The core oligosaccharide component from mannheimia (pasteurella) haemolytica serotype A1 lipopolysaccharide contains L-glycero-D-manno-and D-glycero-D-manno-heptoses: Analysis of the structure and conformation by high-resolution NMR spectroscopy

Article abstract of DOI:10.1139/v02-114

Previous studies from our laboratory have indicated that the lipopolysaccharide (LPS) from Mannheimia haemolytica serotype A1 contains both L-glycero-D-manno-heptose and D-glycero-D-manno-heptose residues. NMR methods making use of 1D ¹H selective excitation and 2D (¹H, ¹³C) and (¹H, ³¹P) heteronuclear experiments were used for the structural determination of the major core oligosaccharide components of the deacylated low-molecular-mass LPS obtained following sequential treatment with anhydrous hydrazine and aq KOH. The core oligosaccharide region was found to be composed of a branched octasaccharide linked to the deacylated lipid A moiety via a 3-deoxy-4-phospho-D-manno-oct-2-ulosonate residue having the structure, α -D-Glc p-(1↓6 β-D -Galp-(1-7)-D -α-D -Hepp-(1-6)-D -α-D -Hepp-(1-6)-β-D -Glcp-(1-4)-L -α-D -Hepp-(1→3↑L -α-D -Hepp-(1-2)-L -α-D -Hepp-(1 Heterogeneity was found to be present at several linkages. NMR methods were devised to distinguish between the diastereomeric forms of the heptose residues. Synthesized monosaccharides of L-D- and D-D-heptose were used as model compounds for analysis of the ¹H and ¹³C NMR chemical shifts and proton coupling constants. Molecular modeling using a Monte Carlo method for conformational analysis of saccharides was used to determine the conformation of the inner core of the oligosaccharide and to establish the stereochemical relationships between the heptoses.

Full text of DOI:10.1139/v02-114

949
The core oligosaccharide component from
Mannheimia (Pasteurella) haemolytica serotype A1
lipopolysaccharide contains L-glycero-D-manno-
and D-glycero-D-manno-heptoses: Analysis of the
structure and conformation by high-resolution
NMR spectroscopy
Jean-Robert Brisson, Ellen Crawford, Dušan Uhrín, Nam Huan Khieu,
Malcolm B. Perry, Wayne B. Severn, and James C. Richards
Abstract: Previous studies from our laboratory have indicated that the lipopolysaccharide (LPS) from Mannheimia
haemolytica serotype A1 contains both L-glycero-D-manno-heptose and D-glycero-D-manno-heptose residues. NMR
1
methods making use of 1D H selective excitation and 2D (1H, 13C) and (1H, 31P) heteronuclear experiments were used
for the structural determination of the major core oligosaccharide components of the deacylated low-molecular-mass
LPS obtained following sequential treatment with anhydrous hydrazine and aq KOH. The core oligosaccharide region
was found to be composed of a branched octasaccharide linked to the deacylated lipid A moiety via a 3-deoxy-4-
phospho-D-manno-oct-2-ulosonate residue having the structure,
α-D-Glc p-(1
6
-(1
β-D
-Galp-(1-7)-
D
-α-D
-Hepp-(1-6)-
D
-α-D
-Hepp-(1-6)-β-
D
-Glcp-(1-4)-
L
-α-D
-Hepp
3
L
-α-D -Hepp-(1-2)-L -α-D -Hepp-(1
Heterogeneity was found to be present at several linkages. NMR methods were devised to distinguish between the
diastereomeric forms of the heptose residues. Synthesized monosaccharides of L-D- and D-D-heptose were used as model
1
compounds for analysis of the H and 13C NMR chemical shifts and proton coupling constants. Molecular modeling
using a Monte Carlo method for conformational analysis of saccharides was used to determine the conformation of the
inner core of the oligosaccharide and to establish the stereochemical relationships between the heptoses.
Key words: LPS, NMR, conformation, oligosaccharide, heptose. 963
Résumé : Des études antérieures effectuées dans notre laboratoire ont indiqué que le lipopolysaccharide (LPS) obtenu
à partir de Mannheimia haemolytica sérotype A1 contient des résidus de L-glycéro-D-manno-heptose ainsi que de D-
1
glycéro-D-manno-heptose. On a utilisé des méthodes de RMN faisant appel à l’excitation selective du H en 1D et à
des expériences hétéronucléaires (1H,13C) et (1H,31P) en 2D pour déterminer la structure des composants oligosaccha-
rides principaux du noyau du LPS désacétylé, de faible masse moléculaire, obtenu après un traitement séquentiel avec
de l’hydrazine anhydre et du KOH aqueux. On a trouvé que l’oligosaccharide de la région du noyau est formée d’un
Received 10 January 2002. Published on the NRC Research Press Web site at http://canjchem.nrc.ca on 5 August 2002.
Dedicated to the memory of Professor Raymond U. Lemieux who pioneered the applications of NMR spectroscopy for
stereochemical and conformational analyses of carbohydrates.
J.-R. Brisson, E. Crawford, N.H. Khieu, M.B. Perry, and J.C. Richards.1 Institute for Biological Sciences, National Research
Council, Ottawa, ON K1A 0R6, Canada.
W.B. Severn. Agricultural Research, Department of Diseases, PO Box 40063, Ward Street, Upper Hutt, New Zealand.
D. Uhrín. University of Edinburgh, Department of Chemistry, West Mains Road, Edinburgh EH9 3JJ, U.K.
1Corresponding author (e-mail: jim.richards@nrc.ca).
Can. J. Chem. 80: 949–963 (2002)
DOI: 10.1139/V02-114
© 2002 NRC Canada

950
Can. J. Chem. Vol. 80, 2002
oligosaccharide ramifié lié à une portion désacétylée du lipide A par le biais d’un résidu 3-désoxy-4-phospho-D-manno-
oct-2-ulosonate de structure,
α-D-Glc p-(1
6
-(1
β-D
-Galp-(1-7)-
D
-α-D
-Hepp-(1-6)-
D
-α-D
-Hepp-(1-6)-β-
D
-Glcp-(1-4)-
L
-α-D
-Hepp
3
L
-α-D -Hepp-(1-2)-L -α-D -Hepp-(1
On a observé de l’hétérogénéité à plusieurs points de ramification. On a mis au point des méthodes RMN qui
permettent de faire la différence entre les formes diastéréomères des résidus heptoses. On a utilisé des monosaccharides
1
de synthèse de L-D- et D-D-heptoses comme modèles pour l’analyse des déplacements chimiques du H et du 13C et des
constantes de couplage du proton. On a utilisé la modélisation moléculaire basée sur la méthode de Monte Carlo pour
faire l’analyse conformationnelle des saccharides afin déterminer la conformation du noyau de l’oligosaccharide et
établir les relations stéréochimiques entre les heptoses.
Mots clés : LPS, RMN, conformation, oligosaccharide, heptose.
[Traduit par la Rédaction] Brisson et al.
Introduction
A1 is most frequently associated with bovine pneumonic
pasteurellosis (shipping fever), a major cause of sickness
and economic loss to the feed lot industry (11, 12). The
pathogenesis of diseases caused by M. haemolytica in cattle has
recently been reviewed (13). Several virulence factors, in-
cluding a leukotoxin, capsular polysaccharide, and LPS have
been identified. M. haemolytica LPS has been reported to
have a severe damaging effect on bovine endothelial cells
(14, 15).
In our laboratory we have undertaken the structural and
antigenic characterization of M. haemolytica (16, 17) and
P. trehalos (17–20) LPS to understand the role of these mol-
ecules in pathogenesis and protective immunity. Immuno-
chemical studies have revealed that eight of the 12 M.
haemolytica serotypes can elaborate LPS which contain a
common O-polysaccharide side chain (17). The structure of
the O-polysaccharide from serotype A1 is representative and
was determined in detail. It consists of a trisaccharide re-
peating unit having the structure -4)-β-D-Galp-(1–3)-β-D-
Galp-(1–3)-β-D-GalpNAc-(1- (16). A detailed structural
model for the core oligosaccharide of M. haemolytica LPS
serotype A1 core oligosaccharide has not been proposed.
Many carbohydrate samples of biological origin are heter-
ogeneous in nature and difficult to separate. To determine
their three-dimensional structure, complete NMR assign-
ments are desirable even for heterogeneous products. Only
then can conformational analysis (21) be made with the goal
of establishing a relationship between three-dimensional
structure and biological function (22). Towards this goal, 1D
selective NMR methods have been used to determine the
structure of mixtures, where the resonances of specific com-
ponents can be selected to characterize individual compo-
nents (23–28). 1D selective excitation experiments are
analogs of standard 2D and 3D experiments where hard
pulses have been replaced by selective pulses. The original
phase-cycled experiments (29–42) have more recently been
adopted for the use of pulsed-field gradients (43–55). Partic-
ularly useful are 1D analogues of 3D experiments (53–55).
Lipopolysaccharides (LPS) are a complex class of glyco-
lipids that can trigger a cascade of immunological responses
in mammals including endotoxic effects and serum antibody
production (1). These molecules make up the outer leaflet of
the outer membrane of Gram-negative bacteria. LPS has
been found to exhibit a common molecular architecture con-
sisting of at least two distinct regions: (i) an oligo- or
polysaccharide region; and (ii) a lipid moiety referred to as
lipid A (2). In certain bacteria, the carbohydrate containing
region consists of a high-molecular-mass O-specific poly-
saccharide, which is covalently linked to a core oligo-
saccharide unit (3). Other bacteria, notably Haemophilus and
Neisseria spp. only produce short-chain LPS, devoid of O-
polysaccharide chains, in which the carbohydrate region
contains mixtures of low-molecular-mass but structurally di-
verse oligosaccharide components. The saccharide portion of
LPS molecules has been found to be linked to the lipid A
moieties via the eight-carbon sugar acid, 3-deoxy-D-manno-
oct-2-ulosonate (Kdo). The lipid A from several bacterial
species has been found (2) to have a relatively conserved
structure consisting of a bisphosphorylated β-1,6-linked
D-glucosamine (GlcN) disaccharide unit that is substituted
by fatty acyl amide and ester groups.
The bacterium Mannheimia haemolytica is a new member
of the Pasteurellaceae family (4). Originally referred to as
Pasteurella haemolytica, this group of organisms was classi-
fied into two biotypes (designated A and T) which were fur-
ther subdivided into 17 serotypes (5, 6). This biochemical
distinction was based on the ability of the organisms to fer-
ment either arabinose (A-biotype) or trehalose (T-biotype)
(7). On the basis of genetic analyses, the P. haemolytica
biotype T organisms (serotypes T3, T4, T10, and T15) were
reclassified as Pasteurella trehalos in 1990 (8, 9). More re-
cently the biotype A group of organisms was reclassified
into the new genus Mannheimia (10). Twelve M. haemolytica
serotypes are now recognized (A1, A2, A5, A6, A7, A8, A9,
A12, A13, A14, A16, and A17). M. haemolytica serotype
© 2002 NRC Canada

