Salmonella Typhi-specific multifunctional CD8+ T cells play a dominant role in protection from typhoid fever in humans

Background Typhoid fever, caused by the human-restricted organism Salmonella Typhi (S. Typhi), is a major public health problem worldwide. Development of novel vaccines remains imperative, but is hampered by an incomplete understanding of the immune responses that correlate with protection. Methods Recently, a controlled human infection model was re-established in which volunteers received ~103 cfu wild-type S. Typhi (Quailes strain) orally. Twenty-one volunteers were evaluated for their cell-mediated immune (CMI) responses. Ex vivo PBMC isolated before and up to 1 year after challenge were exposed to three S. Typhi-infected targets, i.e., autologous B lymphoblastoid cell-lines (B-LCL), autologous blasts and HLA-E restricted AEH B-LCL cells. CMI responses were evaluated using 14-color multiparametric flow cytometry to detect simultaneously five intracellular cytokines/chemokines (i.e., IL-17A, IL-2, IFN-g, TNF-a and MIP-1b) and a marker of degranulation/cytotoxic activity (CD107a). Results Herein we provide the first evidence that S. Typhi-specific CD8+ responses correlate with clinical outcome in humans challenged with wild-type S. Typhi. Higher multifunctional S. Typhi-specific CD8+ baseline responses were associated with protection against typhoid and delayed disease onset. Moreover, following challenge, development of typhoid fever was accompanied by decreases in circulating S. Typhi-specific CD8+ T effector/memory (TEM) with gut homing potential, suggesting migration to the site(s) of infection. In contrast, protection against disease was associated with low or no changes in circulating S. Typhi-specific TEM. Conclusions These studies provide novel insights into the protective immune responses against typhoid disease that will aid in selection and development of new vaccine candidates. Electronic supplementary material The online version of this article (doi:10.1186/s12967-016-0819-7) contains supplementary material, which is available to authorized users.

Typhi (Quailes strain) [3,4]. This controlled infection study was modeled after the human typhoid challenge studies performed in the 1960s at the University of Maryland. The Maryland studies improved understanding of typhoid fever [5][6][7][8] and resulted in the initiation of the process to license the oral attenuated Ty21a typhoid vaccine [9], but did not identify the immunological correlates of protection. Although substantial data are available on the immune responses after infection in the field or following vaccination, there are no studies that provide insights into the immunological status before wild-type infection and its possible effects on clinical outcome.
The re-establishment of the human challenge model with wt S. Typhi, and the use of cutting-edge multichromatic flow cytometry allowed us, for the first time, to investigate the pre-challenge immunological status and its correlation with the subsequent clinical outcome. Furthermore, it allowed the initiation of detailed studies of the kinetics and characteristics of the immunological responses occurring following infection with wt S. Typhi.
Recent research on the immune responses after oral immunization with Ty21a have revealed persistent multiphasic, multifunctional (simultaneous production of multiple cytokines) responses to antigenic presentation by class Ia HLA and by the more conserved and less polymorphic non-classical HLA-E molecules [13,19,20,22].
In the present study we investigated the relationship between S. Typhi-specific CD8+ T cell responses before exposure to wt S. Typhi and clinical outcome, i.e., whether the participants who were challenged developed disease or not. We also explored S. Typhi-specific CD8+ T cells responses following challenge, as well as their gut homing potential in relationship with typhoid diagnosis. Finally, we identified the dominant multifunctional S. Typhi-specific response patterns associated with clinical outcome by evaluating the simultaneous production of macrophage inflammatory protein (MIP)-1β, IFN-γ, TNF-α, IL-2 and IL-17A, as well as the expression of the cytotoxicity degranulation marker CD107a [23]. These investigations provide evidence that baseline S. Typhispecific responses are related to clinical outcome after wt S. Typhi infection and provide novel insights into the immunological responses involved in protection following natural infection and vaccination.

