Prof. John H. Robinson, Musculoskeletal Research Group, Clinical Medical Sciences, University of Newcastle upon Tyne, Framlington Place, Newcastle upon Tyne, NE2 4HH, UK. Email: firstname.lastname@example.org Senior author: Prof. John H. Robinson
We mapped mouse CD4 T-cell epitopes located in three structurally distinct regions of the V antigen of Yersinia pestis. T-cell hybridomas specific for epitopes from each region were generated to study the mechanisms of processing and presentation of V antigen by bone-marrow-derived macrophages. All three epitopes required uptake and/or processing from V antigen as well as presentation to T cells by newly synthesized major histocompatibility complex (MHC) class II molecules over a time period of 3–4 hr. Sensitivity to inhibitors showed a dependence on low pH and cysteine, serine and metalloproteinase, but not aspartic proteinase, activity. The data indicate that immunodominant epitopes from all three structural regions of V antigen were presented preferentially by the classical MHC class II-restricted presentation pathway. The requirement for processing by the co-ordinated activity of several enzyme families is consistent with the buried location of the epitopes in each region of V antigen. Understanding the structure–function relationship of multiple immunodominant epitopes of candidate subunit vaccines is necessary to inform choice of adjuvants for vaccine delivery. In the case of V antigen, adjuvants designed to target it to lysosomes are likely to induce optimal responses to multiple protective T-cell epitopes.
Yersinia pestis is the causative agent of plague, probably the most devastating disease in human history.1 Eradication has not yet been achieved world-wide and current interest has been aroused by the potential of Y. pestis as a biological weapon.2 The bubonic form of the disease is transmitted by fleas from rats to humans and presents as swollen lymph nodes (buboes) draining the site of the flea bite. Bubonic plague progresses to septicaemia if untreated and may develop into the highly contagious pneumonic form which is spread directly from person to person.3,4
The V antigen of Y. pestis, coded by the Ysc-Yop Lcr (low calcium response) plasmid,5 is a type III secretion system component with multiple potential roles in virulence.6 The capsular Caf1 protein also contributes to virulence by resisting phagocytosis of Y. pestis7 and acquired immunity to infection in mice is thought to be mediated by immunoglobulin G (IgG) antibodies specific for V and Caf1.8–11 Immunization of mice with a combination of Caf1 and V antigens confers more effective protection than either alone against both subcutaneous and aerosol challenge, modelling bubonic and pneumonic plague, respectively.12–14 A subunit vaccine based on recombinant V (rV) and Caf1 is under evaluation in human trials.15
Effective delivery of subunit vaccines based on recombinant antigens in synthetic adjuvants requires a better understanding of how candidate vaccine antigens are processed and presented to helper T cells that are necessary for the induction of protective IgG antibodies.16–18 However, although the mechanisms of antigen processing and presentation of a small number of experimental antigens have been reported,19,20 few candidate vaccine antigens have been studied.21–24 Here we localized the major helper T-cell epitopes of the V antigen of Y. pestis in H-2d mice and studied the predominant mechanisms of V antigen presentation to CD4 T cells by macrophages.
Materials and methods
Antigens and cells
The rV antigen of Y. pestis was expressed as a fusion protein with glutathione-s-transferase in Escherichia coli and was cleaved from the fusion protein with either Factor Xa or PreScission protease as previously described.25 Thirty-two synthetic peptides, 20 amino acids in length and overlapping by 10 amino acids, representing the complete 326 amino acid V sequence from acc. no. P21206 were synthesized by Dr J. Gray, Institute of Cell and Molecular Biosciences, University of Newcastle upon Tyne, UK. The purity of the peptides was confirmed by high-performance liquid chromatography and peptides were shown to be non-mitogenic and non-toxic to proliferating T cells. For some experiments peptides were pooled as follows: pool 1 (peptides 1 and 2: 1–21/12–31), pool 2 (peptides 3–5: 22–41/32–51/42–61), pool 3 (peptides 6–8: 52–71/62–81/72–91), pool 4 (peptides 9–11: 82–101/92–111/102–121), pool 5 (peptides 12–14: 112–131/122–141/132–151), pool 6 (peptides 15–17: 142–161/152–171/162–181), pool 7 (peptides 18–20: 172–191/182–201/192–211), pool 8 (peptides 21–23: 202–221/212–231/222–241), pool 9 (peptides 24–26: 232–251/242–261/252–271), pool 10 (peptides 27–29: 262–281/272–291/282–301), pool 11 (peptides 30–32: 292–311/302–321/307–326).
