Abstract

The lipopolysaccharide (LPS) from eight strains of Yersinia pestis which had been cultured at 28°C appeared to be devoid of an O-antigen when analysed by sodium dodecyl sulfate–polyacrylamide gel electrophoresis. LPS isolated from three of these strains which had been cultured at 37°C also appeared to be devoid of an O-antigen. When the LPS from Y. pestis strain CO92 was purified and analysed by matrix-assisted laser desorption-ionisation time-of-flight mass spectrometry, the observed signals were in the mass range predicted for molecules containing lipid A plus the core oligosaccharide but lacking an O-antigen. The nucleotide sequence of Y. pestis strain CO92 revealed the presence of a putative O-antigen gene cluster. However, frame-shift mutations in the ddhB, gmd, fcl and ushA genes are likely to prevent expression of the O-antigen thus explaining the loss of phenotype.

Introduction

Yersinia pestis is one of 11 species of bacteria within the genus Yersinia. Three members of the genus are pathogenic to humans; Y. pestis, Yersinia pseudotuberculosis and Yersinia enterocolitica. Y. pestis is the causative agent of plague, a disease which is usually contracted by humans as a result of the bite of an infected flea [1]. Two to three days after infection the individual develops a range of symptoms including fever, chills, weakness and headache [2], and there is often gastrointestinal involvement with nausea, vomiting and diarrhoea [3]. The lipopolysaccharide (LPS) of Y. pestis is believed to play a key role in the late stages of infection [2,4], when bacterial levels in the blood can reach 107 cfu ml−1[5].

LPS is a major component of the outer membrane of Gram-negative bacteria and invariantly contains lipid A, which is embedded in the membrane, and a polysaccharide core. Frequently, the LPS from Gram-negative bacteria also contains an O-antigen polysaccharide which extends beyond the cell surface. In many bacterial species the O-antigen has been shown to be an essential virulence factor [6] and is thought to be involved in protecting the bacteria against the activity of complement [7], by causing the membrane-attack complex to form at a distance from the bacterial surface. The LPS of Y. pseudotuberculosis and Y. enterocolitica have been shown to possess an O-antigen which is an essential virulence determinant [8]. However, the composition of the O-antigen can vary between strains; in Y. pseudotuberculosis seven different O-antigen polysaccharides have been identified and these differences can be exploited to serotype different strains of the bacteria [9]. The structure of the O-antigen of Y. enterocolitica is thermo-regulated, with longer chain lengths predominating when the bacteria were cultured at temperatures below 37°C [10]. Thermo-regulation of the Y. pseudotuberculosis O-antigen structure has not been shown.

The LPS of Y. pestis has been less well studied. Some workers have reported that it lacked an O-antigen [11,12] and consisted of lipid A bound to the core oligosaccharide by 3-deoxy-d-manno-octulosonic acid (KDO) [13]. Others have reported that both rough and smooth forms of LPS were produced [14] or that the synthesised O-antigen was shed into the growth media [15]. There have also been suggestions that the fine structure of Y. pestis LPS is thermo-regulated. In comparison with the LPS from bacteria which had been cultured at 28°C, LPS isolated from bacteria which had been cultured at 37°C had a lower proportion of KDO and phosphate, as well as increased mobility in sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) gels [16].

We have set out to investigate whether different strains of Y. pestis produce LPS containing an O-antigen and whether expression of the O-antigen is temperature-regulated. We have related this to the genetic make-up of the O-antigen coding region of Y. pestis strain CO92 using the available genome sequence.

Materials and methods

Bacterial strains and growth conditions

Bacterial strains used in this study are shown in Table 1 and were grown at 28 or 37°C. Y. pestis was cultured on blood base II agar (BABII, Oxoid, 40 g l−1) containing 0.25% (w/v) hemin. All other Yersinia strains were grown on nutrient agar (Oxoid).

