![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
Previous Article | Next Article 
Clinical Microbiology Reviews, January 2004, p. 14-56, Vol. 17, No. 1
0893-8512/04/$08.00+0 DOI: 10.1128/CMR.17.1.14-56.2004
Copyright © 2004, American
Society for
Microbiology. All Rights Reserved.
Institut für Medizinische Mikrobiologie und Hygiene, Medizinische Fakultät Carl Gustav Carus, Technische Universität Dresden, Dresden,1 Institut für Klinische Mikrobiologie, Immunologie und Hygiene, Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany2
SUMMARY INTRODUCTION STRUCTURE OF PAI VIRULENCE FACTORS ENCODED BY PAI PROTEIN SECRETION SYSTEMS ENCODED BY PAI Type I Systems Type II Systems Type III Systems Type IV Systems Type V Systems REGULATION OF PAI-ENCODED VIRULENCE FUNCTIONS EVOLUTION AND TRANSFER OF PAI Natural Transformation PAI and Plasmids Transduction INTEGRATION SITES OF PAI PAI OF GRAM-NEGATIVE PATHOGENS Helicobacter pylori cag PAI. Pseudomonas aeruginosa PAGI-1. PAGI-2 and PAGI-3. Glycosylation island. Shigella spp. SHI-O. SHI-1. SHI-2. SRL. SHI-3. Other Shigella islands. Yersinia spp. HPI. Chromosomally encoded type III secretion systems. Vibrio cholerae VPI. VPI-2. VSP-I and VSP-II. Salmonella spp. SPI-1. SPI-2. SPI-3. SPI-4. SPI-5. Major PAI. SGI-1. Other SPI. Enteropathogenic E. coli LEE of EPEC. EspC PAI. LEE of rabbit-pathogenic E. coli. Enterohemorrhagic E. coli LEE of EHEC. Other genetic islands. TAI. Other Intestinal E. coli Groups HPI of pathogenic E. coli. LPA of Stx2d-producing E. coli. Enterotoxigenic E. coli Enteroaggregative E. coli Enteroinvasive E. coli Uropathogenic E. coli PAI I536 to PAI IV536. PAI IJ96 and PAI IIJ96. PAI ICFT073 and PAI IICFT073. Other Extraintestinal E. coli Strains Pathogenic Neisseria spp. Other Gram-Negative Pathogens LEE of Citrobacter rodentium. PAI of Bacteriodes fragilis. rag locus of Porphyromonas gingivalis. Hrp island of Pseudomonas syringae. vap loci of Dichelobacter nodosus. PAI OF GRAM-POSITIVE PATHOGENS Listeria monocytogenes LIPI-1. LIPI-2. Internalin islets. Other loci. Staphylococcus aureus Staphylococcus cassette chromosome mec. PAI encoding toxic shock syndrome toxins. {nu}Sa families of PAI. (i) {nu}Sa1 to {nu}Sa4. (ii) {nu}Sa{alpha}. (iii) {nu}Saß. etd PAI. Streptococcus spp. Enterococcus faecalis PaLoc of Clostridium difficile CONCLUDING REMARKS Approaches to the Identification of New PAI Evolution of PAI Bacterial Pathogens without PAI? OUTLOOK ACKNOWLEDGMENTS REFERENCES
| SUMMARY |
|---|
 Top NextReferences |
|---|
| INTRODUCTION |
|---|
|
TopPreviousNextReferences |
|---|
The process by which the content and organization of genetic information of a species changes over time is known as genome evolution. This process includes four forms of changes: point mutations and gene conversions, rearrangements (e.g., inversion or translocation), deletions, and insertions of foreign DNA (e.g., plasmid integration, transposition). Gene loss and acquisition are genomic changes that can rapidly and radically alter the life-style of a bacterium in "quantum leaps" (122). These latter mechanisms seem to be the primary forces by which bacteria genetically adapt to novel environments and by which bacterial populations diverge and form separate, evolutionary distinct species. Acquisition of foreign genes is obviously coupled with gene loss because genome growth is not unlimited. The balance between selective gene acquisition and secondarily imposed gene loss implies that addition of a foreign gene increases the probability of loss of some resident function of lower selective value (207, 260). Mechanisms of horizontal gene flux include mobile genetic elements such as conjugative plasmids, bacteriophages, transposons, insertion elements, and genomic islands, as well as the mechanism of recombination of foreign DNA into host DNA (128, 236).
