Invasive Salmonellosis in Humans

GEMMA C. LANGRIDGE, JOHN WAIN,* AND SATHEESH NAIR

[SECTION EDITOR: GORDON DOUGAN]

Revised version posted August 18, 2008

(replaces August 26, 2005 archived version)

The Wellcome Trust Sanger Institute, Wellcome Trust Genome Campus, Hinxton, Cambridge CB10 1SA, United Kingdom

*Corresponding author. Mailing address: The Wellcome Trust Sanger Institute, Wellcome Trust Genome Campus, Hinxton, Cambridge CB10 1SA, United Kingdom. Phone: 44 (0) 1223 494828, Fax: 44 (0)1223 494919, E-mail: This e-mail address is being protected from spambots. You need JavaScript enabled to view it

INTRODUCTION

Salmonellosis in the human host is generally associated with Salmonella enterica from subspecies enterica (also termed subspecies I). Acute infections can present in one of four ways: enteric fever, gastroenteritis, bacteremia, and extraintestinal focal infection. As with other infectious diseases, the course and outcome of the infection depend on a variety of factors, including inoculating dose, immune status of the host, and genetic background of both host and infecting organism. Here, we review the salmonellae, for which invasive disease in humans is an important aspect of their population biology, and we focus on the bacterial factors that may contribute to the "invasiveness" of different serovars in the human host.

S. enterica causing human infection can be broadly divided into two groups: the enteric fever (typhoidal) group and the nontyphoidal salmonellae (NTS) (Fig. 1). The NTS typically cause gastroenteritis but have the capacity to cause invasive disease, under certain conditions. Enteric fever is associated with five serovars of S. enterica, the most common of which are serovars Typhi and Paratyphi A; serovars Paratyphi B, Paratyphi C, and Sendai are less frequently isolated. Serovar Typhi is the causative agent of typhoid fever, while serovars Paratyphi A, B, and C are the causative agents of paratyphoid fever. Collectively, typhoid and paratyphoid fever are known as enteric fever. While serovar Typhi on its own, and serovars Paratyphi A and Sendai together, form genetically homogenous groups, serovars Paratyphi B and Paratyphi C are genetically heterogeneous (91). These enteric fever salmonellae appear to be intrinsically invasive in the human host as, upon infection, these strains usually cause systemic disease and are rarely associated with gastroenteritis.

Fig. 1Diagram representing the entire Salmonella genus. Salmonella species and subspecies have been defined by biotyping, DNA-DNA hybridization (25), 16S RNA analysis, and multilocus enzyme electrophoresis (MLEE) (85). Serotyping is used for differentiation beyond the level of subspecies. Numbers in brackets indicate the total number of serotypes included in each subspecies (10). *Common serotypes are listed, but other serotypes may cause bacteremia or focal infection; subsp., subspecies.

The NTS, on the other hand, are more typically associated with gastroenteritis and rarely cause extraintestinal (EI) focal infections in the human host. This difference in clinical outcome is thought to be partly due to the way the different serovars interact with the gut epithelium, with the NTS causing an acute inflammatory reaction that prevents systemic spread of the bacteria. Although rarely, NTS can cause systemic disease, typically when the host’s defense is compromised by underlying disease or immunodeficiency, or when there are lesions that provide a favorable environment for colonization. The highest number of EI NTS infections are caused by the serovars prevalent in the food chain and, consequently, most frequently associated with gastroenteritis (e.g., Salmonella enterica serovar Typhimurium and serovar Enteritidis). However, when the number of cases of bacteremia is expressed as a percentage of the number of cases of gastroenteritis (invasive index) within a defined geographical region, it becomes apparent that certain serovars are more invasive than others (Table 1). While serovars Typhimurium and Enteritidis are frequently isolated as a cause of Salmonella gastroenteritis, their invasive indexes are usually low. For example, between 2002 and 2005 the National Reference Centre for Salmonella and Shigella in Belgium received reports of 158 invasive cases of 7,628 culture-confirmed cases of serovar Typhimurium (1.8% invasive index) and 603 invasive cases of 23,900 culture-confirmed cases of serovar Enteritidis (2.5%). Similarly, a 10-year registry-based study in Denmark revealed an invasive index of 1.8% (126 invasive of 7,021 culture-confirmed) for serovar Typhimurium and 1.8% (261 invasive of 14,533 culture-confirmed) for serovar Enteritidis (47). In contrast, serovar Choleraesuis, a less prevalent serovar, has been reported in various countries with an average invasive index of 66.1% (18, 62, 97, 106). Other serovars, including serovar Virchow, serovar Newport, and serovar Schwarzengrund, have been reported with invasive indexes as high as 10.3%, 17.4%, and 25%, respectively (62). Serovar Dublin has been documented in Brazil at 65.1% (56 invasive of 86 culture-confirmed) (33) and in Belgium at 40.5% (15 invasive of 37 culture-confirmed); reports from England and Wales and the United States also show serovar Dublin to be invasive in the human host (97, 106).

