Synthesis and conformational analysis of a pentasaccharide corresponding to the cell-wall polysaccharide of the Group A Streptococcus

Article abstract of DOI:10.1016/S0008-6215(02)00218-5

The synthesis and conformational analysis of a pentasacccharide corresponding to a fragment of the cell-wall polysaccharide (CWPS) of the bacteria Streptococcus Group A are described. The polysaccharide consists of alternating α-(1->2)- and α-(1->3)-linke

Full text of DOI:10.1016/S0008-6215(02)00218-5

Carbohydrate Research 337 (2002) 2023–2036
www.elsevier.com/locate/carres
Synthesis and conformational analysis of a pentasaccharide
corresponding to the cell-wall polysaccharide of the Group A
Streptococcus
Christer H o¨ o¨ g, Archimede Rotondo, Blair D. Johnston, B. Mario Pinto*
Department of Chemistry, Simon Fraser Uni6ersity, Burnaby, British Columbia, Canada V5A 1S6
Received 26 February 2002; accepted 29 May 2002
This paper is dedicated to Professor Derek Horton on the occasion of his 70th birthday
Abstract
The synthesis and conformational analysis of a pentasaccharide corresponding to a fragment of the cell-wall polysaccharide
(
CWPS) of the bacteria Streptococcus Group A are described. The polysaccharide consists of alternating a-(1“2)- and
a-(1“3)-linked
L-rhamnopyranose (Rhap) residues with branching 2-acetamido-2-deoxy-D-glucopyranose (GlcpNAc) residues
linked b-(1“3) to alternate rhamnose rings. The pentasaccharide is of interest as a possible terminal unit on the CWPS, for use
in a vaccine. The syntheses employed a trichloroacetimidate glycosyl donor. Molecular dynamics (MD) calculations of the
pentasaccharide with the force fields CVFF and PARM22, both in gas phase and with explicit water present, gave different
predictions for the flexibility and preferred conformational space. Metropolis Monte Carlo (MMC) calculations with the HSEA
force field were also performed. Experimental data were obtained from 1D transient NOE measurements. Complete build-up
curves were compared to those obtained by full relaxation matrix calculations in order to derive a model of the conformation.
Overall, the best fit between experimental and calculated data was obtained with MMC simulations using the HSEA force field.
Molecular dynamics and MMC simulations of a tetrasaccharide corresponding to the Group A-variant polysaccharide, which
differs in structure from Group A in lacking the GlcpNAc residues, were also performed for purposes of comparison. © 2002
Elsevier Science Ltd. All rights reserved.
Keywords: Streptococcus Group A; Oligosaccharide synthesis; Conformations; NMR; Molecular dynamics simulations; Metropolis Monte Carlo
calculations
1
. Introduction
decade, there has been an alarming increase in the
incidence of streptococcal toxic shock syndrome and
other invasive infections such as necrotizing fasciitis
The b-hemolytic Streptococcus pyogenes or Group A
1b–d
(
GAS) is one of the primary infective agents in humans,
and myositis.
causing streptococcal pharyngitis (strep throat), some
forms of pneumonia, toxic shock syndrome, and necro-
tizing fasciitis or flesh-eating disease. Further compli-
The need for culturing methods to ensure accurate
diagnosis, together with the threat of antibiotic-resis-
1
6
tant bacteria in the future, make a vaccine protocol an
7
cations arise because of the risk to certain individuals
from the sequelae to bacterial infection such as
rheumatic fever, rheumatic carditis, heart valve disease,
attractive alternative to the present antibiotic protocol.
A vaccine strategy would also be preferred over the
current therapy of invasive diseases such as streptococ-
cal toxic shock syndrome and necrotizing fasciitis which
involves high doses of antibiotics, aggressive surgery,
2
–5
and acute glomerulonephritis. In addition, in the last
1
b,c
and intravenous administration of immunoglobulins.
We have focused our attention on development of a
carbohydrate-based vaccine based on the GAS cell-wall
polysaccharide. A vaccine consisting of heat-killed,
*
Corresponding author. Tel.: (604) 291-4327; fax: (604)
2
91-5424
E-mail address: bpinto@sfu.ca (B.M. Pinto).
0
008-6215/02/$ - see front matter © 2002 Elsevier Science Ltd. All rights reserved.
PII: S0008-6215(02)00218-5

2
024
C. H o¨ o¨ g et al. / Carbohydrate Research 337 (2002) 2023–2036
pepsin-treated bacteria has been used to immunize
two
(B(C)A%B%(C%)A) 9 (see Fig. 1).
prepared glycoconjugates of 3, 4, 7, and 9 for use as
hexasaccharides,
(A(C)BA%B%C%)
8
and
8
,9
15–17
mice, rabbits, and humans. The immune responses
are primarily directed against the exposed carbohydrate
antigens and are characterized by their restricted het-
In addition, we have
18–20
antigens and immunoadsorbents,
and have used
8
erogeneity. The immune response in vivo switches
the entire panel of compounds in the selection and
characterization of polyclonal sera and mAbs with
from IgM to IgG, indicating the action of a T cell-me-
1
8–22
8
,9
binding profiles for different epitopes.
munochemical work
Our im-
diated response.
18,21,22
has indicated that the branch
A patent that describes the preparation of GAS
polysaccharide conjugates for use as vaccines has been
point and the size of the GAS antigen are crucial
elements of the epitope recognized by both polyclonal
and mAbs that also bind the native polysaccharide
antigen.
1
0
issued. The bactericidal activity of human sera (con-
taining antibodies to GAS polysaccharide) against sev-
eral strains of GAS in vitro was demonstrated.
Although this work demonstrates the viability of an
approach based on a polysaccharide vaccine, the pre-
sentation of a high density of well-defined oligosaccha-
ride epitopes in a glycoconjugate vaccine might be
advantageous to elicit discriminating immune
Conformational analysis of the oligosaccharides 4–9
has shown that the branch point represented in the
trisaccharide 4 is a well-defined conformational feature
both in the free ligand and when bound to an anti-
23–26
body.
The polysaccharide exists in a helix in which
the rhamnosyl residues form the core of the helix and
1
1,12
responses.
the N-acetylglucosamine residues are disposed on the
Towards this end, we have described the synthesis of
portions of the cell-wall polysaccharide that might be
candidates for incorporation into vaccines. The repeat-
ing unit of the GAS cell-wall polysaccharide (CWPS)
23,24
periphery.
charide motif, as well as an accessible extended
surface.
The helix presents the branched trisac-
22
We now describe the synthesis of a hitherto unde-
scribed pentasaccharide (B(C)A%B%A) 10, that might
represent the chain terminus of the CWPS, as a suitable
candidate for incorporation into a glycoconjugate vac-
cine. We also present its conformational analysis to
gain an understanding of the possible topographies
presented to the immune system.
consists of a poly a-
posed of alternating (1“2)- and (1“3)-linkages to
which N-acetyl-b- -glucosamine residues are attached
at the 3-position of the rhamnose residue (see Fig.
L-rhamnopyranosyl backbone com-
D
1
3,14
1).
We have synthesized the haptens: disaccharide
(
CB) 2, linear trisaccharide (ABC) 3, branched trisac-
charide (B(C)A%) 4, two tetrasaccharides (AB(C)A%) 5
and (B(C)A%B%) 6, pentasaccharide (B(C)A%B%C%) 7, and
2
. Results and discussion
Synthesis.—The pentasaccharide 10 was synthesized
by slight modifications of the procedures developed for
17
the preparation of related oligosaccharides. The pro-
tection strategy involved only ester protecting groups
and was therefore suitable for preparation of the final
oligosaccharide in the form of an allyl glycoside. We
have previously found that allyl glycosides can provide
a convenient tether for subsequent attachment of
20
oligosaccharides to solid supports or for conjugation
19
to proteins.
The trisaccharide acceptor 11 was prepared as previ-
17
ously described. The required, selectively protected,
rhamnopyranosyl trichloroacetimidate 12 has been re-
27
cently reported without characterization by NMR,
which has therefore been dealt with here. Glycosylation
of the acceptor 11 by donor 12, with activation by a
catalytic amount of triethylsilyl triflate, proceeded
smoothly in CH Cl2 to yield the tetrasaccharide 13
2
(
Scheme 1). The sole acetate ester at the 2-position of
the A%-ring was then selectively removed by methanoly-
sis using HCl in methanol to generate the tetra-
saccharide glycosyl acceptor 14. Glycosylation with 12,
as before, afforded the required protected pentasaccha-
Fig. 1. Cell-wall polysaccharides of the bacteria Streptococcus
Group A (left) and Group A-variant (right). Synthetic exam-
ples: CB, 2; ABC, 3; B(C)A%, 4; AB(C)A%, 5; B(C)A%B%, 6;
B(C)A%B%C%, 7; A(C)BA%B%C%, 8; B(C)A%B%(C%)A, 9.

