TEXT

Top Abstract Text References

In humans, Shigella spp. cause disease ranging from diarrhea to bacillary dysentery. Four species are officially recognizedShigella dysenteriae, Shigella flexneri, Shigella boydii, and Shigella sonneibut all are sufficiently similar to Escherichia coli to be placed in the same species (6, 10, 14, 35, 40). We will refer to Shigella sonnei as Sonnei, a clone of E. coli.

The O antigen, which contains many repeats of an oligosaccharide unit (O unit), is part of the lipopolysaccharide (LPS) present in the outer membrane of gram-negative bacteria. It contributes major antigenic variability to the cell surface and confers O-serotype specificity. There are 166 known O antigens recognized in the E. coli typing scheme (29), and there are 34 among the Shigella strains (11, 28). There is considerable overlap in O antigens of traditional E. coli and Shigella strains of E. coli, and the total number of unique O antigens in E. coli (including Shigella strains) is 190. The surface O antigen is subject to intense selection by the host immune system, which may account for the maintenance of many different O-antigen forms within species such as E. coli.

The genes for O-antigen synthesis are normally grouped together on the chromosome in a gene cluster which maps close to gnd in both E. coli and Salmonella enterica. We, among others, have undertaken an extensive study of the genetic basis of O-antigen variation by sequencing and identifying the O-antigen genes, mostly in S. enterica and E. coli (see reference 42 for a review). It has been found that, in general, O-antigen genes have a low GC content (usually less than 40%). We suggested that this deviation in GC content from that of typical S. enterica or E. coli genes (51%) indicates that the O-antigen DNA originated in a species other than S. enterica or E. coli and was captured by lateral transfer (21). We and others have previously studied DNA recombination events between O-antigen gene clusters within S. enterica (8, 59) and between E. coli and Klebsiella (49). In this study, we demonstrated interspecies transfer of an entire O-antigen gene cluster.

Sonnei has an O antigen (4, 23) not otherwise found in E. coli which is identical to that of serotype 17 of Plesiomonas shigelloides (41, 50). The O-antigen genes of Sonnei are unusual in that they occur on a plasmid, the invasion plasmid (Pinv), which is more than 180 kb in size and is essential for penetration of host epithelial cells (44, 56, 57). It has been shown that this O-antigen gene cluster can hybridize to the chromosomal O-antigen genes of P. shigelloides serotype O17 (52). Most of the plasmid-borne O-antigen gene cluster has been sequenced, and it has been shown that, for at least a few hundred base pairs, the Sonnei O-antigen gene cluster is identical in sequence to that of P. shigelloides O17 (19, 20).

Sonnei strains which have lost the Pinv plasmid lack O antigen (24), suggesting that O-antigen genes at the chromosomal site are nonfunctional. We have recently cloned and sequenced the chromosomal O-antigen genes from Sonnei and showed that this bacterium has a remnant O-antigen gene cluster on the chromosome; most of the original cluster apparently has been deleted by homologous recombination between manB genes in the adjacent O-antigen and colanic acid gene clusters (26). Sonnei, and indeed all Shigella and many other human-pathogenic clones of E. coli, is thought to have relatively recently adapted to the diarrhea-causing mode of pathogenesis in humans (12, 32). The finding that a presumably typical chromosomal O-antigen gene cluster was lost by deletion indicated that to adapt to a new niche, Sonnei gained new O-antigen genes on a plasmid and that this was followed by inactivation of the original chromosomal O-antigen gene cluster.

In this study, we sequenced the entire O-antigen gene cluster of P. shigelloides O17, as well as part of the Sonnei plasmid-borne O-gene cluster to obtain the full sequence. Comparison of the two sequences showed that they are almost identical, with the exception of the wzz and wbgZ genes. The high level of similarity indicates that the O-antigen gene cluster has recently transferred from one species to the other. We assume that because the Sonnei O-antigen genes are on a plasmid, the transfer has been from P. shigelloides to E. coli; the high level of similarity for most genes suggests that only recently did Sonnei gained its current O antigen. The level of difference between the two wzz and wbgZ genes is higher than that between other genes of the two gene clusters, and there is evidence that the wzz gene on the plasmid has been under selection for change in the new host.

Sequences of O-antigen gene clusters. Figure 1 shows the O-antigen gene clusters of P. shigelloides O17 strain C27 (ATCC 14030) and the Pinv plasmid of E. coli Sonnei strain 53G1. The segment from positions 733 to 11750 of plasmid Pinv has already been sequenced (19), and the corresponding region in strain C27 (positions 733 to 10598 [Fig. 1]) was PCR amplified and sequenced by using oligonucleotide primers synthesized based on the published sequence of Pinv. Because these two sequences are highly similar (see below), we also resequenced this region from Pinv to make sure that variations observed were not due to sequencing errors. We found some discrepancies between our sequence of Pinv and the sequence obtained by Houng and Venkatesan (19). We have sequenced each of the relevant bases at least three times and are confident that our sequence is correct.

