Basic Processes in Salmonella-Host Interactions: Within-Host Evolution and the Transmission of the Virulent Genotype

ABSTRACT

Transmission and virulence are central aspects of pathogen evolution. However, in many cases their interconnection has proven difficult to assess by experimentation. Here we discuss recent advances from a mouse model for Salmonella diarrhea. Mouse models mimic the enhanced susceptibility of antibiotic-treated individuals to nontyphoidal salmonellosis. In streptomycin-pretreated mice, Salmonella enterica subspecies 1 serovar Typhimurium efficiently colonizes the gut lumen and elicits pronounced enteropathy. In the host’s gut, S. Typhimurium forms two subpopulations that cooperate to elicit disease and optimize transmission. The disease-causing subpopulation expresses a set of dedicated virulence factors (the type 3 secretion system 1 [TTSS-1]) that drive gut tissue invasion. The virulence factor expression is “costly” by retarding the growth rate and exposing the pathogen to innate immune defenses within the gut tissue. These costs are compensated by the gut inflammation (a “public good”) that is induced by the invading subpopulation. The inflamed gut lumen fuels S. Typhimurium growth, in particular that of the TTSS-1 “off” subpopulation. The latter grows up to very high densities and promotes transmission. Thus, both phenotypes cooperate to elicit disease and ensure transmission. This system has provided an experimental framework for studying within-host evolution of pathogen virulence, how cooperative virulence is stabilized, and how environmental changes (e.g., antibiotic therapy) affect the transmission of the virulent genotype.

INTRODUCTION

In line with Koch’s postulates, studying virulence typically translates into identifying and characterizing the molecular determinants that underlie colonization of a host by a pathogen and the subsequent appearance of symptoms (1). In this conceptual framework, the presence of pathogens implies damage to the host whose intensity is proportional to the virulence of the pathogen. This approach is based on the observation that damage is often related to the expression of specific features of the pathogen, i.e., the virulence factors (2).

It has also become clear that the role of the host in defining virulence of pathogens is absolutely central. The degree of sensitivity of the host is thought to determine (to a large extent) whether pathogenic organisms are effectively virulent or if they remain innocuous upon colonization. Indeed, such innocuous behavior can often be observed in cases of virulence factor-studded pathogens (3) (for examples, see Table 1). One should therefore consider virulence as a probabilistic property of organisms in given hosts and virulence factors as genetic determinants that increase the chance to cause damage. This also means that transmission of virulence does not necessarily rely on triggering symptoms. This is of importance for the vast majority of pathogens and their evolution.

TABLE 1.

Asymptomatic carriage of bona fide pathogenic bacteria in humans

Pathogen	Asymptomatic carriagea	Disease	Disease incidence	Reference(s)
Mycobacterium tuberculosis	10%b	Tuberculosis	10–>500 per 100,000b	4
Nontyphoidal Salmonella spp.	≈1%	Diarrhea	∼400 per 100,000	5
Neisseria meningitidis	5–10%	Meningitis	0.3–0.5 per 100,000	6, 7
Chlamydia trachomatis	3.8%c	Sexually transmitted diseases	426 per 100,000	8
Staphylococcus aureus	≈30%	Bacteremiad	10–30 per 100,000e	9, 10

^a In the entire “healthy” population.

^b TB incidence is particularly high in developing countries; here, up to 80% of the population can test positive in the tuberculin test.

^c In men aged 18 to 26.

^d S. aureus can also cause numerous other diseases.

^e S. aureus bacteremia.

The dual nature of virulence, depending on one side on a pathogen that may express or encode but not express (or successfully deploy) virulence factors, and on the other side on the sensitivity of the host to the presence of the pathogen, implies that its evolutionary dynamics is particularly challenging to generalize. Nevertheless, understanding virulence beyond its purely mechanistic aspects should not be overlooked. Determining why and in which circumstances the virulence of a pathogen can be an adaptive trait should ultimately allow us to predict and to control its evolution.

In order to achieve such a goal, one should ideally consider the complete life cycles of pathogens, including within-host growth, survival in the environment, and modes of transmission to new hosts. The impact of virulence on the reproductive success of pathogens, meaning their ability to repeat their life cycles host after host, should be precisely addressed (11). In this respect, adaptive virulence is synonymous with maintenance and efficient transmission of the virulent genotype from one host to the other.

