Two Salmonella OmpC K^b-Restricted Epitopes for CD8⁺-T-Cell Recognition

Abstract

We report the identification of two peptides from Salmonella OmpC porin that can bind to major histocompatibility complex class I K^b molecules and are targets of cytotoxic T lymphocytes from Salmonella-infected mice. These peptides are conserved in gram-negative bacterial porins and are the first Salmonella porin-specific epitopes described for possible CD8⁺-T-cell elimination of infected cells.

Some of the most-studied protective immunogens of Salmonella are the outer membrane proteins (OMPs) called porins. Vaccination with purified porins can protect mice against the challenge of virulent Salmonella strains (4, 5, 6). In these protection processes, antiporin antibodies and T-cell-mediated immunity are required, and these T cells have been described as being Th2 type with predominant interleukin-4 and immunoglobulin G1 production (1, 2, 10, 11, 19). The role of CD8⁺ T cells during Salmonella infection is unclear but has gained interest because humans and mice immunized with attenuated Salmonella strains can generate a CD8⁺-T-cell response (3, 8, 15, 18). Also, the generation of major histocompatibility complex class I (MHC-I) molecule-restricted peptides from infected cells is not well characterized. We previously described that macrophages infected with Salmonella enterica serovar Typhimurium and activated with gamma interferon (IFN-γ) can secrete peptides that stabilize K^b molecules on the surface of RMA-S cells (9). Lo et al. have described an MHC class Ib-restricted response during infection with virulent Salmonella (7), and Pasetti et al. found an L^d-restricted response after intranasal administration with attenuated Salmonella strains (14).

In the present study, we searched for MHC-I epitopes from Salmonella OmpC porin, one of the two components of the porin preparation used in vaccine trials (5).

K^b binding peptide search and chemical synthesis.

In order to identify CD8⁺-T-cell epitopes derived from Salmonella serovar Typhimurium OmpC porin, we searched the sequence for eight-amino-acid-long peptides with the anchor amino acid motif described for peptides that bind to K^b molecules. The first analysis of peptides was done with the ProPred-I program (http://www.imtech.res.in/raghava/propred1/index.html), which screens for MHC-I binding peptides that can be generated by proteosome cleavage of the original protein (17). Four peptides from the serovar Typhimurium OmpC (gi 7428872) sequence were displayed by this analysis: 132-RNTDFFGL, 73-TRVAFAGL, 343-NTDDIVAL, and 159-ENTNGRSL. A second analysis of the OmpC sequence was done with the Parker HLA peptide motif search program (http://bimas.dcrt.nih.gov/molbio/hla_bind/), which estimates the half time of disassociation of an MHC molecule containing the predicted peptides (13). According to these two criteria, the peptides 132-RNTDFFGL (OmpC-132) and 73-TRVAFAGL (OmpC-73) had the highest chances to be natural K^b epitopes and so were chosen for further analysis. Peptides were synthesized by solid phase by using an ABI 430A automated synthesizer (Applied Biosystems Inc., Foster City, Calif.) and purified to >90% by reverse-phase high-performance liquid chromatography with a C₁₈ column (Millipore, Bedford, Mass.).

Peptide binding assays and flow cytometry.

