The structures of the lipooligosaccharide and capsule polysaccharide
of
Campylobacter jejuni
genome sequenced strain NCTC 11168
Frank St. Michael, Christine M. Szymanski, Jianjun Li, Kenneth H. Chan, Nam Huan Khieu,
Suzon Larocque, Warren W. Wakarchuk, Jean-Robert Brisson and Mario A. Monteiro
Institute for Biological Sciences, National Research Council of Canada, Ottawa, Canada
Campylobacter jejuni infections are one of the leading causes
of human gastroenteritis and are suspected of being a pre-
cursor to Guillain–Barre
´
and Miller–Fisher syndromes.
Recently, the complete genome sequence of C. jejuni NCTC
11168 was described. In this study, the molecular structure of
the lipooligosaccharide and capsular polysaccharide of
C. jejuni NCTC 11168 was investigated. The lipooligosac-
charide was shown to exhibit carbohydrate structures anal-
ogous to the GM1a and GM2 carbohydrate epitopes of
human gangliosides (shown below):
The high M
r
capsule polysaccharide was composed of
b-
D
-Ribp, b-
D
-GalfNAc, a-
D
-GlcpA6(NGro), a uronic
acid amidated with 2-amino-2-deoxyglycerol at C-6, and
6-O-methyl-

In the pioneering studies carried out by Aspinall and
coworkers on the cell-surface carbohydrates from
Campylobacter species, it was observed that insoluble gels
from phenol-water extractions of bacterial cells yielded
mainly low M
r
LOS, with core oligosaccharide linked to
lipid A, and the aqueous phases from such extractions
furnished high M
r
glycans with extended polymers with
no attachment to lipid A as seen in the teichoic acid-like
P/PEtn
GM1a GM2 fl
6
b-Gal-(1fi3)- b-
D
-GalNAc-(1fi4)-b-
D
-Gal-(1 fi3)-b-
D
-Gal-(1fi3)-
L
-a-
D
-Hep-(1fi3)-
L
-a-
D
-Hep-(1fi5)Kdo

@
0
B
B
B
B
@
1
C
C
C
C
A
Correspondence to J R. Brisson, Institute for Biological Sciences,
National Research Council of Canada, Ottawa, Canada, K1A 0R6.
Fax: + 1 613 952 9092, Tel.: + 1 613 990 3244,
E-mail: ; M. Monteiro, Wyeth Vaccines
Research, 211 Bailey Road, West Henrietta, NY, 14586, USA.
Fax: + 1 585 273 751, Tel.: + 1 585 273 7667,
E-mail:
Abbreviations: CE, capillary electrophoresis; ESI-MS, electron spray
ionization mass spectrometry; FAB, fast-atom bombardment;
HMBC, heteronuclear multiple bond coherence; HMQC, heteronu-
clear multiple quantum coherence; HR-MAS, high-resolution magic
angle spinning; HSQC, heteronuclear single quantum coherence;
KmR, kanamycin resistance; LOS, lipooligosaccharide; LPS, lipo-
polysaccharide; OS, oligosaccharide.
Dedication: The authors would like to dedicate this manuscript to
Professor Gerald Aspinall.
(Received 3 July 2002, revised 16 August 2002,

case of human enteritis [18] and later sequenced by Parkhill
et al.[8].E. coli DH10B (Invitrogen) was used as the host
for the cloning experiments. Plasmid pPCR-Script Amp
(Stratagene) was used as the cloning vector.
Media and growth conditions
C. jejuni NCTC 11168 was routinely grown on Mueller
Hinton agar (Difco) under microaerophilic conditions at
37 °C. E. coli clones were grown on Luria S-gal agar
(Sigma) at 37 °C. When appropriate, antibiotics were added
to the following final concentrations: kanamycin,
30 lgÆmL
)1
and ampicillin, 150 lgÆmL
)1
.
Generation of LOS and capsule
Campylobacter biomass was harvested from overnight
liquid cultures by centrifugation. Carbohydrates were
isolated by the hot water/phenol extraction of bacterial
cells [19] as a gel-like pellet upon ultracentrifugation of the
aqueous phase. The LOS pellet was lyophilized, then
purified on a column of Bio-Gel P-2 (1 cm · 100 cm) with
water as eluent. Some of the LOS preparation was then
treated with 1% acetic acid at 100 °C for 1 h with
subsequent removal of the insoluble lipid A by centrifuga-
tion (5000 g) to yield the core oligosaccharide (OS).
The supernatant from ultracentrifugation was purified on
Bio-Gel P-2 with water as eluent and lyophilized to obtain
the capsule polysaccharide (PS-1).
Sugar composition and methylation linkage analysis

1 : 3 and 3 kV as the tip voltage.
Smith degradation
The polysaccharide sample ( 5 mg) was oxidized with
40 m
M
sodium metaperiodate in 0.1
M
sodium acetate at
4 °C for 72 h [23]. The product was isolated on a Bio-Gel
P-2 as per above then reduced with NaBD
4
and acidified
with cation exchange resin (J. T. Baker). The product was
then hydrolyzed with 1
M
trifluoroacetic acid at 45 °C for
1 h, reduced with NaBD
4
and re-acidified. Then, finally the
sample was fractionated on a Bio-Gel P-2 column and
fractions analyzed.
CE-ESI-MS and CE-ESI-MS/MS
A crystal Model 310 capillary electrophoresis (CE)
instrument (AYI Unicam) was coupled to an API 3000
mass spectrometer (Perkin-Elmer/Sciex) via a microIon-
spray interface. A sheath solution (isopropanol/methanol
2 : 1) was delivered at a flow rate of 1 lLÆmin
)1
to a low
dead volume tee (250 lm internal diameter, Chromato-