C. H o¨ o¨ g et al. / Carbohydrate Research 337 (2002) 2023–2036
2025
conditions regenerated the pure pentasaccharide 10
Scheme 3).
NMR analysis.—The H NMR spectrum of the pen-
(
1
tasaccharide 10 exhibited a set of five downfield dou-
blets corresponding to the five anomeric protons. One
of these doublets, at l 4.69, was unique in having the
larger J1,2 coupling constant (8.4 Hz) expected for a
b-N-acetylglycosamine glycoside. This resonance was
assigned to H-1 of the C-ring and provided a conve-
nient starting point for assignment of the other proton
resonances, particularly in the crowded region between
l 4.4 and l 3.4. Analysis of the COSY spectrum
1
provided complete assignment of the C-ring H reso-
nances. A TOCSY spectrum allowed resonances of all
1
four rhamnose residues to be associated with their H
resonances, which appeared as four well-resolved dou-
blets in the l 5.2–4.8 region, each having the small J1,2
coupling constant (1.4–1.7 Hz) characteristic of a-L-
rhamnopyranosides.
Assignment of the four sets of resonances to individ-
ual rhamnose rings proceeded from analysis of
NOESY, ROESY and 1D-NOE difference spectra. The
6
00-MHz NOESY spectrum (mixing time 250 ms) of 1
Scheme 1.
exhibited only a few, very small, negative NOE correla-
tions. The size of pentasaccharide 1 is such that the
Scheme 2.
ride 15 (Scheme 2). Deprotection by successive treat-
ment with NaOMe–MeOH and then ethylenediamine
removed all of the protecting groups, and the free
amine at the 2-position of ring-C was then selectively
acetylated with acetic anhydride in methanol to give the
crude pentasaccharide 10. In order to facilitate purifica-
tion, it proved necessary to re-acetylate all free hy-
droxyl groups to give the protected pentasaccharide 16,
which was more amenable to purification by chro-
matography on silica gel. Deprotection under Zempl e´ n
Scheme 3.

2
026
C. H o¨ o¨ g et al. / Carbohydrate Research 337 (2002) 2023–2036
correlation times resulting from molecular tumbling
result in only small NOE values. ROESY spectra (ob-
tained with continuous spin–lock irradiation times
varying between 100 and 500 ms) were similar, except
that the correlations were now positive with respect to
the negative diagonal peaks. Despite their low intensity,
several inter-ring correlations were obvious. Two of the
four rhamnose H-1 resonances exhibited cross-glycosyl
NOE correlations with H-2 resonances of other rham-
nose residues. These H-1 resonances were assigned to
either the A% or the A rings. One rhamnose residue
showed an H-1 to H-3 cross-glycosyl NOE and was
thus assigned as ring B%. The remaining rhamnose H-1
program. In the case of the pentasaccharide, the calcu-
lated NOE data and corresponding H– H distances
were compared with experimental NOE data.
1
1
signal was correlated with the OCH resonances of the
2
allyl group and, therefore, must correspond to H-1 of
the B-ring. ROESY correlations between non-anomeric
resonances in the l 4.2 to l 3.5 region were obscured by
strong TOCSY breakthrough peaks.
Differentiation of the A and A% rhamnose resonances
was achieved by 1D transient NOE difference experi-
(
a) Computations. Results from the molecular dynam-
ics simulations of 10 are presented in scatterplots about
the ƒ and „ torsion angles. Fig. 2 indicates highly
populated regions of the ƒ/„-space, as shown as black-
ened regions on the plot. The scatterplots from a 10 ns
molecular dynamics simulation with the CVFF force
field (simulation IV), in vacuo, is shown in Fig. 2a: two
roughly equally populated states can be discerned for
the non-reducing end, i.e., between residue A and B%.
State A is characterized by an average ƒA“B% torsion
angle of 55° and „A“B%=11° for the terminal a-
Rhap-(1“2)-a- -Rhap linkage, whereas for state B, the
values are 10 and −46°, respectively (Table 1).
The next linkage, between monosaccharide B% and A%,
an a- -Rhap-(1“3)-a-
ments. Inversion of the H-2 resonance with a selective
B
Gaussian-shaped pulse showed the expected enhance-
ments for the H-1B and H-3B signals and a slightly
greater enhancement for a single rhamnose H-1 reso-
nance at l 5.17. This was assigned as H-1 . The
A%
remaining possibility, the rhamnose H-1 resonance at l
4.97, showed no such enhancement, and, therefore, this
L-
resonance must be that of H-1 . Corroborative evi-
A
L
dence for this assignment was provided when inversion
of the H-1A resonance resulted in enhancement of the
H-2B% signal and had no effect on the H-2 resonance.
B
L
L-Rhap linkage, ƒB%“A% shows
Since TOCSY correlations allowed the association of
each rhamnose H-1 with all other proton resonances on
two different states, while „B%“A% remains in the
−gauche conformation. The time scales for persistence
of these states are in the nanosecond region (Fig. 3).
1
the same ring, the H NMR spectrum assignment was
completed. The spectral dispersion was such that only
the H-6C and H-3C resonances exhibited significant
second-order effects due to overlap of H-4 and H-5
The second a-
L
-Rhap-(1“2)-a- -Rhap linkage, A%“B,
L
shows different behavior compared to that (A“B%) at
the non-reducing end. There is now only one populated
conformer, and it does not correspond to either state A
or B. The average value for ƒA%“B, which shows the
largest change, is −5°, while for „A%“B it is −16°. The
b-(1“3)-linkage between the N-acetylglucosamine
C
C
coupled resonances.
Analysis of the C– H correlated spectrum then
1
3
1
1
3
allowed complete assignment of the C NMR spec-
trum. The one-bond, JC1,H1 coupling constants of the
rhamnopyranose residues were measured from a cou-
1
3
1
(
GlcNAc) and rhamnose (Rha) residues (C“B) ex-
pled C– H correlated spectrum and varied from 173
to 178 Hz, such values being consistent with the a-
hibits two conformations, in which flexibility is dis-
played at the „C“B torsion angle, and an energy barrier
is located at −50°. Simulation V, employing a 20-ns
MD simulation with the CVFF force field, shows the
population of the anti conformer, ƒC“B= ꢀ180°, dur-
ing the last 2 ns and no transition back to the +gauche
conformation is observed.
The molecular dynamics simulations employing the
CHARMM-based force field (simulation I–III) give a
different picture of the preferred conformations of 10
(Fig. 2b). While the molecular dynamics simulations
with InsightII/Discover using the force field CVFF show
two equally populated conformations at the non-reduc-
2
8
configuration.
Conformational analysis.—The repeating units of the
cell-wall polysaccharide of Streptococcus Group A and
variant A strain are shown in Fig. 1. The conforma-
tions of the pentasaccharide 10 and tetrasaccharide 17,
corresponding to fragments of the cell-wall polysaccha-
rides of the Streptococcus Group A and variant A
bacteria, respectively (see Fig. 1), were analyzed with
molecular dynamics simulations employing the CVFF
and CHARMM force fields. Metropolis Monte Carlo
(
MMC) simulations of the pentasaccharide 10 and tetra-
saccharide 17 were also performed with the GEGOP

C. H o¨ o¨ g et al. / Carbohydrate Research 337 (2002) 2023–2036
2027
Fig. 2. Scatter plots of ƒ versus „ for compound 10 from (a) simulation IV (CVFF), (b) simulation III (PARM22), and (c) simulation
VI (HSEA). For the CVFF simulation IV, conformational states were averaged over the following time span: (A) 0.0-2.0 ns, (B)
3.0–6.0 ns.
Table 1
Average torsion angles from the molecular dynamics (MD) and Metropolis Monte Carlo (MMC) simulations for compound 10,
with the rmsd values presented in parentheses
PARM22
CVFF
HSEA
a
a
Sim. I
20° (21°)
−55° (26°)
−13° (16°)
−29° (18°)
12° (13°)
13° (32°)
−10° (16°)
−34° (12°)
Sim. II
Sim. III
21° (21°)
−49° (20°)
−13° (14°)
−5° (31°)
9° (11°)
27° (13°)
1° (32)
−26° (9°)
Sim. IV
39° (28°)
−9° (31°)
−2° (33°)
−37° (20°)
−5° (21°)
−16° (18°)
31° (15°)
Sim. V
45° (24°)
0° (28°)
11° (30°)
−32° (26°)
–11° (25°)
−16° (19°)
45° (51°)
Conf. A
Conf B
Sim. VI
ƒA“B%
„A“B%
ƒB%“A%
„B%“A%
ƒA%“B
„A%“B
ƒC“B
„C“B
31° (14°)
−41° (19°)
−30° (20°)
−60° (38°)
2° (10°)
31° (11°)
−11° (10°)
−32° (10°)
55° (11°)
11° (16°)
17° (21°)
−36° (21°)
−11° (18°)
−15° (17°)
33° (16°)
10° (27°)
−46° (12°)
−37° (14°)
−41° (13°)
8° (17°)
−17° (17°)
27° (14°)
−37° (11°)
40° (19°)
2° (23°)
41° (22°)
8° (24°)
40° (15°)
11° (20°)
53° (22°)
4° (22°)
−66° (23°)
−15° (28°)
−23° (24°)
a
The conformational states were averaged over the following time span: (A) 0.0–2.0 ns, (B) 3.0–6.0 ns, both from simulation
IV (CVFF).

