Reticulate Evolution Among the Group I Salmonellae : An Ongoing Role for Horizontal Gene Transfer

Salmonella is an extremely diversified genus, infecting a range of hosts, and comprised of two species: enterica and bongori. This group is made up of 2579 serovars, making it versatile and fascinating for researchers drawing their attention towards different properties of this microorganism. Salmonella related diseases are a major problem in developed and developing countries resulting in economic losses, as well as problems of zoonoses and food borne illness. Moreover, the emergence of an ever increasing problem of antimicrobial resistance in salmonella makes it prudent to unveil different mechanisms involved. This book is the outcome of a collaboration between various researchers from all over the world. The recent advancements in the field of salmonella research are compiled and presented.


Introduction
Salmonella enterica is responsible for 1.4 million cases of foodborne salmonellosis in the United States annually making it the number one causative agent of bacterial foodborne illnesses (CDC, 2007). Infection can occur after eating undercooked meat, poultry and eggs that have been contaminated with Salmonella (CDC, 2007). In recent years several outbreaks have occurred in the United States that were associated with Salmonella contamination of produce, the most recent being a S. enterica Saintpaul outbreak associated with tomatoes, jalapeño and serrano peppers that sickened over 1400 individuals (CDC, 2008). The movement of several serovars of Salmonella into previously naïve niches (i.e., producegrowing environs) suggests that the pathogen is readily adapting to new environments. An understanding of the reticulate evolutionary mechanisms that underpin the acquisition and composition of the requisite genetic and phenotypic features of Salmonella is essential to more accurate risk assessment of this pathogen (Hohmann, 2001).
It is now widely accepted that horizontal gene transfer (HGT) has driven the emergence of more aggressive and virulent strains of Salmonella in the environment, on the farm, and in the food supply. Such assault by various salmonellae has fueled the in-depth examination of specific genotypes and conditions that permit reticulate evolutionary change and the rise of deleterious phenotypes (LeClerc et el., 1996;Cebula and LeClerc, 1997). The hypermutable phenotype represents one scheme by which reticulate evolution of the bacterial chromosome may occur (Trobner and Piechoki, 1984;Haber et al., 1988;Haber and Walker, 1991;LeClerc et al., 1996;Matic et al., 1997;Radman et al., 1999;2000;Funchain et al., 2000). Methyl-directed mismatch repair (MMR) defects, leading to a mutator or hypermutable phenotype, are found in more than 1% of the isolates within naturally-occurring populations of Salmonella enterica (LeClerc et al., 1996) and at even greater frequencies in the food supply where oxidative and other anti-microbial stressors are applied (Cebula et al., 2001). Up to 73% of the MMR defects found in feral settings are due to lesions within the mutS gene, resulting in increased nucleotide substitution rates, enhanced DNA transposition, and, perhaps most importantly, a relaxation of the internal barriers that normally restrict homeologous recombination following HGT of foreign DNA (Cebula and LeClerc, 1997;Radman et al., 1999). This latter role, as a major sentinel for recombination, led to a substantial focus on the genetics and evolution of the mutS gene and its adjacent sequences located at 63 min on the Salmonella chromosome (Brown et al, 2002;Kotewicz et al., 2003;. Phylogenetic analyses of mutS alleles from strains of the SAR (Salmonella reference) collections (i.e., SARA, SARB, and SARC)-largely taken to represent the extent of genetic variability within the species (Boyd et al., 1993;Beltran et al., 1991)-have revealed striking levels of phylogenetic discordance between trees derived from mutS alleles and wholechromosome trees of the same strains based on MLEE (multilocus enzyme electrophoresis) analysis (Brown et al., 2002. These differences were interpreted as numerous examples of HGT among mutS alleles in Salmonella. Similar observations have been made among sequences abutting the mutS gene in Salmonella, E. coli, and Shigella spp Brown et al., 2001b). Our laboratory showed previously that the 61.5 min mutS-rpoS region retains a novel and highly polymorphic 2.9 kb sequence in the genome of all E. coli O157:H7 strains, Shigella dysenteriae type 1, and several other E. coli strains (LeClerc et al., 1999) but not in Salmonella enterica (Kotewicz et al., 2003). This highly polymorphic stretch of DNA (previously coined the mutS-rpoS "unusual region") is varied in its distribution among enteric bacterial lineages and is absent in others entirely (Kotewicz et al., 2003). Sequence analysis of the region revealed an IS1 insertion element in place of the prpB gene in S. dysenteriae type 1 suggesting the existence of a recombinational crossover in the mutS-rpoS region for this strain (LeClerc et al., 1999). Evidence for additional crossovers in the same region were also obtained for other E. coli strains (Brown et al., 2001b). These findings support the notion that HGT helped forge current relationships among Salmonella and other enteric pathogens in this region and throughout numerous other locales in the Salmonella chromosome.
