Evolution and Ecology of Salmonella

Mollie D. Winfield and Eduardo A. Groisman*

[Section Editor: Monica Riley]

Posted February 27, 2004

Department of Molecular Microbiology, Howard Hughes Medical Institute, Washington University School of Medicine, 660 S. Euclid, St. Louis, MO 63110

*Corresponding author: Phone: (314) 362-3692, Fax: (314) 747-8228, E-mail: This e-mail address is being protected from spambots. You need JavaScript enabled to view it

INTRODUCTION

Salmonella is a major cause of disease worldwide as a result of ingestion of contaminated food or water, with disease manifestations ranging from gastroenteritis to bacteremia and typhoid fever. Since its divergence from the ancestor that would give rise to the nonpathogenic commensal species Escherichia coli 120 to 160 million years ago (56, 87), Salmonella has undergone numerous genetic events, resulting in variants with the ability to infect and often cause disease in >100 different species, including mammals, birds, reptiles, and insects.

The genus Salmonella comprises two species: Salmonella enterica and Salmonella bongori (96). S. enterica is further divided into seven subspecies: enterica (I), salamae (II), arizonae (IIIa), diarizonae (IIIb), indica (IV), houtenae (VI), and several serovars previously assigned to group IV (VII) (Fig. 1). Strains belonging to subspecies I are responsible for 99% of the cases of human salmonellosis (11, 124).

Fig. 1Phylogeny of the genus Salmonella. The genus Salmonella is divided into two species, S. enterica and S. bongori. S. enterica is further subdivided into seven subspecies. The two groups that are closest to the ancestral form, S. bongori and S. enterica subsp. IIIa, are typically associated with cold-blooded hosts and are not capable of flagellar phase variation. All other S. enterica subgroups are associated with warm-blooded hosts and harbor the genetic machinery required for flagellar phase variation.

The genome sequences of two S. enterica strains have been determined (68, 89); together with the near completion of the sequencing of several other Salmonella strains, this allows whole-genome comparisons of isolates differing in host range and virulence (21, 91). These comparisons provide insight into the evolutionary path of the genus Salmonella and attempt to explain when and how the genetic and phenotypic differences between the groups came about. It has been proposed that the evolution of the genus Salmonella occurred in five stages (91). Each of these five evolutionary time points marks genomic alterations (i.e., gene acquisition, rearrangements, and the emergence of pseudogenes) that generated genetic variation allowing expansion of particular strains into new niches.

In this chapter, we use the five time points in the history of Salmonella as the framework to discuss the genetic and phenotypic variation within the genus and how this variation correlates with host specificity. Finally, we consider a set of Salmonella core genes presumed to be necessary for the lifestyle of all serovars in nonhost environments, which helps our understanding of the ecology and evolution of the genus.

DIVERGENCE FROM E. COLI

Following the speciation event from a common ancestor (56, 87), E. coli evolved as a commensal of mammals. (Several pathogenic E. coli types have emerged that differ from commensal strains in harboring large virulence gene clusters called pathogenicity islands and lacking other genetic material [127]). On the other hand, the ancestral Salmonella remained associated with reptiles (61) until the divergence of the Salmonella lineages ~30 million years ago (9). The two most ancestral groups recognized today, S. bongori and S. enterica subsp. IIIa, have maintained an association with cold-blooded hosts, while other S. enterica subspecies, particularly subspecies I, expanded their host ranges to warm-blooded animals (21) (Fig. 1).