Brisson et al.
951
Fig. 1. Proton NMR spectrum at 600 MHz of the core oligosaccharide component from M. haemolytica serotype A1 LPS in D2O
(300 K, pH 3). The HOD resonance is at 4.756 ppm. Resonances in the anomeric region up from 4.4 ppm are labelled along with the
Kdo methylene resonances. The determined structure designated as A1 and nomenclature for the residues are also shown. Heterogeneity
due to partial truncation of the backbone oligosaccharide is indicated by dotted arrows at linkage sites. In the text, the minor compo-
nent for f due to heterogeneity is indicated by f′. HepI, HepII, and HepIII are L-D-heptose while HepIV and HepV are D-D-heptoses. Other
residues have the D-configuration.
c
A1
GlcII
α1-6
g
f
i
k
h
a
β1-4
α1-2
β1-6
d
α1-6
α1-6
α1-5
α2-6
HepV
HepIV
GlcNI α PO4
-
HepI
α1-3
GlcI
Kdo
4
GlcNII
4
β1-7
PO4
PO4
Gal
HepIII
HepII
e
b
j
d1
e1
g1
f1
i1
a1
c1
h1
j1
Kdo
k3eq k3ax
k4
b1
HOD
5.5
5.0
4.5
4.0
3.5
3.0
2.5
ppm
An advantage of 1D selective methods is that the multiplet
pattern for each resonance in the 1D selective experiments
can be clearly observed due to the same digital resolution as
a 1D 1H spectrum. Also, 1D selective experiments are some-
times preferable to 2D methods for the structural analysis of
oligosaccharides as was recently demonstrated in the charac-
terization of some enzymatically synthesized oligosaccha-
rides (56, 57). Also, with selective experiments, incursion in
the ring region of carbohydrates is possible. The 1D
TOCSY-NOESY permits the isolation of spin systems of in-
terests and the detection of NOEs from proton resonances
within the ring allowing NOE constraints to be obtained
which cannot be otherwise determined from 2D experiments
(58, 59). Recently several reviews have dealt with carbohy-
drate structural determination by NMR spectroscopy (60–
64).
Preliminary studies by chemical, 1H NMR, and mass
spectrometric (MS)-based methods have indicated that the
LPS core region of M. haemolytica A1 has the composition
Hex2-3Hep5Kdo. The Hex2Hep5Kdo species is the major core
oligosaccharide component in the O-chain deficient serotype
A8 LPS (17). Analyses of the LPS from these serotypes in-
dicate the core oligosaccharide to contain D-glucose, D-galactose
and both L-glycero-D-manno-heptose (L-D-heptose) and D-
glycero-D-manno-heptose (D-D-heptose) residues. In the
present investigation the structure of the core oligo-
saccharide of M. haemolytica serotype A1 LPS was deter-
mined in detail. The sequence, linkage positions, and
relative stereochemistries of the heptose-containing oligo-
saccharide were established by NMR spectroscopic tech-
niques. On the basis of NMR analyses and Monte Carlo
calculations a conformational model is now proposed for the
core oligosaccharide region of the LPS.
Results and discussion
Our previous analysis of the LPS of M. haemolytica
serotype A1 revealed Hex3Hep5Kdo as the major core
oligosaccharide component. This was determined by 1D H
1
NMR and FAB-MS analysis of the oligosaccharide fraction
obtained following mild acid hydrolysis (1% HOAc, 100°C,
3 h) of the LPS sample (17). The core oligosaccharide was
found to contain D-glucose, D-galactose, L-D-heptose, and D-
D-heptose in a molar ratio of 2:1:3:2 in the major fraction as
determined by GLC–MS analysis of their alditol acetate and
2-butyl glycoside derivatives. The D-Gal residue was found
to be a terminal nonreducing moiety from methylation anal-
ysis and was absent in the minor component of the core
oligosaccharide fraction, that had the composition of
Hex2Hep5Kdo. The occurrence of Kdo in the LPS was estab-
lished by colorimetric analysis.
Treatment of M. haemolytica LPS with anhydrous
hydrazine followed by strong alkali afforded water-soluble,
deacylated LPS oligosaccharides. The deacylated LPS sam-
ple was representative of the intact backbone oligo-
saccharide of the native material containing core and lipid A
oligosaccharide moieties. This was confirmed by electro-
spray ionization ESI-MS which gave molecular ions corre-
sponding to Hex3Hep5Kdo1HexN2(H2PO3)3 as the major
oligosaccharide component (see Experimental section).
From the proton spectrum (Fig. 1) and HMQC and COSY
1
spectra (Fig. 2), 10 anomeric H NMR resonances were ob-
served, as well as methylene proton resonances. The
anomeric resonances were labeled a–j in decreasing order of
1
their H NMR chemical shifts and k3eq and k3ax assigned to
the methylene protons of Kdo. The integral for the b1 and
de1 and j1 anomeric peaks were 40% less than those for the
© 2002 NRC Canada

952
Can. J. Chem. Vol. 80, 2002
Fig. 2. 2D NMR experiments for A1 showing the HMQC spectrum of the anomeric region (a), and the COSY (b) and NOESY
(c) spectra of the anomeric and ring region. The arrow indicates the position of the b2 resonance observed near k5.
j1
j2
4.5
4.5
(a)
(b)
i1,k4
k5
HOD
i2
g2
h1
g1
f2
5.0
5.0
h2
f1
de1
F2
F1
d2 e2
c1
c2
5.5
5.5
b2
b1
a1
a2
ppm
ppm
95
100
ppm
4.0
3.5
ppm
F2
F1
j1
4.5
(c)
k5
k6
d4,6
i5,3
k4,i1
h1
h3,5
f2
g2
g1
i6
5.0
f1
g6
d2
de1
c1
F1
d6e2
d3
k5
b2
c2
5.5
b1
a2
a1
ppm
5.5
5.0
4.5
4.0
3.5
ppm
F2
other anomeric resonances confirming heterogeneity in the
sample. There was also some heterogeneity at h1 with the
appearance of a downfield doublet at 4.91 ppm near h1.
Standard homo- and hetero-nuclear 2D-NMR analyses
was undertaken. From the COSY spectrum in Fig. 2, the H-2
resonances could be located. It was realized that for residues
b, d, e, f, and g, J1,2 was small, typical of manno-heptoses,
and that a 2D-TOCSY could not be used to transfer magneti-
zation from H-1 past H-2. From the HMQC spectrum in
Fig. 3, the H-2 of the heptose residues overlapped with other
resonances making 2D-TOCSY difficult to analyze for these
residues. The 2D-NOESY spectrum also showed an unusu-
ally high number of NOEs especially for the b1, de1 reso-
nances. Also, there was overlap of several resonances in the
anomeric region. Hence, due to the complexity of the spectra
and heterogeneity of the sample, 1D selective methods were
used to extract spectral parameters that could not be ob-
tained using standard 2D methods, thus permitting the reso-
lution of the structure and conformational analysis. HMQC-
TOCSY and HMBC experiments were also very important
in determining the complete assignments, especially for the
heptose units. Using this approach, the complete assignment
with the assignments made here. The 1D-TOCSYs for resi-
dues a and h permitted assignment of the resonances and
measurement of proton coupling constants (Fig. 4a, 4b). Lo-
cation of phosphate groups was confirmed from a 31P
HMQC experiment as done previously (65). Assignment of
the 13C NMR chemical shifts was then made from HMQC,
HMQC-TOCSY, and HMBC spectra. This information led
to the definition of residues a and h as the α-D-GlcN and β-
D-GlcN pyranosyl units of the lipid A moiety. Most of the
proton, 31P and 13C NMR chemical shifts were similar to
those previously reported in a similar structural element
(65). In the 1D-TOCSY for h1, the h′5 peak at 3.62 ppm is
due to hydrolysis of the Kdo-(2-6)-β-D-GlcNII glycosidic
linkage. After several months in solution, the linkage was
completely hydrolyzed with h′6 and h′6 appearing at 3.82
and 3.92 ppm, respectively. The sharp anomeric signal at
3.91 ppm was also found to be due to h′1 of this
disaccharide.
For Kdo, the H-4 and H-5 resonances were assigned from
1D-TOCSY experiments with selective excitation of H-3eq
or H-3ax. The H-4 resonance is shifted downfield due to a
phosphate group at C-4, confirmed by a 31P HMQC. A small
J5,6 < 1 Hz impeded the TOCSY transfer past H-5 (Fig. 4c).
However, a strong NOE is observed between k4 and k6 in
Fig. 2c in accord with the X-ray structure of Kdo (66). 1D
NOESY-TOCSY(k4, k6) was used to complete the assignment
1
for the H and 13C NMR chemical shifts for the major back-
bone oligosaccharide (A1) was possible (Table 1).
The Kdo-GlcNII-GlcNI sequence and partial assignments
are known from previous studies (65) and are consistent
© 2002 NRC Canada

