Culling of activated CD4 T cells during typhoid is driven by Salmonella virulence genes

Abstract

Pathogen-specific CD4 T cells are activated within a few hours of oral Salmonella infection and are essential for protective immunity. However, CD4 T cells do not participate in bacterial clearance until several weeks after infection, suggesting that Salmonella can inhibit or evade CD4 T cells that are activated at early time points. Here, we describe the progressive culling of initially activated CD4 T cells in Salmonella-infected mice. Loss of activated CD4 T cells was independent of early instructional programming, T cell precursor frequency, and antigen availability. In contrast, apoptosis of antigen-specific CD4 T cells was actively induced by live bacteria in a process that required Salmonella Pathogenicity Island-2 (SPI-2) and correlated with increased expression of PD-L1. These data demonstrate efficient culling of initially activated antigen-specific CD4 cells by a microbial pathogen and document a novel strategy for bacterial immune evasion.

Introduction

Control and clearance of a microbial pathogen requires the generation of an appropriate effector response by coordinated activation of innate and adaptive immunity. In order to prolong survival within the infected host, many successful pathogens have evolved countermeasures to avoid initial immune activation or modulate downstream effector responses (1–3). These strategies are collectively referred to as “immune evasion” and can have a profound effect on pathogen virulence (4–7).

Salmonella enterica is an extremely effective pathogen that causes human gastroenteritis, typhoid fever, nontyphoidal bacteremia, and associated veterinary infections, each of which have significant health and economic impacts (8–11). Salmonella is a facultative intra-macrophage pathogen and the generation of a robust Th1 response for macrophage activation is a prerequisite for protective immunity. In the mouse model of typhoid, clearance of Salmonella enterica serovar typhimurium requires production of IFN-γ by CD4 T cells (12, 13). Similarly, patients with genetic deficiencies in IFN-γ and IL-12 receptor signaling, or receiving anti-IL-12 therapy, display increased susceptibility to Salmonella (14–16), indicating that Th1 responses are protective in human Salmonellosis.

Given the importance of Th1 cells for bacterial clearance, it seems likely that Salmonella have evolved mechanisms to inhibit cellular immunity. Indeed, numerous in vitro studies demonstrate that Salmonella can inhibit activation of naïve CD4 T cells (17–24). In contrast, in vivo studies report efficient activation of polyclonal and monoclonal Salmonella-specific CD4 T cells following oral or intravenous infection (25–28). The discrepancy between in vitro and in vivo data could represent an intrinsic limitation of in vitro systems to model immunity to Salmonella. However, it is also possible that inhibitory effects of live bacteria on CD4 T cells actually do occur in vivo, but primarily influence the maturation of a Th1 response rather than the initial activation events.

Indeed, there is circumstantial in vivo evidence for Salmonella inhibition of CD4 responses. Salmonella-specific Th1 cells are activated rapidly and secrete IFN-γ, but surprisingly do not contribute to bacterial clearance until several weeks after infection. For example, mice with a deficiency in CD4 T cells or the development of Th1 cells can control bacterial loads similarly to Wild-type mice during the first weeks of Salmonella infection (12, 13). The discordance between early pathogen-specific CD4 immune activation and delayed CD4-mediated bacterial clearance has not yet been examined in any detail.

Here, we examined the effect of Salmonella infection on the maturation of CD4 T cell responses in vivo and can confirm that Salmonella infection does not interfere with early T cell activation. However, Salmonella infection causes profound depletion of activated Salmonella-specific CD4 T cells such that very few of these cells survive to enter the CD4 effector/memory pool. “Culling” of activated CD4 T cells in infected mice was not dependent on T cell precursor frequency, availability of antigen, or early activation events in vivo. In contrast, “culling” involved enhanced apoptosis of clonally expanded CD4 T cells, required expression of Salmonella Pathogenicity Island (SPI)-2 genes, and correlated with enhanced PD-L1 expression by activated Salmonella-specific T cells. Together these data demonstrate the depletion of activated Th1 cells via the expression of Salmonella virulence genes in vivo.