View larger version (21K): [in this window] [in a new window]   FIG. 1.   Comparison of the Sonnei plasmid-borne O-antigen gene cluster with that of P. shigelloides O17.

The O-antigen gene cluster is located between a JUMPstart sequence and an aquaporin gene. The JUMPstart sequence is located upstream of O-antigen gene clusters. We used a primer which binds to it to amplify and sequence the 5' ends of the gene clusters, and for both gene clusters the first base downstream of the JUMPstart sequence is treated as position 1.

O-antigen genes. There are 10 genes in the gene cluster of strain C27, and there are 10 genes plus an insertion sequence (IS) in the gene cluster of plasmid Pinv; all genes have the same transcriptional direction from JUMPstart to the aquaporin gene (Fig. 1). The nucleotide and amino acid sequences were used to search available databases for indications of possible function. MULTICOMP software (43) was used to do pairwise comparisons of DNA and derived amino acid sequences and to estimate the number of synonymous substitutions per synonymous site (K_S) and the number of nonsynonymous substitutions per nonsynonymous site (K_A).

(i) orf1. The deduced amino acid sequence of the protein encoded by orf1 exhibits about 46% similarity and 26% identity to that of the wzz gene of E. coli O111 (3). Wzz proteins are characterized by two transmembrane domains, located in the amino-terminal and carboxy-terminal regions, and a large hydrophilic central domain located in the periplasm (34). Orf1 has a similar hydrophobic profile, and we have named orf1 wzz.

(ii) orf2 and orf3. The deduced amino acid sequences of the proteins encoded by orf2 and orf3 have significant homology to a variety of UDP-glucose/GDP-mannose dehydrogenase and nucleotide sugar epimerase genes, respectively. WcdA and WcdB of the S. enterica serovar Typhi Vi antigen gene cluster (16, 53) have the highest levels of identity to Orf2 (63%) and Orf3 (64%), respectively. orf2 and orf3, which are thought to be sugar pathway genes, have been named wbgT and wbgU, respectively.

(iii) orf4. The orf4 gene is predicted to encode an integral inner-membrane protein with 12 transmembrane segments, suggesting that it is the wzx gene. The predicted protein also has the conserved motif found near the amino-terminal end of Wzx (48). orf4 has been named wzx.

(iv) orf5. The orf5 gene encodes a protein containing eight predicted transmembrane segments with a large cytoplasmic loop (83 amino acids). This inner-membrane topology is a characteristic feature of all known O-antigen polymerases (33), and we believe that orf5 is the O-antigen polymerase gene, wzy.

(v) orf6. The deduced amino acid sequence of the protein encoded by orf6 shows similarity to many NADH dehydrogenases. It is thought to be a sugar pathway gene and has been named wbgV.

(vi) orf7. The deduced amino acid sequence of the orf7-encoded protein exhibits 54.2% similarity to WaaV, a glucosyl transferase involved in the synthesis of the LPS core in E. coli (17). It is highly likely that this gene, which has been named wbgW, encodes a transferase.

(vii) orf8. The deduced amino acid sequence of the protein encoded by orf8 exhibits 75.5% similarity and 56.5% identity to that of the wlbF gene of the Bordetella pertussis O-antigen gene cluster, which shows similarity to a group of nucleotide sugar aminotransferase genes (2). orf8 is also similar to this group of genes, which includes the per gene (36.2% identity and 59.3% similarity at the amino acid level) of the E. coli O157 O-antigen gene cluster. This gene, which is highly likely to be involved in the synthesis of 4n-FucNAc (see below), has been named wbgX.

(viii) orf9. The deduced amino acid sequence of the orf9-encoded protein shows 63% similarity and 39% identity to the C-terminal half of that of wbaP of the S. enterica O4 gene cluster (21). WbaP is the galactosyltransferase catalyzing the first step of S. enterica O4-antigen synthesis (21), and it has been shown that it is the C-terminal half that encodes the transferase function (54). We have shown that all or C-terminal half of WbaP exhibits similarity to a protein family whose members all catalyze the first step in polysaccharide synthesis (54). orf9, which has been named wbgY, is thought to encode the transferase catalyzing the first step of O-antigen synthesis.

(ix) orf10. The deduced protein product of orf10 is homologous to the members of the WbcP family of proteins, which are required for the biosynthesis of surface polysaccharides in numerous bacterial species and may function as nucleotide sugar epimerases or dehydratases (9). Among the proteins of this family, WbcP of the Yersinia enterocolitica LPS outer-core gene cluster (47) is most similar to Orf10, exhibiting 82.4% similarity and 68.5% identity. orf10, a putative sugar pathway gene, has been named wbgZ.

