- Volume 148, Issue 11, 2002

To determine the function of the waaE gene in the biosynthesis of the inner-core LPS of Klebsiella pneumoniae, a waaE non-polar mutant has been constructed. Data obtained from the comparative chemical analysis of LPS samples obtained from the wild-type, the mutant strain and the complemented mutant demonstrated that the waaE gene is involved in substitution of α-L-glycero-D-manno-heptopyranose I (L,D-HeppI) at the O-4 position by a β-D-glucopyranose (β-D-Glcp) residue. In addition, DNA amplification and nucleotide sequence determination studies revealed that waaE homologues located between the waaA and coaD genes are present in clinical isolates of Enterobacteriaceae containing the structure β-D-Glcp-(1→4)-α-L,D-HeppI (K. pneumoniae, Proteus mirabilis and Yersinia enterocolitica), as well as in strains of Serratia marcescens and Enterobacter aerogenes of unknown LPS-core structures. Complementation studies using non-polar waaE mutants prove that all the waaE homologues perform the same function. Furthermore, K. pneumoniae, Ser. marcescens and P. mirabilis non-polar waaE mutants showed reduced adhesion and pathogenicity. In addition, the Ser. marcescens and P. murabilis waaE mutants showed reduced swarming motility and ability to form biofilms in vitro. All these characteristics were rescued by reintroduction of the waaE gene independently of its origin. An easy DNA amplification method to detect this gene was established, which also helps in finding the potential presence of this structural feature [β-D-Glcp-(1→4)-α-L,D-HeppI] in the inner-core LPS of Enterobacteriaceae members with unknown LPS-core structures.

FhuA is a multifunctional protein in the outer membrane of Escherichia coli that actively transports Fe3+–ferrichrome and the antibiotics albomycin and rifamycin CGP 4832, and serves as a receptor for the unrelated phages T5, T1, ϕ80 and UC-1, colicin M and microcin J25. The energy source for active transport is the proton-motive force of the cytoplasmic membrane, which is required for all FhuA functions except infection by phage T5, and is thought to be mediated to the outer-membrane receptor FhuA by the TonB protein. The crystal structure of FhuA consists of a β-barrel that is closed by a globular domain. The proximal region carries the TonB box (residues 7–11), for which genetic evidence exists that it interacts with the region around residue 160 of TonB. However, deletion of the TonB box along with the globular domain results in a protein, FhuAΔ5–160, that still displays TonB-dependent active ferrichrome transport across the outer membrane and confers sensitivity to the FhuA ligands. In this study synthetic nonapeptides identical in sequence to amino acids 150–158, 151–159, 152–160, 153–161 and 158–166 of TonB were shown to reduce ferrichrome transport of cells via wild-type FhuA and the corkless derivative FhuAΔ5–160, which suggests that this TonB region is involved in the interaction of TonB with the β-barrel of FhuA. TonB missense mutants reduced the activity of FhuA and FhuAΔ5–160. TonB proteins of different Enterobacteriaceae activated FhuA and FhuAΔ5–160 to a similar degree. TonB of Pantoea agglomerans displayed low activity in an E. coli tonB mutant. Sequencing of the tonB gene of P. agglomerans revealed differences from E. coli TonB in the region around residue 160 of the deduced protein; these differences might contribute to the lower activity of the P. agglomerans TonB protein when coupled to the E. coli FhuA protein. The data support the theory that the β-barrel receives the energy from the cytoplasmic membrane via TonB and responds to the energy input and thus represents the transporting domain of FhuA.

The truB gene of Escherichia coli encodes the pseudouridine-55 (ψ55) synthase and is responsible for modifying all tRNA molecules in the cell at the U55 position. A truB null mutant grew normally on all growth media tested, but exhibited a competitive disadvantage in extended co-culture with its wild-type progenitor. The mutant phenotype could be complemented by both the cloned truB gene and by a D48C, catalytically inactive allele of truB. The truB mutant also exhibited a defect in survival of rapid transfer from 37 to 50 °C. This mutant phenotype could be complemented by the cloned truB gene but not by a D48C, catalytically inactive allele of truB. The temperature sensitivity of truB mutants could be enhanced by combination with a mutation in the trmA gene, encoding an m5U-methyltransferase, modifying the universal U54 tRNA nucleoside, but not by mutations in trmH, encoding the enzyme catalysing the formation of Gm18. The truB mutant proteome contained altered levels of intermediates involved in biogenesis of the outer-membrane proteins OmpA and OmpX. The truB mutation also reduced the basal expression from two σE promoters, degP and rpoHP3. Three novel aspects to the phenotype of truB mutants were identified. Importantly the data support the hypothesis that TruB-effected ψ55 modification of tRNA is not essential, but contributes to thermal stress tolerance in E. coli, possibly by optimizing the stability of the tRNA population at high temperatures.