CTTTCATCATTTTAAACGCTCTT-3¢)andfclR51
(5¢-TACAGCATTGGTAGAAAACTTACAA-3¢). For
construction of the kpsM mutant, gene Cj1448c was PCR
amplified with: kpsMF771 (5¢- TACCGCCGTTAAAGCT
TGTCTATTA-3¢) and kpsMR73B (5¢- TATATATGGGT
AGTTGGGGAGCCTA-3¢). For construction of the
Cj1439c mutant, gene Cj1439c was PCR amplified with:
glfF1081 (5¢-TTTTACAAAATAATAATGCCGATCT-3¢)
and glfR6 (5¢-TGATTATTTAATTGTTGGTTCTGG
A-3¢). The PCR products were ligated to pPCR-Script
Amp according to the manufacturer’s instructions. A blunt-
ended kanamycin resistance (KmR) cassette from pILL600
[24] was inserted into the filled-in BglII restriction site of
Cj1428c to create pCSc28, into the NheI restriction site
of kpsM to create pCSc48, and into the BsaBI restriction
site of Cj1439c to create pCSc39. The orientation of the
KmR cassette was determined by sequencing with the
ckanB primer (5¢-CCTGGGTTTCAAGCATTAG-3¢)
using terminator chemistry and AmpliTaq DNA polym-
erase FS cycle sequencing kits (Perkin Elmer-Applied
Biosystems) and analyzed on an Applied Biosystems 373
DNA sequencer. The mutated plasmid DNA was used for
electroporation into C. jejuni NCTC 11168 [25] and KmR
transformants were characterized by PCR to confirm that
the incoming plasmid DNA had integrated by a double
cross-over event.
Reverse transcriptase-polymerase chain reaction
It has previously been shown that gene insertion of the
Campylobacter KmR cassette in a nonpolar orientation has
no effect on transcription of downstream genes [26].

Nuclear magnetic resonance
NMR experiments were acquired on Varian Inova 600,
500 and 400 MHz spectrometers using a 5-mm triple
resonance probe with the
1
H coil nearest to the sample
and with a Z gradient coil. All measurements were made
at 25 °C on 2–5 mg of sample dissolved in 0.6 mL of
D
2
O, pH 6–7. Experiments in 90% H
2
O were carried
out at pH 4–5. The methyl resonance of acetone was
used as an internal reference at 2.225 p.p.m. for
1
H
spectra and 31.07 p.p.m. for
13
C spectra. Standard
homo and heteronuclear correlated 2D techniques were
used for general assignments: COSY, TOCSY, NOESY,
HMQC or HSQC, HMQC-TOCSY and HMBC.
Selective 1D experiments were performed for the
determination of accurate coupling constants and NOEs
andtoperformthe1Danalogofa3D-TOCSY-
NOESY experiment [33]. High resolution magic angle
spinning (HR-MAS) experiments were performed using
a gradient 4 mm indirect detection nano-NMR probe
(Varian) with a broadband decoupling coil. Proton

Vacuum MD simulation were performed with the
DISCOVER
-3 program from Accelrys Inc. (MSI) using the
AMBER FORCEFIELD
version 1.0–1.6, with Homans’ param-
eters applicable to saccharides [37], on a SGI Indigo 2 Solid
Impact R10000 195 MHz. As before, the (O5–C5–C6–O6)
torsion angle for residue D was restrained. The initial
structure was subjected to a 300-step energy minimization
(BFGS method), followed by 50 ps dynamics simulation at
298 K. Initial velocities were generated from a Maxwell–
Boltzmann distribution, and the temperature was controlled
by the direct velocity scaling method. The Verlet algorithm
of integrating the Newton’s equations of motion was
applied with 1 fs timestep for the simulation where a
distance-dependent dielectric of the form er (r ¼ the
distance between atoms) was used. Distances were extracted
Ó FEBS 2002 LOS and capsule structures of C. jejuni NCTC 11168 (Eur. J. Biochem. 269) 5121
from a trajectory file of 5000 frames stored after each MD
run.
RESULTS
Structural determination of the LOS
The alditol acetate derivatives [20] of
D
-glucose (
D
-Glc),
D
-galactose (
D

D
-GalNAc, trace amounts of 3-substituted
D
-GalNAc,
one unit of 2,3-disubstituted
LD
-Hep, and trace amounts
of 3,4-disubstituted
LD
-Hep. A parallel linkage analysis on
the liberated core OS, after removal of lipid A with 1% acetic
acid, afforded the same sugar linkage types, but in addition it
showed a significant decrease of 3,4-disubstituted
D
-Gal and
a greater amount of 4-substituted
D
-Gal (Table 1).
To gain a quick insight into the overall composition of the
LOS, a series of ESI-MS experiments were performed on
the core OS. The ESI-MS spectrum (Fig. 1a) of the core OS
showed a heterogeneous mixture (Table 2) with the pres-
ence at the reducing-end of the anhydro form of 3-deoxy-
manno-octolusonic acid (Kdo) as a distinct marker. The
primary molecular ion at m/z 1759 [)18 (H
2
O)] correspon-
dedtoacompositionofHex
5
,Hep