Materials and Methods

Mouse strains

C57BL/6 mice were purchased from the National Cancer Institute (Frederick, MD) and used at 6–16 weeks of age. Rag-deficient, CD45.1 or CD90.1 congenic, SM1 T cells specific for Salmonella flagellin have been described (28, 29), and were intercrossed in our animal facility. Rag-deficient, CD90.1 congenic, OT-II (30) and TEa (31) TCR transgenic mice were also bred in our facility. All mice were cared for in accordance with University of Minnesota Research Animal Resource guidelines and experiments were approved by the Institutional Animal Care and Use Committee.

Bacterial strains, antigens, and peptide

BRD509 AroA⁻D⁻(32) was provided by Dr. D. Xu, University of Glasgow, U.K. Recombinant Salmonella strains expressing OVA or Eα have been described (33, 34). Salmonella mutants with deficiencies in the phoP/Q two-component regulatory system (PhoPC), ssaV, an essential component of SPI2 type II secretion system apparatus (HH109, ssaV::aphT) (35), or spv, located on the virulence plasmid (P5D10, spvA::mTn5) (36), were provided by Dr. A. Gewirtz, Emory University, Atlanta, GA and Dr. D. Holden, Imperial College London, London, UK. Heat-killed Salmonella serovar typhimurium (HKST) was prepared by re-suspending bacteria in PBS at a concentration of 5×10⁸/ml, heating at 75°C for 1 hour, and plating to confirm the absence of live bacteria. Flagellin peptide (427–441) recognized by SM1 T cells (37) was purchased from Invitrogen (Carlsbad, CA).

Salmonella infection

Salmonella strains were cultured in LB broth, bacterial numbers estimated from OD600, recovered by centrifugation, and re-suspended in PBS at the appropriate concentration. Mice were injected intravenously in the lateral tail vein with 5×10⁵ Salmonella and monitored daily. In every experiment, the administered bacterial dose was confirmed by plating serial dilutions onto MacConkey agar plates and incubating overnight at 37°C.

Adoptive Transfer of Salmonella-specific CD4 T cells

Spleens and lymph nodes were pooled from CD45.1 or CD90.1 congenic SM1, OT-II, or TEa TCR transgenic mice and stained for 10 minutes at 37°C with CFSE before adoptive transfer into C57BL/6 mice. Recipient mice were injected IV with 2×10²–2×10⁷ TCR transgenic cells depending on the experiment.

Flow Cytometric Analysis

Spleens were recovered from infected and immunized mice and a single cell suspension generated. Spleen cells were incubated for 20–45 minutes at 4°C in Fc block (spent culture supernatant from the 24G2 hybridoma, 2% rat serum, 2% mouse serum, and 0.01% sodium azide) in the presence of primary antibodies. Fluoroscein isothiocyanate-(FITC), phycoerytherin-(PE), CyChrome-, PE-Cy5-, or allophycocyanin-conjugated antibodies specific for CD4, CD11a, CD45.1, CD45RB, CD62L, CD69, CD90.1, CD127, PD-1, PD-L1, PD-L2, CTLA-4, and isotype control antibodies were purchased from eBioscience (San Diego, CA) and BD Biosciences (San Diego, CA). After surface staining, cells were fixed and analyzed by flow cytometry using a FACS Canto (BD Biosciences). Data was analyzed using FlowJo software (TreeStar, San-Carlos, CA). Poly-caspase staining was detected by flow cytometry using the FLICA apoptosis detection kit according to manufacturers instructions (Immunochemistry Technologies, Bloomington, MN).

Results

Depletion of activated Salmonella-specific T cells in mice infected with live bacteria

Several reports have documented Salmonella inhibition of antigen presentation and/or naïve T cell activation using in vitro culture systems (17–24). In order to examine this process in vivo, we monitored the initial activation of Salmonella-specific SM1 T cells in response to live or Heat-killed Salmonella (HKST). SM1 T cells increased surface expression of the early activation marker CD69 six hours after exposure to either live Salmonella or HKST (Fig. 1A), and expanded in the spleen of both groups of mice three days later (Fig. 1B). Therefore, in contrast to in vitro culture systems, efficient antigen presentation and naïve T cell activation occurs in Salmonella-infected mice.