The most apparent difference between nonpathogenic commensal E. coli and pathogenic Salmonella is the presence of species-specific genes (85). There are >1,100 genes in the S. enterica serovar Typhimurium LT2 genome that are absent from the E. coli K-12 genome, and >800 E. coli K-12 genes lack homologues in the serovar Typhimurium LT2 genome (68). Serovar Typhimurium LT2 and E. coli K-12 share 55% of their proteins, which is higher than the 39.2% of proteins shared among E. coli K-12, enterohemorrhagic E. coli EDL933, and uropathogenic E. coli CFT073 (127). A majority of Salmonella subspecies harbor 10 distinct regions, called Salmonella pathogenicity islands (SPI), not found in E. coli, along with numerous smaller genomic islets (35, 43, 66, 89). The ancestral Salmonella lineage evolved following the acquisition of several genetic elements required for host cell invasion and induction of diarrhea in the infected animal (7, 14, 61). These include the 50-gene SPI-1 encoding a type III secretion apparatus and several secreted effectors, as well as the effectors SopB (SPI-5), SopD, and SopE, which are encoded outside SPI-1 (32, 72, 73). Likewise, the mgtC gene in the SPI-3 island is necessary for the survival of Salmonella in low-magnesium environments, such as inside host cells (12). Phylogenetic analysis has demonstrated that the SPI-1 genes, and a five-gene cluster of SPI-3 that includes mgtC, are present in all the salmonellae, suggesting that they were acquired early in Salmonella evolution (73, 84, 91) whereas other pathogenicity islands were acquired more recently. Microarray analysis examining gene content has established that Salmonella gained a total of 513 genes following its divergence from E. coli, 61% of which are genes of unknown function (91).

Measurement of Genetic Variation

Numerous techniques have been used to measure the genetic variation within the genus Salmonella. These include serotyping (58), biotyping (58), phage typing (48), multilocus enzyme electrophoresis (MLEE) (11, 15, 60, 96, 103), DNA hybridization (55), and nucleotide sequencing (21, 91).

Serotyping.

The Kauffmann-White serotyping scheme, which is based on the antigenic variation in surface molecules, has been used to identify and classify Salmonella strains for over 60 years (112). The extensive antigenic variation in the somatic lipopolysaccharide (O) antigen, the phase 1 and phase 2 flagellar (H1 and H2, respectively) antigens, and the capsular (Vi) antigen (which is specific to serovar Typhi) can be used to distinguish >2,300 serovars (58, 60). The O antigen is synthesized by enzymes encoded in the rfb locus, whereas the fliC and fliB genes encode the phase 1 and phase 2 flagellins, respectively (60). The lipopolysaccharide O antigen varies in the structure of the trisaccharide backbone and side chains, resulting in 46 distinguishable O serotypes (28, 126). Antibodies specific for the numerous O antigens are used to classify Salmonella into six serogroups designated A, B, C1, C2, D, and E (102). Further classification of serotypes requires analysis of the antigenicity of the flagellar H antigen. Flagellins are highly conserved in the carboxy and amino termini. However, the central regions exhibit high sequence variation (112), allowing the identification of 50 H1 serotypes and 30 H2 serotypes (111). Antigenic variation at the Vi antigen locus is used to characterize isolates of serovar Typhi (102, 104).

Serotyping is a convenient method for strain identification and epidemiology. However, distantly related isolates may have the same serotype and closely related isolates may have different serotypes, suggesting horizontal gene transfer and recombination between strains (11, 60), which has been shown to occur via viral transduction and bacterial conjugation (133). Thus, the horizontal transfer of surface antigen genes and recombination between distantly related strains do not allow an accurate assessment of the evolutionary relationships among strains and serovars by this method.

Multilocus enzyme electrophoresis (MLEE).

MLEE measures genetic variation by assessing changes in the electrophoretic mobilities of specific enzymes as a result of allelic polymorphisms. Unlike serotyping, the method provides a measure of genetic variation within and between populations (104). MLEE analysis of 24 enzymes revealed that the mean diversity per enzyme locus within the salmonellae is ~0.260 (11, 15), whereas the total diversity between Salmonella groups per locus is ~0.560 (15, 96). These data mean that there is a 26% probability that a locus will be heterozygous within groups (i.e., different strains of the same serovar) and there is a 56% probability that a given locus will be heterozygous between groups (i.e., different serovars): the higher the value the higher the genetic diversity. The total diversity among the salmonellae is similar to the total diversity that has been detected among E. coli strains and among Haemophilus spp. (0.573 and 0.52, respectively) but much greater than the variation observed among Shigella spp., Legionella pneumophila strains, and Bordetella spp. (0.29, 0.313, and 0.284, respectively) (96). The phylogenetic data produced by MLEE analysis have been invaluable in elucidating the evolutionary relationships within the genus Salmonella (11, 15, 96, 104), which have been corroborated by comparative genomic analyses (21, 91).