Brisson et al.
953
1
Fig. 3. Selected plots from the heteronuclear H-13C 2D NMR spectra of A1. (a) In the HMQC spectrum of the ring region some as-
signments are labelled where possible. (b) In the HMQC-TOCSY spectrum, the C7-H7s-C6-H6, C4-H4-C5-H5 TOCSY correlations
and those for b2 are labelled. (c) In the HMBC spectrum, the H1-C1-O1-Cx inter-glycosidic correlations are shown for the anomeric
proton resonances along with their intra-residue correlations.
de1
c6
c6'
g7'
j66'
f'7
g1
h1
(b)
(a)
g7
f1
g6
c1
j1
a1
i1
b1
d7
d7'
k8
d6
b6
h6
b7
e6
65
65
e7
e4
(c)
i6
i6
i6'
e5
g4
b4
b5
f5
f4
a3
c4
b6
a6
k7
a6
i4
70
70
a4
i2
g2
b3
e6
d2
f6
c5
c3
2j
c2
f7
F1
F1
b5
f7
a5
e2
b5
d5
k5
j3
f5
c3
a5
75
80
75
d4
i5
h5
d4
h4
d3
d3
d3
j5
g6
g6
g7 g7'
i3
b5
b34
b2
80
b2
b2
5.0
d6
d6
4.2
4.0
3.8
3.6 ppm
4.2
4.0
3.8
3.6 ppm
4.5 ppm
5.5
F2
F2
F2
(Fig. 4d). As shown later, for the HepI-(1-5)-Kdo linkage,
the d1-k7 NOE was also observed, typical of a substitution
at C-5 of Kdo (67). Using this NOE, the Kdo resonances de-
tected from the 1D-NOESY-TOCSY (de1, k7) (Fig. 4e) had
the same chemical shifts and similar multiplet patterns as
those found in the previous experiment, thus confirming the
termined from the H-5 multiplet observed for the i1-i5 NOE
in Fig. 5f. It was apparent that both J5,6 and J5,6′ had small
values <2 Hz since the H-5 multiple appeared as a doublet
dominated by the large J4,5 coupling of 10 Hz. This showed
that both H-6 and H-6′ were gauche to H-5 with the O6-C6-
C5-O5 = –60° rotamer being preferred in solution.
1
Kdo H NMR assignments. The 13C NMR assignments for
Residue c was determined to be a terminal α-D-glucose
denoted as GlcII. From the 1D-TOCSY for c1 (Fig. 4f ) reso-
nances up to H-5 were detected, indicating that residue c
was an α-glucose based on the measured proton coupling
constants. From the HMQC spectra and comparison with
chemical shifts of terminal glucose model compounds (68,
Kdo where then obtained from the HMQC spectrum and
confirmed from the HMQC-TOCSY spectrum.
Residue i was determined to be 6-substituted β-D-glucose
denoted as GlcI. The anomeric resonance for residue i over-
lapped with the H-4 resonance of Kdo. In the 1D-TOCSY
for these two overlapping resonances, it was possible to dis-
tinguish the resonances for residue i because of the small J5,6
coupling constant for Kdo (Fig. 4c). All resonances up to H-
6s could be detected and coupling constants of the multiplets
could be measured. Due to its β-D configuration the i1-i3
and i1-i5 NOE were also observed in the NOESY spectrum
(Fig. 2 and Fig. 5f ). 13C NMR assignments for GlcI were
then obtained from the HMQC spectrum and confirmed with
the HMQC-TOCSY spectrum. From a comparison with
chemical shifts of terminal glucose (68, 69), a glycosidation
down-field shift of 3.7 ppm was observed for the C-6 reso-
nance, indicating its substitution at that position. A substan-
tial shift of –2.2 ppm was also observed for C-5. The
rotamer distribution about the C-5—C-6 bond could be de-
1
69), all the H and 13C NMR assignments could be com-
pleted.
The terminal galactose, residue j, denoted as Gal, was
identified from the 1D-TOCSY of j1 and resonances up to
H-4 were observed (Fig. 4j). To get past the small J4,5 cou-
pling constants and identify H-5, the 1D-TOCSY-NOESY
(j1, j4) was done (Fig. 4h). From the HMQC spectra and
comparison with chemical shifts of terminal galactose model
1
compounds (68, 69), all the H and 13C NMR assignments
were completed.
To assign the heptose residues, model compounds were
1
synthesized to obtain accurate H and 13C NMR chemical
shifts and JH,H values. These were crucial in assigning the
terminal heptose units by chemical shifts comparisons. Also,
© 2002 NRC Canada

954
Can. J. Chem. Vol. 80, 2002
Table 1. NMR chemical shifts (ppm) of the core oligosaccharide component from Mannheimia (Pasteurella) haemolytica serotype A1
LPS.
C-3
H-3
H-3′
C-6
H-6
H-6′
C-7
H-7
H-7′
C-8
H-8
H-8′
δC
δH
C-4
H-4
C-5
H-5
Residue
C-1 H-1
C-2 H-2
HepI
D
100.3
5.17
71.1
4.13
75.3
4.01
74.9
4.17
72.5
4.17
81.2
4.11
63.5
4.02
3.89
64.1
3.70
3.60
HepII
B
100.7
5.57
80.5
4.26
70.6
3.89
67.6
3.90
72.2
3.64
69.6
4.05
HepIII
E
102.5
5.16
71.5
4.01
71.5
3.87
67.0
3.83
72.4
3.78
70.2
4.04
64.5
3.83
3.68
HepIV
G
99.8
4.95
70.5
4.08
71.7
3.81
67.9
3.78
71.8
3.94
77.0
4.14
61.2
3.91
3.80
HepV
F
99.0
5.04
70.8
3.93
71.5
3.85
68.1
3.82
73.7
3.99
71.7
4.22
71.6
4.16
3.85
HepV
f′
99.0
5.04
70.8
3.93
71.5
3.85
68.3
3.77
73.7
3.97
72.8
4.05
62.9
3.79
3.73
GlcI
I
104.0
4.64
74.2
3.51
77.7
3.45
70.2
3.64
74.6
3.54
65.5
4.06
3.69
60.3
3.96
3.76
GlcII
C
102.4
5.22
72.9
3.58
73.9
3.83
69.3
3.58
72.4
3.95
Gal
J
104.3
4.46
71.7
3.56
73.5
3.67
69.5
3.93
76.0
3.71
61.8
3.80
3.76
GlcNI
A
92.6
5.75
54.8
3.45
70.2
3.93
70.5
3.51
73.4
4.13
69.7
4.28
3.89
GlcNII
H
99.7
4.88
56.3
3.17
72.2
3.90
75.0
3.94
74.6
3.78
63.4
3.81
3.64
Kdo
K
99.7
34.5
2.37
2.13
70.8
4.61
72.7
4.32
73.2
3.87
69.9
3.78
64.2
3.94
3.73
Note: Measured at 600 MHz (1H) in D2O, 25°C and pH 3 from HMQC and HMBC data with the CH3 signal of external acetone set at 2.225 ppm for
1H NMR and 31.07 ppm for 13C NMR. Average error of ±0.2 ppm for δC and ±0.02 ppm for δH. The minor component for f due to heterogeneity is
indicated by f′. The CH2 spin systems, (H, H′), are in decreasing order of chemical shifts. For Kdo, H-3 and H-3′ are assigned to H-3eq and H-3ax.
upon glycosidation, the 13C NMR resonance of the substi-
tuted carbon experiences a substantial down-field glyco-
sidation shift (56, 57). The proton spectra for D-D-heptose
and D-L-heptose are shown in Fig. 6. In solution both the α
Residues b, d, e, f, and g in A1 were identified as hep-
toses from their narrow anomeric resonance due to the small
J1,2 coupling and lack of transfer of magnetization beyond
H-2 from H-1 in a TOCSY experiment. All the heptoses in
A1 had the α-D configuration since the only intra-residue
NOE from H-1 was to H-2 (Fig. 2). 1D-TOCSY-TOCSY (H-
1, H-2) was used to assign all the heptose residues. The first
TOCSY step from H-1 transfers the magnetization to H-2
and further transfer is impeded due to the small J1,2 value of
1.8 Hz. However, the second step at H-2 will transfer the
magnetization further to higher spins due to the larger J2,3
value of 3 Hz. However, for L-D-heptose, transfer of magne-
tization stops at H-5 due to the small J5,6 value of 1.6 Hz.
For D-D-heptose, transfer of magnetization is less impeded
due to a large J5,6 value of 3.2 Hz. However, relaxation
1
and β forms are present for each compound. Their H NMR
spectra were assigned using 1D-TOCSY experiments. To
obtain accurate coupling constants and chemical shifts, spin
simulations of the proton spectra were done. The simulated
spectra are shown in Fig. 6. The simulated spectra repro-
duced exactly the observed spectra, especially the strong
coupling for H-3-H- in D-α-L-heptose (Fig. 6b). 13C NMR
chemical shifts were assigned using HMQC. The NMR data
for the heptose monosaccharides are given in Table 2. The
chemical shifts for L-D-heptose are the same as those of D-L-
heptose since they are enantiomers of each other.
© 2002 NRC Canada