Figure 1. Salmonella-specific CD4 T cells are depleted in mice infected with live Salmonella.

(A) C57BL/6 mice were adoptively transferred with 1×10⁶ CFSE-stained CD90.1 SM1 T cells and infected intravenously the following day with 5×10⁵ Salmonella, BRD509 or immunized with 5×10⁵ HKST. Six hours later, the percentage of SM1 T cells (CD4⁺CD90.1⁺), surface CD69 expression, and CFSE-dye dilution was examined in the spleen of uninfected (Transfer Only), infected, or immunized mice. Lower plots show CFSE and CD69 staining after gating the boxed SM1 gate in upper plots. Bar graph shows the mean percentage +/− SD CD69⁺ SM1 cells of 3–4 mice per group. (B) C57BL/6 mice were adoptively transferred with 1×10⁶ CD45.1 SM1 T cells and infected or immunized with HKST. Three or ten days later, the percentage of SM1 T cells (CD4⁺CD45.1⁺) was determined in the spleen. (C) C57BL/6 mice were adoptively transferred with 1×10⁶ CD90.1 OT-II T cells and infected with Salmonella-OVA or immunized with HKST-OVA. Three or ten days later, the percentage of OT-II T cells (CD4⁺CD90.1⁺) was determined in the spleen. Boxes and numbers in all plots show SM1 or OT-II T cells as a percentage of all spleen cells. All plots are representative of 3 mice per group and at least 2 separate experiments.

In mice immunized with HKST, SM1 T cells expanded, contracted, and were detected at elevated frequency 10 days later (Fig. 1B). In contrast, very few surviving SM1 T cells were found in mice 10 days after infection with live Salmonella (Fig. 1B). SM1 T cells recognize a peptide of Salmonella flagellin (37), a bacterial antigen that is expressed at low levels in vivo (38, 39). We examined whether loss of flagellin-specific T cells was unique to this particular antigen by tracking the response of OVA-specific OT-II and Eα-specific TEa T cells to recombinant Salmonella expressing OVA or Eα. As with SM1 cells, OT-II and TEa T cells expanded after infection with Salmonella-OVA or Salmonella-Eα, but did not persist at later time points (Fig. 1C, and data not shown). In marked contrast, OT-II and TEa T cells persisted in mice immunized with HKST-OVA or HKST-Eα (Fig. 1C and data not shown). Together, these data demonstrate that activated Salmonella-specific CD4 T cells do not survive in infected mice, that depletion is independent of T cell antigen specificity, and that it occurs after the peak of clonal expansion.

Culling of activated CD4 T cells in infected mice is unaffected by initial clonal frequency

The initial frequency of naïve antigen-specific CD4 and CD8 T cells can profoundly affect the development of memory responses (40–42). We examined whether initial naïve frequency influenced survival of SM1 T cells in infected mice. C57BL/6 mice were adoptively transferred with SM1 T cells across a 100,000-fold range (200-2×10⁷) prior to infection with live Salmonella or immunization with dead bacteria. CFSE⁺ SM1 T cells were detected in the spleen of naive mice transferred with 2×10⁵–2×10⁷ SM1 cells but were below the limit of detection in mice receiving lower transfers (Fig. 2-Transfer Only). Expanded SM1 T cells were detected 20 days after HKST immunization in mice that had been transferred with 2×10³–2×10⁶ SM1 cells (Fig. 2-HKST). However, in agreement with previous reports (40, 41), the transfer of 10-fold more naive SM1 cells paradoxically reduced the frequency of surviving T cells after immunization (Fig. 2-HKST). As with HKST responses, clonal expansion of SM1 T cells to live Salmonella was detected at day 3 or day 7 in mice transferred with 2×10³–2×10⁷ SM1 cells. However, at day 20 few surviving SM1 cells were detected in any mouse that had been infected with live Salmonella, whether transferred with high or low numbers of SM1 cells (Fig. 2-Salmonella Day 20). These data suggest that naïve clonal frequency does not explain loss of activated SM1 T cells in Salmonella-infected mice.

