Polyfunctional CD4+ T Cell Responses to Immunodominant Epitopes Correlate with Disease Activity of Virulent Salmonella

Salmonella enterica serovars are intracellular bacteria capable of causing typhoid fever and gastroenteritis of significant morbidity and mortality worldwide. Current prophylactic and therapeutic treatment is hampered by the emergence of multidrug-resistant (MDR) strains of Salmonella, and vaccines provide only temporal and partial protection in vaccinees. To develop more effective Salmonella vaccines, it is important to understand the development of protective adaptive immunity to virulent Salmonella. Here we report the identification of novel CD4+ T cell peptide epitopes, which are conserved among Salmonella serovars. Immunization of Salmonella-infected mice with these peptide epitopes reduces the burden of Salmonella disease. Furthermore, we show that distinct polyfunctional (interferon-γ+, tumor necrosis factor+, and interleukin-2+) Salmonella-specific CD4+ T cell responses develop with respect to magnitude and kinetics. Moreover, we found that CD4+ T cell responses against immunodominant epitopes are predictive for active Salmonella disease. Collectively, these data could contribute to improved diagnosis of Salmonella-related diseases and rational design of Salmonella vaccines.


Introduction
Salmonella enterica serovars are intracellular bacteria capable of causing localized and systemic disease of significant morbidity and mortality [1,2,3]. Natural acquisition of Salmonella via contaminated water or food introduces the bacteria to the gastrointestinal tract, where Salmonella colonizes underlying mucosal tissue and disseminates to spleen and liver [4,5]. Salmonella enterica serovar Typhi causes typhoid fever in humans with an estimated 21 million cases annually, resulting in more than 200,000 deaths per year in endemic areas [6,7]. The closely related Salmonella enterica serovar Typhimurium and Enteriditis cause gastroenteritis, a foodborne disease which constitutes a major public health burden and represents a significant cost to society in many countries [2,8]. Thus, infections with Salmonella (salmonellosis) are not only a health concern in developing nations but are also an important cause of gastrointestinal infections in developed nations where contaminated food products are rapidly and widely distributed.
Despite the use of antibiotics in recent years, salmonellosis remains a major public health problem, both in terms of incidence and severity of cases. Many typhoid patients in developing countries fail to recover due to lack of treatment or a substantial delay in antibiotic administration [1,3]. In addition, multidrug-resistant (MDR) strains of Salmonella often emerge [9]. Although subunit vaccines comprising antigens conserved across different Salmonella serovars have the potential to be a safe and cost-effective prophylactic measure in combating Salmonella-related diseases, the current available vaccines provide only 50-60% protection in vaccinated individuals [3,7]. Moreover, typhoid vaccines lose effectiveness after several years. Importantly, the immune mechanisms mediating protection by such vaccination are not well understood, and which antigens are important for the induction of immunity to Salmonella is largely unknown.
In experimental models of Salmonella infection it was shown that mice develop a typhoid-like disease following S. Typhimurium infection [10]. Mice lacking MHC class II-restricted CD4 + T cells and B cells due to knockout mutations or antibody depletion are highly susceptible to both attenuated and virulent Salmonella [11,12,13,14], indicating important functions of these cells for immunity. In contrast, mice lacking class I-restricted CD8 + T cells or cd T cells can resolve infection with attenuated strains [12]. The mechanisms by which Salmonella-specific CD4 + T cells contribute to protective immunity are incompletely understood [15] and attempts to transfer immunity with Salmonella-specific T cell lines have failed to identify the specificity of the response [16,17]. The repertoire of Salmonella epitopes recognized by CD4 + T cells during infection is largely unknown with the exception of epitopes in flagellin (FliC) [18,19,20] and the type-III-secretion system [21].
A hurdle to the development of effective vaccines against (chronic) infectious diseases is the identification of antigens capable of eliciting polyfunctional T cell responses that provide immune correlates for disease activity and/or vaccine-mediated protection. In this study, we have identified novel CD4 + T cell epitopes that are conserved among Salmonella serovars. Furthermore, we show that diverse kinetics and polyfunctional profiles of Salmonellaspecific CD4 + T cell responses in mice exist, and that the CD4 + T cell responses to immunodominant epitopes are predictive for disease activity. These data could lead to improved diagnosis of Salmonella-related diseases and enhance rational design of vaccines against Salmonella serovars.