2
028
C. H o¨ o¨ g et al. / Carbohydrate Research 337 (2002) 2023–2036
Table 2
Average torsion angles from the molecular dynamics (MD
)
and Metropolis Monte Carlo (MMC) simulations for com-
pound 17, with the rmsd values presented in parentheses
PARM22
CVFF
HSEA
ƒA“B%
„A“B%
ƒB%“A%
„B%“A%
ƒA%“B
„A%“B
14° (25°)
41° (15°)
−19° (24°)
26° (32°)
−14° (30°)
37° (21°)
−17° (26°)
40° (18°)
2° (23°)
41° (21°)
8° (24°)
37° (20°)
3° (23°)
−47° (20°)
−9° (14°)
10° (36°)
24° (22°)
−51° (26°)
age, the major populated conformer shows a ƒC“B
gauche angle. The torsion angle ƒB%“A% shows a
−
negative value, in contrast to the CVFF calculations
where the −gauche and +gauche states are equally
populated.
A Metropolis Monte Carlo (MMC) simulation was
also performed with the HSEA force field at 600 K
(
simulation VI), a temperature which has previously
1
1
been found to reproduce experimental H– H distances
30
well. The results are illustrated in Fig. 2c and Table 2.
The ƒ-torsion angles for the glycosidic linkage between
two rhamnose units, i.e., A“B%, B%“A% and A%“B,
are ꢀ40° and for the C“B linkage 53°. The „ torsion
angles range from 2 to 11°. The torsion angle ƒC“B
shows an anti conformation populated to the extent of
2
% which is slightly lower than previous findings for
31,32
smaller oligosaccharides.
MMC simulations with the
HSEA force field have previously shown the population
24
of an anti conformer at ƒC“B
.
Comparison of the conformations of the pentasac-
charide 10 and tetrasaccharide 17 indicates that the
largest difference is between the A% and B residues,
where the lack of a GlcNAc residue in the tetra-
saccharide increases the available conformational space
for the a-L-Rhap-(1“2)-a-L-Rhap linkage and makes
the scatterplots for the A“B% and A%“B linkages
more similar (Fig. 4a–c and Table 2). The MD simula-
tion with the CVFF force field also shows that the
Fig. 3. Time series of torsion angles for the non-reducing end
1
1
of 10 and the corresponding H– H distances from simula-
tion IV, showing the time scales for population of states A
and B.
a-L
-Rhap-(1“3)-a- -Rhap (B%“A%) linkage is affected
L
as the flexibility for cB%“A% is increased. The conforma-
tion at the terminal non-reducing end seems to be
affected only to a minor extent because of the lack of
residue C. The resulting proton–proton distances from
the different simulations are listed in Tables 3 and 4.
(b) Experimental. One-dimensional transient NOE
spectra were measured at different mixing times, and
the resulting inter-residue build-up curves are shown in
Fig. 5, together with the corresponding calculated
build-up curves calculated from simulation III and a
combination of states A and B found in simulations IV
and V. The calculated build-up curves were generated
ing end (A“B%), the MD simulations with PARM22 show
one major and one minor populated state, together
roughly corresponding to region B. One striking differ-
ence between the results obtained with the force fields
PARM22 and CVFF is that the latter shows greater
flexibility about ƒ than „, whereas the former shows
the opposite, in agreement with the exo-anomeric ef-
29
fect. Simulation I differs from simulation III in that a
gauche state is also populated for the „A%“B linkage
Table 1). For the b- -GlcpNAc-(1“3)-a- -Rhap link-
−
(
D
L

C. H o¨ o¨ g et al. / Carbohydrate Research 337 (2002) 2023–2036
2029
Fig. 4. Scatter plots of ƒ versus „ for compound 17, from simulations with the force fields (a) CVFF, (b) PARM22 and (c) HSEA
.
by full relaxation matrix calculations of averaged con-
considering that it is well known that molecules tend to
31
formations with the program CROSREL
,
as described
minimize their accessible surface in gas phase simula-
23,34
36
in our earlier publications.
The correlation time and
tions, and in the case of HSEA, the shape is due to the
leakage rate, R , obtained by fitting data for intra-
lack of attractive electrostatic forces. The most impor-
tant torsion angles that govern the overall shape of 10
are between the central residues, i.e., the torsion angles
L
residue proton pairs that have defined distances, were
found to be 737 ps and 0.20 Hz, respectively, for the
CVFF-generated conformations, A+B, and 596 ps and
between B%“A% and A%“B. Simulation III (PARM22
)
0
.0003 Hz, respectively, for the force field PARM22, with
and simulation IV (HSEA) give overall better fits be-
tween experimental and calculated H– H distances
than the combination A+B (CVFF) (Table 4).
1
1
explicit water present, i.e., simulation III. The correla-
3
5
tion times are reasonable for a pentasaccharide. The
agreement between calculated and the normalized ex-
perimental peak intensities is expressed by a weighted
residual RW factor, inspired by X-ray crystallography.
CROSREL permits determination of the ratio of two or
more conformations and the best RW factor was
reached when region A was populated to the extent of
Table 3
a
Calculated proton–proton distances for compound 17
r (A
,
)
PARM22
CVFF
HSEA
8
2% and B to the extent of 18%. In general, for the
non-reducing end, the combination of states A and B
ꢀ) gives the best fit to the experimental curves with
H-1
A
–H-1B%
2.45
2.20
3.10
2.16
3.79
2.42
2.26
2.95
3.31
2.33
2.53
2.36
3.99
3.26
2.34
2.62
3.32
2.28
2.34
2.33
3.84
3.32
2.27
2.38
H-1 –H-2
A
B%
H-5 –H-1
(
A
B%
H-1 –H-3
B%
A%
the exception of the short inter-residue distances over
the glycosidic linkages.
H-1 –H-4
B%
A%
H-1 –H-1B
A%
The differences in overall conformation of the pen-
tasaccharide 10, which is important for biological
recognition, suggested by the three force fields are as
follows: CVFF, in vacuo, results in a compact sphere
while CHARMM, in water, and HSEA, in vacuo, give a
more extended structure. This divergence is expected
H-1 –H-2B
A%
H-5 –H-1B
A%
a
−6 −1/6
The calculated distances are averaged as r=Žr

using the proton pair H-1–H-2 in rhamnose as the reference
distance.

2
030
C. H o¨ o¨ g et al. / Carbohydrate Research 337 (2002) 2023–2036
Table 4
Experimental and calculated proton–proton distances for compound 10
a
r (A
,
)
Exp.
CHARMM
CVFF
HSEA
A+B
Sim. I
Sim. II
Sim. III
Sim. IV
Sim. V
Sim. VI
H-1 –H-1
3.50
2.16
2.46
2.14
3.70
3.84
2.37
4.36
3.70
2.52
3.25
2.27
2.34
2.26
3.09
2.16
4.34
3.16
2.09
2.95
4.80
3.24
2.56
2.21
2.69
2.16
2.68
2.54
3.94
4.02
2.08
2.91
4.62
3.42
2.65
2.20
2.47
2.21
2.95
2.13
4.01
4.00
2.08
3.08
3.37
3.36
2.81
2.19
2.98
2.45
2.55
2.37
4.48
2.91
2.25
4.67
5.18
3.59
3.14
2.30
3.23
2.47
2.47
2.32
4.26
2.83
2.29
4.81
4.88
3.65
3.29
2.37
3.34
2.29
2.34
2.34
3.85
3.57
2.29
4.38
3.09
2.41
3.64
2.39
3.35
2.48
2.40
2.33
4.42
2.89
2.25
3.82
5.13
3.68
3.17
2.30
A
B%
B%
H-1 –H-2
A
H-1 –H-5
B%
A
H-1 –H-3
B%
A%
A%
H-1 –H-4
B%
H-1 –H-1
A%
B
B
C
C
H-1 –H-2
A%
H-1 –H-1
A%
H-1 –H-2
A%
H-1 –H-5
B
A%
H-1 –H-2
C
B
B
H-1 –H-3
C
a
−6 −1/6
The calculated distances are averaged as r=Žr

using the proton pair H-1–H-2 in rhamnose or H-1–H-3 in GlcNAc
as the reference distance.
The experimental distances in Table 4 were derived
agreement. The calculated distance of 3.09 A
H-1 –H-2 in the MMC simulation VI of 10 is consis-
,
between
by the isolated spin–pair approximation (ISPA) (Eq.
A%
C
3
7
(
1)), with the reference distance, rH1,H2=2.52 A
,
,
tent with previous measurements on a corresponding
hexasaccharide that has an additional GlcpNAc residue
taken from the rhamnose residues in the CHARMM
simulations.
23
at the 3-position on the B% unit. For the pentasaccha-
ride 10, none of the force fields used gives a good fit
with experimental data for the H-1 –H-2 contact,
1
r =r (| /| )
6
(1)
ij
ref ref ij
A%
C
although for the longer H-1 –H-1 distance, the MMC
A%
C
The last column in Table 4 shows the effective dis-
simulation VI (HSEA) shows excellent agreement,
whereas PARM22 and CVFF do not. The longer experi-
mental value for H-1 –H-2 for 10 is difficult to ex-
tances in the pentasaccharide 10 when region A is
populated to the extent of 82% and B to 18%, as
suggested by CROSREL. These distances are derived
from the following two state relationship (Eq. (2)),
where the right-hand side of the equation is given in
Table 4 and x is the population of state A.
A
C
plain since the other distances across the torsion angles
that define the three-dimensional structure for the link-
ages between residue A%“B and C“B give reasonable
agreement with those obtained in simulation VI. We
could not obtain evidence to support the population of
an anti-conformation at ƒC“B, because spectral over-
lap did not permit measurement of NOEs between
−
6
−6
−6
A+B
(
x)ꢀr
ꢁ+(1−x)ꢀrState Bꢁ=r
(2)
State A
The distances obtained from the combined popula-
tion, A+B, give good agreement for the terminal non-
reducing end, with the exception of the inter-glycosidic
distance H-1 –H-2 where both CHARMM and HSEA
H-2 and H-3 .
C
B
A
B%
provide better agreement (see also the build-up curves,
Fig. 5). However, PARM22, which only populates region
B, gives the worst fit for this linkage. The experimental
3
. Experimental
General methods.— H NMR and C NMR spectra
1
13
distances between H-1 –H-1 (3.5 A) and H-5 –H-1
,
A
B%
A B%
,
) are in close agreement with values found in a
-Rhap-
-Rhap-OMe. For the central linkages, B%“
were recorded with a Bruker AMX-400 or with a
Bruker AMX-600 NMR spectrometer for solutions in
CDCl
(
2.46 A
conformational study of the disaccharide a-
L
3
8
1
(
1“2)-a-
L
3
(internal standards, for H: residual CHCl3,
l
13
A% and A%“B%, the MMC simulation VI (HSEA) gives the
best agreement with experiment; for the linkage B%“A%,
simulation III (PARM22) also gives reasonable agree-
ment. The linkage between the GlcpNAc and rhamnose
residues is best described by the CVFF force field, with
the combined conformations, A+B; the distances re-
sulting from simulation VI (HSEA) also gives reasonable
7.24 and for C: CDCl
3
, l 77.0) or in D
O (internal
2
1
13
standard, for H: HOD, l 4.78; and for C: MeOH, l
49.50). First-order chemical shifts and coupling con-
stants were obtained from one-dimensional spectra, and
all assignments of proton and carbon resonances were
based on COSY, NOESY, TOCSY and
HMQC experiments. The h/i stereochemistry of the
1
3
1
C– H