In this review, we focus on a group of mobile genetic elements whose discovery has influenced and revised our thinking about genomic stability and the species concept in prokaryotes. These elements play a pivotal role in virulence of bacterial pathogens and are also essential for virulence in pathogens of animals and plants. We review a subgroup of genomic islands, the pathogenicity islands (PAI). Since excellent reviews and original papers have already been published on the molecular structure and evolution of these genetic elements (27, 68, 127-129), this review emphasizes the contribution of PAI to the development of disease and to the virulence of bacterial pathogens carrying them.
The concept of PAI was founded in the late 1980s by Jörg Hacker and colleagues in Werner Goebel's group at the University of Würzburg, Würzburg, Germany, who were investigating the genetic basis of virulence of UPEC strains 536 and J96 (126, 186). The group observed a genetic linkage of determinants encoding P fimbriae, P-related fimbriae, and hemolysins in these strains and could also detect a codeletion of these linked genes (126). Similar DNA segments with more than one linked virulence gene were described earlier and were termed virulence gene blocks in concordance with the names given by other authors (125, 151, 151, 215). However, the observation that a single deletion event results in the loss of two linked virulence gene clusters together with additional DNA segments more than 30 kb apart led to the definition of the epithet "pathogenicity DNA islands" and later on to "pathogenicity islands" (PAI) (26, 126). Hacker and colleagues showed that deletion of a PAI led to a nonpathogenic phenotype of E. coli strain 536, and it has been suggested that such deletions are a genetic mechanism to modulate bacterial virulence. In a later study, the size and genetic structure of these PAI found in E. coli strain 536 were investigated in detail (126, 131, 187). The regions carrying genes for hemolysin production (hly) and P-related fimbriae, termed PAI I and II, were mapped at centisomes 82 and 97 in the E. coli chromosome. It was also shown that the tRNA loci selC and leuX are located at the junction to the chromosome and that direct repeats of 16 and 18 nucleotides flank these PAI. After this initial discovery of PAI, these genetic structures have been found increasingly in other E. coli groups and also in other bacterial species (129).
| STRUCTURE OF PAI |
|---|
|
TopPreviousNextReferences |
|---|
|
|
(ii) PAI are present in the genomes of a pathogenic bacterium but absent from the genomes of a nonpathogenic representative of the same species or a closely related species.
(iii) PAI occupy relatively large genomic regions. The majority of PAI are in the range of 10 to 200 kb.
(iv) PAI often differ from the core genome in their base composition and also show a different codon usage. The base composition is expressed as percentage of guanine and cytosine (G+C) bases, and the average G+C content of bacterial DNA can range from 25 to 75%. Most pathogenic bacterial species have G+C contents between 40 and 60%. The reasons for that variation are not known, but the conservation of a genus- or species-specific base composition is a remarkable feature of bacteria. It is considered that the horizontally acquired PAI still has the base composition of the donor species. On the other hand, it is also observed that the base composition of horizontally acquired DNA will gravitate to the base composition of the recipient's genome during evolution. Thus it is difficult to explain why "ancient" PAI still show a different base composition. Further factors such as DNA topology or specific codon usage of the virulence genes in PAI may also account for the maintenance of the divergent base composition.
(v) PAI are frequently located adjacent to tRNA genes. This observation gave rise to the hypothesis that tRNA genes serve as anchor points for insertion of foreign DNA that has been acquired by horizontal gene transfer. The frequent insertion at tRNA loci may be explained by the observation that genes encoding tRNAs are highly conserved between various bacterial species. After acquisition by horizontal gene transfer, a DNA fragment that contains a tRNA gene can insert into the recipients genome by recombination between the tRNA genes. The second observation is that certain bacteriophages use tRNA genes as specific insertion points in the host genome. tRNA genes may represent specific anchor points for the integration of foreign DNA.