Figure 2. SM1 T cell clonal frequency does not affect loss in infected mice.

C57BL/6 mice were adoptively transferred with 2×10²–2×10⁷ CD90.1 SM1 T cells and infected or immunized intravenously the following day. Three days later, the frequency of SM1 T cells (CD4⁺CD90.1⁺) was determined in the spleen of uninfected (Transfer Only), infected, or immunized mice. Boxes and numbers in plots show SM1 cells as a percentage of all spleen cells. All plots are representative of 3 mice per group and 3 separate experiments.

Depletion of SM1 T cells is not driven by antigen availability in vivo

Salmonella flagellin is poorly expressed during intracellular bacterial growth in the infected host (39, 43). We examined whether additional antigen would allow greater survival of SM1 memory T cells in infected mice. However, injection of flagellin peptide at days 3, 5 and 7 after infection did not increase the frequency of surviving SM1 T cells (Fig. 3A-Infected+peptide). We also constructed a recombinant Salmonella strain that over-expresses flagellin under the control of a constitutive promotor. However, infection with this Salmonella-Flag strain did not increase survival of SM1 cells following clonal expansion (Fig. 3B). Therefore, providing additional antigen does not reverse the loss of activated SM1 T cells in Salmonella-infected mice. We also considered whether excessive antigen stimulation during live infection could adversely affect activated SM1 cell survival. However, repeated HKST immunization mildly enhanced SM1 memory at day 20, demonstrating that prolonged presentation of cognate peptide/MHC is not deleterious to SM1 survival in vivo (Fig. 3C). Thus, insufficient or excessive antigen presentation is unlikely to be the cause of SM1 T cell loss in mice infected with live Salmonella.

Figure 3. Modulation of antigen availability does not influence SM1 survival in infected mice.

(A) C57BL/6 mice were adoptively transferred with 1×10⁶ CD90.1 SM1 T cells and infected, immunized, or infected and immunized intravenously the following day. Some infected mice were injected intravenously with 200μg of flagellin peptide, 3, 5, and 7 days after infection (Infected + peptide). Twenty days later, the frequency of SM1 T cells (CD4⁺CD90.1⁺) was determined in the spleen. (B) C57BL/6 mice were adoptively transferred with 1×10⁶ CD45.1 SM1 T cells and infected the following day with Salmonella (BRD509) or recombinant Salmonella over-expressing flagellin (Salmonella-Flag). Three and seven days later, the frequency of SM1 T cells (CD4⁺CD90.1⁺) was determined in the spleen. (C) C57BL/6 mice were adoptively transferred with 1×10⁶ CD90.1 SM1 T cells and immunized the following day with HKST or immunized with HKST every day for 6 days (HKSTx6). Twenty days later, the frequency of SM1 T cells (CD4⁺CD90.1⁺) was determined in the spleen. Boxes and numbers in all the above plots show SM1 cells as a percentage of all spleen cells. All plots are representative of 3 mice per group and at least 2 separate experiments.

Loss of Salmonella-specific T cells is due to enhanced apoptosis during clonal contraction

Signals delivered during early activation can determine the extent of proliferation and memory development for CD8 T cells (44–46), and CD127 or CD8αα expression have been used to identify activated cells that will survive to memory (47, 48). We considered the possibility that loss of SM1 cells in infected mice was due to a deficiency in early activation events. First, we examined the expression of surface markers on SM1 cells at the peak of clonal expansion. However, SM1 cells responding to either HKST or live bacteria displayed similar levels of CFSE-dye dilution, and expression of CD11a, CD62L, CD127 and CD45RB at this time (Fig. 4). Next we tested directly whether SM1 survival was determined by signals delivered before or after clonal expansion. C57BL/6 mice were adoptively transferred with CD90.1⁺ or CD45.1⁺ SM1 T cells and immunized with HKST or infected with live Salmonella. T cells were purified from mice three days later, mixed together in a 1:1 ratio, and retransferred into infected or HKST-immunized mice (Fig. 5A). Twelve days after infection or immunization we examined whether CD45.1 or CD90.1 SM1 T cells had survived in each group. As expected, SM1 T cells from HKST-immunized mice that had been transferred back into HKST-immunized mice had a detectable SM1 memory population, while SM1 T cells from Salmonella-infected mice that were returned to Salmonella-infected mice did not (Fig. 5B). Conversely, SM1 cells from HKST-immunized mice that were transferred to Salmonella-infected mice were not detectable at day 12, while SM1 cells from Salmonella-infected mice but transferred to HKST-immunized mice were detected at an elevated frequency (Fig. 5B). The absolute number of SM1 T cells recovered from Salmonella-infected mice was also substantially lower, indicating that low frequency was not simply a result of Salmonella-induced splenomegaly (Fig. 5C). These data demonstrate that the signals causing the loss of Salmonella-specific CD4 T cells are delivered after clonal expansion has occurred in infected mice.

