Nature Reviews Microbiology | AOP, published online 9 March 2015; doi:10.1038/nrmicro3428
Now you see me, now you don’t: the interaction of Salmonella with innate immune receptors A. Marijke Keestra-Gounder, Renée M. Tsolis and Andreas J. Bäumler
Abstract | Salmonella enterica serovars are associated with an estimated 1 million deaths annually and are also useful model organisms for investigating the mechanisms of host– bacterium interactions. The insights gained from studies on non-typhoidal Salmonella (NTS) serovars have provided a fascinating overview of the mechanisms by which the innate immune system detects and responds to bacterial pathogens. However, specific virulence factors and changes in virulence gene regulation in S. enterica subsp. enterica serovar Typhi alter the innate immune responses to this pathogen. In this Review, we compare and contrast the interactions of S. Typhi and NTS serovars with host innate immune receptors and discuss why the disease manifestations associated with S. Typhi infection differ considerably from those associated with the closely related NTS serovars.
Serovars Distinct variants of bacterial species. Each of the more than 2,500 serovars that can be distinguished in the genus Salmonella is defined by an antigen formula, which lists O factors (antigenic epitopes in the O‑antigen), H factors (antigenic epitopes in flagellin) and fermentative characteristics.
Mesenteric lymph nodes Lymph nodes in the mesentery that drain the afferent lymphatic vessels from the small and large intestine.
Department of Medical Microbiology and Immunology, School of Medicine, University of California, Davis, One Shields Avenue, Davis, California 95616, USA. Correspondence to A.J.B. e‑mail: [email protected]
doi:10.1038/nrmicro3428 Published online 9 March 2015
Salmonella enterica serovars are important model organisms that are often used to establish new principles in bacterial pathogenesis, cellular microbiology and innate immunity research. For example, non-typhoidal Salmonella (NTS) serovars, such as the S. enterica subsp. enterica serovars Typhimurium and Enteritidis, are natural pathogens of mice, and mice infected with these serovars are excellent models in which to study host– microorganism interactions in vivo (reviewed in REF. 1). These animal models serve as ‘litmus tests’ to establish the in vivo relevance of a host–microorganism interaction, which is one reason that NTS serovars are often the pathogens of choice for testing new concepts in bacterial pathogenesis or innate immunity. Extrapolating insights gained from mouse models to Salmonella disease in humans is important, because Salmonella serovars are not only model pathogens but are also a leading cause of human morbidity and mortality worldwide (BOX 1). S. enterica serovars are associated with three distinct clinical syndromes: typhoid fever, which is caused by S. enterica subsp. enterica serovar Typhi; gastroenteritis, which is caused by NTS serovars in immunocompetent individuals; and bacteraemia, which is caused by NTS serovars in immunocompromised individuals (BOX 1). Following ingestion, S. enterica serovars invade epithelial cells preferentially in the terminal ileum. After traversing the epithelial layer, the pathogen is phagocytosed and can persist within the Salmonella-containing vacuole (SCV;
also known as the Salmonella-containing phagosome) of macrophages. Although gastroenteritis is a localized infection of the intestinal mucosa and mesenteric lymph nodes, S. enterica serovars disseminate throughout the body during typhoid fever and bacteraemia. The host responses elicited during bacteraemia are classically thought of as consequences of lipopolysaccharide (LPS) being detected by the pattern recognition receptor Toll-like receptor 4 (TLR4) expressed on monocytes, which triggers a cytokine storm that can lead to sepsis2. However, recent work on how the innate immune system detects NTS serovars has established several new paradigms that do not follow the current textbook models. Furthermore, it has become apparent that S. Typhi expresses virulence mechanisms that enable it to evade many of the host pathways that are used to detect NTS serovars. Here, we review these recent advances and discuss the relevance of findings that have been obtained using mouse models of infection for disease caused by S. enterica in humans.