(vi) PAI are frequently associated with mobile genetic elements. They are often flanked by direct repeats (DR). DR are defined as DNA sequences of 16 to 20 bp (up to 130 bp) with a perfect or nearly perfect sequence repetition. DR might have served as recognition sites for the integration of bacteriophages, and their integration resulted in the duplication of the DR. Furthermore, DR act as recognition sequences for enzymes involved in excision of mobile genetic elements, thus contributing to the instability of a PAI flanked by DR. Deletion of a PAI is probably promoted by the same mechanisms that contribute to the loss of antibiotic resistance factors in the absence of selective pressure. In both situations, the deletion results in a reduction in genome size leading to a reduced generation time that is of advantage in competition with other microbes. PAI often carry cryptic or even functional mobility genes such as integrases or transposases. Integrases, which may have been derived from lysogenic bacteriophages, mediate the integration of the phage genome into the genome of the host bacteria, as well as the excision needed to enter a lytic cycle. Such genes are still functional in certain PAI, and the encoded proteins can mediate the excision of the PAI and its loss. The role of bacteriophages in transfer of PAI is described later in this review. Other PAI contain genes that are similar to integrase and resolvase genes of transposons. These mobile genetic elements can change their location within the chromosome, but transposons can also jump from a chromosomal location into a plasmid and vice versa. Insertion sequence (IS) elements are frequently observed in PAI. Insertion of IS elements can result in the inactivation of genes, but the combination of two or more IS elements can also result in the mobilization of larger portions of DNA. PAI can also represent integrated plasmids, conjugative transposons, bacteriophages or parts of these elements (127).
(vii) PAI often are unstable and delete with distinct frequencies. Virulence functions encoded by certain PAI are lost with a frequency that is higher than the normal rate of mutation. Genetic analyses showed that such mutations are caused not by defects in individual virulence genes within the PAI but, rather, by loss of the large portions of a PAI or even the entire PAI. These mutations can be observed during cultivation of pathogens in vitro, but they are also found in isolates obtained from infected individuals, for example during persistent infections. This indicates that such PAI have an intrinsic genetic instability. The same genetic mechanisms allowing the distribution of PAI by horizontal gene transfer also determine their genetic instability. Several characteristic elements, such as integrases, transposases, and IS elements, have been identified that contribute to mobilization and as well as to instability as described above.
(viii) PAI often represent mosaic-like structures rather than homogeneous segments of horizontally acquired DNA. Some PAI represent an insertion of a single genetic element. Others show a more complex structure, since elements of different origin are present. During evolution, several genetic elements have been acquired independently at different time points and from different hosts. However, these DNA acquisitions integrated at the same position into the chromosome of the recipient bacterial cell. This will result in the accumulation of horizontally acquired elements at a certain location of the chromosome, and the same target structures (e.g. tRNA genes) served repeatedly for the integration of the various elements.
These properties apply to numerous PAI, but with the acquisition an increasing amount of genomic sequence information, it became clear that the genomes of prokaryotes are highly diverse mosaic structures. Besides a core genome, which mostly demonstrates homogeneous G+C content and codon usage, there exists a flexible gene pool that is formed by mobile genetic elements. Although the majority of the genes of the flexible gene pool confer selective advantages to their bacterial recipients, a few represent selfish DNA only promoting their own spread. The latter elements are insertion elements, particularly prophages and restriction/modification systems. PAI are part of that flexible gene pool. Sequencing of entire bacterial genomes revealed a more ubiquitous occurrence of such islands than was previously thought, and this represents a paradigm of more general genetic entities that are present in the genome of many bacteria. Therefore, the designation "pathogenicity islands" has been extended to "genomic islands," which can encode a wide range of functions. Most genomic islands carry genes useful for the survival and transmission of microbes. In a recent review by Hacker and Carniel (128), the authors propose a model for the development of specialized genomic islands. In this model, a bacterial cell first acquires blocks of genes. Selection processes than may favor the maintenance and development of genomic islands that increase fitness. "Fitness islands" may then specialize as ecological, saprophytic, or symbiosis islands or PAI. This model is based on Darwinian laws and explains the actual situation most satisfactorily.
Genomic islands encode many different functions, which depend largely on the environmental context in which the bacterium grows. PAI, which are the best understood genomic islands known to date, carry clusters of virulence genes whose products contribute to the pathogenicity of the bacterium. In the case of E. coli, such islands have allowed the bacteria to adapt to specific environments and to cause disease (Fig. 2). The division of fitness islands into the different subtypes is based not only on their genetic composition but also on their effects in a specific ecological niche and within a particular organism. This means that the same island may fulfill different functions. For example, the yersiniabactin iron uptake system of Yersinia spp. and several pathogenic enterobacteria is also present in soil bacteria. In the latter group, the yersiniabactin iron uptake system appears as a "fitness islands" allowing life under iron-limiting conditions. If this system is present in a bacterium that colonizes a host organism, together with other factors this locus will become a PAI (128).
