Structure of the O-polysaccharide and classification
of
Proteus mirabilis
strain G1 in
Proteus
serogroup O3
Zygmunt Sidorczyk
1
, Krystyna Zych
1
, Filip V. Toukach
2
, Nikolay P. Arbatsky
2
, Agnieszka Zablotni
1
,
Alexander S. Shashkov
2
and Yuriy A. Knirel
2
1
Department of General Microbiology, Institute of Microbiology and Immunology, University of Lodz, Poland;
2
N.D. Zelinsky
Institute of Organic Chemistry, Russian Academy of Sciences, Moscow, Russian Federation
The O -chain polysaccharide of the lipopolysaccharide (LPS)
of a previously nonclassi fied strain o f Proteus m irabilis
termed G1 was studied by sugar analysis and
1
Hand

of P. mirabilis strains from two other serogroups contain-
ing
D
-GalA6(
L
-Lys) in the O-polysacch aride or in the core
region.
Keywords: Proteus mirabilis; O-polysaccharide; lipopoly-
saccharide; N
a
-(
D
-galacturonoyl)-
L
-lysine; serogroup.
Much has been written about the taxonomy o f Proteus since
the original pub lication by Hauser in 1885 who established
the genus [1]. Currently, the genus Proteus consists of
five named species (P. mirabilis, P. penneri, P. vulgaris,
P. myxofaciens and P. hauseri) and three unnamed
genomospecies 4, 5 and 6 [2,3]. Proteus rods are widespread
in the environment a nd make up part of the normal flora of
the human gastrointestinal tract. Proteus ran ks third (after
Escherich ia and Klebsiella) as the cause of uncomplicated
cystitis, pyelonephritis and prostatitis, particularly, in hos-
pital-acquired cases [4]. P. mirabilis accounts for approxi-
mately 3% of nosocomial infections in the United States
where, together w ith P. penneri, it may play a role in s ome
diarrhoeal diseases [5]. Recently, it has been suggested that
P. mirabilis may p lay an e thiopathogenic role in rheumatoid

Perch scheme of Proteus [7]. Biochemical properties of both
strains were checked in API 20E test, which showed 99.9%
identity with the P. mirabilis species. For other s trains used
in this work, P. mirabilis O28 (51/57) was purchased from
the Czech National Collection of Type Cultures ( CNCTC,
Institute of Epidemiology and Microbiology, Prague,
Czech Republic), and P. mirabilis S1959 (O3) and its R14
mutant (T-like form) came from the collection of the
Correspondence to Z. Sidorczyk, D epartment of Ge neral Microbio-
logy, Institute of Microbiology and Immun o logy, University of Lodz,
90–237, Lodz, Poland. Fax a nd Tel.: + 48 42 635 44 67,
E-mail:
Abbreviations: EIA, enzyme i mmunosorbent assay;
D
-GalA(l-Lys),
N
a
-(
D
-galacturonoyl)-
L
-lysine;
D
-GlcA,
D
-glucuronic acid; HMBC,
heteronuclear multiple-bond correlation; LPS, lipopolysaccharide;
ROESY, rotating-frame NOE spectroscopy.
(Received 15 O ctober 2001, revised 2 January 2002, accepted
11 January 2002)

Rabbit polyclonal anti-(O-polysaccharide) sera against
P. mirabilis G1 and P. mirabilis S1959 were obtained by
intravenous imm unization of rabbits every 5 days with 0.25,
0.5 and 1.0 mL bacterial suspension ( 1.5 · 10
10
c.f.u.ÆmL
)1
),
boiled at 100 °C f or 2 h. One week after t he last injection,
rabbits were bled. The obtained antisera were stored at
)20 °C. For passive immunohemolysis, the antisera w ere
inactivated a t 56 °C f or 30 min and absorbed with sheep red
blood cells. O ne hemolytic unit of anti-(O-polysaccharide)
serum was defined as the antibody dilution yielding 50%
lysis of sheep red b lood cells.
Agglutination test
Agglutination i n tubes was performed w ith a suspension of
heat-killed Proteus bacteria incubated (24 h at 50 °C) with
diluted P. mirabilis G1 anti-(O-polysaccharide) serum.
Passive immunohemolysis, inhibition of passive
immunohemolysis and absorption
Sheep red blood cells were sensitized with a growing
concentration of a lkali-treated LPS fo r 30 min at 37 °C, then
washed with NaCl/P
i
pH 7.2 (15 m
M
Na
2
HPO