SPECIATION WITHIN SALMONELLA

The split of Salmonella into two species, S. enterica and S. bongori, marks the second evolutionary time point in the history of Salmonella. This event is characterized by S. enterica’s gain of 111 genes (59% uncharacterized), some of which may have provided the key innovations for its adaptation as a pathogen of warm-blooded hosts. These include the intracellularly expressed type III secretion system encoded by SPI-2 (84, 91), which is required for the survival and proliferation of salmonellae inside macrophages (35, 86, 107).

S. bongori

S. bongori, originally isolated from a lizard in Chad, is associated with cold-blooded animals (58). The inability of S. bongori to survive within mammalian host cells is due in part to the absence of SPI-2 (21, 46, 84, 91). SPI-2 was acquired by S. enterica following the split from S. bongori (84). The absence of SPI-2 from the S. bongori genome is one of the important genetic differences marking the speciation event within the genus Salmonella. The split between S. enterica and S. bongori is estimated to have occurred 25 to 40 million years ago (9). S. bongori shows a genetic distance of 0.57 from all other Salmonella subgroups (96). This amount of diversity defines a new Salmonella species, following the criterion that a genetic distance between 0.50 and 0.60 represents significant divergence for discrimination between bacterial species (105). Moreover, genomic subtraction estimates the amount of nonhomologous DNA between serovar Typhimurium and S. bongori to be 20%, much higher than the genomic difference of serovar Typhimurium from a second isolate of serovar Typhimurium, from serovar Meuchen (subspecies I), and from serovar Typhi (2, 5, and 9%, respectively) (55).

MONOPHASIC VERSUS DIPHASIC SEROVARS

The salmonellae can be divided into two groups based on whether they display one or more flagellar phase types, termed monophasic or diphasic, respectively. Salmonella serovars capable of diphasic H antigen expression, known as flagellar phase variation, harbor two genes encoding the filament protein flagellin, fliB and fliC (77, 109). Phase variation occurs through the reversible inversion of a 970-bp region upstream of the fliB gene that contains the fliB promoter. When this region is in the forward orientation, FliB, the H2 antigen, as well as FliA, a repressor of FliC, are expressed, and when the region is flipped to the opposite orientation, FliC, the H1 antigen, is expressed (77, 110). In this way, only one flagellin surface protein is expressed at any given time. The switch from expressing one flagellin type to expressing a different one occurs with a frequency of 10⁻³ to10⁻⁵ per bacterium per generation (109). This variation in antigen presentation creates a polymorphic population and increases the amount of genetic variation (109). It has been suggested that this mechanism for creating antigenic diversity plays a role in evasion of the warm-blooded-host immune system (60); however, data supporting this hypothesis are not currently available.

S. enterica subsp. IIIa (arizonae) is characteristically monophasic and, like S. bongori, lacks the fliA and hin genes (91). Therefore, subspecies IIIa and S. bongori do not have the genetic machinery required for the diphasic switch between the H1 and H2 flagellar antigens (Fig. 1). Similar to S. bongori, the original strain of subspecies IIIa was isolated from a reptile and is typically found in association with cold-blooded animals (58).

Subspecies I, II, IIIb, and VI are diphasic strains of Salmonella (91) which have been found to cause disease in warm-blooded organisms (Fig. 1). This suggests that acquisition of the genetic components for flagellar phase variation may have been a key step in the expansion of Salmonella as a pathogen of warm-blooded hosts (61), marking the third evolutionary time point (91). Genetic analysis has revealed that, while initially biphasic, the subspecies I serovar Typhi reverted to the monophasic phenotype, possibly as a result of its extremely narrow host range, as it is a pathogen only of humans (77, 91).