Participants and study design
Twenty-one healthy, male or female participants aged 18-60 years were recruited by the Oxford vaccine Group, Department of Paediatrics, Oxford, UK, to participate in this dose-escalation study. Any participant with a history of typhoid fever or immunization against typhoid fever, or who lived in a typhoid-endemic region for longer than 6 months, was excluded from participation. Only participants with low risk of becoming chronic carriers (including those without gall stones, determined by ultrasound examination of the gallbladder) were included. Participants were challenged orally with a dose of 1-5 × 10 3 CFU of wt S. Typhi (Quailes strain) suspended in sodium bicarbonate. The S. Typhi Quailes strain, which was used extensively for human challenge studies in the 1960s/1970s was developed by the University of Maryland and used to establish a master cell bank in Oxford. Participants were monitored closely throughout the study. A positive typhoid fever diagnosis was defined based on the following criteria: either a positive blood culture for Salmonella Typhi from day 5 post-challenge or, oral temperature ≥38 °C, persisting continuously for at least 12-h in the absence of anti-pyretic medication, occurring from 72-h after challenge. At the point of typhoid fever diagnosis (TD, as determined by S. Typhi bacteremia or development of a fever >38 °C for ≥12 h) participants were treated with a 2-week course of antibiotics (ciprofloxacin, 500 mg BD). Participants who did not developed typhoid fever (NoTD) received a 2-week course of antibiotics at day 14 post-challenge. Peripheral blood mononuclear cells (PBMC) obtained from 16 selected participants based on cell availability were analyzed in these studies (Additional file 1: Table S1). These included all nine who were not diagnosed with typhoid (NoTD) and seven participants were diagnosed with typhoid (TD).

Ethics statement
Written informed consent was obtained and all procedures approved by National Research Ethic Service (NRES), Oxfordshire Research Ethics Comittee A (10/ H0604/53) and conducted in accordance with the principles of the International Conference of Harmonisation Good Clinical Practice guidelines.

Specimen collection and isolation of PBMC
Blood was collected from each participant and routine blood hematology was assessed on alternate days after challenge and at typhoid diagnosis. PBMC were separated by Lymphoprep gradient centrifugation (Axis-Shield, Oslo, Norway) cryopreserved in liquid nitrogen following standard techniques within 4 h of initial blood draw within 4 h of initial blood draw and stored in liquid N 2 . Upon thawing, viability and recovery were measured using trypan blue exclusion and a Guava easyCyte ™ flow cytometer (Merck KGaA, Darmstadt, Germany), and cells were rested overnight in complete RPMI (cRPMI: RPMI 1640 media (Gibco, Carlsbad, CA) supplemented with 100 U/mL penicillin (Sigma), 100 µg/mL streptomycin (Sigma), 50 μg/mL gentamicin (Gibco), 2 mM l-glutamine (Gibco), 10 mM HEPES buffer (Gibco) and 10 % heat-inactivated fetal bovine serum (Gemini Bioproducts, West Sacramento, CA) to serve as effector cells in CMI assays.

Stimulator cells
Autologous Epstein-Barr virus (EBV)-transformed lymphoblastoid cell line (B-EBV cells) and autologous blasts were generated from the PBMC of each participant isolated before challenge. B-EBV cells were obtained by incubation of the PBMC with EBV-containing supernatant from the B95-8 cell line (ATCC CRL1612) and cyclosporine (0.5 ug/mL; Sigma-Aldrich, Saint-Louis, MO) at 37 °C with 5 % CO 2 for 2-3 weeks. PHA-activated blasts were prepared by incubating PBMC with 1 μg/ ml PHA (Sigma-Aldrich, St. Louis, MO) in cRPMI for 24 h, followed by washing and culture in cRPMI containing 20 IU/ml recombinant human IL-2 (rhIL-2; Roche, Indianapolis, IN) for 7 days. The HLA classical class I-defective B cell line 721.222.AEH which expresses nonclassical class-I HLA-E molecules was provided by Dr. D. Geraghty [14] and cultured in cRPMI supplemented with 200 mU/ml hygromycin B (Sigma-Aldrich).

