The HS:19 serostrain of Campylobacter jejuni has a
hyaluronic acid-type capsular polysaccharide with a
nonstoichiometric sorbose branch and O-methyl
phosphoramidate group
David J. McNally, Harold C. Jarrell, Nam H. Khieu, Jianjun Li, Evgeny Vinogradov,
Dennis M. Whitfield, Christine M. Szymanski and Jean-Robert Brisson
Institute for Biological Sciences, National Research Council of Canada, Ottawa Ontario, Canada
Campylobacter jejuni is one of the leading causes of
human gastroenteritis and surpasses Salmonella,Shig-
ella and Escherichia in some regions as the primary
cause of gastrointestinal disease [1–3]. There is also a
convincing body of evidence linking C. jejuni infections
to the onset of Guillain–Barre
´syndrome [4–11].
Although of relatively rare occurrence, this syndrome
is the most common cause of acute neuromuscular
paralysis since the eradication of polio. It is character-
ized by weakness in the limbs and respiratory muscles,
with paralysis generally occurring 1–3 weeks after
infection [12,13]. Penner’s passive haemagglutination
Keywords
Campylobacter jejuni; capsular
polysaccharide; high-resolution magic angle
spinning (HR-MAS) NMR; phosphoramidate;
sorbose
Correspondence
J.-R. Brisson, Institute for Biological
Sciences, National Research Council of
Canada, 100 Sussex Drive, Ottawa Ontario,
Canada, K1A 0R6
Fax: +1 613 9529092
Tel: +1 613 9903244
E-mail: jean-robert.brisson@nrc-cnrc.gc.ca
(Received 3 April 2006, revised 2 June
2006, accepted 29 June 2006)
doi:10.1111/j.1742-4658.2006.05401.x
A recent study that examined multiple strains of Campylobacter jejuni
reported that HS:19, a serostrain that has been associated with the onset of
Guillain–Barre
´syndrome, had unidentified labile, capsular polysaccharide
(CPS) structures. In this study, we expand on this observation by using
current glyco-analytical technologies to characterize these unknown groups.
Capillary electrophoresis electrospray ionization MS and NMR analysis
with a cryogenically cooled probe (cold probe) of CPS purified using a gen-
tle enzymatic method revealed a hyaluronic acid-type [-4)-b-d-GlcA6NGro-
(1–3)-b-d-GlcNAc-(1-]
n
repeating unit, where NGro is 2-aminoglycerol. A
labile a-sorbofuranose branch located at C2 of GlcA was determined to
have the lconfiguration using a novel pyranose oxidase assay and is the
first report of this sugar in a bacterial glycan. A labile O-methyl phosphor-
amidate group, CH
3
OP(O)(NH
2
)(OR) (MeOPN), was found at C4 of Glc-
NAc. Structural heterogeneity of the CPS was due to nonstoichiometric
glycosylation with sorbose at C2 of GlcA and the nonstoichiometric, vari-
ably methylated phosphoramidate group. Examination of whole bacterial
cells using high-resolution magic angle spinning NMR revealed that the
MeOPN group is a prominent feature on the cell surface for this sero-
strain. These results are reminiscent of those in the 11168 and HS:1 strains
and suggest that decoration of CPS with nonstoichiometric elements such
as keto sugars and the phosphoramidate is a common mechanism used by
this bacterium to produce a structurally complex surface glycan from a lim-
ited number of genes. The findings of this work with the HS:19 serostrain
now present a means to explore the role of CPS as a virulence factor in
C. jejuni.
Abbreviations
CPS, capsular polysaccharide; CE-ESI-MS, capillary electrophoresis electrospray ionization mass spectrometry; HMBC, heteronuclear
multiple-bond correlation; HR-MAS NMR, high-resolution magic angle spinning nuclear magnetic resonance spectroscopy; HSQC,
heteronuclear single-quantum coherence; MeOPN, O-methyl phosphoramidate CH
3
OP(O)(NH
2
)(OR).
FEBS Journal 273 (2006) 3975–3989 ª2006 The Authors Journal compilation ª2006 FEBS 3975
assay [14] was used to show that 81% of C. jejuni iso-
lates from patients with Guillain–Barre
´syndrome in
Japan belonged to Penner’s HS:19 serotype. In the
United States, 33% of C. jejuni isolates from such
patients were classified as HS:19 [11,15,16].
