contours were located near the methyl of the C-5 acetamido group
within the active site.
It is noteworthy that no favourable contours were seen around the equatorial position at C-4 of
Neu5Ac. A similar GRID study of influenza virus sialidase24 identified favourable interactions for a
nitrogen probe in a large cavity found around the equatorial C-4 position. This led to the synthesis26 of

62
J. C. Wilson et al. / Tetrahedron: Asymmetry 11 (2000) 53–73
Fig. 8. Contour map showing the most favourable interactions (in blue) at −18 kcal mol−1 for the amino group probe with
2-α-Neu5Ac modelled in the active site
Neu5Ac2en derivatives containing amino and guanidino groups at C-4, for example 2, which interacted
with acidic active site residues located in this region,12,25 and that proved to be potent and selective
inhibitors of influenza virus sialidase.25,27,29 Analysis of the crystal structures of the bacterial sialidases
from V. cholerae17 and S. typhimurium15 shows that there is no similar cavity at C-4 and that a guanidino
group would be too sterically demanding to be able to fit in this position. This observation, which was
predicted,27 but unconfirmed until the crystal structure analysis of V. cholerae sialidase was completed,
explains why 4-deoxy-4-guanidino-Neu5Ac2en (Zanamivir, Relenza®) was found to exhibit far less
inhibition of the bacterial sialidases than of influenza virus sialidase.27
2.1.1.5. Aliphatic/aromatic based probes. Hydrophobic probes such as methyl (Fig. 9), methylene and
phenyl were investigated and in general were contoured at much higher interaction energies (e.g. −3
kcal mol−1). Clearly from Fig. 9 it can be seen that there are significant regions of favourable interaction

J. C. Wilson et al. / Tetrahedron: Asymmetry 11 (2000) 53–73
63
Fig. 9. Contour map showing the most favourable interactions (in red) at −3 kcal mol−1 for the methyl group probe with
2-α-Neu5Ac modelled in the active site
energies (in red) corresponding to the positioning of the methyl of the C-5 acetamido group as well as
around the C-3 methylene of 2-α-Neu5Ac.
In the cavity around C-5 there is a conserved tryptophan residue (also found in S. typhimurium15 and
influenza virus8 sialidases) which, in the case of influenza virus sialidase, makes important hydrophobic
contacts (4.0 Å) to the methyl of the C-5 acetamido group of Neu5Ac.11 Mutation of this residue in
influenza virus sialidase led to a properly expressed, transported and folded protein which, however,
retained no sialidase activity.53 In V. cholerae sialidase, this conserved Trp311 is situated ca. 4.1 Å from
the methyl of the C-5 acetamido group of Neu5Ac.
2.1.2. Conformational searching routines for C-5-modified Neu5Ac2en derivatives
The GRID analysis of the active site cavity of V. cholerae sialidase showed that there were possibilities
for investigating modifications to the C-5 acetamido group and substitution at the C-4-epi position
with a view to improving the binding of potential inhibitors of the sialidase. To further investigate the
influence that modifications to the C-5 acetamido group would have on binding, a series of C-5-modified
Neu5Ac2en derivatives were examined in the active site of V. cholerae sialidase. Neu5Ac2en 3 was

64
J. C. Wilson et al. / Tetrahedron: Asymmetry 11 (2000) 53–73
chosen as the starting template for the molecular modelling protocol as it is, as previously mentioned, a
µM inhibitor of V. cholerae sialidase.27
C-5-Modified Neu5Ac2en derivatives were built-up, situated in the active site of the fully minimized V.
cholerae sialidase structure and then subjected to a conformational searching protocol based on molecular
dynamics calculations. These calculations were performed to identify potential binding modes of the
modified Neu5Ac2en derivatives within the active site pocket. Choice of the modification at C-5 was
guided to some extent by the aforementioned GRID studies. Although many different functional group
possibilities were examined in the GRID study only those that showed interesting interactions with the
active site will be discussed here.
2.1.2.1. N-Glycolyl-Neu2en 5. The N-glycolyl derivative interacts with several residues in the vicinity
of the C-5 binding domain of the active site cavity. Specifically, hydrogen bonding interactions (3.4 Å)
were seen between the amide carbonyl of Gln317 and the hydroxyl group of the N-glycolyl moiety. Some
conformers from the searching routine were also seen with interactions between the N-glycolyl hydroxyl
group and Asp292 at 2.9 Å and Asn545 at 3.1 Å.
2.1.2.2. N-Chloroacetyl-Neu2en 6. Chain extension of the N-acetyl by a chlorine atom seemed to have
little overall impact on the binding of the analogue compared to the Neu5Ac2en starting template.
Initially, it was anticipated that the chlorine would form a hydrogen bond to the amine groups of the
asparagine residues that sit above and below the methyl of the C-5 acetamido group. The conformational
searching routine results do not support this however, and it seems more likely that the chlorine interacts
with the hydrophobic Trp311 situated 3.8 Å away.
2.1.2.3. Aliphatic chain extensions [N-propanoyl- 7, N-butanoyl- 8, and N-(acryloyl)- 9 Neu2en]. The
effects of aliphatic, one and two carbon, chain extensions to the N-acetyl group were also investigated by
the conformational searching protocol. The conformational search revealed that the C-5 binding domain
of the active site could accommodate an N-propanoyl or N-acryloyl side chain but chain extension to the
N-butanoyl group was not accommodated. Similarly, replacing the methyl of the N-acetyl with a phenyl
ring giving N-benzoyl-Neu2en was also not possible on steric grounds. These results indicated that there
was little scope for improving inhibition through chain extension, even though adjacent to and below the
C-5 acetamido group there is quite an extensive cavity. The C-5 binding domain of the active site is able to
accommodate differences in flexibility with the N-propanoyl and N-acryloyl groups being accommodated
equally well. The N-acryloyl derivative participates in hydrophobic interactions with Trp311 (3.3. Å) and
Pro251 (3.7 Å) in the C-5 binding domain.