Experimental attempts to tackle this problem remain scarce for multiple reasons. Considering, for instance, pathogens infecting humans, animal models do not always exist, and when they do, they do not necessarily allow reconstitution of all the infection steps. Moreover, the full life cycle of a pathogen can be complex and often includes phases in various environments or in more than one type of host. And lastly, our knowledge of pathogen life cycles is most of the time incomplete and strong selective pressures in favor of or against virulence could remain concealed.

In this chapter, we discuss recent advances in addressing the adaptive value of virulence of the archetypal enteropathogenic bacterium Salmonella enterica serovar Typhimurium, a pathogen well characterized in laboratory model hosts such as mice. Rodents are natural hosts for S. Typhimurium and tractable surrogates in which the progression of the disease mimics pathogenesis in humans (12). Please note that most of experimental results described here were obtained by making use of the antibiotic-pretreated mouse model. The antibiotic pretreatment opens the intestinal niche for S. Typhimurium, which is otherwise occupied by the host’s microbiota. This allows very reproducible colonization but comes at the price of abrogating the initial competition between Salmonella and the microbiota, which certainly plays a role in the evolution of the pathogen.

S. Typhimurium infects hosts via the fecal-oral route by the uptake of contaminated food or water. In immunocompetent humans, S. Typhimurium causes a self-limiting diarrhea that lasts for 5 to 7 days. The diarrhea is accompanied by inflammation that is thought to provide an advantage to Salmonella over the protective microbiota (13). However, within-host growth seems to have a negative impact on the evolution of virulence (14). This paradox will be addressed in this chapter. We will also discuss processes that may help to maintain virulence.

Evidence for direct transmission from person to person is rare, and this route of transmission seems restricted to situations of intense contact, such as between hospitalized patients and staff members (15). Patients can asymptomatically carry nontyphoidal Salmonella (NTS) for years with sporadic relapses, which could favor spreading of Salmonella in the environment and subsequent transmission (16). As described below, Salmonella spp. are facultative intracellular pathogens that are able to invade the host tissues, where they could remain dormant, making the disease chronic. However, in immunocompromised hosts, such as HIV-infected patients, unrestricted intracellular bacteria can spread and cause life-threatening bacteremia requiring antibiotic treatment (17).

The reservoirs for NTS are mainly domesticated birds, swine, and cattle, but also wild rodents and plants (18). S. Typhimurium is therefore able to colonize a wide variety of hosts and environments in which functions such as virulence factors could have dissimilar adaptive values (19). Consequently, the regulation of gene expression by environmental cues is key to ensure the maintenance of virulence (20). Due to the broad range of hosts that can serve as reservoirs, utilization of “uniform” environmental cues that are present in equivalent sites within the different hosts would seem essential for regulation. This point will be discussed in more detail later below.

SALMONELLA IN THE GUT LUMEN: WITHIN-HOST SELECTION FOR ATTENUATED DEFECTORS

The Competition for the Gut

To initiate host colonization right after ingestion, S. Typhimurium must survive the acidic pH of the stomach. This requires the acid response, while virulence factors are not needed. Then the bacteria are faced with a high density of physical and chemical host defenses and a very dense microbiota population, which deplete nutrients, produce inhibitors, and occupy mucosal binding sites, thereby effectively prohibiting S. Typhimurium growth, a phenomenon called colonization resistance (CR).

We recently discovered how S. Typhimurium can profit from metabolic intermediates of the microbiota metabolism (i.e., molecular H₂ released by microbiota fermentation) to fuel the initial growth in this niche (hyb H₂ hydrogenase and fumarate reductase are two key gut colonization factors [21, 22]). The archetypical symptoms that seem to promote the selection for virulence of this pathogen are only triggered once the bacteria have reached sufficiently high densities in the gut lumen (∼10⁸ CFU per g of stool) (23). Beyond this threshold, pathogen invasion into intestinal epithelial cells is frequent enough to elicit the acute mucosal innate immune response (i.e., gut inflammation), which fosters the growth of S. Typhimurium (13) (Fig. 1).

FIGURE 1.

The different steps of the gut infection cycle by Salmonella Typhimurium. Red arrows depict the potential for S. Typhimurium (S. Tm) transmission to the next host during each step. Blue- and red-colored cells depict healthy and inflamed guts, respectively. PMN, polymorphonuclear neutrophils; DC, dendritic cells; MΦ, macrophages; IgA/G, immunoglobulin A/G produced as part of the host’s adaptive immune response 2 weeks postinfection—this follows the regrowth of Salmonella (2nd bloom) after a population bottleneck inflicted by the innate immune response (at day 2 postinfection); IL-18, interleukin-18; Casp-1, caspase-1.