CD4 + T cell responses control virulent Salmonella infection
We initially set up a natural course of virulent Salmonella infection in the susceptible mouse strain C57BL/6 with the Salmonella enterica serovar Typhimurium (S. Typhimurium) strain LT2. After oral infection, which mimics the natural route of infection, with 1610 7 S. Typhimurium the absolute number of splenocytes and splenic weight in Salmonella infected mice progressively increased until week 3 post-infection ( Figure 1A). At week 4 post-infection, the total splenocyte number and splenic weight declined ( Figure 1A). The majority of the mice became moribund at later timepoints (5-6 weeks post-infection), and had to be euthanized because of ethical reasons. In parallel, we evaluated the magnitude and activation status of CD4 + and CD8 + T cell populations during the course of infection. The absolute number of CD4 + and CD8 + T cells followed similar kinetics as the total number of splenocytes ( Figure 1B). The expression of killer cell lectin-like receptor G1 (KLRG1), which identifies antigenexperienced T cells [22,23], was progressively increased on the cell surface of both CD4 + (Foxp3 2 ) and CD8 + T cells and showed similar kinetics as the absolute lymphocyte counts ( Figure 1B and 1C). Other indicators of T cell activation such as the up-regulation of the early activation marker CD69 and the down-regulation of the lymph node homing molecule CD62L were also observed during the course of infection ( Figure 1D and 1E). Thus virulent Salmonella infection expands T cell populations that display evidence of activation and likely represents development of Salmonella-specific T cell responses.
The bacterial burden increased until week 2 post-infection after which the number of live bacteria remained relatively high ( Figure 2A). To examine the importance of activated CD4 + and CD8 + T cell populations for controlling virulent Salmonella infection, we selectively depleted these subsets by using depleting monoclonal antibodies. The depletion of CD4 + T cells after the onset of infection resulted in significantly increased numbers of Salmonella bacteria, whereas CD8 + T cell depletion did not ( Figure 2B). Thus, although both CD4 + and CD8 + T cell subsets are activated, CD4 + T cells play a particular important role in the control of virulent Salmonella infection. Together these data establish the tempo of persistent infection, the accompanied activation status of the T cells, and the importance of CD4 + T cells in our experimental virulent Salmonella infection model and reflects the findings of other Salmonella infection models [11,12,13,24,25,26,27].
Identification of peptide epitopes to track Salmonellaspecific CD4 + T cell responses Given the importance of CD4 + T cells during Salmonella infection, we predicted genome-wide the MHC class II (H2 I-A b ) binding peptides encoded by S. Typhimurium strain LT2 [28] and S. Typhi strain CT18 [29] using optimized computer-based algorithms [30,31]. Since the amino acid homology of the S. Typhimurium and Typhi is .95% analogous, we selected only conserved epitopes. For the screening of the Salmonella epitopes, we tested by interferon (IFN)-c ELISPOT the top 1443 predicted epitope candidates pooled in 180 mixtures of 8 peptides each for their ability to restimulate splenic CD4 + T cells of C57BL/6 mice orally infected 18 days earlier with virulent S. Typhimurium. To minimize background levels of ELISPOT assays, we loaded in vitro cultured dendritic cells (DCs) from naive mice with the peptides and used these peptide-loaded cells to restimulate the splenic CD4 + T cells that were purified by magnetic beads. In this screen, we identified three positive pools. Analysis of the individual peptide epitopes of these pools yielded three peptides with positive responses. The identified CD4 + T cell epitopes were all novel and were located in different gene products of Salmonella namely in the alkyl hydroperoxide reductase subunit C (AhpC), the ethanolamine ammonia-lyase small subunit (EutC) and in STM1540, a putative hydrolase related to choloylglycine hydrolase (Table 1). Previous work showed that expression of AhpC is induced by the oxidative burst of macrophages and that this Salmonella enzyme plays a role in the stimulation of T cells [32,33]. A study using mass spectometry-based proteomics confirmed the obvious presence of AhpC in Salmonella and additionally detected expression of EutC [34]. A comprehensive study that performed proteomics and transcriptional analyses showed that expression of AhpC was altered post-transcriptionally through the RNA-binding protein SmpB while post-transcriptional regulation of STM1540 was via the RNA-binding protein Hfq [35]. Blast searches of the amino acid sequences using The National Center for Biotechnology Information (NCBI) web pages and databases revealed that the three newly identified CD4 + T cell epitope sequences are not only conserved in S. Typhi and Typhimurium but are also conserved among many other Salmonella serovars. The AhpC 154-168 epitope is fully conserved in other microbes as well such as Pseudomonas, Burkholderia, Citrobacter and Polaromonas. Also the EutC 243-257 epitope sequence is 100% conserved in other micro-organisms including Klebsiella, Shigella, Citrobacter, Escherichia coli and Aspergillus but the amino acid homology of the STM1540 262-276 epitope is 80% or lower with other bacteria (e.g. Geobacter and Xanthomonas).
Tracking the Salmonella-specific CD4 + T cell response over time revealed that the most immunodominant response throughout infection was against the STM1540 262-276 epitope, followed by AhpC 154-168 and EutC 243-257 respectively ( Figure 3A and 3B). The response against the previous identified FliC 429-443 epitope [19] epitope was roughly similar in magnitude as the EutC 243-257 epitope. Besides displaying variation in magnitude, the kinetics of Salmonella-specific CD4 + T cell responses were also diverse: whereas STM1540 262-276 -specific CD4 + T cell responses increase until day 21, the responses against the AhpC 154-168 and FliC 429-443 epitopes contracted after day 14 post-infection. Moreover, the percentage of EutC 243-257 -specific CD4 + T cells gradually increased until day 28 post-infection ( Figure 3B). The STM1540 262-276 and AhpC 154-168 -specific CD4 + T cell responses were also clearly detectable in the mesenteric lymph nodes of Salmonellainfected mice at day 14, 21 and 28 post-infection ( Figure 3C and data not shown). Together, these data show that CD4 + T cell epitopes to virulent Salmonella can be identified and that Salmonella-