Adult female BALB/c (H-2d), B10.D2 (H-2d), C57BL/6 (B10, H-2b) and B10.BR (H-2k) mice were purchased from Harlan Ltd, Cirencester, UK. All cells were grown in cell culture medium containing RPMI-1640 supplemented with 3·0 mm l-glutamine, 0·05 mm 2-mercaptoethanol, and 10% fetal bovine serum (all from Sigma-Aldrich, Gillingham, UK). Bone marrow macrophages were used as antigen-presenting cells (APC) and were generated as described previously26 by culture of femoral bone marrow cells; they were shown to be at least 90% enriched by flow cytometry using fluorescein isothiocyanate-conjugated anti-F4/80 monoclonal antibody (Serotec, Oxford, UK). Briefly, bone marrow cells were cultured for 6 days in bacteriological Petri dishes in the culture medium described above supplemented with 5% horse serum, 1 mm sodium pyruvate, 10 mm HEPES buffer (Sigma-Aldrich) and 10% of a culture supernatant from the L929 cell line as a source of macrophage-colony-stimulating factor.
Mice were immunized in one footpad with 25 μg rV in Titermax adjuvant (Sigma-Aldrich) and after 7 days popliteal lymph node cells were stimulated with 25 μg/ml rV and then restimulated with irradiated spleen cells and 25 μg/ml rV twice at 2-weekly intervals to generate T-cell lines. T-cell hybridomas were generated by polyethylene-glycol fusion of T-cell lines with BW5147 (TCRα–β–) cells (provided by Dr P. Marrack, Denver, CO). All four T-cell hybridomas used in this study expressed CD4, CD3ε and TCRαβ by flow cytometry. Major histocompatibility complex (MHC)-restriction was determined by synthetic peptide presentation by L cells transfected with either Aαdβd or Eαdβd (a gift from Dr R. Germain, National Institutes of Health, Bethesda, MD), and those used in this study were Ad-restricted.
Lymph node proliferation assay
The proliferation assay was performed on popliteal lymph node cells from rV-immunized mice in 200 μl culture medium in 96-well round-bottom microtitre plates (3 × 105 cells/well) for 72 hr with 20 μg/ml rV or synthetic peptides. Cells were labelled with 14·8 KBq tritiated thymidine (TRA310, specific activity 74 GBq/mmol, Amersham International, Buckinghamshire, UK) for the final 18 hr of culture before harvesting on glass-fibre membranes and quantification of radioactivity by direct beta-counting (Matrix 9600, Packard Instrument, Meriden, CT). The results are shown as mean counts per min (c.p.m.) of triplicate wells ± SD and were considered significant if antigen stimulation exceeded twice the c.p.m. in the absence of antigen.
Antigen presentation assay
Bone marrow macrophages (4 × 104/well) were allowed to adhere to flat-bottom 96-well microtitre plates overnight in the presence of 1 ng/ml interferon-γ (R & D Systems, Abingdon, UK) at 37° in a humidified CO2 incubator and then used untreated or treated with inhibitors (all from Sigma, Dorset UK, except Nα-p-tosyl-l-lysine chloromethyl ketone from Calbiochem, La Jolla CA) before use as APC. Inhibitor doses were titrated and the lowest dose showing maximum inhibition of presentation of rV but no effect on synthetic peptide presentation was used. A dose range of rV or synthetic peptides was added for a further 5-hr incubation before the macrophages were washed and fixed with 1% paraformaldehyde for 4 min. Fixation was stopped by adding 0·05% Gly-Gly (Sigma) for 4 min before washing three times with phosphate-buffered saline to remove the fixative and any remaining inhibitors. In some experiments macrophages were fixed and washed before the addition of antigen.
T-cell hybridoma cells (4 × 104/well) were added and plates were incubated at 37° in a CO2 incubator for 24 hr before culture supernatants were collected and stored frozen. T hybridoma responses were determined by the interleukin-2 content of hybridoma supernatants diluted 1 in 2, measured as the proliferative response of 3 × 104 CTLL-2 cells cultured for 24 hr in triplicate by adding 14·8 KBq tritiated thymidine for the final 18 hr of culture as described for lymph node proliferation assay. Results are shown as mean c.p.m. of triplicate wells ± SD and experiments were repeated at least twice.