1

Bacterial species and strains used in this study

Species and strain Characteristics 
Y. pestis 195/P virulent 
Y. pestis 1255 virulent 
Y. pestis CO92 virulent 
Y. pestis EV76 attenuated 
Y. pestis EVAA attenuated 
Y. pestis GB virulent 
Y. pestis Java9 virulent, F1-antigen negative 
Y. pestis TW virulent 
Y. pseudotuberculosis NCTC 8580 serogroup 0:4 
Y. pseudotuberculosis NCTC 1779 serogroup 0:2 
Y. enterocolitica NCTC 10461 serogroup 0:3 
Y. enterocolitica NCTC 10598 serogroup 0:1 
E. coli O111 serotype O111 
Species and strain Characteristics 
Y. pestis 195/P virulent 
Y. pestis 1255 virulent 
Y. pestis CO92 virulent 
Y. pestis EV76 attenuated 
Y. pestis EVAA attenuated 
Y. pestis GB virulent 
Y. pestis Java9 virulent, F1-antigen negative 
Y. pestis TW virulent 
Y. pseudotuberculosis NCTC 8580 serogroup 0:4 
Y. pseudotuberculosis NCTC 1779 serogroup 0:2 
Y. enterocolitica NCTC 10461 serogroup 0:3 
Y. enterocolitica NCTC 10598 serogroup 0:1 
E. coli O111 serotype O111 

Purification of LPS

For the extraction of LPS, a modification of the phenol–chloroform–petroleum ether (PCPE) method described by Galanos [17] was used. Instead of homogenising the bacteria, they were mixed with the extraction mixture using a pestle and mortar. After washing with ether the extract was allowed to dry in a fume hood rather than under vacuum. Proteinase K minipreparations of LPS were produced according to the method of Chart [18].

Gel electrophoresis and staining

Glycine gel electrophoresis was performed according to the method of Laemmli [19] using a 12.5% separating gel with a 4.5% stacking gel. Gels were silver-stained according to the method of Chart [18].

Protein determination

The protein content of the isolated LPS was determined by using a BCA protein assay supplied by Pierce UK. The detection limit for protein in this kit was 10 μg ml−1.

Mass spectrometry

Matrix-assisted laser desorption-ionisation time-of-flight mass spectrometry (MALDI-TOF MS) was performed using a Perseptive Biosystems Voyager Elite mass spectrometer with Delayed Extraction. Native LPS was dissolved in 90% dimethylformamide in water and aliquots (0.5 μl) of the resulting solution were analysed using a matrix of 2,5-dihydrobenzoic acid. Angiotensin I was employed as an external calibrant.

Nucleotide sequencing and analysis

To determine the nucleotide sequence of the Y. pestis strain CO92 genome, a random library of 1.5–2.5-kb DNA fragments were first cloned into pUC18 [20]. This library was used for sequencing with 10-fold coverage of the genome. A preliminary sequence assembly from 86 345 sequencing reads was completed. The sequence of the Y. pestis O-antigen cluster was extracted from this publically available database which is also available at the Sanger Centre www site (http://www.sanger.ac.uk/Projects/Y_pestis). This section of sequence was subsequently checked and finished to standard criteria (each base sequenced on both strands, or with different sequencing chemistries). Analysis was performed with the program Artemis (http://www.sanger.ac.uk/Software/Artemis) which allows the visualisation of BLASTN, BLASTX, FASTA, BLASTP, Pfam and Prosite searches, along with hydrophobicity and other analyses, in the context of the sequence and of the predicted proteins.

Results and discussion

Comparison of Y. pestis, Y. pseudotuberculosis and Y. enterocolitica LPS

Y. pseudotuberculosis (serogroups 0:2 and 0:4), Y. enterocolitica (serotypes 0:1 and 0:3) and Escherichia coli strain 0111 were cultured at 37°C and the LPS was isolated. When analysed by SDS–PAGE with silver staining, a characteristic ladder pattern, indicative of an O-antigen side chain was seen (Fig. 1). This ladder was not visible when LPS, isolated from Y. pestis strain CO92 which had been cultured at 28°C, was analysed (Fig. 1, track 1).