2.2. Synthesis of C-5-modified Neu5Ac2en derivatives
The data from the molecular modelling studies supported the notion that N-glycolyl-Neu2en 5 and
N-chloroacetyl-Neu2en 6 should be accommodated by the enzyme’s active site, and in particular the
C-5 substituents, N-glycolyl and N-chloroacetyl, may be able to access favourable interactions with
a number of active site residues. These interactions could potentially result in improved inhibition of
the sialidase when compared with the benchmark compound Neu5Ac2en. The inhibition of V. cholerae
sialidase by both of these compounds has previously been reported by others,39,43 although in each case
different assays were used to evaluate sialidase inhibition making it difficult to compare the data. As
a result of these differences we decided to re-investigate these compounds so that we could obtain

J. C. Wilson et al. / Tetrahedron: Asymmetry 11 (2000) 53–73
65
a direct comparison of inhibition between them and Neu5Ac2en in a fluorescence assay procedure.
Furthermore, and as explained in the molecular modelling section, we concluded that the active site
architecture around the C-5 binding domain should be able to accommodate some hydrophobic electron
rich functionalities such as an acryloyl group. The preparation of the N-acryloyl derivative 9 was achieved
from the reactive intermediate N-(3-chloropropanoyl)-Neu2en. In addition and for completeness, N-
trichloroacetyl-Neu2en 10 was synthesised to observe the effects of introducing a very bulky and
electronically demanding group.
The procedure used for the preparation of the C-5-modified Neu5Ac2en derivatives is outlined in
Scheme 1.
Scheme 1. Reagents and conditions: (a) Ba(OH)2, H2O, 90°C; (b) (i) Et3N, THF, H2O, rt; (ii) R0COOH, DCC, rt; (c) Et3N,
MeOH, R0COCl, −60°C
Methyl
5-acetamido-4,7,8,9-tetra-O-acetyl-2,6-anhydro-3,5-dideoxy-D-glycero-D-galacto-non-2-
enoate54 11 was deprotected and de-N-acetylated by reaction with aqueous barium hydroxide55 at ca.
90°C for several hours to give an 85% yield of Neu2en 12. This yield is similar to the 89% yield reported
by Isecke and Brossmer for the de-N-acetylation of the fully protected 2-α-methyl ketoside of Neu5Ac
with barium hydroxide.55 N-Acylation of Neu2en,44,56,57 and neuraminic acid derivatives,55,58–62 with
a variety of agents has been previously reported. In this work the reaction was carried out by means of
either a dicyclohexylcarbodiimide (DCC)-mediated coupling with a carboxylic acid, or by reaction with
an acid chloride.
2.3. Inhibition of Vibrio cholerae sialidase
Using a well known fluorometric assay,27 inhibition constants (Ki values) were determined for the C-5-
modified Neu5Ac2en derivatives synthesised, as well as Neu5Ac2en itself, and these data are presented
in Table 2. The inhibition constants for both N-glycolyl-Neu2en43 5 and N-chloroacetyl-Neu2en39 6,
relative to Neu5Ac2en, are comparable with those found in previously reported inhibition studies as
further elaborated in the following discussion. Furthermore, we have taken the opportunity to now

66
J. C. Wilson et al. / Tetrahedron: Asymmetry 11 (2000) 53–73
Table 2
Inhibition data against Vibrio cholerae sialidase for Neu5Ac2en and C-5-modified derivatives of
Neu5Ac2en
interpret, at the molecular level, some of the biological data38,39,43,44 that has been reported in the past
for a range of C-5-modified Neu5Ac2en derivatives.
2.3.1. N-Glycolyl-Neu2en
The GRID study had indicated contours for favourable interactions of the hydroxyl group probe in the
vicinity of the methyl of the C-5 acetamido group of Neu5Ac. The molecular dynamics investigation
of the unsaturated N-glycolyl derivative 5 indicated conformers with hydrogen bonding interactions
between the glycolyl hydroxyl group and several residues in the active site. The inhibition data obtained
here, and previously (Nöhle et al.43 reported IC50 values against V. cholerae sialidase of 1.0×10−5
M
for Neu5Gc2en 5, and 1.5×10−5 M for Neu5Ac2en 3), is supportive of the proposal that the N-glycolyl
group improves binding into the active site compared to the N-acetyl group.
2.3.2. N-Haloacetyl-Neu2en derivatives
The most potent sialic acid-based inhibitor of V. cholerae sialidase reported to date is the N-
trifluoroacetyl-Neu2en derivative 13,39,44 with a 10-fold increase in inhibition over Neu5Ac2en. In
addition, both the N-monofluoroacetyl- and N-difluoroacetyl-Neu5Ac2en derivatives show improved
inhibitory activity over Neu5Ac2en.39 Inhibition data (IC50) determined for N-haloacetyl-Neu2en de-
rivatives against V. cholerae sialidase by Meindl et al.39 is presented in Table 3. The GRID study tallies
with these findings, with a fluoro probe showing favourable interactions in the vicinity of the methyl of
the C-5 acetamido group of Neu5Ac. No evidence of hydrogen-bonding interactions that could contribute
to this increased inhibition was observed in the molecular dynamics calculations and therefore the likely
increase in inhibition may result from interaction with the hydrophobic residues of the C-5 binding
domain.