Inferred amino acid sequences of the methyl coenzyme-M reductase (mcrA) gene from five different methanogen species were aligned and two regions with a high degree of homology flanking a more variable region were identified. Analysis of the DNA sequences from the conserved regions yielded two degenerate sequences from which a forward primer, a 32-mer, and a reverse primer, a 23-mer, could be derived for use in the specific PCR-based detection of methanogens. The primers were successfully evaluated against 23 species of methanogen representing all five recognized orders of this group of Archaea, generating a PCR product between 464 and 491 bp. Comparisons between the mcrA and 16S small subunit rRNA gene sequences using PHYLIP demonstrated that the tree topologies were strikingly similar. Methods were developed to enable the analysis of methanogen populations in landfill using the mcrA gene as the target. Two landfill sites were examined and 63 clones from a site in Mucking, Essex, and 102 from a site in Odcombe, Somerset, were analysed. Analysis revealed a far greater diversity in the methanogen population within landfill material than has been seen previously.

In this study, high-affinity maltose- and glucose-binding activities in cell-free extracts of Thermotoga maritima were detected; these activities were distinct and specific. At the gross level, the expression of binding-protein activities was repressed by growth of T. maritima in the presence of the cognate sugar. Growth of the organism in the presence of maltose reduced maltose-binding activity but not glucose-binding activity, while growth in the presence of glucose reduced glucose-binding activity but not maltose-binding activity. In competition assays, these binding activities showed distinct patterns of substrate specificity: whereas the maltose-binding activity showed specificity for α-linked glucosides, the glucose-binding activity showed a broader specificity. All maltose- and glucose-binding activity was found in the supernatant retrieved following centrifugation (100000  g ) of the cell-free extracts prepared by French-pressure-cell treatment; no activity was found in an octyl-glucoside-treated extract of the membrane fraction. The maltose-binding-protein activity was recovered from the periplasmic fraction by selective release of the periplasmic contents of T. maritima cells using a newly developed freeze–thaw procedure. Annotation of the complete genome sequence of T. maritima suggests that there may be at least two maltose-binding proteins, MalE1 and MalE2, encoded in the genome. The maltose-binding activity corresponded to a protein of 43 kDa, which was consistent in size with either of the putative proteins. These data demonstrate that the hyperthermophilic bacterium T. maritima possesses separate maltose- and glucose-binding-protein activities that are freely soluble in its periplasm, in contrast to the membrane-bound sugar-binding proteins found in archaeal hyperthermophiles.

GTP-binding proteins are found in all domains of life and are involved in various essential cellular processes. With the recent explosion of available genome sequence data, a widely distributed bacterial subfamily of GTP-binding proteins was discovered, represented by the Escherichia coli Era and the Bacillus subtilis Obg proteins. Although only a limited number of theGTP-binding proteins belonging to the subfamily have been experimentally characterized, and their function remains unknown, the available data suggests that many of them are essential to bacterial growth. When the complete genomic sequence of B. subtilis was surveyed for genes encoding GTP-binding proteins of the Era/Obg family, nine such genes were identified. As a first step in elucidating the functional networks of those nine GTP-binding proteins, data presented here indicates that six of them are essential for B. subtilis viability. Additionally, it is shown that the six essential proteins are able to specifically bind GTP and GDP in vitro. Experimental depletion of the essential GTP-binding proteins was examined in the context of cell morphology and chromosome replication, and it was found that two proteins, Bex and YqeH, appeared to participate in the regulation of initiation of chromosome replication. Collectively, these results suggest that members of the GTP-binding Era/Obg family are important proteins with precise, yet still not fully understood, roles in bacterial growth and viability.

