Dynamics of the bacterial genome: deletions and integrations as mechanisms of bacterial virulence modulation

Pathogenicity islands: a molecular toolbox for bacterial virulence

Pathogenicity islands (PAIs) are distinct genetic elements on the chromosomes of a large number of bacterial pathogens. PAIs encode various virulence factors and are normally absent from non-pathogenic strains of the same or closely related species. PAIs are considered to be a subclass of genomic islands that are acquired by horizontal gene transfer via transduction, conjugation and transformation, and provide 'quantum leaps' in microbial evolution. Data based on numerous sequenced bacterial genomes demonstrate that PAIs are present in a wide range of both gram-positive and gram-negative bacterial pathogens of humans, animals and plants. Recent research focused on PAIs has not only led to the identification of many novel virulence factors used by these species during infection of their respective hosts, but also dramatically changed our way of thinking about the evolution of bacterial virulence.

Use of diagnostic microarrays for determination of virulence gene patterns of Escherichia coli K1, a major cause of neonatal meningitis

Forty Escherichia coli strains isolated primarily from neonatal meningitis, urinary tract infections and feces were screened for the presence of virulence genes with a newly developed microarray on the array tube format. A total of 32 gene probes specific for extraintestinal as well as intestinal E. coli pathotypes were included. Eighty-eight percent of the analyzed strains were positive for the K1-specific probe on the microarray and could be confirmed with a specific antiserum against the K1 capsular polysaccharide. The gene for the hemin receptor ChuA was predominantly found in 95% of strains. Other virulence genes associated with K1 and related strains were P, S, and F1C fimbriae specific for extraintestinal E. coli, the genes for aerobactin, the alpha-hemolysin and the cytotoxic necrotizing factor. In two strains, the O157-specific catalase gene and the gene for the low-molecular-weight heat-stable toxin AstA were detected, respectively. A total of 19 different virulence gene patterns were observed. No correlation was observed between specific virulence gene patterns and a clinical outcome. The data indicate that virulence genes typical of extraintestinal E. coli are predominantly present in K1 strains. Nevertheless, some of them can carry virulence genes known to be characteristic of intestinal E. coli. The distribution and combination of virulence genes show that K1 isolates constitute a heterogeneous group of E. coli.

Instability of pathogenicity islands in uropathogenic Escherichia coli 536

The uropathogenic Escherichia coli strain 536 carries at least five genetic elements on its chromosome that meet all criteria characteristic of pathogenicity islands (PAIs). One main feature of these distinct DNA regions is their instability. We applied the so-called island-probing approach and individually labeled all five PAIs of E. coli 536 with the counterselectable marker sacB to evaluate the frequency of PAI-negative colonies under the influence of different environmental conditions. Furthermore, we investigated the boundaries of these PAIs. According to our experiments, PAI II536 and PAI III536 were the most unstable islands followed by PAI I536 and PAI V536, whereas PAI IV536 was stable. In addition, we found that deletion of PAI II536 and PAI III536 was induced by several environmental stimuli. Whereas excision of PAI I536, PAI II536, and PAI V536 was based on site-specific recombination between short direct repeat sequences at their boundaries, PAI III536 was deleted either by site-specific recombination or by homologous recombination between two IS100-specific sequences. In all cases, deletion is thought to lead to the formation of nonreplicative circular intermediates. Such extrachromosomal derivatives of PAI II536 and PAI III536 were detected by a specific PCR assay. Our data indicate that the genome content of uropathogenic E. coli can be modulated by deletion of PAIs.

S-layer variation in Bacillus stearothermophilus PV72 is based on DNA rearrangements between the chromosome and the naturally occurring megaplasmids

Bacillus stearothermophilus PV72 expresses different S-layer genes (sbsA and sbsB) under different growth conditions. No stretches of significant sequence identity between sbsA and sbsB were detected. In order to investigate S-layer gene regulation in B. stearothermophilus PV72, we characterized the upstream regulatory region of sbsA and sbsB by sequencing and primer extension analysis. Both genes are transcribed from unique but different promoters, independently of the growth phase. Localization of sbsB in the sbsA-expressing strain PV72/p6 revealed that the coding region of the second S-layer gene sbsB is located not on the chromosome but on a natural megaplasmid of the strain, whereas the upstream regulatory region of sbsB was exclusively detected on the chromosome of PV72/p6. For sbsB expression, the coding region has to be integrated into the chromosomally located expression site. After the switch to sbsB expression, the sbsA coding region was removed from the chromosome but could still be detected on the plasmid of the sbsB-expressing strain PV72/p2. The sbsA upstream regulatory region, however, remained on the chromosome. This is the first report of S-layer variation not caused by intrachromosomal DNA rearrangements, but where variant formation depends on recombinational events between the plasmid and the chromosome.