Table 1Invasive index of nontyphoidal salmonellae

Understanding the bacterial factors responsible for an invasive phenotype is a field of knowledge that is still growing. Genetic comparisons of the different serovars, including plasmid carriage (89), have identified some genes that may be important for systemic spread in the human host (Table 2). There are now studies in the literature looking at the expression of these genes, which may also give clues as to the importance of various virulence determinants (3, 83). Recently, there has been some effort to ascertain expression and proteomic profiles from in vivo and in vitro studies of Salmonella in macrophages. Selective capture of transcribed sequences (SCOTS) has been used to identify genes transcribed by serovar Typhi while growing inside human macrophages (26, 32). Total cDNA is prepared from infected cells, then bacterial cDNA is selectively captured by hybridizing to biotinylated bacterial genomic DNA. This approach has revealed intracellular expression of 36 serovar Typhi genes that are not present in serovar Typhimurium, including 25 found on Salmonella pathogenicity islands (SPIs) and prophage elements (31). Of these, five genes of uncharacterized function were found on prophage ST18. This is of particular interest because this prophage has been found only in serovar Paratyphi A (full-length element) and in serovars Paratyphi C and Choleraesuis (partial element) (80). Another group has used liquid chromatography-mass spectrometry-based proteomics to analyze serovar Typhimurium and identified a novel protein STM3117 that contributes to the replication of the serovar inside murine macrophages (93). This protein is part of a pathogenicity "islet" STM3117-9, which is predicted to be involved in biosynthesis and modification of cell wall peptidoglycan. On a wider scale, whole-genome expression profiling has also contributed to this field, revealing new insights into Salmonella-macrophage interactions, especially in the case of serovar Typhimurium in the murine model of pathogenesis (29).

Table 2Specific genetic traits of Salmonella serovars

Immunoscreening of sera from acute and convalescent typhoid patients has been used to look for immunogenic proteins expressed by serovar Typhi during human infection (45). This in vivo-induced antigen technology (IVIAT) identified four antigens, two of which were specific to acute-phase patients. One of these was PagC (PhoP/PhoQ activated gene C), an outer membrane invasion protein implicated in survival inside the macrophage (69, 82) and the other was TcfB, a fimbrial subunit of the tcf operon found in few Salmonella serovars (Table 2). Another PhoP/PhoQ-regulated gene, PagN, may also have a role in the invasive phenotype. This gene is found on SPI-6 in serovar Typhi, as is the tcf operon, and encodes a putative outer membrane protein that binds epithelial cells (S. Smith [Trinity College, Dublin, Ireland], personal communication). The exact mode of binding has yet to be determined but it is also believed to interact with OmpD. Performing a BLAST search using the nucleotide sequence of PagN reveals that this gene has homologues in serovars Paratyphi A, Choleraesuis, and Typhimurium.

ENTERIC FEVER SALMONELLAE

Serovar Typhi causes typhoid fever, which is the most common form of enteric fever; it was first described by both Eberth and Koch in 1880 and subsequently cultured by Gaffki in 1884. Agglutination with specific sera and the ability to ferment sugars allow serovar Typhi to be differentiated from nonpathogenic Escherichia coli. An inverse relationship between pathogenicity and the ability to ferment alternative sugars has been described, with isolates from food poisoning being intermediate in both respects. Isolates resembling the food-poisoning bacteria have also been isolated from cases of enteric fever but these can be differentiated from other enteric-fever-causing organisms by culture on lead acetate media. These isolates were termed "B. paratyphosus" (114). An improved form of the lead acetate test is still in use (2, 65).