Brisson et al.
955
Fig. 5. 1D selective experiments to detect dipolar interactions in
A1. 1D NOESY spectra of the resonances in the anomeric re-
gion with a 200 ms mixing time for b1, c1, de1, f1, g1, i1/k4
and a 1D ROESY (f1, 500 ms) for j1 in (a)–(g), respectively.
The selected anomeric resonances and the linkage are indicated
on the left of the spectra.
Fig. 4. 1D selective experiments for assignment of the non
heptose residues in A1. (a) 1D TOCSY (a1, 180 ms); (b) 1D
TOCSY (h1, 180 ms); (c) 1D TOCSY (i1 k4, 180 ms); (d) 1D
NOESY-TOCSY (i1 k4, 400 ms; k6, 180 ms); (e) 1D NOESY-
TOCSY (de1, 400 ms; g3 k7, 180 ms); ( f ) 1D TOCSY (c1,
180 ms); (g) 1D TOCSY (j1, 180 ms); and (h) 1D TOCSY-
NOESY (j1, 180 ms; j4, 400 ms). Minor components for f and h
due to heterogeneity are indicated by f ′ and h′, respectively. The
resonance for the first selective step is underlined and the one
for second selection step is doubly underlined, where applicable.
Selected resonances in the anomeric region are indicated on the
left of the spectra.
j5
j3
f7'
j1
j(1-7)g
j2
f7
(g)
(f)
d4
i5
d6
k5
i3
k4/i1
i(1-4)d
k6
j4
j5
g2
i6
j1
j1
(h)
(j)
i6'
g1
g(1-6)i
b3
f2
b2
b2
(e)
(d)
j4
j2
j3
g6
g5
f1
f(1-6)g
c2,4
c5
c3
g3
c1
k7
g3
(f)
k5
de1
d(1-5)k
e(1-2)b
e2
d2
g2
i4
(c)
k7
k8
k8
k6
g2
k8'
(e)
de1
d6
c2
c1
c(1-6)d
i1
k4
k6
(b)
k7
k8'
(d)
(c)
b2
d3
i5
e7 e5
k5
a6
b1
b(1-3)d
i3
d4
i4
e7'
i1
k4
i2
d2
i2
(a)
i6
i6'
4.3
4.2
4.1
4.0
3.9
3.8
3.7
3.6
3.5 ppm
h3
h5
h6
h6'
h'5
h4
h'6
h1
a1
(b)
(a)
(Fig. 3). Comparison of chemical shifts with those of L-α-D-
heptose indicated a 9 ppm down-field shift for C-2 and an
up-field shift of –0.8 ppm for C-3 indicative of a substitution
at C-2. The C-4 to C-7 chemical shifts were within 0.6 ppm
of those of the monosaccharide.
a3
a4
a2
a6'
a5
Residue e was determined to be a terminal L-α-D-heptose
denoted as HepIII. Although the d1 and e1 resonances over-
lap, this is of no concern since their H-2 resonances did not
overlap. The 1D-TOCSY-TOCSY (de1, e2) identified the e3,
e4, and e5 spins (Fig. 7c). From the b1 NOE in Fig. 5a, the
e5 resonance was also observed. The high digital resolution
of the 1D selective experiments permits accurate matching
of resonances between different experiments due to the ob-
servation of their multiplet pattern. In Fig. 5a, the b1-e7 and
b1-e7′ NOEs were also observed. As seen later, these NOEs
are due to the close proximity of b1 proton to the e5 and the
e7 and e7′ protons. The H-6 resonance was located in the
4.3
4.2
4.1
4.0
3.9
3.8
3.7
3.6
3.5 ppm
effects can also impede transfer of magnetization and one
cannot use a lack of transfer beyond H-5 as proof of L-D-
heptose identification.
Residue b was determined to be the 2-substituted L-α-D-
heptose denoted as HepII. The 1D-TOCSY-TOCSY (b1, b2)
identified resonances at 3.64 ppm and 3.9 ppm (Fig. 7a).
The multiplet pattern at 3.64 ppm was indicative of a H-5
resonance while those at 3.9 ppm were similar to the
strongly coupled H-3 and H-4 resonances observed in the
monosaccharide (Fig. 6b). To identify H-6 a 1D-TOCSY-
NOESY (b2, b5) was done (Fig. 7b). In the first TOCSY
step for b2, the a6 resonance was also irradiated but the sec-
ond step only selected the b5 resonance. NOEs from b2
were observed on b3 and b6 along with an inter-residue
NOE on d2. Once b6 was located, the HMQC-TOCSY from
the C-7-H-7s-C-6-H-6 was used to assign the H-7 and
H-7′ resonances (Fig. 3b). Assignments for the other reso-
nances were then obtained from the HMQC spectrum and
confirmed from the HMQC-TOCSY and HMBC spectra
1
HMQC-TOCSY from C-7-H-7s-C-6-H-6. H and 13C NMR
assignments were completed with HMQC and HMQC-
1
TOCSY. H and 13C NMR chemical shifts of residue e were
similar to those of L-α-D-heptose.
Residue d was identified as a 3,4,6-trisubstituted L-α-D-
heptose denoted as HepI. The 1D-TOCSY-TOCSY (de1, d2)
identified a narrow multiplet at 4.17 and a broader one at
4.01 ppm (Fig. 7d). The multiplets at 4.01 ppm and
4.17 ppm (d2) were also observed in the 1D-NOESY for b1
(Fig. 5a). The resonance at 4.17 ppm was also observed in
© 2002 NRC Canada

956
Can. J. Chem. Vol. 80, 2002
Table 2. NMR data for D-L-heptose and D-D heptose monosaccharides.
δC
C-1
H-1
J1,2
C-2
H-2
J2,3
C-3
H-3
J3,4
C-4
H-4
J4,5
C-5
H-5
J5,6
C-6
H-6
J6,7
C-7
δH
H-7
H-7′
Heptose
JH,H
J6,7′
J7,7′
D-α-L-
95.00a
5.166
1.8
71.45
3.914
3.1
72.02
3.927
3.2
71.32
3.905
3.4
71.84
3.921
3.3
71.36
3.838
10.0
74.09
3.649
10.0
71.38
3.812
9.4
74.11
3.622
9.5
67.05
3.845
9.7
66.69
3.788
9.8
68.35
3.756
10.1
68.16
3.682
9.9
71.81
3.749
1.6
75.43
3.329
1.8
73.42
3.865
3.2
77.22
3.423
3.3
69.63
4.024
7.3
69.53
3.98
7.5
72.72
4.018
3.3
72.623
4.025
3.4
63.84
3.689
5.5
63.57
3.713
5.8
62.75
3.797
7.6
62.58
3.788
7.4
3.652
–11.7
D-β-L-
D-α-D-
D-β-D-
94.74b
4.866
1.0
3.695
–11.7
94.86a
5.151
1.8
94.79b
4.851
1.1
3.708
–12.0
3.725
–12.0
1
Note: Measured at 600 MHz (1H) in D2O at 25°C from H spin simulations and from the 13C spectra (150 MHz) with the CH3 signal of acetone set at
1
2.225 ppm for H NMR and 31.07 ppm for 13C NMR. δC and δH are in ppm with an average error of ±0.005 ppm for δC and ±0.003 ppm for δH. JH,H
values are in Hz with an average error of ±0.2 Hz. H-7 and H-7′ are in decreasing g order of chemical shifts.
aJC1,H1 = 171 ± 1 Hz.
bJC1,H1 = 160 ± 1 Hz measured from the undecoupled 13C NMR spectrum.
that the crosspeaks at (74.9, 4.17) and (75.3, 4.01) were sub-
Fig. 6. Proton spectra of heptose monosaccharides. Resolution
enhanced spectrum of D-L-heptose at 600 MHz in D2O, 300 K
(a) and its simulated spectrum for the α (b) and β forms (c).
Resolution enhanced spectrum of D-D-heptose at 600 MHz in
D2O, 300 K (d) and its spin simulated spectra for the α (e) and
β forms ( f ). The anomerics and H-5β resonances are not shown.
Note that strong coupling for the H-3, H-4 and virtual coupling
for H-5 resonances are reproduced correctly for (b).
ject to a 13C NMR down-field glycosidation shift. Since
these three crosspeaks are from d3, d4, or d5, only the d5
crosspeak will not experience a large glycosidation shift
since substitutions on heptose cannot occur at C-5. Hence,
the crosspeak at (72.5, 4.17) was assigned to (C-5, H-5). The
crosspeak at (75.3, 4.01) was assigned to (C-3, H-3) based
on the HMBC (d3, d1) and (d3, b1) correlations. The
crosspeak at (74.9, 4.01) was thus assigned as (d4, d1), con-
4
sistent with the observation of the HMBC (d4, i1) correla-
tion and (i1, d4) NOE (Fig. 3 and Fig. 5f ). The d6
resonance was assigned from the 1D-NOESY from i1, since
for the GlcI-(1-4)-HepI linkage, a strong NOE to H-6 is also
expected only if residue d is L-D-heptose. The d6 resonance
was also observed for the 1D-NOESY for c1 (Fig. 5b). The
H-7 and H-7′ resonances were located from the HMQC-
TOCSY (C-7, H-6) crosspeak (Fig. 3) and from 1D-
NOESY-TOCSY (i1, d6) (not shown). Comparison of chem-
ical shifts with those of L-α-D-heptose indicated a 3.9, 7.9,
and 11.6 ppm down-field glycosidation shift for C-3, C-4,
and C-6, respectively.
As shown below, there should be a 4,6-disubstituted L-α-
D-heptose (d′) present due to heterogeneity on the branching
heptose (Fig. 1). From the HMQC and COSY spectra for the
anomeric region, no separate signal could be observed either
for the H-1 or H-2 due to overlap with other resonances. It is
suspected that the proton chemical shifts of d′1 were the
same as the one for d1, and that d′2 had a similar chemical
shift to e2.
3
2
7
7'
6
(f)
4
7
2
7'
3
6
5
(e)
(d)
4
7 7'
7
2
3
6
(c)
3
4
7'
2
6
5
(b)
(a)
Residue g was determined to be a 6-substituted D-α-D-
heptose denoted as HepIV. The 1D-TOCSY-TOCSY (g1, g2)
identified the g3, g4, and g5 spins (Fig. 7e). The multiplet
pattern at 3.94 was indicative of a H-5 resonance. To iden-
tify H-6, an 1D-TOCSY-NOESY (g2, g5) was acquired
(Fig. 7f ). NOEs from g2 were observed on g3 and g6 along
with a putative inter-residue NOE on i6. Once g6 was lo-
cated, the HMQC-TOCSY from the C-7-H-7s-C-6-H-6
(Fig. 3b) and 1D-NOESY-TOCSY (f1, g6) (Fig. 7g) were
4.00
3.95
3.90
3.85
3.80
3.75
3.70
3.65 ppm
the 1D-NOESY for i1 (Fig. 5f ). In the HMQC spectrum two
narrow crosspeaks at (δc, δh) (74.9, 4.17), (72.5, 4.17) and a
broader one at (75.3, 4.01) were identified. HMQC-TOCSY
correlations were also observed between these crosspeaks.
From a comparison of the 13C NMR chemical shifts with
those of the α-D-heptopyranoses in Table 2, it was obvious
© 2002 NRC Canada