|
|
| VIRULENCE FACTORS ENCODED BY PAI |
|---|
|
TopPreviousNextReferences |
|---|
|
| PROTEIN SECRETION SYSTEMS ENCODED BY PAI |
|---|
|
TopPreviousNextReferences |
|---|
-hemolysin (see also
Uropathogenic E. coli
below). Although termed "secretion systems," the main function of T3SS is not the secretion of proteins into the medium but rather, the translocation across a third membrane, i.e., the membrane of a eukaryotic host cell. Translocation of such effector proteins into eukaryotic host cells is the basis for specific interference with eukaryotic cells functions, resulting in host cell invasion, inactivation of phagocytic cells, apoptosis, and interference with intracellular transport processes. This form of protein translocation requires contact between the pathogen and the target cell. T3SS-dependent translocation can be observed in extracellular pathogens via the cytoplasmic membrane as well as by intracellular pathogens via the phagosomal membrane. Gene clusters encoding T3SS can be found on virulence plasmids, for example in Yersinia and Shigella spp., as well as in PAI. PAI encoding T3SS include SPI-1 and SPI-2 of Salmonella enterica and the locus of enterocyte effacement (LEE) in enteropathogenic E. coli. Further reference of the genetic and biochemical features of T3SS can be found in recent reviews (59, 162).
| REGULATION OF PAI-ENCODED VIRULENCE FUNCTIONS |
|---|
|
TopPreviousNextReferences |
|---|
Most frequently, regulators belong to the AraC/XylS family or to the two-component response regulator family. Alternative sigma factors and histone-like proteins are also involved in PAI regulation. Regulatory cascades, in which PAI-encoded regulators of PAI-located virulence genes are modulated by systems encoded outside the PAI, include VPI of Vibrio cholerae, SPI-1 and SPI-2 of S. enterica, the Yop virulon of pathogenic Yersinia spp., and the LEE of enteropathogenic E. coli (EPEC) and EHEC. The regulation of SPI-1, SPI-2, and LEE of E. coli have been investigated in a number of studies. Although many details are not yet known, a good overview is available (Fig. 3).
|
SPI-1 encodes a number of transcriptional regulators: HilA plays a central role in controlling SPI-1 gene expression, HilD and HilC interact with a DNA sequence upstream of the HilA promoter, presumably displacing a repressor from this site, and InvF controls expression of the genes encoding the substrate proteins of SPI-1. In a current model, SPI-1 regulation involves a cascade of transcriptional activation in which HilD and HilC, HilA, and InvF (Fig. 3A, dark grey bars) act sequentially to activate T3SS genes. First, HilD and HilC bind to several sites within PhilA and derepress hilA transcription. Then HilA binds to invF and prgH transcription start sites and activates expression. This results in expression of the genes encoding T3SS components. InvF is also required for the expression of sptP, so it is possible that sicP sptP may be cotranscribed with the sip genes. In addition to expression of SPI-1 genes, loci outside of SPI-1 encoding T3SS effectors SigD (SopB), SopD, SopE, SopE2, and presumably yet unidentified factors are also expressed from InvF/SicA-dependent promoters (for reviews, see references 214 and 216). The regulation of the LEE is described in detail below in the discussion of PAI of EPEC.
| EVOLUTION AND TRANSFER OF PAI |
|---|
|
TopPreviousNextReferences |
|---|
Many PAI are too large to be transferred as passengers in bacteriophage genomes. For example, gene clusters on PAI encoding T3SS or T4SS comprise 25 to 40 kb DNA, which is almost equivalent to the total genome size of a bacteriophage. In these cases, other mechanisms are conceivable. Certain bacteriophages are capable of generalized transduction. Normally, for the replication of the phage within the host bacterium, copies of the phage genome are packaged into phage heads. During replication, the host DNA is fragmented. Occasionally, the enzymes involved in packaging the phage genome erroneously pack a fragment of the host genome into the phage head. Since the resulting particles are still able to infect a new bacterial host, a fragment of the bacterial DNA can be transduced. Given sufficient sequence similarity, recombination may occur and the transduced fragment is integrated into the genome of the new host.
PAI do not occur only in human pathogens; they have also been found in animal and plant pathogens. Examples are the hrp islands of Pseudomonas syringae and Xanthomonas campestris, and islands in animal pathogenic salmonellae and staphylococci. They are distributed throughout the bacterial world, and horizontal transfer may be facilitated by plasmids and phages or by bacteria, which are competent for the uptake of free DNA by natural transformation.
| INTEGRATION SITES OF PAI |
|---|
|
TopPreviousNextReferences |
|---|
C
stem-loop region of a tRNA up to the conserved CCA end
(156). The
molecular basis of the use of tRNA genes as integration sites is not
fully understood, but three hypotheses are plausible.
|
A second hypothesis would include the presence of multiple copies of tRNA genes, providing multiple insertion sites and amplification of pathogenicity factors. This is, however, not true for selC and leuX, which occur in single copies.