the plate was incubated (37 °C for 1 h) and the 50%
inhibition value of hemolysis was read.
In the absorption test, 1 mL anti-(O-polysaccharide)
serum d iluted with N aCl/P
i
(1 : 50) was treated with 100 lL
sheep red blood cells (0.2 mL) sensitized w ith the respective
antigen (200 lgLPS)for30mininanicebath.After
centrifugation, the level of antibodies was evaluated using
passive immunohemolysis test.
Enzyme immunosorbent assay (EIA) and inhibition
of the reaction in EIA
Maxi Sorb microtiter plates (U-bottom form, Nunc,
Denmark) were coated with LPS (50 ng per well) diluted
with NaCl/P
i
at 4 °C for 16 h and washed with water. Plates
were blocked with 2.5% casein in NaCl/P
i
(incubation with
NaCl/P
i
/casein for 1 h at 37 °C followed by two washing
cycles with NaCl/P
i
) and anti-(O-polysaccharide) serum
diluted appropriately with NaCl/P
i
/casein was added. After
incubation at 37 °C for 1 h and washing, peroxidase-

with the same buffer to give A
405
of 1.0–1.6 without adding
the inhibitor. After incubation at 37 °C for 15 min, the
mixture was transf erred t o EIA plates coated with LPS, and
further steps were performed a s described above.
SDS/PAGE and Western blot
SDS/PAGE and W estern immunoblots w ere carried out
according to Laemmli [16]. Briefly, LPS in sample buffer
(4 lL per lane) were separated using 3.5% polyacrylamide
stacking gel and 12.5% running gel a nd then transferred to
a nitrocellulose membrane. The membrane was blocked
with 10% s kimmed milk in dot-blot buffer pH 7.4 (50 m
M
Tris/HCl and 200 m
M
NaCl)at20 °C for 1 h and incubated
with anti-(O-polysaccharide) serum diluted 1 : 300 with the
same buffer for 16 h. The reaction was developed with
alkaline phosphatase-conjugated goat anti-(rabbit IgG) Ig
(Dianova, Germany) diluted 1 : 500 with blotting buffer
supplemented with dried skim milk at 20 °Cfor2h.
5-Bromo-4-chloro-3-indoylphosphate p-toluidine and
p-nitroblue tetrazolium chloride (Bio-Rad, Poland) were
used as substrate.
Sugar analysis
The polysaccharide was hydrolysed with 3
M
CF
3

spectrometer equipped with an SGI INDY computer
workstation. 2D NMR experiments were performed using
standard Bruker software, and
XWINNMR
program (Bruker)
was used to acquire and process data. A mixing time of 200
and 300 ms was used in TOCSY and ROESY experiments,
respectiv ely.
RESULTS AND DISCUSSION
Structural studies
The O-polysaccharide was prepared by mild acid degrada-
tion of P. mirabilis G1 LPS followed by gel-permeation
chromatography on Sephadex G-50. Sugar analysis of the
polysaccharide after a cid hydrolysis revealed glucuronic a cid
(GlcA)andgalacturonicacid(GalA)intheratio 1:5.
Analysis on an amino-acid analyser showed the presence o f
2-amino-2-deoxygalactose and lysine. The
D
configuration
of GalA and GalN was determined by GLC of the
acetylated (S)-2-butyl glycosides and the
L
configuration of
lysine by GLC of the acetylated (S)-2-butyl ester. The
D
configuration of GlcA was established b y a nalysis of
13
C
NMR chemical shift data of the polysaccharide (see below).
The