Brisson et al.
957
Fig. 7. 1D selective experiments for assignment of the heptose
residues in A1. (a) 1D TOCSY-TOCSY (b1, 75 ms; b2, 75 ms)
to detect b3, b4 and b5; (b) 1D TOCSY-NOESY (b2, 150 ms;
b5, 400 ms) to detect the b5-b6 and b5-b3 NOEs. Note also the
strong b5-d2 NOE due to the b(1-3)d linkage; (c) 1D TOCSY-
TOCSY (de1, 75 ms; e2, 75 ms) to detect e3, e4 and e5; (d) 1D
TOCSY-TOCSY (de1, 75 ms; d2, 75 ms) to detect d3, d4 and
d5; (e) 1D TOCSY-TOCSY (g1, 90 ms; g2, 150 ms) to detect g3
to g5; ( f ) 1D TOCSY-NOESY (g2, 150 ms; g5, 400 ms) show-
ing the g5-g6 and g5-g3 NOEs and g5-i6 NOE in accord with
the g(1-6)i linkage; (g) 1D NOESY-TOCSY (f1, 400 ms;
g6 180 ms) to detect the g7 resonances from the f1-g6
interglycosidic NOE in accord with the f(1-6)g linkage; (h) 1D
TOCSY-TOCSY (f1, 90 ms; f2, 150 ms) for detection of f and f′
resonances up to H-6. Note the clear multiplet pattern for f5 and
f′5; (i) 1D TOCSY (f6, 40 ms) to detect the f7 resonances only
due to the short spin lock time. The resonance for the first selec-
tive step is underlined and the one for second selection step is
doubly underlined, where applicable. Selected resonances in the
anomeric region are indicated on the left of the spectra.
Residue f was determined to be a 7-substituted D-α-D-
heptose and denoted as HepV. The 1D-TOCSY-TOCSY (f1,
f2) identified spins up to f6 indicating that residue f was a
D-D-heptose. The f5 multiplet pattern was quite clear. From a
1D-TOCSY starting on f6, the f7, and f7′ resonances were
identified. The 13C NMR assignments were then obtained
from the HMQC spectrum and confirmed with the HMQC-
TOCSY spectrum. Comparison of chemical shifts with those
of α-D-D-heptose indicated a 8.8 ppm down-field shift for
C-7. A terminal D-α-D-heptose, denoted f′, was only detected
due to heterogeneity at this linkage site. As seen in the 1D-
TOCSY-TOCSY (f1, f2), the f′5 and f′6 resonance can be
detected, along with crosspeaks in the HMQC and HMQC-
TOCSY spectra which correspond to those of a terminal D-
α-D-heptose similar to those for D-α-D-heptose in Table 2.
The sequence of the core octasaccharide was established
from the 1D-NOESY spectrum for the anomeric resonances
presented in Fig. 5 and from the HMBC spectrum of the
anomeric proton resonances (Fig. 3c). Inter-residue NOEs
were also observed in 1D-TOCSY-NOESY experiments in
Fig. 7b and 7f. The results are tabulated in Table 3 and the
structure shown in Fig. 1. From integration of the anomeric
resonances and appearance of new resonances with time, the
linkage sites where heterogeneity occurs were determined.
The k(2-6)h linkage and the b(1-3)d linkage hydrolyzed over
a period of months in solution at pH 3, while the j(1-7)f
linkage was stable.
An unusually large number of long-range NOEs were ob-
served spanning up to five sequential residues. For an α-D
sugar, the H-1-H-2 intra-residue NOE is expected. For a β-D
sugar, the H-1-H-3 and H-1-H-5 intra-residue NOEs are ex-
pected. The H-1-C-1-O-1-C-x-H-x interglycosidic NOEs are
expected for (1-x) linkage. Inter-residue anomeric NOEs be-
tween two linked sugars on the H-x ± 1 and H-x ± 2 can also
occur. Other inter-residue NOEs, not in the vicinity of the
glycosidic linkage, are deemed to be long-range NOEs as
listed in Table 3.
To explain the large number of long-range NOEs,
conformational analysis was done using the Metropolis
Monte Carlo (MMC) method to vary the glycosidic linkage
angles and sample the multiple conformations of the mole-
cule. Using this method for the inner core oligosaccharide
with Kdo at the reducing end, it was found that all the ob-
served long-range NOEs could be explained by the close
proximity of the e-b branch to the g-i branch. These long-
range NOEs were possible due to restriction of the inner
core residues brought about by the three branching points on
HepI.
f7'
f6
f7
(i)
f2
f5
f3
f'5
f4
f1
f1
f'6
f6
(h)
g6
g7'
g7
(g)
(f)
g2
g3
g6
i6
g5
g5
g2
g3,4
g1
(e)
(d)
d2
d4,5
de1
d3
e2
e3,4
e5
de1
b1
(c)
(b)
(a)
b2
b2
b3
d2
b6
b5
b3,4
3.9
b5
A minimum energy conformer obtained from the calcula-
tion is shown in Fig. 8. The various interproton distances
measured from these coordinates are given in Table 3. As
can be observed the occurrence of most NOEs can be ex-
plained by the short inter-proton distances. The long-range
NOEs for g1-b2 and g1-b3 are not consistent with distances
obtained from the molecular model drawn in Fig. 8. Al-
though there is restriction due to the 3,4,6)-HepI substitution
pattern, there is flexibility about the glycosidic linkage and
this must be taken into account. As shown in Fig. 9 for the
g1-b2 and g1-b3 distances, multiple conformations having
short inter-proton distances are sampled consistent with the
observed NOEs. The same situation was applicable for all
4.3
4.2
4.1
4.0
3.8
3.7
3.6
3.5 ppm
used to assign the H-7 and H-7′ resonances. Assignments for
the other resonances were then obtained from the HMQC
spectrum and confirmed with the HMQC-TOCSY spectrum
(Fig. 3). In the 1D-NOESY (f1), the g6 and g5 resonances
were observed (Fig. 5d). The NOE f1-g5 NOE can only be
possible if residue g is a D-D-heptose from molecular model-
ing studies (see below). Comparison of chemical shifts with
those of D-α-D-heptose indicated a 4.3 ppm down-field shift
for C-6.
© 2002 NRC Canada