1

SDS–PAGE analysis of LPS isolated from Y. pestis strain CO92 (track 1), Y. pseudotuberculosis IV (track 2), Y. pseudotuberculosis II (track 3), Y. enterocolitica 0:1 (track 4), Y. enterocolitica 0:3 (track 5) and E. coli 0111 (track 6). The dye front is indicated by the arrow.

1

SDS–PAGE analysis of LPS isolated from Y. pestis strain CO92 (track 1), Y. pseudotuberculosis IV (track 2), Y. pseudotuberculosis II (track 3), Y. enterocolitica 0:1 (track 4), Y. enterocolitica 0:3 (track 5) and E. coli 0111 (track 6). The dye front is indicated by the arrow.

Comparison of the LPS from different strains of Y. pestis cultured at 28 or 37°C

Eight strains of Y. pestis (Table 1) were cultured at 28°C and LPS isolated using the proteinase K method. These strains included virulent and attenuated isolates, and a strain of Y. pestis (Java9) which has been reported to lack the cell-surface capsular F1-antigen. When the preparations were analysed by SDS–PAGE and silver-stained they each revealed one strongly stained band with similar mobilities (data not shown). There was no evidence of an O-antigen side chain.

To investigate whether growth temperature affected O-antigen presence, three strains of Y. pestis (GB, EV76 and Java9) were cultured at 28 or 37°C, and LPS was isolated. All of the preparations revealed a single major band and lacked the banding patterns typical of an O side chain when analysed by SDS–PAGE and silver-stained (data not shown). These findings suggest that the failure to produce LPS O-antigen is a common feature of Y. pestis strains and that growth temperature does not influence this phenotype.

LPS purification and mass spectrometry

The PCPE extraction method was used to purify LPS from Y. pestis strain CO92 cultured at 28°C producing a yield of 0.5% w/w (10.3 mg from 2 g dry weight of cells). One mg of this preparation contained less than 10 μg of protein (the limit of detection of the assay). When analysed by SDS–PAGE and silver-staining, one intensely stained band and one lower molecular mass weakly stained band were visible (data not shown). No bands were observed after staining similar SDS–PAGE gels with Coomassie brilliant blue. The LPS was analysed by MALDI-TOF MS in the mass range m/z 9000 to m/z 2000. A major cluster of signals was present near m/z 3000 (Fig. 2). No significant signals were observed above m/z 3400. The observed signals are in the mass range predicted for molecules containing lipid A plus the core oligosaccharide but lacking an O-antigen. Therefore, the mass spectrometry data presented here confirms that LPS isolated from Y. pestis strain CO92 lacks an O-antigen.

2

MALDI-TOF trace of LPS isolated from Y. pestis strain CO92 grown at 28°C.

2

MALDI-TOF trace of LPS isolated from Y. pestis strain CO92 grown at 28°C.

The O-antigen of LPS is thought to play several roles in infection. It is believed to hinder the access of complement to the bacterial surface and thereby hamper the formation of a membrane-attack complex close to the cell surface [21]. Bacteria lacking an O-antigen are often killed by serum complement [22]. Y. pestis, but not Y. pseudotuberculosis or Y. enterocolitica, produces a capsular antigen (F1-antigen), which has been linked with protection against complement [23] and it is possible that the F1-antigen is a functional replacement for the O-antigen. However, when we examined the LPS from a naturally occurring F1-antigen negative strain of Y. pestis (Java9) this was also devoid of an O-antigen. This suggests, that either the F1-antigen is not a functional replacement for the absent O-antigen or that in strain Java9 an alternative surface antigen performs this function.