The genes norCBQD that encode the bc-type nitric oxide reductase from Bradyrhizobium japonicum USDA110 have been isolated and characterized. norC and norB encode the cytochrome c-containing subunit II and cytochrome b-containing subunit I of nitric oxide reductase, respectively. norQ encodes a protein with an ATP/GTP-binding motif, and the predicted norD gene product shows similarity with NorD from other denitrifiers. Mutational analysis indicates that the two structural norC and norB genes are required for microaerobic growth under nitrate-respiring conditions. A mutant strain lacking a functional norC gene also lacked the 16 kDa c-type cytochrome that is normally detectable by haem-staining of proteins from membranes of microaerobically grown wild-type cells. Expression of a transcriptional fusion of the nor promoter region to the reporter gene lacZ (PnorC–lacZ) was not detected in aerobically grown cells of USDA110, but the fusion was induced threefold when the cells were cultured under microaerobic conditions (1% O2) with either nitrite or nitric oxide, and about 18-fold when nitrate was the N oxide present in the medium. The PnorC–lacZ fusion was not expressed in the B. japonicum fixK 2 mutant strain 9043, but complementation of the mutant with the fixK 2 gene restored β-galactosidase activity to levels similar to those found in the parental strain. The promoter region of the norCBQD genes has been characterized by primer extension. A major transcript initiates 45·5 bp downstream of the centre of a putative binding site for the transcription factor FixK2.

Nisin produced by Lactococcus lactis subsp. lactis is a 34-residue antibacterial polypeptide and belongs to a group of post-translationally modified peptides, lantibiotics, with dehydrated residues and cyclic amino acids, lanthionines. These modifications are supposed to be made by enzymes encoded by lanB and lanC genes, found only in biosynthetic operons encoding lantibiotics. To analyse the extent of modification, His-tagged nisin precursors were expressed in nisB and nisC mutant strains. The His-tagged nisin precursors were purified from the cytoplasm of the cells, as lack of NisB or NisC activity impaired translocation of the nisin precursor. The purified His-tagged polypeptides were analysed with trypsin digestion followed by nisin bioassay, SDS-PAGE, N-terminal sequencing and mass spectroscopy. According to the results, nisin precursors from the strain lacking NisB activity were totally unmodified, whereas nisin precursors from the strain lacking NisC activity, but having NisB activity, were dehydrated and devoid of normal lanthionine formation. This is the first experimental evidence showing that NisB is required for dehydration and NisC for correct lanthionine formation in nisin maturation.

A detailed study on the geographic distribution, molecular diversity and evolutionary relationships of 24 closely related variants of the Tn5041 transposon found among 182 mercury resistant environmental Gram-negative strains from the IMG-Hg Reference Collection is reported here. RFLP analysis, followed by the determination of partial DNA sequences, identified 14 distinct types of these transposons, which differed from each other by 1–7 single-event DNA polymorphisms. No polymorphisms were detected at the right arm of the transposons except an insertion of a new mobile DNA element carrying a mer operon (named the mer2 cassette) within the Tn5041 mer operon. According to the model presented here, the insertion occurred via homologous recombination with a circular form of the mer2 cassette. A total of 8 point mutations, 1 internal deletion, 2 end-involving deletions, 3 mosaic regions and 2 insertions were detected at the left arm of the transposons. The insertions were a transposon closely related to Tn21 but lacking the integron and a new group II intron (named INT5041C). Inspection of the geographic distribution of the Tn5041 variants suggested that at least three long-distance waves of dissemination of these variants had occurred, accompanied by homologous recombination between different Tn5041 lineages. Movements of circular DNAs by homologous recombination as a source of mosaic genes and new mer genes, and formation of unusual mosaics ending or beginning at the Tn5041 att site are discussed.