It is now widely accepted that the major antigenic
component of Penner’s serotyping system for C. jejuni
is capsular polysaccharide (CPS) [17]. This was not
always the case, as lipopolysaccharide was for many
years thought to be the basis for Penner’s classification
system [8]. Because CPS is the outermost structure on
the bacterial cell, it plays a key role in the interaction
between the pathogen, host, and environment [18] and
is generally thought to be important for bacterial survi-
val and persistence in the environment [19]. By mimick-
ing host cell antigens and through structural variation,
CPSs also convey evasion from host immune responses
and are therefore considered important virulence fac-
tors. CPS production in C. jejuni remained unnoticed
until the genome sequencing of C. jejuni NCTC11168
in 2000 and the identification of genes implicated in
CPS biosynthesis [20]. Since this discovery, genetic, bio-
chemical and microscopy studies have demonstrated
CPS production in different strains of C. jejuni
[17,21,22]. Our laboratory has shown that it is possible
to study CPS directly on the surface of intact C. jejuni
cells using high-resolution magic angle spinning NMR
spectroscopy (HR-MAS NMR) [18,23,24]. On the basis
of its role in epithelial cell invasion, diarrhoeal disease,
serum resistance and maintenance of bacterial cell sur-
face hydrophilicity [17,22], CPS is thought to play a
critical role in pathogenesis for C. jejuni.
The precise mechanisms by which CPS conveys viru-
lence to C. jejuni are poorly understood; however,
structural variation of this surface glycan is emerging
as a possible mechanism [23,24]. The CPSs produced
by C. jejuni are structurally diverse and there are more
than 60 serostrains described for this bacterium, exclu-
ding nontypeable strains, each having a different CPS
structure [25]. For each serostrain, there are often
phase-variable structural modifications such as the
incorporation of methyl, ethanolamine and amino-
glycerol groups on CPS sugars [18,23,24]. The most
unusual of these modifications is the O-methyl phos-
phoramidate CH
3
OP(O)(NH
2
)(OR) (MeOPN), which
is a highly labile phosphorylated structure that was
first described on the GalfNAc CPS sugar for the
11168 strain [18]. By using mild extraction conditions
to purify CPS and HR-MAS NMR to study CPS
in vivo, it was recently shown that the HS:1 serostrain
also expresses this phosphorylated modification on two
labile fructofuranose branches. By placing variable
MeOPN groups on labile branches, the HS:1 strain
was shown to produce a structurally variable and
therefore highly complex CPS despite having a small
CPS biosynthetic locus containing 11 genes [23,25].
Sequencing the CPS biosynthetic regions for several
strains of C. jejuni uncovered evidence for multiple
mechanisms of CPS variation, including exchange of
capsular genes and entire clusters by horizontal trans-
fer, gene duplication, deletion, fusion and contingency
gene variation [25]. Of particular interest, the CPS
locus for the HS:19 serostrain was shown to contain
only 13 genes, including a udg homologue responsible
for producing b-d-GlcA6NGro [25]. This finding corre-
lated well with the CPS structure reported for the
HS:19 serostrain, which was shown to consist of a [-4)-
b-d-GlcANGro-(1–3)-b-d-GlcNAc-(1-]
n
repeating unit
[5,6,12]. The CPS locus of the HS:19 strain was also
shown to contain HS19.07, a homologue of cj1421 in
the 11168 strain, which is speculated to be a MeOPN
transferase responsible for adding MeOPN to CPS
sugars [24]. By examining a partially purified CPS sam-
ple prepared from HS:19 cells, the latter study
observed unidentified labile groups and speculated that
one of these was a MeOPN modification similar to the
one reported for other strains of C. jejuni [18,23–25].
In this study, we thoroughly investigate the complete
CPS structure for the HS:19 serostrain of C. jejuni by
using the latest glyco-analytical technologies to charac-
terize these unknown labile groups. Initially, HR-MAS
NMR was used to examine CPS directly on the surface
of whole HS:19 cells. To study the structure of the
CPS in greater detail, we isolated it using a mild enzy-
matic extraction method that preserves labile groups
[23]. In previous studies [5,6,12], the classic hot
water ⁄phenol extraction method [26] was used, as this
structure was originally thought to be a high-molecu-
lar-mass lipopolysaccharide. High-resolution NMR at
600 MHz (
1
H) with an ultra-sensitive cryogenically
cooled probe (cold probe), and capillary electro-
phoresis electrospray ionization mass spectrometry
(CE-ESI-MS) with in-source collision-induced dissoci-
ation [27] were then used to determine the structure of
purified CPS. Herein we report the complete structure
for the CPS of the HS:19 serostrain of C. jejuni and
discuss the biological significance of these new struc-
tural findings for this organism.