Figure 4. Comparison of activation markers on SM1 cells in infected and immunized mice.

C57BL/6 mice were adoptively transferred with 1×10⁶ CFSE-labeled CD90.1 SM1 T cells and infected or immunized intravenously the following day. Three days later, the expression of CD11a, CD127, CD62L, and CD45RB was determined on SM1 T cells in each group. (A) Plots show CFSE staining and activation marker expression after gating on SM1 cells. All plots are representative of 3 mice per group and 3 separate experiments. (B) Bar graph shows the mean percentage +/− SD of surface marker expression for CD11a, CD127, CD62L, and CD45RB on SM1 T cells for 3–4 mice per group.

Figure 5. Depletion of SM1 T cells occurs after the peak of clonal expansion.

(A) Model of adoptive transfer and sorting experiment. (B) C57BL/6 mice were adoptively transferred with 1×10⁶ CD90.1 SM1 T cells and immunized intravenously the following day. Another group of mice were adoptively transferred with 1×10⁶ CD45.1 SM1 T cells and infected intravenously the following day. Three days later, spleen and lymph node cells were harvested and CD4 T cell enriched using magnetic beads. Enriched CD4 cells were stained with antibodies to CD45.2 and CD90.2 to label endogenous cells and the negative fraction containing SM1 cells FACS sorted and transferred back into infected or immunized mice. Twelve days later SM1 T cells were detected in the spleen of infected or immunized mice as described above. Boxes and numbers in plots show SM1 cells as a percentage of all spleen cells. Plots are representative of 3 mice per group and 2 separate experiments. (C) Mean percentage and absolute numbers of SM1 T cells +/− SD for the experiment described above.

We also examined whether there was a difference in the number of apoptotic SM1 T cells in mice infected with live Salmonella versus HKST. Adoptively transferred mice were infected with Salmonella or immunized with HKST and the ex-vivo poly-caspase FLICA stain was used to examine apoptotic SM1 T cells during the contraction phase of the response. Over 50% of detectable SM1 T cells in Salmonella-infected mice stained positive with the FLICA stain compared to less than 10% in mice immunized with HKST (Fig. 6). Thus, the loss of SM1 T cells in Salmonella-infected mice is due to enhanced apoptosis of cells following the peak of clonal expansion.

Figure 6. Enhanced apoptosis of SM1 T cells in infected mice.

C57BL/6 mice were adoptively transferred with 1×10⁶ CFSE-labeled CD90.1 SM1 T cells and infected or immunized intravenously the following day. Five days later, polycaspase staining was detected in SM1 T cells recovered from the spleen of each group of mice using a FLICA detection kit. Plots show FLICA and CFSE staining after gating on SM1 cells. All plots are representative of 3 mice per group and 3 separate experiments.

Culling of activated SM1 cells requires expression of Salmonella virulence genes and correlates with elevated PD-L1 expression in vivo

Given the kinetics of activated T cell depletion, it seemed possible that the increased apoptosis of activated CD4 T cells was an active process driven by Salmonella virulence genes in vivo. Therefore, we examined whether live bacteria cause depletion of SM1 T cells in mice co-immunized with HKST. C57BL/6 mice were adoptively transferred with SM1 T cells and immunized with HKST, infected with live Salmonella, or simultaneously immunized and infected. SM1 T cells did not survive in mice co-administered HKST and live bacteria (Fig. 3A-HKST+Salmonella), demonstrating that culling of activated SM1 T cells is an active process that hinders the response to immunization.

