Salmonella polarises peptide-MHC-II presentation towards an unconventional Type B CD4+ T-cell response

Distinct peptide-MHC-II complexes, recognised by Type A and B CD4+ T-cell subsets, are generated when antigen is loaded in different intracellular compartments. Conventional Type A T cells recognize their peptide epitope regardless of the route of processing, whereas unconventional Type B T cells only recognise exogenously supplied peptide. Type B T cells are implicated in autoimmune conditions and may break tolerance by escaping negative selection. Here we show that Salmonella differentially influences presentation of antigen to Type A and B T cells. Infection of bone marrow-derived dendritic cells (BMDCs) with Salmonella enterica serovar Typhimurium (S. Typhimurium) reduced presentation of antigen to Type A T cells but enhanced presentation of exogenous peptide to Type B T cells. Exposure to S. Typhimurium was sufficient to enhance Type B T-cell activation. Salmonella Typhimurium infection reduced surface expression of MHC-II, by an invariant chain-independent trafficking mechanism, resulting in accumulation of MHC-II in multi-vesicular bodies. Reduced MHC-II surface expression in S. Typhimurium-infected BMDCs correlated with reduced antigen presentation to Type A T cells. Salmonella infection is implicated in reactive arthritis. Therefore, polarisation of antigen presentation towards a Type B response by Salmonella may be a predisposing factor in autoimmune conditions such as reactive arthritis.


Introduction
Salmonella enterica is an intracellular pathogen that survives and replicates in phagocytic cells within specialised compartments known as Salmonella-containing vacuoles (SCV) [1]. Following oral ingestion, Salmonella crosses the intestinal epithelium by encode type III secretion systems (T3SS) that inject bacterial effector proteins into host cells. T3SS-1 is encoded within SPI1 and is required for invasion of host cells, whereas T3SS-2 is encoded by SPI2 and contributes to immune evasion and maintenance of the SCV by intracellular Salmonella [4]. Salmonella enterica serovars such as Typhimurium (S. Typhimurium) and Enteritidis cause rapid-onset gastroenteritis in a range of species, whereas serovars such as Typhi and Paratyphi cause systemic typhoid fever in humans. Salmonella Typhi can establish lifelong infection of the gall bladder in 1-4% of patients. These typhoid carriers exhibit normal antibody responses to Salmonella Typhi antigens but have an impaired cell-mediated immune response [5].
MHC-II molecules play an essential role in the cell-mediated immune response by presenting antigenic peptides to CD4 + T cells. Immature MHC-II molecules are assembled in the ER and are composed of α and β chains in complex with preformed trimers of invariant chain (Ii) [6]. Ii occupies the peptide-binding groove of MHC-II to prevent premature peptide binding and chaperones the MHC-II complex from the ER to the endocytic pathway. Entry into the endocytic pathway is predominantly by clathrinmediated endocytosis from the plasma membrane [7], but can also be direct from the trans-golgi network [8]. Once inside the endosomal compartments, Ii is degraded by lysosomal proteases until only CLIP is left bound in the MHC-II peptide-binding groove. HLA-DM exchanges CLIP for antigenic peptides in late endosomal compartments and mature peptide-MHC-II (pMHC-II) complexes are then exported to the cell surface [9]. In DCs, ubiquitination of a conserved lysine residue in the β chain cytoplasmic tail regulates surface expression and targeting of pMHC-II into late endosomal multi-vesicular bodies (MVBs) [10].
Formation of pMHC-II conformers from native protein occurs primarily in HLA-DM + late endosomes and generates stable complexes that are recognised by conventional Type A CD4 + T cells. In contrast, loading of exogenous peptide can occur throughout the endosomal pathway or at the cell surface and can generate pMHC-II conformers that are recognised by conventional Type A and unconventional Type B CD4 + T cells [11]. Type B T cells only recognise exogenous peptide and not the identical peptide when processed from protein. As a consequence, Type B T cells escape negative selection and are implicated in autoimmune conditions. In the NOD mouse model, Type B insulin-reactive T cells are pathogenic and trigger diabetes in adoptive transfer experiments [12]. Type B T cells constitute 30-50% of the T-cell repertoire [13], and phenotypically may resemble either Th1 or Th2 CD4 + T cells [12].
Salmonella is reported to interfere with MHC-II antigen processing and presentation to CD4 + T cells [14][15][16][17]. The relevance of these mechanisms in vivo is not clear as CD4 + T-cell priming has also been observed in mouse models of Salmonella infection [18][19][20][21]. We have previously shown that Salmonella infection of human DCs results in polyubiquitination and reduced surface expression of MHC-II [15,22]. In this study, we investigate how Salmonella influences MHC-II trafficking and presentation of antigen to Type A and B CD4 + T cells.