Peptide vaccination reduces the burden of virulent Salmonella infection
To assess whether the newly identified peptide epitopes could elicit Salmonella-specific CD4 + T cell responses that contribute to immune control, we vaccinated wild-type mice at day 5 post Salmonella infection with a mixture of the peptides AhpC 154-168 , EutC 243-257 and STM1540 262-276 (8 mg (,5 nmol) or 40 mg (,25 nmol) of each peptide). Quantification of the bacterial colonization in the spleens and livers at day 14 post-infection revealed a significant decrease of the bacterial burden in the peptide vaccinated groups as compared to non-vaccinated mice ( Figure 4A). The positive impact of the peptide vaccine on Salmonella disease was also reflected in the decreased serum level of alanine aminotransferase (ALT) ( Figure 4B), which is known to be indicative for liver damage [36] and previously reported to be elevated in human salmonellosis [37,38]. Analysis of the CD4 + T cell responses by ELISPOT revealed that the peptide vaccination resulted in a two-fold increase of the AhpC-specific CD4 + T cell response but the STM1540-specific CD4 + T cell response was decreased as compared to unvaccinated mice ( Figure 4C). Vaccination with the same mix of peptides in a prophylactic setting (i.e. 14 days before Salmonella challenge) however, did not result in reduction of bacterial load or ALT levels ( Figure 4 and 4B). Together these data demonstrate that boosting of CD4 + T cell responses during infection with peptide-based vaccines improves the clearance and burden of virulent Salmonella bacteria.