Results
The results generated by HR-MAS NMR and high-
resolution NMR, CE-ESI-MS and chemical ⁄enzymatic
analyses revealed a hyaluronic acid-type CPS with a
[-4)-b-d-GlcA6NGro-(1–3)-b-d-GlcNAc-(1-]
n
repeating
unit (Fig. 1). This finding is in agreement with
Campylobacter jejuni HS: 19 CPS D. J. McNally et al.
3976 FEBS Journal 273 (2006) 3975–3989 ª2006 The Authors Journal compilation ª2006 FEBS
previous studies that examined CPS in the HS:19 sero-
strain of C. jejuni [5,6,12]. However, the complete
structure of the CPS was found to be more complex
because of a nonstoichiometric a-l-sorbose branch
located at C2 of b-d-GlcA and a variably methylated
nonstoichiometric MeOPN group at C4 of b-d-Glc-
NAc.
HR-MAS NMR spectroscopy of whole cells
Examination of CPS on the surface of whole HS:19
cells with HR-MAS NMR revealed multiple signals
originating from the cell surface (Fig. 2). The
1
H
HR-MAS NMR spectrum exhibited broad lines, how-
ever; a doublet characteristic of a MeOPN group at d
H
3.77 p.p.m. was observed to protrude above the broad
CPS signals (Fig. 2A). A scalar coupling of 12.0 Hz
was measured for this doublet, which is in good agree-
ment with
3
J
H,P
couplings determined for the MeOPN
in the 11168 and HS:1 serostrains [18,23,24]. A 1D
31
P
heteronuclear single-quantum coherence (HSQC)
experiment that specifically selects for the MeOPN
group (
31
P decoupled, MeOPN-filtered
31
P HSQC) con-
firmed that this doublet originated from a MeOPN
group (Fig. 2B). The chemical shift of the MeOPN
signal at d
P
14.7 p.p.m., determined using a 2D
31
P-
HSQC experiment (Fig. 2C), is highly unique to a
phosphoramidate bond and is consistent with MeOPN
signals observed in other strains of C. jejuni, which
range from d
P
13.1 p.p.m. to 14.7 p.p.m. [18,23,24]. For
HR-MAS NMR of whole C. jejuni cells, we typically
observe only the
1
H-
31
P correlation between the phos-
phorus and the sharp methyl group resonance of the
MeOPN. The correlation between phosphorus of the
MeOPN and the ring protons of pyranose sugars is not
observed [24], probably because of short T
2
transverse
O
H
O
H
H
H
H
O
OH
O
NH
O
H
NH
H
H
H
H
O
O
CH2OH
O
CH2OHHOH2C
P
O
ONH2
O
CH3
H
O
OH
OH
HOH2C
CH2OH
H
HH
CH3
A
B
C
- -Sor
Lf
- -GlcA NGro
D6
- -GlcNAc
D
D
MeO N
P
1
3
4
6
2
1
2
3
4
5
6
788
1
2
3
4
5
6
Fig. 1. The complete structure of the repeating unit for the CPS
of the HS:19 serostrain of C. jejuni. The repeating unit of the
CPS consists of [-4)-b-D-GlcA6NGro-(1–3)-b-D-GlcNAc-(1-]
n
with an
a-L-sorbofuranose branch located at C2 of GlcA and an MeOPN
group at C4 of GlcNAc. Structural heterogeneity is due to the non-
stoichiometric sorbofuranose branch and the variably methylated
nonstoichiometric MeOPN group. Residue A represents b-D-glucu-
ronic acid-6-N-glycerol, B is 2-acetamido-2-deoxy-b-D-glucose, C is
a-L-sorbofuranose, and D is MeOPN.
A
B
C
Fig. 2. HR-MAS NMR spectroscopy of intact
C. jejuni HS:19 cells. (A)
1
H-HR-MAS NMR
spectrum (256 transients) showing a doublet
originating from a MeOPN group. (B) 1D
31
P-HSQC ‘MeOPN-filtered’ HR-MAS NMR
spectrum (
31
P-decoupled, 256 transients,
1
J
H,P
¼10 Hz) showing a broad signal ori-
ginating from a MeOPN group. (C) 2D
31
P-HSQC HR-MAS NMR spectrum showing
two MeOPNs (256 transients, 128 incre-
ments,
1
J
H,P
¼10 Hz).