Table 3
Inhibition data against Vibrio cholerae sialidase for Neu5Ac2en and N-haloacetyl-Neu2en derivatives;
N-C(O)R-Neu2en39
The GRID results with the chloro probe led us to conclude that a chloroacetyl group should be
accommodated in the C-5 binding domain and moreover participate in favourable interactions with amino
acid residues in this domain. However, in the molecular dynamics investigation no apparent improvement

J. C. Wilson et al. / Tetrahedron: Asymmetry 11 (2000) 53–73
67
of interactions of the N-chloroacetyl analogue 6, when compared with Neu5Ac2en, was found. It is
interesting to note that the molecular dynamics findings are indeed reflected in the comparable inhibition
constants (Table 2) for 6 and Neu5Ac2en 3.
The larger iodo probe showed no favourable interaction energies in the area of the C-5 pocket,
presumably due to steric restrictions. In agreement with this observation, the N-iodoacetyl-Neu2en
derivative 14 was reported39 to produce a slightly weaker inhibition when compared to Neu5Ac2en.
Predictably, therefore, based on steric bulk and electronic demands, substitution of the N-acetyl group by
an N-trichloroacetyl group leads to a dramatic (∼100-fold) loss of inhibition (Table 2).
2.3.3. Effects of altering the aliphatic chain of the N-acyl group
The GRID study and conformational searching routines indicated that extension of the N-acetyl
side chain by one carbon to give the N-propanoyl or N-acryloyl derivatives could be accommodated,
but further aliphatic extension or introduction of a phenyl group could not. The molecular dynamics
investigation also suggested that chain extensions by one carbon from the parent N-acetyl group may
actually be favourable, taking advantage of hydrophobic interactions with Trp311. Indeed the N-acryloyl
derivative 9 produced a slight improvement in inhibition (Table 2). Likewise, Meindl et al.38 reported a
similar trend in inhibition results with an increase in inhibition by the N-propanoyl derivative 7 relative
to Neu5Ac2en (Table 4), but found that increasing the length of the acyl chain beyond propanoyl led
to a reduction in the inhibition of V. cholerae sialidase. It is not clear at this stage if unsaturation in the
acyl chain improves the level of inhibition because it is difficult to compare absolute inhibition constants
across the two assays. However, it is not unreasonable to speculate that the alkenyl moiety may participate
in some π–π interactions with Trp311.
Table 4
Inhibition data against Vibrio cholerae sialidase for Neu5Ac2en and C-5-modified derivatives with
different acyl-group chain lengths; N-C(O)R-Neu2en38
Other work by Meindl and Tuppy demonstrated that the replacement of the N-acetyl group by N-
formyl resulted in an approximately 30-fold drop in inhibition of the sialidase (IC50 values: 8.0×10−4
M for N-formyl-Neu2en 15, vs 3.0×10-5 M for Neu5Ac2en 3).39 We believe that this finding can be
rationalized from the present molecular modelling study. It is not unreasonable to suggest that the loss of
the methyl group would lead to a concomitant decrease in the hydrophobic interaction with Trp311 and
therefore result in poorer binding affinity.
3. Summary
Although there is a large cavity in the active site pocket of V. cholerae sialidase around the C-5 position,
and a large number of amino acids available to participate in electrostatic interactions, generally only
marginal increases in inhibition over N-acetyl-Neu2en have been achieved by substitutions at C-5 to date.
The small improvement in inhibition seen with the N-glycolyl substituent could be accounted for as being
due to potential interaction with several active site residues. The slight increases in inhibition observed
for N-haloacetyl-, N-propanoyl- and N-acryloyl-Neu2en derivatives appear to stem from interaction with
the hydrophobic areas of the side chain pocket.

68
J. C. Wilson et al. / Tetrahedron: Asymmetry 11 (2000) 53–73
The major constraint on which groups will be tolerated in the C-5 pocket appears to be steric. The
cavity in the active site leading away from the C-5 position, allows for extension of the N-acetyl group
by one carbon, or other atom, but a longer aliphatic chain or a more sterically demanding group is not
accommodated.
In conclusion, we have investigated the active site architecture of V. cholerae sialidase using the
previously determined crystal structure17 and GRID software. From these results we have designed and
synthesized a number of neuraminic acid-based probes that appear to be well accommodated by the active
site. These results provide important information regarding the active site pocket of V. cholerae sialidase
that may facilitate the further design and synthesis of potent inhibitors of this enzyme.
4. Experimental
Professor Garry Taylor (St. Andrews University) kindly provided the crystal structure coordinates of
V. cholerae sialidase at 2.3 Å resolution.
4.1. Molecular modelling
GRID calculations: Calculations were carried out on silicon graphics R8000 and R4000 workstations
using GRID version 11.50 The results and the crystal structures were viewed using InsightII (MSI, San
Diego, CA). GRID calculations were performed on a cube of 35 Å per side, centred on the active site, with
a GRID spacing of 0.5 Å. The interaction energy between the probe and every atom within the protein
structure was evaluated at each grid point. A dielectric constant of 80 was used to simulate a bulk aqueous
phase, while areas as determined by GRID to be excluded from solvent were assigned a dielectric constant
of 4 (e.g. the interior of the protein). The utility program GRIN was used to automatically assign atom
types and charges for the protein using the standard parameter file provided with GRID. The output was
converted (using GINI, supplied with GRID) into a suitable form for input into the MSI utility Contour.