The site-specific integration of genetic elements may modulate thermostable protease production, a virulence factor in Dichelobacter nodosus, the causative agent of ovine footrot

The gram-negative anaerobe Dichelobacter nodosus is the causative agent of footrot in sheep. The authors have previously characterized two genetic elements, the intA (vap) and intB elements, which integrate into the genome of D. nodosus. In the virulent strain A198 there are two copies of the intA element. One copy is integrated into the 3' end of the tRNA-serGCU gene, close to the aspartokinase (askA) gene, and the second copy is integrated into the 3' end of the tRNA-serGGA gene, next to the polynucleotide phosphorylase (pnpA) gene. In this study, a new genetic element was identified in the benign strain C305, the intC element, integrated into the 3' end of the tRNA-serGCU gene, next to askA. The intC element was found in most D. nodosus strains, both benign and virulent, which were examined, and was integrated into tRNA-serGCU in most strains. Between the askA and tRNA-serGCU genes, a gene (designated glpA), was identified whose predicted protein product has very high amino acid identity with RsmA from the plant pathogen Erwinia carotovora. RsmA acts as a global repressor of pathogenicity in E. carotovora, by repressing the production of extracellular enzymes. In virulent strains of D. nodosus the intA element was found to be integrated next to pnpA, and either the intA or intC element was integrated next to glpA. By contrast, all but one of the benign strains had intB at one or both of these two positions, and the one exception had neither intA, intB nor intC at one position. The loss of the intC element from the virulent strain 1311 resulted in loss of thermostable protease activity, a virulence factor in D. nodosus. A model for virulence is proposed whereby integration of the intA and intC genetic elements modulates virulence by altering the expression of glpA, pnpA, tRNA-serGCU and tRNA-serGGA.

A Vibrio cholerae pathogenicity island associated with epidemic and pandemic strains

The bacterial species Vibrio cholerae includes harmless aquatic strains as well as strains capable of causing epidemics and global pandemics of cholera. While investigating the relationship between pathogenic and nonpathogenic strains, we identified a chromosomal pathogenicity island (PAI) that is present in epidemic and pandemic strains but absent from nonpathogenic strains. Initially, two ToxR-regulated genes (aldA and tagA) were studied and were found to be associated with epidemic and pandemic strains but absent in nontoxigenic strains. The region containing aldA and tagA comprises 13 kb of previously unidentified DNA and is part of a PAI that contains a regulator of virulence genes (ToxT) and a gene cluster encoding an essential colonization factor and the cholera toxin phage receptor (toxin-coregulated pilus; TCP). The PAI is 39.5 kb in size, has low %G+C (35%), contains putative integrase and transposase genes, is flanked by att sites, and inserts near a 10Sa RNA gene (ssrA), suggesting it may be of bacteriophage origin. We found this PAI in two clinical non-O1/non-O139 cholera toxin-positive strains, suggesting that it can be transferred within V. cholerae. The sequence within this PAI includes an ORF with homology to a gene associated with the type IV pilus gene cluster of enteropathogenic Escherichia coli, a transposase from Vibrio anguillarum, and several ORFs with no known homology. As the PAI contains the CTXPhi receptor, it may represent the initial genetic factor required for the emergence of epidemic and pandemic cholera. We propose to call this island VPI (V. cholerae pathogenicity island).

Escherichia coli CoIV plasmid pRK100: genetic organization, stability and conjugal transfer

Uropathogenic Escherichia coli strains express chromosomal and plasmid-encoded virulence-associated factors such as specific adhesins, toxins and iron-uptake systems. A CoIV plasmid (pRK100) of a uropathogenic strain and its host KS533 were studied. The host strain encodes the K1 capsule, and P and S fimbriae, but neither haemolysin nor the cytotoxic-necrotic factor CNF1, indicating that this strain does not harbour a larger pathogenicity island. A restriction map of pRK100 was constructed on the basis of hybridization experiments and nucleotide sequencing. pRK100 harbours CoIV, the conserved replication region RepFIB, the aerobactin-uptake system, a RepFIC replicon and additionally Colla as well as transposon Tn5431. The location of the RepFIC replicon was similar to that in plasmid F. CoIV plasmids and F thus share a region spanning more than half the length of plasmid F. Even though their replication and transfer regions are homologous, CoIV plasmids are found only in E. coli strains. Among the four other species tested, conjugal transfer of pRK100 was demonstrated, with low frequency, only to Klebsiella pneumoniae, suggesting that a natural barrier effectively bars transfer. In vitro stability of the plasmid with integration into the chromosome to ensure maintenance in the presence of an incompatible plasmid was demonstrated.