Indeed, as evidenced from global efforts involving whole-genome sequencing, microarray, and multi-locus sequence typing, the substantial impact that HGT has played in structuring the chromosome of Salmonella enterica is now indisputable (Porwollik and McClelland, 2003;Fricke et al., 2011;Kelly et al., 2009;Hall, 2010). Previous estimates indicate that at least onequarter of the Salmonella genome may have been forged through HGT and reticulate evolutionary events (Porwollik and McClelland, 2003), although this number seems conservative from current views. In addition to the 61.5 min region surrounding mutS, HGT has played a key role in structuring many other regions of the Salmonella chromosome. Notably, SPI elements (Salmonella pathogenicity islands) have likely been acquired through HGT Ochman et al., 2000;Hacker and Kaper, 2000;Baumler et al., 1997). For example, the SPI-1 pathogenicity island, comprising the genes encoding a type III secretion system, was probably acquired early in Salmonella evolution (Kingsley and Baumler, 2000;Li et al., 1995), yet several inv-spa alleles seem to have converged horizontally more recently between S. enterica groups IV and VII Brown et al., 2002). Additionally, type 1 pilin genes that encode fimbrial adhesins retain unusually low GC contents and aberrant DNA sequence phylogenies relative to other fim genes (Boyd and Hartl, 1999). Other studies focusing on numerous housekeeping gene loci have reported evolutionary histories for these genes that are strikingly decoupled from S. enterica strain history Thampapillae et al., 1994;Brown et al., 2002; The now incontrovertible connection between horizontal transfer and MMR gene evolution has led to the thesis that genetic exchange of mutS alleles could simultaneously quiet the mutator phenotype while rescuing adaptive changes from the population (LeClerc et al., 1996;Denamur et al., 2000). Consistent with this hypothesis, the mutS gene is evolutionarily scrambled by HGT in subspecies I Salmonella enterica. Our laboratories documented the prevalence of horizontal gene transfer (HGT) among strains of Salmonella enterica (Brown et al., 2002;. In comparing across and within subspecies of Salmonella, a recombination gradient was noted wherein the incidence of HGT was inversely correlated with the genetic diversity separating individual strains. It appears that a genetic threshold exists that tolerates free exchange of sequences within a framework delimited by sequence variation and niche diversity of individual strains. We demonstrated this through identification of intragenic (patch-like) recombination as the primary outcome across disparate Salmonella subspecies and assortative (whole-allele) recombination which caused extensive reassortment of alleles among more genetically homogeneous populations of group I Salmonella pathogens, all sharing a common niche restricted to warm-blooded mammals.
A torrent of scientific information has accrued over the past decade to support the important role of HGT in the genetic and evolutionary diversification of S. enterica subspecies, serovars, and individual pathogenic clones (McQuiston et al., 2008;Octavia and Lan, 2006;Lan et al., 2009;Fricke et al., 2011). Our understanding in reconstructing the horizontal acquisitions of important features including those involved in virulence, drug resistance, and other adaptations that foster an enhanced fitness for Salmonella persistence in foods, animals, and people is expanding at a pace which we could not have foreseen even a decade ago (Sukhnanand et al., 2005). It is important to recall however that reticulate evolutionary pressures do not subside once selectively advantageous traits are gained. Rather, horizontal exchange likely continues to dapple the evolutionary landscape between even the most closely related salmonellae . Here, we provide results of several previously unreported phylogenetic studies that evidence (i) the continued role of HGT in the intra-operon shuffling of SPI-1 alleles among subspecies I S. enterica strains; (ii) the often under-appreciated role for HGT and recombination in the homogenization of allele structure in a closely related population of S. enterica; and (iii) the panmictic and reticulate nature of restriction-modification (R-M) genes among group I salmonellae. This last finding, noting free exchange of R-M (i.e., hsd) alleles, provides phylogenetic evidence of the compatibility of S. enterica subspecies I R-M complexes, likely accounting for the documented successful HGT of entire gene sequences among closely (e.g., intra-subspecies) related strains as DNA exchange between strains that shared or recently shared common R-M alleles would not be subject to substantial restriction (Sharp et al., 1992).