S. ENTERICA SUBSP. I RADIATION

Host range defines an important aspect of Salmonella pathogenesis. On the one hand, S. bongori and S. enterica subsp. II and III are members of the normal flora of cold-blooded hosts (58) as well as pathogens, because human infections occasionally occur as a result of contact with reptiles (33, 101). On the other hand, certain serovars of S. enterica subsp. I cause different diseases in warm-blooded hosts, depending on the genetic complement of the serovar and the ability of the pathogen to interact with the host (27). For example, serovar Typhi is not able to cause disease in mice (123). Conversely, a mouse infected with serovar Typhimurium will develop a severe systemic disease resulting in death (102). Furthermore, serovars Pullorum and Gallinarum are specifically adapted to cause disease in poultry and are unable to infect mammals, therefore presenting little threat to public health (108). It appears that S. enterica subsp. I serovars have evolved to cause disease in specific vertebrate hosts.

S. enterica subsp. I represents a heterogeneous collection of strains exhibiting 79 to 100% relatedness by DNA hybridization (24). The serovars belonging to this group harbor additional segments of DNA, including SPI-4 and SPI-5, which are absent from other Salmonella subspecies (1). SPI-4 contains 18 open reading frames, three of which appear to encode products involved in the secretion of toxins (35). SPI-5 is a 7-kb region encoding six putative proteins, four of which are involved in pathogenesis in calves (128). The diversification within subspecies I marks the fourth evolutionary time point in the history of Salmonella (91).

The radiation of S. enterica subsp. I separates typhoid (i.e., the human-adapted strains causing typhoid fever) and nontyphoid (i.e., strains that cause gastroenteritis) strains. It also distinguishes groups on the basis of host specificity: host-restricted serovars, host-adapted serovars, and non-host-adapted serovars (96, 124). Host-restricted serovars are those specifically adapted to cause disease in a particular animal species (124) and include serovars Typhi (103), Gallinarum (62), and Abortusovis, which cause systemic disease in humans, birds, and sheep, respectively (124). Host-adapted serovars, including Dublin (106), Choleraesuis (21), and Pullorum (63), typically cause disease in cattle, pig, and bird populations, respectively, but have also been associated with disease in other vertebrates, including humans (124). The non-host-adapted serovars, such as Typhimurium and Enteritidis, are able to infect virtually all animals, both vertebrates and invertebrates (71, 124).

S. enterica Serovar Typhi

S. enterica serovar Typhi is the etiological agent of typhoid fever in humans, a disease which continues to be a major health concern in developing nations that lack proper water sanitation methods (89; http://www.who.int/). Natural populations of serovar Typhi demonstrate low genetic heterogeneity, as determined by biochemical (88), restriction fragment length polymorphism (65), and MLEE (96) analyses. The mean genetic diversity per locus is reported to be 0.083 (103), suggesting global distribution of a single clone. However, chromosomal analyses, including pulsed-field gel electrophoresis (121), partial chromosomal digestion (64), and total genome sequencing (89), reveal genomic plasticity between strains. In contrast to the highly conserved genome organization displayed by most members of the genus Salmonella, the serovar Typhi genome organization varies as a result of frequent recombination among its seven rRNA (rrn) genes (64). These genomic rearrangements are responsible for creating new serovar Typhi ribotypes (83) but appear to be strongly selected against in the other serovars (64). Moreover, the plasticity of the serovar Typhi genome has resulted in significant variation in genome size among isolates, ranging from 4,000 to 4,900 kb (121).