S. Typhi infection of stimulator cells
Stimulating cells were infected by incubation for 3 h at 37 °C with wt-S. Typhi strain ISP1820 at a multiplicity of infection of 7:1 in RPMI without antibiotics, washed three times with cRMPI and incubated overnight with cRPMI containing 150 μg/ml gentamicin. Cells were washed and infection with S. Typhi was confirmed by staining with anti-Salmonella common structural Ag (CSA-1) (Kierkegaard and Perry, Gaithersburg, MD) and analysis by flow cytometry as previously described [10].

Ex-vivo stimulation of effector cells
PBMC were thawed and rested in cRPMI overnight at 37 °C in 5 % CO 2 before stimulation with S. Typhiinfected stimulating cells. Uninfected target cells and Staphylococcal enterotoxin B (SEB; 10 μg/ml) were used as negative and positive controls, respectively. Target cells were gamma-irradiated (6000 rad) and co-cultured with PBMC (effector:stimulator ratio 5:1) in the presence of the FITC-conjugated anti-CD107a (BD Biosciences) monoclonal antibody (mAb). After 2 h, the protein transport blockers monensin (1 µg/ml, Sigma) and brefeldin A (2 µg/ml; Sigma) were added to the PBMC and cells were incubated overnight at 37 °C in 5 % CO 2 . For the kinetics analysis, because of the variable responsiveness at baseline in the different participants, the net values (day x post-challenge-day 0) were used to normalize the data.

Statistical analyses
Mann-Whitney tests and linear regression analysis were performed using Prism v7.02 (Graphpad software, La Jolla, CA). Areas under the curve were measured using the trapezoidal method (GraphPad Prism v7.02). P values <0.05 were considered significant. Some comparisons were based on multiple outcomes from the same individual as indicated in the text. In these cases the same individual provided information on cytokine production and/or CD107a expression levels after stimulation with three types of stimulations (EBV-B, AEH, and blasts), and these responses we evaluated with regard to clinical outcome (i.e., TD vs. NoTD patients). In these analyses we accounted for the correlation between multiple outcomes from the same person using a mixed effects model fit by maximum likelihood, including a random effect for person, using SAS 9.3 (Cary, NC). We found by performing simulations that this correlation model provided more accurate statistical inference than models with more complex correlation structure given the small sample sizes.

Baseline S. Typhi-specific CD8+ T cell responses correlate with clinical outcome after challenge
CD8+ T cells have been reported to be a major component of the CMI response against S. Typhi [11-13, 16, 22]. Thus, we first explored whether CD8+ T cell responses in healthy participants at baseline could predict clinical outcome after challenge with 1-5 × 10 3 CFU of wt S. Typhi. All nine volunteers who did not develop typhoid disease (NoTD group) and seven participants who did develop typhoid disease (TD group) were evaluated. Peripheral blood mononuclear cells (PBMC) isolated from volunteers before challenge (day 0) were stimulated in vitro with S. Typhi-infected autologous B-EBV, S. Typhi-infected autologous blasts or S. Typhi-infected AEH cells (the latter to measure HLA-E-restricted CMI). PBMC were evaluated using 14-color multiparametric flow cytometry and CD8+ T cells were divided based on their expression of CD62L and CD45RA into naïve T (TN; CD62L+CD45RA+), T central memory (T CM ; CD62L+CD45RA−), T effector memory (T EM ; CD62L− CD45RA−), and T effector memory CD45RA+ (T EMRA ; CD62L−CD45RA+) subsets. No significant differences were observed at baseline between the TD and NoTD groups in the absolute number of white cells, lymphocytes, CD3, CD8 and CD4 memory subsets (Additional file 1: Figure S1). S. Typhi-specific responses were further characterized by co-expression of the T cell activation marker CD69 and the cytotoxicity degranulation marker CD107a, as well as the production of IFN-γ, TNF-α, MIP-1β, IL-17A and IL-2.
Most participants who did not develop infection (NoTD) showed higher S. Typhi-specific T EM and T EMRA responses at baseline than participants diagnosed with typhoid disease (TD) (Fig. 1). Since these differences were observed after stimulation with the different S. Typhiinfected cells, we combined them for the analyses shown in Fig. 1. Consistent with our previous findings [17,19,20,24], T EM cells (T EM and T EMRA ) represented the major subsets exhibiting intracellular chemokine/cytokine production. Significant differences were observed between the TD and NoTD groups for the expression of CD107a in T EM , as well as the production of TNF-α, MIP-1β and IL-2 in T EM and T EMRA . Similar trends, albeit not significant, were also observed in CD107a expression and IFN-γ production by T EMRA and IL-17A production by T EM , T EMRA and T CM . No differences were observed in the T CM subset, except for IL-17A production but this difference did not reach significance.
Interestingly, among the volunteers diagnosed with typhoid, we also found a positive relationship (significant in many cases) between S. Typhi-specific baseline levels of all markers in T EM and delay in time of disease onset (Fig. 2). These relationships were significant for CD107a expression, and IFN-γ and MIP-1β production. These data show that the presence of higher levels of S. Typhispecific T cells before challenge were associated with protection against typhoid fever and delayed time to disease onset.