Escherichia coli LexA protein is the repressor of a gene network whose members are directly involved in the repair of damaged DNA and in the survival of bacterial cells until DNA lesions have been eliminated. The lexA gene is widely present in bacteria, although the sequences of only three LexA-binding sites are known: Gram-positive, alpha Proteobacteria and some members of gamma Proteobacteria represented by E. coli. Taking advantage of the fact that the genome sequence of the plant-pathogenic bacterium Xylella fastidiosa has been determined, its lexA gene has been cloned and overexpressed in E. coli to purify its product. After demonstration that X. fastidiosa lexA and recA genes are co-transcribed, gel mobility shift assays and directed mutagenesis experiments using the promoter of the lexA–recA transcriptional unit demonstrated that the X. fastidiosa LexA protein specifically binds the imperfect palindrome TTAGN6TACTA. This is the first LexA binding sequence identified in the gamma Proteobacteria differing from the E. coli-like LexA box. Although a computational search has revealed the presence of TTAGN6TACTA-like motifs upstream of X. fastidiosa genes other than lexA, X. fastidiosa LexA only binds the promoter of one of them, XF2313, encoding a putative DNA-modification methylase. Moreover, X. fastidiosa LexA protein does not bind any of the other genes whose homologues are regulated by the LexA repressor in E. coli (uvrA, uvrB, ssb, ruvAB, ftsK, dinG, recN and ybfE). RT-PCR quantitative analysis has also demonstrated that lexA–recA and XF2313 genes, as well as the X. fastidiosa genes which are homologues to those of E. coli belonging to the LexA regulon, with the exception of ssb, are DNA damage-inducible in X. fastidiosa.

Differential expression of the Streptococcus salivarius 57.I urease operon in response to pH is effected by repression of transcription from a proximal promoter, PureI. To localize the cis-acting elements involved in the regulation of the urease operon, the intact promoter region and its derivatives were generated and fused to a promoterless chloramphenicol acetyltransferase (cat) gene. The promoter–cat fusions were established in the lacZ gene of S. salivarius by using a newly constructed integration vector. CAT-specific activities were examined in batch-grown cells at pH 7·5 and 5·5. The results indicated that a 21 bp region immediately 5′ to the −35 element was required for efficient repression of PureI at neutral pH and that the 39 bp (−57 to −95) 5′ to this region contained sequences required for optimal expression of PureI. A potential secondary repressor-binding site was tentatively identified further upstream of the −35 element (−96 to −115). To further analyse the cis-acting elements, base changes were introduced into two AT-rich repeats within the primary repressor-binding site. One such derivative, S. salivarius M1, with five base substitutions immediately 5′ to the −35 element, expressed 20-fold more CAT-specific activity at neutral pH than the strain carrying wild-type PureI–cat. Also, the pH sensitivity of strain M1 was greatly reduced, suggesting that this AT-rich region is crucial for repression of the urease operon. Deletion of three consecutive 15- or 16-base segments from −52 to −96 in the S. salivarius M1 background resulted in lower activities compared to strain M1, confirming the presence of sequences required for optimal expression of the operon. All of the PureI–cat fusions were also integrated into the gtfG gene of Streptococcus gordonii DL1, a non-ureolytic oral Streptococcus sp. Repression of PureI was observed at neutral pH in S. gordonii and the effects of the various mutations of the repressor-binding site largely paralleled those seen in S. salivarius, suggesting that the cis-elements may be a target for a global regulatory circuit that controls gene expression in streptococci in response to pH.

Four potential binding sites for LexA were identified upstream of the Mycobacterium tuberculosis lexA gene. A mutational analysis of these sites in a lexA–lacZ reporter construct revealed that only one of these SOS boxes was required for DNA-damage-mediated regulation of lexA expression. A novel DNA-damage-inducible gene, Rv2719c, was identified that was divergently transcribed relative to lexA; the other three SOS boxes were found to be involved in regulating expression of this novel mycobacterial-specific gene. The SOS boxes lay in the respective promoter regions of the genes that they regulated.

‘Pseudomonas butanovora’ is capable of growth with butane via the oxidation of butane to 1-butanol, which is catalysed by a soluble butane monooxygenase (sBMO). In vitro oxidation of ethylene (an alternative substrate for sBMO) was reconstituted in the soluble portion of cell extracts and was NADH-dependent. Butane monooxygenase was separated into three components which were obligately required for substrate oxidation. The N-terminal sequences of the peptides associated with butane monooxygenase led to the cloning and sequencing of the 5797 nucleotide bmo gene cluster. Comparisons of the deduced amino acid sequences with other multicomponent monooxygenases suggest that sBMO is a multimeric hydroxylase with 61, 45 and 19 kDa subunits encoded by bmoXYZ, a 40 kDa oxidoreductase encoded by bmoC, and a 15 kDa regulatory protein encoded by bmoB. A sixth structural gene (bmoD) encodes a 9·6 kDa protein with similarity exclusively to mmoD (orfY), a putative metal centre assembly protein of the soluble methane monooxygenases. Insertional inactivation of bmoX resulted in a mutant ‘P. butanovora’ strain incapable of growth with butane. A putative promoter element characteristic of promoters associated with σ54-dependent transcription initiation was located upstream of the bmo genes. Expression of all six genes was detected in butane-induced cells. Butane monooxygenase from ‘P. butanovora’ aligns most closely with non-haem carboxylate-bridged diiron monooxygenases and, moreover, contains the characteristic iron-binding motif. The structural and mechanistic implications of the high sequence identity (up to 64%) between the peptides of butane monooxygenase and methane monooxygenases are discussed.