than expec ted relative con tent o f GlcA i n t he polysaccharide
hydrolysate could be accounted for by its retention in
oligosaccharides with GalN (see the polysaccharide struc-
ture below).
The
1
Hand
13
C NMR spec tra o f the polysaccharide w ere
assigned using 2D COSY, TOCSY, ROESY,
1
H,
13
C
HMQC, and HMQC-TOCSY experiments (Table 1 ). The
TOCSY spectrum showed correlations between H1 and
H2–H5 for GlcA and GalA and between H1 and H2–H4
for both GalNAc residues (GalNAc
I
and GalNAc
II
). The
signals for H5 and H6 of GalNAc
I
were assigned by H4/H5
correlation in the R OESY spectrum a nd H5/H6 c orrelation
in the COSY spectrum. The corresponding
13
CNMR
signals were found by

CHMQC
experiment. The signals for the carboxyl groups (C6 of
GlcA and GalA and C1 of Lys) were assigned by H5/C6
and H2/C1 correlations, respectively, observed in the
HMBC spectrum. The spectrum showed also a correlation
between H2 o f Lys and C 6 o f G alA, thus demonstrating th e
presence of N
a
-galacturonoyllysine (GalA6Lys). This con-
clusion was confirmed by t ypical
13
C NMR chemical shifts
for the free carboxyl group of lysine (d 177.9) and the
amidated carboxyl group of GalA (d 172.3) (compare
published data [21,23]).
Relatively large J
1,2
coupling constant values of 7–8 Hz
determined from the
1
H NMR spectrum for the anomeric
protons at d 4.51–4.55 showed that GlcA and both GalNAc
residues are b-linked. The a-linkage was suggested for a
poorly resolved H1 signal of GalA that appeared downfield
at d 5.20, and was confirmed by t he
13
C NMR chemical shift
data (Table 1, compare to published data [23]).
Significant d ownfield displacements of the signals for C3
of GalNAc

II
H1/GlcA C4 and GlcA H1/
GalNAc
I
C3 correlations. In addition, GalA C1/GalNAc
II
H4 and GalNAc
II
C1/GlcA H4 cross-peaks were present at
Table 1. 500-MHz
1
H and 125-MHz
13
C NMR ch emical shifts (d, p.p.m.) of the O-polysaccharide o f P. mirabilis G1. Additional chemical sh ifts for
NAc are d
H
2.02 and 2.05; d
C
23.6 (2 Me), 175.9 a nd 176.2 (both CO).
Sugar or amino-acid residue H1 H2 H3a, 3b H4 H5 H6a, 6b C1 C2 C3 C4 C5 C6
fi 3)-b-
D
-GalpNAc
I
-(1 fi 4.55 4.02 3.85 4.12 3.70 3.78, 3.78 102.4 52.3 81.3 69.0 76.1 62.4
fi 6)-b-
D
-GalpNAc
II
-(1 fi 4.51 3.89 3.80 4.03 3.85 3.89, 4.16 102.8 53.2 71.1 75.9 73.4 66.7

configuration.
On the basis of the data obtained, it was concluded that
the O-polysaccharide P. mirabilis G1 has the structure
shown in Fig. 3. This structure is similar to that of the
O-polysaccharide of P. mirabilis S1959 and OXK from
serogroup O3 [22,23], the r epeating unit of P. mirabilis G1
differing on ly in the absence of the lateral a-
D
-Glcp residue
(Fig. 3).
Serological studies
Rabbit polyclonal anti-(O-polysaccharide) serum against
P. mirabilis G1 was tested in immunohemolysis with LPS
from the complete s et of Proteus stra ins, including 37 strains
of P. mirabilis and 28 strains of P. vulgaris belonging to 49
Proteus O-serogroups as well as 133 strains of P. penneri.
From 188 tested LPS, anti-(O-polysaccharide) serum
against P. mirabilis G1 reacted only with the homologous
LPS and LPS of P. mirabilis S1959, O28, and a mutant of
S1959 (R14, T-like form).
In enzyme immunosorbent assay (EIA), P. mirabilis G1
and P. mirabilis S1959 anti-(O-polysaccharide) sera showed
the strongest reaction w ith LPS of both P. mirabilis G1 and
S1959, whereas LPS of P. mirabilis O28 and R14 reacted
markedly weaker (Fig. 4). The specificity of the cross-
reactions was confirmed b y i nhibition of the reaction in EIA
Fig. 3. Structures of the O-polysaccharides of the cross-reactive LPS of
P. mirabilis G1, S1959, and O28.
Fig. 4. Reactivity of anti-(O-polysaccharide) sera against P. mirabilis
G1 (A) and S1959 (B) in EIA. j,LPSofP. mirabilis G1; d,LPSof