To test the binding of OmpC peptides to K^b molecules, the murine TAP 2 (transporter associated with antigen processing type 2)-deficient RMA-S cell line was cultured with 50 and 100 μM concentrations of each peptide for 6 to 8 h at 37°C. K^b binding of ovalbumin amino acids 257 to 264 (SIINFEKL peptide; OVA_257-264) was used as a reference for positive peptide binding. Following extensive washes with phosphate-buffered saline (PBS), cells were independently surface stained with purified monoclonal antibodies (MAbs) against K^b (Y3 [12]), D^b (28-14-8S; ATCC HB-27), and K^d (SF1-1.1.1; ATCC HB-159) at 4°C. After two washes with PBS, counterstaining with fluorescein isothiocyanate-conjugated goat anti-mouse antibody (Sigma, St. Louis, Mo.) was performed under the same conditions. A total of 10⁴ RMA-S cells for each treatment was analyzed by flow cytometry on a FACSort (Becton Dickinson, Mountain View, Calif.). Figure 1A shows the K^b profiles for RMA-S cells loaded with the predicted Salmonella OmpC peptides, and as expected OmpC-132 and OmpC-73 can bind to K^b molecules and stabilize alpha chains on the surfaces of RMA-S cells in a dose-dependent manner (a 3.8-fold increase in the mean fluorescence [mf] with 100 μM OmpC-73 and a 2.1-fold increase in the mf with 100 μM OmpC-132). OmpC peptide binding was K^b specific because high-dose OmpC peptides did not stabilize D^b molecules (the mf values were 16.26 in the sample without peptide, 21.19 in the sample with OmpC-73, and 25.84 in the sample with OmpC-132). Although this result was clear for the OmpC peptides, the OVA_257-264 peptide showed some binding to D^b (mf, 62.73) when RMA-S cells were cultured with a 100 μM dose of peptide (Fig. 1B). In all cases, absence of staining was observed with the isotype control antibody against K^d (Fig. 1C).

FIG. 1.

Salmonella OmpC peptides bind specifically to K^b molecules. RMA-S cells were cultured with 50 and 100 μM concentrations of OmpC-73, OmpC-132, and OVA_257-264 (OVA) peptides for 6 h, and then MHC-I molecules were surface stained and analyzed with a fluorescence-activated cell sorter. The results for anti-K^b MAb (A), anti-D^b MAb (B), and anti-K^d MAb (C) are shown. Data shown are representative of three reproducible independent experiments.

Can these OmpC peptides elicit a CTL response after immunization?

To test whether the OmpC peptides can elicit a cytotoxic T-lymphocyte (CTL) response in vivo, we immunized C57BL/6J mice with an emulsion of incomplete Freund’s adjuvant containing 100 μg of each peptide, which was applied subcutaneously at the base of the tail. This regimen was repeated three times at intervals of 1 week. At day 31, the mice were sacrificed for spleen removal and in vitro stimulation with the peptides. Briefly, splenocytes were cultured in 10% fetal calf serum-supplemented Dulbecco minimal essential medium and a 50 μM concentration of the corresponding peptide. Cells surviving after 5 days of culture were tested for cytolytic activity against ⁵¹Cr-labeled RMA-S cells loaded with peptide, as reported previously (20, 23). In Fig. 2A, a representative experiment is shown in which CTLs from OmpC-73- and OmpC-132-primed mice specifically kill RMA-S cells loaded with the corresponding peptide but not OVA_257-264-loaded cells, indicating that the CTL activity observed was not due to another type of cell such as NK cells. It is worthwhile to mention that OmpC peptides bind and stabilize primarily to K^b and not to D^b (Fig. 1A and B) molecules on RMA-S cells. Therefore, CTLs from Salmonella-infected mice recognize these epitopes only in a K^b-restricted manner on this cell line. In addition, OmpC-132 induced a stronger CTL response than the one derived from OmpC-73 immunization (P of <0.002 for effector-to-target cell ratios greater than 25:1), which correlates with their levels of peptide binding to RMA-S cells shown in Fig. 1. This finding confirms that C57BL/6J mice can generate a K^b-restricted CTL response against OmpC-73 and OmpC-132 peptides.

FIG. 2.

Cytotoxic response against Salmonella OmpC peptides. (A) C57BL/6 mice were immunized with OmpC-73 or OmpC-132 peptides. After 4 weeks, splenocytes were stimulated in vitro with the indicated peptide (expansion), and an assay of cytotoxic activity against RMA-S cells loaded with peptide (target) was performed. (B) Mice were orally infected with serovar Typhimurium SL3261 (S.t SL3261), and after 6 weeks splenocytes were tested as described for panel A. (C) Mice received an intraperitoneal load of serovar Typhi given in 5% mucin or in PBS (S.t mucin or S.t PBS), and splenocytes from both groups of mice were tested as described for panel A. All curves show the mean percentages of specific lysis at different effector-to-target cell ratios for triplicate wells; the bars indicate standard deviations. In all cases, results for a representative experiment out of three experiments with similar results are shown. Comparisons of cytotoxicity levels were performed by Student's t test with SigmaStat software (version 9.0; SPSS Science Software Products, Chicago, Ill.). P values of <0.05 were considered significant. OVA, OVA_257-264 peptide.

