A unique variant of streptococcal group O-antigen (C-polysaccharide)
that lacks phosphocholine
Niklas Bergstro¨m
1
, Per-Erik Jansson
1
, Mogens Kilian
2
and Uffe B. Skov Sørensen
2
1
Clinical Research Centre, Analytical Unit, Karolinska Institute, Huddinge Hospital, Sweden;
2
Department of Medical Microbiology
and Immunology, University of Aarhus, Denmark
Streptococcus mitis strain SK598, which represents a sub-
group of biovar 1, possesses a unique variant of the
C-polysaccharide found in the cell wall of all strains of
Streptococcus pneumoniae and in some strains of S. mitis.
This new variant lacks the choline methyl groups in contrast
to the previously characterized forms of C-polysaccharide,
which all contain one or two choline residues per repeat. The
following structure of the repeating unit of the SK598
polysaccharide was established:
where AAT is 2-acetamido-4-amino-2,4,6-trideoxy-
D
-
galactose.
This structure is identical to the double choline-substi-
tuted form of C-polysaccharide, except that it is substituted
with ethanolamine instead of choline. This extends the

forms to four.
Materials and methods
Bacterial strain
The S. mitis biovar 1 strain SK598 used for preparation of
polysaccharide was from our own strain collection. This
strain was selected as it was negative for the presence of
phosphocholine, although it seemed to possess a C-poly-
saccharide like molecule when examined by ELISA and by
immunoelectrophoresis [1]. Strain SK598 was characterized
and identified as previously described [1,2]. It belongs to
Lancefield serogroup O as an extract from SK598 reacts
with streptococcal group O-antiserum purchased from
Statens Serum Institut, Copenhagen, Denmark.
Preparation of polysaccharide
The S. mitis biovar 1 strain SK598 was cultured overnight
at 37 °C in 5 L laboratory flasks each containing 2.5 L
Todd-Hewitt broth (CM189, Oxoid, Basingstoke, UK).
The bacterial cells were harvested by centrifugation
Correspondence to P E. Jansson, Karolinska Institute,
Clinical Research Centre, Novum, Huddinge University Hospital,
S-141 86 Huddinge, Sweden.
Fax: + 46 8585 83820, Tel.: + 46 8585 83821,
E-mail:
(Received 5 September 2002, revised 6 March 2003,
accepted 13 March 2003)
Eur. J. Biochem. 270, 2157–2162 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03569.x
(10 000 g, 30 min) and pooled from a total of 30 L broth
culture. The cells were washed twice in saline and suspended
in 50 mL of lysis buffer [0.1
M

13
C NMR spectra were recorded with a JEOL JNM
ECP500 spectrometer, using standard pulse sequences.
Spectra of samples in 20 m
M
phosphate buffers of pD 7.4
were recorded at 35 °C. Chemical shifts are reported
in p.p.m., using sodium 3-trimethylsilylpropanoate-d
4
(d
H
0.00) or acetone (d
C
31.00) or aqueous 2% phosphoric acid
(d
P
0.00) as internal references. For
13
Cand
31
P the reference
measurement was made with a separate tube before the
actual measurement. Chemical shifts were taken from 1D
spectra when possible, or else from
1
H,
1
H-correlated 2D
NMR spectra, i.e.
1

or NaBD
4
,and
acetylation. For methylation analysis, methylation was
performed with methyl iodide in the presence of sodium
methyl sulfinyl methanide, and the methylated products
were purified using Sep-Pak C
18
-cartridges. For GLC, a
Hewlett-Packard 5890 instrument fitted with a flame-
ionization detector was used. Separation of alditol acetates
was performed on a DB-5 capillary column (30 m ·
0.25 mm) using a temperature program 160 °C(1min)fi
250 °Cat3°CÆmin
)1
. GLC-MS (EI) was performed on
a Hewlett-Packard 5890/Nermag R10–10H quadrupole
instrument. Partially methylated alditol acetates were
separated on a DB-5 capillary column (25 m · 0.20 mm),
using the same temperature program as described for alditol
acetates. The absolute configurations of the sugar residues
were determined by GLC-MS of the trimethylsilylated
(+)-2-butyl glycosides [5], using the same temperature
program as described for alditol acetates.
HF degradation
A solution of the crude cell wall polysaccharide (69 mg) in
aqueous 48% HF (1 mL) was kept for 48 h at 18 °C, blown
to dryness with dry air and residual traces of acid were
neutralized with 1
M