View larger version (10K): [in this window] [in a new window]   FIG. 2.   Proposed pathway for conversion of UDP-GlcNAc to UDP-4n- FucNAc.

The plasmid-borne Sonnei O-antigen gene cluster has an IS630 element. Houng and Venkatesan (19) found that between the wzy and wbgV genes in the O-antigen gene cluster of plasmid Pinv there is a 1,153-bp IS (positions 6215 to 7367) which is absent in the O-antigen gene cluster of strain C27 (Fig. 1). Comparison of the two gene clusters shows that the IS is within the sixth codon from the end of the wzy gene, making the wzy gene in plasmid Pinv four codons shorter than that in the C27 cluster and generating a duplication of the two bases (TA) upstream of the insertion site (Fig. 3). The IS, which is an IS630 element, is 1,153 bp in length and is almost identical (99.4% identity) to the IS630 element of Sonnei identified by Matsutani et al. (31). There is a transposase gene in the IS and a 29-bp inverted repeat at each end as described by Matsutani et al. (31).

View larger version (9K): [in this window] [in a new window]   FIG. 3.   The IS630 integration site within the Sonnei plasmid-borne O-antigen gene cluster. Tpase, transposase.

Relationship between the O-antigen gene clusters of P. shigelloides O17 and E. coli Sonnei. The functional O-antigen genes of Sonnei are on a plasmid, and we have previously shown that this bacterium has a remnant O-antigen gene cluster on its chromosome (26). In the present study, we showed that the plasmid-borne Sonnei O-antigen gene cluster is almost identical to that of P. shigelloides O17. The gene order is conserved between the two gene clusters, except for the presence of an IS630 element between wzy and wbgV in Sonnei. The level of DNA identity is generally high between corresponding genes of the two gene clusters, ranging from 99.1 to 100%, except for wzz and wbgZ. Thus, the transfer of this gene cluster from one species to the other must have been a recent event. Because the gene cluster is located on a plasmid and has an IS630 element inserted in the case of Sonnei, it is highly likely that the direction of the transfer was from P. shigelloides to E. coli.

The chromosomal wzz gene in Sonnei. The E. coli wzz gene is usually located outside of the O-antigen gene cluster in a region between the gnd gene and the his operon (3). We showed previously that most of the Sonnei chromosomal O-antigen gene cluster had been deleted, but we did not look at the wzz gene.

Chain length of the Sonnei O antigen. There is evidence of an association between O-antigen chain length and pathogenicity in E. coli, and it has been suggested that the specific chain length range of the O antigen may be an important virulence factor (13). Sonnei produces an LPS with a basic chain length of 18 to 25 O units (51). The Sonnei chromosomal wzz gene encodes a typical intermediate-length Wzz (for a chain length of 10 to 18 O units), and we believe that the O-antigen chain length is determined by the protein encoded by the plasmid wzz gene. Comparison of the Sonnei plasmid-borne wzz with that of P. shigelloides O17 showed a relatively high level of nonsynonymous substitutions. Sonnei has a shorter O-antigen chain length than P. shigelloides O17 (51), and we suggest that a shorter chain length better suits the niche occupied by Sonnei and that there has been selection pressure on the Sonnei plasmid-borne wzz gene to change.

Time of transfer of the O-antigen gene cluster to Sonnei. The sequence variation at synonymous sites (Fig. 1) was used to estimate when the last common ancestor existed. wzz, wbgZ, and aqpZ were excluded for reasons given above. Two synonymous clock rates have been proposed for E. coli. Based on the proposal that E. coli and S. enterica last shared a common ancestor about 140 million years ago and that 95% of synonymous sites have been exchanged since then, Whittam calculated the rate of synonymous-polymorphism accumulation to be 6 × 10⁹ per year (58). A higher rate of 3 × 10⁸ was calculated by Guttman and Dykhuizen on the basis that the mutation rate is about 10¹⁰ and that E. coli undergoes approximately 300 divisions per year under natural conditions (15). The mean K_S value for these eight genes is 0.01339, and the date since divergence of the two O-antigen gene clusters was estimated to be 45,000 to 220,000 years, using the two clock rates. This estimate assumes similar clock rates for P. shigelloides and E. coli. Furthermore, the average generation time for Shigella may be different from that of other E. coli strains since Shigella forms have a rapid buildup of numbers during infection. Thus, the figures obtained here can only be thought of as estimates.

Nucleotide sequence accession number. The DNA sequences have been deposited in GenBank under accession no. AF285970 (the O-antigen gene cluster of P. shigelloides O17), AF285971 (the plasmid-borne O-antigen gene cluster of Sonnei), and AF295304 (the chromosomal wzz gene of Sonnei).