MHC-II accumulates in MVBs in Salmonella-infected cells
MHC-II is specifically removed from the surface of Salmonellainfected cells and accumulates in intracellular vesicles that resemble HLA-DM + LAMP-1 + EEA − peptide-loading compartments [15,22]. To better define the nature of these compartments, MHC-II localisation was assessed in Salmonella-infected MelJuSo cells, as the endocytic pathway is well characterised in this human epithelial-like melanoma cell line [23]. Cell surface HLA-DR was labelled with the monoclonal antibody L243 and after internalisation was visualised by cryo-immunoelectron microscopy.
HLA-DR was predominantly detected at the cell surface at 12 h post-infection in both uninfected (data not shown) and Salmonella-infected cells (Fig. 1A). Between 12 and 20 h postinfection, HLA-DR was endocytosed and distributed within early endosomes, MVBs and at the cell surface in uninfected cells ( Fig. 1C and F). In Salmonella-infected cells, there was a twofold greater accumulation of HLA-DR in MVBs compared with uninfected cells (Fig. 1B, D and E). The internalised MHC-II was not significantly associated with the SCV but localised to MVBs that most likely represent conventional MHC-II containing compartments found in the Salmonella-infected cells. There were fewer MVBs in uninfected cells suggesting that Salmonella may enlarge this compartment through accumulation of intracellular HLA-DR (data not shown). Since Salmonella infection results in polyubiquitination of MHC-II, and ubiquitination regulates sorting of MHC-II at MVBs [10,15], these results may suggest that Salmonellainduced ubiquitination of MHC-II enhances accumulation in MVBs to prevent recycling of mature MHC-II to the cell surface.

MHC-II down-regulation by Salmonella requires clathrin but not invariant chain-directed trafficking
To determine whether Ii-directed trafficking of MHC-II is required by Salmonella to regulate MHC-II surface expression, we generated HeLa cell transfectants stably expressing HLA-DR, but lacking endogenous Ii. There was no significant difference in the extent of HLA-DR down-regulation by Salmonella in HeLa cells expressing CIITA (Ii-positive) and HeLa cells transduced with HLA-DR (Ii-negative) ( Fig. 2A). As expected, HLA-DR dimers that lacked the DRβ cytoplasmic tail (DRα-219 ,β-223 and DRα,β-223 ) or with a lysine to arginine mutation in the β chain ubiquitination site (DRα,β-K 225 R), were not down-regulated by Salmonella ( Fig. 2A).
Endocytosis of pMHC-II is clathrin, AP-2 and dynamin independent [24]. To examine whether HLA-DR down-regulation by Salmonella requires AP-2 and clathrin, Ii-negative HeLa cells stably expressing HLA-DR were transfected with AP-2 and clathrin siRNA oligonucleotides and surface expression of HLA-DR was assessed by flow cytometry. In the absence of Ii, siRNA knockdown of clathrin, but not AP-2, reduced HLA-DR down-regulation by Salmonella (Fig. 2B, right panel). These data show that down-regulation of pMHC-II surface expression by Salmonella requires clathrin but was independent of Ii-directed trafficking and AP-2.