obtained by a harsher treatment with 5% acetic acid, to
intentionally remove any acid labile Neu5Ac, yielded the
same primary ions as discussed above, but no ions
containing sialic acid. Taking into account the previous
observed variation between 3,4-disubstituted Gal and
3-substituted Gal in the core OS, before and after mild acid
treatment, and the detection of the acid sensitive sialic acid
in ESI-MS, suggested that Neu5Ac may be attached at O-3
of the 3,4-disubstituted Gal.
The CE-MS/MS spectra for the components having a
total mass of m/z 2050 or precursor ions at m/z 1026 (doubly
protonated) are presented in Fig. 1b. The fragment ions
observed at m/z 1848 and 1760 clearly indicated that one
HexNAc residue and one Neu5Ac residue were present as
terminal units. As shown in the Fig. 1b, the fragment ion of
m/z 1848, arising from the loss of HexNAc, subsequently
loses one hexose (m/z 1686), one Neu5Ac (m/z 1394.5), three
hexoses (m/z 1232, 1070, 908) and finally one heptose
residue (m/z 716). The fragment ion at m/z 366 suggested
that the HexNAc was attached to a Hex unit, and the
fragment ion at m/z 454 was indicative a Hex-Neu5Ac
disaccharide. Moreover, fragment ion m/z 554 suggested the
existence of PEtn-Hep-Kdo moiety. It was also observed
that a minor glycoform, with an extra Hex (composition of
Hex
6
HexNAc
1
PEtn
1

3
)
+
, m/z
1029 (Hex
2
, GalNAc, Neu5Ac)
+
, m/z 1120 (GalNAc, Hex
3
,
Hep)
+
, m/z 1233 (GalNAc, Hex
3
,Neu5Ac)
+
, m/z 1324
(GalNAc, Hex
4
,Hep)
+
, m/z 1407 (Hex
4
,Hep
2
,P)
+
, m/z
1477 (Hex

(Table 1) and from the selective ESI-MS experiments
(Fig. 1, Table 2). The following provisional structural
arrangement for the core OS region can be proposed
(Hex ¼ Glc or Gal):
Table 1. Methylation linkage analysis of C. jejuni NCTC 11168 intact
LOS, core OS and Smith degradation products.
Linkage type LOS OS
Smith degradation
product
Glc-(1fi 22
Gal-(1fi 113
fi3)-Gal-(1fi 1
fi4)-Gal-(1fi Traces < 1 Traces
fi2,3)-Gal-(1fi 11
fi3,4)-Gal-(1fi 1 < 1 Traces
GalNAc-(1fi 1 1 Traces
fi3)-GalNAc-(1fi Traces Traces
fi2,3)-Hep-(1fi 11
– fi3,4)-Hep (1fi Traces Traces
fi3)-Man-(1fi 3
fi5)-3d-Hexitol
a
0.5
fi5)-3d-Hexitol
b
0.5
a,b
Two isomeric forms of 3-deoxy-1,1,2,6-tetra-
2
H-5-O-acetyl-

these two derivatives originated from an extended molecule
that contained a hexose at the nonreducing terminus of
the core {Hex-(1fi4)-GalNAc-(1fi4)[Neu5Ac-(1fi3)]-
Gal… inner core}.Theisomerichexitolderivativeswere
recognized as originating from a modified 5-linked Kdo
termini as seen in all C. jejuni strains. The backbone Gal and
LD
-Hep units are thus joined by 1fi3 linkages and the inner
most
LD
-Hep is linked to O-5 of Kdo. The FAB-MS
Fig. 1. Electron spray ionization-mass spectr-
ometry. C. jejuni NCTC 11168 core OS
showing a heterogeneous mixture (a). CE-MS/
MS (+ ion mode, produces ions of m/z 1026)
analysis of C. jejuni NCTC 11168 core OS (b).
CE-MS/MS (+ ion mode, produces ions of
m/z 1005) analysis of LOS from C. jejuni
NCTC 11168 core OS (c).
P/PEtn
fl
[
D
-Hex-(1fi3)]±-
D
-GalNAc-(1fi3 or 4)-
D
-Gal-(1fi2 or 3)-
D
-Gal-(1fi2 or 3)-

degradation (shown below), one was terminated by a GM2
structure, and the major product was a linear Gal-Man-
Man-3dhex (shown below) backbone:
Two unresolved anomeric resonances, characteristic of
sugars with the mannose configuration, were observed at d
5.207 and d 5.07 in the
1
H nuclear magnetic resonance
spectrum of the Smith degradation product. The same
spectrum also showed one b anomeric resonance at 4.55
(J
1,2
7.2 Hz). The
1
H NMR data just described indicated
that the Man (
LD
-Hep) units possessed an a anomeric
configuration, whereas the 2,3-disubstituted
D
-Gal had a b
anomeric configuration. Therefore, at this time, the follow-
ing structure for the core OS region, where Hex represents
Glc or Gal, was proposed:
The sugar linkage analysis (Table 1) performed on the
core OS suggested the presence of slightly more than one
unit of terminal Gal and two units of terminal Glc. Given
the fact that the hexose at the nonreducing terminus was
only present in trace amounts, as observed by linkage
analysis (traces of 3-substituted GalNAc), ESI-MS and