The third, and most plausible, hypothesis suggests that the conserved structure in tRNA genes provides structural motifs that facilitate the integration and excision of PAI and also phages (290). This emphasizes that integration and excision are catalyzed by integrases.
Hou (156) proposed a fourth hypothesis, in which the 3' end of tRNA plays a major role. In his hybrid theory, the conserved CCA ends provide the initial site for integration by an integrase. The 3' end of a tRNA hybridizes to one strand of a duplex DNA during recombination. This stabilizes the separation of the DNA duplex for recombination (for details, see reference 156). Whether this theory or one of the others is correct has yet to be elucidated. Nevertheless, it is apparent that phages and PAI use conserved genes as integration sites. These conserved genes might confer safety to the mobile genetic element that they can integrate in any genome of members of a given population. This need could be, in an evolutionary biology point of view, important to maintain pathogenicity factors in a bacterial population (329).
| PAI OF GRAM-NEGATIVE PATHOGENS |
|---|
|
TopPreviousNextReferences |
|---|
H. pylori organisms are curved, rod-shaped bacteria with a group of polar flagella and are covered by a membrane sheath. Motility is an important virulence factor and enables the bacteria to penetrate the mucin layer of the gastric epithelium (171). The bacteria also produce urease. This enzyme catalyzes the formation of CO2 and ammonia that can neutralize the acidic pH in the vicinity of the bacteria. Cultivation of H. pylori requires a microaerophilic atmosphere and complex media.
Clinical isolates of H. pylori have been classified into type I and type II strains, which are associated with different clinical outcomes ranging from gastric ulcer to asymptomatic colonization. There are also various forms of intermediate virulence. Type I strains carry genes encoding both, the cytotoxins CagA and VacA, while type II strains contain vacA genes only (376). VacA is a secreted toxin that induces extensive vacuolation in epithelial cells, cell death, and destruction of epithelial integrity.
The attachment of type I strains to gastric epithelial cells induces the synthesis and secretion of several chemokines, and the secretion of interleukin-8 (IL-8) is frequently assayed in model systems. It has also been observed that the infection of epithelial cells by H. pylori leads to dramatic rearrangements of the host cell actin cytoskeleton and the formation of pedestals (318) that are reminiscent of EPEC-induced pedestals, as well as to changes in the gross morphology of host cells (hummingbird phenotype). These phenotypes are associated with alterations in the signal transduction pathways of the host cell and the presence of a tyrosine-phosphorylated protein (317).
cag PAI. Detailed analysis of the cagA loci in type I and type II strains indicated that the latter group showed deletions of a large chromosomal region. This locus had the typical characteristics of a PAI and was termed the cag PAI (49). Censini et al. (49) characterized this locus and showed that the cag PAI had a size of 37 to 40 kb, flanked by direct repeats of 31 bp (Fig. 5A). The locus has a G+C content of 35%, in contrast to the 39% observed for the core genome (349). A gene for a tRNA has not been identified at the point of integration, but the glr gene (glutamate racemase) was disrupted by insertion of the PAI. There are no genes associated with DNA mobility within the cag PAI of type I strains. However, the presence of an IS 605 element within the cag PAI of strains with an intermediate virulence phenotype was observed. In strains of intermediate virulence, various forms of deletions with the cag PAI were detected, and in certain strains the locus was separated into two portions, referred to as cagI and cagII (1, 49). These observations support a correlation between the presence and integrity of the cag PAI and the severity of disease. Studies with a mouse model have shown that an association between cag PAI-negative H. pylori strains and cag PAI-positive strains that are mouse adapted and have modulated their ability to activate a proinflammatory response can better colonize mice than the parental strains do, indicating that the cag PAI of type I strains may become lost during colonization of infected animals (272). In addition to large deletions and chromosomal rearrangements of the cag PAI, there are indications that point mutations in the PAI genes result in ablation of CagA translocation and IL-8 induction (92). This effect can be explained by loss of function of the T4SS.
|
Sequence analysis of the cag PAI indicated that a T4SS is encoded by this locus and that CagA is a translocated substrate of the secretion apparatus. Only 6 of the 27 to 29 predicted open reading frames (ORF) in the cag PAI show significant sequence similarity to components of the T4SS of other bacteria, and the contribution of other genes to formation of a translocation apparatus is not clear (Fig. 5A).
