Role of fimbriae as antigens and intestinal colonization factors of Salmonella serovars

The two most important human disease syndromes caused by Salmonella serovars are typhoid fever and enterocolitis. Typhoid fever is a systemic infection caused by Salmonella serovars which are strictly adapted to humans or higher primates, including Salmonella enterica serovars Typhi and paratyphi A, B, and C. Human enterocolitis is an infection localized to the intestine and mesenteric lymph nodes which can be caused by any of more than 2000 Salmonella serovars, however, S. enterica serovars Typhimurium and Enteritidis together account for approximately 50% of cases in the USA. The animal model most frequently used to study Salmonella pathogenesis is the typhoid fever-like disease caused by serovar Typhimurium in mice. The pathology in the intestine of mice infected with serovar Typhimurium is characterized by a predominantly mononuclear leukocyte infiltrate, as well as edematous villi which become shortened in height [1]. Capillary thrombosis, hemorrhage, and ulcerations may be present at areas of Peyer's patches. The intestinal pathology caused by serovar Typhimurium in mice is similar to that observed in typhoid fever patients. That is, intestinal biopsies taken from volunteers infected with serovar Typhi or from typhoid fever patients reveal a predominantly mononuclear leukocyte infiltrate [2,3]. Furthermore, colonization of human Peyer's patches by serovar Typhi commonly produces capillary thrombosis, which can result in hemorrhage, necrosis and ulceration [4]. In humans, these lesions can progress to intestinal perforation, a potentially fatal complication which usually occurs in the third week of infection [4]. In contrast, rapid bacterial multiplication in the liver and other internal organs of mice causes death within 6–10 days post serovar Typhimurium infection, while intestinal perforation has not been reported in this model. Since intestinal perforation is an important cause of death during typhoid fever [4], it is likely that the intestinal phase of serovar Typhi infections is of greater importance than suggested by the mouse typhoid model. While serovar Typhimurium causes a typhoid fever-like disease in mice, infections of humans or cattle commonly manifest as enterocolitis. Recent studies have thus focused on serovar Typhimurium infection of calves as an animal model for human enterocolitis [5]. Calves develop a necrotizing enterocolitis characterized by a severe diffuse infiltration of polymorphonuclear leukocytes associated with necrosis of the upper mucosa and pseudomembrane formation [6]. Comparison of serovar Typhimurium mutant phenotypes in mice and calves illustrates that the intestinal phase of infection is of much greater importance during enterocolitis than suggested by data obtained using the mouse typhoid model. That is, mutations in Salmonella pathogenicity island (SPI) 1, a virulence factor required for intestinal invasion, result in the absence of diarrhea and avirulence during oral infection of calves but cause only modest attenuation (between 0- and 50-fold) in the mouse [6]. The importance of intestinal perforation during typhoid fever and the fact that during bovine enterocolitis the most severe pathological changes are restricted to the intestine illustrate that the intestinal phase of infection plays a critical role during disease caused by Salmonella serovars. Attachment is thought to be an important first step during intestinal colonization and has been studied extensively in non-invasive enteric pathogens, such as Escherichia coli. However, in Salmonella pathogenesis, this area of research has received little attention in the past, which may be largely because the extensive use of mice to study Salmonella pathogenesis has diverted attention from mechanisms of intestinal colonization to genes required for bacterial multiplication in the liver and other internal organs. Three pathways for the assembly of fimbrial adhesins have been described in enteric pathogens. These include the chaperone/usher-dependent assembly pathway, the nucleator-dependent assembly pathway and the assembly pathway for type IV fimbriae. Six chaperone/usher-dependent fimbrial operons are present in serovar Typhimurium, including fim (type 1 fimbriae), pef (plasmid-encoded fimbriae), lpf (long polar fimbriae), stf (S. enterica serovar Typhimurium fimbriae), bcf (bovine colonization factor) and saf (Salmonella fimbriae). In addition, the serovar Typhimurium genome contains an operon of the nucleator-dependent assembly pathway, termed agf (aggregative fimbriae). The agf operon and three chaperone/usher-dependent fimbrial operons, including fim, pef and sef (S. enterica serovar Enteritidis fimbriae), are present in S. enterica serovar Enteritidis (for a recent review and related