Comparison of the serovar Typhi and Typhimurium genomes demonstrates conservation of overall gene order (89) but differences in gene content. S. enterica serovar Typhi strain CT18 contains 601 genes absent from serovar Typhimurium strain LT2, and serovar Typhimurium strain LT2 has 479 genes not found in serovar Typhi strain CT18 (89). Examples of the serovar Typhi-specific genes include those encoding four putative chaperone-usher fimbrial systems (staA to -G, tcfA to -D, steA to -G, and stgA to -G), a putative hemolysin (STY1498), putative toxins (STY1886, STY1890, and STY1891), and a putative polysaccharide acetyltransferase, as well as SPI-7, SPI-8, and SPI-10 (89). In addition to differences in gene content, serovar Typhi CT18 differs from serovar Typhimurium LT2 in that it harbors 204 pseudogenes, 71% of which are functional genes in serovar Typhimurium (89). One explanation for the minimization of bacterial genomes, particularly those of symbiotic species, has been the "use it or lose it" hypothesis, which states that bacterial species that inhabit relatively unchanging environments, such as the obligate intracellular pathogens Mycoplasma spp., have undergone a reduction in genome size (25, 76). Although serovar Typhi is by no means a symbiotic species and is subjected to a number of environments during the course of infection, it is a pathogen specific to humans, perhaps suggesting that a host-specific lifestyle may not require as many genes as a non-host-specific lifestyle. The ability to detect pseudogenes in the serovar Typhi genome suggests that these mutations occurred relatively recently without sufficient time for elimination of the pseudogenes from the genome. The pseudogenes appear to derive from genes involved in a variety of functions: 37% are associated with housekeeping functions, 13% appear to be vestiges of mobile genetic elements, and 23% are involved in host interactions and pathogenicity (89). The last group includes genes from both SPI-1 and SPI-2 and the slrP gene, which is important for serovar Typhimurium host adaptation, as well as other genes encoding surface structures (89, 123).

SEROVAR TYPHIMURIUM RADIATION

S. enterica serovar Typhimurium is a member of the non-host-adapted serovars and is the predominant cause of human gastroenteritis. The radiation of serovar Typhimurium marks the fifth evolutionary time point in the history of Salmonella (91), resulting in at least 17 distinct globally distributed electrophoretic types with a mean genetic diversity per locus of 0.114 (11). Microarray analysis suggests that 144 new genes were acquired at this stage, 89% of which have unknown functions (91).

Given the public health concerns caused by the continued reemergence of this food-borne pathogen, there is great interest in defining the genetic mechanisms responsible for its virulence and its broad host range. Horizontal gene transfer has been implicated as one of the major mechanisms for generating the high genetic variation within the genus Salmonella, which can result in a 20% difference in gene content between subspecies (55). Approximately 10 to 15% of the serovar Typhimurium genes lack homologues in other serovars (91). For example, 11% are not found in the serovar Typhi genome and 29% are missing from E. coli K-12 (68). Moreover, it has been reported that 1% of S. enterica isolates are mutator strains which, being deficient in methyl-directed mismatch repair (57), demonstrate an increased (up to 1,000-fold) rate of mutation and recombination (94). Cells lacking the capacity to correct errors in DNA synthesis (74) and to prevent the recombination of nonhomologous DNA (67, 95) may gain increased genetic variation for natural selection to act upon, which has been suggested to provide a selective advantage for the lifestyle of pathogenic bacteria (120).

Serovar Typhimurium harbors numerous genes believed to be important for its ability to infect a variety of organisms. These include virulence genes within four prophages: Gifsy-1, Gifsy-2, Fels-1, and Fels-2, which are specific to the serovar Typhimurium genome (27, 29, 68). The mosaic structure and mobile abilities of these phages suggest a potential role in shuffling virulence elements between isolates, thereby providing this serovar with a collection of polymorphic loci that may enhance the pathogenic capacity of the organism (29, 44). For example, phylogenetic analysis of effectors translocated by the type III secretion system encoded in SPI-2 demonstrates a variable distribution of these virulence determinants throughout the genus Salmonella. Serovar Typhimurium and serovar Enteritidis, which is also a ubiquitous serovar, harbor all the known SPI-2 effectors, whereas serovar Typhi does not (44), raising the possibility that the presence or absence of particular effectors may be correlated with the host range.