Salmonella down-regulates murine MHC-II surface expression and antigen presentation to CD4 + T cells
To examine the effect of Salmonella on T-cell presentation, we first exposed murine BMDCs to GFP-expressing Salmonella and examined surface I-A and I-E expression. Exposure to Salmonella increased overall I-A k and I-E k surface expression, consistent with BMDC activation and maturation [25] (data not shown). Infection of BMDCs with WT Salmonella reduced surface expression of I-A k and I-E k . I-A k and I-E k down-regulation was not detected following infection with ssaV Salmonella (Fig. 3A), as observed previously for HLA-DR [15]. This indicated that the SPI2 effector system was also required to regulate MHC-II surface expression in murine cells. Comparable I-A and I-E down-regulation was also seen for the b and d haplotypes (data not shown). MHC-II down-regulation was not detected in either human monocyte-derived macrophages, or a murine macrophage cell line, RAW264.7-CIITA (Supporting Information Fig. 2). The reason for this is unknown but  Fig. 1A and B for gating strategy and representative flow cytometry data. Graph shows percent of normal HLA-DR surface expression in uninfected (GFP-negative) cells combined from at least four independent experiments. (B) HeLa cells stably expressing HLA-DR WT (DRα,β)(Ii negative) were transfected with AP-2, clathrin or control siRNAs. Cells were infected with invasive GFP-S. Typhimurium after 5 days of AP-2 or clathrin depletion and surface HLA-DR was assessed as described in (A). Western blot shows AP-2 and clathrin depletion from representative cell lysates after 5 days of siRNA treatment. The loading control is β-actin. Graph shows percent of normal surface HLA-DR expression in uninfected (GFP negative) cells combined from four independent experiments. Comparison of distributions was performed by unpaired (A) or paired (B) two-tailed t-tests. may reflect functional differences between DCs and macrophages [26].
To assess antigen presentation in the context of Salmonella infection we analysed I-A k -dependent presentation of the model antigen hen egg lysozyme (HEL) by BMDCs to a CD4 + T-cell hybridoma expressing a HEL-specific TCR (3A9). T-cell hybridomas do not require co-stimulation and therefore pMHC-II levels should directly correlate with the extent of antigen presentation. BMDCs were used because they can be generated in large quantities and they resemble the myeloid CD11b + DCs present in the sub-epithelial dome of murine Peyer's patches where Salmonella internalise early after oral infection in vivo [3,27].
Incubation of BMDCs with exogenous HEL protein resulted in dose-dependent HEL-specific T-cell activation, as measured by IL-2 production (data not shown). After infection of BMDCs with Salmonella a reduction in T-cell activation was observed (Fig. 3B), in line with previous observations using exogenous antigen [14,28]. The reduction in T-cell activation was SPI2 dependent, although the effect was subtle. A MOI of ten bacteria to one BMDC was used as this does not induce significant NO production by the BMDCs (confirmed by Griess assay; data not shown) [14]. Infection with an equal number of heat-killed (HK) Salmonella had no influence on T-cell activation confirming that viable bacteria are required to inhibit antigen presentation in the absence of NO. These data show that down-regulation of MHC-II surface expression by Salmonella correlates with reduced presentation of antigen to CD4 + T cells.

Salmonella enhances presentation of exogenous peptide to Type B CD4 + T cells
To determine whether Salmonella also influenced presentation of antigen to Type B CD4 + T cells, we compared I-A k -dependent presentation of exogenous HEL protein and HEL 46-61 peptide by BMDCs to a Type A T-cell hybridoma (3A9) and a Type B T-cell hybridoma (11A10) with identical peptide specificity.
In line with previous publications, incubation of BMDCs with exogenous HEL protein or HEL 46-61 peptide resulted in dosedependent HEL-specific Type A T-cell activation, whereas only incubation with exogenous HEL 46-61 peptide resulted in equivalent activation of Type B T cells (Fig. 4A, open circles) [11]. Infection of BMDCs with WT Salmonella inhibited presentation of both exogenous HEL protein and HEL 46-61 peptide to Type A T cells. Intriguingly, WT Salmonella infection caused a dramatic increase in the presentation of exogenous HEL 46-61 peptide to Type B T cells, but had little effect on presentation of HEL protein (Fig. 4A). Unlike inhibition of Type A T-cell activation by Salmonella, enhanced presentation of HEL 46-61 peptide to Type B T cells was not SPI2 (ssaV) dependent (Fig. 4B). Furthermore, infection with an equal number of HK Salmonella had no effect on Type A T-cell activation but subtly increased Type B T-cell activation (Fig. 4B). These data show that Salmonella influenced antigen presentation in several distinct ways. Most dramatically Salmonella infection resulted in elevated presentation of exogenous peptide to Type B T cells. This was associated with a reduction in presentation of peptide or protein antigen to Type A T cells.