Enteric fever is a febrile, systemic illness. Fever itself is the only consistent symptom (77, 78), although abdominal pain or discomfort, muscle and/or joint pain, and headache are frequently observed. All these resolve relatively quickly if a course of an appropriate antibiotic is administered, but they can persist for several weeks in the untreated patient (77, 78). Some typhoid patients develop complications, the most severe of which is gastrointestinal (GI) hemorrhage and perforation (14, 30), a condition that requires both surgical and antimicrobial intervention and carries a high risk of mortality (13). A minority of infected persons become asymptomatic chronic carriers and continue to intermittently excrete serovar Typhi in their stools for prolonged periods (>1 year).

SEROVAR TYPHI

Serovar Typhi is restricted and adapted to the human host. It is identified by its antigenic formula, 9,12, [Vi]: d:- and specific biochemical reactions. It is less saccharolytic than most other members of the S. enterica species and forms a discrete group when analyzed by the phylogenetically relevant method of multilocus sequence typing (MLST) (58). The genome of serovar Typhi CT18 is predicted to contain more than 4,000 coding sequences and 204 pseudogenes. These are largely conserved in the second sequenced isolate serovar Typhi Ty2 (27), which supports the concept that genome degradation has contributed to the host restriction of serovar Typhi (76). When serovar Typhi became isolated in the human host, the function of genes required for survival in other hosts could have been lost in the absence of selective pressure. The presence of host adaptation, the ability to cause invasive disease, may also be the result of a loss of gene function. By reducing the ability to cause a secretory response in the human gut (e.g., loss of gene function on the CS54 island, see "Serovar Paratyphi A" below) the invasive nature of serovar Typhi infection may be favored. The acquisition of DNA has also played an important role in the evolution of salmonellae (59). The most noticeable genetic addition in serovar Typhi when compared with other salmonellae is the presence of Salmonella Pathogenicity Island (SPI) 7 (107). This island contains the genes responsible for the synthesis and transport of the Vi capsular polysaccharide (termed the ViaB region) (46), the SPI-1 type III secretion system (TTSS) effector protein sopE (44), and the pil genes that encode a type IVB pilus implicated in bacterial self-association (72, 100) and interactions with epithelial cells (116). SPI-7 is present in a very similar form in serovar Paratyphi C and some isolates of serovar Dublin (79). Subtle differences, however, may affect the properties encoded by regions of the island, for example, serovar Paratyphi C, unlike serovar Typhi, cannot self-associate via the products of the pil region. This is because sequence variation in the repeat regions flanking the serovar Paratyphi C shufflon cause the shufflon to be inactive, resulting in the stable production of PilV, a potential minor pilus protein that has been shown to have a negative effect on the self-association of serovar Typhi (96). The importance of pil-mediated serovar Typhi self-association in the pathogenesis of typhoid fever has yet to be demonstrated, although transfer of conjugative plasmids in vitro is enhanced by the ability to self-associate (72). Serovar Typhi CT18 has a unique repertoire of eight fimbrial operons (99). These are all present in various combinations with other fimbrial operons in several salmonellae but a simple correlation between fimbrial operons and host specificity, or disease syndrome, has not been demonstrated. Serovar Typhimurium LT2 has 13 fimbrial operons (68) and it is possible that a lack of fimbriae is associated with host restriction because of the increased need for immune evasion associated with dependence on a single host. Certainly, of the human restricted salmonellae, none possess the same major lipopolysaccharide (LPS) group, a fact interpreted as showing that the host immune response to the major serogroups (O-antigen epitopes) selects against less transmissible host-restricted S. enterica of the same serovar (103) (Table 3). This may also explain the variation in fimbriae expressed by different serovars. In serovar Paratyphi A, SopE from the serovar Typhi specific SPI-7 is present, but not in the same site as in serovar Typhi, while the rest of SPI-7 is entirely absent (79). SPI-7 can be missing from serovar Typhi in culture collections (73), but it is almost always present in fresh clinical isolates (108). This suggests that SPI-7 plays an important role during the infection process. The absence of most of SPI-7 from serovar Paratyphi A, which causes a disease very similar to typhoid in humans, is therefore of interest. An H-NS-regulated cytotoxin is one virulence factor shared by, and apparently restricted to, serovar Paratyphi A and serovar Typhi, the expression of which is enhanced in the serovar Typhi vaccine strain Ty21a (75).