C. H o¨ o¨ g et al. / Carbohydrate Research 337 (2002) 2023–2036
2031
glycosidic bonds was confirmed by measurement of the
JC1,H1 coupling constants. Optical rotations were
zoic acid matrix. Microanalyses were performed by the
SFU Department of Chemistry microanalytical service.
NMR spectroscopy. The pentasaccharide 10 (13 mg)
1
28
measured on a Rudolph Research Autopol II polarime-
ter. TLC was performed on aluminum plates precoated
with Silica Gel 60 F254 (E. Merck) and detected with
UV light and/or spraying with a solution containing 1%
Ce(SO ) and 1.5% molybdic acid in 10% aq H SO
4
was lyophilized twice from D O in order to exchange
2
the hydroxyl protons; it was then redissolved in of D O
2
(0.6 mL). NMR spectra were measured on a 600-MHz
Bruker AMX spectrometer equipped with an inverse
triple probe. Chemical shifts, referenced to external
3-trimethylsilyl-1-propanesulfonic acid (DSS), were as-
signed from 2D TOCSY, ROESY and HMQC experi-
ments at 298 K. For the TOCSY spectra, 32 transients
4
2
2
followed by heating. Compounds were purified by flash
chromatography on Silica Gel 60 (E. Merck, 230–400
mesh). MALDI mass spectra were obtained on a
PerSeptive Biosystems, Voyager DE time-of-flight spec-
trometer for samples dispersed in a 2,5-dihydroxyben-
of 2048 points were accumulated for 512 t -increments,
1
1
1
Fig. 5. Theoretical and experimental proton–proton cross-relaxation build-up curves obtained from 1D H– H NOE spectra:
experimental (ꢁ), combination A+B from CVFF (ꢀ) and the average conformation from simulation III, PARM22 (ꢂ).

2
032
C. H o¨ o¨ g et al. / Carbohydrate Research 337 (2002) 2023–2036
with the MLEV composite spin–lock pulse, and for the
ROESY and HMQC experiments, 48 transients were
used at every increment.
simulations of 10 with PARM22 had different starting
conformations that were either derived from the MD
simulations with the consistent valence force field
One-dimensional transient NOEs were measured at
(CVFF), state A or B, or with all ƒ torsion angles set to
6
–8 mixing times to derive the cross-relaxation rates
60° and „=0°. Simulations with the CHARMM molecu-
lar mechanics program on an SGI Octane (single pro-
cessor) used a CPU time of ꢀ30 h for 100 ps.
and thereby, proton–proton distances with the isolated
spin–pair approximation (ISPA) procedure. Errors of
about 5–10% in integrated intensities were estimated.
Selective excitation was achieved by use of Gaussian-
shaped pulses. These pulses were defined with the shape
program and 1024 digital points. Seven different proton
resonances were saturated with pulse durations of 80,
The InsightII/Discover package was used for the
molecular dynamics simulations with the force field
46
CVFF
.
Energy minimization with steepest descent was
followed by conjugate gradient and a quasi-Newton–
Raphson method (BFGS) until a maximum derivative
−
1
−1
2
00, and 300 ms, and relative power attenuations were
of 0.01 kcal mol
A
,
was reached. Subsequent
found to be 71, 78, and 81 dB, respectively.
heating and equilibration for 50 ps, with a target tem-
perature of 298 K, were performed, followed by simula-
tions of 10 ns (simulation IV) or 20 ns (simulation V).
The time step was set to 1 fs, and no cutoffs were
employed. The dielectric constant was set to r, where r
is the distance between two non-bonding partially
charged atoms. Coordinate sets were saved every 0.1 ps
for later analysis.
Computer simulations. Metropolis Monte Carlo
(MMC) and molecular dynamics (MD) simulations were
performed for the pentasaccharide (10) and the tetra-
saccharide (17). The torsion angles ƒ and „ are defined
as follows, ƒ=H-1–C-1–O-x–C-x and „=C-1–O-x–
C-x–H-x, where x is the linkage position. For the MMC
runs, the program GEGOP39 (version 2.7), with its force
4
0
field HSEA
,
was employed. The simulations in vacuo
Simulations I–III were performed with the PARM22
force field, with explicit water present, using the
CHARMM program, simulations IV–V in the gas phase
with the consistent valence force field (CVFF), and simu-
lation VI with the HSEA force field using the program
GEGOP. The simulations for the tetrasaccharide 17 are
only denoted by the respective force field used.
NOE build-up curves were calculated from average
conformations derived from the simulations using the
6
at 600 K employed 2×10 macrosteps with a total
acceptance ratio of 45 and 55% for 10 and 17, respec-
tively. The maximum step length for the glycosidic
torsion angles was set to 20°.
4
1
The molecular mechanics program CHARMM
,
ver-
sion 27b4, was used for MD simulations of 10 and 17,
with explicit water present, with the CHARMM-based
force field, PARM22 (Accelrys Inc., San Diego, CA,
USA). The oligosaccharide was placed in a previously
33
CROSREL program.
rhamnopyranosyl trichloroacetimidate (12).—A solu-
tion of 2-O-acetyl-3,4-di-O-benzoyl-a- -rhamnopyran-
ose (2.70 g, 6.52 mmol) in CH Cl (50 mL) was cooled
2-O-Acetyl-3,4-di-O-benzoyl-h-L-
equilibrated cubic water box of length 34.137 A
,
con-
taining 1331 TIP3P water molecules, and those waters
that were closer than 2.5 A to any solute atom were
42
L
27
,
2
2
removed. This procedure resulted in a system with the
oligosaccharide and 125492 (10) and 1274 (17) waters
which was minimized in energy using steepest descent
in an ice bath. Trichloroacetonitrile (2.0 mL, 20 mmol)
and DBU (0.25 mL, 1.7 mmol) were added, and the
mixture was stirred for 0.5 h under an N atmosphere.
2
(
250 steps) and adopted basis Newton–Raphson (250
The cooling bath was removed, and the mixture al-
lowed to warm to rt over 15 min. Volatile material was
removed by rotary evaporation and the residue was
purified by column chromatography (3:2 hexanes–
EtOAc) to give 4 as a pale-yellow foam (2.74 g, 75%).
steps) with the oligosaccharide kept fixed. After releas-
ing the constraints, further energy minimizations were
performed with steepest descent (400 steps), followed
by adopted basis Newton–Raphson until the root-
mean-square gradient was less than 0.01 kcal mol
A
heating in 5 K increments during 8 ps to 300 K, where
the system was equilibrated for 100 ps. Each simulation
was run for 1 ns. Berendsens weak-coupling algorithm
−
1
1
The H NMR spectrum indicated that the product was
−
1
,
. Velocities were initialized at 100 K, followed by
\95% pure a-trichloroacetimidate, and this material
was used directly in glycosylation reactions. Prolonged
2
2
storage at rt resulted in partial hydrolysis. [h] +16°
D
43
1
(c 1.4, CH Cl ). H NMR (400 MHz, CDCl ): l 8.80
2
2
3
was used to keep the temperature constant during the
simulations. Minimum-image boundary conditions
were used with a heuristic non-bond frequency update
(bs, 1 H, NH), 8.00–7.30 (m, 10 H, Ar), 6.85 (d, 1 H,
J1,2 1.9 Hz, H-1), 5.76 (dd, 1 H, J2,3 3.5, J3,4 10.2 Hz,
H-3), 5.68 (dd, 1 H, H-2), 5.65 (dd, 1 H, J4,5 10.1 Hz,
4
4
and a force shift cutoff acting to 15 A
,
, using a
H-4), 4.33 (dq, 1 H, H-5), 2.19 (s, 3 H, OAc), 138 (d, 3
45
13
dielectric constant of unity. SHAKE
,
with a tolerance
H, J5,6 6.3 Hz, CH , H-6). C NMR (100 MHz,
3
−
4
gradient of 10 , was used to restrain hydrogen-heavy
atom bond stretch; the time step was accordingly set to
CDCl ): l 169.65 (CꢀO, OAc), 165.68, 165.45 (2×
3
CꢀO, OBz), 160.04 (CꢀN), 133.45–128.34 (10 C, Ar),
2
fs. Data were saved every 0.2 ps for analysis. All MD
94.75 (C-1), 70.79 (C-4),