Genetic and Phenotypic Variation

Historically, the lack of sexual reproduction suggested that bacterial species were clonal populations with little genetic diversity (26). Despite the basically clonal structures of natural populations of Salmonella, the genus displays high within- and between-group variation, a result of recombination of genetic material from other Salmonella strains, as well as other bacterial species (11). This bacterial "sex" has led to shuffling of genetic material between highly divergent groups of Salmonella. Recombination has had a particularly important role in generating the diversity in the phase 1 flagellin gene that, in turn, has been pivotal in the generation and characterization of new Salmonella serovars (111, 112). Although first thought to be a result of mutation and genetic drift, the variation observed in the central region of the fliC alleles is now believed to be a consequence of recombination (111). The incongruence between the phylogeny of the fliC sequences of the Salmonella strains tested and the phylogeny of the Salmonella chromosomes themselves suggests that recombination has occurred at the fliC locus (17, 111).

Whole-genome comparison across the genus Salmonella using microarray technology demonstrates phylogenetic clustering of human-specific serovars (i.e., serovars Typhi, Paratyphi A, Paratyphi C, and Sendai), host-adapted serovars (i.e., serovars Dublin, Enteritidis, and Pullorum), and serovar Typhimurium strains (21). These clusters reflect not only genetic relatedness but also similarity of lifestyles (21). The amount of genetic diversity appears to correlate with the host range: host-adapted serovars demonstrate lower genetic diversity per locus than non-host-adapted serovars (104). Thus, the phenotypic variation generated by genetic variation may allow survival in a variety of hosts.

Two possible explanations have been proposed for the relationship between genetic variation and host range (104). First, according to the neutral-mutation theory (54), genetic variation at a single locus depends on the effective population size (N_e). Therefore, a small effective population will have less genetic variation than a large effective population (104). Serovar Typhi has a small effective population size and displays less genetic heterogeneity than serovar Typhimurium (11). Second, under the assumption that allelic diversity at particular loci is adaptive, a narrow host range selects for genetic homogeneity. Adaptation to a diversity of niches (i.e., a broad host range), therefore, requires selection for many different genotypes (104). Neither of these hypotheses has been tested directly, and it may be difficult to do so (104); moreover, these two mechanisms are not mutually exclusive. Furthermore, Salmonella serovars, like any other group that has undergone an adaptive radiation, are in a balance with each other. This means that different serovars are specifically adapted to different ecological niches (i.e., resource partitioning [119]). Alterations in this balance, due to natural competition between strains or to human attempts to eradicate a particular serovar, may lead to the emergence of new serovars filling the empty niche (6).

Competitive Exclusion of Salmonella

The principle of competitive exclusion states that two species cannot occupy the same environment if they use it in the same way (119). Thus, if two species attempt to occupy the same niche, one of the species will inevitably outcompete the other (119). The effects of this principle have had both negative and positive outcomes for Salmonella epidemics (6, 69, 93, 125). Prior to the 1970s, the avian-adapted serovars Gallinarum and Pullorum were endemic in poultry populations in Europe and the Americas, resulting in major economic losses to the poultry industry (18). As domestic and aquatic fowl are the only animal reservoirs for these two Salmonella serovars, "test-and-slaughter" methods led to complete eradication of the pathogens from commercial poultry flocks (18, 117). At the same time, the number of cases of human food-borne illness due to serovar Enteritidis began to increase (2, 3, 47, 90, 97, 118). The coincidence of the eradication of serovars Gallinarum and Pullorum from poultry and the increased incidence of egg-associated serovar Enteritidis (23, 47, 118) led to the hypothesis that serovar Enteritidis filled the ecological niche vacated by serovar Gallinarum (6).

Serovars Gallinarum, Pullorum, and Enteritidis have the same immunodominant surface O antigen (O9) (5). Immunization of chickens with serovar Gallinarum specifically protects against infection with serovar Enteritidis (79), thereby limiting transmission of serovar Enteritidis to humans via infected eggs and poultry (6, 93). Immunoprotection is not observed for serovars carrying alternative O antigens, such as serovar Typhimurium O4 (50). The coexistence of serovars Gallinarum, Pullorum, and Enteritidis in natural populations led to competition for the same ecological niche, resulting from the shared O9 antigen (6, 93). Elimination of the avian-specific pathogens led to loss of O9 antigen immunity in the poultry populations and allowed the expansion of serovar Enteritidis from rodent species into fowl with subsequent transmission to humans (6, 93). It has been suggested that control of the present serovar Enteritidis epidemic may require restoration of a natural ecological competitor, most practically through vaccination of chickens with a live attenuated serovar Gallinarum strain (93).