Table 3Serotypes implicated in human invasive disease: antigenic formulas and indicators of variation

SEROVAR PARATYPHI A

Serovar Paratyphi A is the second most common cause of enteric fever. The number of reported cases in several Asian countries is currently on the increase, although the risk factors for acquiring serovar Paratyphi A may differ from those for typhoid fever (105). If this is the case, then, even though the pathogenicity of serovar Typhi and serovar Paratyphi A are very similar, there may be some differences in transmission routes.

Serovar Paratyphi A is part of a clonally related group of bacteria (90, 91), which also includes serovar Sendai (9). The completed genome sequence of serovar Paratyphi A revealed a very close relationship to serovar Typhi (67). There are 173 pseudogenes in serovar Paratyphi A compared with 204 in serovar Typhi. Of the 168 that could be identified as probable orthologues, 28 were pseudogenes in both (67). Of these, 27 had different mutations causing the predicted loss of the open reading frame. Thus, these two pathogens, if adapting to the niche of the human host by loss of gene function, seem to be doing so along a path of convergent evolution. Three pseudogenes, shdA, ratB, and sivH, are on a single island, CS54, and are shared by serovar Typhi and serovar Paratyphi A. ShdA binds fibronectin and is involved in the persistence of serovar Typhimurium in the mouse gut (61). All three genes are involved in colonization of the mouse gut (Peyer’s patch or cecum), and ratB and shdA, but not sivH, are required for fecal shedding of serovar Typhimurium LT2 in mice (60). The combined effect of knocking out all three genes together has not been reported to date. The importance of these three pseudogenes to the clinical picture seen in enteric fever, i.e., little early colonization of the gut and a lack of inflammatory response (50), is unclear. The other two genes present in CS54, ratA and sivI, have no detectable phenotype in serovar Typhimurium mouse models of infection.

The type III secretion systems, SPI-1 and SPI-2, are intact in both serovar Typhi and serovar Paratyphi A, although some genes encoding effector proteins (sopA, sopD2, sseJ, and slrP) are predicted to be pseudogenes in both serovars; once again the significance of this is unknown. One effector protein, SopE, is present in both serovar Typhi and serovar Paratyphi A and only in some epidemic strains of serovar Typhimurium (35). This protein is very closely related to SopE2, which is intact in serovar Paratyphi A but present as a pseudogene in serovar Typhi. SopE would therefore seem worthy of further investigation.