In order to define this process of bacterial driven T cell depletion in more detail, we examined whether “culling” of SM1 cells occurred in mice administered Salmonella mutants that lacked certain virulence genes. Many of these mutants are not fully attenuated, limiting the number of time points that can be used to examine SM1 survival. However, SM1 cells expanded in response to all Salmonella mutants examined but were still profoundly depleted by day 6 post-infection in mice administered AroA-D-deficient, PhoP-deficient, or Spv-deficient bacteria (P5D10) (Fig. 7A). In contrast, an elevated frequency of SM1 T cells was detected in mice infected with a Salmonella strain lacking the SPI-2 needle complex (Fig. 7A and B, HH109). The HH109 mutant is not constructed on a BRD509 background, complicating a direct comparison of these strains and the exact role of SPI2 genes in the loss of SM1 cells. Unfortunately, we do not have an SPI2 mutant strain on the BRD509 background. However, enhanced survival of T cells was not due to differences in the inflammatory response accompanying different bacterial loads since at day 6 post infection HH109-infected mice had approximately 10 times more bacteria that BRD509-infected mice but 10 times less than PhoPC-infected mice (data not shown). Thus, “culling” of activated SM1 T cells is an active process requiring in vivo expression of an SPI-2 encoded Type-III secretion system.

Figure 7. Depletion of SM1 T cells in infected mice requires Salmonella SPI2 genes and correlates with elevated PD-L1 staining.

(A) C57BL/6 mice were adoptively transferred with 1×10⁶ CFSE-labeled CD45.1 SM1 T cells and infected the following day with various Salmonella mutants. Three and six days later the frequency of SM1 T cells was determined in the spleen of each group of mice. Boxes and numbers in plots show SM1 cells as a percentage of spleen cells. Plots are representative of 3 mice per group and 3 separate experiments. (B) Bar graph showing mean percentage +/− SD of surviving SM1 T cells at day 6 for 3–4 mice per group. Star indicates mean SM1 percentage in HH109-infected mice is statistically greater than other groups (p<0.05 by students unpaired t test). (C) C57BL/6 mice were adoptively transferred with 1×10⁶ CD90.1 SM1 T cells and infected the following day with BRD509 (Salmonella) or HH109 (SPI2-deficient Salmonella), or immunized with HKST. Three days later the expression of PD-L1 was examined on SM1 T cells recovered from the spleen of each group of mice. Shaded grey histograms show PD-L1 staining on naïve SM1 T cells while black lines show staining on SM1 cells recovered from infected or immunized mice. Plots are representative of 3 mice per group and 3 separate experiments.

Recent reports demonstrate that T cells are negatively regulated by signals delivered via Programmed cell death-1 (PD-1) and ligands for this molecule (49, 50). Given the inhibition of CD4 responses in vivo, we examined whether Salmonella infection induced altered expression of inhibitory molecules on SM1 T cells. Surface expression of PD-1, PD-L2, and CTLA-4 was similar on expanded SM1 T cells recovered from infected or immunized mice (data not shown). However, the inhibitory receptor PD-L1 was highly expressed on SM1 T cells in Salmonella-infected mice compared with HKST-immunized mice (Fig. 7C). Furthermore, this elevated expression on SM1 T cells required expression of SPI-II genes by Salmonella in vivo (Fig. 7C). Therefore, live Salmonella induce expression of the inhibitory ligand PD-L1 on Salmonella-specific CD4 T cells in a process that requires expression of the SPI-2 virulence genes. Together, these data suggest a model where live Salmonella can induce “culling” of activated CD4 T cells via the function of the SPI-2 encoded type-III secretion system and the induction of PD-L1.

Discussion

Microbial pathogens adopt a wide array of strategies to limit initial activation or the effector function of pathogen-specific T cells (4–7). Protective immunity to an intra-macrophage pathogen requires development of a functional CD4 Th1 response, suggesting that these cells could be targeted during these infections.