Exposure to Salmonella is sufficient to enhance presentation of exogenous peptide to Type B T cells
At the MOI used in the above experiments (MOI = 10), only 10-20% of the BMDCs were infected with Salmonella. This sug-gested that direct infection may not be required and that a soluble factor produced by infected BMDCs could be influencing neighbouring cells. To screen for potential soluble factors produced by Salmonella-infected BMDCs, culture supernatant was harvested from infected BMDCs at 20 h post-infection and incubated with fresh BMDCs, Type B T hybridoma cells and HEL 46-61 peptide.
Incubation of fresh BMDCs with culture supernatant from Salmonella-infected BMDCs was sufficient to enhance presentation of exogenous peptide to Type B T cells (Fig. 5A). Clearance of the supernatant using a 0.45 μm filter prior to incubation with fresh BMDCs (Fig. 5A) or separation of infected BMDCs from fresh BMDCs and T cells using 0.45 μm transwells (Supporting Information Fig. 4) abrogated the effect. This suggested that whilst a component of culture supernatant from Salmonella-infected BMDCs can influence uninfected BMDCs in trans, the factor responsible was not smaller than 0.45 μm.
Salmonella are rod-shaped bacteria, 0.5-1.5 μm in diameter and 2-5 μm in length. Clearance of culture supernatant using a 0.45 μm filter would therefore remove any intact Salmonella present. To determine whether direct exposure of BMDCs to Salmonella was sufficient to enhance presentation of exogenous peptide to Type B T cells, BMDCs were infected with GFPexpressing Salmonella and then sorted at 20 h post-infection to separate infected from exposed but uninfected populations. Presentation of HEL 46-61 peptide to Type B T hybridoma cells was compared for BMDCs that were unexposed, exposed but uninfected, or exposed and infected with Salmonella, respectively.
Exposure of BMDCs to Salmonella was sufficient to enhance presentation of exogenous peptide to Type B T cells (Fig. 5B). There was no significant difference in the extent of Type B T-cell activation between sorted BMDCs that were exposed to, or infected with Salmonella. There was a consistent trend towards reduced presentation in the infected BMDCs, although the effect was subtle (Fig. 5B). Notably, presentation of peptide to Type B T cells was enhanced by HK Salmonella if the MOI was increased (Fig. 5C). This suggested that whilst viable bacteria contribute more significantly to the enhanced presentation of exogenous HEL 46-61 peptide to Type B T cells observed, viability is not essential.