2,3
2 Hz)], typical of
a-
D
-Man configurations and were thus assigned to the
L
-a-
D
-Hep residues, as were observed in the Smith
degradation
1
H NMR product described above. All other
anomeric resonances detected possessed b anomeric con-
figuration and could be seen at d 4.99 (J
1,2
)7Hz),d 4.88
(J
1,2
)7Hz),d 4.69 (J
1,2
)7Hz),d 4.66 (J
1,2
)7Hz)andd
4.62 (J
1,2
)7 Hz). To situate the side-branch hexoses, a 2D
1
H–
1
H NOESY experiment was performed and conclusive

core region (Fig. 3) composed of basal core OS units, Gal,
LD
-Hep and Glc. The GM2 and GM1a epitopes completed
a core OS similar to that present in C. jejuni serogroup
HS:1 [6]. In addition, the innermost
LD
-Hep was phos-
phorylated by a monoester phosphate or by a 2-amino
ethyl phosphate.
Structure of the capsule polysaccharide
The polysaccharide, obtained from the aqueous phase after
ultracentrifugation, was purified on a Bio-Gel P-2 (PS-1).
Alditol acetate analysis revealed the presence of
D
-Glc,
Table 2. Negative ion ESI-MS data and proposed compositions for
C. jejuni NCTC 11168 core OS and de-O-acylated LOS (masses include
the addition of water)
a
.
Core OS
Observed molecular
mass (Da)
Proposed
structure
1635 ()18) HexNAcÆHex
5
ÆHep
2
ÆKdo

a
Residues used and their molecular mass: Neu5Ac, 291; HexNAc,
203; Hex, 162; Hep, 192; PEtn, 123; P, 79; Kdo, 220.
P/PEtn
fl
[
D
-Hex-(1fi3)]±-
D
-GalNAc-(1fi4)-
D
-Gal-(1fi3)-b-
D
-Gal-(1fi3)-
L
-a-
D
-Hep-(1fi3)-L-a-
D
-Hep-(1fi5)-Kdo
32 2 4
›› › ›
21 1 1
Neu5Ac
D
-Hex
D
-Hex
D
-Hex

-gal-
actosamine (
D
-GalNAc). From the alditol acetate analysis
it was observed that PS-1 was slightly contaminated with
LOS, additional efforts at purification were not successful.
The methylation linkage analysis revealed the presence
of 2-substituted
D
-Rib, 4-substituted
D
-GalNAc, and a
terminal heptose unit. No substituted heptose was detected,
and thus the 6-O-Me-Hep was present as a side chain
residue.
For further characterization of the structure of PS-1, the
sample was acid hydrolyzed under mild conditions (1
M
HCl, 100 °C for 5 min) and the resultant hydrolysate was
purified on a Bio-Gel P-2 column. The sample was then
analyzed by CE-ESI-MS (Fig. 4a) and gave rise to two
components having a mass of 791 as the major product and
a minor mass of 762. The MS/MS spectra of m/z 791
(Fig. 4b) revealed fragments m/z 588 and 585. This showed
the loss of either 6-O-Me-Hep or GalNAc, respectively; thus
illustrating that they were terminal units in this OS from
acid hydrolysis of PS-1. This finding was consistent with
the previous observation in the linkage analysis that the
6-O-Me-Hep was a terminal unit. The fragment ion at 382
arose from the loss of both the 6-O-Me-Hep and GalNAc,

mutant was denoted as PS-2 (Fig. 6). The proton spectrum
of PS-2 was more homogeneous and had sharper lines
than the proton spectrum of PS-1. As the spectrum of PS-2
was less complex than the spectrum of the native PS-1,
its structural determination was first undertaken.
NMR methods, as outlined before [33,38,39], were used
for the structural determination of polysaccharides 1 and
Fig. 2. FAB-MS spectra of the methylated
C. jejuni NCTC 11168 core OS.
Ó FEBS 2002 LOS and capsule structures of C. jejuni NCTC 11168 (Eur. J. Biochem. 269) 5125
Table 3. Interpretation of m/z ions in the FAB-MS spectrum of the methylated core OS from C. j ejuni NCTC 11168.
Primary m/z ion Secondary m/z ion Double cleavage m/z ion Proposed structure
260 228 (260–32) GlcNAc
+
376 344 (376–32) Neu5Ac
+
464 432 (464–32) Gal-(1fi3)-GalNAc
+
547 Glc-(1fi4)-Hep
+
›
P
668 GalNAc-(1fi4)-Gal-(1fi3)-Gal
+
826 GalNAc-(1fi4)-Gal
+
3
›
2
Neu5Ac

1407 P
fl
Gal-(1fi3)-Hep-(1fi3)-Hep
+
22 4
›› ›
11 1
Gal Glc Glc
1477 PEtn
fl
Gal-(1fi3)-Hep-(1fi3)-Hep
+
224
›› ›
111
GalGlc Glc
1685 GalNAc-(1fi4)-Gal-(1fi3)-Gal-(1fi3)-Hep
+
322
›››
211
Neu5Ac Gal Glc
5126 F. St. Michael et al. (Eur. J. Biochem. 269) Ó FEBS 2002
2 obtained from C. jejuni. 1D selective NMR methods
were also used to characterize individual components [33].
1D-TOCSY experiments were used to detect the coupled
spin systems and spin simulation was used to obtain
accurate coupling constants (Figs 7a–d). HMQC was used
to assign CH, CH
2

+
3224
››› ›
211 1
Neu5Ac Gal Glc Glc
Atomic mass units of the residues discussed above: Glc/Gal GalNAc Hep Neu5Ac P PEtn
Terminal 219 260 263 376 94 166
Monosubstituted 204 245 248
Disubstituted 189 233
Fig. 3. The complete structure of C. jejuni
NCTC 11168 LOS.
Fig. 4. CE-ESI-MS and CE-MS/MS analysis. CE-ESI-MS of PS-1
after acid hydrolysis (1
M
HCl, 100 °C for 5 min) and Bio-Gel P-2
purification(a),CE-MS/MSofm/z 791 (b), CE-MS/MS of m/z 762 (c).
Fig. 5. Proton spectra of C. jejuni whole cells and isolated polysaccha-
rides. HR-MAS spectrum of C. jejuni whole cells (a) and its isolated
polysaccharide (b). HR-MAS spectrum of C. jejuni Cj1428c mutant
whole cells (c) and its isolated polysaccharide (d). The anomeric
resonances are labeled according to the structures shown in Fig. 6.
Ó FEBS 2002 LOS and capsule structures of C. jejuni NCTC 11168 (Eur. J. Biochem. 269) 5127
As the absolute configuration of the sugars was known from
the chemical analysis, from a comparison of chemical shifts
and coupling constants with those of monosaccharides,
residue A was assigned as b-
D
-Ribf, residue B as the amide
of a-
D