Contour maps were built-up using steps of 2 kcal mol−1. The contour map was then viewed superimposed
on the crystal structure of the active site of V. cholerae sialidase using the program InsightII. The GRID
results were conveniently displayed using contour maps of the interaction energies superimposed on the
crystal structure of the sialidase, together with a solvent accessible surface overlaid on a 15 Å radius of
the active site residues and N-acetylneuraminic acid (Neu5Ac) positioned in the binding pocket.
Binding of C-5-modified Neu5Ac2en derivatives: The method used to calculate binding modes of
the C-5-modified Neu5Ac2en derivatives has been described in detail by Taylor and von Itzstein.49,63
Firstly, a complete minimization of the V. cholerae sialidase crystal structure with Neu5Ac bound,
crystallographic waters and a 15 Å cap of solvent water over the active site was performed. The effects of
hydration were modelled by partially solvating the active site of the complex and by using a distance
dependent dielectric. The minimization was carried out within Discover (MSI) using the Consistent
Valence Force Field with a cut-off distance of 15 Å. A multistep protocol was used to ensure the
minimization of the strained interactions arising from the crystal structure refinement. First, solvent water
molecules were energy minimized with all other atoms fixed, and then all protein backbone and Neu5Ac
non-hydrogen atoms were included in the minimization process, followed by the protein side chains, and
finally the entire system was free to move. Calculations were done using the algorithms steepest descents
(in all constrained minimizations for up to 500 iterations) and conjugate gradients (down to a maximum
atomic root mean square derivative of 0.01 kcal mol−1).

J. C. Wilson et al. / Tetrahedron: Asymmetry 11 (2000) 53–73
69
Conformational space of the enzyme–ligand complex within the active site was explored using a
combination of constrained molecular dynamics and molecular mechanics calculations. Beginning with
the energy minimized sialidase/Neu5Ac complex a ligand (previously built-up using InsightII) was
docked into the active site with its position and conformation as similar to Neu5Ac as possible. Solvent
molecules were then added or subtracted from the area around the docked ligand depending on the space
requirements. The complex was constrained throughout all calculations; only the ligand and protein
residues within a 15 Å radius of the active site (centred on the C-2 position of sialic acid) and the water
molecules around the active site (crystallographic and soak water) were free to move. In addition to the
atom constraints, solvent molecules within the active site were subjected to harmonic restraints, with a
force constant of 0.05 kcal mol−1, to accurately model solvent mobility. An initial energy minimization
was performed to relieve steric strain associated with the newly docked ligand. Molecular dynamics
calculations were then performed at 350 K using the leapfrog algorithm in Discover. Dynamics were
equilibrated for 2 ps with time steps of 1 fs and then continued for 2 ps with time steps of 2 fs. The
resulting structure was extracted and energy minimized. A further 2 ps of dynamics with a time step of 2
fs were performed, with the final structure being again subjected to energy minimization. This protocol
was performed 15 times beginning at the molecular dynamics initialization stage.
4.2. Syntheses
General: 1H and 13C NMR spectra were recorded in D2O using a Bruker AM-300 spectrometer. Chem-
ical shifts are given in ppm relative to external Me4Si. ESI mass spectra were obtained using a Micromass
Platform II electrospray spectrometer, and HRMS were obtained using a Bruker BioApex II FTMS.
Optical rotations were measured using a Jasco DIP-370 digital polarimeter. Reactions were monitored by
TLC on aluminium plates coated with silica gel 60 F254 (Merck). Amines were visualized with ninhydrin.
Dowex 50WX8 (400, H+ form) and Dowex 1X8 (400, Cl− form) ion exchange resins were washed and
activated before use. Methyl 5-acetamido-4,7,8,9-tetra-O-acetyl-2,6-anhydro-3,5-dideoxy-D-glycero-D-
galacto-non-2-enoate 1154 was prepared in-house. 2-Acetoxyacetic acid was prepared by acetylation of
glycolic acid (Aldrich) with acetyl chloride and was recrystallized from toluene (mp 66–68.5°C, lit.64
66–68°C). Chloroacetic acid (Koch-Light) was recrystallized from benzene. Dicyclohexylcarbodiimide
(DCC) (Fluka), trichloroacetyl chloride (Aldrich) and 3-chloropropionyl chloride (Fluka) were distilled
before use.