C. H o¨ o¨ g et al. / Carbohydrate Research 337 (2002) 2023–2036
2033
6
9.49 (C-3), 68.45 (C-2), 66.73 (C-5), 20.73 (CH , OAc),
(CH , OAc), 17.61, 17.55 and 17.32 (3×CH , C-6A%,
3 3
3
1
7.60 (CH , C-6).
C-6B%, C-6B). MALDI MS Calcd for C H NNaO :
3
93 83 28
Allyl 2-O-acetyl-3,4-di-O-benzoyl-h-
osyl-(1“3)-2,4-di-O-benzoyl-h- -rhamnopyranosyl-(1
2)-[3,4,6-tri-O-benzoyl-2-deoxy-2-phthalimido-i-
glucopyranosyl-(1“3)]-4-O-benzoyl-h- -rhamnopyran-
oside (13).—A solution of the acceptor 11 (1.95 g, 1.54
L
-rhamnopyran-
m/z 1685.7; found m/z 1685.6. Anal. Calcd for
L
C H NO : C, 67.18; H, 5.03; N, 0.84. Found: C,
93
83
28
“
D-
67.05; H, 4.98; N, 0.86.
L
Allyl 3,4-di-O-benzoyl-h- -rhamnopyranosyl-(1“3)-
L
2,4-di-O-benzoyl - h -
tri-O-benzoyl-2-deoxy-2-phthalimido-i-
osyl-(1“3)]-4-O-benzoyl-h- -rhamnopyranoside (14).
L
- rhamnopyranosyl - (1“2)-[3,4,6-
1
7
mmol) and the trichloroacetimidate 12 (1.04 g, 1.86
D
-glucopyran-
mmol) in anhyd CH Cl2 (30 mL) was stirred with
L
2
freshly-activated crushed 4 A
,
molecular sieves (2.0 g)
—A solution of HCl in MeOH was prepared by adding
acetyl chloride (2.0 mL) dropwise to dry methanol (40
mL). This solution was used to dissolve the tetra-
saccharide 13 (1.93 g, 1.16 mmol), and the homoge-
neous solution was warmed in a 40–45 °C water bath
for 3.5 h. The mixture was cooled to rt and poured into
under N for 10 min at room temperature. The mixture
2
was cooled in a −40 °C bath, and triethylsilyl triflate
30 mL, 0.13 mmol) was added. The temperature was
allowed to rise to −20 °C over a period of 1 h, then
the cooling bath was removed, and the mixture was
allowed to warm to room temperature. Triethylamine
(
a cold, stirred, satd aq NaHCO solution (100 mL). The
3
(
50 mL) was added, and the molecular sieves were
aqueous mixture was extracted twice with CH Cl (100
mL, 50 mL), and the combined extracts were dried over
2
2
removed by filtration through Celite with the aid of
additional CH Cl (100 mL). The filtrate was washed
MgSO and evaporated to leave a colorless foam (1.57
2
2
4
with satd NaHCO (30 mL), dried over MgSO and
concentrated to a syrup. Flash chromatography (1:1
hexanes–EtOAc) gave the tetrasaccharide 13 (2.12 g,
g, 83%) that was essentially pure by NMR analysis and
was used in the next reaction without purification. An
analytical sample was purified by column chromatogra-
phy (5:1 toluene–EtOAc) to give compound 14 as an
3
4
8
3%) as a colorless, hard foam containing a trace of
1
20
1
trichloroacetamide ( H NMR, broad singlet at l 6.6
amorphous solid. [h]D +56° (c 0.96, CHCl ). H NMR
3
ppm). An analytically pure sample was obtained by
(600 MHz, CDCl ): l 8.43–6.88 (m, 44 H, Ar), 6.24
3
2
0
re-chromatography on silica gel. [h] +58° (c 0.76,
(dd, 1 H, J2C,3C 10.5, J3C,4C 9.2 Hz, H-3C), 6.03 (dd, 1
H, J4C,5C 9.7 Hz, H-4C), 5.83–5.75 (m, 1 H, CHꢀCH2),
5.75–5.72 (m, 2 H, H-2A%, H-1C), 5.68 (t, 1 H,
J3A%,4A%=J4A%,5A%=9.8 Hz, H-4A%), 5.62 (d, 1 H, J1A%,2A%
1.6 Hz, H-1A%), 5.53 (t, 1 H, J3B%,4B%=J4B%,5B%=9.8 Hz,
H-4B%), 5.47 (dd, 1 H, J2B%,3B% 3.0, J3B%,4B% 10.1 Hz,
H-3B%), 5.30 (d, 1 H, H-1B%), 5.29 (dd, 1 H, J1C,2C 8.1
D
1
CHCl ). H NMR (400 MHz, CDCl ): l 8.43–6.88 (m,
3
3
4
4 H, Ar), 6.20 (dd, 1 H, J2C,3C 10.5, J3C,4C 9.1 Hz,
H-3C), 6.06 (dd, 1 H, J4C,5C 9.9 Hz, H-4C), 5.85–5.75
m, 1 H, CHꢀCH ), 5.75 (d, 1 H, J1C,2C 8.3 Hz, H-1C),
(
2
5
1
.69 (t, 1 H, J3A%,4A%=J4A%,5A%=9.9 Hz, H-4A%), 5.67 (d,
H, J1A%,2A% 1.9 Hz, H-1A%), 5.65 (dd, 1 H, J2A%,3A% 3.1
Hz, H-2A%), 5.59 (dd, 1 H, J2B%,3B% 3.5, J3B%,4B% 10.0 Hz,
H-3B%), 5.45 (t, 1 H, J3B%,4B%=J4B%,5B%=9.9 Hz, H-4B%),
Hz, H-2C), 5.28–5.25 (m, 1 H, CHꢀCH ), 5.19–5.17
2
(m, 1 H, CHꢀCH ), 5.17 (t, 1 H, J3B,4B=J4B,5B=10.0
2
5
5
.35 (dd, 1 H, H-2C), 5.31–5.25 (m, 1 H, CHꢀCH2),
Hz, H-4B), 4.83 (d, 1 H, J1B,2B 1.5 Hz, H-1B), 4.71 (dd,
1 H, J6C,6%C 12.1, J5C,6C 2.7 Hz, H-6C), 4.67 (dd, 1 H,
J2A%,3A% 3.1 Hz, H-3A%), 4.55 (dd, 1 H, J5C,6C 5.7 Hz,
H-6%C), 4.41 (dd, 1 H, J2B,3B 3.3 Hz, H-2B), 4.31–4.18
(m, 5 H, H-2B%, H-5B%, H-5A%, H-5C, H-3B), 4.13–4.09
.29 (dd, 1 H, J1B%,2B% 1.3, H-2B%), 5.21–5.17 (m, 1 H,
CHꢀCH ), 5.19 (d, 1 H, H-1B%), 5.17 (t, 1 H, J3B,4B=
2
J4B,5B=9.8 Hz, H-4B), 4.80 (d, 1 H, J1B,2B 1.3 Hz,
H-1B), 4.71 (dd, 1 H, J6C,6%C 12.2, J5C,6C 2.6 Hz, H-6C),
4
.65 (dd, 1 H, H-3A%), 4.50 (dd, 1 H, J5C,6C 5.6 Hz,
(m,
1
H, OCH CHꢀCH ), 3.97–3.93 (m,
1
H,
6.3 Hz, H-5B),
5B,6B
2
2
H-6%C), 4.40 (dd, 1 H, J2B,3B 3.7 Hz, H-2B), 4.34 (dq, 1
H, J5B%,6B% 6.2 Hz, H-5B%), 4.29 (ddd, 1 H, H-5C), 4.22
OCH CHꢀCH ), 3.84 (dq, 1 H, J
1.27 (d, 3 H, J5%B,6%B 6.3 Hz, H-6B%), 1.25 (d, 3 H, J
6.3 Hz, H-6A%), 1.12 (d, 3 H, H-6B); C NMR (150
MHz, CDCl ): l 168.03 and 166.43 (2×CꢀO, NPhth),
166.13–164.65 (8 C, 8×CꢀO, OBz), 133.61–122.61 (55
C, Ar, CꢀCH ), 117.75 (CꢀCH ), 100.42 (C-1B%), 99.77
(C-1C), 99.69 (C-1A%) 98.19 (C-1B), 80.32 (C-2B), 77.60
(C-3B), 73.97 (C-3A%), 73.64 (C-4A%) 72.63 (C-2A%),
72.47 (C-4B), 72.34 (C-3B%), 72.10 (C-5C), 71.51 (C-
4B%), 70.96, (C-3C), 69.79 (C-4C), 69.02 (C-2B%), 68.15
(OCH CꢀCH ), 67.69 (C-5B%), 67.14 (C-5A), 66.27 (C-
2
2
5%A,6%A
13
(
4
dd, 1 H, H-3B), 4.17 (dq, 1 H, J5A%,6A% 6.2 Hz, H-5A%),
.14–4.08 (m, 1 H, OCH CHꢀCH ), 3.98–3.92 (m, 1 H,
2
2
3
OCH CHꢀCH ), 3.85 (dq, 1 H, J
1
3
6.2 Hz, H-5B),
2
2
5B,6B
.84 (s, 3 H, CH , OAc), 1.33 (d, 3 H, H-6B%), 1.23 (d,
3
2
2
13
H, H-6A%), 1.13 (d, 3 H, H-6B); C NMR (100 MHz,
CDCl ): l 169.05 (CꢀO, OAc), 168.08 and 166.04 (2×
CꢀO, NPhth), 166.13–164.49 (8 C, 8×CꢀO, OBz),
1
9
3
33.61–122.83 (55 C, Ar, CꢀCH ), 117.71 (CꢀCH ),
2
2
9.77 and 99.72 (C-1C and C-1A%), 99.10 (C-1B%), 98.30
2
2
(
C-1B), 80.08 (C-2B), 77.53 (C-3B), 73.94 (2 C, C-4A
and C-3A%), 73.33 (C-2A), 72.47 (C-4B), 72.07 (C-5C),
1.77 (C-4B%), 71.14 (C-3C), 69.74 (3 C, C-2B%, C-3B%
and C-4C), 68.14 (OCH CꢀCH ), 67.59 (C-5B%), 67.04
5B), 62.93 (C-6C), 54.40 (C-2C), 17.69, 17.46 and 17.41
(3×CH , C-6A%, C-6B%, C-6B). MALDI MS Calcd for
3
7
C H NNaO : m/z 1643.6; found m/z 1643.3. Anal.
91
81
27
Calcd for C H NO : C, 67.44; H, 5.04; N, 0.86.
Found: C, 67.65; H, 5.02; N, 0.74.
2
2
91 81
27
(
C-5A), 66.29 (C-5B), 62.82 (C-6C), 54.40 (C-2C), 20.43