D. J. McNally et al.Campylobacter jejuni HS: 19 CPS
FEBS Journal 273 (2006) 3975–3989 ª2006 The Authors Journal compilation ª2006 FEBS 3977
relaxation times and the lower intensity of the broad
GlcNAc H4 resonance due to homonuclear couplings
and structural heterogeneity of the CPS. A second
minor MeOPN signal observed at d
P
14.3 p.p.m. indi-
cated structural heterogeneity for CPS on the cell sur-
face. Spectral lines originating from CPS on the surface
of HS:19 cells were too broad to draw additional con-
clusions regarding its structure, necessitating its isola-
tion for further study.
Isolation of CPS
Previous studies that have examined CPS in the HS:19
serostrain used Westphal’s classic hot water ⁄phenol
extraction method to isolate the polysaccharide
[5,6,12,26]. Our results using this method closely
agreed with these studies in terms of the quantity and
purity of CPS obtained. However, CE-ESI-MS and
NMR analyses of hot water ⁄phenol-purified CPS
revealed that labile groups such as the MeOPN and
sorbose branch were mostly absent (not shown). On
the basis of these observations, we concluded that the
hot water ⁄phenol method was responsible for remov-
ing labile CPS groups, and that removal of these labile
modifications by this method resulted in them being
overlooked by previous studies. These findings support
a recent study that found that the hot water ⁄phenol
method hydrolyzed labile groups that are present on
the CPS of the HS:1 serostrain of C. jejuni [23]. With a
less harsh enzymatic purification method, 5 mg CPS
was obtained from a 6-L culture of HS:19 cells (7 g
cells, wet pellet mass). CPS isolated with this gentler
technique contained a moderately higher concentration
of nucleic acid and protein impurities. However, this
enzymatic extraction method preserved the sought
after labile groups, facilitating the characterization of
the complete CPS structure.
High-resolution NMR spectroscopy
of purified CPS
Examination of purified CPS with NMR at 600 MHz
(
1
H) with a cold probe revealed a [-4)-b-d-GlcA6N-
Gro-(1–3)-b-d-GlcNAc-(1-]
n
backbone and showed
the unknown labile CPS structures to be a MeOPN
group and an a-sorbofuranose branch (Table 1,
Fig. 3). The proton spectrum of purified CPS closely
resembled CPS on the cell surface in that a doublet
originating from a MeOPN group (d
H
3.77 p.p.m.,
3
J
H,P
¼12.0 Hz) was observed to protrude above
broad CPS signals (Fig. 3A). Selective 1D TOCSY and
1D NOESY experiments were used to assign the pro-
ton resonances for GlcA, GlcNAc and sorbose. 1D
TOCSY of the sorbose H3 resonance revealed overlap-
ping H4 and H5 signals (Fig. 3B), and simultaneous
excitation of the H4 and H5 resonances showed signals
corresponding to H6 ⁄H6¢(Fig. 3C). As C2 of sorbose
does not have a proton, a 1D NOESY experiment of
the sorbose H3 resonance was used to assign H1 ⁄H1¢
resonances (Fig. 3D).
A
1
H-
31
P correlation observed between the MeOPN
OCH
3
group and H4 of GlcNAc at d
P
14.7 p.p.m.
indicated the location of the MeOPN at C4 of Glc-
NAc (Fig. 3E). Carbon assignments were determined
from
13
C-
1
H correlations observed using a
13
C-HMQC
experiment (Fig. 3F). With the exception of sorbose
resonances, signals within the
13
C-HMQC spectrum
for purified HS:19 CPS were generally broad and
therefore weak. In particular,
13
C-
1
H correlations for
C3 and C4 of GlcNAc and C1 of GlcA and GlcNAc
were visible only at higher temperature (40 C) and
at 600 MHz (
1
H) with a cryogenically cooled probe.
Proton and carbon resonances determined from
13
C-HMQC and heteronuclear multiple-bond correla-
tion (HMBC) experiments were consistent with those
reported for b-GlcA6NGro and b-GlcNAc [5,6,12], a
MeOPN group [18,23,24] and a-sorbofuranose [28–30].
Structural heterogeneity generated by the sorbose
branch and MeOPN group was indicated by two sets
of
13
C-
1
H correlations for C2 and C3 of GlcA, as well
Table 1. NMR proton and carbon chemical shifts d(p.p.m.) for CPS
purified from HS:19 serostrain of C. jejuni.