4.2.1. 5-Amino-2,6-anhydro-3,5-dideoxy-D-glycero-D-galacto-non-2-enoic acid Neu2en 12
A suspension of methyl 5-acetamido-4,7,8,9-tetra-O-acetyl-2,6-anhydro-3,5-dideoxy-D-glycero-D-
galacto-non-2-enoate 11 (657 mg, 1.39 mmol) in saturated aqueous barium hydroxide55 (30 ml) was
heated at ca. 90°C for 18 h. TLC (15:4:0.5 iPrOH:H2O:AcOH); Neu5Ac2en 3 Rf 0.41, Neu2en 12
(ninhydrin +ve) Rf 0.25. The reaction was cooled, filtered through a Celite plug, and the brown residue
was washed thoroughly with water. The filtrate was neutralized by the addition of small pieces of solid
carbon dioxide, with vigorous stirring, and then the fine precipitate of BaCO3 was filtered-off and washed
thoroughly with water. The precipitate was resuspended in water and stirred for a further 10 min before
refiltration. This procedure was repeated three times. The combined filtrates were treated with portions of
Dowex 50WX8 (H+) resin (ca. 10 ml total) until the supernatant tested negative for the presence of barium
ions.55 The resin was filtered-off into a small column and washed thoroughly with water. Evaporation of
the washings and lyophilization of the residue gave a red–brown gum. TLC analysis showed numerous
components including 12 and Neu5Ac2en 3. The majority of the product 3 was eluted from the resin
with 1 to 2 M aqueous ammonia solution. Evaporation and lyophilization of the appropriate fractions

70
J. C. Wilson et al. / Tetrahedron: Asymmetry 11 (2000) 53–73
gave Neu2en 12 as an off-white hygroscopic solid. Further purification of the mixture obtained by elution
with water was carried out by passage through a column of Dowex 50WX8 (H+) (7 ml). Elution with
1.5 M aqueous ammonia solution gave 12 (overall yield 294 mg, 85%); [α]D −11 (c 0.53, H2O) [lit.56
[α]D −12.5 (c 0.50, H2O); lit.44 [α]D −10 (c 0.26, H2O)]; 1H NMR δ 3.39 (1H, app.t, H-5), 3.65 (1H,
dd, J9,8 5.7, J9,9 12.0 Hz, H-9), 3.71 (1H, d, J7,8 9.2 Hz, H-7), 3.83 (1H, d, J9 ,9 12.0 Hz, H-90), 3.94
(1H, m, H-8), 4.35 (1H, d, J6,5 10.1 Hz, H-6), 4.48 (1H, br.d, H-4), 5.65 (1H, s, H-3) [this spectrum is in
agreement with that reported by Schreiner et al.44]; m/z (FAB) 272 [(M+Na)+, 41%], 250 [(M+H)+, 40].
0
0
4.2.2. 2,6-Anhydro-3,5-dideoxy-5-(2-hydroxyacetamido)-D-glycero-D-galacto-non-2-enoic acid
[N-glycolyl-Neu2en] 5
DCC (62 mg, 0.3 mmol) was suspended in THF (3 ml) and a solution of 2-acetoxyacetic acid (32 mg,
0.3 mmol) in THF (2 ml) was added with vigorous stirring. The resulting mixture was stirred for 5 min at
rt before addition to a solution of Neu2en 12 (50 mg, 0.2 mmol) in water (2 ml), triethylamine (83µl, 0.6
mmol) and THF (3 ml). Further aliquots of triethylamine were added to the reaction to maintain a pH of
ca. 8 (a total of ca. 10 equiv. was often added). The reaction was monitored by TLC, and after 1 h further
portions of Et3N, DCC and 2-acetoxyacetic acid were added under the same conditions as initially used.
The reaction was worked-up by addition of ca. 10 ml of water, filtration of the mixture, and evaporation
of the filtrate under reduced pressure. The crude N-(2-acetoxyacetyl)-Neu2en was dissolved in water
(3 ml), the solution treated with 0.1 M NaOH solution to pH ∼9, and the reaction left at rt for 2 h.
The reaction was neutralized with 1 M AcOH, concentrated and the crude N-(2-hydroxyacetyl)-Neu2en
subject to ion-exchange chromatography. An aqueous solution of the crude product was passed through
a column of Dowex 50WX8 (H+) resin (ca. 10 ml). The N-acylated product was eluted with water while
any unreacted 12 could be recovered by elution with 1.5–2 M aqueous ammonia solution. The fractions
containing the N-acylated product were applied to a column of Dowex 1X8 (acetate) resin (ca. 12 ml)
and elution with a gradient of acetic acid was carried out (traces of the reagent carboxylic acid remaining
after this procedure could be removed by repeating the purification on the anionic resin). Elution with
4 M AcOH, followed by evaporation and repeated lyophilization gave 5 as a white solid (60 mg, 95%);
[α]D +21 (c 0.21, H2O); 1H NMR δ 3.61 (1H, d, J7,8 8.8 Hz, H-7), 3.64 (1H, m, H-9), 3.84 (1H, br.d,
J9 ,9 12.0 Hz, H-90), 3.89 (1H, m, H-8), 4.15 (3H, m, H-5, glycolyl CH2), 4.35 (1H, d, J6,5 10.8 Hz, H-6),
0
4.57 (1H, d, J4,5 8.9 Hz, H-4), 5.99 (1H, s, H-3); 13C NMR δ 52.0 (C-5), 63.7 (C-9*), 65.7 (glycolyl
CH2*), 69.7 (C-7†), 70.6 (C-4†), 72.6 (C-8†), 78.5 (C-6†), 115.1 (C-3), 146.5 (C-2), 168.3 (C-1), 178.2
(glycolyl CONH) [*,† tentative assignments]; m/z 368 [(M·Na+K)+, 8%], 352 [(M·Na+Na)+, 100], 330
[(M+Na)+, 50]; HRMS: calcd for C11H17NO9·H 308.09816, found 308.09740.
4.2.3. 2,6-Anhydro-3,5-dideoxy-5-(2-chloroacetamido)-D-glycero-D-galacto-non-2-enoic acid
[N-chloroacetyl-Neu2en] 6
Prepared by reaction of Neu2en 12 with DCC/chloroacetic acid as described above for the preparation
of N-(2-acetoxyacetyl)-Neu2en in the synthesis of 5. Elution from Dowex 1X8 (acetate) resin with 1 to 2
M AcOH, followed by evaporation and repeated lyophilization gave 6 as a white solid (90%); [α]D +14.6
(c 0.26, H2O) [lit.56 [α]D +33.2 (c 0.50, H2O); 1H NMR δ 3.73 (2H, m, H-7, H-9), 3.96 (1H, dd, J9 ,8
0
2.7, J9 ,9 11.3 Hz, H-90), 4.01 (1H, ddd, J8,7 9.0, J8,9 6.3, J8,9 2.7 Hz, H-8), 4.26 (1H, br.app.t, H-5), 4.29
(2H, s, ClCH2CO), 4.48 (1H, d, J6,5 10.8 Hz, H-6), 4.67 (1H, dd, J4,3 2.4, J4,5 8.9 Hz, H-4), 6.14 (1H,
d, J3,4 2.4 Hz, H-3); m/z 386 [(M·Na+K)+, 40%], 370 [(M·Na+Na)+, 100], 348 [(M+Na)+, 37]; HRMS:
calcd for C11H16NO8Cl·NH4 343.09082, found 343.08979.