Virulent Salmonella infection elicits polyfunctional CD4 + T cell responses
The identification of differential cytokine expressing CD4 + T cells could lead to novel predictors of disease activity of virulent Salmonella infection as has recently been found for other pathogens such as HIV and Mycobacterium tuberculosis [39,40,41]. Furthermore, it has been shown that the quality of a CD4 + T cell cytokine response is able to define correlates of protection against pathogens [42,43]. To functionally characterize the CD4 + T cell response to virulent Salmonella, we determined by intracellular cytokine staining and polychromatic flow cytometry [44] whether Salmonella-specific CD4 + T cells can simultaneously produce IFNc, tumor necrosis factor (TNF), interleukin (IL)-2, IL-10 and/or IL-17.
The analysis of intracellular cytokine production by CD4 + T cells in which total splenocyte cultures were restimulated with peptide revealed that the background (peptide unstimulated) IFNc and TNF staining was very high (data not shown), which is likely attributed to the persistent presence of virulent Salmonella bacteria in antigen presenting cells. By using the same stimulation procedure as used in the ELISPOT assays to stimulate the Salmonella-specific CD4 + T cells (i.e. stimulation of purified CD4 + T cells by peptide-loaded DCs) the IFN-c and TNF background staining decreased considerably. A DC-T cell ratio of 1:5 and 1:10 was optimal, but ratios of 1:20 and 1:40 also elicited positive responses ( Figure S1). Importantly, at two, three and four weeks post-infection a population of high IFN-c producers was found exclusively in the peptide-stimulated samples, which were more easily visible by costaining for TNF ( Figure 5A and data not shown). This indicates that the IFN-c + TNF + CD4 + T cells most likely identify the epitope-specific population, and confirms the findings of Srinivasan et al [45]. Further examination of the polyfunctionality of the Salmonella-specific CD4 + T cells revealed that most of the IFN-c + TNF + cells co-produced IL-2 ( Figure 5B), demonstrating a predominant development into polyfunctional T cells with a T-helper type 1 (T H 1) signature [46]. Consistent with the immunodominance hierarchy, the mean fluorescence intensity (MFI) of IFN-c within the triple IFN-c + TNF + IL-2 + and double IFN-c + TNF + CD4 + T cells was the highest for the reactivity of these cells against the STM1540 262-276 epitope followed by AhpC 154-168 , EutC 243-257 and FliC 429-443 ( Figure 5C). The IFNc MFI of the single IFN-c + CD4 + T cells was consistently lower than the triple IFN-c + TNF + IL-2 + and double IFN-c + TNF + populations. Intracellular expression of IL-10 and IL-17A was neither found in the double producers nor found in the total CD4 + T cell population ( Figure 5D). Thus, polyfunctional CD4 + T cells that develop during Salmonella infection produce simultaneously IFN-c, TNF and IL-2 and the differential amounts of IFN-c that is being produced by these cells correlates with the immunodominance hierarchy.

Polyfunctional CD4 + T cell responses correlate to disease activity
To determine statistical correlations between the Salmonellaspecific CD4 + T cell responses and the disease activity of Salmonella, we plotted the CD4 + T cell frequency against the recoverable Salmonella CFU in the spleen. There was a significant correlation between the frequency of STM1540 262-276 -specific IFN-c + TNF + CD4 + T cells and the splenic bacterial load ( Figure 6A). Also the CD4 + T cell response against AhpC 154-168 but not the EutC 243-257 epitope correlated with the bacterial colonization (data not shown). Plotting of the STM1540 262-276specific IFN-c + TNF + CD4 + T cell frequency against the number of IFN-c + spots as determined by ELISPOT showed a strong correlation between these two assays for enumerating antigenspecific T cells ( Figure 6B). Accordingly, the Salmonella-specific CD4 + T cells responses as determined by IFN-c ELISPOT correlated to the disease activity of Salmonella ( Figure 6C). In conclusion, these data provide evidence that immunodominant CD4 + T cell responses that develop during virulent Salmonella infection are predictive for the disease activity.