2
034
C. H o¨ o¨ g et al. / Carbohydrate Research 337 (2002) 2023–2036
Allyl 2-O-acetyl-3,4-di-O-benzoyl-a-
osyl-(1“2)-3,4-di-O-benzoyl-h- -rhamnopyranosyl-(1
3)-2,4-di-O-benzoyl-h- -rhamnopyranosyl-(1“2)-
3,4,6-tri-O-benzoyl-2-deoxy-2-phthalimido-i- -gluco-
pyranosyl-(1“3)]-4-O-benzoyl-h- -rhamnopyranoside
15).—Glycosylation of the tetrasaccharide 14 (1.26 g,
L
-rhamnopyran-
4-O-acetyl-h-L-rhamnopyranoside (16).—The protected
pentasaccharide 15 (1.20 g, 0.595 mmol) was dissolved
L
“
[
L
in CH Cl (5 mL) and MeOH (50 mL). The mixture
2
2
D
was stirred, and a solution of 1 M NaOMe in MeOH
(2.5 mL) was added. After 10 h at rt, an additional
aliquot of NaOMe–MeOH solution (1 mL) was added,
L
(
0
0
.777 mmol) with the trichloroacetimidate 12 (0.522 g,
.934 mmol) using the procedure previously described
and the reaction was kept under N for a further 36 h
2
at rt. Trituration with hexanes (ꢀ20 mL) produced a
white solid that was collected by filtration and washed
thoroughly with hexanes (2×30 mL). The dry solid
was dissolved in a mixture of EtOH (50 mL) and
for the preparation of 13 yielded the protected pen-
tasaccharide 15 as a colorless foam (1.38 g, 88%) after
purification by column chromatography (9:1 toluene–
2
0
1
EtOAc). [h]D +63° (c 0.74, CHCl ). H NMR (400
ethylenediamine (5 mL) and refluxed under N for 24 h.
3
2
MHz, CDCl ): l 8.45–6.90 (m, 54 H, Ar), 6.25 (dd, 1
The solvents were removed, and toluene (2×50 mL)
3
H, J2C,3C 10.5, J3C,4C 9.1 Hz, H-3C), 6.02 (dd, 1 H,
J4C,5C 10.1 Hz, H-4C), 5.85–5.75 (m, 1 H, CHꢀCH2),
was added and evaporated. The residue was treated
with Ac O (5 mL) and MeOH (50 mL) for 2 h at rt and
2
5
.74 (d, 1 H, J1C,2C 8.1 Hz, H-1C), 5.72 (d, 1 H, H-1A%),
.71 (dd, 1 H, J2B%,3B% 3.6, J3B%,4B% 10.0 Hz, H-3B%), 5.70
concentrated to give crude deprotected pentasaccharide
10. Purification by flash chromatography (6:3:1
5
(
1
dd, 1 H, J3A,4A 10.0 Hz, H-3A), 5.66 (dd, 1 H, J1A%,2A%
.6 Hz, H-2A%), 5.64 (t, 1 H, J3A%,4A%=J4A%,5A%=9.8 Hz,
H-4A%), 5.58 (t, 1 H, J3A,4A=J4A,5A=9.8 Hz, H-4A),
EtOAc–MeOH–H O) was not successful at completely
2
removing salts and aromatic impurities as evidenced by
1
H NMR, so the impure product was acetylated to
5
1
.50 (dd, 1 H, J1A,2A 2.0, J2A,3A 3.4 Hz, H-2A), 5.34 (dd,
H, H-2C), 5.34 (t, 1 H, J3B%,4B%=J4B%,5B%=9.7 Hz,
facilitate purification. After treatment with Ac O (15
2
mL), pyridine (15 mL) and DMAP (50 mg) for 3 h at
45 °C, the reagents were removed in vacuo, and the
residue was dissolved in EtOAc (75 mL) and washed
with water (30 mL), 1 M aq HCl (10 mL), satd
H-4B%), 5.33 (d, 1 H, J1B%,2B% 1.5 Hz, H-1B%), 5.31–5.25
(
(
m, 1 H, CHꢀCH ), 5.21–5.17 (m, 1 H, CHꢀCH ), 5.20
t, 1 H, J3B,4B=J4B,5B=9.8 Hz, H-4B), 4.80 (d, 1 H,
2
2
J1B,2B 1.3 Hz, H-1B), 4.70 (dd, 1 H, J6C,6%C 12.1, J5C,6C
NaHCO solution (30 mL) and satd NaCl solution (10
3
2.8 Hz, H-6C), 4.67 (dd, 1 H, J2A%,3A% 3.0 Hz, H-3A%),
4.63 (d, 1 H, J1A,2A 1.8 Hz, H-1A), 4.49 (dd, 1 H, J5C,6C
5.5 Hz, H-6%C), 4.40 (dd, 1 H, J2B,3B 3.5 Hz, H-2B), 4.32
mL). Purification by flash chromatography (EtOAc)
gave the acetylated pentasaccharide 16 as a colorless
20
1
foam (0.403 g, 51%). [h]D −50° (c 0.88, CHCl ). H
3
(
dq, 1 H, H-5A), 4.29 (ddd, 1 H, H-5C), 4.23 (dd, 1 H,
NMR (400 MHz, CDCl ): l 6.50 (br d, 1 H J
6.9
3
NH,C2
J3B,4B 9.7 Hz, H-3B), 4.20–4.08 (m, 3 H, 2B%, H-5A%,
OCH CHꢀCH ), 4.02 (dq, 1 H, H-5B%), 3.98–3.93 (m, 1
Hz, NHAc), 5.97 (dd, 1 H, J2C,3C 10.7, J3C,4C 9.1 Hz,
H-3C), 5.94–5.83 (m, 1 H, CHꢀCH ), 5.37 (d, 1 H,
2
2
2
H, OCH CHꢀCH ), 3.87 (dq, 1 H, H-5B), 1.97 (s, 3 H,
J
1C,2C
8.1 Hz, H-1C), 5.34 (dd, 1 H, J1A%,2A% 1.9, J2A%,3A%
2
2
CH , OAc), 1.41 (d, 3 H, J
6.2 Hz, H-6A), 1.18 (d,
3.4 Hz, H-2A%), 5.32 (dd, 1 H, J2A,3A 3.4, J3A,4A 10.0
3
5A,6A
3
H, J5A%,6A% 6.3 Hz, H-6A%), 1.13 (d, 3 H, J
6.3 Hz,
Hz, H-3A), 5.32–5.26 (m, 1 H, CHꢀCH ), 5.26 (dd, 1
5B,6B
1
2
3
H-6B), 0.92 (d, 3 H, J5B%,6B% 6.2 Hz, H-6B%); C NMR
100 MHz, CDCl ): l 169.11 (CꢀO, OAc), 168.13 and
H, J1A,2A 1.8 Hz, H-2A), 5.23–5.19 (m, 1 H, CHꢀCH2),
5.12–5.03 (m, 5 H, H-1A%, H-4A%, H-3B%, H-4B%, H-
4A), 5.00 (t, 1 H, J3B,4B=J4B,5B=9.9 Hz, H-4B), 4.97
(d, 1 H, J1B%,2B% 1.7 Hz, H-1B%), 4.90 (dd, 1 H, J4C,5C 10.1
Hz, H-4C), 4.76 (d, 1 H, J1A,2A 1.7 Hz, H-1A), 4.72 (d,
1 H, J1B,2B 1.4 Hz, H-1B), 4.23 (dd, 1 H, J6C,6%C 12.2,
J5C,6C 4.6 Hz, H-6C), 4.19 (dd, 1 H, J2A%,3A% 3.4, J3A%,4A%
9.9 Hz, H-3A%), 4.17–4.13 (m, 1 H, OCH CHꢀCH ),
(
1
3
65.85 (2×CꢀO, NPhth), 166.11–164.42 (10 C, 10×
CꢀO, OBz), 133.55–122.52 (67 C, Ar, CꢀCH ), 117.73
2
(
CꢀCH ), 99.93 (C-1B%), 99.78 (C-1C), 99.65 (C-1A%)
2
98.94 (C-1A), 98.26 (C-1B), 79.61 (C-2B), 77.49 (C-3B),
76.58 (C-2B%), 74.15 (C-4A%), 73.53 (C-3A%), 73.25 (C-
2A%), 72.37 (C-4B), 71.98 (C-5C), 71.94 (A-4), 71.65
2
2
(
C-4B%), 71.09 (2 C, C-3A and C-3C), 69.72 (2 C, C-2A
4.14 (dd, 1 H, J3B,4B 9.7 H-3B), 4.07 (dd, 1 H, J5C,6C 2.5
Hz, H-6%C), 4.02 (dq, 1 H, H-5B%), 4.01–3.95 (m, 1 H,
and C-4C), 69.59 (C-3B%), 68.10 (OCH CꢀCH ), 67.94
2
2
(
C-5A), 67.41 (C-5B%), 66.86 (C-5A%), 66.32 (C-5B),
OCH CHꢀCH ), 4.00 (dd, 1 H, J
3.3 Hz, H-2B),
2
2
2B,3B
6
1
2.81 (C-6C), 54.37 (C-2C), 20.62 (CH , OAc), 17.61,
3.91 (dd, 1 H H-2B), 3.89 (dq, 1 H, H-5A%), 3.83 (m, 1
H H-5A), 3.76 (ddd, 1 H, H-5C), 3.73 (dq, 1 H, H-5B),
2.74 (ddd, 1 H, H-2C), 2.25–1.87 (12 s, 36 H, 12×
3
7.53, 17.41 and 17.20 (4×CH , C-6A, C-6A%, C-6B%,
3
C-6B). MALDI MS Calcd for C113H101NNaO : m/z
34
2040.0; found m/z 2040.3. Anal. Calcd for
CH , 12 OAc), 1.24 (d, 3 H, J
6.1 Hz, H-6A), 1.22
3
5A,6A
C113H101NO : C, 67.29; H, 5.05; N, 0.69. Found: C,
(d, 3 H, J5B%,6B% 6.5 Hz, H-6B%), 1.17 (d, 3 H, J5A%,6A% 6.5
34
1
3
67.31; H, 5.06; N, 0.72.
Hz, H-6A%), 1.15 (d, 3 H, J5B,6B 6.4 Hz, H-6B);
C
Allyl 2,3,4-tri-O-acetyl-h-
L-rhamnopyranosyl-(1“2)-
NMR (100 MHz, CDCl ): l 171.61–169.50 (12 C,
3
3
,4-di-O-acetyl-h-
- rhamnopyranosyl - (1“2)-[3,4,6-tri-O-ace-
tyl-2-deoxy- 2-acetamido-i- -glucopyranosyl-(1“3)]-
L
-rhamnopyranosyl-(1“3)-2,4-di-O-
12×CꢀO, NHAc and 11×OAc), 133.51 (CꢀCH2),
acetyl - h -
L
117.70 (CꢀCH ), 99.56 (2 C, C-1A and C-1B%), 99.24
2
D
(C-1A%), 98.36 (C-1C), 98.13 (C-1B), 78.40 (C-2B),