Systematic mutagenesis approaches to analysis of the functions of the genes in the cag PAI were performed by Fischer et al. (92) and Selbach et al. (320). The individual inactivation of 27 putative genes and phenotypic analysis identified a subset of 17 genes that are required for the translocation of CagA into host cells and a subset of 14 genes that are required for the stimulation of IL-8 synthesis in host cells. Although the assembly of T4SS is not understood in full detail, these observations indicate that the majority of genes within the cag PAI are required for the formation of a functional T4SS, by encoding either structural components or protein important for the assembly and regulation of the system. Neither approach resulted in the identification of mutant strains that were deficient in CagA translocation but capable of inducing IL-8 secretion. These observations indicate the absence of a further translocated protein responsible for IL-8 induction within the cag PAI or a direct effect of the T4SS in IL-8 induction. The secretion of VacA is not dependent on the cag PAI, and so far no further proteins translocated by cag PAI-encoded T4SS have been identified.
The observation that the cag PAI is absent or partially deleted in H. pylori strains with low virulence might suggest that the function of the cag PAI-encoded T4SS is not compatible with a long-lasting colonization of the gastric epithelium. The inflammatory response elicited by H. pylori after contact-dependent translocation could lead either to a clearance of the infection or to a severe immunopathology. However, epidemiological data indicate that the frequency of cag+ isolates of H. pylori is much higher in the Asian population than in the Western population, indicating that further host and pathogen factors are involved in colonization (21).
The genome of H. pylori is characterized by a high flexibility, and an extremely high frequency of recombination was observed (89, 341). The DR flanking the cag PAI probably function as sites for recombination and deletion of the locus.
The 6,264,403-bp chromosome of P. aeruginosa strain PAO1 has been sequenced completely (340). Besides a large gene repertoire involved in the catabolism, transport, and efflux of organic compounds, which basically is responsible for its metabolic versatility, P. aeruginosa possesses 10 islands of 3.0 kb and larger, which have a significantly lower G+C content than the rest of the chromosome (66.6% for the rest of the chromosome) and show an unusual codon usage (203).
Some islands carry apparently dispensable genes that are not present in all P. aeruginosa strains, such as genes encoding toxins, pyocins, and proteins with unknown functions. Other islands encode cellular appendages and elements of the outer membrane such as lipopolysaccharide (LPS) biosynthesis enzymes. These islands are referred to as PAI-like structures, since analysis to determine the role of these elements have not been performed so far.
P. aeruginosa displays interclonal heterogeneity. Comparison of the genomes of P. aeruginosa strain PAO1 and P. aeruginosa strain C has shown that the latter possesses 700 kb of additional DNA and carries 11 regions of 5 kb and larger (20 to 160 kb) which are not present on PAO1. These regions have a lower G+C content and may be considered to be PAI.
In P. aeruginosa strain PAO1, a T3SS is encoded by the 25,670-bp exoenzyme S regulon, with subunits displaying a high level of sequence similarity to the components of the Yersinia Yop virulon. The T3SS permits contact-mediated translocation of the antihost factors ExoS, ExoT, ExoU, and PcrV. ExoS and ExoT are related proteins that have 75% amino acid identity (101). The ExoS protein uncouples the Ras-mediated signal transduction pathway. While the C terminus of ExoS possesses ADP-ribosyltransferase activity, the N-terminal domain is responsible for the disruption of actin and is therefore cytotoxic (101, 269). ExoT also possesses ADP-ribosyltransferase activity, but only 0.2% of the amount in ExoS. Recently, it has been shown that ExoT functions as anti-internalization factor which prevents uptake by multiple cell lines and is able to modify the host cytoskeleton (112). Expression of another antihost factor, ExoU, correlates with acute cytotoxicity and lung injury. ExoU was also demonstrated to cause necrosis of macrophages. Moreover, the T3SS transports at least another macrophage-killing activity, which is independent of ExoU and causes apoptosis (139). PcrV is only poorly characterized. It is thought to be involved in modulation of the host cytokine response. Apparently, the T3SS also causes ExoU-independent oncosis of macrophages and polymorphonuclear leukocytes, cellular and nuclear swelling, disintegration of the plasma membrane, and absence of DNA fragmentation. The ExoS regulon may be considered an ancient PAI that became irreversibly fixed in the genome and has lost all elements of mobility.