Discussion
Our results indicate that polyfunctional CD4 + T cell responses, which develop during Salmonella infection, can predict important elements of disease activity. Although both CD4 + and CD8 + T cell populations were activated, the depletion of CD4 + but not CD8 + T cells affected bacterial clearance. The importance of CD4 + T cell activation was further demonstrated by the efficacy of therapeutic vaccination with the CD4 + T cell peptide epitopes during virulent Salmonella infection. The latter observation supports the use of T cell peptide vaccination as a potential strategy to combat ongoing virulent Salmonella infection. Our finding that the frequency of immunodominant Salmonella-specific CD4 + T cell responses can be used as an immunological predictor for the disease activity opens new avenues for diagnosis of Salmonella infection and may aid in the rational design of vaccines.
An important role for CD4 + T cell immunity is also found in other persistent infections that are bacterial (e.g. Mycobacterium tuberculosis), parasitic (e.g. Leishmania major) or viral (e.g. cytomegalovirus) [47,48,49,50]. Thus far, little is known about CD4 + T cell responses generated in response to virulent Salmonella. Here we identified novel CD4 + T cell epitopes to facilitate the study of Salmonella-specific T cell immunity. More than 1400 epitopes were predicted, of which, three elicited significantly elevated responses in Salmonella-infected C57BL/6 mice. To our knowledge, we provide here for the first time the identification of epitopes that are able to monitor CD4 + T cell responses during virulent Salmonella infection directly ex vivo. These epitopes were found in different Salmonella enzymes. Since these epitopes are located in conserved enzymes that are also present in other microbes, it might be that in these microbes the same epitopes elicit specific CD4 + T cell responses. Based on high in vivo expression levels Rollenhagen et al. identified Salmonella antigens for protective immunity but the epitope specificity remained unknown [51]. Other studies have identified epitopes in FliC (15) and recently two epitopes in the type III secretion system were reported [21]. FliC is transcriptionally switched off once Salmonella is inside macrophages, which renders this response not suitable to track T cell responses during virulent infection [52,53]. Yet, immunization with FliC protein results in development of protective CD4 + T cell immunity [19,54]. The epitopes within the type III secretion system proteins were discovered using a non-virulent Salmonella model and the CD4 + T cell responses to these epitopes in this model displayed similar kinetics as compared to the EutC 243-257 -specific response during virulent Salmonella infection [21].
One reason for the gap in knowledge regarding development of T cell immunity to Salmonella is that traditional CD4 + T cell antigen screening and epitope mapping methods are hampered by the large genome size of bacteria including Salmonella. Bioinformatic predictions of peptide:MHC class II binding have been successfully used to identify peptide epitopes genome wide in viruses such as vaccinia virus and mouse cytomegalovirus [31,55], and also in the known B cell antigen targets of the intracellular bacterium Coxiella burnetti [56]. The rate of identified CD4 + T cell epitopes in the tested Salmonella peptides was only ,0.2% (3/ 1443), which clearly is a low efficiency, and in fact lower than the identified CD4 + T cell epitopes in vaccinia virus (,0.7% hit rate) and much lower than the identification in mouse cytomegalovirus and the focused search in Coxiella burnetti (both ,7% hit rate). It is likely that additional T cell epitopes recognized in virulent Salmonella exist, as our screen was directed at identifying the most dominant peptide responses, which was necessitated by the relatively high level of background responses in the animals with virulent infection. Also, the limitation to screen only the top 0.5% of peptides will necessarily exclude some responses. Finally, the algorithms used were based on a limited training dataset, and if repeated nowadays would have picked a very different and presumably better set of candidate peptides, as evidenced by the fact that the FLiC peptide was not included in our screen, but would now be included. In summary, we believe that the responses identified are some of the most dominant, but we believe that multiple additional T cell epitopes exist. Another possible limitation to our screen is that it selected only IFN-c producing T cells. Additional CD4 + T cell epitopes may exist that do not induce IFN-c and/or TNF, but do trigger production of other cytokines.
The role of CD4 + T cells during Salmonella infection is likely twofold: 1) direct effector cell activity and 2) providing help for B cells and CD8 + T cells. Since IFN-c is a critical cytokine for controlling Salmonella infection, it is likely that CD4 + T cell derived IFN-c plays a particularly important role [12]. Besides producing IFN-c, direct killing activity of CD4 + T cells, which has been described in lymphocytic choriomeningitis virus [57], West Nile virus [58] and dengue virus [59] infection, might also be a function of these cells.
Since MHC class II-expressing macrophages are an important reservoir of Salmonella [60,61], it might be that these cells are the targets. Further studies are required to discriminate between these different effector functions.
In conclusion, we demonstrate the feasibility of MHC class II binding predictions to identify CD4 + T cell responses during virulent Salmonella infection. The prospect of predicting the burden of Salmonella infections based on the responses against these newly identified CD4 + T cell epitopes could lead to new diagnostic tools and reagents for tracking and analyzing Salmonella-specific T cell responses during Salmonella disease or after vaccination. While the exact epitopes identified in mice are not likely to be recognized in humans due to differences in MHC specificity, their source proteins are more likely to be prevalently recognized in humans [62]. Further studies that aim to predict human CD4 + T cell epitopes (e.g. HLA DR1/3/4) could translate into direct clinical applications. The results obtained provide novel insights into the mechanisms of T cell immunity to virulent Salmonella and could therefore aid in the diagnostics of Salmonella infection and in a rational development of (subunit) vaccines to Salmonella serovars.