2
,Rib,GlcA)
+
, m/z 1101 (GalNAc
2
,Rib
2
,
GlcA)
+
, m/z 1620 (GalNAc
3
,Rib
2
,GlcA
2
)
+
, m/z 1781
Fig. 6. Structure of polysaccharides PS-1 and PS-2 from C. jejuni,and
labeling of the residues and atoms. Residue A is b-
D
-Ribf,residueBis
the amide of a-
D
-GlcpA with ethanolamine at C-6 for PS-2 and with
2-amino-2-deoxyglycerol at C-6 for PS-1, residue C is b-
D
-GalfNAc,
and residue D is

and m/z 2301 (GalNAc
4
,Rib
3
,
GlcA
3
)
+
.
Similarly to PS-2, the 1D-TOCSY experiments were used
to identify the spin systems for each residue and pendant
groups of PS-1. The 1D-TOCSY for anomeric resonance of
residue D is shown in Fig. 9a. The HMQC spectrum was
used to assign the
13
C resonances (Fig. 9g). Residues A, B
and C were found to be the same as in PS-2 except for the
pendant group at C-6B. The sequence of the sugars for PS-1
was established from detection of the inter-residue NOE
between the anomeric resonance and aglycon resonance.
Because the (H-1A, H-5C) (H-1C, H-4B) (H-1B, H-2A)
NOEs were present, the polymeric sequence (-A-C-B-)
n
was
thus found to be the same as in PS-2. Residue D was found
to be linked at C-3 of residue B from the presence of the
(H-1D, H-3B) NOE. From the (H-7B, C-8,9B) HMBC
correlations (Fig. 9h), the 1D-TOCSY for H-9B, and the
NOEs for the NH resonances at C-6B (results not shown),

C
, ± 0.01 p.p.m. for d
H
,and±0.5HzforJ.J-values were obtained from the spin-simulated spectra. Vicinal
3
J values are
positive and geminal
2
J values are negative.
Atom
PS-2 PS-1
Type d
C
d
H
J Type d
C
d
H
1 A CH 105.9 5.37 1.0 CH 106.0 5.36
2 A CH 81.1 4.18 5.0 CH 81.2 4.22
3 A CH 71.0 4.26 7.2 CH 70.8 4.32
4 A CH 84.0 4.12 2.6 CH 84.0 4.13
5A CH
2
63.2 3.86
3.67
6.5
)12.3
CH

6B NH 8.51 5.7 NH 8.33
1C CH 106.8 4.87 2.0 CH 105.2 5.05
2C CH 64.0 4.14 4.5 CH 63.0 4.08
3C CH 76.5 4.17 7.0 CH 74.7 4.24
4C CH 82.9 4.22 3.2 CH 82.1 4.24
5C CH 77.6 3.96 4.0 CH 77.9 3.89
6C CH
2
62.2 3.80
3.75
6.7
)12.3
CH
2
62.0 3.81
3.77
7C C¼O 174.9 C¼O 175.2
8C CH
3
23.0 2.02 CH
3
23.0 2.05
2C NH 8.29 NH 8.21
1D CH 98.2 5.60
2D CH 72.3 3.52
3D CH 73.8 3.74
4D CH 70.4 3.57
5D CH 72.1 4.15
6D CH 79.6 3.80
7D CH

NOEs for H-3D. (g) HMQC spectrum and
assignment of
13
C resonances. (h) HMBC
spectra for location of -NH-CH-(CH
2
OH)
2
substituent on residue B and OMe group on
residue D. (i) NOESY spectrum and deter-
mination of the sequence from inter-residue
NOE correlations.
5130 F. St. Michael et al. (Eur. J. Biochem. 269) Ó FEBS 2002
determined to be glycero-a-gluco-heptopyranose. The
NMR data for PS-1 are given in Table 4.
The absolute configuration of residues A, B and C could
be determined from MS based methods [21,41]. The
absolute configuration of residue D was determined by
NOEs (Fig. 9) in conjunction with molecular modeling.
This method has been previously used to establish the
absolute configuration of sugars in cases where conven-
tional methods are not amenable [42]. The use of selective
experiments, especially the 1D-TOCSY-NOESY, permitted
the detection of NOEs from proton resonances within the
ring. This allowed NOE constraints to be obtained which
cannot be otherwise determined from 2D experiments [33].
NOEs are highly dependent of interproton distances
(r · 10
)6
) and thus on the conformation of the molecule.

is due to flexibility, mainly about the glycosidic bonds. Only
for residue D having the
D
-glycero-
L
-gluco-heptopyranose
absolute configuration are short inter-residue interproton
distances possible, consistent with the observed NOEs.
Average distances obtained from a Metropolis Monte-
Carlo calculation gave similar results. Minimum energy
conformers from the Metropolis Monte-Carlo calculations
are shown in Fig. 10. When residue D has the
L
-glycero-
D
-
gluco-heptopyranose absolute configuration, the range of
inter-residue interproton distances being sampled are not
consistent with the observed NOEs, especially for the 5D-
2C NOE.
Genetic analysis of the capsule polysaccharide
Using gene mutation to manipulate the capsular structure in
combination with the use of HR-MAS NMR, we were able
to compare the complex capsular carbohydrate structure of
wild-type NCTC 11168 with the simpler capsule of the
Cj1428c mutant (Figs 5c and 11a). Cj1428c is homologous
to fcl of E. coli whose product is involved in the conversion
of GDP-4-keto-6-deoxy-
D
-mannose to GDP-4-keto-6-