Is the OmpC-specific CTL response present after Salmonella invasion?

C57BL/6J is an Ity^s mouse strain that cannot resist infection with virulent Salmonella serovar Typhimurium. To overcome this problem and evaluate the natural course of disease, we infected C57BL/6J mice orally with 10⁶ PFU of attenuated aroA^−/− Salmonella serovar Typhimurium SL3261. This protocol elicits a specific immune response for bacterial clearance in mice (21). After 6 weeks, mouse spleens were harvested, stimulated with peptides, and evaluated for specific CTL activity as described above. Mice given a Salmonella serovar Typhimurium dose that self-limits its replications after invasion can generate specific CTLs against OmpC-132 and OmpC-73 (Fig. 2B).

Based on the known homology of the Salmonella serovar Typhimurium OmpC sequence with other OMPs of gram-negative bacteria, we searched for OmpC-132 and OmpC-73 epitopes in the protein family of porins (Table 1). All Salmonella strains share the exact sequence for OmpC-73 and OmpC-132 in the OmpC porin. Interestingly, other strains and different porins share similar peptides with the anchor residues for binding to K^b molecules. Normally, S. enterica serovar Typhi does not cause systemic disease in mice. Therefore, to further confirm that the invasion of Salmonella into macrophages was required to generate the CTL response, we used mucin to apply serovar Typhi to the peritoneum of C57BL/6J mice (16). For this reason, 5,000 PFU of serovar Typhi 9,12,Vi:d in 5% mucin was administered, and control mice received the same bacterial dose in PBS. Six weeks later, splenocytes were screened for peptide CTL activity as described above. As shown in Fig. 2C, only mice that received serovar Typhi and mucin were able to generate a CTL response against OmpC peptide-loaded RMA-S cells (P of <0.02 for effector-to-target cell ratios greater than 50:1). Thus, Salmonella needs to colonize macrophages in order to generate a CTL response against porin peptides. This result also suggests that infected macrophages contribute to the generation of MHC-I peptides; further investigation will clarify the contribution of other antigen-presenting cells (dendritic cells and B cells) to the presentation of Salmonella antigens (22, 24, 25).

TABLE 1.

OmpC predicted epitopes in gram-negative bacterial OMPs

GenBank accession no.a	Protein	Bacterial origin of proteinb	Similarity toc:
			OmpC-73 (TRVAFAGL)	OmpC-132 (RNTDFFGL)
16761195	OmpC	Serovar Typhimurium	+	+
47797	OmpC	Serovar Typhi	+	+
8953564	OmpC	Serovar Minnesota	+	+
19743624	OmpC	Serovar Dublin	+	+
19743622	OmpC	Serovar Gallinarum	+	+
26248604	OmpC	Escherichia coli	+	+
24113600	Omp1b	Shigella flexneri	+	RNSDFFGL
16760442	OmpS	Serovar Typhi	TRLAFAGL	+
3273514	OmpN	Escherichia coli	TRLAFAGL	+
16764916	OmpD	Serovar Typhimurium	TRLAFAGL	+
16764875	OmpC2	Yersinia pestis	TRLGFAGL	+
151149831	OmpK36	Klebsiella pneumonie	TRLAFAGL	RNSDFFGL

^aGenBank, National Center for Biotechnology Information.

^bAll serovars are those of Salmonella enterica.

^c+, identity to OmpC peptide. Amino acids shown in bold type are those that differ from the sequence in parentheses at top of column.

Further analysis of a specific CTL response directed against OmpC peptides will clarify the participation of CD8⁺ T cells in the elimination of infected cells, as well as some unanswered questions on the processing and presentation of Salmonella antigens.