Use of a novel approach, termed island probing, identifies the Shigella flexneri she pathogenicity island which encodes a homolog of the immunoglobulin A protease-like family of proteins

The she gene of Shigella flexneri 2a, which also harbors the internal enterotoxin genes set1A and set1B (F. R. Noriega, GenBank accession no. U35656, 1995) encodes a homolog of the virulence-related immunoglobulin A (IgA) protease-like family of secreted proteins, Tsh, EspC, SepA, and Hap, from an avian pathogenic Escherichia coli, an enteropathogenic E. coli, S. flexneri 5, and Haemophilus influenzae, respectively. To investigate the possibility that this locus was carried on a larger deletable element, the S. flexneri 2a YSH6000T she gene was insertionally disrupted by allelic exchange using a Tn10-derived tetAR(B) cassette. Then, to detect loss of the she locus, the tetracycline-resistant derivative was plated onto fusaric acid medium to select for tetracycline-sensitive revertants, which were observed to arise at a frequency of 10(-5) to 10(-6). PCR and pulsed-field gel electrophoresis analysis confirmed loss of the she::tetAR(B) locus in six independent tetracycline-sensitive isolates. Sample sequencing over a 25-kb region flanking she identified four insertion sequence-like elements, the group II intron-like sequence Sf.IntA, and the 3' end of a second IgA protease-like homolog, sigA, lying 3.6 kb downstream and in an orientation inverted with respect to she. The deletion was mapped to chromosomal NotI fragment A and determined to have a size of 51 kb. Hybridization with flanking probes confirmed that at least 17.7 kb of the 51-kb deletable element was unique to the seven she+ strains investigated, supporting the conclusion that she lay within a large pathogenicity island. The method described in this study, termed island probing, provides a useful tool to further the study of pathogenicity islands in general. Importantly, this approach could also be of value in constructing safer live attenuated bacterial vaccines.

Large genome rearrangements discovered by the detailed analysis of 21 Pseudomonas aeruginosa clone C isolates found in environment and disease habitats

In order to determine primary genetic events which occur during the diversification of a Pseudomonas aeruginosa clone in natural habitats, comparative genome analysis of 21 isolates of a predominant clone, called clone C, derived mainly from patients with cystic fibrosis (CF) and the aquatic environment, was carried out. Physical chromosome maps were constructed for the restriction enzymes SpeI, PacI, SwaI and I-CeuI by one and two-dimensional pulsed-field gel electrophoresis and by comparison with the existing strain C map. The positioning of 26 genes generated the genetic maps. Chromosome size varied between 6345 and 6606 kilobase-pairs (kb). A plasmid of 95 kb was detected in the strains of non-CF origin and, in addition, was found to be integrated into the chromosome of all strains but one CF isolate. Four subgroups of clone C strains were discriminated by the acquisition and loss of large blocks of DNA that could cover more than 10% of the chromosome size. The exchange of DNA blocks which ranged in size from 1 kb to 214 kb occurred preferentially around the terminus of replication region which is poor in biosynthetic genes. Genetic material which was additionally introduced into strain C in comparison with strain PAO seems to be a target of mutational processes in clone C strains. Within and among subgroups CF isolates frequently exhibited large inversions affecting the whole chromosomal structure. We concluded that the exchange of DNA blocks by mechanisms of horizontal transfer and large chromosomal inversions are major factors leading to the divergence of a clone in the species P. aeruginosa.

Characterization of Escherichia coli strains isolated from environmental water habitats and from stool samples of healthy volunteers

This study was undertaken to determine the frequency of pathogenic Escherichia coli strains among wild-type E. coli strain isolates from the microbial flora of healthy volunteers and from natural residential water habitats of a defined geographic area. In total, 131 stool and 95 water isolates as well as 14 E.coli K12 strains were examined for DNA sequences specific for 20 different genes encoding E. coli pathogenicity factors, including adherence factors, toxins, invasins, capsules and iron uptake systems. The expression of the corresponding pathogenicity factors was also investigated. No pathogenicity factors were found to be present in the tested E. coli K12 strains. In contrast, 41.0% of the water samples and 63.4% of the stool samples contained pathogenicity factors specific for extraintestinal E. coli pathogens. While no virulence determinants specific for intestinal E. coli pathogens were found among the investigated environmental water isolates, 4.5% of the stool samples contained either only intestinal or both intestinal and extraintestinal virulence genes. Both the prevalence of the virulence genes and the expression of the corresponding pathogenicity factors were, in general, higher in stool than in water samples. These findings might indicate the prevalence of different clonal types and/or differential regulation of pathogenicity factor expression in diverse ecological niches.