Reticulate evolution in SPI-1 of Salmonella enterica subspecies I
Salmonella pathogenicity island 1 (SPI-1) specifies a type III secretion system essential for host cell invasion and macrophage apoptosis (Galan and Curtiss, 1989;Galan and Collmer, 1999). SPI-1 comprises a cluster of virulence genes (e.g., the inv/spa gene cluster) that encode, in part, the "needle complex", a key delivery component for transporting virulence associated effector molecules into the host cell (Galan and Collmer, 1999). The disparate phylogenetic distribution, lack of chromosomal synteny, and diverse base compositions of SPI-1 and its homologues indicate that these sequences were obtained independently across enteric species of bacteria. It is presumed that SPI-1 was present in the last common ancestor of all Salmonella lineages. Horizontal acquisition of the inv/spa gene cluster, however, is thought to have been a pivotal event for the emergence of Salmonella as a pathogenic species Groisman and Ochman, 2000). The gene complex lies adjacent to the polymorphic mutS-rpoS region of the chromosome. We and others previously presented phylogenetic evidence for intragenic recombination of sequences within several SPI-1 invasion loci Brown et al., 2002), primarily among S. enterica subspecies IV and VII. However, in order to determine the extent to which HGT may have disrupted SPI-1 evolution across the more ecologically and genetically homologous group I salmonellae, we examined nine SPI-1 invasion loci from nearly half of the SARB reference collection of strains (Boyd et al., 1993), composed exclusively of subspecies I Salmonella serovars.

SPI-1 gene evolution is decoupled from Salmonella chromosome evolution
Using a cladistic approach (Forey et al., 1992;Allard et al., 1999;Bell et al., 2011), the nucleotide sequences from nine invasion gene sequences were subjected to phylogenetic analysis. The resultant invasion gene phylogenies were then compared to phylogenetic groupings from the mdh gene, a chromosomal anchor locus that is taken largely to reiterate chromosome evolution within subspecies I (Boyd et al., 1994) and MLEE (multi-locus enzyme electrophoresis), also applied here as a metric of strain/chromosome evolution for the group I salmonellae (Boyd et al., 1993). As shown in Fig. 1, strains composing single SARB mdh and MLEE lineages were, for the most part, distributed across disparate inv/spa gene clades for all nine invasion genes tested indicating that many of these strains, although linked tightly in chromosome evolution, retain invasion gene alleles with unrelated evolutionary histories, presumably as a result of HGT.
Evolutionary incongruence between inv/spa genes and the Salmonella chromosome was affirmed using the ILD (incongruence length difference) test, which evaluates the likelihood of a common evolutionary history between genes (Farris et al., 1995;LeCointre et al., 1998;Brown et al., 2001a). Seven of the nine invasion genes yielded significant ILD scores (p < 0.05), indicating that a hypothesis of congruence could be rejected for these strains and further reinforcing the discordance evident in the clade comparisons. The only exceptions were invB (p = 0.08) and spaP (p = 0.59), albeit both still retained cladistic signatures of HGT from broken clade structures in the tree analysis.

SPI-1 gene evolution is decoupled from mutS gene evolution
The mutS gene, downstream and adjacent to SPI-1 in S. enterica, has been shuffled extensively by HGT . In order to determine whether mutS may have been linked in the recombination now evident among SPI-1 genes, cladistic comparisons were made between mutS phylogeny and inv/spa gene phylogeny revealing substantial incongruence between inv/spa trees and mutS trees. Six of these comparisons are shown in the form of tanglegrams (Fig. 2). Again, strains composing SARB mutS clades were distributed across disparate inv/spa gene clades for all nine invasion genes tested, and seven of nine inv/spa genes were further www.intechopen.com Reticulate Evolution Among the Group I Salmonellae: An Ongoing Role for Horizontal Gene Transfer 213 confirmed as discordant with mutS based on ILD testing. Taken together, these findings indicate that inv/spa gene sequences and mutS sequences from the same strains are decoupled in their evolution. These data suggest that reticulate evolution has repeatedly forged this contiguous region of the Salmonella chromosome such that different strains appear to have been affected by assortative (allelic) HGT between the two loci. Fig. 1. Phylogenetic discordance between SPI-1invasion genes and the Salmonella chromosome. mdh and MLEE comparisons are shown to each of nine different inv/spa genes indicated. Identical letters denote strains from the same mdh or MLEE lineage. It is important to note that letters are only relevant to their respective data column and do not cross-over columns. The column to the left of the dividing line designates mdh clade assignments for the respective S. enterica strain while the column of letters to the right of the divider corresponds to MLEE clade assignments. The number at the base of each tree denotes the ILD score (p-value) relative to a comparison for congruence between the respective inv/spa gene and the mdh gene sequence alignment for the same strains. Trees shown were rooted using S. bongori as an outgroup. Nucleotide sequence alignments were performed using CLUSTAL X (Thompson et al., 1998). Most parsimonious trees were generated in PAUP* v.10 (Swofford et al., 2002). Tanglegrams of several invasion gene and mutS revealing the phylogenetic incongruence between inv/spa genes and the mutS, which lies adjacent to SPI-1 on the Salmonella chromosome. Lines connect the discordant, potentially recombinagenic (incongruent) strains. inv/spa to mutS comparisons with an ILD score of p < 0.01 were displayed. Trees shown were again rooted using S. bongori as an outgroup taxa.