Work from the past few years has provided initial insights into why S. Typhimurium growth is enhanced in the lumen of the inflamed gut, supporting the view that this pathogen constructs its niche by triggering the host innate immune defenses. The gut inflammation represents a major ecological perturbation (for a review, see reference 24): antimicrobial proteins that preferentially inhibit the growth of Salmonella competitors are secreted (e.g., RegIIIb and lipocalin-2); high-energy nutrients sensed by the chemotactic receptors of Salmonella are released (e.g., galactose-rich glycoconjugates [25]); and an alternative electron acceptor (tetrathionate) is formed from endogenous thiosulfate by reactive oxygen species mainly generated by neutrophils massively recruited in the gut lumen (26). S. Typhimurium growth is further fueled by ethanolamine that is massively released into the lumen of the inflamed gut (27).

The link between gut inflammation and efficient gut luminal pathogen growth is quite striking. In the absence of inflammation, S. Typhimurium fails to sustain the colonization of conventional mice. This was demonstrated by making use of antibiotic-pretreated mice in which the intestinal microbiota and CR are transiently disrupted. In these conditions, S. Typhimurium strains resistant to the antibiotic reach in a few hours a population size of 10⁹ CFU/g of stool. Thanks to gut inflammation, virulent wild-type strains can maintain themselves for 2 to 3 weeks, provided that the mouse does not die from systemic infection (e.g., in resistant Nramp^+/+ mice) (28, 29). In contrast, avirulent strains unable to elicit inflammation are excluded from the intestine by the regrowing microbiota already by 3 to 4 days post-antibiotic treatment (13). Moreover, mice carrying a simplified microbiota (e.g., the altered Schaedler flora) can durably shed high loads of avirulent strains of Salmonella (30), and introducing additional microbiota strains can fortify CR in such models (31). This clearly demonstrates the protective role of the intestinal microbiota against enteropathogens as well as the adaptive value of enteric virulence.

Of note, the pronounced inflammation triggered by fully virulent S. Typhimurium strains leads to a strong mucosal defense that transiently reduced the pathogen load in the gut lumen from 10⁹ CFU/g at day 1 to <10⁴ CFU/g of stool at day 2 postinfection (32). Attenuated strains do not provoke such a stark decimation. The underlying mechanisms will be an interesting topic for future work, and this shows that gut inflammation has two sides: it benefits Salmonella over all but remains costly (Fig. 1).

The niche construction strategy could be a general property of enteropathogenic bacteria, maximizing their transmission. Besides S. Typhimurium, Citrobacter rodentium is a second experimentally validated case of an enteropathogen able to profit from inflammation, which provides evidence of convergent evolution. C. rodentium was discovered in the 1960s after outbreaks in laboratory mouse colonies in the United States and Japan (33, 34). This pathogen is a model for the study of human pathogens that induce intestinal attaching-and-effacing lesions, e.g., enteropathogenic (EPEC) and enterohemorrhagic Escherichia coli (EHEC) (35). The major virulence determinants of C. rodentium are located at the locus of enterocyte effacement (LEE), which is conserved in EPEC and EHEC and encodes a type 3 secretion system (TTSS) and secreted effectors. Effectors are injected via the TTSS into epithelial cells, and provoke the formation of an actin pedestal on the apical side of the host cell where the bacteria are attached. The infection by C. rodentium is self-limiting, as bacteria stop expressing ler, the main positive regulator of LEE expression, after a few days (36). Inflammation then resolves, and the regrowing microbiota eventually excludes C. rodentium.

Similarities with S. Typhimurium are striking also with regard to the molecular mechanisms that underlie the triggering and exploitation of gut inflammation to compete against the protective host microbiota. Just like C. rodentium, S. Typhimurium expresses a TTSS, the Salmonella pathogenicity island 1 (SPI-1)-encoded TTSS-1, and injects proinflammatory effector proteins into the cytosol of enterocytes. This provokes cytoskeletal rearrangements and allows the entry of Salmonella into these nonphagocytic cells (37–41). From the enterocytes, Salmonella can translocate into the lamina propria, get phagocytized by monocytes and possibly also infect other cell types (42), and reach systemic organs (lymph nodes, spleen, and liver) (40). The intracellular niche plays an important role in transmission of Salmonella, which will be discussed in the sections below.