958
Can. J. Chem. Vol. 80, 2002
Table 3. HMBC and NOE data for A1 and distances from a minimum energy conformer of the core heptasaccharide shown in Fig. 8.
HMBC
H-1-C-x
Intra-residue
NOE
Inter-residue
NOE
Long range
NOE
Linkage
b(1-3)d
r (Å)
r (Å)
r (Å)
b1-d3
b1-b2
b5-b3
b5-b6
2.6
2.5
2.4
b1-d3
b1-d4
b5-d2
k3ax-d5
c1-d6
d1-k5
d1-k7
e1-b2
e1-b1
2.6
3.7
2.3
3.2
2.0
2.2
2.9
2.1
2.8
b1-e5
b1-e7
b1-e7′
b1-i2
c1-i1
3.3
2.9
2.6
2.2
2.4
c(1-6)d
d(1-5)k
c1-d6
e1-b2
c1-c2
d1-d2
2.4
2.6
e(1-2)b
e1-e2
f1-f2
2.6
2.6
e1-g2
e1-g3
e1-i4
4.2
2.3
2.0
f(1-6)ga
g(1-6)i
f1-g6
g1-i6
f1-g6
f1-g5
g1-i6
g1-i6′
g5-i6
i1-d4
i1-d6
j1-f7
j1-f7′
2.5
2.6
2.6
2.6
3.0
2.8
2.1
2.7
2.5
g1-g2
g5-g3
g5-g6
i1-i3
i1-i5
j1-j2
j1-j3
j1-j5
2.6
2.6
2.5
2.6
2.4
3.1
2.7
2.4
g1-b2
g1-b3
4.2
4.9
i(1-4)d
j(1-7)fa
i1-d4
j1-f7
i1-c1
2.4
aFor the j and f residues in the Gal-(1-7)-HepV-(1-6)-HepIV sequence (j-f-g), the average distances from MMC calculations are given.
Fig. 8. Molecular model of a minimum-energy conformer ob-
tained from a MMC calculation of the inner core heptasaccharide
of A1. Oxygens are depicted as the larger spheres and hydroxyl
protons are removed. Relevant protons are labelled. Note the
close proximity of the e-b branch to the g-i branch and the close
proximity of the exocyclic chain of residue e to the anomeric
proton of residue b consistent with the long range NOEs ob-
served. Also, note the differences in orientation of the exocyclic
chain between L-D heptose (residues d, b, e) and D-D heptose
(residue g).
Fig. 9. Variation of inter-proton distances (r) vs. macro move in
a MMC calculation of the inner core heptasaccharide of A1 for
b1-e5, b1-e7, e1-i4, e1-g2, g1-b2, g1-b3 in (a)–(g), respectively.
Occurrences of interproton distances in the 2–4 Å range are con-
sistent with the observed long range b-e, e-i, and e-g NOEs.
the other long-range and inter-residue NOEs. The long-range
NOEs for g1-b2 and g1-b3 and e1-i4 span four residues with
mobility about the three glycosidic (φ, ψ) angles for a total
of 6 df. The long-range NOEs between e1-g2 and e1-g3 that
span five residues (e-b-d-i-g) vary even more due to 8 df
© 2002 NRC Canada

Brisson et al.
959
about the glycosidic angles, not counting possible flexibility
about the C-5—C-6 bond for the g(1-6)i linkage. The b1-e5,
b1-e7, and b1-e7′ are dependent on mobility about the
e(1-2)b glycosidic bond and rotation about C-6-C-7 of resi-
due e. In all cases, the occurrence of multiple conformations
that have short inter-proton distances is consistent with the
observed inter-residue and long-range NOEs.
In the present study it was established that the lipid A re-
gion of the M. haemolytica A1 LPS molecule contains a bis-
4,4′-phosphorylated β-1,6 linked D-glucosamine disaccharide
moiety. The nature and substitution patterns of the fatty acyl
groups attached to this disaccharide have not been reported.
The fully acylated LPS molecule makes up the outer most
leaflet of the bacterial membrane and is essential for main-
taining membrane integrity. The glycose portion of the mol-
ecule typically extends out and away from the plane defined
by the bacterial membrane. The LPS oligosaccharide portion
is important in virulence (70) and is involved in eliciting
host immune responses (17). To gain a perspective of the
relative sizes and orientation of the core oligosaccharide and
membrane-anchoring lipid A regions of this LPS molecule, a
molecular model was constructed using the lipid A fatty acid
haemolytica serotype A8 lacks the terminal Gal unit in the
core oligosaccharide, a residue which possibly provides a
site for O-chain attachment (16). An understanding of LPS
structural differences among M. haemolytica serotypes could
provide the basis for a vaccine against M. haemolytica
bovine pasturellosis. Serotype A1 is the principal microor-
ganism responsible for this disease, accounting for 30% of
total cattle deaths globally (11). The use of a less virulent
serotype or antigens thereof in a vaccine formulation could
provide an effective disease management strategy.
Experimental
Preparation of LPS from M. haemolytica
Pasteurella (Mannheimia) haemolytica serotype A1
(NRCC 4212) was obtained from the Veterinary Infectious
Diseases Organization (VIDO), Saskatoon, SK, Canada. LPS
was extracted and purified from fermenter-grown bacteria
culture by the hot aqueous phenol method as previously de-
scribed (16).
Deacylation of LPS to give backbone oligosaccharides
Backbone oligosaccharide was prepared as previously de-
scribed (74) by modification of the deacylation procedure of
Holst et al. (75). Ester-bound fatty acids were removed from
the lipid A of the LPS by treatment of a sample (400 mg)
with anhydrous hydrazine at ambient temperature (20 mL,
37°C, 30 min). Excess hydrazine was destroyed by addition
of acetone (three volumes) to the cooled reaction mixture
(0°C). The precipitated O-deacylated LPS product was iso-
lated by centrifugation (5000 rpm, 10 min), washed with ac-
etone, and lyophylized from water. Removal of N-acyl
groups was achieved by heating the O-deacylated sample
(200 mg) in aq KOH (4 M, 10 mL) at 100°C for 20 h. The
reaction mixture was cooled (0°C), neutralized with 4 M
HCl, and the precipitated fatty acids removed by
centrifugation (12 000 × g, 30 min). The supernatants were
filtered in an Amicon concentration cell with a 500-
molecular-weight-cutoff membrane (Amicon; YC05) and
washed with deionized water until the eluant was free of
chloride ions (as determined with aq AgNO3). The dialyzed
material was lyophylized to give deacylated LPS (ca.
80 mg). The oligosaccharide was purified by anion-exchange
chromatography on DEAE A-25 (74). The backbone
oligosaccharide fraction (ca. 50 mg) showed a doubly
charged ion at m/z 1124.8 in the positive ion ESI-MS as the
substitution pattern found in
a
related organism,
Haemophilus influenzae. This is shown in Fig. 10. The H.
influenzae lipid A has been reported (71) to have six fatty
acyl groups in which the amide group (C-2) and C-3
hydroxyl groups of the reducing glucosamine residue are
acylated by 3-hydroxytetradecanoic acid while the C-2′ am-
ide and C-3′ hydroxyl groups of the nonreducing residue are
acylated by 3-tetradecanoyloxytetradecanoic acid.2 As ob-
served in Fig. 10, due to the inner core which provides a
fairly rigid structure, the glycose portion projects out from
the lipid A moiety which defines the bacterial membrane.
More mobility was observed for the terminal residues, espe-
cially for the terminal Gal residue. In the NOESY spectrum
in Fig. 2, no NOEs for the anomeric resonance were ob-
served consistent with increased mobility of this residue.
The core structures of several LPS have been found to be
highly branched (3) and this can result in well-defined con-
formations for the inner core. In a previous study, the LPS of
Moraxella catharrhalis was found to adopt an unusual con-
formation in which a very rare anti conformer (72) was ob-
served (73). For a highly branched 3,4,6-trisubstituted D-
glucose residue in this LPS molecule, a dihedral angle (C-1′-
O-1′-C-4-H-4) near 180° was detected for β-D-Glcp-(1–4)-D-
Glcp linkage to the branched glucose unit.
The results of the present study have provided a detailed
picture of the structure of the LPS core oligosaccharide re-
gion of M. haemolytica serotype A1. We have previously
shown that the oligosaccharide region of M. haemolytica
LPS is immunogenic in mice, sheep, and cattle and that
eight of the 12 M. haemolytica serotypes (i.e., serotypes A1,
A5, A6, A7, A8, A12, A14, and A16) share common core
oligosaccharide determinants (17). The core oligosaccharides
major ion which corresponded to
a composition of
Hex3Hep5 Kdo HexN2(H2PO3)3 (M, 2246.9). MS–MS of the
doubly charged ion gave characteristic fragments at m/z 500
(HexN2(H2PO3)2) and 801 (KdoHexN2(H2 PO3)2) corre-
sponding to deacylated lipid A and the Kdo-P substituted
units, respectively.
D-glycero-L-manno-Heptose
1
from serotypes A1 and A8 show almost identical H NMR
The heptose was prepared by the condensation of D-
galactose with nitromethane in alkaline methanol (76). The
crystalline 1-deoxy-1-nitro-D-glycero-L-manno-heptitol obtained
from the aqueous solution of the deionized products (mp
spectra (17) establishing the presence of a common basal
structure. Based on the results of the present study it is ap-
parent that the O-chain deficient LPS elaborated by M.
2 ES-MS of O-deacylated LPS from M. heamolytica serotype A1 is consistant with each glucosamine residue containing an amide-linked
3-hydroxytetradecanoic acid group (data not shown).
© 2002 NRC Canada

960
Can. J. Chem. Vol. 80, 2002
Fig. 10. Molecular model of the LPS of M. haemolytica A1 constructed using a lipid A having the fatty acyl substitution patterns re-
ported (71) for LPS from H. influenzae. The lipid A moiety is colored with the atoms C in black, O in red, and H in white. The sugar
moiety is colored with Kdo in gray, heptoses in red or purple, glucose in green, and galactose in blue. The PO4 groups are yellow.
Hydroxyl protons have been removed.
158°C, 21% yield) was converted to its sodium salt. On
treatment with dilute sulphuric acid (35°C), followed by
deionization with Rexyn 101(H+) and Amberlite A4(OH–)
ion-exchange resins, the solution was lyophilized to give the
heptose as a syrup (80% yield). The product was fraction-
ated by cellulose column chromatography using butan-1-ol–
water (1:10 v/v) as the eluant. Fractions containing the heptose
were concentrated to dryness under reduced pressure. The D-
glycero-L-manno-heptose having [ꢀ]D –14° (c 0.2, water) was
pure by paper chromatography and its reduced (NaBH4) and
acetylated product on GLC gave a single peak correspond-
ing in retention time with that of authentic hepta-O-acetyl-
D-glycero-L-manno-heptitol, establishing its purity.
D-glycero-D-manno-Heptose
The heptose was prepared from D-altrose via its condensa-
tion with nitromethane to yield 1-deoxy-1-nitro-D-glycero-D-
manno-heptitol (mp 109°C, [ꢀ]D –7.0° (c 1.1, EtOH), 27%
yield)) whose aqueous sodium salt solution on slow dropwise
addition to stirred 20% (v/v) sulfuric acid maintained at 0°C
© 2002 NRC Canada