D
-glycero-
L
-gluco-heptopyranose in (a) and an
L
-glycer o-
D
-gluco-
heptopyranose in (b). Residue B was modeled as a glucuronic acid. The
hydroxyl protons are not shown and the exocyclic chain of residue C is
not shown in (a). The absolute configuration of residue D was estab-
lished to be
D
-glycero-
L
-gluco-heptopyranose, consistent with the
5D-77¢D NOEs and the inter-residue 3D-2C, 3D-4C, 5D-2C, 5D-4C
NOEs.
Table 5. Experimental NOEs and interproton distances (average
± standard deviation) obtained from molecular dynamics for the BCD
trisaccharide with different absolute configurations for the heptose
residue D.
H–H NOE
a
DL
-Hep
LD
-Hep
r (A
˚

GlcNAc [40]. Comparison of NCTC 11168 and HS:19
capsular loci sequences demonstrates that HS:19 also
contains a Cj1441c homolog (unpublished results).
Analysis of the genome sequence of NCTC 11168
demonstrates that the strain contains two gene clusters
involved in heptose biosynthesis [8]. One cluster, located in
the LOS gene cluster, is similar to that found in other Gram-
negative bacteria and is necessary for the biosynthesis of the
L
-glycero-
D
-manno-heptopyranoses commonly found in
LPS of many Gram-negative bacteria. The second cluster,
found in the capsule gene cluster, is homologous to
genes recently identified in the Gram-positive, A. thermo-
aerophilus, involved in the biosynthesis of
D
-glycero-
D
-manno-heptopyranose in the S-layer glycoprotein [46].
Campylobacter species have been shown to produce both
LD
-heptoses in their LOS cores and other heptose isomers in
their capsular polysaccharides (for examples see [6]).
Therefore, it was not surprising to find
L
-glycero-
D
-manno-
heptopyranose in the LOS core and 6-O-Me-Hep in the

(results not shown). We have isolated a natural phase
variant that produces the capsular polysaccharide without
the methyl modification and this variant is currently under
investigation.
DISCUSSION
The LOS outer core region of C. jejuni NCTC 11168 has
been found to consist predominately of a GM2 ganglioside
mimic, but also in smaller amounts, of a GM1a ganglioside
mimic lacking the terminal b-1,3-linked galactose (Fig. 3).
In the strain investigated here, wlaN encodes the b-1,3-
galactosytransferase which adds the terminal Gal to
GalNAc. Linton et al. showed that the homopolymeric
C-tract within wlaN was subject to frequent on/off switching
through a mechanism of slipped-strand mispairing during
DNA replication [17]. The authors predominantly observed
thein-frameversionofwlaN while our laboratory observes
Fig. 11. Mutagenesis and analysis of capsule mutants of C. jejuni NCTC 11168. (a) Schematic of the capsule gene cluster with gray arrows
representing genes mentioned in the text. The kps genes involved in capsule transport and assembly are shown along with possible hypervariable
regions within the cluster (due to homopolymeric C tracts or a 21-bp repeat, R) that may be responsible for the structural variability described. The
constructs used in this study along with the direction of the KmR cassette insertion are shown. (b) Silver-stained deoxycholate-PAGE of proteinase
K whole cell digests of wild-type NCTC 11168 and isogenic mutants. Lane 1: wild-type NCTC 11168; lane 2: kpsM mutant; lane 3: 1428c mutant;
lane 4: 1439c mutant. The capsular repeat is indicated by the arrow.
5132 F. St. Michael et al. (Eur. J. Biochem. 269) Ó FEBS 2002
mostly the truncated version of the gene (results not shown).
Examination of LOS cores isolated from growth of single
colonies of our NCTC 11168 by deoxycholate-PAGE and
MS yielded structures with and without the terminal
galactose (results not shown). These results demonstrate
that the population analyzed contained a mixture of GM1a
and GM2 core types with the GM2 core type mimic

similar to group II/III capsule loci with genes involved in
capsule biosynthesis being flanked by the kps genes involved
in the transport of the sugar polymer (Fig. 11a) [9–11]. Class
II/III capsules have been shown to be attached to the
membrane by glycerophosphate anchors [9]. However, we
were unable to conclusively demonstrate the linkage of the
capsule polysaccharide to its membrane anchor and are
currently trying to determine the nature of the membrane
linkage.
It is currently unknown whether the polysaccharide that
we describe is polymerized by a processive or blockwise-
dependent method, although the ability to visualize a ladder
by silver staining (Fig. 11b) would suggest block transfer of
sugars differing by increments of one repeat unit [51]. Also,
mutation of Cj1439c, which we believe is involved in
GalfNAc biosynthesis, resulted in an acapsular phenotype
again suggesting a blockwise polymerization mechanism, in
contrast to the processive polymerization described for
group II/III capsular polysaccharides [9]. Deoxycholate-
PAGE (Fig. 11b) and HR-MAS NMR (Fig. 5c) analysis of
the Cj1428c mutant demonstrated that the capsule repeats
were still produced in the absence of the 6-O-Me-Hep
branch. These results suggest that NCTC 11168 possesses a
promiscuous polymerase which assembles the carbohydrate
repeats even in the absence of the branch sugar. This
observation is similar to that described for the E. coli K4
chondroitin polymerase which assembles the capsular
polysaccharide backbone of GlcAb(1–3)-GalNAc b(1–4)
even in the absence of the b-fructose branch at position C-3
of the glucuronosyl residue [52].