when antisera were absorbed with the homologous LPS
(Table 2). Absorption of P. mirabilis G1 anti-(O-polysac-
charide) serum with P. mirabilis S1959 LPS significantly
decreased the serum titre in the homologous system and
completely removed all cross-reactive antibodies against
LPS of P. mirabilis S1959, R14 and O28. Absorption with
LPS from each of two last strains decreased the reactivity
level with L PS of P. mirabilis G1 a nd S 1959 and c ompletely
abolished the reactivity with LPS of P. mirabilis R14 and
O28. Similar results were obtained with P. mirabilis S1959
anti-(O-polysaccharide) serum absorbed with P. mirabilis
G1 LPS (Table 2).
These results suggested the presence in P. mirabilis G1
and S1959 anti-(O-polysaccharide) sera of cross-reactive
antibodies of at least two types. Antibodies of the first type
bound to an epitope on LPS of P. mirabilis G1 and S1959.
Antibodies of the other type bound to another epitope
shared by the homologous LPS and LPS of all cross-
reactive strains.
In Western b lot, P. mirabilis G1 anti-(O-polysaccharide)
serum recognized slow migrating bands of three LPS
(without R14) and fast migrating bands of P. mirabilis
G1, O 28, and R14 LPS (Fig. 5A). These bands correspond
to high- and low-molecular-mass LPS species consisting of
the core-lipid A moiety with or without an O-chain
polysaccharide attached, respectively. The lack of reactivity
of fast migrating b ands of P. mirabilis S1959 LPS indicated
a difference between the core structures in P. mirabilis
S1959 and G1. Anti-(O-polysaccharide) serum against
P. mirabilis S1959 recognized fast migrating bands of all

including LPS of P. mirabilis S1959 [28]. A partial epitope
O3b is evidently linked t o a lateral a-
D
-Glcp residue which is
present in P. mirabilis S1959 but absent from P. mirabilis
G1, and a p artial epitope O3c in P. mirabilis G1 may b e an
extended epitope that is masked by the a-
D
-Glcp residue in
P. mirabilis S1959.
Comparison of the structures of t he O-chain polysaccha-
rides a nd core oligosaccharides [21–23,27,29,30] enabled
suggestion that
D
-GalpA6(
L
-Lys) is responsible for the
cross-reactivity of not only P. mirabilis G1 and S1959 but
also P. mirabilis O28 and R14. Indeed, LPS of P. mirabilis
O28 is characterized by the presence of a-
D
-GalpA6(
L
-Lys)
in the O-chain polysaccharide [21] (Fig. 3) and
b-
D
-GalpA6(
L
-Lys) in the core oligosaccharide [29]. No