NTS interactions with the innate immune system In immunocompetent individuals, infection with NTS serovars is associated with gastroenteritis, a typical diarrhoeal disease that remains localized to the terminal ileum, the mesenteric lymph nodes and (less frequently) the colon (reviewed in REF. 3). The pathogens most commonly associated with this diarrhoeal disease are the
NATURE REVIEWS | MICROBIOLOGY
ADVANCE ONLINE PUBLICATION | 1 © 2015 Macmillan Publishers Limited. All rights reserved
REVIEWS Box 1 | The global burden of human Salmonella disease The combined morbidity and mortality caused by Salmonella enterica serovars worldwide is considerable. Typhoid fever is responsible for approximately 21.65 million cases annually117, which, on the basis of a recent global estimate of a 2% case fatality rate118, results in 433,000 deaths per year. The global burden of non-typhoidal Salmonella (NTS) gastroenteritis is estimated to account for 93.8 million cases of disease annually, with 155,000 deaths119. NTS bacteraemia is a disease manifestation that is commonly observed in immunocompromised individuals, such as adults with advanced HIV disease or children with severe malaria or malnutrition. Between 2% and 8.5% of the approximately 20.2 million HIV-positive adults living in sub-Saharan Africa develop NTS bacteraemia annually120–122, resulting in a case fatality rate of 20%123 (that is, approximately 212,000 deaths per year). The incidence of child NTS bacteraemia (175–388 cases per 100,000 individuals124–126) suggests that among the 389 million children living in sub-Saharan Africa, approximately 1.09 million develop this bloodstream infection each year. On the basis of an estimated in‑hospital case fatality rate of 18.1%5, this means that approximately 197,000 child deaths result from NTS bacteraemia annually. Thus, the different clinical syndromes associated with S. enterica serovars are responsible for a combined global toll of approximately 1 million deaths annually, which makes these pathogens one of the leading causes of human mortality worldwide. This total is likely to be an underestimate, because the global mortality of NTS gastroenteritis was calculated by assuming that the disease is responsible for only 4,100 deaths per year in Africa and was extrapolated from data from returning travellers rather than Africans living in rural areas or those on a low income119. A recent study shows that all healthy children in Malawi develop Salmonella-specific antibodies by the time they are 16 months old127, suggesting that there is universal exposure to this pathogen at an early age. A survey from Kenya has shown that NTS serovars are present in 22% of stools from children who die of diarrhoeal disease, compared to 11% for Shigella spp., 9% for rotavirus and 5% for Campylobacter spp.128. These observations indicate that NTS gastroenteritis might substantially contribute to the approximately 2 million annual diarrhoeal deaths among African children.
Lipopolysaccharide (LPS). Decorates the surface of Salmonella enterica subsp. enterica serovar Typhimurium. A lipid A moiety anchors the LPS molecule in the outer membrane. The lipid A moiety is linked to the oligosaccharide core, which connects to the O‑antigen repeat units that extend from the bacterial surface.
Pattern recognition receptor A protein of the innate immune system that can sense an infection by identifying pathogen-associated molecular patterns or pathogenassociated processes.
Zoonotic Describes a pathogenic microorganism that spreads from an animal reservoir to cause disease in humans.
Lipid A A phosphorylated glucosamine disaccharide carrying multiple fatty acid chains that anchor lipopolysaccharide (LPS) in the outer membrane of most Gram-negative bacteria. The lipid A moiety is responsible for the toxicity of LPS, which is also known as endotoxin.
zoonotic S. Typhimurium and S. Enteritidis (reviewed in REF. 4).
By contrast, in immunocompromised individuals, NTS serovars are associated with an invasive bloodstream infection termed NTS bacteraemia (reviewed in REFS 5,6). NTS bacteraemia causes an acutely ill, febrile state, and the symptoms of gastroenteritis are often absent 7–9. NTS bacteraemia. Mice infected with S. Typhimurium develop an invasive bloodstream infection, which arguably best resembles the bacteraemia that this pathogen causes in immunocompromised individuals (BOX 2). The outcome of this invasive bloodstream infection depends on the genetic background of the mice10. In inbred mouse strains that are genetically resistant to S. Typhimurium infection (such as CBA mice), S. Typhimurium disseminates to the liver and spleen, where host defences limit bacterial replication and the pathogen is eventually cleared. By contrast, inbred mouse strains that are genetically susceptible to S. Typhimurium infection (such as C3H/HeJ, BALB/c and C57BL/6 mice) are unable to limit bacterial growth in the liver and spleen, and succumb to their illness. Importantly, there is no consensus as to whether susceptible or resistant mouse lineages are better models for studying the mechanisms involved during human infection (BOX 2). Mouse susceptibility to S. Typhimurium infection is determined by various mutations. For example, C3H/HeJ mice are genetically susceptible to S. Typhimurium infection owing to a mutation in the gene encoding TLR4, which, together with its co‑receptors CD14 and