SEROVAR PARATYPHI B

Serovar Typhi and serovar Paratyphi A are both restricted to humans, whereas the serovar Paratyphi B group is more like the NTS in causing a range of infections in both humans and animals. In most clinical microbiology laboratories Salmonella with the antigenic formula 4:b:1,2 isolated from the blood of patients with suspected enteric fever would be typed as serovar Paratyphi B. Salmonella with this serovar, however, are more frequently isolated from the stools of patients with gastroenteritis. The ability to ferment d-tartrate (dT) has become the diagnostic method to distinguish between the enteric fever serovar Paratyphi B, which does not ferment d-tartrate (dT⁻), and the NTS-like serovar Paratyphi B var. java, which can ferment d-tartrate (dT⁺) (56). The ability to produce a slime wall has also been associated with the typhoidal strains; however, in a very detailed study of United Kingdom isolates, Chart et al. demonstrated that this distinction is not absolute and slime wall production can be the same for both dT⁺ and dT^– strains. Furthermore dT^– serovar Paratyphi B is more often associated with gastroenteritis than with enteric fever (15). Of 367 dT^– serovar Paratyphi B isolates, only 12.3% were isolated from blood, which suggests that, although dT^– serovar Paratyphi B is more invasive than serovar Paratyphi B var. java (1.3% of 1,007 cases), it is much less invasive than serovar Typhi or serovar Paratyphi A. Chart also noted that patients with invasive disease had a history of travel and suggested that there may be an invasive strain of serovar Paratyphi B circulating in Asia that could be adapted to the human host. The observation that serovar Paratyphi B (SPB) is heterogeneous genetically could explain the apparent low invasive index of SPB as a group. Pulse-field gel electrophoresis (PFGE) has been used to distinguish between serovar Paratyphi B and serovar Paratyphi B var. java, but it could not discriminate between invasive and noninvasive isolates in Malaysia (37). However, another study, using a comprehensive German reference collection, has shown an association between certain chromosomal digest patterns and an ability to cause invasive disease (81). Characterization of the serovar Paratyphi B group by multilocus enzyme electrophoresis (MLEE) revealed 23 electrophoretotypes (ETs) (90). All dT^– strains fall into a single ET, known as Pb1, whereas dT⁺ strains are found in all ETs. The restriction of dT^– strains to a single phylogenetic group underlines the importance of this discriminatory test. Tartrate metabolism in E. coli has been described in detail (84). Metabolism is oxygen sensitive and depends on two genes, ttdA and ttdB, which have orthologues in Salmonella; annotated as STM3355 and STM3354 in the serovar Typhimurium LT2 sequence, and STY3535 and STY3534 in the serovar Typhi CT18 sequence. A direct link between tartrate metabolism and invasive disease is unlikely but a rapid molecular test for the recognition of the genetic lesion should allow large-scale epidemiology to be carried out. Malorny et al. (65) performed a detailed analysis of this chromosomal region in serovar Paratyphi B and serovar Paratyphi B var. java. This led to the description of a point mutation in the putative start codon of an adjacent possible membrane transport protein STM3356 which correlates perfectly with the dT^– phenotype. This suggests that the protein may be essential for tartrate metabolism, although the function of this protein remains unknown. Within the sequence around the putative coding sequence (CDS) STM3356 there are several possible start codons, and in the annotation for serovar Typhi CT18 (STY3536) one such alternative is chosen. The biological significance of the single-base-pair difference in this CDS has yet to be determined. Given that serovar Paratyphi B isolates from the MLEE group Pb1 are more often associated with gastroenteritis, even though the group contains strains that cause systemic disease, it seems likely that the ability to use tartrate has phylogenetic but not necessarily biological relevance. It is probable that there are invasive and noninvasive variants within the SPB Pb1 clonal group of salmonellae. The analysis of very closely related strains from a single MLEE group (Pb1) of S. enterica that have totally different disease characteristics in the human host, such as the invasive and noninvasive strains of serovar Paratyphi B, may greatly facilitate the identification of a set of genes required for invasion across the human gut, invasion of the deeper tissues, and/or systemic dissemination and survival. So far, comparisons of these two sets of isolates have revealed no differences in siderophores, serum resistance, or surface structures (15), but invasive strains do possess a bacteriophage-encoded sopE, lack avrA, and have normal levels of SopB, but not SopD, production (81). Within the SPB serovar, strains adapted to poultry have been described (104), and there is extensive literature on infection in humans. This heterogeneous group of organisms should provide a fruitful area for the investigation of host adaptation in salmonellae.

SEROVAR PARATYPHI C

By using biochemical tests, S. enterica isolates with the antigenic formula 6,7:c:1,5 can be subdivided into several different subtypes: serovar Paratyphi C, serovar Typhisuis, serovar Choleraesuis, serovar Choleraesuis var. Kunzendorf, and serovar Choleraesuis var. Decatur (102). Serovar Paratyphi C is probably restricted to humans, but difficulties in strain identification make this assertion unsure.

Serovar Typhisuis is a typical host-restricted serovar that is rarely isolated from humans. Infection in pigs, the natural host, ranges from enterocolitis (86) to chronic paratyphoid fever (101). Serovar Typhisuis is auxotrophic and forms a monophyletic group by MLEE (9, 91). Differentiation from serovar Paratyphi C is based on the prototrophic nature of the latter. The use of fermentation tests for utilization of arabinose and trehalose have been described (102). By using MLEE, two distinct groups of serovar Paratyphi C have been described, one of which is closely related to serovar Typhisuis and the other to serovar Miami (91) or serovar Montevideo (9). With serovar Paratyphi C strains SARB48 and SARB49, it appears that the two MLEE groups can also be differentiated by ribotyping (102), suggesting that they represent distinct phylogenetic groups. However, recent MLST data reveal that these two strains are in fact clonally related (S. Nair, personal communication). Extensive analyses of serovar Paratyphi C have been carried out with PFGE (54) and ribotyping (102), but an association between strain type and difference in disease has not been described. Sequencing of a serovar Paratyphi C isolate is underway (Peking University Health Science Center, Beijing, China).