In other bacterial species the absence of an O-antigen reduces virulence. A spontaneous Y. enterocolitica mutant lacking an O-antigen was 50-fold less virulent than the wild-type strain expressing an O-antigen when given orally [24]. However, it is possible that the requirement for an O-antigen is linked to the route of infection. For example, a Salmonella enterica O-antigen mutant was virulent when inoculated intravenously or intraperitoneally but did not cause disease when given orally [7]. Y. pestis is thought to be descended from Y. pseudotuberculosis [25] but, unlike Y. pseudotuberculosis, Y. pestis is not an enteric pathogen. Therefore, it is possible that the O-antigen is not a virulence determinant in Y. pestis and has become redundant over a period of time.

Sequence analysis of the O-antigen pseudocluster

The putative O-antigen gene cluster was identified in the Y. pestis strain CO92 genome sequence using genes from related Yersiniae as DNA probes. The Y. pestis strain CO92 gene cluster was flanked by the adk and ushA genes (Fig. 3). However, scrutiny of this gene cluster identified several mutations providing a probable explanation for the failure of Y. pestis strain CO92 to produce O-antigen. Frame-shift mutations were identified within a C(11) tract in the ddhB gene, in a non-repetitive region in the gmd gene and in a G(6) tract in the fcl gene. The ushA gene contained two frame-shift mutations, the first of which was in an A(6) tract and the second in a non-repetitive sequence. An additional open reading frame was found to be present between the manB and manC genes when this region was compared with similar O-antigen clusters in other bacteria. Three of the frame-shift mutations found in the Y. pestis O-antigen cluster were the result of repeated A, C or G bases. In Haemophilus influenzae, O-antigen expression is regulated by repeat units of the tetranucleotide CAAT in the sequence of the O-antigen operon [26]. The O-antigen is switched on or off depending on the number of repeats. This process of slipped-strand mispairing results in the gene being transcribed in and out of frame. It is possible that these homopolymeric tracts are variable, and that the culture conditions used in this study failed to switch on the necessary genes for O-antigen production. However, the frequency and nature of the mutations in the Y. pestis gene cluster suggests that the simultaneous resolution of all five mutations is unlikely.

3

The genetic organisation of the O-antigen pseudocluster in Y. pestis strain CO92, lower panel. The G+C content of the O-antigen pseudocluster is shown in the upper panel.

3

The genetic organisation of the O-antigen pseudocluster in Y. pestis strain CO92, lower panel. The G+C content of the O-antigen pseudocluster is shown in the upper panel.

Whilst this work was in progress the sequence of the putative O-antigen coding region in Y. pestis strain EV76 was reported [27] and some comparisons were made with the strain CO92 data which is now reported in full here. This analysis indicated that strain EV76, which we have also shown to lack an O-antigen, possesses similar mutations to strain CO92 in the O-antigen gene pseudocluster.

Origin of the O-antigen pseudocluster

A recent analysis of the O-antigen pseudocluster from Y. pestis strain EV76 indicated that it was similar in organisation to the functional O-antigen gene cluster in Y. pseudotuberculosis 0:1b [27]. This finding supports the suggestion by Achtman et al. [25] that Y. pestis has evolved from Y. pseudotuberculosis. The average G+C content across the Y. pestis strain CO92 O-antigen pseudocluster was 41.05% and showed a pattern (Fig. 3) which was similar to that reported in other Gram-negative bacteria [28,29]. The G+C content of the entire genome was 47.6%, whereas, the region between ddhD and gsk had a G+C content of 36.7%, and the G+C content within the 7.4-kb region between prt and gmd was 30.3%. It has been suggested that the lower G+C content of an O-antigen gene cluster provides evidence of interspecific gene transfer [29]. The observation that the Y. pestis O-antigen pseudocluster contained regions with different G+C contents suggests that the ancestral (functional) gene cluster might have been assembled from different sources.

Acknowledgements

This work was supported by Beowulf Genomics, Biotechnology and Biological Sciences Research Council (BBSRC) and the Wellcome Trust (Grants 030825 and 046294). P.G.H. is a recipient of a BBSRC CASE award. We would also like to acknowledge the staff at the Sanger centre for sequencing of Y. pestis strain CO92 and Dr. H. Chart for advice on the analysis of LPS by SDS–PAGE.