references on fimbriae of Salmonella serovars see [7]). While the data accumulated by these studies indicate that genomes of Salmonella serovars contain a large number of fimbrial operons, identification of the complete repertoire of fimbrial biosynthesis genes became possible only with the advent of genome sequencing. The first genomic analysis of fimbrial gene sequences reveals that the serovar Typhi chromosome contains 14 putative fimbrial operons [8]. These include an operon encoding type IV fimbriae, the agf operon, and 12 chaperone/usher-dependent fimbrial operons (Fig. 1) [8]. Fimbrial gene sequences of the chaperone/usher-dependent assembly class present in the serovar Typhi strain CT18 chromosome. Fimbrial operons of serovar Typhi and their map locations have been reported previously [8]. Open reading frames encoded clockwise or counterclockwise are indicated by arrows which point to the right or left, respectively. Broken arrows indicate pseudogenes. Hybridization analysis suggests that the genomes of other Salmonella serovars contain many of the fimbrial gene sequences identified in serovars Typhi, Typhimurium or Enteritidis [8,9]. Thus, the presence of a large number of fimbrial gene sequences appears to be a common feature of the genomes of Salmonella serovars. However, Southern hybridization further indicates that the genome of each Salmonella serovar contains a different combination of fimbrial gene sequences [8]. It is currently an enigma, which selective pressures are responsible for generating and maintaining these different repertoires of attachment factors, thereby illustrating that our understanding of the genetic design of Salmonella serovars with regard to fimbrial gene sequences is rather limited. Fimbrial attachment may serve several functions during the intestinal phase of infection with Salmonella serovars. Fimbriae mediate attachment of serovar Typhimurium to various epithelial cell lines in vitro. Furthermore, this attachment is required for invasion of tissue culture cells in vitro [10–,12]. The role of attachment during invasion appears to be the selection of a target cell type for entry. Consistent with this idea, the type III secretion system encoded by SPI1 is required for invasion of a large number of cell lines, whereas each fimbrial operon appears to be only required for attachment to and invasion of an individual cell line [10]. Thus, fimbriae may allow serovar Typhimurium to distinguish between different epithelial cell types. The finding that mutations in lpf and pef reduce bacterial numbers associated with murine Peyer's patches and villous intestine, respectively, suggests that the binding specificity observed with cell lines may result in tissue tropism in the murine intestine [13,14]. Southern hybridization suggests that the tcf fimbrial operon is restricted to serovar Typhi and may thus be involved in its adaptation to the human host [15]. A more recent hybridization analysis performed with a larger number of Salmonella serovars indicates, however, that this operon is also present in a number of non-typhoidal serovars, thereby making it unlikely that a simple correlation exists between the host range and the presence of a single fimbrial operon [8]. Nonetheless, an involvement of fimbriae in host adaptation is suggested by a recent comparative signature-tagged mutant screen performed in mice and calves which identified bcf as a serovar Typhimurium fimbrial operon required for colonization of bovine but not murine Peyer's patches [16]. However, further studies are needed to firmly establish the role of fimbriae in determining host specificity of Salmonella serotypes. Studies using the mouse model of serovar Typhimurium infection reveal that inactivation of an individual fimbrial operon results in only modest (5-fold or less) attenuation. Given the large number of genetic adherence determinants (i.e. 14 fimbrial operons in serovar Typhi), it is possible that alternate attachment factors may compensate for the loss of an individual fimbrial operon, thereby resulting in only modest attenuation of mutants lacking a single adhesin. Consistent with this idea, simultaneous inactivation of the lpf, pef and agf operons reduces mouse virulence of serovar Typhimurium more strongly (30-fold attenuation) than inactivation of each individual operon alone [17]. The large number of partially redundant attachment factors present in the genomes of Salmonella serovars illustrates the need for further characterizing the role of fimbriae during pathogenesis. Furthermore, much work remains to be done to understand the role of fimbriae in the pathogenesis of enterocolitis. Attachment mediated by fimbriae may be beneficial during infection, but expression of these surface structures could at the same time be potentially harmful to the pathogen since fimbrial