on sugars has not been described except for its presence in
phosphoethanolamine, a common variable modification on
bacterial lipid As, including that of C. jejuni.
In summary, we have presented the complete structures
of the LOS and capsule polysaccharide of the genome
sequenced strain C. jejuni NCTC 11168. The outer core
LOS of NCTC 11168 has structural homology with the
human gangliosides, GM2 and GM1a. As demonstrated
previously in NCTC 11168 [16,17,48] and in 81–176 [12],
C. jejuni can exhibit variable ganglioside mimics due to
variation of the LOS core. The mimicry of human cell-
surface glycolipids and glycoproteins appears to be a general
trend of mucosal pathogens such as Haemophilus, Neisse-
ria, Helicobacter and Campylobacter (for reviews see
[7,61]). The peculiar property of producing structures
analogous to cell-surface molecules of the host and having
the ability to vary these structures plays an important role in
pathogenesis and survival within the environment.
In the capsular polysaccharide of NCTC 11168, we have
identified two new sugars that have not been described
before. As far as we are aware, this is the first report of
GalfNAc in the literature. Molecular modeling also dem-
onstrated that the absolute configuration of the heptopyr-
anose is
D
-glycero-
L
-gluco-heptopyranose. This is the first
demonstration of the
L

1. Allos, B.M. (2001) Campylobacter jejuni infections: update on
emerging issues and trends. Clin. Infect. Dis. 32, 1201–1206.
2. Altekruse, S.F., Stern, N.J., Fields, P.I. & Swerdlow, D.L. (1999)
Campylobacter jejuni ) an emerging foodborne pathogen. Emerg.
Infect. Dis. 5, 28–35.
3. Kaldor, J. & Speed, B.R. (1984) Guillain–Barre
´
syndrome and
Campylobacter jejuni:aserologicalstudy.Br.Med.J.Clin.Res.
288, 1867–1870.
4. Rhodes, K.M. & Tattersfield, A.E. (1982) Guillain–Barre
´
syn-
drome associated with Campylobacter infection. Br.Med.J.Clin.
Res. 285, 173–174.
5. Jacobs, B.C., Endtz, H., van der Meche, F.G., Hazenberg, M.P.,
Achtereekte, H.A. & van Doorn, P.A. (1995) Serum anti-GQ1b
IgG antibodies recognize surface epitopes on Campylobacter jejuni
from patients with Miller Fisher syndrome. Ann. Neurol. 37,260–
264.
6. Moran, A.P., Penner, J.L. & Aspinall, G.O. (2000) Campylobacter
lipopolysaccharides. In Campylobacter (Nachamkin,I.&Blaser,
M.J., eds), pp. 241–257. American. Society for Microbiology,
Washington, D.C.
7. Moran, A.P. & Prendergast, M.M. (2001) Molecular mimicry in
Campylobacter jejuni and Helicobacter pylori lipopolysaccharides:
contribution of gastrointestinal infections to autoimmunity.
J. Autoimmun. 16, 241–256.
8. Parkhill, J., Wren, B.W., Mungall, K., Ketley, J.M., Churcher, C.,
Basham, D., Chillingworth, T., Davies, R.M., Feltwell, T.,

lipopolysaccharides of Campylobacter fetus serotype A strains.
Infect. Immun. 69, 7596–7602.
15. Guerry, P., Ewing, C.P., Hickey, T.E., Prendergast, M.M. &
Moran, A.P. (2000) Sialylation of lipooligosaccharide cores affects
immunogenicity and serum resistance of Campylobacter jejuni.
Infect. Immun. 68, 6656–6662.
16. Gilbert, M., Brisson, J.R., Karwaski, M.F., Michniewicz, J.,
Cunningham, A.M., Wu, Y., Young, N.M. & Wakarchuk, W.W.
(2000) Biosynthesis of ganglioside mimics in Campylobacter jejuni
OH4384. Identification of the glycosyltransferase genes, enzymatic
synthesis of model compounds, and characterization of nanomole
amounts by 600-MHz
1
Hand
13
C NMR analysis. J. Biol. Chem.
275, 3896–3906.
17. Linton, D., Gilbert, M., Hitchen, P.G., Dell, A., Morris, H.R.,
Wakarchuk,W.W.,Gregson,N.A.&Wren,B.W.(2000)Phase
variation of a beta-1,3 galactosyltransferase involved in generation
of the ganglioside GM1-like lipo-oligosaccharide of Campylo-
bacter jejuni. Mol. Microbiol. 37, 501–514.
18. Ahmed, I.H., Manning, G., Wassenaar, T.M., Cawthraw, S. &
Newell, D.G. (2002) Identification of genetic differences between
two Campylobacter jejuni strains with different colonization po-
tentials. Microbiology 148, 1203–1212.
19. Westphal, O. & Jann, K. (1965) Bacterial lipopolysaccharide.
Extraction with phenol–water and further applications of the
procedure. Methods Carbohydr. Chem. 5, 88–91.
20. Sawardeker, D.G., Sloneker, J.H. & Jeanes, A. (1965) Quantita-