At the end of the acute phase of the disease, the host’s adaptive immune system mounts a protective secretory immunoglobulin A (sIgA) antibody response. After 10 to 15 days, the antibodies can thereby promote remission and the regrowth of a normal microbiota (29). Thus, the pathogen elimination from the gut at the end of an infection is attributable to the tipping of a delicate balance between the pathogen’s virulence, the microbiota, and the host response (Fig. 1).

Microevolution of S. Typhimurium in the Gut

In a favorable intestinal niche, Enterobacteriaceae such as S. Typhimurium can grow quickly (25-min to 2-h doubling time) (43) and maintain high densities (10⁹ bacteria/g of intestinal content) for several days. This is particularly true in the antibiotic-pretreated mouse model but also in conventional mice provided that S. Typhimurium manages to trigger intestinal inflammation (13). Consequently, within-host microevolution of the pathogen occurs at observable rates, allowing us to experimentally address dynamics of horizontal gene transfer (HGT) and mutation accumulation and their impact on the transmission of virulence.

Population density, inflammation, sIgA, and HGT

Horizontal transfer of mobile genetic elements is a major driving force of virulence evolution (44). The murine model of S. Typhimurium infection provides a potent model to study the dynamics of HGT in vivo.

It is demonstrated that the conjugative plasmid pCol1B9 is efficiently transmitted from S. Typhimurium SL1344 to strains of E. coli that are able to bloom in the inflamed gut (45). The conjugative transfer of the derepressed pCol1B9 being only contact dependent, the role of the inflammation is nevertheless indirect. Indeed, in antibiotic-pretreated mice, avirulent strains of S. Typhimurium are equally able to exchange pCol1B9 in absence of gut inflammation (46), the limiting factor being the population densities of donor and recipient strains.

On the other hand, the inflammation directly stimulates temperate bacteriophage transfer (lysogenic conversion) between strains of S. Typhimurium in the gut (46). Several lines of evidence suggest that stresses encountered by S. Typhimurium SL1344 within the inflamed gut trigger the SOS response, which controls the expression of the lytic cycle genes of the SopEΦ prophage. Thereby, the host’s innate immune defense increases the amounts of free virions in the gut and the acquisition of the bacteriophage by a coinfecting recipient strain (S. Typhimurium 14028S) (46).

We were able to limit both conjugation and lysogenic conversion by vaccinating the mice against S. Typhimurium (43, 46). Mice treated with dead S. Typhimurium 4 weeks before infection can be highly colonized by virulent strains while not showing any sign of intestinal inflammation (47). The presence of specific sIgA in the gut lumen enchains the rapidly dividing Salmonella, which therefore form monoclonal clumps (43). This enchained growth prevents triggering of inflammation and thereby the transfer of SopEΦ. Moreover, the monoclonal nature of the S. Typhimurium clumps renders contact-dependent conjugative transfer less frequent (43).

The rise of defectors

Among key features that make Salmonella such a successful pathogen are the TTSS-1 and effector proteins that are injected into the cytosol of host cells, where they manipulate host-cellular responses and eventually trigger gut inflammation (12, 48). As described above, inflammation is a shared resource that helps S. Typhimurium to outcompete the microbiota and to maximize the transmission to new hosts via the fecal-oral route (13) (Fig. 2). On the other hand, the expression of TTSS-1 and the induction of inflammation represent a significant fitness burden for Salmonella, namely, (i) strong growth retardation (14, 49) and (ii) killing by the innate immune system when the bacteria invade the host tissues (50). Bacteria that express TTSS-1 are growing only half as fast as S. Typhimurium cells that do not express it (14, 49), and 90% of the bacteria that reach the intracellular niche are eliminated by the host innate immune response (50) (Fig. 2B). Thus, the elicitation of gut inflammation (a “public good”) is associated with high costs to the TTSS-1-expressing bacteria. We analyzed if such a strategy is sustainable in the long run, as the costly production of public goods could be unstable. Indeed, we found that within-host evolution of S. Typhimurium results in the evolution of avirulent mutants (Fig. 2C). In other words, the virulence of S. Typhimurium is a cooperative trait that is prone to the selection of defectors, i.e., clones that profit from a public good without contributing to its production (14, 51). In this case, defectors are mutants with a defect in a central positive regulator (hilD) of ttss-1 expression. Thus, the defectors have lost the ability to express TTSS-1 and therefore profit from the inflammation without paying any fitness cost (growth defect or intracellular killing). Within-host competition drives the evolution of S. Typhimurium, and therefore defectors increase in frequency during long-term infections. When the frequency of defectors becomes too high, the inflammation cannot be sustained and the whole Salmonella population drops (presumably by microbiota regrowth), thus further accelerating the recovery of the host. The rise of the defectors turns out to be detrimental to the long-term colonization of the host.