0
0

J. C. Wilson et al. / Tetrahedron: Asymmetry 11 (2000) 53–73
71
4.2.4. 2,6-Anhydro-3,5-dideoxy-5-acrylamido-D-glycero-D-galacto-non-2-enoic acid [N-(acryloyl)-
Neu2en] 9
To a solution of Neu2en 12 (50 mg, 0.2 mmol) in MeOH (2 ml) and Et3N (0.5 ml) at −60°C under
N2 was added 3-chloroproprionyl chloride (58 µl, 0.6 mmol). The reaction was allowed to warm to room
temperature when it was concentrated, the residue dissolved in water, and the solution subjected to ion-
exchange chromatography as described above for 5. Elution from Dowex 1X8 (acetate) resin with 2 M
AcOH, followed by evaporation and repeated lyophilization gave 9 as a white solid (62%); 1H NMR δ
3.61 (2H, m, H-7, H-9), 3.84 (1H, d, J9 ,9 13.2 Hz, H-90), 3.89 (1H, m, H-8), 4.15 (1H, br.app.t, H-5),
0
4.31 (1H, d, J6,5 10.7 Hz, H-6), 4.54 (1H, d, J4,5 8.5 Hz, H-4), 5.78 (1H, d, J 8.9 Hz, olefinic-H trans-
CONH), 5.98 (1H, s, H-3), 6.21 (1H, d, J 17.4 Hz, olefinic-H cis-CONH), 6.29 (1H, dd, J 8.8, J 17.4
Hz, olefinic-H gem-CONH); m/z 364 [(M·Na+K)+, 38%], 348 [(M·Na+Na)+, 100], 326 [(M+Na)+, 45];
HRMS: calcd for C12H17NO8·H 304.10324, found 304.10283.
4.2.5. 2,6-Anhydro-3,5-dideoxy-5-(2,2,2-trichloroacetamido)-D-glycero-D-galacto-non-2-enoic acid
[N-trichloroacetyl-Neu2en] 10
Prepared by reaction of Neu2en 12 with trichloroacetyl chloride as described above for the synthesis
of 9. Elution from Dowex 1X8 (acetate) resin with 6 M AcOH, followed by evaporation and repeated
lyophilization gave 10 as a white solid (21%); 1H NMR δ 3.58 (2H, m, H-7, H-9), 3.83 (1H, br.d, J9 ,9
0
12.2 Hz, H-90), 3.90 (1H, m, H-8), 4.15 (1H, br.app.t, H-5), 4.41 (1H, d, J6,5 11.0 Hz, H-6), 4.57 (1H, d,
J4,5 8.2 Hz, H-4), 5.88 (1H, s, H-3); m/z 454 [(M·Na+K)+, 27%], 438 [(M·Na+Na)+, 67], 416 [(M+Na)+,
33]; HRMS: calcd for C11H14NO8Cl3·NH4 411.01287, found 411.01271.
4.3. Sialidase assay
Vibrio cholerae sialidase was obtained by over-expression of the V. cholerae nanH gene (kindly
provided by Dr. E. Vimr, Illinois) in E. coli and subsequent purification of the enzyme as described
elsewhere.65,66 Sialidase activity was assayed using the previously reported fluorometric assay.27 The
substrate used for the enzyme assay, 2-α-(40-methylumbelliferyl)-N-acetylneuraminic acid, was prepared
using published methods.67 The concentration of 2-α-(40-methylumbelliferyl)-N-acetylneuraminic acid
and inhibitor was varied between 50–250 µM and 5–100 µM, respectively, for the Ki determinations.
Assays were performed in at least duplicate.
It should be noted that there is an almost 10-fold difference in the Ki value reported by different
research groups for Neu5Ac2en against V. cholerae sialidase; for example, Schreiner et al.44 2.5×10−5
M, Holzer et al.27 3×10−6 M. This difference is presumably carried over into the Ki values measured for
other sialic acid derivatives.27,37 It is possible that slightly different methods of data analysis has led to
this anomoly. Schreiner et al.44 calculated constants by fitting data with a non-linear regression program
to a model of partially mixed inhibition, with Ki values representing the competitive part of the model in
order to facilitate comparison with older data. Holzer et al.27 used a model of competitive inhibition. The
precise effect these differences may have on inhibition data has not yet been determined. The original
inhibition data for Neu5Ac2en and derivatives reported by Meindl et al. was presented either as Ki38 or
IC5039 values. Comparison of inhibition of V. cholerae sialidase by Neu5Ac2en derivatives has therefore
been based on trends observed within sets of data.

72
J. C. Wilson et al. / Tetrahedron: Asymmetry 11 (2000) 53–73
Acknowledgements
Dr. Donald Angus is thanked for assistance with the preparation of this manuscript. We gratefully
acknowledge the financial support of the Australian Research Council and GlaxoWellcome, and the
Australian Government for an Australian Postgraduate Award to PF.