PAGI-1. Differential hybridization of a P. aeruginosa isolate from a patient with urinary tract infection and strain PAO1 resulted in the discovery of P. aeruginosa genomic island 1 (PAGI-1). It consists of 48,893 bp of DNA and contains 51 ORF. The G+C content of PAGI-1 has an asymmetric distribution. In the first three-quarters of the sequence, from ORF 1 to ORF 30, the G+C content is 63.7%, and the remaining portion of the PAI, from ORF 31 to ORF 51, has a G+C content of 54.7%, which is lower than that of the P. aeruginosa core chromosome. Approximately 50% of ORF located on PAGI-1 encode hypothetical proteins with unknown functions. Among the genes that could be assigned to putative functions, the most notable are remnants of transposable elements, putative transcriptional regulators, and a number of genes encoding various dehydrogenases. At the left end of PAGI-1 are found two ORF encoding proteins homologous to paraquat-inducible proteins of E. coli. In E. coli, a pair of paraquat-inducible genes, pqiA and pqiB, are under the control of the SoxS and SoxR regulators, which respond to redox-cycling agents capable of generating superoxide radicals in the cell. The precise role of PqiA and PqiB in detoxification of radicals is not known. The presence of these genes in P. aeruginosa and the large number of dehydrogenases with putative functions in the detoxification indicates a role for this island in the detoxification of reactive oxygen species, i.e., in protection against oxidative damage (209). Biological studies to confirm this suggestion are under way.
PAGI-2 and PAGI-3. Further analyses of P. aeruginosa strains C (isolate from the airways of a patient with cystic fibrosis), SG17M (an isolate from the aquatic environment), and PAO1 (an isolate from a patient with cystic fibrosis) revealed two strain-specific islands that are integrated into tRNAGly genes, which are located within a cluster comprising one tRNAGlu gene followed by two identical tRNAGly genes. The first island is designated PAGI-2(C), is present in P. aeruginosa C, and consists of 104,996 bp of DNA; the second is designated PAGI-3(SG), is present in SG17M, and consists of 103,304 bp of DNA. Analysis of PAGI-2(C) and PAGI-3(SG) sequences revealed 111 and 106 ORF, respectively (204). In both strains, the gene islands are partitioned into two larger blocks. The cluster adjacent to the attL site harbors strain-specific genes. The other gene block predominantly contains hypothetical genes with various homologues, e.g., to genes in Xylella fastidiosa islands. Some of the strain-specific genes of PAGI-2(C) putatively encode proteins for complexation and transport of heavy metal ions, essential proteins for cytochrome c biogenesis and related thiol disulfide exchange proteins. Moreover, proteins for cation transport, a two-component regulatory system, transcriptional regulators, a transport system conferring mercury resistance, and other transporters are present. The role of these proteins is unclear. The authors hypothesized that cytochrome biogenesis may facilitate iron uptake and inactivation of peroxides and thus may confer advantage to the bacteria in persistence in the lungs of cystic fibrosis patients, where they suffer from iron limitation and oxidative stress (204). Other genes, such as those responsible for copper homeostasis and mercury resistance, may be important not for pathogenesis but for survival in an environment with high heavy-metal concentrations. PAGI-3(SG) of the environmental isolate SG17M is thought to be a metabolic island containing genes related to metabolism and transport of amino acids, coenzymes, porphyrins, and a number of putative enzymes (204).
Besides PAGI-1, PAGI-2(C), and PAGI-3(SG), large (100-kb) islands that are derived from plasmids and reversibly recombine with either of the two tRNALys genes have been found in isolates of P. aeruginosa (184).
Glycosylation island.
Recently, a glycosylation island
was discovered in P. aeruginosa strain PAK
(7). This island is
obviously present in all strains that express a-type flagellin and
mediates posttranslational glycosylation of this type of flagellin by
covalent attachment of glycosyl moieties to one or several sites within
the polypeptide. This island is 
16 kb in size and contains 14
ORF; most of these may be involved in glycosylation and various
biosynthetic pathways, and some are of unknown function. The precise
role of glycosylation in flagellar function is not known. Glycosylation
is not necessary for motility under laboratory conditions, but the
strong association of this island with certain pathogenic strains of
P. aeruginosa suggests that there are some functions for
glycosylation in
vivo.