Residue Position d
H
d
C
b-D-GlcA6NGro (A) A1 4.65 101.0
A2 3.71 73.8
A3 3.71 73.8
A4 3.90 79.0
A5 3.94 75.1
A6 170.7
A7 4.07 53.9
A8 ⁄A8¢3.65 ⁄3.75 61.5
b-D-GlcNAc (B) B1 4.62 100.7
B2 3.97 56.1
B3 4.24 75.8
B4 4.26 74.2
B5 3.61 75.5
B6 ⁄B6¢3.74 ⁄3.93 61.3
B7 175.0
B8 2.11 23.5
a-L-Sorf(C) C1 ⁄1¢3.64 ⁄3.73 61.5
C2 104.3
C3 4.17 79.2
C4 4.41 76.1
C5 4.39 79.2
C6 ⁄C6¢3.69 ⁄3.79 62.9
MeOPN (D) D1 3.77 54.8
Campylobacter jejuni HS: 19 CPS D. J. McNally et al.
3978 FEBS Journal 273 (2006) 3975–3989 ª2006 The Authors Journal compilation ª2006 FEBS
as for C4 and C5 of GlcNAc (Fig. 3F, indicated with
asterisks). The chemical shifts of these extra resonances
were in excellent agreement with those reported for
nonphosphoramidated, nonsorbosylated CPS [5,6,12].
Compared with nonphosphoramidated CPS, our
results indicated that the MeOPN caused a downfield
shift in the H4 resonance of GlcNAc (0.71 p.p.m.).
This observation is consistent with the effects of phos-
phoramidation reported for the 11168 and HS:1
strains, where the presence of the MeOPN group
caused the signals for neighbouring protons to shift by
0.6–0.8 p.p.m. [18,23,24].
MS analysis of purified CPS
Because of the large molecular mass of the HS:19
CPS, a high orifice voltage (+ 200 V) was used to
promote in-source collision-induced dissociation [27]
to facilitate its analysis with CE-ESI-MS (Fig. 4).
CE-ESI-MS analysis of purified CPS corroborated the
hyaluronic acid-like structure deduced from chemical
analyses and NMR spectroscopy (Fig. 4A, Table 2).
Interestingly, ions observed at m⁄z514.4, 532.4, 694.5
and 984.8 indicated the presence of an O-phosphor-
amidate group OP(O)(NH
2
)(OR) without O-methyla-
tion, suggesting that the nonstoichiometric MeOPN
group is also variably methylated for the HS:19 sero-
strain. CE-ESI-MS ⁄MS analysis of m⁄z708.5, corres-
ponding to one complete repeat of the CPS, revealed
fragment ions at m⁄z297.0 and m⁄z412.3, confirming
the location of the MeOPN group on GlcNAc and
the presence of a single sorbose branch on GlcA
(Fig. 4B).
Determination of absolute configuration
for CPS sugars
By comparing the GC retention times of the R- and
S-butyl glycosides of authentic standards with the
R-butyl glycosides prepared from a purified CPS sam-
ple, b-GlcA and b-GlcNAc were shown to have the
dconfiguration (not shown). As the chiral alcohol
Fig. 3. High-resolution NMR spectroscopy
of CPS purified from the HS:19 serostrain of
C. jejuni. (A)
1
H NMR spectrum (1024 tran-
sients) showing a doublet originating from
the MeOPN group. (B) 1D TOCSY (30 ms)
of Sor H3. (C) 1D TOCSY (30 ms) of Sor H4
and H5. (D) 1D NOESY (400 ms) of Sor H3.
(E)
31
P-HSQC spectrum (64 transients, 32
increments,
1
J
H,P
¼7 Hz). (F)
13
C-HMQC
spectrum (128 transients, 128 increments,
1
J
C,H
¼150 Hz). For selective 1D experi-
ments, excited resonances are underlined.
Residues with asterisks correspond to struc-
tural heterogeneity generated by the non-
stoichiometric MeOPN group and
a-L-sorbofuranose branch. Annotations for
residues are the same as Fig. 1 and sm
represents sorbose monosaccharide.
D. J. McNally et al.Campylobacter jejuni HS: 19 CPS
FEBS Journal 273 (2006) 3975–3989 ª2006 The Authors Journal compilation ª2006 FEBS 3979