www.intechopen.com
Reticulate Evolution Among the Group I Salmonellae: An Ongoing Role for Horizontal Gene Transfer 215

Intra-island HGT within the SPI-1 region of subspecies I Salmonella strains
In order to determine the presence and extent to which HGT has shuffled individual alleles within SPI-1 among more closely related subspecies I strains, a pairwise ILD approach was adopted wherein congruence was scored for individual comparisons of all nine of the inv/spa genes included in this study (Fig. 3). Several findings were noteworthy. Although no individual invasion gene showed unanimous evolutionary discordance with its neighbors, three inv/spa loci (invA, invB, and spaP) were incongruent (p < 0.10) with a significant majority of other genes. invA and invB showed discordance with all other loci except spaN and spaQ, while spaP showed discordance to all but spaM and spaQ. Conversely, with the exception of spaQ, no inv/spa gene was congruent with every other. Thus, a hypothesis of extensive intra-island shuffling begins to emerge with an evolutionary decoupling of individual invasion loci one from another. Additional tree comparisons buttressed this conclusion. Akin to the selfish operon theory (Lawrence and Roth, 1996), these data suggest that the SPI-1 region is a chromosomal mosaic, composed of inv/spa gene sequences that have converged within this island but with each retaining unique evolutionary paths. Fig. 3. ILD test results for intragenic comparisons among inv/spa invasion genes. ILD tests (Farris et al., 1995) were performed with 1000 partitions using the Partition Heterogeneity command in PAUP* v.10 (Swofford et al., 2002). A p-value of 0.05 or less allows for a rejection of the null hypothesis of congruence (vertical evolution) and accepts the alternative hypothesis of incongruence which is interpreted among bacterial phylogeny as evidence for HGT (LeCointre et al., 1998).

Key observations
i. The inv/spa complex of S. enterica subspecies I appears to have undergone extensive intra-island allelic shuffling due to HGT. This suggests that the SPI-1 region is a mosaic composed of SPI-1 gene sequences with distinct evolutionary origins. ii. Invasion genes within this Salmonella population are not only decoupled phylogenetically from mutS and other flanking sequences but also from the chromosomes of group I S. enterica strains, suggesting that these genes have been reassorted by HGT. iii. Much of the recombination observed here appears to be assortative transfer, a finding that contrasts to the inv genes in S. enterica as a whole, where tree structure was largely intact with HGT limited mostly to subspecies IV and VII Brown et al., 2002). iv. Allele shuffling appears to be most prominent within the subspecies I taxonomic boundary and not across other subspecies of S. enterica. This finding is consistent with a relaxed and compatible restriction-modification system among more closely related Salmonella strains .

HGT homogenizes the mutS gene among 'Typhimurium' complex strains
Here, we present phylogenetic and genetic analyses of Salmonella reference collection A (SARA), also known as the Typhimurium strain complex-the most homogeneous S. enteric reference collection, consisting solely of five closely related subspecies I serovars (Typhimurium, Paratyphi B, Muenchen, Saintpaul, and Heidelberg) (Beltran et al., 1991). Given the evolutionary similarity shared among these pathogens and trend noted previously that highlight the inverse relationship between Salmonella diversity and recombination, one would expect to observe an even greater role for HGT in the population structure of the S. enterica SARA collection of pathogens.