The tetrathionate (Ttr) and thiosulfate (Phs) reductases of Salmonella enterica LT2, together with the polysulfide reductase (Psr) of Wolinella succinogenes, are unusual examples of enzymes containing a molybdopterin active-site cofactor since all formally catalyse sulfur–sulfur bond cleavage. This is in contrast to the oxygen or hydrogen transfer reactions exhibited by other molybdopterin enzymes. Here the catalytic specificity of Ttr and Phs has been compared using both physiological and synthetic electron-donor systems. Ttr is shown to catalyse reduction of trithionate but not sulfur or thiosulfate. In contrast, Phs cannot reduce tetrathionate or trithionate but allows whole cells to utilize elemental sulfur as an electron acceptor. Mechanisms are proposed by which the bacterium is able to utilize an insoluble sulfur substrate by means of reactions at the cytoplasmic rather than the outer membrane.

Legionella pneumophila has been shown to induce apoptosis within macrophages, monocytic cell lines and alveolar epithelial cells. The mechanisms and significance of L. pneumophila-associated apoptosis are not well understood. It has been speculated that L. pneumophila may induce apoptosis through ligation of death receptors by bacterial surface components or by secreted bacterial factors. Translocation of apoptotic factor(s) through the Dot/Icm secretion machinery followed by direct activation of caspases within the cytosol is discussed as another possible mechanism of apoptosis induction by L. pneumophila. Here, it is shown that L. pneumophila induced the mitochondrial release of cytochrome c in CD95 (Fas/Apo-1)-negative monocytic Mono Mac 6 cells, indicating that Legionella-induced apoptosis is mediated via the mitochondrial signalling pathway. In addition, blocking of the death receptor pathway at distinct stages using CD95-, FADD- or caspase-8-deficient Jurkat cells did not affect induction of apoptosis by L. pneumophila. Conversely, inhibition of the mitochondrial death pathway by overexpression of the anti-apoptotic protein Bcl-2 potently inhibited the processing of caspases and the induction of apoptosis. Therefore, these findings support a model in which the induction of apoptosis by L. pneumophila is mediated by activation of the intrinsic mitochondrial death pathway in the absence of external death receptor signalling.

Total genomic DNA from samples of intact mouse small intestine, large intestine, caecum and faeces was used as template for PCR amplification of 16S rRNA gene sequences with conserved bacterial primers. Phylogenetic analysis of the amplification products revealed 40 unique 16S rDNA sequences. Of these sequences, 25% (10/40) corresponded to described intestinal organisms of the mouse, including Lactobacillus spp., Helicobacter spp., segmented filamentous bacteria and members of the altered Schaedler flora (ASF360, ASF361, ASF502 and ASF519); 75% (30/40) represented novel sequences. A large number (11/40) of the novel sequences revealed a new operational taxonomic unit (OTU) belonging to the Cytophaga–Flavobacter–Bacteroides phylum, which the authors named ‘mouse intestinal bacteria’. 16S rRNA probes were developed for this new OTU. Upon analysis of the novel sequences, eight were found to cluster within the Eubacterium rectale–Clostridium coccoides group and three clustered within the Bacteroides group. One of the novel sequences was distantly related to Verrucomicrobium spinosum and one was distantly related to Bacillus mycoides. Oligonucleotide probes specific for the 16S rRNA of these novel clones were generated. Using a combination of four previously described and four newly designed probes, approximately 80% of bacteria recovered from the murine large intestine and 71% of bacteria recovered from the murine caecum could be identified by fluorescence in situ hybridization (FISH).