FIGURE 2.

The division of labor and the rise of defectors. (A) (Left) Bimodal expression of ttss-1. The population of S. Typhimurium is divided into cells that express ttss-1 and cells that do not. (Right) Microscopy picture showing microcolonies on an agar pad of slow-growing ttss-1 “on” cells (expressing green fluorescent protein [GFP] under the control of PsicA, the promoter controlling the SPI-1 operon sicAsipBCDA) and fast-growing “off” cells. Reproduced with permission from reference 49. (B) The ttss-1 “on” cells enter into the mucosa and trigger inflammation. Most of these cells are killed by the mucosal innate immune response. Moreover, ttss-1 expression correlates with a substantial growth retardation. The ttss-1 “off” cells grow quickly in the lumen, ensuring the transmission of the virulent genotype. The inflammation is a public good shared among all cells in the lumen. (C) Colony blot obtained and described in reference 14. Within-host evolution of S. Typhimurium leads to the rise of avirulent mutants (defectors), which are clones that do not express ttss-1. The frequency of defectors was increasing between day 2 and day 10 postinfection (p.i.).

All isolated defectors carried point mutations in the hilD gene, which encodes the main positive regulator of TTSS-1 expression. When coinoculated with an isogenic virulent wild-type clone of S. Typhimurium, a synthetic hilD deletion mutant wins the competition and protects the host from full-blown inflammation. We hypothesized hilD mutants to grow faster than wild-type strains, as up to 40% of the wild-type population experience the growth defect associated with virulence expression (14). However, the HilD regulon comprises numerous functions (52), and alternative explanations, such as a reduced death rate or a better-adapted metabolism, are still open to investigation. In any case, defectors could potentially be used as probiotics in order to competitively exclude virulent strains of Salmonella. This could be a good alternative to less and less efficient antibiotics against NTS (53, 54).

The selection for attenuated clones carrying mutations in transcriptional regulators of virulence, as observed in laboratory mice, was recently reported in patients persistently infected by NTS (16). Mutations were discovered, for instance, in hilD and barA, both positively regulating the expression of TTSS-1. This suggests that the fitness costs associated with virulence expression are a constraint on its evolution in a wide range of ecological contexts (i.e., in vitro, in mice, and in human hosts).

Phenotypic Variability as an Adaptive Trait: the Division of Labor Theory

In spite of the inherent instability of S. Typhimurium virulence, most patient isolates express TTSS-1. This raises the question of how defectors are kept at bay in nature. Intriguingly, in isogenic populations of S. Typhimurium and upon homogeneous inducing conditions, not all bacterial cells express TTSS-1 (14, 50). The expression of Salmonella virulence is tightly regulated and ttss-1 is expressed in a bimodal fashion (Fig. 2). Yet poorly understood regulatory mechanisms allow the coexistence of cells expressing TTSS-1 (TTSS-1 “on”) with a pool of phenotypically avirulent cells (TTSS-1 “off”). The “off” cells carry the virulent genotype as well as the “on” cells and are able to switch on the expression of ttss-1 stochastically when encountering inducing cues from the environment. As mentioned before, the “on” cells grow just half as fast as the “off” cells (14, 49). The reason for the growth defect is still unclear although probably multifactorial, as many genes are coregulated with TTSS-1 (i.e., the HilD regulon) (52, 55).

Accordingly, a mathematical model of S. Typhimurium population dynamics predicted that decreasing the initial proportion of TTSS-1 “off” cells, by manipulating the regulation of ttss-1 expression, should lead to faster fixation of mutants that never switch on the expression of ttss-1 (TTSS-1 locked “off” defectors) (14). This model was simulating competitions between populations of wild-type S. Typhimurium (TTSS-1 “on” or “off”), defectors (locked “off”), and the microbiota in the gut. In this model, the inflammation was triggered above a certain population size of ttss-1-expressing cells, as observed in vivo (14, 49). Varying the frequency of TTSS-1 “on” versus “off” cells in the wild-type population revealed an optimum that allows S. Typhimurium on one hand to trigger inflammation and to outcompete the microbiota and on the other hand to limit the rise of defectors. This optimal theoretical TTSS-1 “on” frequency was ∼35%, very close to the frequency observed in the wild-type population in inducing conditions (14).