&unnamedProteus genomospecies 4, 5 and 6. Int. J. Syst. Evol.
Microbiol. 50, 1869–1875.
3. O’Hara, C.M., Brenner, F.W. & Miller, J.M. (2000) Classification,
identification, and clinical significance of Pr oteus, Providencia,and
Morganella. Clin. Microbiol. Rev. 13, 534–546.
4. Stamm, W.E. (1999) Urinary tract infection. In Clinical Infection
Diseases: a P ractical Approach (Root, R.K, ed.), pp. 549–656.
Oxford University Press, Oxford, UK.
5. Mu
¨
ller, H.E. (1986) Occurrence and pathogenic role of Morga-
nella-Proteus-Providencia group bacteria in human feces. J. Clin.
Microbiol. 23, 404–405.
6. Wilson, C., Thakore, A., Isenberg, D. & Ebringer, A. (1997)
Correlation b etween anti-Proteus antibo dies and isolation rates o f
P. mirabilis in rheumatoid arthritis. Rheumatol. Intern. 16,
187–189.
7. Larsson, P. (1984) Serology of Proteus mirabilis and Proteus
vulgaris. Methods Microbiol. 14, 187–214.
8. Penner, J.L. & Hennessy, J.N. (1980) Separate O-grouping
schemes for serotyping clinical isolates of Proteus vulgaris and
Proteus mirabilis. J. Clin. Microbiol. 12, 304–309.
9. Zych, K., Kocharova, N.A., Kowalczyk, M., Toukach, F.V.,
Kaminska, D., Shashkov, A.S., Knirel, Y.A. & Sidorczyk, Z.
(2000) Structure of the O-specific polysaccharide of Proteus pen-
neri 71 and classification of cross-reactive P. penneri str ai ns t o a
new proposed serogroup O64. Eur. J. Biochem. 267, 808–814.
10. Shashkov, A.S. , Kondakova, A.N., Senchenkova, S.N., Zych, K .,
Toukach, F.V. , Knirel, Y.A. & Sidorczyk, Z. (2000) Structure of a
2-aminoethyl phosphate-containing O-specific polysaccharide of

and
L
configuration of neutral mono -
saccharides by high-resolution capilary G.L.C. Carbohydr. Res.
62, 349–357.
19. Shashkov, A.S., Senchenkova, S.N., Nazarenko, E.L., Zubkov,
V.A., Gorshkova, N.M., Knirel, Y.A. & Gorshkova, R.P. (1997)
Structure of phosphorylated polysaccharide from Shewanella
putrefaciens strain S29. Carbohydr. Res. 303, 333–338.
20. Patt, S.L. & Shoolery, J.N. (1982) Attached proton test for
carbon-13 NMR. J. Magn. Reson. 46, 535–539.
21. Radziejewska-Lebrecht, J., Shashkov, A.S., Vinogradov, E.V.,
Grosskurth, H., Bartodziejska, B., Rozalski, A., Kaca, W.,
Kononov, L.O., Chernyak, A.Y., Mayer, H., Knirel, Y.A. &
Kochetkov, N .K. (1995) Structure and e pitope characterisation of
the O-specific polysaccharide of Proteus mirabilis O28 containing
amides of
D
-galacturonic a cid with
L
-serine and
L
-lysine. Eur J.
Biochem. 230, 705–712.
22. Ziolkowski, A ., Shashkov, A.S., Swierzko, A.S., Sen chenkova,
S.N.,Toukach,F.V.,Cedzynski,M.,Amano,K.I.,Kaca,W.&
Knirel, Y.A. ( 1997) S tructures o f the O -antigens o f Pro teus bacilli
belonging to OX group (serogroups O1–O3) used in Weil-Felix
test. FEBS Lett. 411, 221–224.
23. Kaca, W., Knirel, Y.A., Vinogradov, E.V. & Kotelko, K. (1987)

of the carbohydrate backbone of the core –lipid A region of the
lipopolysaccharide from Proteus mirabilis serotype O28. Carbo-
hydr. Res. 329, 351–357.
30. Vinogradov, E.V., Radziejewska-Lebrecht, J. & Kaca, W. (2000)
The structure of t he carbohydrate backbone of core–lipid A region
of the lipopolysaccharides from Proteus mirabilis wild-type strain
S1959 (serotype O3) and its Ra mutant R110/1959. Eur. J. Bio-
chem. 267, 262–269.
1412 Z. Sidorczyk et al. (Eur. J. Biochem. 269) Ó FEBS 2002