The principle of competitive exclusion has recently been exploited as a means to prevent contamination of agricultural animals (particularly chickens) with food-borne human pathogens (69, 70, 125). These methods take advantage of the protective immunity imparted by live attenuated Salmonella strains (70), as well as the rapid establishment of adult-type intestinal microflora in newly hatched chicks (69, 125). Both of these strategies are designed to exclude the pathogenic species as a result of competition for the same ecological niche.

Salmonella Core Genes

As discussed above, the genus Salmonella displays high genetic and phenotypic heterogeneity. Nonetheless, microarray data have been used to define a set of genes found in all Salmonella species and serovars (21, 91). This core group of genes comprises ~54% of the serovar Typhimurium LT2 genes and includes those required for virulence and those involved in basic cellular functions, as well as several with unknown functions (21). A small subset of these genes is Salmonella specific, suggesting a possible role for the genes in aspects unique to the lifestyle of Salmonella (91). These include genes participating in virulence, such as seven genes present in SPI-1 that are found in all Salmonella subspecies (32, 35, 91).

The lifestyle of Salmonella also requires genes involved in survival under the harsh conditions of nonhost environments, such as soil and water. The majority of nonhost environments are characterized by low nutrient availability. As a result, Salmonella must sense its environment and subsequently acquire nutrients, suggesting the participation of some of the Salmonella-specific core genes in survival outside animal hosts (Table 1). For example, the pcgL gene encodes a periplasmic dipeptidase required for cleavage of and growth on the peptidoglycan-derived substrate d-alanyl-d-alanine (49), which originates during peptidoglycan remodeling and would also be released from the cell walls of dead bacterial cells (49, 59, 78). Inactivation of the pcgL gene renders Salmonella defective for survival under nutrient-poor conditions (78), such as those that Salmonella might encounter outside an animal host. Salmonella subspecies appear to have retained the pcgL gene as a result of the contribution the gene makes to fitness in nonhost environments (78).

Table 1Table 1. Salmonella-specific genes with putative functions in survival outside the animal host

The alternative electron acceptor tetrathionate facilitates the anaerobic growth of Salmonella on ethanolamine or propanediol using endogenously synthesized vitamin B₁₂ (92, 100). This metabolic process requires genes making up ~2% of the Salmonella genome, the majority of which are found in all Salmonella isolates but not in E. coli (92). These genes include sulfur reduction operons (ttr, phs, and asr) (19, 22, 51), the 1,2-propanediol degradation operon (pdu) (68), the propionate operon (prp, which is also found in E. coli), the ethanolamine degradation operon (eut, which is also found in E. coli) (13, 68), and vitamin B₁₂ synthesis genes (cob and cbi) (92, 122). These data have led to the hypothesis that Salmonella may frequently encounter an anaerobic, tetrathionate-rich environment containing ethanolamine and/or propanediol as a carbon source, as this B₁₂-dependent metabolism appears to be an important aspect of the Salmonella-specific lifestyle (92). It has been suggested that Salmonella subspecies acquired the genetic components necessary to synthesize B₁₂ and catabolize propanediol by horizontal gene transfer ~70 million years ago (92). The sulfur degradation operons (ttr, phs, and asr) appear to be ancestral genes lost from the E. coli lineage (92). The pdu and cob genes are transcriptionally activated in the presence of 1,2-propanediol (99). Furthermore, both of these genes activate the prpBCDE operon, which is needed for growth on either 1,2-propanediol or propionate (123). As the Salmonella-specific CobB protein is also expressed in the absence of 1,2-propanediol, propionate can be catabolized independently of the presence of 1,2-propanediol (122). Cumulatively, these data suggest that the salmonellae harbor the unique ability to grow on propionate as a result of the ability to synthesize B₁₂ and to reduce sulfur-containing compounds in anaerobic environments.