Predictions from this model were experimentally verified by following the evolution of an S. Typhimurium mutant that formed a reduced proportion of TTSS-1 “off” cells (higher TTSS-1 “on” frequency) at the beginning of the experiment. The mutant was a knockout of hilE, a gene encoding a negative regulator of ttss-1 expression. Thus, hilE mutants do form populations with increased fractions of “on” cells. As predicted by the model, hilD mutants (TTSS-1 locked “off” defectors) were increasing in frequency much faster in the hilE mutant background than in the wild-type background during within-host growth. This demonstrated that the avirulent phenotype indeed serves to slow down the rise of defectors, probably by occupying the same ecological niche (14). These results suggested that fine-tuned bimodal gene expression could be adaptive. We hypothesized that it could allow division of labor between two subpopulations of Salmonella: the TTSS-1 “on” cells express virulence factors and trigger inflammation, while the TTSS-1 “off” cells compete with fast-growing defector mutants, thus ensuring the transmission of the virulent genotype.

S. TYPHIMURIUM IN THE TISSUES, A HETEROGENEOUS RESERVOIR FOR THE VIRULENT GENOTYPE

S. Typhimurium can invade the epithelium using its flagella, the SiiE adhesin, and the SPI-1-encoded TTSS-1 (12, 56–58). This invasion is balanced by a swift innate host defense that reduces epithelial pathogen loads in an NLRC4 (NOD-like receptor subfamily C 4) inflammasome-dependent fashion (NLRC4-inflammasome driven expulsion of infected enterocytes) (41). The exact mechanisms driving the pathogen’s expulsion from this site remain poorly understood.

Within host cells, S. Typhimurium resides in the Salmonella-containing vacuole (SCV) and expresses the SPI-2-encoded TTSS-2. Thirty different effector proteins are then secreted through the membrane of the SCV into the host cell cytosol (for a review about their functions, see reference 59). This process is driven by the conditions inside the SCV, i.e., low nutrients and low pH. TTSS-2 is essential to ensure the systemic survival of S. Typhimurium and full virulence. Although the importance of SPI-2 effectors in intraepithelial growth in mice seems rather limited (58; B. Felmy, personal communication), they play a significant role in survival and division inside macrophages (60).

Depending on the host cell type, various proportions of intracellular bacteria can escape the SCV and divide in the cytosol (a phenomenon reviewed in reference 61). Transient coexpression of TTSS-1 and TTSS-2 upon entry could favor destabilization of the nascent SCV by TTSS-1 (62–64). Once in the cytosol, S. Typhimurium expresses TTSS-1 and the flagella. Luminal release of epithelial cells loaded with bacteria that are primed to reinvade the mucosa is thought to help intestinal spreading and sustaining of the gut inflammation (64, 65).

The heterogeneity in S. Typhimurium physiological states and death rates tends to further increase with the multiplicity of cell types hosting the bacteria within systemic organs such as the spleen (66, 67). This aspect of the infection process in Salmonella spp. is a fairly general feature of intracellular pathogens, e.g., Yersinia pseudotuberculosis (68) or Mycobacterium tuberculosis (69) (reviewed in reference 70). Such heterogeneity may promote chronic infections and impair the efficacy of antibiotics in treating diseases, with implications for the evolution of virulence. Mouse-to-mouse experimental transmissions of clones isolated from spleen or liver have demonstrated that growth conditions within host tissues select for increased virulence (71) and pathoadaptive mutations (72). As discussed in the next section, systemic dissemination of NTS is not (always) an evolutionary dead end. During prolonged infections in hosts able to contain systemic spread, extraintestinal sites represent a reservoir for virulent genotypes. Sporadic reseeding of the gut from the tissues (i.e., in relapses) can also promote the fecal shedding of virulent NTS and their natural transmission to new hosts.

IMPACT OF ANTIBIOTICS ON THE TRANSMISSION OF THE VIRULENT GENOTYPE

The cooperative virulence of S. Typhimurium is inherently unstable in the gut lumen, from where the pathogen is excreted into the environment and can reach new hosts. Within this niche, avirulent defector mutants evolve. When the frequency of defectors is high enough, they can eventually impair the transmission of virulent genotypes to the next host (73) (Fig. 3). This happens particularly in the absence of ecological disturbance in the gut lumen, where competition for resources favors defectors, although different selection regimes in variable environments can influence the rate of defector fixation (74).