Typhi-specific responses after challenge
After identifying correlations between baseline responses and clinical outcome, we examined the kinetics of immune responses until day 28 after challenge. Decreases in the absolute numbers of CD3+, CD8+ were observed in TD participants before diagnosis (Additional file 1: Figure S2A), a finding consistent with the decreased lymphocyte counts previously reported [3]. Decreases in the absolute numbers of S. Typhi-specific T EM expressing CD107a and cytokine-producing cells were also observed (Additional file 1: Figure S2B) in TD subjects. In contrast, these decreases in the absolute numbers of S. Typhi-specific T EM expressing CD107a or producing IFN-γ were not observed in NoTD participants (Additional file 1: Figure S2B). We also investigated whether challenge with wt S. Typhi elicited multiphasic responses such as those observed after vaccination with Ty21a [19,20]. Soon following challenge and before the onset of disease, 5 out of 6 TD evaluable participants showed a decrease in circulating CD107a-expressing and cytokine-producing T EM cells following stimulation with S. Typhi-infected B-EBV cells (Fig. 3a, Additional file 1: Figure S3A). In most TD subjects, this early drop was followed by a sharp rebound above baseline levels after disease onset. In contrast, no changes were observed in 4 out of the 6 NoTD participants (Additional file 1: Figure S3A). Of note, in three NoTD participants who did not reach the criteria for typhoid disease diagnosis, S. Typhi was detected either in blood by PCR or in stool (Additional file 1: Figure  S3A). Interestingly, these NoTD volunteers showed similar decreases after challenge to those observed in TD volunteers. Of note, variability in the magnitude of the decrease after challenge was noticed among TD participants (Additional file 1: Figure S3A). We identified a positive trend for all biomarkers between the magnitude of the decrease below baseline (area under the curve (AUC)) and their baseline levels. Significant correlations in these measurements were observed for IFN-γ, TNF-α, IL-17A and IL-2 in T EM cells (Fig. 3b). Notably, T EM cells showed a more pronounced decrease for all makers, followed by T EMRA and T CM cells (Additional file 1: Figure  S3B).