Mapping CD4 T-cell epitopes of rV
CD4 T-cell epitopes of the V antigen of Y. pestis were mapped in two strains of H-2d inbred mice, BALB/c and B10.D2 immunized in the footpad with rV antigen in adjuvant. Popliteal lymph node cells were assayed for their proliferation responses to overlapping synthetic peptides representing the complete V sequence. Lymph node T-cell responses were recalled in vitro to pools of synthetic peptides, and all rV-immunized BALB/c mice responded to rV as well as pools 4, 6, 7 and 8, but not to the remaining peptide pools (Fig. 1a). Two or more of the rV-immunized B10.D2 mice also responded to rV and the same pools of peptides as BALB/c mice (Fig. 1b). Lymph node cells from further groups of rV-immunized BALB/c and B10.D2 mice were assayed for in vitro responses to individual peptides from the stimulatory pools, and all mice responded to rV as well as to peptides 11 (102–121), 17 (162–181) and 22 (212–231) (Fig. 1c,d). Some B10.D2 mice also responded to peptide 16 (152–171). A small but significant response was also detected to peptides 18 (172–191) and 19 (182–201) from pool 7. Groups of three B10.D2 mice were immunized with each of the three stimulatory peptides 102–121, 162–181 and 212–231 and two or more mice from each group responded to rV and the immunizing peptide (data not shown), confirming the immunodominance of the three epitopes, and their localization largely to the V antigen sequences 102–121, 162–181 and 212–231 in H-2d mice.
In similar experiments, we localized the major epitopes recognized by rV-immunized B10(H-2b) mice to V peptides 72–91, 102–121 and 162–181 (data not shown), which completely included the three immunodominant H-2b-restricted epitopes 73–84, 103–114 and 166–177 that were identified in a recent report.27 Thus, both H-2d-restricted and H-2b-restricted epitopes are located within V antigen peptides 102–121 and 162–181, which may define regions of V that are immunogenic for multiple MHC haplotypes.
T-cell hybridomas were generated using rV-specific T-cell lines from BALB/c (H-2d) and BALB.B (H-2b) mice and the majority were specific for one of the epitopes recognized by lymph node cells from rV-immunized mice. The Ad-restricted T-cell hybridomas used in this study all recognized intact rV and were 2H1-specific for peptide 11 (102KEFLESSPNTQWELRAFMAV121) only, 1E10·2- and 1B9-specific for peptide 17 (162AELKIYSVIQAEINKHLSSS181) with limited response to the flanking peptide 16 (152KLREELAELTAELKIYSVIQ171), and 1B4-specific for peptide 22 (212EYKILEKMPQTTIQVDGSEK231) only, a similar pattern of specificity to the lymph node responses shown above.
Mechanisms of V antigen processing
Figure 2 shows a ribbon diagram of the V antigen modified from the published structure.28 Peptides including immunodominant epitopes recognized by H-2d mice are highlighted. The three epitopes are located one each in the two globular domains as well as the ‘grip’ region of the ‘dumbell’ structure of V antigen. The responses of H-2d-restricted T-cell hybridomas to these three epitopes from rV were studied to gain an understanding of the dominant mechanisms of antigen processing and presentation of rV by macrophages. Fixation of macrophages before incubation with rV prevented presentation of all three epitopes (Fig. 3a–c), demonstrating a dependence on uptake and/or intracellular processing for presentation of epitopes in all regions of V antigen. The response to each epitope was detected within an hour and reached a plateau 3–4 hr after exposure of macrophages to rV (Fig. 3d–f). Depleting endosomal compartments of newly synthesized MHC class II molecules by treatment of macrophages with cycloheximide to prevent short-term protein synthesis (Fig. 4a–c) or with brefeldin A to disrupt Golgi transport (Fig. 4d–f) reduced presentation of all three epitopes, although 212–231 less profoundly than the other two.
Treatment of macrophages with ammonium chloride or monensin, agents that raised endosomal pH, profoundly reduced antigen presentation of the three H-2d-restricted epitopes of rV (Fig. 5a–c), suggesting that processing of rV or loading peptide epitopes was dependent on an acidic endosomal compartment. We treated macrophages with enzyme inhibitors of each of the major families of proteolytic enzymes to determine the pattern of antigen processing of rV. Presentation of all three epitopes showed a profound sensitivity to E-64d and DCI, broad-spectrum inhibitors of cysteine and serine proteinases (Fig. 5d–f). Inhibitors selective for both trypsin-like (Nα-p-tosyl-l-lysine chloromethyl ketone) and chymotrypsin-like (Nα-p-tosyl-l-phenylalanine chloromethyl ketone) serine proteinases also profoundly reduced the response (data not shown). Presentation of two of the epitopes was less sensitive to the metallo-proteinase inhibitor 1,10-phenanthroline, although the response to the epitope within 212–231 was more profoundly inhibited. The aspartic proteinase inhibitor pepstatin A had no effect, other than to modestly enhance the response to epitope 102–121 (Fig. 5d–f). Collectively, the data show a similar pattern of involvement of multiple enzyme families for presentation of epitopes in distinct regions of V.
Responses of H-2b-restricted T-cell hybridomas to the two shared epitopes on peptides 102–121 and 162–181 were essentially the same as the H-2d-restricted responses shown in Figs 3–5 (data not shown). Collectively the data suggest that epitopes from structurally distinct regions of V required uptake and/or processing and loading of newly synthesized MHC class II molecules by the classical pathway for the induction of helper T-cell responses.