FIGURE 3.

Antibiotic treatments select for virulent clones of S. Typhimurium able to form persisters in the host tissues. (Left) In the absence of antibiotics, defectors can reach fixation and their transmission to the next host prevents disease. (Right) Antibiotics kill all cells in the lumen: defectors (hilD mutants) and virulent wild-type (wt) cooperators. However, wt cells survive in the tissues and can reseed the lumen upon antibiotic withdrawal. This leads to successful transmission of the virulent genotype to the next hosts. Reproduced with permission from reference 73.

One of the strongest disturbances that can occur in the ecological niche of a pathogen is the presence of antibiotics. In the gut lumen, pathogen cells are growing and are highly sensitive to and efficiently killed by antibiotics. However, in systemic organs as well as in the cells of the intestinal mucosa, it was observed that S. Typhimurium is relatively tolerant to antibiotics (67, 75). In infected mice treated with the fluoroquinolone ciprofloxacin, the gut luminal population of S. Typhimurium SL1344 drops below the detection limit within a few hours after the onset of the therapy, while a substantial population remains viable in the intestinal mucosa (73) (Fig. 3) and other systemic organs (e.g., in the mesenteric lymph nodes) (75). These pathogen cells can survive for at least 10 days under continuous ciprofloxacin therapy. The ciprofloxacin is able to diffuse systemically and penetrate host cells. This means that intracellular Salmonella is indeed exposed to this antibiotic. The observed resistance is due not to mutation but rather to the expression of a transient tolerant phenotype called “persister” (76), which seems linked here to the slow replication rate of a subset of intracellular bacteria (67, 75, 77). Persisters can be isolated from the host’s tissues and normally cultivated in vitro. The reisolated clones present a sensitivity to ciprofloxacin that is identical to that of the ancestor clone, before passage in mice and exposure to the antibiotic. This indicates that ciprofloxacin survival is due to phenotypic adaptation, not to selection of genetically resistant mutants.

Upon clearance of the antibiotic, fully virulent clones of S. Typhimurium were able to reseed the gut lumen directly from the mucosa and to reestablish acute inflammation. Moreover, when transmitted to naive mice, the reseeding population was able to trigger colitis, while a population invaded by defectors (i.e., stool from untreated mice) was incapable of causing any symptoms (Fig. 3).

This demonstrates that in the presence of antibiotics, tissue invasion represents a potent selective advantage that offsets the fitness cost associated with TTSS-1 expression. In other words, antibiotic treatments can artificially inflict a cost of cheating (which was absent in our original experiments where we detected the upgrowth of defectors in the gut luminal pathogen population) that is higher than the benefit from not expressing virulence. This observation revealed two important aspects of the pathogen-host interaction: first, an important detrimental effect of antibiotic-based therapies against NTS, namely the selection for virulence; and second, a possible role of the persister phenotype in the evolution of facultative intracellular pathogens.

PERSPECTIVES: CAN THE PROCESS OF TRANSMISSION IN ITSELF MAINTAIN A VIRULENT GENOTYPE?

The life cycle of enteropathogens is not limited to within-host growth. It also includes transmission steps often associated with passages through harsh extrahost environments that can dramatically alter the structure of the pathogen population, a factor influencing the emergence of cooperation in general (78) and of cooperative virulence in particular (79). Transmission events often imply random sampling of bacteria reaching the next host (population bottleneck). This reduces the genetic variability previously generated within the donor host and can favor clones that are not necessarily more adapted (genetic drift) (80). The impact of population bottlenecks depends on the size of these bottlenecks (i.e., the effective population size reaching the next host) and on the genetic diversity of the population when they occur. Transmission could therefore help maintain virulence when virulent clones are overrepresented in the donor and have a higher chance to pass through the bottleneck without defectors. Once released into the environment, defectors could also be more or less efficient than virulent clones in reaching new hosts. In the most extreme case, i.e., clonal transmission, defectors have no chance to trigger inflammation by themselves, a scenario that should strongly favor the evolution (or maintenance) of virulence. Further investigations are required to understand the evolutionary dynamics of NTS virulence after successive passages from hosts to hosts.