Iron is an essential nutrient for a majority of microorganisms and a critical environmental signal for the regulation of a wide variety of genes important for the survival of bacteria (38). Salmonella, as well as many other gram-negative bacteria, harbor two iron-responsive global transcriptional regulators, PmrA and Fur. The PmrB protein is a sensor for periplasmic Fe³⁺, which activates the PmrA protein in the presence of high levels of iron (129). The PmrA regulon confers resistance to the antimicrobial peptide polymyxin B (98), resistance to iron-mediated killing (129), virulence in mice (39), invasion of chicken macrophages (132), and survival in soil (20). The Fur protein, on the other hand, senses intracellular Fe²⁺ levels, largely acting as a repressor when iron concentrations are high inside the cell (4). Low Fe²⁺ derepresses the Fur regulon, activating the transcription of genes important for iron acquisition and resistance to oxidative stress (4, 41, 42, 45, 131). The availability of Fe²⁺ in the environment is severely limited by the oxidation of Fe²⁺ to Fe³⁺ and the low solubility of Fe³⁺ at neutral pH (16, 38). To overcome this difficulty, bacteria secrete iron uptake molecules, called siderophores, which themselves are regulated by intracellular iron levels (16, 38, 80, 81). Salmonella harbor an iron-responsive, Fur-regulated locus called iroA composed of iroBCDE and iroN (10, 30, 31, 130), which displays wide phylogenetic distribution among Salmonella serovars and is absent only from S. bongori (10); indeed the iroB gene has been used for rapid PCR identification of Salmonella strains (7). This locus exhibits an average G+C content higher than that characteristic of S. enterica, suggesting acquisition via horizontal gene transfer (8). The products of both iroDE and iroN display identity to gene products associated with siderophore utilization: the E. coli enterochelin utilization protein Fes and the E. coli enterochelin receptor FepA, respectively (8). The IroN protein facilitates the growth of S. enterica in soil (8), raising the possibility that acquisition of the iroA locus provided Salmonella with enhanced iron uptake capability, thereby facilitating its ability to survive in nonhost environments.

The phoN gene of Salmonella encodes a periplasmic nonspecific acid phosphatase (53) that exhibits sequence similarity to an acid phosphatase from Citrobacter sp. strain N14, which has been implicated in detoxification of heavy metals that accumulate in metal-contaminated soil (75). The phoN gene is transcriptionally regulated by the PhoP/PhoQ system, which is activated under magnesium starvation conditions (34, 53).

Mg²⁺ is the most abundant divalent cation in living organisms and plays numerous roles in cell physiology (113). Salmonella harbors three magnesium transporters: CorA, MgtA, and MgtB (52, 68, 115, 116), and an Mg²⁺-responsive PhoP/PhoQ two-component regulatory system (34). The mgtCB operon, which encodes the Mg²⁺ transporter MgtB (34, 115, 116) and the MgtC membrane protein, which is required for normal growth under Mg²⁺-limiting conditions (12), is specific to Salmonella. In contrast to the mgtC gene, the mgtB gene is not required for virulence, suggesting that it may enhance the survival of Salmonella in nonhost environments (36). Moreover, MgtA and MgtB demonstrate differing affinities for other divalent cations, which may result in the inhibition of Mg²⁺ transport by one of the systems, depending on the growth conditions outside the host (114), emphasizing the importance of both systems for the survival of Salmonella. Likewise, the putative copper binding protein ScsA and the putative thiol-disulfide interchange proteins ScsBCD are necessary for copper tolerance (40, 68), and the putative nickel transporter NxiA (68) may contribute to survival outside animal hosts.

CONCLUSION

The genus Salmonella is both genetically and phenotypically diverse. The evolution of this genus has occurred as a result of selection acting on genetic variation produced by a variety of mechanisms. During the course of its history, Salmonella has acquired genes important for survival both inside and outside host environments. Comparative genomic studies will continue to shed light on the evolution and ecology of this genus.