References
1. Schauer, R.; Kamerling, J. P. In Glycoproteins II; Montreuil, J.; Vliegenthart, J. F. G.; Schachter, H., Eds. Chemistry,
biochemistry and biology of sialic acids. Elsevier Science: Amsterdam, 1997; pp. 243–402.
2. Biology of the Sialic Acids; Rosenberg, A., Ed.; Plenum Press: New York, 1995.
3. Reuter, G.; Gabius, H.-J. Biol. Chem. Hoppe-Seyler 1996, 377, 325–342.
4. Varki, A. FASEB J. 1997, 248–255.
5. Izard, T.; Lawrence, M. C.; Malby, R. L.; Lilley, G. G.; Colman, P. M. Structure 1994, 2, 361–369.
6. Taylor, G. Curr. Opin. Struct. Biol. 1996, 6, 830–837.
7. Saito, M.; Yu, R. K. In Biology of the Sialic Acids; Rosenberg, A., Ed. Biochemistry and function of sialidases. Plenum
Press: New York, 1995; pp. 261–313 and references cited therein.
8. Colman, P. M. Protein Sci. 1994, 3, 1687–1696 and references cited therein.
9. Varghese, J. N.; Laver, W. G.; Colman, P. M. Nature 1983, 303, 35–40.
10. Varghese, J. N.; Colman, P. M. J. Mol. Biol. 1991, 221, 473–486.
11. Varghese, J. N.; McKimm-Breschkin, J. L.; Caldwell, J. B.; Kortt, A. A.; Colman, P. M. Proteins: Struct., Funct., Genet.
1992, 14, 327–332.
12. Varghese, J. N.; Epa, V. C.; Colman, P. M. Protein Sci. 1995, 4, 1081–1087.
13. Burmeister, W. P.; Ruigrok, R. W. H.; Cusack, S. EMBO J. 1992, 11, 49–56.
14. Burmeister, W. P.; Henrissat, B.; Bosso, C.; Cusack. S.; Ruigrok, R. W. H. Structure 1993, 1, 19–26.
15. Crennell, S. J.; Garman, E. F.; Laver, W. G.; Vimr, E. R.; Taylor, G. L. Proc. Natl. Acad. Sci. USA 1993, 90, 9852–9856.
16. Crennell, S. J.; Garman, E. F.; Philippon, C.; Vasella, A.; Laver, W. G.; Vimr, E. R.; Taylor, G. L. J. Mol. Biol. 1996, 259,
264–280.
17. Crennell, S.; Garman, E.; Laver, G.; Vimr, E.; Taylor, G. Structure 1994, 2, 535–544.
18. Gaskell, A.; Crennell, S.; Taylor, G. Structure 1995, 3, 1197–1205.
19. Luo, Y.; Li, S.-C.; Chou, M.-Y.; Li, Y.-T.; Luo, M. Structure 1998, 6, 521–530.
20. Luo, Y.; Li, S.-C.; Li, Y.-T.; Luo, M. J. Mol. Biol. 1999, 285, 323–332.
21. White, C. L.; Janakiraman, M. N.; Laver, W. G.; Philippon, C.; Vasella, A.; Air, G. M.; Luo, M. J. Mol. Biol. 1995, 245,
623–634.
22. Watowich, S. J.; Skehel, J. J.; Wiley, D. C. Structure 1994, 2, 719–731 and references cited therein.
23. Graves, B. J.; Crowther, R. L.; Chandran, C.; Rumberger, J. M.; Li, S.; Huang, K.-S.; Presky, D. H.; Familletti, P. C.;
Wolitzky, B. A.; Burns, D. K. Nature 1994, 367, 532–538.
24. von Itzstein, M.; Dyason, J. C.; Oliver, S. W.; White, H. F.; Wu, W.-Y.; Kok, G. B.; Pegg, M. S. J. Med. Chem. 1996, 39,
388–391.
25. von Itzstein, M.; Wu, W.-Y.; Kok, G. B.; Pegg, M. S.; Dyason, J. C.; Jin, B.; Phan, T. V.; Smythe, M. L.; White, H. F.;
Oliver, S. W.; Colman, P. M.; Varghese, J. N.; Ryan, D. M.; Woods, J. M.; Bethell, R. C.; Hotham, V. J.; Cameron, J. M.;
Penn, C. R. Nature 1993, 363, 418–423.
26. von Itzstein, M.; Wu, W.-Y.; Jin, B. Carbohydr. Res. 1994, 259, 301–305.
27. Holzer, C. T.; von Itzstein, M.; Jin, B.; Pegg, M. S.; Stewart, W. P.; Wu, W.-Y. Glycoconjugate J. 1993, 10, 40–44.
28. Chandler, M.; Bamford, M. J.; Conroy, R.; Lamont, B.; Patel, B.; Patel, V. K.; Steeples, I. P.; Storer, R.; Weir, N. G.; Wright,
M.; Williamson, C. J. Chem. Soc., Perkin Trans. 1 1995, 1173–1180.
29. Kiefel, M. J.; von Itzstein, M. Prog. Med. Chem. 1999, 36, 1–28.
30. Calfee, D. P.; Hayden, F. G. Drugs 1998, 56, 537–553.
31. Waghorn, S. L.; Goa, K. L. Drugs 1998, 55, 721–725.
32. Stewart-Tull, D. E. S.; Ollar, R. A.; Scobie, T. S. J. Med. Microbiol. 1986, 22, 325–333.
33. Galen, J. E.; Ketley, J. M.; Fasano, A.; Richardson, S. H.; Wasserman, S. S.; Kaper, J. B. Infect. Immun. 1992, 60, 406–415
and references cited therein.