filaments, which are composed of approximately 1000 copies of a major subunit, likely represent good targets for the host immune response. Consistent with a role of fimbriae as antigens during an infection, seroconversion to SefA, PefA and FimA is observed during serovar Enteritidis infection in chickens [18–,,21] and to LpfA during serovar Typhimurium infection in mice [22]. Furthermore, a strong T cell response against SefA is elicited in mice upon exposure to serovar Enteritidis as indicated by delayed-type hypersensitivity reactions and stimulation of in vitro proliferation and cytokine release by T cells [23]. Adaptive immunity against serovar Typhimurium infection of mice involves both B cell-dependent and T cell-dependent effector mechanisms. For instance, B cell-deficient mice immunized with serovar Typhimurium are more susceptible to a challenge with virulent serovar Typhimurium than mice with intact B cells [24,25]. Similarly, depletion of T cells in serovar Typhimurium vaccinated mice increases their susceptibility to a challenge with virulent serovar Typhimurium [26]. Thus, the induction of B cell-dependent and T cell-dependent responses by exposure to fimbrial proteins during infection with Salmonella serovars suggests that exposure to these antigens elicits adaptive immunity. Transcription of three fimbrial operons of serovar Typhimurium, including lpf [27], pef [28], and fim [29], is regulated by phase variation, an oscillation between phase ON and phase OFF expression states, mediated by a random heritable switching mechanism. Whether expression of the other fimbrial operons present in Salmonella serovars is regulated by phase variation has not yet been investigated. Phase variation results in expression of fimbriae in only a fraction of bacterial cells present in a culture, a fact commonly ignored when comparing phenotypes produced by a culture of a fimbrial mutant with that of a culture of an isogenic wild-type using animal or tissue culture models. Furthermore, it is currently not known which fraction of cells in a culture carries fimbrial operons in the phase ON expression state, and whether non-fimbriated bacteria that may be present in the culture are avirulent. The immune response elicited against fimbrial antigens may be the reason why expression of fimbriae is commonly regulated by phase variation. However, it is not clear how fimbrial phase variation may benefit a Salmonella serovar during infection of a host with recent exposure to this pathogen because it is not able to evade immunity against constitutively expressed surface structures, such as the O-antigen repeat of its lipopolysaccharide (LPS). For instance, mice immunized with a serovar Enteritidis aroA vaccine are resistant against a subsequent challenge with virulent serovar Enteritidis despite fimbrial phase variation [30]. Immunization of mice with a serovar Enteritidis aroA vaccine, on the other hand, does not confer protection against a challenge with a virulent serovar Typhimurium strain [30]. The evasion of cross-immunity between serovars Enteritidis and Typhimurium is due to expression of different immunodominant O-antigens by these pathogens. Although O-antigen repeat units of serovars Enteritidis and Typhimurium share the trisaccharide backbone (mannose–rhamnose–galactose), which is termed the O12-antigen, they differ by immunodominant sugar residues branching from their mannose residues in the trisaccharide backbone. In case of serovar Enteritidis the branching sugar is tyvelose (O9-antigen), while serovar Typhimurium O-antigen has abequose attached as a side unit to mannose (O4-antigen). Indeed, cross-immunity between both serovars is observed when mice immunized with a serovar Enteritidis aroA vaccine are challenged with a virulent serovar Typhimurium mutant that is genetically engineered to express the O9-antigen instead of the O4-antigen [30]. These data provide convincing experimental evidence that O-antigen polymorphism is a mechanism to evade cross-immunity between serovars Enteritidis and Typhimurium against LPS. Recent studies suggest that phase variation of the lpf operon is a mechanism to evade cross-immunity between serovars Enteritidis and Typhimurium against LpfA [22,31]. The primary structures of LpfA from serovars Enteritidis and Typhimurium are highly conserved, and both proteins are serologically cross-reactive [31]. When mice immunized with a serovar Typhimurium aroA vaccine are subsequently challenged with a virulent serovar Enteritidis strain, a selection against lpf phase ON variants is observed during recovery of the challenge strain from organs or feces of infected animals [31]. These data suggest that previous exposure to serovar Typhimurium LpfA results in cross-immunity against serovar Enteritidis LpfA. However, no cross-protection is observed and similar total