myeloid differentiation factor 2 (MD2; also known as LY96), detects the lipid A moiety of S. Typhimurium LPS as a pathogen-associated molecular pattern (PAMP)11 (FIG. 1). This historical observation highlights the crucial role that the TLR4–CD14–MD2 receptor complex has in orchestrating immune responses that protect the host during NTS bacteraemia. Although lipid A dominates the response during S. Typhimurium bacteraemia, other PAMPs are sensed by host cells as well: TLR2 and TLR9, which detect lipoproteins and non-methylated CpG dinucleotides, respectively, contribute to the responses elicited during disseminated S. Typhimurium infection in mice12. By contrast, TLR5, which detects flagellin, has a small role during bacteraemia in mice that have been orally inoculated with S. Typhimurium, because flagella are no longer being expressed by the time the bacterium disseminates through the bloodstream to the liver and spleen. Downregulation of flagellum expression might therefore be an example of innate immune evasion by NTS serovars13,14. Although TLR4‑deficient mice readily succumb to infection with S. Typhimurium15–18, the absence of all TLR-mediated responses in mice that are deficient in TLR2, TLR4 and TLR9 partially restores resistance to infection. The explanation for these paradoxical observations is that expression of a major virulence determinant, the type III secretion system (T3SS) T3SS2, which is encoded on the Salmonella pathogenicity island (SPI) SPI‑2, requires TLR-dependent acidification of the SCV12; this acidification still occurs in TLR4‑deficient mice owing to activation of TLR2 or TLR9 by the pathogen. Thus, TLR signalling can both increase pathogen survival (owing to the TLR2- or TLR9‑induced expression of T3SS‑dependent virulence factors) and decrease pathogen survival (owing to the activation of TLR4‑dependent host defences). As the activation of TLR4 dominates, the overall outcome of the detection of S. Typhimurium PAMPs by TLRs is the control of bacterial growth in the liver and spleen. The pro-inflammatory cytokine responses that result from TLR4 signalling are amplified through the complement system, a humoral pattern recognition system that is activated (through the alternative pathway) by reactive hydroxyl groups in the O‑antigen repeat unit of LPS (BOX 3; FIG. 1). Complement activation leads to the deposition of complement component 3 fragment b (C3b) on the bacterial surface, which enhances opsonophagocytosis. The long O‑antigen chains (16–35 repeat units) present in S. Typhimurium LPS create a physical distance between the outer membrane and the location where the terminal membrane attack complex (C5b–C6–C7–C8– C9) is formed, thereby conferring resistance to complement-mediated lysis19–21 (BOX 3). However, the soluble cleavage product (C5a) that is released from the bacterial surface during complement deposition functions as a potent attractant for neutrophils22. Despite these beneficial effects for the host, activation of the complement system can also have detrimental consequences. The most important harmful mediator is C5a, an anaphylatoxin that contributes to the development of LPS-induced septic shock23. C5a enhances the
2 | ADVANCE ONLINE PUBLICATION
www.nature.com/reviews/micro © 2015 Macmillan Publishers Limited. All rights reserved
REVIEWS Box 2 | The mouse model and its relevance for human disease Although it is relatively straightforward to establish the biological significance of a new concept using a mouse model of Salmonella infection, it is not always immediately apparent for which human clinical syndrome a given microorganism–host interaction is relevant. Despite this limitation, mouse models have been invaluable in establishing the in vivo relevance of observations made in tissue culture. However, when extrapolating data on Salmonella disease in mice to Salmonella disease in humans, some limitations of the mouse models must be considered. First, as non-typhoidal Salmonella (NTS) serovars do not cause typhoid fever in humans, it is not clear to what extent infection of mice with NTS serovars resembles human infection with Salmonella enterica subsp. enterica serovar Typhi. Furthermore, the disseminated bloodstream infection caused by NTS serovars in mice more closely resembles NTS bacteraemia than NTS gastroenteritis in humans. Thus, it seems that some mucosal barrier functions that help to prevent dissemination of NTS serovars beyond the mesenteric lymph node in humans are weakened in mice. Possible candidates to explain the differences in barrier function between mice and humans are the defences orchestrated by professional phagocytes, particularly neutrophils. The population of the United States has a mean neutrophil count of approximately 4 × 109 cells per litre129. A benign neutropenia with a neutrophil count of