Enhanced gut homing potential of S. Typhi-specific CD8+ T cells in participants diagnosed with typhoid disease
The mucosal immunity elicited at the site of entry for S. Typhi is expected to be a key factor in protection against typhoid fever after vaccination with live oral typhoid vaccines [14,17]. Therefore, we assessed whether the decrease in S. Typhi-specific responses after challenge were at least in part due to the selective migration of CD8+ T cells to the small intestine, as determined by the expression of the gut homing molecule integrin α 4 β 7 (Fig. 4a) in S. typhi-specific T EM cells. We found no difference in the baseline proportions of integrin α 4 β 7expressing cells between the TD and NoTD groups in total CD8+ T EM cells (Fig. 4a). Interestingly, a substantial decline in the proportion of integrin α 4 β 7 + CD8+ T EM was observed in circulation around the time of diagnosis (d6-d14) for TD participants, while this proportion remained unchanged for NoTD participants (Fig. 4b). We next measured the S. typhi-specific expression of integrin α 4 β 7 on activated (CD69+) CD8+ T EM after stimulation with S. typhi-infected cells (AEH cells, B-EBV cells or blasts). Early after challenge, TD participants showed similar decreases in both integrin α 4 β 7 − and integrin α 4 β 7 + S. typhi-specific T EM , and both populations rebounded over baseline levels after disease diagnosis (Fig. 4c). In contrast, no changes were observed in NoTD participants (Fig. 4c).

S. Typhi-specific CD8+ T M cells are primarily multifunctional (MF)
Concomitant production of cytokines/chemokines and/ or expression of CD107a by single cells (multifunctional cells) appear to be central for protective immune responses [19,20,[25][26][27][28]. Thus, we next studied the characteristics of all 64 possible combinations of the six functional biomarkers measured in CD8+ T EM cells at baseline. We first grouped them into single positive (1+) or multifunctional (MF) cells, i.e., cells positive for two or more biomarkers (Fig. 5a). Notably, at baseline, the MF populations were significantly higher in participants who did not develop typhoid disease. We further categorized the MF populations based on the number of biomarkers they exhibited, and observed that double (2+), triple (3+) and quadruple (4+) positive populations comprise the dominant populations of S. Typhi-specific MF CD8+ T EM in TD and NoTD participants (Fig. 5b). Moreover, we observed that quadruple (4+) and quintuple-sextuple (5-6+) positive MF cells were significantly higher in the NoTD group, while strong trends were also observed for double and triple positive cells. We further characterized the individual MF populations by identifying the six dominant (i.e., highest frequency) MF populations, which together represented 65-85 % of the total S. Typhi-specific MF cells in the NoTD participants (Fig. 5c). The production of the chemokine MIP-1β was a major characteristic in all populations, followed by CD107a expression and production of IFN-γ and TNF-α while IL-2 was detected at considerably lower frequencies. The CD107a+IFN-γ+TNF-α+MIP-1β+, CD107a+TNF-α+MIP-1β+ and CD107a+MIP-1β+ populations were predominant and significantly higher in the NoTD participants. IL-17A was detected at lower levels, and was not produced by any of the 6 dominant populations. Of note, several of the dominant MF populations (e.g., CD107a+MIP-1β+) showed significant correlations with delayed time to disease diagnosis. The MF cells were similarly dominant in both integrin α 4 β 7 − and integrin α 4 β 7 + T EM in NoTD participants (Additional file 1: Figure S4A, B), however, the TNF-α+MIP-1β+ and CD107+TNF-α+MIP-1β+ populations were significantly higher in integrin α 4 β 7 − T EM subsets (Additional file 1: Figure S4C). No differences were observed between integrin α 4 β 7 − T EM and integrin α 4 β 7 + T EM in TD participants.
Finally, we observed that S. Typhi-specific MF responses were also prevalent at day 7 post challenge in NoTD participants (Fig. 6a, b), and no differences were found in the percentages of the six dominant populations compared to baseline (Fig. 6c). Interestingly, in contrast to what we observed before challenge, in TD subjects we recorded a strong trend towards a dominance of MF cells 48 h after typhoid diagnosis (Fig. 6a). This dominance of MF populations in TD participants at 48 h was more evident in integrin α 4 β 7 + than in integrin α 4 β 7 − S. Typhi-specific T EM subsets (Additional file 1: Figure S5A, B). TNF-α+MIP-1β+, CD107+MIP-1β+ and CD107a+IFN-γ+TNF-α+MIP-1β+ populations in NoTD participants were significantly higher in the integrin α 4 β 7 − T EM subsets (Additional file 1: Figure S5C). However, no differences were observed between integrin α 4 β 7 − T EM and integrin α 4 β 7 + T EM in TD participants. Taken together, these results point to the importance of MF S. Typhi-specific CD8+ T EM with gut, as well as extra-intestinal tissue homing potential in protection from disease.