minor amounts of other monosaccharides.
The latter major fraction, PSII was hydrolyzed with 4
M
hydrochloric acid and showed glucose and galactosamine
in the proportions 1 : 4.5. This hydrolysis enhances amino
sugars but ribitol is not detected. The absolute configuration
of the sugars was
D
, as demonstrated by GLC of the
trimethylsilylated (+)-2-butyl glycosides. In order to main-
tain a constant pD to get reproducible spectra in the NMR
studies, the solution of PSII was buffered at pD 7.4
(pH 7.0). The
1
H-NMR spectrum of PSII showed five
major peaks in the anomeric region corresponding to
approximately one proton each, and some smaller signals
(Fig. 1). The five large signals in the anomeric region
appeared at d 5.17, 4.94, 4.77, 4.64 and 4.62 (Table 1). This
could be recognized as closely similar but not identical to
signals in the anomeric region from the C-polysaccharide
purified from S. pneumoniae [1,3,6–8]. Four of the signals
could be shown to be anomeric and appeared at d 5.17 (J
1,2
3.5 Hz, 1H), 4.94 (J
1,2
3.5 Hz, 1H), 4.64 (J
1,2
7.3 Hz, 1H),
and 4.62 (J

was indicated by a correlation in the HMQC spectrum to a
signal at d 50.1. The C-5 signal was identified from a
correlation from H-1 in the HMBC spectrum. H-5 and H-6
were both identified by a correlation to C-4 in the HMBC
spectrum; correlations between H-5/C-6 and H-6/C-5
verified the assignments. Substituted positions in the residue
were indicated from the large glycosylation shifts, 7.8 and
1.9 p.p.m., for the C-4 and C-6 signals, respectively, when
compared to unsubstituted a-
D
-GalNAc. Residue B (H-1
Fig. 1.
1
H NMR spectrum (35 °C, 500 MHz) of the cell wall polysaccharide from S. mitis SK598. A–D refer to anomeric protons as described in the
text.
Table 1.
1
H- and
13
C-NMR data for the C-polysaccharide (PSII) of S. mitis. SK598 obtained at pD 7.4.
Sugar residue
Chemical shifts (p.p.m.)
1 2345 6a6b
fi6)-a-GalpNAc(1fi A 5.17 [3,5]
a
4.32 3.93 4.11 4.01 4.02 4.02
4 94.2 50.1 67.5 77.4 71.3 64.0
›
fi3)-a-AATp(1fi B 4.98 [3,5] 4.23 4.39 3.94 4.77 1.24
98.9 49.0 75.6 55.3 63.7 16.0

b-Glc residue as all ring proton couplings in the ring system
were large, thereby demonstrating an all-axial proton
relation and the anomeric configuration was b,astheJ
1,2
-
value was 7.3 Hz. In the NOESY spectrum H-3 and H-5
signals could be assigned from correlations to H-1. Further
assignments were obtained from the HMQC-TOCSY spec-
trum, where correlations H-2/C-3 and C-4/H-5 were evident.
The residue was determined to be 6-substituted because of a
large glycosylation shift, 3.2 p.p.m., for the C-6 signal.
Residue D (H-1 d 4.62) was assigned to a 3,6-disubsti-
tuted GalNAc residue with the b configuration (J
1,2
-value of
7.3 Hz) and the galacto-configuration being evident with a
small coupling between H-3 and H-4. The C-2 was linked to
nitrogen indicated by a correlation in the HMQC spectrum
to signal at d 51.1. In the NOESY spectrum H-3 and H-5
signals were assigned from correlations to H-1. C-6 was
determined from a correlation to H-5 and H-6 was
confirmed by a correlation to C-5, both in the HMQC-
TOCSY spectrum. The 3,6-disubstitution was indicated by
the chemical shifts of the C-3 and C-6 which were shifted 3.0
and 3.1 p.p.m., respectively.
Residue E was determined to be a 1,5-disubstituted ribitol
residue as all proton and carbon signals could be assigned
with the aid of the COSY, NOESY and HMQC spectra by
which a pentitol residue was evident. A good correspon-
dence with previous data from C-polysaccharide was also

structural element CBADE.
The
31
P-NMR spectrum showed three signals of equal
intensity at d 1.33, 0.33, and )0.04 (Fig. 2). All three signals
could be assigned to a polysaccharide similar to
C-polysaccharide. The signal at d 1.33 was assigned to a
phosphate group bridging the ribitol and Glc residues. The
value is close to that observed for C-polysaccharide. Thus,
correlations in the H,P-HMQC spectrum from phosphorus
to protons with signals at d 4.15, 4.09 (H-6a and H-6b of
residue C), 4.06, and 3.98 (H-5a and H-5b in the ribitol,
Fig. 2.
31
PNMRspectrum(35°C, 200 MHz) of the cell wall poly-
saccharide from S. mitis SK598.
Table 2. Inter-residue connectivities observed in HMBC and NOESY
spectra for C-polysaccharide of S. mitis SK598.
Residue
Chemical shifts (H/C)
Anomeric
nucleus Inter-residue correlations
d (
1
H) d (
13
C) d (
1
H) d (
13