Shigella spp. are able to invade epithelial cells via M-cells to gain entry into the colonic epithelium. After transcytosis, the bacteria enter lymphoid follicles containing resident tissue macrophages. The cells are engulfed by macrophages but not destroyed. After liberation from the phagosome into the host cell cytoplasm, they cause caspase-1-mediated apoptosis and release of IL-1ß and IL-8. The inflammatory response to these cytokines damages the colonic mucosa and exacerbates the infection. After apoptotic release of bacteria from infected macrophages, Shigella gains access to adjacent enterocytes via the basolateral side. This pathogen cannot enter polarized epithelial cells from the apical side.
The genes for invasiveness are located on the large invasion plasmid (225, 299). This plasmid contains a 30-kb DNA region encoding a T3SS including Mxi (for "membrane excretion of Ipa") and Spa (for "surface presentation of invasion plasmid antigen") proteins, and a second operon encoding effector proteins such as Ipa (for "invasion plasmid antigens"). In addition to this plasmid, chromosomal genes are needed for the full array of virulence phenotypes caused by Shigella spp. So far, five PAI have been identified in Shigella spp. Although these PAI carry virulence genes, their contribution to pathogenesis is not yet fully understood (164, 217, 279, 282, 361).
SHI-O. LPS is an important virulence factor of Shigella spp. (210). Since the immune response to Shigella spp. is O-antigen specific, an immune response to a specific O antigen does not protect against infection with other serotypes. Therefore, the capacity to alter serotypes may be of advantage for Shigella spp. in the infectious process.
More than 13 S. flexneri serotypes are known, basically defined by differences in the O-antigen structure. The difference in serotype is caused by glucosylation and O acetylation of the basic O antigen. These modifications are catalyzed by enzymes, which are encoded predominantly by bacteriophages. The only known exceptions are the genes for determination of O antigen in serotype 1, which are obviously located on a PAI termed SHI-O (164).
The gene composition and order indicate that this PAI is derived from an ancestral bacteriophage which has lost its ability to be excised from the bacterial chromosome. A larger part of the phage genome is deleted. The remaining genome is flanked by two typical phage attachment sites which are located in an unusually short distance of 6.5 kb. Interestingly, the G+C content of SHI-O is around 40%, much lower than the G+C content of the rest of the Shigella chromosome (49 to 53%). Besides phage gene remnants and insertion elements, the SHI-O island contains three ORF whose products have high sequence identity to proteins encoded by other serotype-converting bacteriophages. The respective gene products are putatively involved in glucosylation reactions necessary for serotype conversion. The first gene, gtrA, encodes a highly conserved, hydrophobic 120-amino-acid integral membrane protein with unknown function. It may function as flipase for the UndP-glucose precursor (124). The gtrB gene encodes a bactoprenol glucosyltransferase which may catalyze the transfer of glucose from UDP-glucose to bactoprenol phosphate, and the third gene, gtr, determines a serotype-specific glucosyltransferase (5).
SHI-1. The second PAI, first described in S. flexneri serotype 2a, has been completely sequenced from strain YSH6000T. This PAI was originally termed "she" because it contains the she gene (282). It consists of 46,603 bp, is located directly downstream of the pheV tRNA gene, and includes an imperfect repeat of the 3'-end 22 bp of the pheV gene at the right boundary of that PAI. SHI-1 contains a bacteriophage P4-like integrase gene, intact and truncated mobile genetic elements, plasmid-related sequences, ORF similar to those found in the EHEC LEE and SHI-2 and in the sigA, pic (she), set1A, and set1B genes, as well as two novel ORF (4, 282). Several of the proteins encoded by SHI-1 are thought to be virulence factors. SigA and Pic (also known as ShMU) belong to the autotransporter family of proteins (143). The Shet1 protein is an enterotoxin of the CT/LT-like toxin family. SigA (Shigella IgA-like protease homologue) is a 139.6-kDa protein exhibiting protease activity and cytopathic effects on HEp2-cells. Mutants with mutations in sigA caused 30% less fluid accumulation in a rabbit loop model of infection, suggesting that this protein plays a role in Shigella pathogenesis.
The Pic (for "protein involved in intestinal colonization") protein is a serine protease of 109.8 kDa, which is able to cleave gelatin as well as bovine and murine mucin. The mucus layer overlying the mucosal surface is considered to be a protective barrier against enteric infections (283). Some enteric pathogens have developed strategies to penetrate this layer, such as flagella motion or production of mucus-degrading enzymes (81, 103). It has been speculated that Pic could be involved
Sal PAI of MRSA is shown. A remarkable
feature of PAI in S. aureus is the presence of a large number
of genes with related functions, such as genes for enterotoxin (dark
grey) or lipoproteins (grey). Modified from reference
9.