29. Komuro, T. & Galanos, C. (1988) Analysis of Salmonella lipo-
polysaccharides by sodium deoxycholate–polyacrylamide gel
electrophoresis. J. Chromatogr. 450, 381–387.
30. Tsai, C.M. & Frasch, C.E. (1982) A sensitive silver stain for de-
tecting lipopolysaccharides in polyacrylamide gels. Anal. Biochem.
119, 115–119.
31. Fomsgaard, A., Freudenberg, M.A. & Galanos, C. (1990) Mod-
ification of the silver staining technique to detect lipopoly-
saccharide in polyacrylamide gels. J. Clin. Microbiol. 28, 2627–
2631.
32. Guard-Petter, J., Lakshmi, B., Carlson, R. & Ingram, K. (1995)
Characterization of lipopolysaccharide heterogeneity in Salmo-
nella enteritidis by an improved gel electrophoresis method. Appl.
Environ. Microbiol. 61, 2845–2851.
33. Uhrı
´
n, D. & Brisson, J.R. (2000) Structure determination of
microbial polysaccharides by high resolution NMR spectroscopy.
In NMR in Microbiology: Theory and Applications (Barbotin, J.N.
& Portais, J.C., eds), pp. 165–210. Horizon Scientific Press,
Wymondham, UK.
34. Jachymek, W., Niedziela, T., Petersson, C., Lugowski, C., Czaja,
J. & Kenne, L. (1999) Structures of the O-specific polysaccharides
from Yokenella regensburgei (Koserella trabulsii)strainsPCM
2476, 2477, 2478, and 2494: high-resolution magic-angle spinning
NMR investigation of the O-specific polysaccharides in native
lipopolysaccharides and directly on the surface of living bacteria.
Biochemistry 38, 11788–11795.
35. Peters,T.,Meyer,B.,Stuike,P.,Somorjai,R.&Brisson,J.R.
(1993) A Monte Carlo method for conformational analysis of

saccharide from lipopolysaccharide of Vibrio cholerae strain H11
(non-O1). Eur. J. Biochem. 210, 491–498.
42. Sadovskaya, I., Brisson, J.R., Khieu, N.H., Mutharia, L.M. &
Altman, E. (1998) Structural characterization of the lipopoly-
saccharide O-antigen and capsular polysaccharide of Vibrio ordalii
serotype O: 2. Eur. J. Biochem. 253, 319–327.
43. Andrianopoulos, K., Wang, L. & Reeves, P.R. (1998) Identifica-
tion of the fucose synthetase gene in the colanic acid gene cluster of
Escherichia coli K-12. J. Bacteriol. 180, 998–1001.
44. Nassau,P.M.,Martin,S.L.,Brown,R.E.,Weston,A.,Monsey,
D., McNeil, M.R. & Duncan, K. (1996) Galactofuranose bio-
synthesis in Escherichia coli K-12: identification and cloning of
UDP-galactopyranose mutase. J. Bacteriol. 178, 1047–1052.
45. Sieberth, V., Rigg, G.P., Roberts, I.S. & Jann, K. (1995) Expres-
sion and characterization of UDPGlc dehydrogenase (KfiD),
which is encoded in the type-specific region 2 of the Escherichia coli
K5 capsule genes. J. Bacteriol. 177, 4562–4565.
46. Kneidinger, B., Graninger, M., Puchberger, M., Kosma, P. &
Messner, P. (2001) Biosynthesis of nucleotide-activated
D
-glycero-
D
-manno-heptose. J. Biol. Chem. 276, 20935–20944.
47. Naess, V. & Hofstad, T. (1984) Chemical studies of partially
hydrolysed lipopolysaccharides from four strains of Campylo-
bacter jejuni and two strains of Campylobacter coli. J. Gen.
Microbiol. 130, 2783–2789.
48. Gilbert, M., Karwaski, M.F., Bernatchez, S., Young, N.M.,
Taboada, E., Michniewicz, J., Cunningham, A.M. & Wakarchuk,
W.W. (2002) The genetic bases for the variation in the lipo-oli-

phosphorylcholine moiety of commensal Neisseria lipopoly-
saccharide. Mol. Microbiol. 43, 437–448.
56. Kooistra, O., Luneberg, E., Knirel, Y.A., Frosch, M. & Zahringer,
U. (2002) N-Methylationinpolylegionaminicacidisassociated
with the phase-variable epitope of Legionella pneumophila ser-
ogroup 1 lipopolysaccharide. Identification of 5-(N,N-dimethyla-
cetimidoyl) amino and 5-acetimidoyl (N-methyl) amino-7-
acetamido-3,5,7,9-tetradeoxynon-2-ulosonic acid in the O-chain
polysaccharide. Eur. J. Biochem. 269, 560–572.
57. Forsberg, L.S., Bhat, U.R. & Carlson, R.W. (2000)
Structural characterization of the O-antigenic polysaccharide of
the lipopolysaccharide from Rhizobium etli strain CE3. A unique
O-acetylated glycan of discrete size, containing 3-O-methyl-
6-deoxy-
L
-talose and 2,3,4-tri-O-methyl-1 fucose. J. Biol. Chem.
275, 18851–18863.
58. Hisatsune,K.,Kondo,S.,Isshiki,Y.,Iguchi,T.&Haishima,Y.
(1993) Occurrence of 2-O-methyl-N-(3-deoxy-
L
-glycero-tetronyl)-
D
-perosamine (4-amino-4,6-dideoxy-
D
-manno-pyranose) in lipo-
polysaccharide from Ogawa but not from Inaba O forms of O1
Vibrio cholerae. Biochem. Biophys. Res. Commun. 190, 302–307.
59. Scha
¨
ffer, C., Wugeditsch, T., Kahlig, H., Scheberl, A., Zayni, S. &