References

[1]
Poland
J.D.
Quan
T.J.
Barnes
A.M.
(
1994
)
Plague
. In:
Handbook of Zoonoses: Section A
 ,
2nd
edn. (
Beran
G.W.
, Ed.), pp.
93
112
.
CRC Press
,
London
.
[2]
Butler
T.
(
1983
)
Plague and Other Yersinia Infections
  (
Greenough
W.B.
III
Merigan
T.C.
, Eds.), pp.
3
210
.
Plenum Medical Book Company
,
New York
.
[3]
Perry
R.D.
Fetherston
J.D.
(
1997
)
Yersinia pestis: etiological agent of plague
.
Clin. Microbiol. Rev.
 
10
,
35
66
.
[4]
Albizo
J.M.
Surgalla
M.J.
(
1970
)
Isolation and biological characterization of Pasteurella pestis endotoxin
.
Infect. Immun.
 
2
(
3
),
229
236
.
[5]
Butler
T.
Levin
J.
Linh
N.
Chau
D.
Adickman
M.
Arnold
K.
(
1976
)
Yersinia pestis infection in Vietnam. II. Quantitative blood cultures and detection of endotoxin in the cerebrospinal fluid of patients with meningitis
.
J. Infect. Dis.
 
133
,
493
499
.
[6]
Binns
M.M.
Vaughan
S.
Timmis
K.N.
(
1985
)
O-antigens are essential virulence factors of Shigella sonnei and Shigella dysenteriae 1
.
Zbl. Bakt. Hyg. I Abt. Orig. A
 
181
,
197
205
.
[7]
Reeves
P.
(
1995
)
Role of O-antigen variation in the immune response
.
Trends Microbiol.
 
3
(
10
),
381
386
.
[8]
Skurnik
M.
Mikkola
P.
Toivanen
P.
Tertti
R.
(
1996
)
Passive immunization with monoclonal antibodies specific for lipopolysaccharide (LPS) O-side chain protects mice against intravenous Yersinia enterocolitica serotype 0 infection
.
Acta Pathol. Microbiol. Immunol. Scand.
 
104
(
3
),
598
602
.
[9]
Kessler
A.C.
Haase
A.
Reeves
P.R.
(
1993
)
Molecular analysis of the 3,6-dideoxyhexose pathway genes of Yersinia pseudotuberculosis serogroups IIA
.
J. Bacteriol.
 
175
(
5
),
1412
1422
.
[10]
al Hendy
A.
Toivanen
P.
Skurnik
M.
(
1991
)
The effect of growth temperature on the biosynthesis of Yersinia enterocolitica O:3 lipopolysaccharide: temperature regulates the transcription of the rfb but not of the rfa region
.
Microb. Pathog.
 
10
,
81
86
.
[11]
Chart
H.
Cheasty
T.
Rowe
B.
(
1995
)
Differentiation of Yersinia pestis and Yersinia pseudotuberculosis by SDS–PAGE analysis of lipopolysaccharide
.
Lett. Appl. Microbiol.
 
20
(
6
),
369
370
.
[12]
Davies
D.A.L.
(
1955
)
A specific polysaccharide of Pasteurella pestis
.
Biochem. J.
 
63
,
105
116
.
[13]
Ellwood
D.C.
(
1960
)
The 3-deoxy-d-manno-octulosonic acid found in the lipopolysacchasride of Pasteurella species
.
Proc. Biochem. Soc.
 
47
48
.
[14]
Bordet
C.
Bruneteau
M.
Michel
G.
(
1977
)
Lipopolysaccharides from a reduced-virulence strain of Yersinia pestis
.
Eur. J. Biochem.
 