Cladistic evidence for horizontal exchange of mutS alleles among 'Typhimurium' complex strains
As was done for SPI-1 gene sequences, a phylogenetic tree was derived from 72 SARA mutS sequences and was compared to phylogenetic trees derived from multi-locus enzyme electrophoresis (MLEE) and mdh (malate dehydrogenase) gene sequences for the same strains. Phylogenies derived from horizontally exchanged sequences display evolutionary discordance (incongruence) when compared to mdh and MLEE trees. In the tree shown, six clades of mutS alleles were observed and compared to the distribution of four mdh and six MLEE multi-strain containing clades (Fig. 4). Two of the four SARA mdh clades were found to be displaced into multiple clades on the mutS tree. Two additional mdh clades were found to have converged into a single mutS clade, suggesting that HGT may have homogenized mutS diversity of these particular mutS lineages. Similarly, strains from five of the six MLEE lineages were displaced into separate clades on the mutS tree. The only exception was a single clade of MLEE SARA strains (A57, A58, A59, and A60), which was also found intact in the mutS tree except for the inclusion of SARA strain A56. Nonetheless, numerous examples of evolutionary discordance between the 1.1 kb mutS segment and the chromosome of the 'Typhimurium' complex strains indicate that horizontal exchanges of mutS alleles have accumulated during the rather shallow radiation of even these highly homogeneous g r o u p I p a t h o g e n s . A s a n a s i d e , i t w a s www.intechopen.com Reticulate Evolution Among the Group I Salmonellae: An Ongoing Role for Horizontal Gene Transfer 217 noteworthy that full-length mutS alleles were horizontally transferred among SARA S. enterica strains, lending further credence to a model for R-M compatibility among closely related S. enterica serovars and strains. Fig. 4. Most-parsimonious relationships of SARA mutS alleles. mutS clades are bracketed and numbered to the right of the tree. Distributions of mutS, mdh, and MLEE clades are presented in column form. Note that strains originating from the same clade retain a common shape and common internal shading. Bootstrap nodal support values (Felsenstein et al., 1985) are presented on the mutS tree as follows: ^, 76-100%; *, 51-75%; +, 26-50%; o, 1-25%. In this case, mdh and MLEE are taken to represent the evolution of the strain in general (Boyd et al., 1994;Beltran et al., 1991). The tree shown is rooted with two E. coli outgroups.

Homogenization of mutS sequence diversity among S. Typhimurium and S. Heidelberg strains
Curiously, a single clade in the SARA mutS tree was found to comprise three distinct Salmonella serovars. In this clade, every strain representing S. Typhimurium (n=21) and S. Heidelberg (n=11), along with a single strain of S. Saintpaul, converged into a single evolutionary lineage of mutS alleles. In the SARA mdh tree (Fig. 5), mdh alleles for these same SARA serovars formed three disparate clades in the tree such that S. Typhimurium strains clustered only with other S. Typhimurium and S. Heidelberg strains only with other S. Heidelberg. S. Saintpaul strains formed a single lineage at the tip of the tree with strains of S. Muenchen and a single S. Paratyphi B. It should be noted that these distinct clades retained substantial statistical support with bootstrap values around 90% (Felsenstein, 1985). Thus, phylogenetic comparison of mutS and mdh sequences supported the notion that these serovars have converged into a single mutS clade, possibly as a result of the repeated HGT of only one or a few preferred mutS alleles. For sample sizes greater than one, multiple strains of the same serovar are depicted as a cone on the tree terminal nodes. Note that strains originating from the same clade are designated by a common bracket and letter. Bootstrap nodal support values are presented on the mdh tree as follows: ^, 76-100%; *, 51-75%; +, 26-50%; o, 1-25%. Note the bifurcations between specific clusters in the tree, signaling sequence diversity among distinct serovars using the mdh gene.
In order to further investigate the genetic structure of this converged clade, we examined mutS sequence homogeneity across the strains composing this lineage as well as the remaining mutS alleles of the SARA collection (Fig. 6). Evaluation of polymorphic positions in the mutS alignment revealed several findings consistent with homogeneous clade structure surrounding these serovars. First, five substitutions were observed across the entire 1,115 bp sequence for all 33 strains that define this mutS clade (#2). Second, with the exception of the polymorphism at position 913 in SARA strains 12 and 13, no clade #2 substitution was retained by more than one strain. Thus, none of the substitutions present within this clade partitioned any member serovar from another. The near structural uniformity of this clade at the nucleotide level further suggests that HGT has homogenized mutS alleles among these particular serovars. This is consistent with the thesis of Dykhuizen and Green (1991) who reminded that recombination can not only diversify the genome but can also homogenize it as well. Fig. 6. mutS nucleotide sequence homogeneity among S. enterica serovars Typhimurium, Heidelberg, and a strain of Saintpaul. Periods indicate exact nucleotide identity to the reference sequence at the top of the alignment while listed nucleotides represent actual substitutions. The synonymous/nonsynonymous status (blackened ovals indicate synonymous change) of each substitution is noted below the alignment. Nucleotide sequences were generated using a PCR-based approach and automated CE-sequencing technology.