The third member of this serovar, serovar Choleraesuis, is a more diverse group of organisms containing three biotypes and five MLEE groups (9, 91). Serovar Choleraesuis is a host-adapted, but not restricted, serovar that causes swine paratyphoid fever (19). When contracted by humans, it causes a highly invasive disease with little involvement of the gastrointestinal tract (23) and infection is often associated with underlying diseases in the patient (19, 21, 110). There are reports in Taiwan, the United States, and England and Wales of invasive indexes ranging from 52% to 74% (18, 62, 97, 106). The genome sequencing of serovar Choleraesuis strain SC-B67 and its two plasmids (pSCV50 and pSC138) has been published (20). Serovar Choleraesuis SC-B67 contains 151 pseudogenes, of which 17 are shared with serovar Typhi CT18. Two pseudogenes, SCPS112 (shared with serovar Typhi CT18) and SCPS30 are derived from genes that may partly explain why diarrhea is more rarely a symptom of serovar Choleraesuis. SCPS112 is annotated as shdA in serovar Typhimurium and encodes a Peyer’s path colonization and shedding factor, while SCPS30 encodes an invasin C (Yersinia) homologue, which may cause adhesion to intestinal epithelial cells to be impaired.

NONTYPHOIDAL SALMONELLOSIS

Improvements in diagnosis (16), the influence of HIV (17, 38), and/or the appearance of more virulent strains (42) may be responsible for the growing number of NTS serovars reported as causing invasive disease syndromes in humans (Fig. 1). Serovars of NTS are widespread and are commonly associated with specific animals. They typically cause a self-limiting gastroenteritis in the human host, the symptoms of which include fever, diarrhea, vomiting, and stomach cramps. The illness can be accompanied by prolonged fecal shedding (>1 month). NTS gastroenteritis carries a significant burden of mortality, largely because of the association with extraintestinal (EI) infections (106). EI NTS infections can present as gastroenteritis with secondary bacteremia or as a primary bacteremia or focal infection with no history of gastroenteritis (50). EI infection carries a significant mortality that is highest for primary bacteremia and focal infections (94). A variety of focal infections has been documented, including meningitis, septic arthritis, osteomyelitis, and infectious endarteritis (53). In the otherwise healthy, immunocompetent host EI NTS infections are relatively rare; however, certain groups of individuals, under 2 and over 65 years (94, 111), are at increased risk. Age-related differences in the type of EI infection also exist; children are more likely to develop secondary bacteremia and meningitis, whereas adults are more at risk of a primary bacteremia or focal infection, including endarteritis (4, 52). Focal infections have been linked to a variety of intrinsic and iatrogenic host factors that affect immune status, e.g., diabetes and malignancy, HIV infection, and receiving therapeutic immunosuppression (11, 51, 52). Several hypotheses have been proposed for the association between immunodeficiency and EI NTS infection, including reactivation of latent infection or hemolytic disease (115) or malaria-associated anemia (41, 43). The high risk of death in the elderly is most likely due to an increased prevalence of atherosclerotic plaques and aneurysms in this age group, which increases the risk of infectious endarteritis. The mortality rate for EI NTS infection in HIV-positive individuals is high, with recurrent infection occurring in a significant proportion of the survivors (38). An association with helminth infections suggests gut damage may predispose to invasive NTS (28). Enteric fever has a lower mortality than systemic NTS infections (39) and, although HIV-positive patients may be at a greater risk of acquiring typhoid fever (40), few cases of co-infection (66) have been reported and the expected increase in cases of typhoid fever has not taken place in regions where coinfection with HIV could occur (78).