numbers of serovar Enteritidis are recovered during a challenge from fecal pellets and organs of immunized and naive mice [31]. These data suggest that during a challenge, serovar Enteritidis lpf phase OFF variants are able to evade cross-immunity generated by exposure to serovar Typhimurium LpfA and establish infection. On the contrary, cross-protection between serovars Typhimurium and Enteritidis is observed if mice immunized with a serovar Typhimurium aroA vaccine are challenged with a serovar Enteritidis strain which expresses the lpf operon constitutively [22]. Thus, serovar Enteritidis is not able to evade cross-immunity elicited by exposure to serovar Typhimurium LpfA once lpf fimbrial phase variation has been eliminated. These data directly support the hypothesis that phase variable expression of the lpf operon is a mechanism to evade cross-immunity between serovars Enteritidis and Typhimurium. The significance of evading cross-immunity is that this process may provide a significant selective advantage for serovars Enteritidis and Typhimurium during coexistence in their rodent animal reservoir. That is, the ability of a Salmonella serovar to persist and spread within an animal population is dependent on the density of susceptible hosts. The host population will be less susceptible if it was previously exposed to a serovar expressing similar antigen, because a fraction of animals will be removed from the susceptible population by cross-immunity. Cross-immunity will thus reduce transmissibility, thereby resulting in between-serovar competition if two serovars which share similar antigens circulate in the same host population. Serovars Enteritidis and Typhimurium evade this between-serovar competition during coexistence in their rodent animal reservoir since they do not elicit cross-immunity. Evasion of cross-immunity may represent one of the driving forces for the regulation of fimbrial expression by phase variation, and for maintaining O-antigen polymorphism among Salmonella serovars. For a more in depth discussion on the epidemiological consequences of cross-immunity, the reader is referred to a recent review article [32]. The picture emerging from these studies is that genomes of Salmonella serovars contain a large number of fimbrial operons, some or all of which may act synergistically during infection. Although there is no evidence to suggest that fimbrial operons are co-regulated, their transcriptional regulation appears to share a common feature, phase variation. Recent data show that fimbrial phase variation is a mechanism to evade cross-immunity between Salmonella serovars elicited by shared fimbrial antigens. A corollary of the notion that fimbrial phase variation is a mechanism for immune evasion is that bacteria lacking an individual fimbrial structure (phase OFF variants) can only evade a host immune response and establish infection if they retain considerable virulence because they express alternate attachment factors. Partial functional redundancy of fimbriae is thus predicted to be a prerequisite for effectively using phase variation as a mechanism for immune evasion. This hypothesis hence provides an attractive explanation as to why genomes of Salmonella serovars contain a large number of fimbrial operons that are regulated by phase variation and exhibit partial functional redundancy. The presence of a large number of fimbrial adhesins, which oscillate between phase ON and phase OFF expression states and act synergistically during infection, illustrates that attachment is a complex trait. Since fimbriae share functional (attachment) and regulatory (phase variation) features, we propose that fimbrial operons should be considered components of a complex virulence factor, the serovar specific fimbrial virulon. In light of this hypothesis, it appears logical to expand the conventional approach to study one individual fimbrial operon to an analysis which includes all fimbrial operons present in the genome of a serovar to investigate the role of its serovar specific fimbrial virulon. For instance, the phenotype caused by inactivation of a serovar specific fimbrial virulon is currently unknown although this step usually forms the beginning of analyzing complex virulence factors, such as type III or type IV secretion systems. It is not possible to inactivate a serovar specific fimbrial virulon by inactivating a single gene without causing a general defect in protein transport. Thus, to determine the importance of fimbrial attachment for Salmonella pathogenesis, the contribution of each individual fimbrial operon present in the genome needs to be considered. Work in A.B.'s laboratory is supported by the Texas Advanced Research (Technology) Program under Grant number 000089-0051-1999, Public Health Service Grant #AI40124 and Public Health Service Grant #AI44170.