Distinct roles for HGT across various taxonomic tiers of S. enterica
With the inclusion of the SARA analysis reported here, we have been able to define varying roles for HGT across three taxonomically distinct populations of S. enterica (SARA, B, and C) (Fig. 7). Within S. enterica as a whole, a model for HGT begins to emerge that tolerates nearfree HGT among closely-related subspecies I strains. As genetic divergence increases across serovars, however, the extent of HGT appears to decrease. The analysis reported here suggested two unique findings for SARA, the most genetically monomorphic population.  to have been influenced more extensively by HGT than SARC possibly because they are not s o d i v e r g e d t h a t e x c h a n g e i s i n h i b i t e d due to extreme niche or R-M (restrictionmodification) system variability. Moreover, it is also possible that much of the HGT among SARA strains have gone undetected here since identical alleles would leave no phylogenetic footprint following an exchange event.

Key observations
i. Horizontal gene transfer of mutS alleles in Salmonella appears to play a prominent role in the evolutionary structure of the five closely-related serovars representing the SARA ('Typhimurium' complex) collection, a finding consistent with extensive HGT that has been documented among subspecies I serovars in general . ii. Cladistic analysis of SARA strains revealed the first example of a substantial convergence of mutS alleles from disparate serovars into a single clade. This suggests that HGT is homogenizing allele diversity among certain Salmonella strains and serovars-an observation reminiscent of allele homogenization observed for the E. coli polA gene (Patel and Loeb, 2000). iii. Among closely related 'Typhimurium' complex strains, mutS alleles appear to have shuffled largely as single units rather than in intragenic segments. One explanation for this might be a more recent evolutionary divergence of the five serovars composing the highly homogeneous 'Typhimurium' strain complex. Alternatively, recombination of highly homologous mosaic segments of the mutS gene would do little to obscure phylogeny and likely go undetected in these analyses. iv. Retrospective comparison of SARA HGT patterns with that of SARB and SARC strains yields a gradated model for HGT whereby different taxonomic tiers of Salmonella are subject to different HGT effects. The differences appear coupled to the extent of genetic diversity that defines these three different "tiers" of Salmonella population structure.

HGT among restriction-modification (R-M) genes of subspecies I salmonellae
The restriction and modification (R-M) system is a defense mechanism developed by bacteria to protect the bacterial genome from invasion by foreign DNA (Bullas et al., 1980). Foreign sequences entering the cell are cleaved by restriction enzyme(s), while the bacterial DNA itself is modified by methylase(s), thus providing protection from its own restriction enzyme (Murray, 2000). R-M systems are composed of genes that encode a specific restriction endonuclease and modification methylase. There are several types of R-M systems, namely type I (e.g., EcoKI), type II (e.g., EcoRI), and type III (e.g., Sty LTI) ( Barcus et al., 1995). Types of R-M systems are classified on the basis of their composition and cofactor requirements, the nature of the target sequence, and the site of DNA cleavage with respect to the target sequence (Murray, 2000;Naderer et al., 2002).
Compatibility of R-M systems among strains was proposed as one explanation to account for contrasting recombination rates . In this model, compatible R-M complexes would permit the successful transfer of larger gene segments among closely related Salmonella pathogens; crosses between strains with identical R-M systems would not be subject to restriction (Sharp et al., 1992). A gradation in the size limits of DNA segments exchanged would depend on the polymorphic character of R-M systems in natural strains.
Here, we investigate this model by examining the molecular evolutionary relationships of hsd genes encoding R-M complexes among closely related pathogenic Salmonella strains (i.e., the 'Typhimurium' complex). If, indeed, hsd alleles are freely exchanged themselves among strains that display a substantial tolerance for HGT and recombination of diverged DNA sequences, then an explanation accounting for observed tolerance to extensive HGT begins to emerge for S. enterica group I serovars.