SEROVAR TYPHIMURIUM AND SEROVAR ENTERITIDIS

In both developed and less developed countries, serovar Enteritidis (in particular, Phage Type 4) and serovar Typhimurium infection are the most commonly reported cause of EI NTS (11, 55). These systemic infections are most frequently seen in individuals with predisposing factors (11, 24, 43, 88, 106, 109), but bacterial factors including plasmid-borne multidrug resistance (MDR) to three or more antimicrobial agents (55) and chromosomally mediated quinolone resistance (48) are also involved. Many NTS serovars, including serovar Typhimurium and serovar Enteritidis, carry a plasmid not usually associated with resistance, the virulence plasmid, which encodes the spv genes that are essential for infection in laboratory rodents. The role of the virulence plasmid in gastroenteritis and invasive disease in humans is unclear. Some reports suggest that the spv genes promote dissemination of serovar Typhimurium from the gut (34), although in vitro experiments to date have not demonstrated a role for these genes in invasion of epithelial cells (12). Harboring a virulence plasmid may facilitate systemic virulence in some S. enterica subspecies I serovars, but the virulence level of the invasive disease will almost certainly vary with the bacterial strain and the host involved. The study of spv-containing pathogens needs to be complemented with epidemiological studies to determine whether virulence plasmids are important in the pathogenesis of NTS infections in humans (34).

SEROVAR DUBLIN

The natural host for serovar Dublin is cattle. Infections in humans are not very common, although when reported, the invasive index is typically high. A study undertaken in Brazil showed that in a collection of 3,554 isolates over 8 years, 51 invasive cases of 86 culture-confirmed cases occurred (65%) (33). The most recently published data from the United States (106) indicates that serovar Dublin had an average invasive index of 71% from 1996 to 1999, although the low prevalence of the serovar precludes conclusions on temporal trends. Reports in Europe vary; serovar Dublin has been reported as causing invasive cases in England and Wales and Belgium but does not seem to be a problem in Spain, where no cases of serovar Dublin were reported among the 970 Salmonella isolates from humans collected between 1991 and 1996 (87). A similar pattern is observed in New Zealand where only one (noninvasive) case of serovar Dublin has been reported from the Public Health Surveillance Enteric Reference Laboratory in the past 5 years, which is perhaps unsurprising given that no bovine isolates of the serovar were isolated over the same period. However, serovar Dublin is commonly isolated from Australian cattle, yet no human cases were reported between 1995 and 2003, which raises interesting questions about interspecies transmission of this serovar. There are few, if any reports of invasive cases of serovar Dublin being reported from Africa and Asia. Whether this patchy distribution of human cases is because of differences in reporting of cases, human factors, food animals, or variation in the bacterial population has not been studied in depth. Variation in the bacterial population has been demonstrated; by using MLEE, serovar Dublin can be divided into three clonal types that are very closely related to serovar Enteritidis (92). Though both serovars are similar at the genetic level, serovar Dublin is largely host adapted for cattle and serovar Enteritidis has a wider range of hosts, the strongest association being with poultry and humans. Geographical differences also occur; type Du1 is globally distributed, whereas Du3 is restricted to France and the United Kingdom and contains all the capsulate strains (Vi positive) (92). The epidemiological data linking bacterial variation and human disease is limited but there is an association between ribotype and the ability of serovar Dublin to cause human invasive disease (22), and with strain type when using a combination of plasmid profiles, ribotyping, and PFGE (64). The data taken together suggest that some serovar Dublin strains may be more infectious for humans. Which genetic factors may be responsible for this has also been investigated. Vi-positive strains contain the SPI-7 type IV pili, used by Typhi for self-association (79), and disruption of this system reduces invasion across human cultured intestinal cells (71). Serovar Dublin also carries an spv-containing plasmid that promotes enhanced intracellular proliferation in intestinal tissues and at extraintestinal sites in the natural host, cattle (63); however, no association has been found with disease in humans (22, 74).

Serovar Dublin therefore seems to be a polyphyletic serovar with two main clonal groups. One of these groups contains strains that are Vi positive and may be associated with the ability to cause invasive disease in humans.