Heterogeneity of genome sizes among natural isolates of Escherichia coli

Comparisons of the genetic maps of Escherichia coli K-12 and Salmonella typhimurium LT2 suggest that the size and organization of bacterial chromosomes are highly conserved. Employing pulsed-field gel electrophoresis, we have estimated the extent of variation in genome size among 14 natural isolates of E. coli. The BlnI and NotI restriction fragment patterns were highly variable among isolates, and genome sizes ranged from 4,660 to 5,300 kb, which is several hundred kilobases larger than the variation detected between enteric species. Genome size differences increase with the evolutionary genetic distance between lineages of E. coli, and there are differences in genome size among the major subgroups of E. coli. In general, the genomes of natural isolates are larger than those of laboratory strains, largely because of the fact that laboratory strains were derived from the subgroup of E. coli with the smallest genomes.

The evolution of invasion by enteric bacteria

Despite differences in disease pathologies and host range, many enteric pathogens, including Salmonella and Shigella spp., utilize a remarkably similar machinery to secrete proteins that promote their entry into host cells. Analogous structures are required for the export of virulence proteins in other animal and plant pathogens. While the structure and organization of the gene complexes specifying these secretory pathways are broadly conserved, their phylogenetic distribution and genomic locations suggest that these sequences arose independently in divergent pathogens.

Molecular epidemiology of Candida isolates from AIDS patients showing different fluconazole resistance profiles

Thirty Candida isolates obtained from the oropharynxes of three AIDS patients were genotypically characterized. In vitro fluconazole MIC determination revealed increasing fluconazole resistances during treatment, thereby confirming the in vivo situation. Pulsed-field gel electrophoresis karyotyping, randomly amplified polymorphic DNA analysis, and hybridizations with Candida albicans repetitive element 2 were used to determine possible genotypic changes. The isolates from two patients showed genetic homogeneity, suggesting the selection for resistant variants. One patient experienced a strain switch to Candida krusei. Horizontal spread of identical strains between the patients could be excluded. However, the molecular methods used might not be sufficient to detect the underlying genetic basis of resistance to fluconazole.

Adhesin regulatory genes within large, unstable DNA regions of pathogenic Escherichia coli: cross-talk between different adhesin gene clusters

The uropathogenic Escherichia coli strain 536 possesses two large, unstable DNA regions on its chromosome, which were termed pathogenicity islands (pais). Deletions of pais, which occur with relatively high frequency in vitro and in vivo, lead to avirulent mutants. The genetic determinants for production of haemolysin (Hly) and P-related fimbriae (Prf) are located in one of these islands. Deletion of this pathogenicity island (paill) not only removes the hly- and prf-specific genes, but also represses S fimbriae (Sfa), although the sfa genes of this virulence factor are not located on paill. We have identified two regulatory genes, prfB and prfl, of the prf gene cluster that are homologous to the sfa regulatory genes sfaB and sfaC, respectively. Mutations in sfaB and sfaC that inhibit transcription of the major fimbrial subunit gene sfaA were complemented by the homologous prf genes, suggesting communication between the two fimbrial gene clusters in the wild-type strain. Chromosomal mutagenesis of the two prf regulators in strain 536 repressed transcription of sfaA, detected by Northern hybridization and a chromosomal sfaA-lacZ fusion. In addition, haemagglutination assays measured a lower level of S fimbriae in these mutants. Expression of the cloned prf regulators in trans reversed the effect of the mutations; furthermore, constitutive expression of prfB or prfl could also over-come the repression of S fimbriae in a strain that had lost the pathogenicity islands. Virulence assays in mice established that the prf mutants were less virulent than the wild-type strain. The results demonstrate that cross-regulation of two unlinked virulence gene clusters together with the co-ordinate loss of large DNA regions significantly influences the virulence of an extraintestinal E. coli wild-type isolate.

Excision of large DNA regions termed pathogenicity islands from tRNA-specific loci in the chromosome of an Escherichia coli wild-type pathogen

Uropathogenic Escherichia coli 536 (O6:K15:H31) carries two unstable DNA regions, which were shown to be responsible for virulence. These regions, on which the genes for hemolysin production (hly) and P-related fimbriae (prf) are located, are termed pathogenicity islands (PAI) I and II, and were mapped to positions 82 and 97, respectively, on the E. coli K-12 linkage map. Sequence analysis of the PAI region junction sites revealed sequences of the leuX and selC loci specific for leucine and selenocysteine tRNAs. The tRNA loci function as the targets for excision events. Northern (RNA) blot analysis revealed that the sites of excision are transcriptionally active in the wild-type strain and that no tRNA-specific transcripts were found in the deletion mutant. The analysis of deletion mutants revealed that the excision of these regions is specific and involves direct repeats of 16 and 18 nucleotides, respectively, on both sides of the deletions. By using DNA long-range mapping techniques, the size of PAI I, located at position 82, was calculated to be 70 kb, while PAI II, mapped at position 97, comprises 190 kb. The excision events described here reflect the dynamics of the E. coli chromosome.