Discussion
Studies on typhoid fever immunity in humans have been largely restricted to patients residing in endemic areas and subjects immunized with licensed and experimental typhoid vaccines. In contrast, virtually no information is available on the immune status, particularly regarding CMI, prior to infection and on the immunological correlates of protection following oral exposure to wt S. Typhi. The re-establishment of the human challenge model with wt S. Typhi provided a unique opportunity to directly assess the impact of S. Typhi-specific CMI responses before pathogen exposure to the subsequent development of, or protection from, typhoid disease. Herein we show that higher S. Typhi-specific baseline responses are associated with protection against typhoid fever and delayed time to diagnosis and characterized these responses in great detail. If confirmed by future studies involving additional volunteers, these novel observations suggest an important role of anti-S. Typhi specific multifunctional CMI responses in protection from disease. This information has the potential to greatly accelerate the development of novel new generation typhoid vaccines. It has been recently described that the multifunctional quality of the responses is critical for protection against pathogens [23,[25][26][27][28]. In the current study we provide evidence that S. Typhi-specific MF CD8+ T EM and T EMRA cells are the major effector subsets associated with protection against typhoid fever, as well as delayed time to disease onset. Of note, while MF S. Typhi-specific CD8+ cells in NoTD participants were dominant both before and after challenge, this dominance was seen in TD participants only after challenge. MIP-1β was shown to play a role in CTL activity and in controlling infection in HIV non-progressors [29][30][31].
We previously described the co-production of MIP-1β with IFN-γ, TNF-α, IL-17A and IL-2 following vaccination with Ty21a [20] and MIP-1β was reported to be produced by PBMC obtained from S. Typhi-infected convalescent patients [32]. Herein, we show that production of MIP-1β is a major feature of S. Typhi-specific MF populations suggesting that co-production of MIP-1β with other cytokines is a key component in protection against S. Typhi. These results highlight that a strong MF component of CD8+ T EM responses both at baseline and after challenge, is associated with disease protection and delayed time to diagnosis. Following challenge we observed decreases in the percentages of circulating S. Typhi-specific T EM cells in TD participants, suggesting that these cells might be leaving the systemic circulation during the incubation phase of the disease. Similar to the decreases in lymphocyte counts reported post-challenge in TD participants [3], we also observed a decline in the absolute numbers of CD3+, CD8+ and CD4+ T cells. Integrin α 4 β 7 expression plays an essential role in the selective homing of T cells to the gut, the site of entry for S. Typhi [16,[33][34][35][36]. After challenge, TD participants show a decrease in total integrin α 4 β 7 + CD8+ T EM and in both integrin α 4 β 7 − and α 4 β 7 + S. Typhi-specific T EM cells. However, 48 h after diagnosis, the proportion of S. Typhi-specific MF integrin α 4 β 7 + T EM in TD subjects was higher than in integrin α 4 β 7 − cells. This shift in the balance of singlepositive vs. MF S. Typhi-specific CD8+ T EM in circulation might be the result of activation and expansion of MF S. Typhi-specific integrin α 4 β 7 + CD8+ T EM cells due to the ongoing infection in the gut and other lymphoid tissues. Taken together, these results highlight that strong MF CD8+ T EM responses with the potential to home to the gut as well as to other lymphoid tissues are associated with disease protection and delayed time to diagnosis. These results complement and expand our previous findings showing that vaccination against S. Typhi elicited integrin α 4 β 7 + and integrin α 4 β 7 − CD8+ effector T cells [14,16,17,24]. These observations also suggest that S. typhi-specific T EM cells migrate not only to mucosal sites, but also presumably to secondary lymphoid tissues, where they may reduce S. typhi replication during the incubation phase, delaying disease onset. Moreover, post-challenge decreases in circulating S. Typhi-specific T cells are proportional to the levels present at baseline, suggesting that the higher the pool of S. Typhi-specific T cells available, the higher the number of these cells that are recruited to the sites of inflammation.