Discussion
We show that Salmonella infection influences MHC-II antigen presentation to CD4 + T cells by two distinct mechanisms. Intracellular replication of Salmonella resulted in reduced expression of pMHC-II complexes at the cell surface and altered presentation of antigen to CD4 + T cells. Most importantly, exposure of BMDCs to Salmonella resulted in enhanced presentation of exogenous peptide to Type B CD4 + T cells, which have been linked to autoimmune disease progression [12].
We first examined the influence of intracellular Salmonella on MHC-II trafficking and localisation. Using Ii-negative HeLa cells, we showed that down-regulation of MHC-II by Salmonella was independent of Ii-directed trafficking and AP-2, but required clathrin. As AP-2 is the principal adaptor protein required for formation of clathrin-coated pits at the plasma membrane [29], it is unlikely that MHC-II down-regulation by Salmonella requires the formation of clathrin-coated pits. Distinct clathrin coats are also present at the cytoplasmic face of MVBs and are proposed to concentrate cargo for subsequent incorporation into luminal vesicles [29]. In addition, sorting of pMHC-II into luminal vesicles at MVBs is regulated by ubiquitination [30]. Therefore, the requirement for clathrin by Salmonella may be related to clathrindependent sorting of ubiquitinated cargo at MVBs. We next confirmed that Salmonella down-regulated surface expression of I-A and I-E in BMDCs, similar to what has been observed in human cells [15,22]. This validated the use of murine T-cell reagents to assess antigen presentation following Salmonella infection of BMDCs. Salmonella infection of murine DCs has been reported to inhibit presentation of antigen to CD4 + T cells [14,28].
Here we compared the influence of Salmonella on presentation of antigen to Type A and B CD4 + T-cell subsets. In contrast to the suppressive effect of Salmonella infection on presentation of both exogenous protein and peptide antigen to Type A T cells, presentation of peptide to Type B T cells was significantly enhanced. Salmonella infection did not significantly alter presentation of exogenous protein antigen to Type B T cells. This contrasts with the recent data of Strong and Unanue showing that TLR ligands have no effect on the presentation of peptide via the Type B conformer in splenic DCs [31]. This may reflect differences in antigen handling between DC subsets as reported by Lovitch et al. for presentation of native antigen to Type B T cells using LPS-stimulated BMDCs [32].
The relevance of reduced Type A presentation in relation to immunity to Salmonella infection in vivo is not clear. Whilst chronic typhoid carriers exhibit impaired humoral and cellular immunity [5], rapid priming of CD4 + T cells in mouse models of Salmonella infection was reported to elicit effective Th1 responses [21]. Regardless, priming of Type B T-cell responses does not imply any perturbation in the Th1/Th2 balance and could occur in the presence or absence of effective anti-Salmonella responses. Taken together, these data suggest that Salmonella infection polarises antigen presentation by stabilising or enhancing formation of the pMHC-II conformer recognised by Type B T cells, leading to increased Type B T-cell activation. In heavily infected BMDCs, this polarisation may be further promoted by reduced presentation of pMHC-II conformers recognised by Type A T cells.
Mere exposure to HK Salmonella or supernatant from Salmonella-infected BMDCs was sufficient to enhance presentation of exogenous peptide to Type B T cells. This phenotype is unlikely to be caused by soluble TLR ligands such as LPS and flagellin, as these are shed by Salmonella in significant amounts [33] and the effect was lost when the culture supernatant was filtered or when infected BMDCs were spatially separated from uninfected BMDCs and T cells. It is unlikely to be due to secretion of a soluble cytokine as this would also remain in the filtered supernatants. Direct contact between Salmonella and BMDCs is required. We are currently attempting to identify the bacterial components responsible for this effect.
The potential for Type B T-cell activation to lead to autoimmune disease is established [12]. Infection of the gastrointestinal tract with Salmonella, as well as Yersinia, Campylobacter and Shigella [34], is frequently associated with reactive arthritis in humans [35] and mice [36]. Salmonella infection in humans caused incidence rates of between 6 and 30% (with variable severity), reflecting the propensity of different Salmonella species to induce arthritis. The enhanced activation of Type B T cells Typhimurium-infected BMDCs were sorted from the exposed but uninfected population. Refer to Supporting Information Fig. 1A for representative gating strategy. Presentation of 0.5 μM HEL 46-61 peptide to 11A10 (Type B) T cells was compared for BMDCs that were unexposed, exposed but uninfected, or infected with S. Typhimurium. Comparison of distributions was performed by unpaired twotailed t-tests. (A-C) Data shown are the mean + SD and are representative of one out of at least four independent experiments.
observed with Salmonella infection is not limited to pathogenspecific T cells as the antigens used in these experiments were not derived from Salmonella. Therefore, it is possible that infected DCs in vivo may incidentally present self-peptide-associated Type B conformers, leading to activation of potentially autoreactive Type B T cells. Processes such as inflammation may increase the supply of exogenous peptide and thereby facilitate the generation of Type B pMHC-II conformers. In fact, several immune cell types, including neutrophils [37] and DCs [38,39], are known to generate exogenous antigenic peptides. In Type 1 diabetes, related mechanisms are thought to generate peptides from the insulin B chain, which when presented on MHC-II are specifically recognised by diabetogenic Type B T cells, leading to disease [12].
This study is the first to show that exposure of BMDCs to Salmonella enhances the presentation of exogenously supplied peptides to Type B T cells. It suggests a mechanism by which Salmonella infection could lead to a breakdown in immunological tolerance. Further studies will be required to identify the factor responsible for this alteration in peptide presentation and to evaluate the role of Type B T cells in infection and autoimmunity.