C. H o¨ o¨ g et al. / Carbohydrate Research 337 (2002) 2023–2036
2035
7
7
7
7.48 (C-2B%), 74.32 (C-3B), 73.25 (C-3A%), 72.84 (C-4),
2.51 (C-4B), 71.42 (2 C, C-2A% and C-5C), 71.04 (C-4),
0.89 (C-4), 70.21 (C-3B%), 70.07, (C-3C), 69.90 (2 C,
3B), 78.70 (C-2B%), 77.25 (C-2B), 77.19 (C-3A%), 76.40
(C-5C), 74.51 (C-3C), 72.76 (C-4B%), 72.63 (C-4A),
72.37 (C-4A%), 71.86 (C-4B), 70.69 (C-3A), 70.64 (C-
3B%), 70.58 (C-2A), 70.41 (2 C, C-2A% and C-4C), 69.91
(C-5B), 69.73 (3 C, C-5A, C-5B% and C-5A%), 68.82
(OCH CꢀCH ), 61.43 (C-6C), 56.49 (C-2C), 17.44,
C-2A, C-4C), 68.59 (C-3A), 68.01 (OCH CꢀCH ),
2
2
6
5
1
7.60 (C-5A), 67.20 (C-5B%), 67.00 (C-5A%), 66.49 (C-
B), 62.38 (C-6C), 57.75 (C-2C), 23.14–20.70 (12 C,
2
2
2×CH , NHAc and 11×OAc), 17.51, 17.38 (2 C),
17.23 (2 C), 17.14 (4 C, 4×CH , C-6A, C-6A%, C-6B%,
3
3
and 17.30 (4×CH , C-6A, C-6A%, C-6B%, C-6B).
C-6B). MALDI MS Calcd for C H NNaO : m/z
3
35 59
22
MALDI MS Calcd for C H NO : m/z 1330.5; found
868.8; found m/z 868.5. Anal. Calcd for C H NO :
5
7
81
33
35 59 22
m/z 1330.9. Anal. Calcd for C H NNaO : C, 52.33;
C, 49.70; H, 7.03; N, 1.66. Found: C, 49.40; H, 7.28; N,
1.65.
5
7
81
33
H, 6.24; N, 1.07. Found: C, 52.60; H, 6.19; N, 1.00.
Allyl h- -rhamnopyranosyl-(1“2)-h- -rhamnopyran-
osyl-(1“3)-h- -rhamnopyranosyl-(1“2)-[2-deoxy-2-
acetamido-i- -glucopyranosyl-(1“3)]-h- -rhamnopy-
ranoside (10).—The acetylated pentasaccharide 16
0.322 g, 0.246 mmol) was dissolved in dry MeOH (20
L
L
L
4
. Conclusions
D
L
The efficient synthesis of a pentasaccharide 10, corre-
(
sponding to a portion of the cell-wall polysaccharide of
the Group A Streptococcus, has been accomplished. A
sufficient number of experimental inter-glycosidic NOE
contacts for every linkage in the pentasaccharide 10
made the determination of its conformation possible.
The pentasaccharide 10 was found to populate one
major conformation with an overall extended shape.
The conformation calculated by MMC simulations with
the HSEA force field showed the best fit with experimen-
tal NOE data, as in previous investigations of the
conformations of oligosaccharides. It can also be
concluded that the CVFF force field gives better agree-
ment with experiment at the terminal residues, whereas
the CHARMM-based force field, PARM22, gives more
satisfactory results for the more rigid, central part of
the molecule.
mL) and placed under N . A 1.0 M solution of NaOMe
2
in MeOH (1.0 mL) was added, and the mixture was
kept at rt for 3.5 h. A second aliquot of the NaOMe–
MeOH solution (1.0 mL) was then added, and the
reaction was allowed to continue for a further 2 h. The
base was neutralized by the addition of an excess of
+
Rexyn 101 (H ) resin, and the mixture was filtered and
concentrated to a syrupy residue that was purified by
flash chromatography (6:3:1 EtOAc–MeOH–H O) to
2
give a colorless solid. The solid was dissolved in water
47
(
(
(
4 mL) and freeze-dried to give the pentasaccharide 10
2
0
0.188 g, 88%) as a fluffy amorphous solid. [h] −76°
D
1
c 0.82, MeOH). H NMR (600 MHz, D O): l 6.00–
2
5
5
.93 (m, 1 H, CHꢀCH ), 5.39–5.35 (m, 1 H, CHꢀCH ),
2
2
.33–5.30 (m, 1 H, CHꢀCH ), 5.20 (d, 1 H, J
1.4
2
1B%,2B%
Hz, H-1B%), 5.18 (d, 1 H, J1A%,2A% 1.7 Hz, H-1A%), 4.97
(
d, 1 H, J1A,2A 1.7 Hz, H-1A), 4.87 (d, 1 H, J1B,2B 1.7
Hz, H-1B), 4.69 (d, 1 H, J1C,2C 8.4 Hz, H-1C), 4.26–
.22 (m, 1 H, OCH CHꢀCH ), 4.18 (dd, 1 H, J 3.0
Acknowledgements
4
2
2
2B,3B
Hz, H-2B), 4.11–4.07 (m, 1 H, OCH CHꢀCH ), 4.09
We are grateful to the Natural Sciences and Engi-
neering Research Council of Canada for financial sup-
port of this work and the Wenner–Gren Foundation,
Sweden, for a postdoctoral fellowship (to C.H.).
2
2
(
2×dd, 2 H, J2A,3A 3.4, J2A%,3A% 3.2 Hz, H-2A and
H-2A%), 4.07 (dd, 1 H, J2B%,3B% 3.4 Hz, H-2B%), 3.96 (dd,
H, J3B%,4B% 9.9 Hz, H-3B%), 3.95–3.91 (2nd order m, 1
H, H-6C), 3.87 (dd, 1 H, J3B,4B 9.8 Hz, H-3B), 3.84 (dd,
H, J3B%,4B% 9.9 Hz, H-3A%), 3.82 (dq, 1 H, H-5B%), 3.81
dd, 1 H, J3A,4A 9.8 Hz, H-3A), 3.79–3.75 (2nd order
1
1
(
References
m, 1 H, H-6%C), 3.74 (dq, 1 H, H-5B), 3.73 (dd, 1 H,
J2C,3C 10.2, H-2C), 3.73 (dq, 1 H, H-5A%), 3.71 (dq, 1 H,
H-5A), 3.58 (dd, 1 H, J4A%,5A% 9.7 Hz, H-4A%), 3.57–3.52
1
. (a) Streptococcal Diseases and the Immune Response;
Read, S. E.; Zabriskie, J. B., Eds.; Academic Press: New
York, 1980;
(
2nd order m, 1 H, H-3C), 3.52 (dd, 1 H, J4B,5B 9.7 Hz,
H-4B), 3.50 (dd, 1 H, J4B%,5B% 9.8 Hz, H-4B%), 3.48–3.44
2nd order m, 2 H, H-4C and H-5C), 3.45 (dd, 1 H,
J4A,5A 9.8 Hz, H-4A), 2.03 (s, 3 H, CH , NHOAc), 1.33
(b) Weiss, K. A.; Laverdiere, M. Can. J. Surg. 1997, 40,
18–25;
(
(
c) Stevens, D. L. Immunol. In6est. 1997, 26, 129–150;
d) Nowak, R. Science 1994, 264, 1665.
(
3
2
. Bisno, A. L. Nonsuppurative Poststreptococcal Sequelae:
Rheumatic Fever and Glomerulonephritis. In Principles
and Practice of Infectious Diseases; Mandell, G. L.; Dou-
glas, R. G.; Bennett, J. E., Eds., 2nd ed.; Wiley: New
York, 1985; pp 1133–1142.
. (a) World Health Organization: Community Control of
Rheumatic Heart Disease in Developing Countries. 1. A
Major Public Health Problem, WHO Chronicle, 1980, 34,
336–345;
(
d, 3 H, J5B%,6B% 6.3 Hz, 6B%-CH ), 1.30 (d, 3 H, J
6.2
3
5B,6B
Hz, 6B-CH ), 1.28 (d, 3 H, J
1
MHz, D O): l 175.09 (CꢀO, NHAc), 133.68 (CꢀCH ),
1
(
1
6.3 Hz, 6A%-CH ),
.27 (d, 3 H, J5A,6A 6.2 Hz, 6A-CH3); C NMR (100
3
5A%,6A%
3
1
3
2
2
3
1
19.37 (CꢀCH ), 103.22 ( J
159 Hz, C-1C), 102.93
173 Hz, C-1A%),
178 Hz, C-1B%), 98.17 (C-1B), 80.56 (C-
2
C–H
1
1
JC–H 173 Hz, C-1A), 101.99 ( J
C–H
1
01.31 ( J
C–H