Mice
C57BL/6 mice were purchased either from The Jackson Laboratory (Bar Harbor, ME) and maintained at the Department of Laboratory Animal Care at LIAI or were purchased from Charles River (France) and maintained at the Central Animal Facility of LUMC. All mice were housed in specific pathogen-free conditions and used at 8-12 weeks of age.

Salmonella infection and quantification of disease activity
The virulent Salmonella enterica serovar Typhimurium (S. Typhimurium) strain LT2 was obtained from the American Type Culture Collection (ATCC). For infections, S. Typhimurium was grown to mid-log phase in Luria-Bertani media at 37uC with shaking and subcultures were setup to grow to early-log phase (OD 600 nm of ,0.25). The bacterial concentration was estimated using a spectrophotometer and the bacteria were diluted in PBS. Mice were orally infected with 1610 7 Salmonella colony forming   CD4 + T cell epitope prediction The entire Salmonella proteomes of S. Typhimurium strain LT2 [28] and S. Typhi strain CT18 [29] were scanned for conserved 15-mer peptide sequences that are predicted, using a consensus approach as described [31], to have a high-affinity binding capacity for the MHC class II molecule H2 I-A b . Briefly, predictions were obtained from the ARB and SMM-align tool on the immune epitope database (IEDB) website [63], and all peptides were ranked according to their predicted affinity by each method. To construct a consensus from the prediction methods, the median of the ranks was used to select peptides for screening. A set of 1443 different peptides with the highest median ranks corresponding to the top ,0.5% scoring peptides were selected for screening and were synthesized as crude material (A&A Labs, San Diego, CA) and used in initial screening experiments. Peptides AhpC 154-168 , EutC 243-257 , STM1540 262-276 (see Table 1 for amino acid sequences), and FliC 429-443 (VQNRFNSAITNLGNT [19]) used in subsequent experiments were re-synthesized as purified material.

Preparation of peptide-loaded dendritic cells
Bone marrow cells were harvested from wild-type C57BL/6 mice and resuspended in culture medium (Iscove's Modified Dulbecco's Medium (IMDM), 10% heat inactivated fetal calf's serum, 2 mM L-Glutamine, and 2-mercaptoethanol), which was supplemented with 200 ng/ml Flt-3 ligand (Peprotech). After 10 days, dendritic cells were harvested and either loaded with peptides (10 mg/ml) for 90 minutes at 37uC or not loaded with peptides (medium control). After extensive washing, peptideloaded and unloaded dendritic cells were used for ELISPOT assays and intracellular cytokine staining.