Evidence for HGT of R-M alleles among Salmonella enterica group I strains
DNA sequences from three hsd type I R-M genes were subjected to cladistic analysis. The resultant invasion gene phylogenies were then compared to phylogenetic groupings from the mdh gene and from the Salmonella MLEE data. Cladistic comparisons of hsd genes to markers of stable Salmonella chromosome evolution revealed several findings, and the data for hsdS is shown (Fig. 8). For hsdS section S1, SARA 56 is removed from neighboring strains when compared to mdh or mutS. For hsdS section S2, the collapsing of numerous clades into a single conserved clade was observed. It should be noted that such collapsing was observed in many of the trees reported here and suggests that HGT may be homogenizing hsd alleles. Moreover, a distinct allele that has no homology with its sister allele in a neighboring clade can be seen on the tree. Finally, the hsdS section S3 tree breaks up clades from both MLEE and mdh. In addition, this tree has three distinct allele types that can be seen phylogenetically, as in the case of S2. Compatibility among R-M systems has been proposed to account for the extensive levels of HGT documented among subspecies I Salmonella pathogens. Since the mutS gene appears to have been shuffled among this group of strains, we examined the phylogenetic relationship of mutS to type I R-M hsd genes. Incongruence was observed between hsd genes and mutS phylogeny, suggesting that patterns of HGT for hsd alleles differ from those for mutS alleles. hsd segments S1 and S3 each retained at least one incongruent strain between these gene phylogenies. In addition, several hsd genes collapsed divergent mutS clades into single hsd lineages in the trees. For instance, three mutS clades composed a single hsdM clade, a pattern that held true for other hsd genes, including hsdS segments 2 and 3. These data indicate distinct roles for HGT between most R-M genes and mutS. Nonetheless, observed homogenization of type I R-M loci among subspecies I Salmonella strains suggests they are compatible systems, allowing additional genes like mutS to be transferred within this population in its entirety.

Evidence for intra-operon HGT of R-M alleles
Intra-operon evolutionary incongruence between hsd genes was further examined using the ILD (incongruence length difference) test, which evaluates the likelihood of a common evolutionary history between genes. The ILD comparisons yielded more notable incongruence between genes than did the tanglegram analysis, suggesting that small patches of sequence within individual genes may be responsible for much of the observed incongruence. Intragenic patterns of HGT have been noted previously for more diverse subspecies (Brown et al., 2002). In the ILD comparisons ( Fig. 9), eight of the ten hsd data set comparisons yielded significant Fig. 9. Pairwise incongruence length difference (ILD) test results for several genes of the Type I Restriction -Modification system in Salmonella. ILD tests were performed with 1000 partitions using the Partition Heterogeneity command in PAUP*v.10 (Swofford et al., 2002). Pairwise ILD comparisons were made among the three hsd genes R, M, and S including the three sub-regions that were amplified from hsdS (i.e., S1, S2, and S3). As in Fig. 3, a p-value of 0.05 or less allows for a rejection of the null hypothesis of congruence (vertical evolution) and accepts the alternative hypothesis of incongruence which is interpreted among bacterial phylogeny as evidence for HGT (LeCointre et al., 1998).
ILD scores (p < 0.05) such that a hypothesis of congruence could be rejected for these intragene comparisons. The only exceptions were the hsdS2-hsdM comparison (p = 1.00) and the hsdS3-hsdS2 comparison (p = 1.00). It is noted that, with the exception of hsdR, all of the hsd data matrices were also incongruent with mutS. When examined in total, the data suggest that the Type-1 R-M operon is a mosaic comprising hsd gene sequences that have converged evolutionarily within this operon, but with each possessing a unique phylogenetic path.
It was also noteworthy that hsdS segments S2 and S3, however, retained groups of alleles that shared little or no identified homologies. That is, hsdS2 yielded two unique sequence cassettes, one of which was found in strains of serovar S. Paratyphi B in the SARA complex. hsdS3 yielded three distinct cassette types within the alignment, all of which shared no homology with their counterparts. A cassette retained by SARA strain 56 showed homology to an hsdS variant in E. coli, suggesting that this sequence has resulted from HGT between these lineages. The other unique cassette, retained by SARA strains 49, 50, and 51, showed no homology to any other hsd sequence, indicating that it may be been transferred into S. Paratyphi B from a yet unidentified source. The examples of unique cassette formation within this gene reinforce the role that HGT has played in the intra-operon and intragenic evolution of the Type-1 R-M gene system. These data also reveal the exchangeable nature of hsd gene sequences in these loci as a result of HGT.