79
,
443
449
.
[15]
Gremyakova
T.A.
Fomchenkov
V.M.
Fursova
N.K.
Volkovoi
K.I.
(
1996
)
Physicochemical properties of the lipopolysaccharides of Yersinia pestis and R mutants of Enterobacteria: a comparative study
.
Microbiology
 
65
,
665
669
.
[16]
Darveau
R.P.
Charnetzky
W.T.
Hurlbert
R.E.
Hancock
R.E.W.
(
1983
)
Effects of growth temperature, 47-MDa plasmid, and calcium deficiency on the outer membrane protein porin and lipopolysaccharide composition of Yersinia pestis EV76
.
Infect. Immun.
 
42
(
3
),
1092
1101
.
[17]
Galanos
C.
Luderitz
O.
Westphal
O.
(
1969
)
A new method for the extraction of R lipopolysaccharides
.
Eur. J. Biochem.
 
9
,
245
249
.
[18]
Chart
H.
(
1994
)
LPS: isolation and characterisation
. In:
Methods in Practical Laboratory Bacteriology
  (
Raton
B.
Arbor
A.
, Eds.), pp.
11
20
.
CRC Press
,
London
.
[19]
Laemmli
U.K.
(
1970
)
Cleavage of structural proteins during the assembly of the head of bacteriophage T4
.
Nature
 
227
,
680
685
.
[20]
Fleischmann
R.D.
Adams
M.D.
White
O.e.
(
1995
)
Whole-genome random sequencing and assembly of Haemophilus influenzae Rd
.
Science
 
269
,
496
512
.
[21]
Whitfield
C.
Amor
P.A.
Koplin
R.
(
1997
)
Modulation of the surface architecture of Gram-negative bacteria by the action of surface polymer:lipid A-core ligase and by determinants of polymer chain length
.
Mol. Microbiol.
 
23
,
629
638
.
[22]
Skurnik
M.
Toivanen
P.
(
1993
)
Yersinia enterocolitica lipopolysaccharide: genetics and virulence
.
Trends Microbiol.
 
1
,
148
152
.
[23]
Williams
R.C.
Gewurz
H.
Quie
P.G.
(
1972
)
Effects of fraction 1 from Yersinia pestis on phagocytosis in vitro
.
J. Infect. Dis.
 
126
,
235
241
.
[24]
al Hendy
A.
Toivanen
P.
Skurnik
M.
(
1992
)
Lipopolysaccharide O side chain of Yersinia enterocolitica O:3 is an essential virulence factor in an orally infected murine model
.
Infect. Immun.
 
60
,
870
875
.
[25]
Achtman
M.
Zurth
K.
Morelli
G.
Torrea
G.
Guiyoule
A.
Carniel
E.
(
1999
)
Yersinia pestis, the cause of plague, is a recently emerged clone of Yersinia pseudotuberculosis
.
Proc. Natl. Acad. Sci. USA
 
96
,
14043
14048
.
[26]
Moxon
E.R.
Maskell
D.
(
1992
)
Haemophilus influenzae lipopolysaccharide: the biochemistry and biology of a virulence factor
. In:
Molecular Biology of Bacterial Infection
  (
Hormaeche
C.E.
Penn
C.W.
Smyth
C.J.
, Eds.), Vol. 49, pp.
75
96
.
Cambridge University Press
,
Victoria
.
[27]
Skurnik
M.
Peippo
A.
Ervela
E.
(
2000
)
Characterisation of the O-antigen gene cluster of Yersinia pseudotuberculosis and the cryptic O-antigen gene cluster of Yersinia pestis shows that the plague bacillus is most closely related to and has evolved from Y. pseudotuberculosis serotype O1b
.
Mol. Microbiol.
 
37
,
316
330
.
[28]
Jiang
X.M.
Neal
B.
Santiago
F.
Lee
S.J.
Romana
L.K.
Reeves
P.R.
(
1991
)
Structure and sequence of the rfb (O-antigen) gene cluster of Salmonella serovar typhimurium (strain LT2)
.
Mol. Microbiol.
 
5
,
695
713
.
[29]
Reeves
P.
(
1993
)
Evolution of Salmonella O-antigen variation by interspecific gene transfer on a large scale
.
Trends Genet.
 
9
,
17
22
.