ELISPOT Assay
Antigen-specific IFN-c recall responses were measured by enzyme-linked immunospot (ELISPOT). 2610 5 CD4 + T cells, purified by CD4 magnetic microbeads (Miltenyibiotec), were cultured with 4610 4 peptide-loaded dendritic cells in a total volume of 100 ml culture medium in Multiscreen 96-well plates (Millipore) that were precoated overnight at 4uC with 5 mg/ml rat- anti-mouse IFN-c. Background values were determined by culturing purified CD4 + T cells with unloaded dendritic cells. After 16 h incubation at 37uC, the plates were washed with PBS containing 0.05% Tween 20, followed by incubation with 1 mg/ml biotinylated IFN-c antibody at room temperature for two hours. After washing, extravidine (Sigma) and BCIP/NBT were used to visualize spots. The frequency of IFN-c producing cells was quantified by ELISPOT Readers (Autoimmun Diagnostika GmbH, Strassberg, Germany). Wells were considered positive if they yielded values .2 times above mean background plus two standard deviations.

Flow cytometric analysis and intracellular staining
Spleens were harvested and single-cell suspensions were prepared by mincing through 70 mm cell strainers (BD Biosciences). Erythrocytes were lysed in a hypotonic (0.82%) ammonium chloride buffer. For cell surface staining, cells were resuspended in staining buffer (PBS +1% FCS +0.05% sodium azide) and incubated with fluorescent conjugated antibodies for 30 minutes at 4Cu in presence of Fc block (monoclonal antibody to mouse CD16/CD32). To determine the activation status on conventional CD4 + T cells, regulatory T cells were excluded by staining for intracellular Foxp3 using the anti-mouse Foxp3 staining kit (eBioscience). For determination of intracellular cytokine expression, 4610 5 purified CD4 + T cells were cultured with 8610 4 peptide-loaded dendritic cells in 96-well flat-bottom plates in a total volume 200 ml culture medium for 8 h of which the last 6 h were in presence of 1 mg/ml brefeldin A. Background was determined by culturing purified CD4 + T cells with unloaded dendritic cells (medium control). After incubation, cells were transferred to Ubottom 96-well plates, and the cell surface was stained with fluorescent conjugated antibodies at 4Cu for 0.5 h in staining buffer. After washing, cells were fixed with Fix/perm buffer for 1 h followed by intracellular cytokine staining at 4Cu for 1 h in Perm/ Wash buffer (BD Biosciences). Cells were washed and resuspended in staining buffer before analysis. Flow cytometric acquisition was performed with a BD LSR II flow cytometer and data was analyzed using FlowJo software (Tree Star). Fluorochrome-conjugated monoclonal antibodies specific for CD3, CD4, CD8, CD62L, CD69, KLRG1, IFN-c, TNF, IL-2, IL-10 and IL-17 were purchased from BD Biosciences (San Diego, CA) or eBioscience (San Diego, CA).

Statistical analysis
Statistical significance between groups was determined with the Mann-Whitney test and correlations were analyzed with linear regression. We used Prism GraphPad Software for all statistical analyses. Figure S1 Optimization of cytokine expression in Salmonella-specific CD4 + T cells. C57BL/6 mice were infected orally with 1610 7 virulent Salmonella Typhimurium and at day 14 post-infection the IFN-c and TNF expression in Salmonellaspecific CD4 + T cells was determined by intracellular staining after restimulation with peptide-loaded dendritic cells (DCs). (A) Titration of peptide dose. CD3 + CD4 + T cells were stimulated with DCs (DC:T cell ratio = 1:5) that were loaded with different concentrations of STM1540 262-276 peptide or not loaded with peptide (medium). Graph shows the percentage of IFN-c + TNF + cells within the CD3 + CD4 + T cell population plotted against the indicated peptide concentrations. (B) Titration of DC:T cell ratio. 4610 5 CD3 + CD4 + T cells were stimulated with different amounts of DCs that were loaded with 10 mg/ml STM1540 262-276 peptide or not loaded with peptide (medium). Graph shows the percentage of IFN-c + TNF + cells within the CD3 + CD4 + T cell population versus the different DC:T cell ratios. Shown are representative analyses of two independent experiments with three mice per experiment.