H
4.02 (H-6a and
H-6b, of A)andasthesignalatd
P
– 0.04 correlated to
d
H
4.13 (H-1a and H-1b, of G)andtod
H
4.07 (H-6a and
H-6b, of D). The two phosphoethanolamine groups are
therefore linked to the 6-positions of residues A and D.
Two peaks were obtained in the chromatogram when the
crude material was treated with aqueous 48% HF for 48 h
at )18 °C and fractionated on a column of Bio-Gel P-4. The
first peak contained PSI and was a polymeric fraction eluted
at 1.2 void volumes. The second peak was an oligosaccha-
ride fraction eluted at 1.4 void volumes (PSII-OLS). From
the
1
H-NMR spectrum it was clear that the oligosaccharide
fraction was a mixture and that it was the same mixture as
that obtained from pneumococcal C-polysaccharide when
treated with aqueous 48% HF under same conditions. The
phosphoethanolamine and phosphate groups were absent
as a result of that all phosphate ester linkages were broken.
From the
1
H-NMR spectrum it was clear that the fraction
contained a major and a minor compound, where the major

We previously interpreted reactivity of a streptococcal
cell wall polysaccharide preparation with both of two
monoclonal antibodies that detect phosphocholine and
the backbone of pneumococcal C-polysaccharide, respect-
ively, as an indication of the presence of an antigen
identical or closely similar to C-polysaccharide. This
interpretation was validated for S. mitis strain SK137 [1].
Structural analysis demonstrated that this S. mitis biovar 1
strain possesses a true C-polysaccharide in addition to
a unique glycan. The C-polysaccharide found in all
S. pneumoniae strains and in most S. mitis biovar 1
strains was shown to represent the streptococcal sero-
group O antigen [1].
We have now investigated the structure of a polysaccha-
ride prepared from another S. mitis biovar 1 strain that
differs from the previously examined strain by failing to
react with the monoclonal antibody against phospho-
choline. As expected, the predominant polysaccharide
demonstrated in strain SK598 was found to be a cell wall
polysaccharide similar but not identical to pneumococcal
C-polysaccharide. The structures are identical except
that the characteristic phosphocholine residues of pneumo-
coccal C-polysaccharide are absent from the new S. mitis
C-polysaccharide, which is substituted with ethanolamine
(structure 1).
Choline is a strict nutritional requirement for pneumo-
cocci although mutant strains that have acquired the ability
to grow in the absence of choline have been described [9,10].
When grown in a chemically defined medium containing
ethanolamine but no choline, such strains generate phos-

the following main structure or ÔbackboneÕ:
6Þ-b-d-Glcp-ð1!3Þ-a-AATp-ð1!4Þ-a-d-GalpNAc-
ð1!3Þ-b-d-GalpNAc-ð1!1Þ-ribitol-5-P-ðOÞ
in which one or both Gal are amino sugars that may or
may not be N-acetylated.
The finding of a new C-polysaccharide structure extends
the number recognized of C-polysaccharide variants. The
first found contained only one phosphocholine group and
one GalNH
2
residue, which is normally N-acetylated [6].
Subsequently, a polysaccharide with only N-acetylated
GalN residues and with two phosphocholine residues was
reported [7]. More recently a polysaccharide with the same
backbone but with one phosphocholine group was
identified [3]. The polysaccharide with two phosphoetha-
nolamine groups described in this communication extends
the list to four. We suggest that streptococcal strains,
including pneumococci, which possess one of these
C-polysaccharide variants are referred to as Lancefield
serogroup O [1].
Acknowledgements
This work was supported by grants from the Karolinska institutets
fonder (to P.E.J.) and by the Danish Medical Research Council grant
# FOR 9702265 (to M.K.).
References
1. Bergstro
¨
m, N., Jansson, P.E., Kilian, M. & Skov Sørensen, U.B.
(2000) Structures of two cell wall-associated polysaccharides of a

(1993) Teichoic acid and lipoteichoic acid of Streptococcus pneu-
moniae possess identical chain structures. A reinvestigation of
teichoic acid (C-polysaccharide). Eur. J. Biochem. 215, 851–857.
9. Yother,J.,Leopold,K.,White,J.&Fisher,W.(1998)Generation
and properties of a Streptococcus pneumoniae mutant which does
not require choline or analogs for growth. J. Bacteriol. 180,
2093–2101.
10. Fisher, W. (2000) Phosphocholine of pneumococcal teichoic acids:
role in bacterial physiology and pneumococcal infection. Res.
Microbiol. 151, 421–427.
2162 N. Bergstro
¨
m et al. (Eur. J. Biochem. 270) Ó FEBS 2003