J. C. Wilson et al. / Tetrahedron: Asymmetry 11 (2000) 53–73
73
34. Merritt, E. A.; Sarfaty, S.; van den Akker, F.; L’Hoir, C.; Martial, J. A.; Hol, W. G. J. Protein Sci. 1994, 3, 166–175 and
references cited therein.
35. Holmquist, L. FOA Reports 1975, 9, 1–20.
36. Corfield, A. P.; Schauer, R. In Sialic Acids — Chemistry, Metabolism and Function; Schauer, R., Ed. Metabolism of sialic
acids. Springer: Wien, 1982; pp. 195–261.
37. Zbiral, E. In Carbohydrates — Synthetic Methods and Applications in Medicinal Chemistry; Ogura, H.; Hasegawa, A.;
Suami, T., Eds. Synthesis of sialic acid analogs and their behavior towards the enzymes of sialic acid metabolism and
hemagglutinin X-31 of influenza A-virus. VCH: Weinheim, 1992; pp. 304–339.
38. Meindl, P.; Tuppy, H. Hoppe-Seyler’s Z. Physiol. Chem. 1969, 350, 1088–1092.
39. Meindl, P.; Bodo, G.; Palese, P.; Schulman, J.; Tuppy, H. Virology 1974, 58, 457–463.
40. Khorlin, A. Ya.; Privalova, I. M.; Zakstelskaya, L. Ya.; Molibog, E. V.; Evstigneeva, N. A. FEBS Lett. 1970, 8, 17–19.
41. Brossmer, R.; Bürk, G.; Eschenfelder, V.; Holmquist, L.; Jäckh, R.; Neumann, B.; Rose, U. Behring Inst. Mitt. 1974, 55,
119–123.
42. Brossmer, R.; Keilich, G.; Ziegler, D. Hoppe-Seyler’s Z. Physiol. Chem. 1977, 358, 391–396.
43. Nöhle, U.; Shukla, A. K.; Schröder, C.; Reuter, G.; Schauer, R.; Kamerling, J. P.; Vliegenthart, J. F. G. Eur. J. Biochem.
1985, 152, 459–463.
44. Schreiner, E.; Zbiral, E.; Kleineidam, R. G.; Schauer, R. Carbohydr. Res. 1991, 216, 61–66.
45. Schreiner, E.; Zbiral, E. Liebigs Ann. Chem. 1990, 581–586 and references cited therein.
46. Miller, C. A.; Wang, P.; Flashner, M. Biochem. Biophys. Res. Commun. 1978, 83, 1479–1487.
47. Furuhata, K.; Sato, S.; Goto, M.; Takayanagi, H.; Ogura, H. Chem. Pharm. Bull. 1988, 36, 1872–1876.
48. Chong, A. K. J.; Pegg, M. S.; Taylor, N. R.; von Itzstein, M. Eur. J. Biochem. 1992, 207, 335–343.
49. Taylor, N. R.; von Itzstein, M. J. Med. Chem. 1994, 37, 616–624.
50. Goodford, P. J. J. Med. Chem. 1985, 28, 849–857.
51. Vasella, A.; Wyler, R. Helv. Chim. Acta 1991, 74, 451–463.
52. Vasella, A.; Wyler, R. Helv. Chim. Acta 1990, 73, 1742–1763.
53. Lentz, M. R.; Webster, R. G.; Air, G. M. Biochemistry 1987, 26, 5351–5358.
54. Okamoto, K.; Kondo, T.; Goto, T. Bull. Chem. Soc. Jpn. 1987, 60, 631–636.
55. Isecke, R.; Brossmer, R. Tetrahedron 1994, 50, 7445–7460.
56. Meindl, P.; Tuppy, H. Monatsh. Chem. 1973, 104, 402–414.
57. Smith, P. W.; Starkey, I. D.; Howes, P. D.; Sollis, S. L.; Keeling, S. P.; Cherry, P. C.; von Itzstein, M.; Wu, W.-Y.; Jin, B.
Eur. J. Med. Chem. 1996, 31, 143–150.
58. Roy, R.; Laferrière, C. A. Can. J. Chem. 1990, 68, 2045–2054.
59. Brossmer, R.; Gross, H. J. Methods Enzymol. 1994, 247, 153–176.
60. Sparks, M. A.; Williams, K. W.; Lukacs, C.; Schrell, A.; Priebe, G.; Spaltenstein, A.; Whitesides, G. M. Tetrahedron 1993,
49, 1–12.
61. Sugata, T.; Kan, Y.; Nagaregawa, Y.; Miyamoto, T.; Higuchi, R. J. Carbohydr. Chem. 1997, 16, 917–925.
62. Choi, S.-K.; Lee, S.; Whitesides, G. M. J. Org. Chem. 1996, 61, 8739–8745.
63. Taylor, N. R.; von Itzstein, M. J. Comput.-Aided Mol. Des. 1996, 10, 236–246.
64. Anschütz, R.; Bertram, W. Ber. 1903, 36, 466–468.
65. Vimr, E. R.; Lawrisuk, L.; Galen, J.; Kaper, J. B. J. Bacteriol. 1988, 170, 1495–1504.
66. Taylor, G.; Vimr, E.; Garman, E.; Laver, G. J. Mol. Biol. 1992, 226, 1287–1290.
67. Myers, R. W.; Lee, R. T.; Lee, Y. C.; Thomas, G. H.; Reynolds, L. W.; Uchida, Y. Anal. Biochem. 1980, 101, 166–174.