Genes encoding pediocin-like bacteriocins are usually co-transcribed with a gene encoding a cognate immunity protein. To investigate the functionality and specificity of immunity proteins, immunity genes belonging to the bacteriocins curvacin A, enterocin A, enterocin P, leucocin A, pediocin PA-1 and sakacin P, as well as a putative immunity gene, orfY, were expressed in three bacteriocin-sensitive lactic acid bacteria (Lactobacillus sake, Carnobacterium piscicola and Enterococcus faecalis). The transformed indicator strains, each containing one of the immunity genes, were tested for sensitivity towards seven different purified bacteriocins (curvacin A, enterocin A, enterocin P, leucocin A, leucocin C, pediocin PA-1 and sakacin P). Cross-immunity was observed almost exclusively in situations where either the bacteriocins or the immunity proteins belonged to the same sequence-based subgroup. In a few cases, the functionality of immunity proteins was strain-dependent; e.g. the leucocin A immunity gene provided immunity to enterocin A, pediocin PA-1 and leucocin A in Ent. faecalis, whereas in the other two indicators, this gene provided immunity to leucocin A only. The orfY gene, which is transcribed without a cognate bacteriocin, was shown to encode a functional immunity protein that expands the bacteriocin resistance of the strain possessing this gene. The results show that the bacteriocin sensitivity of a lactic acid bacterium strain can depend on (1) the presence of immunity genes in connection with its own bacteriocin production, (2) the presence of extra immunity genes and (3) more general properties of the strain such as the membrane composition or the presence of receptors.

Helicobacter pylori is a Gram-negative bacterium that is associated with the development of peptic ulcers and gastric carcinoma in humans. This species appears to be one of the most genetically variable bacteria described to date. The overall level of heterogeneity within strains of this organism was determined by comparing the genome sequences of two reference strains, J99 and 26695. The aim of this study was to measure the genetic diversity within strains of H. pylori by looking for strain-specific genes in nine H. pylori strains isolated from patients suffering from chronic gastritis (n=3), duodenal ulcers (n=3) or gastric cancer (n=3). Seven loci that contained strain-specific genes in strains J99 and 26695 were studied. These regions were subsequently amplified from most of the clinical isolates studied and their sequences were determined. ORFs were predicted from the sequence data and were compared to sequences within the databases. The results showed that the genes flanking the ORFs specific to either strain J99 or strain 26695 were also present in a similar configuration in the genomes of the nine clinical isolates. Moreover, in most regions, ORFs homologous to those found in the corresponding loci in the two reference strains were detected. However, in 10 regions, genes similar to those located at another locus in the genome of J99 or 26695 were found. Finally, six strain-specific genes were identified in three regions of three of the H. pylori strains isolated from patients with duodenal ulcers (n=2) and gastric cancer (n=1). Of these six genes, five were putative genes and one was an orthologue of a gene encoding a transposase in Thermotoga maritima. However, no association with disease was found for these genes.

Acquisition of virulence genes encoded on mobile genetic elements has played an important role in the emergence of pathogenic isolates of Vibrio cholerae, the causative agent of the diarrhoeal disease cholera. The genes encoding cholera toxin (ctxAB), the main cause of profuse secretory diarrhoea in cholera, are encoded on a filamentous bacteriophage CTXϕ. The toxin coregulated pilus (TCP), an essential intestinal colonization factor, was originally designated as part of a pathogenicity island named the Vibrio pathogenicity island (VPI), but this island has more recently been proposed to be the genome of a filamentous phage, VPIϕ. In this study, it is shown that nanH, which encodes neuraminidase, maps within a novel pathogenicity island designated VPI-2. The 57·3 kb VPI-2 has all of the characteristic features of a pathogenicity island, including the presence of a bacteriophage-like integrase (int), insertion in a tRNA gene (serine) and the presence of direct repeats at the chromosomal integration sites. Additionally, the G+C content of VPI-2 (42 mol%) is considerably lower than that of the entire genome (47 mol%). VPI-2 encodes several gene clusters, such as a restriction modification system (hsdR and hsdM) and genes required for the utilization of amino sugars (nan-nag region) as well as neuraminidase. To determine the distribution of VPI-2 among V. cholerae, 78 natural isolates were examined using PCR and Southern hybridization analysis for the presence of this region. All toxigenic V. cholerae O1 serogroup isolates examined contained VPI-2, whereas non-toxigenic isolates lacked the island. Of 14 V. cholerae O139 serogroup isolates examined, only one strain, MO2, contained the entire 57·3 kb island, whereas 13 O139 isolates contained only a 20·0 kb region with most of the 5′ region of VPI-2 which included nanH deleted in these strains.