Brisson et al.
961
was converted to D-glycero-D-manno-heptose (77). The reac-
tion mixture was neutralized with sat. Ba(OH)2 solution, fil-
tered, passed through Rexyn 101(H+) and Duolite A4(OH–)
ion-exchange resins to remove residual ionic material, and
was concentrated to a syrup (92% yield). The heptose prod-
uct, as indicated by paper chromatography, was contami-
nated with altrose and unchanged 1-deoxy-1-nitro-D-glycero-
D-manno-heptitol (ca. 2 to 3%). The product was purified by
cellulose column chromatography using butan-1-ol–water
(1:10 v/v) as the mobile phase. The D-glycero-D-manno-
mannose had [α]D +22° (c 2, MeOH). A reduced (NaBH4)
and acetylated sample on GLC gave a single peak corre-
sponding to authentic hepta-O-acetyl-D-glycero-D-manno-D-
heptitol indicating the identity and purity of the synthesized
heptose.
tion step. As shown in Fig. 1, all the residues are labeled by
letters and the spins are labeled by the atom number, so that
a1 refers to the H-1 resonance of residue a.
On the AMX spectrometers, selective excitation was
achieved by means of half-Gaussian pulses. The mixing time
used for a TOCSY depended on the spin system. Usually a
range of mixing times (spin lock times) from 30 to 180 ms
was used to assign the spin system. The mixing time for a
1D-NOESY depended on the correlation time of the mole-
cule and internal motion about the glycosidic linkage.
NOESY mixing times were in the range from 150 to
400 ms. To detect inter-residue NOEs for the terminal Gal
residue, a 1D-ROESY experiment was done with a mixing
time of 500 ms. For the doubly selective experiments, the se-
lective pulses were kept as short as possible to avoid loss of
signal intensity due to relaxation effects. Each part was opti-
mized one at a time. The 1D-TOCSY could be carried out in
a matter of minutes. The 1D-NOESY took from minutes to
hours depending on the magnitude of the NOE. Some dou-
bly selective experiments took up to 12 h. The 1D-TOCSY-
TOCSY and 1D-TOCSY-NOESY were the most efficient.
As the sample degraded with time, some experiments were
also performed on an Inova 600 spectrometer making use of
pulse field gradients.
Electrospray mass spectrometry
Samples were analyzed on a VG Quattro triple quadrupole
mass spectrometer (Micromass, Manchester, U.K.) fitted
with an electrospray ion source. Backbone oligosaccharide
was dissolved in acetonitrile–water (approx. 1:2 v/v) con-
taining 0.5% acetic acid. Injection volumes were 10 µL and
the flow rate was set at 4 mL min–1. The electrospray tip
voltage was typically 2.7 kV and the mass spectrometer was
scanned from m/z 50 to 2500 with a scan time of 10 s. For
MS–MS experiments, precursor ions were selected using the
first quadrupole and fragment ions formed by collisional ac-
tivation with argon in the RF-only quadrupole cell, were
mass analyzed by scanning the third quadrupole. Collision
energies were typically 60 ev (laboratory frame of refer-
ence).
Molecular modeling
The conformational analysis was done using the Metropo-
lis Monte Carlo method as previously described (79). The
PFOS potential was used (80). Minimized coordinates for
the monosaccharides were obtained using MM3 (92) avail-
able from the Quantum Chemistry Program Exchange
(QCPE). The minimum energy conformation for each
disaccharide was used as the starting conformation. Starting
from the Kdo at the reducing end, calculations were per-
formed for various oligosaccharides up to the complete
structure. For the inner core octasaccharide, 5 × 104 macro
moves were used with a step length of 5° for the glycosidic
linkage and a temperature of 1 × 103 K resulting in an ac-
ceptance ratio of 0.36. The molecular model for the inner
core octasaccharide was generated using the minimum en-
ergy conformer. Distances were extracted from the saved co-
ordinates at each macro move. The complete LPS with lipid
A was generated using the lipid A structure (81) and the
minimum energy conformer for the Kdo linkage. Molecular
drawings were done using Schakal97 from E. Keeler, Uni-
versity of Freiburg, Germany.
Nuclear magnetic resonance
NMR spectra were performed on a Bruker AMX 600,
AMX 500, or a Varian Inova 600 spectrometer using stan-
dard software. All measurements were made at 300 K and at
pH 3, containing 10 mg of sample dissolved in 0.6 mL of
D2O. Measurements were done at pH 3 to improve resolu-
tion of the proton spectrum as done previously (65). Acetone
was used as an internal or external reference at 2.225 ppm
for 1H spectra and 31.07 ppm for 13C spectra. Standard
homo- and heteronuclear correlated 2D techniques were
used for general assignments of the core oligosaccharide:
COSY, TOCSY, NOESY, triple quantum homonuclear corre-
lated experiment, HMQC, HMQC-TOCSY, and HMBC and
31P HMQC (78). Spin simulations for the heptose mono-
saccharides were performed with standard Varian software.
Accurate chemical shifts and coupling constants were ob-
tained from the parameters used to perform the spin simula-
tion and not from the peak listing of the spectra.
Acknowledgements
We thank D.W. Griffith for large-scale production of cells
and D. Krajcarski for ESI-MS analyses.
Selective 1D-TOCSY, 1D-NOESY, 1D-TOCSY-TOCSY,
1D-TOCSY-NOESY, and 1D-NOESY-TOCSY experiments
were performed for complete residue assignment and for de-
1
termination of inter-residue H-1H nuclear Overhauser en-
References
hancements. The nomenclature used to describe the use of
the 1D selective sequences will be 1D EXP (spin, mixing
time) and 1D EXP1-EPX2 (spin1, mixing time; spin2, mix-
ing time), where EXP, EXP1, and EXP2 stand for either
TOCSY or NOESY. Also, in the figures, the selected reso-
nances are underlined. A doubly underlined resonance
means that this resonance was selected as the second selec-
1. E.Th. Rietschel, L. Brade, U. Schade, U. Seydel, U. Zähringer,
S. Kusmoto, and H. Brade. Bacterial endotoxins: properties
and structure of biologically active domains, in surface struc-
tures of microorganisms and their interactions with the mam-
malian host. In Proceedings Workshop Conference, Hoechst,
Verlag Chemie. Edited by E. Schrinner, M.H. Richmond, G.
Seibert, and U. Schwarz. Weinheim, Germany. 1988. p. 1.
© 2002 NRC Canada