SEROVAR VIRCHOW

There have been numerous single-case reports of invasive serovar Virchow infections (6, 7, 98); this serovar is recognized as a significant cause of invasive salmonellosis in Israel (94, 111), Australia (1, 5), New Zealand (Public Health Surveillance), and the United Kingdom (98, 112). The increasing incidence of serovar Virchow PT26 is of particular concern because of its association with more invasive disease in humans (95, 113). In contrast to the United Kingdom and Australia, there have been no reports of invasive serovar Virchow infections from North America (57, 106) and this serovar was not mentioned among the Public Health Laboratory Information System’s (PHLIS) top twenty most frequently reported Salmonella from humans (PHLIS surveillance summaries 1995 to 2005). Little is known about the population structure of this serovar, but if it appears in the animal population, it seems to cause invasive disease in humans.

OTHER SEROVARS ASSOCIATED WITH HUMAN INVASIVE DISEASE

Serovars Panama and Oranienburg have similar invasive indexes of ~12.5% in New Zealand (Enteric Reference Laboratory) while serovar Panama infections in Brazil and England and Wales are less invasive at just under 3%. Serovar Schwarzengrund was reported to be invasive in Taiwan (62) with 15 invasive cases of 60 culture-confirmed cases (25%) occurring between 1998 and 2002. Invasive cases have also been reported in Brazil and the United States (33, 106).

Bacteremia due to serovar Heidelberg has been reported from the United States (97, 106) and Australia (NEPSS 2001 to 2003). It can be associated with eggs (49), and in the United States the invasive index of 11% was higher than for serovar Typhimurium (6%) and serovar Enteritidis (6%) (106). However, the high blood invasion index for serovar Heidelberg is not consistent in different countries: 3.3% in the United Kingdom, 1% from Scotland, and 5.8% from Australia. Limited data on the population structure of this serovar give conflicting views on whether it forms a monophyletic group (9, 80).

In Crete, a study looking at Salmonella infection in immunocompetent children identified seven cases of serovar Newport of which one was invasive (16.7%) (36). This is consistent with a Taiwanese study where 8 invasive cases of 46 culture-confirmed cases were reported (17.4%) (62). However, contradictory reports from Brazil (33), New Zealand (Public Health Surveillance), the United States (106), and Belgium (Reference Lab) all document a lower invasive index, from 1.5% to 4.3%. Comparisons of the genetic content of five serovar Newports suggest that this serovar does form a monophyletic group, so the discrepancy in invasive index may be due to local outbreaks of invasive strains (80).

A number of serovars appear to be invasive in Brazil (33), including serovars Agona, Oranienburg, Saintpaul, Hadar, Javiana, and Mbandaka. Unusually, the majority of Salmonella infections appear to occur in the age range 20-59, although the distribution of invasive infection by age range is not given.

CONCLUSIONS

For serovar Typhi and serovar Paratyphi A there is a clear association between serovar and systemic infection in humans. Comparisons made between sequenced strains of serovar Typhi and serovar Paratyphi A and microarray data investigating regions of intraserovar variation for serovar Typhi (8) and serovar Paratyphi A (67) have begun to shed light on genetic factors that are important in the ability of these related but distinct serovars restricted to humans to cause systemic disease. The acquisition of islands of DNA or single genes may explain, in part, why these two serovars are adapted to humans, but the presence of an overlapping, but not identical, set of pseudogenes caused by different mutational lesions suggests convergent evolution and genome degradation as part of adaptation to a common niche (67). Which of the many pseudogenes or regions of acquired sequence have contributed to the host restriction of both serovar Typhi and serovar Paratyphi A and/or to their ability to cause systemic disease in the human host remains to be discovered. For serovars Paratyphi B and Paratyphi C, a good clinical description of the host and detailed population genetics of the pathogen are necessary before more detailed genetic studies of novel virulence factors can be initiated.

Serovar Typhimurium and serovar Enteritidis are the most common serovars within the food chain and, thus, the high number of invasive infections associated with these serovars is most likely due to exposure rather than to increased virulence of the pathogen. In Africa, however, a closely related group of strains of serovar Typhimurium may have become host adapted to humans (42). Some of the NTS associated with human invasive disease are not single clonal groups (101) and, again, a more detailed molecular analysis of the pathogen will be necessary before we can effectively search for bacterial virulence factors. The similarities between serovars such as serovar Typhi and serovar Typhimurium suggest that they share some mechanisms for invasion and intracellular trafficking (70); however, it is also clear that salmonellae exhibit diversity in their mechanism of adaptation even to a single host, human beings (101).