Several studies in the murine S. typhimurium model have shown that the balance of suppressive regulatory T cell (T reg ) and pro-inflammatory T cell responses influence bacterial clearance or persistence [37]. We have recently evaluated in these challenged volunteers the hypothesis that the development of T reg responses [38] following exposure to wt S. Typhi could be responsible, at least in part, for the decrease of S. Typhi-specific T EM responses. We observed that TD participants exhibited up-regulation of the gut homing molecule integrin α 4 β 7 pre-challenge, followed by a significant down-regulation post-challenge consistent with T reg homing to the gut, as well as up-regulation of activation molecules postchallenge. We also showed that depletion of T reg results in increased S. Typhi-specific cytokine production by CD8+ T EM in vitro. These observations suggest that the tissue distribution of activated T reg , their characteristics and activation status may play a key role in typhoid fever, possibly through suppression of S. Typhi-specific effector T cell responses [39].
In contrast to the results seen in TD participants, we observed that protection against typhoid fever is mostly associated with very low or no changes in circulating S. Typhi-specific T M after challenge. Of note, several of the few NoTD participants showing decreases in S. Typhispecific T EM were PCR positive or stool positive participants despite the fact that they did not meet typhoid disease diagnosis. These findings suggest that the control and clearance of S. Typhi in NoTD participants might be driven by innate and/or adaptive immune responses in the gut microenvironment, precluding S. Typhi from becoming invasive and causing disease. However, the underlying mechanism(s) of control of S. Typhi infection in the gut microenvironment remain unclear. "Innatelike" T cell types such as TCRγ/δ T cells, NK-T cells and mucosal associated invariant T (MAIT) cells might be involved in protection from typhoid disease. MAIT cells are comprised of CD8+CD4− and CD8−CD4− subsets restricted by the MR1 MHC-related molecule which are widely believed to play an important in mucosal immunity [40]. We have recently shown that MAIT cells from healthy individuals are able to produce IL-17A, IFN-γ and TNF-α when exposed to S. Typhi-infected cells [41]. Ongoing experiments directed to characterize MAIT cell responses in the challenged participants are expected to shed light into the role of these cells following exposure to wt S. Typhi.
Because of the very limited data available regarding immune responses prior to typhoid infection, the reasons for the disparity in baseline responses observed among the participants are unclear. Participants were recruited in the UK, a non-endemic region, have not been vaccinated against typhoid, and therefore have likely never encountered S. Typhi. However, differences in baseline responses could be due to cross-reactive memory responses elicited by previous exposure to other Salmonella serovars [42][43][44], or other Enterobacteriaceae, including those present in the normal gut microbiota [22,[45][46][47]. Several studies highlight the importance of the gut microbiota in modulating host immune responses to pathogens or to vaccination [45][46][47]. We have recently shown that vaccination against S. Typhi caused no alterations of the microbiota, however, individuals displaying early multiphasic CMI responses harbored more diverse communities than late responders [47]. We have initiated studies to identify the interplay between the host immune response, the microbiota and clinical outcome in volunteers challenged with wt S. Typhi. In addition to these acquired immune response differences, genetic determinants like HLA molecules can also be critical in defining the variation in immune responses. For example, the presence of the HLA-DRB1*04:05 allele was recently associated with protection against S. Typhi [48].

Conclusions
In summary, we have provided the first direct evidence of an association between higher baseline levels of multifunctional S. Typhi-specific CD8+ T cells and protection, as well as delayed time to disease onset, in an oral challenge model with wt S. Typhi in humans. These studies also revealed some of the immunological responses associated with delayed time to disease onset and defined the homing characteristics of the S. Typhi-specific effector and memory CD8+ T cell populations. This information supports the performance of in depth CMI measurements to aid in the early selection of novel vaccine candidates for further development and evaluation in clinical trials.