962
2. C.R.H. Raetz. Annu. Rev. Biochem. 59, 129 (1990).
3. O. Holst and H. Brade. Chemical structure of the core region
of lipopolysaccharides. In Bacterial endotoxic lipopoly-
saccharides. Edited by D.C. Morrison and J.L. Ryan. CRC
Press. Boca Raton, Fla. 1992. p. 135.
4. S. Pohl. DNA relatedness among members of haemophilus,
pasteurella and actinobacillus. Haemophilus, pasteurella and
actinobacillus. Edited by M. Kilian, W. Frederiksen, and E.L.
Biberstein. Academic Press. New York. 1981. p. 245.
5. E.L. Biberstein. Biotyping and serotyping of Pasteurella
haemolytica. In Methods in enzymology. Edited by T. Bergan
and J.R. Norris. Academic Press. New York. 1978. p. 253.
6. M. Younan and L. Fodor. Res. Vet. Sci. 58, 98 (1995).
7. G.R. Smith. J. Pathol. Bacteriol. 81, 431 (1961).
8. D.P. Bingham, R. Moore, and A.B. Richards. Am. J. Vet. Res.
51, 1161 (1990).
Can. J. Chem. Vol. 80, 2002
30. H. Kessler, H. Oschkinat, and C. Griesinger. J. Magn. Reson.
70, 106 (1986).
31. H. Kessler, U. Anders, G. Gemmecker, and S. Steuernagel. J.
Magn. Reson. 85, 1 (1989).
32. L.D. Hall and T.J. Norwood. J. Magn. Reson. 87, 331 (1990).
33. J. Friedrich, S. Davies, and R. Freeman. J. Magn. Reson. 80,
168 (1988).
34. E. Kupce and R. Freeman. J. Magn. Reson. 100, 208 (1992).
35. X.J. Miao and R. Freeman. J. Magn. Reson. 119, 145 (1996).
36. R. Bazzo, C.J. Edge, R.A. Dwek, and T.W. Rademacher. J.
Magn. Reson. 86, 199 (1990).
37. D. Boudot, C. Roumestand, F. Toma, and D. Canet. J. Magn.
Reson. 90, 221 (1990).
38. L. Poppe and H. van Halbeek. J. Magn. Reson. 96, 185 (1992).
39. D. Uhrín, J.R. Brisson, and D.R. Bundle. J. Biomol. NMR, 3,
367 (1993).
9. P.H.A. Sneath and M. Stevens. Nov. Int. J. Syst. Bacteriol. 40,
148 (1990).
40. S. Holmbeck, P.J. Hajduk, and L. Lerner. J. Magn. Reson. B,
102, 107 (1993).
10. O. Angen, M. Quirie, M. Donachie, and M. Bisgaard. Vet.
Microbiol. 65, 283 (1999).
41. D. Uhrín, J.R. Brisson, G. Kogan, and H.J. Jennings. J. Magn.
Reson. Ser. B, 104, 289 (1994).
11. D. Griffin. In Veterinary clinics of North America: Food ani-
mal practice. Edited by J. Vestweber and G. St Jean. Philadel-
phia. 1997. p. 367.
42. K. Zangger and H. Sterk. J. Magn. Reson. 107, 186 (1995).
43. J. Stonehouse, P. Adell, J. Keeler, and A.J. Shaka. J. Am.
Chem. Soc. 116, 6037 (1994).
12. W.D.G. Yates. Can. J. Comp. Med. 46, 225 (1982).
13. A.W. Confer, K.D. Clinkenbeard, and G.L. Murphy.
Pathogenesis and virulence of Pasteurella haemolytica in cat-
tle: an analysis of current knowledge and future approaches. In
Haemophilus, actinobaccilus and pasteurella. Edited by W.
Donachie, F.A. Lainson, and J.C. Hodgson. Plenum Press.
New York. 1995. p. 51.
14. M.A. Breider, S. Kumar, and R. Corstvet. Vet. Immuno.
Immunopathol. 27, 337 (1991).
15. D.B. Paulsen, D.A. Mosier, K.D. Clinkenbeard, and A.W. Confer.
Am. J. Vet. Res. 50, 1633 (1989).
44. M.A. Bernstein and L.A. Trimble. Magn. Reson. Chem. 32,
107 (1994).
45. P. Adell, T. Parella, F. Sanchez-Ferrando, and A. Virgili. J.
Magn. Reson. Ser. A, 113, 124 (1995).
46. C. Dalvit and G. Bovermann. Magn. Reson. Chem. 33, 156
(1995).
47. T. Facke and S. Berger. J. Magn. Reson. Ser. A, 113, 257
(1995).
48. C. Dalvit. J. Magn. Reson. 113, 120 (1995).
49. P. Adell, T. Parella, F. Sanchez-Ferrando, and A. Virgili. J.
Magn. Reson. Ser. B, 108, 77 (1995).
16. W.B. Severn and J.C. Richards. Carbohydr. Res. 240, 277 (1993).
17. R.P. Lacroix, J.R. Duncan, R.P. Jenkins, R.A. Leitch, J.A. Perry,
and J.C. Richards. Infect. Immun. 61, 170 (1993).
18. M.B. Perry and L.A. Babiuk. Can. J. Biochem. Cell Biol. 62,
108 (1984).
50. T.L. Hwang and A.J. Shaka. J. Magn. Reson. 112, 275 (1995).
51. K. Stott, J. Stonehouse, J. Keeler, T.L. Hwang, and A.J. Shaka.
J. Am. Chem. Soc. 117, 4199 (1995).
52. G. Xu and J.S. Evans. J. Magn. Reson. Ser. B, 111, 183 (1996).
53. T. Parella. Magn. Reson. Chem. 34, 329 (1996).
54. M.J. Gradwell, H. Kogelberg, and T.A. Frankiel. J. Magn.
Reson. Ser. B, 124, 267 (1997).
19. R.A. Leitch and J.C. Richards. Can. J. Biochem. Cell Biol. 66,
1055 (1988).
20. J.C. Richards and R.A. Leitch. Carbohydr. Res. 186, 275 (1989).
21. R.U. Lemieux and K. Bock. Arch. Biochem. Biophys. 221,
125 (1983).
22. R.U. Lemieux. Chem. Soc. Rev. 18, 347 (1989).
23. D.W. Hood, A.D. Cox, M. Gilbert, K. Makepeace, S. Walsh,
M.E. Deadman, A. Cody, A. Martin, M. Mansson, E.K. Schweda,
J.R. Brisson, J.C. Richards, E.R. Moxon, and W.W. Wakarchuk.
Mol. Microbiol. 39, 341 (2001).
24. E.H. Schweda, J.R. Brisson, G. Alvelius, A. Martin, J.N. Weiser,
D.W. Hood, E.R. Moxon, and J.C. Richards. Eur. J. Biochem.
267, 3902 (2000).
25. I. Sadovskaya, J.R. Brisson, P. Thibault, J.C. Richards, J.S.
Lam, and E. Altman. Eur. J. Biochem. 267, 1640 (2000).
26. J.S. Plested, K. Makepeace, M.P. Jennings, M.A. Gidney, S.
Lacelle, J.R. Brisson, A.D. Cox, A. Martin, A.G. Bird, J.C.
Tang, J.C. Richards, and E.R. Moxon. Infect. Immun. 67, 5417
(1999).
55. D. Uhrín and P.N. Barlow. J. Magn. Reson. 126, 148 (1997).
56. W.W. Wakarchuk, D. Watson, F. St Michael, J. Li, Y. Wu, J.R.
Brisson, N.M. Young, and M. Gilbert. J. Biol. Chem. 276,
12 785 (2001).
57. M. Gilbert, J.R. Brisson, M.F. Kawarski, J. Michniewicz, A.M.
Cunningham, Y. Wu, N.M. Young, and W.W. Wakarchuk. J.
Biol. Chem. 275, 3896 (2000).
58. J.R. Brisson, S. Uhrinova, R.J. Woods, M. van der Zwan, H.C.
Jarrell, L.C. Paoletti, D.L. Kasper, and H.J. Jennings. Bio-
chemistry, 36, 3278 (1997).
59. K. Scheffler, J.R. Brisson, R. Weisemann, J.L. Magnani, T.C.
Wong, B. Ernst, and T. Peters. J. Biomol. NMR, 9, 423 (1997).
60. J.O. Duus, C.H. Gotfredsen, and K. Bock. Chem. Rev. 100,
4589 (2000).
61. D. Uhrín and J.R. Brisson. Structure determination of micro-
bial polysaccharides by high resolution NMR spectroscopy. In
NMR in microbiology: theory and applications. Edited by J.N.
Barbotin and J.C. Portais. Horizon Scientific Press.
Wymondham, U.K. 2000. p. 165.
62. G. Kogan and D. Uhrín. Current NMR methods in the struc-
tural elucidation of polysaccharides. In New advances in ana-
lytical chemistry. Edited by Atta-ur-Rahman. Harwood
Academic. Amsterdam. 2000. p. 73.
27. W.W. Wakarchuk, M. Gilbert, A. Martin, Y. Wu, J.R. Brisson,
P. Thibault, and J.C. Richards. Eur. J. Biochem. 254, 626 (1998).
28. A.D. Cox, M.D. Howard, J.R. Brisson, M. van der Zwan, P.
Thibault, M.B. Perry, and T.J. Inzana. Eur. J. Biochem. 253,
507 (1998).
29. D.G. Davis and A. Bax. J. Am. Chem. Soc. 107, 9197 (1985).
© 2002 NRC Canada

Brisson et al.
963
63. D. Uhrín. Concatenation of polarization transfer steps in 1D
homonuclear chemical shift correlated experiments. Applica-
tion to oligo- and polysaccharides. In Methods for structure
elucidation by high-resolution NMR. Edited by Gy. Batta, K.E.
Kövér, and C. Szántay, Jr. Elsevier Science B. V. Amsterdam.
1997. p. 51.
72. S.A. Wacowich-Sgarbi, C.C. Ling, A. Otter, and D.R. Bundle.
J. Am. Chem. Soc. 123, 4362 (2001).
73. H. Masoud, M.B. Perry, J.R. Brisson, D. Uhrín, and J.C. Rich-
ards. Can. J. Chem. 72, 1466 (1994).
74. W.B. Severn, R.F. Kelly, J.C. Richards, and C. Whitfield. J.
Bacteriol. 178, 1731 (1996).
64. C. Roumestand, C. Delay, J.A. Gavin, and D. Canet. Magn.
Reson. Chem. 37, 451 (1999).
75. O. Holst, U. Zahringer, H. Brade, and A. Zamojski.
Carbohydr. Res. 215, 323 (1991).
65. A.D. Cox, J.R. Brisson, V. Varma, and M.B. Perry. Carbohydr.
Res. 290, 43 (1996).
76. J.C. Sowden and D.R. Strobach. J. Am. Chem. Soc. 82, 954
(1960).
66. G.I. Birnbaum, R. Roy, J.R. Brisson, and H.J. Jennings. J.
Carbohydr. Chem. 6, 17 (1987).
77. R.K. Hulyalkar, J.K.N. Jones, and M.B. Perry. Can. J. Chem.
41, 1490 (1963).
67. E. Altman, J.R. Brisson, and M.B. Perry. Eur. J. Biochem. 170,
185 (1987).
78. D. Uhrín, J.R. Brisson, L.L. MacLean, J.C. Richards, and
M.B. Perry. J. Biomol. NMR, 4, 615 (1994).
68. K. Bock and C. Pedersen. Adv. Carbohydr. Chem. Biochem.
41, 27 (1983).
79. T. Peters, B. Meyer, P. Stuike, R. Somorjai, and J.R. Brisson.
Carbohydr. Res. 238, 49 (1993).
69. P.-E. Jansson, L. Kenne, and G. Widmalm. Carbohydr. Res.
188, 169 (1989).
80. I. Tvaroska and S. Pérez. Carbohydr. Res. 149, 389 (1986).
81. Y. Wang and R.I. Hollingsworth. Biochemistry, 35, 5647 (1996).
70. M.D. Potter and R.Y. Lo. FEMS Microbiol. Lett. 129, 75 (1995).
71. I.M. Helander, B. Lindner, H. Brade, K. Altmann, A.A.
Lindberg, E.T. Rietschel, and U. Zahringer. Eur. J. Biochem.
177, 483 (1988).
© 2002 NRC Canada