Key observations
i. These findings demonstrate several instances for the three hsd loci encoding the type I R-M operon in Salmonella to be decoupled phylogenetically from the chromosomes of group I Salmonella strains (i.e., mdh and MLEE), suggesting that certain alleles from these genes have been shuffled by HGT between closely related S. enterica strains. ii. The hsd operon of S. enterica subspecies 1 appears to have undergone intra-operon structuring due to HGT, producing an evolutionary mosaic in the hsd region. iii. The lack of homology within hsdS indicates that these specific segments may have been acquired from distantly related bacterial species. An aberrant GC content for hsdS of 41%, a value far removed from an average value for enteric bacterial genomes of 56%, reinforces this conclusion. iv. The data demonstrate that HGT has been a common occurrence in hsd gene evolution and point to a genetic compatibility among closely-related salmonellae for exchange of hsd alleles that appears to resemble a panmictic genetic structure among these closely related strains. This may explain, in part, why Salmonella known to share homologous genomes and common niches more freely exchange DNA.

Discussion and conclusions
In summary, substantial phylogenetic evidence has been presented for the horizontal transfer of mutS alleles within a pathogenically homogeneous group of subspecies I Salmonella enterica pathogens. Of note, is the observation that mutS clades appear to be undergoing homogenization within the 'Typhimurium' strain complex as a result of the repeated HGT of only a few preferred alleles. Moreover, examination of R-M loci revealed that subspecies I Salmonella readily exchange hsd genes. These findings support the notion that R-M compatibility may be, in part, responsible for the substantial tolerance of HGT and recombined DNA between subspecies I strains.
www.intechopen.com Reticulate Evolution Among the Group I Salmonellae: An Ongoing Role for Horizontal Gene Transfer

225
An overwhelming body of evidence has been compiled that documents the reticulate evolutionary nature of the Salmonella mutS gene and its surrounding sequences. In an analysis of nearly 200 different strains documented here and in several previous reports over the past decade (LeClerc et al., 1998;Brown et al., 2002;Kotewicz et al., 2002;, our laboratory has demonstrated the extent, the chromosomal effects, and the evolutionary history of HGT events that have scrambled this part of the genome in S. enterica. Exhaustive phylogenetic comparisons have been brought to bear on mutS sequences using various chromosomal markers including MLEE, rDNA, several individual housekeeping genes, and an a priori prior agreement MLST data based on concatenation of a three-gene supermatrix (Brown et al., 2002). Puzzling then was a later report that argued a more nonremarkable evolutionary pattern for mutS stating, "mutS is not more recombinogenic than the other genes" (Octavia and Lan, 2006). The authors based this conclusion solely on a modest 15 strain set of subspecies I salmonellae. Albeit, it remains to be seen to what extent additional homogeneous Salmonella populations retain the phylogenetic vestiges of horizontally transferred mutS and hsd alleles. Whatever the final outcome, it is apparent that horizontal transfer has played a prominent role in the current evolutionary structure of mutS and many other genes with virulence, resistance, stress-response, and general housekeeping function, all underscoring recombination as a key mechanism in the generation of genetic diversity among these closely related salmonellae.
Roughly two decades ago, Salmonella enterica was regarded as one of only a few eubacterial species that maintained a "truly clonal" evolutionary structure (Selander et al., 1990;Reeves et al., 1989). Today, armed with whole-genomic analysis, it is now clear that horizontal transfer has shaped and honed unique evolutionary histories for numerous genes, operons, and islands within the Salmonella chromosome. With the complete genome sequences of dozens of Salmonella now available and countless more underway, such analyses of congruence should aid in determining the extent to which recombination has disrupted clonality throughout the entire Salmonella chromosome. Certainly, a greater recognition of precisely how HGT has forged the genomes of Salmonella pathogens should enhance the accuracy of risk assessment strategies for these bacteria as well as provide avenues for better detection and characterization of this devastating foodborne pathogen.

Acknowledgments
The authors would like to thank Mr. David Weingaertner for repeated and excellent graphical assistance. We would also like to acknowledge Drs. M. Kotewicz, B. Li, A. Mukherjee, A. Shifflet, J. Zheng, S. Jackson, and J. Meng for numerous helpful discussions over many years.