Indeed, numerous reports have demonstrated the ability of Salmonella to inhibit CD4 T cell responses (17–24). These experiments used in vitro culture systems to examine Salmonella inhibition of naïve CD4 T cell activation. However, in vivo data presented above clearly demonstrate that live Salmonella do not inhibit naive CD4 T cell activation, thus confirming and extending reports by our laboratory and others (25–28). Why would Salmonella inhibit T cell activation in vitro but not affect the response in the lymphoid tissue of an intact animal? It seems likely that the cultured dendritic cells used for in vitro cultures do not reproduce the diversity of subsets that are found in vivo or do not model other important aspects of lymphoid anatomy within infected tissues (51, 52). Furthermore, local inflammatory responses by innate cells within infected tissues may well enhance antigen presentation and T cell activation, overcoming the bacterial inhibitory effects that are detected in vitro (53). Whatever the reason, conclusions about bacterial inhibition of immunity should be viewed with some degree of caution unless they can be confirmed in vivo.

Despite the initial activation of Salmonella-specific T cells, our data demonstrate that few of these CD4 T cells survive during the clonal contraction phase in infected mice. Several factors that are known to influence development of memory T cells including antigen availability (54), naïve clonal frequency (40–42), and the expression of CD127 (47, 55), had no demonstrable effect on Salmonella-specific CD4 T cell survival in infected mice. Furthermore, the nature of the antigen recognized by SM1 T cells was unimportant, since OVA-specific and Eα-specific CD4 T cells were depleted in Salmonella-infected mice. Our conclusion from these studies is that the process of T cell depletion in Salmonella-infected mice is not simply a consequence of ineffective immune activation but more likely to be an active process induced by live bacteria. Indeed, “culling” of activated Salmonella-specific T cells was due to enhanced apoptosis during the clonal contraction phase and was able to inhibit the development of an effective SM1 response to HKST-immunization. Thus, Salmonella are able to limit the development of cellular immunity by targeted inhibition of CD4 T cells after clonal expansion has occurred. These data may also provide an explanation for the delayed development of CD4-mediated clearance of bacteria and the recent observation that B cell germinal center development is also delayed during Salmonella infection (12, 13, 56).

The T cell “culling” detected in this model required the function of Salmonella SPI2 virulence genes by live bacteria. The SPI2 locus encodes a type III secretion system that is expressed by Salmonella within macrophages to deliver effector proteins into the cytosol from the Salmonella-containing vacuole (57). There are 16–18 known SPI2 effector proteins and the function of many of these is currently unclear. Expression of the SPI-2 type III secretion system was also required for enhanced expression of PD-L1 on the surface of expanded SM1 T cells in infected mice. Increased expression of PD-L1 can be induced in response to inflammatory cytokines (58, 59), and could therefore be a secondary consequence of SPI2-mediated effects on cytokine production by Salmonella-infected macrophages. However, it was recently reported that SPI2 effector proteins are injected into non-phagocytic cells, including CD4 T cells, during Salmonella infection (60), although how this occurs in vivo is currently unknown. Therefore, it is possible that CD4 T cell depletion involves direct or indirect effects of SPI-2 effector proteins in vivo. Although PD-L1 can deliver an inhibitory signal to T cells via the receptor PD-1 (49), recent data suggest that ligation of PD-L1 expressed on T cells can also result in an inhibitory signal (50).

Together, our data suggest a novel model where Salmonella use SPI2 effector proteins to increase PD-L1 expression on activated CD4 T cells, increase the rate of apoptosis during clonal contraction, and therefore “culling” a potent effector population. Future work will be needed to determine which of the SPI2 effector proteins is responsible for this effect in vivo. Given the eventual development of Salmonella-specific CD4 T cells that contain bacterial growth several weeks after infection (12, 13), this potent inhibitory effect of SPI2 effector proteins is only active during the early stages of Salmonella infection or is eventually overcome by innate immune activation (61). However, efforts to block inhibition of CD4 T cell development by Salmonella may be beneficial for the development of effective cellular immunity to typhoid and may therefore aid the development of live vaccine strains.