2
036
C. H o¨ o¨ g et al. / Carbohydrate Research 337 (2002) 2023–2036
(
b) Markowitz, M. Clin. Therapeutics 1981, 4, 240–251.
710–720.
4
5
6
7
. Wannamaker, L. W. Re6. Infect. Dis. 1979, 1, 967–975.
. Zabriskie, J. B. Circulation 1985, 71, 1077–1086.
. Wright, K. Science 1990, 249, 22–24.
. Rotta, J. Streptococci: Prospects for Vaccines. In To-
wards Better Carbohydrate Vaccines; Bell, R.; Torrigiani,
G., Eds.; John Wiley & Sons: New York, 1987; pp
26. Weimar, T.; Harris, S. L.; Pitner, J. B.; Bock, K.; Pinto,
B. M. Biochemistry 1995, 34, 13672–13681.
27. Wang, W.; Kong, F. J. Carbohydr. Chem. 1999, 18,
263–273.
28. Bock, K.; Pedersen, C. J. Chem. Soc., Perkin Trans. 2
1974, 293–297.
29. Lemieux, R. U.; Koto, S. Tetrahedron 1974, 30, 1933–
1944.
30. Peters, T.; Meyer, B.; Stuike-Prill, R.; Somorjai, R.;
Brisson, J.-R. Carbohydr. Res. 1993, 238, 49–73.
31. Bukowski, R.; Morris, L. M.; Woods, R. J.; Weimar, T.
Eur. J. Org. Chem. 2001, 2697–2705.
32. H o¨ o¨ g, C.; Landersj o¨ , C.; Widmalm, G. Chem. Eur. J.
2001, 7, 3069–3076.
33. Leeflang, B. R.; Kroon-Batenburg, L. M. J. J. Biomol.
NMR 1992, 2, 495–518.
34. Diaz-Quijada, G. A.; Weinberg, N.; Holdcroft, S.; Pinto,
B. M. J. Phys. Chem. A 2002, 106, 1277–1285.
35. Rundl o¨ f, T.; Venable, R. M.; Pastor, R. W.; Kowalewski,
J.; Widmalm, G. J. Am. Chem. Soc. 1999, 121, 11847–
11854.
36. Brooks, C. L., III; Karplus, M.; Pettitt, B. M. Ad6ances
in Chemical Physics, Vol. 71: Proteins: A Theoretical
Perspecti6e of Dynamics, Structure and Thermodynamics;
John Wiley & Sons: New York, 1988.
2
03–218.
8
. Braun, D. G.; Schalch, W.; Schmid, I. The Restricted
Antibody Response to Streptococcal Group Antigens in
Animals and Man: A Complex Trait. In Streptococcal
Diseases and the Immune Response; Read, S. E.;
Zabriskie, J. B., Eds.; Academic Press: New York, 1980;
pp 317–333.
. Emmrich, F.; Schilling, B.; Eichmann, K. J. Exp. Med
1
0. Blake, M. S.; Zabriskie, J. B.; Tai, J. Y.; Michon, F. US
Patent 5,866,135 (1999).
1. Benaissa-Trouw, B.; Lefeber, D. J.; Kamerling, J. P.;
Vliegenthart, J. F. G.; Kraaijeveld, K.; Snippe, H. Infect.
Immun. 2001, 69, 4698–4701.
9
985, 161, 547–562.
1
1
1
1
2. Pozsgay, V. Ad6. Chem. Biochem. 2001, 56, 153–199.
3. Coligan, J. E.; Kindt, T. J.; Krause, R. M. Immunochem-
istry 1978, 15, 755–760.
1
4. Huang, D. H.; Krishna, N. R.; Pritchard, D. G. Carbo-
hydr. Res. 1986, 155, 193–199.
1
1
5. Pinto, B. M. ACS Symp. Ser. 1992, 519, 111–131.
6. Marino-Albernas, J.-R.; Harris, S. L.; Varma, V.; Pinto,
B. M. Carbohydr. Res. 1993, 245, 245–257.
37. Keepers, J. W.; James, T. L. J. Magn. Reson. 1984, 57,
404–426.
38. Widmalm, G.; Byrd, R. A.; Egan, W. Carbohydr. Res.
1992, 229, 195–211.
39. Stuike-Prill, R.; Meyer, B. Eur. J. Biochem. 1990, 194,
903–919.
40. Thøgersen, H.; Lemieux, R. U.; Bock, K.; Meyer, B. Can.
J. Chem. 1982, 60, 44–57.
41. Brooks, B. R.; Bruccoleri, R. E.; Olafson, B. D.; States,
D. J.; Swaminathan, S.; Karplus, M. J. Comput. Chem.
1983, 4, 187–217.
42. Jorgensen, W. L.; Chandrasekhar, J.; Madura, J. D.;
Impey, R. W.; Klein, M. L. J. Chem. Phys. 1983, 79,
926–935.
43. Berendsen, H. J. C.; Postma, J. P. M.; van Gunsteren, W.
F.; DiNola, A.; Haak, J. R. J. Chem. Phys. 1984, 81,
3684–3690.
44. Steinbach, P. J.; Brooks, B. R. J. Comput. Chem. 1994,
15, 667–683.
1
1
1
2
7. Auzanneau, F.-I.; Forooghian, F.; Pinto, B. M. Carbo-
hydr. Res. 1996, 291, 21–41.
8. Reimer, K. B.; Gidney, M. A. J.; Bundle, D. R.; Pinto, B.
M. Carbohydr. Res. 1992, 232, 131–142.
9. Auzanneau, F.-I.; Pinto, B. M. Bioorg. Med. Chem. 1996,
4, 2003–2010.
0. Auzanneau, F.-I.; Christensen, M. K.; Harris, S. L.;
Meldal, M.; Pinto, B. M. Can. J. Chem. 1998, 76, 1109–
1
118.
2
2
1. Harris, S. L.; Craig, L.; Mehroke, J. S.; Rashed, M.;
Zwick, M. B.; Kenar, K.; Toone, E. J.; Greenspan, N.;
Auzanneau, F.-I.; Marino-Albernas, J.-R.; Pinto, B. M.;
Scott, J. K. Proc. Natl. Acad. Sci. USA 1997, 94, 2454–
2
459.
2. Pitner, J. B.; Beyer, W. F.; Venetta, T. M.; Nycz, C.;
Mitchell, M. J.; Harris, S. L.; Marino-Albernas, J.-R.;
Auzanneau, F.-I.; Forooghian, F.; Pinto, B. M. Carbo-
hydr. Res. 2000, 324, 17–29.
45. van Gunsteren, W. F.; Berendsen, H. J. C. Mol. Phys.
1977, 34, 1311–1327.
46. Dauber-Osguthorpe, P.; Roberts, V. A.; Osguthorpe, D.
J.; Wolff, J.; Genest, M.; Hagler, A. T. Proteins: Struc-
ture Function and Genetics 1988, 4, 31–47.
47. H o¨ o¨ g, C.; Widmalm, G. Arch. Biochem. Biophys. 2000,
377, 163–170.
2
2
2
3. Kreis, U. C.; Varma, V.; Pinto, B. M. Int. J. Biol.
Macromol. 1995, 17, 117–130.
4. Stuike-Prill, R.; Pinto, B. M. Carbohydr. Res. 1995, 279,
5
9–73.
5. Peters, T.; Pinto, B. M. Curr. Opin. Struct. Biol. 1996, 6,