BMDC preparation and antigen presentation
Mice were maintained according to institutional guidelines at the University of Cambridge. BM was harvested from femurs/tibias of 8-12 week female C3H/HeNCrl mice (Charles River) and passed through a 70 μm strainer in IMDM. BM cells were seeded in 9 cm plates at 1 × 10 6 cells/mL in IMDM, 10% FCS, 2 mM Ultraglutamine (Lonza), 10 ng/mL IL-4 (Peprotech), 20 ng/mL GM-CSF (Peprotech) and penicillin/streptomycin (PAA Laboratories) for ∼30 min to adhere macrophage-precursors. Non-adherent BM cells were reseeded in 6-well plates for the differentiation into BMDCs, with media/cytokine replacement on days 3 and 5. Day 7 BMDCs were harvested by gentle scraping on ice. Differentiated BMDCs were routinely 50-60% CD11c/CD11b + , CD80 hi , CD86 lo and MHC-II lo , as assessed by flow cytometry. For antigen presentation, BMDCs were seeded at 3 × 10 4 cells/well in 96-wellflat-bottomed plates more than 6 h prior to Salmonella infection. For sorts, DCs were seeded at 5 × 10 6 cells/9 cm plate prior to infection. At 20 h post-infection, cells were washed with PBS, then antigen (HEL protein (Sigma) or HEL 46-51 peptide (Cambridge Bioscience)) and T cells were added. T-cell hybridomas were prewashed with DMEM, 5% FCS and were added at a 5:1 T cell:DC ratio (refer to Supporting Information Fig. 3 for titration). Culture supernatants were harvested after 24 h, frozen at −80 • C and then IL-2 quantified by ELISA using mouse IL-2 Ready-SET-Go! kits (eBioscience).

Salmonella strains, infection and flow cytometry
Salmonella Typhimurium 12023 (ATCC) WT and ssaV strains that constitutively express GFP from pFVP25.1 were grown as described previously [15]. MelJuSo, RAW264.7-CIITA and HeLa cells were infected by SPI1-invasion as described previously [15]. BMDCs and monocyte-derived macrophage were infected with stationary phase Salmonella (pre-opsonised in 20% normal mouse serum (PAA Laboratories) or autologous human serum, respectively, for 30 min) for 60 min at a MOI of 10:1. Where specified, Salmonella were HK at 65 • C for 45 min prior to opsonisation. For flow cytometry, cells were harvested by scraping and incubated with appropriate antibodies, in FACS buffer (PBS, 5% FCS) at 4 • C. BMDCs were incubated with Mouse Fc Block (BD Biosciences) for 5 min at 4 • C prior to antibody addition. After washing, cells were fixed in 1% paraformaldehyde and analysed using an FACScan Flow Cytometer and Summit software (BD Biosciences). MHC-II surface expression was calculated as mean fluorescence of infected cells (GFP positive)/mean fluorescence of uninfected cells (GFP negative) ×100.

Western blot
Cells were harvested using Cell Dissociation Buffer (Sigma) and lysed for 30 min at 4 • C in PBS, 1% Nonidet P-40 substitute, 50 mM Tris (pH 7.5), 5 mM EDTA, 150 mM NaCl, protease inhibitor mixture (Roche Diagnostics), and 5 mM Iodoacetamide. Samples were boiled in SDS-PAGE loading buffer before protein separation by SDS-PAGE and transfer to Immobilon-P PVDF membrane (Millipore). Membranes were probed with antibodies and analysed by rapid immuno-detection using ECL. Membranes were blocked with 5% non-fat milk powder/0.05% Tween-20 for subsequent detections.

Cryo-immunoelectron microscopy
MelJuSo cells were infected with GFP-Salmonella in 9 cm plates. At 12 h post-infection, cells were washed and surface HLA-DR was labelled (L243 antibody) at 4 • C. After 20 min, cells were washed and returned to 37 • C. After 0 and 7 h endocytosis, cells were fixed in 0.2% glutaraldehyde/2% paraformaldehyde for 2 h at RT. Fixed cells were harvested, embedded in gelatin, and cryo-sectioned using a Leica FCS [41]. Ultrathin sections (50 nm) were cut at -120 • C using a Cryo-immuno knife (Diatome, Switzerland) and cells were labelled with 10 nm Protein A gold particles (Cell Biology, Medical School, Utrecht University) in PBS/1% BSA. Images were collected using a Philips/FEI CM10 electron microscope.