We mapped helper T-cell epitopes of the V antigen of Y. pestis and identified three major immunodominant MHC class II-restricted epitopes for inbred H-2d mice. The epitopes were distributed throughout the three distinct regions of V antigen, both globular domains and the intervening grip region of the ‘dumbell’ structure of V antigen.28 We and others27 have also shown that two of these epitopes are recognized by H-2b mice. The enteropathogenic yersinae Y. pseudotuberculosis and Y. enterocolitica express V antigens that are highly homologous with that of Y. pestis.9 In particular the amino acid sequence for all three V antigens (from acc. nos. P23994, P08495 and P21206, respectively) are identical or nearly so across the 20 amino acid peptides containing the three epitopes. Infection with either Y. pseudotuberculosis or Y. enterocolitica, or passive transfer of antibody specific for the V antigen of Y. pseudotuberculosis, can protect mice against challenge with Y. pestis.9,29 These authors interpreted the cross-protection as resulting from serological cross-reactivity. However, the degree of conservation of sequence of the T-cell epitopes we mapped across the V family suggests that vaccination with rV of Y. pestis induces helper T-cell responses that contribute to cross-protection against all enteropathogenic yersiniae, although this possibility remains to be demonstrated.
T-cell hybridomas specific for the three epitopes were generated to study how rV was processed and presented to CD4 T cells. The mechanisms of antigen processing and presentation of epitopes from all three regions of V were similar. The pattern of kinetics and dependence on rV uptake, low endosomal pH, protein synthesis and intact Golgi suggested that all epitopes were generated in acidic intracellular compartments for loading onto newly synthesized MHC class II molecules, consistent with presentation of the major immunodominant epitopes of V by the so-called classical antigen presentation pathway (reviewed by Robinson and Delvig30).
The low pH dependence suggested that lysosomal degradation of rV was required for presentation to T cells, which led us to investigate which of the major enzyme families were required for rV processing. Treatment of macrophages with broad-spectrum inhibitors of the major enzyme families during incubation with rV demonstrated an overall dependence on cysteine and serine proteinases, whereas the requirement for metallo-proteinase activity differed depending on the epitope, and aspartic proteinases did not appear to participate in V processing. Cysteine, in preference to aspartic, proteinases are thought to be the main lysosomal enzymes for both antigen and invariant chain processing in the classical pathway.19 We have previously implicated serine proteinases and metalloproteinases in antigen processing of other candidate bacterial vaccine antigens22–24 although their activity is not usually associated with lysosomes. The data reported here are consistent with cleavage of rV by serine and metallo-proteinases at some stage during endosomal transport, followed by processing by lysosomal cysteine proteinases and peptide loading of newly synthesized MHC class II molecules in the classical antigen presentation pathway. The V sequence includes only a single cysteine residue and the structure does not include a disulphide bridge, so the disulphide reductase GILT, a metalloproteinase implicated in antigen processing,31 is unlikely to participate in V processing.
We studied the mechanisms of antigen presentation by bone marrow-derived macrophages. It has long been known that phagosomes of monocytes and macrophages are primary sites of multiplication of Y. pestis early in infection.32,33 It has more recently been shown that Y. pestis infection of human monocytes in vitro induces their differentiation into dendritic cells.34 The APC type that is critical for immunity to plague has not yet been determined, although dendritic cells are considered essential APC in the induction of protective immunity against other intracellular Gram-negative pathogens.35,36 Both dendritic cells and macrophages have been used to study antigen presentation of antigens from bacteria, and the pathways used for particular epitopes are broadly similar.37–39 However, significant differences in antigen processing by macrophages, DC and B cells are emerging from recent studies.40,41
Few studies have determined the pathways used for presentation of multiple epitopes from candidate vaccine antigens.42,43 We have previously shown that CD4 T-cell epitopes from other bacterial proteins include representatives of both the classical and recycling pathways, depending on the structural context of the particular epitope.22–24 Antigen processing and MHC class II loading in the classical pathway has been localized to late endosomes or lysosomes,44–46 in contrast to alternative pathways in which mature MHC class II molecules are loaded at the cell surface47 or in early endosomes.43 Determining the pathways of presentation of multiple epitopes from candidate vaccine antigens in different APC types is of value in the development of adjuvants for vaccine delivery. Strategies for optimizing T-cell responses have been reported, such as the selective targeting of protein antigens to lysosomes by encapsulation within liposomes44,48 or by covalent attachment to the lysosome-seeking lysosomal integral membrane protein LIMP-II.49 However, lysosomal targeting by conjugation to LIMP-II was shown to enhance CD4 T-cell responses to a major epitope on one viral antigen (lymphocytic choriomeningitis virus nucleoprotein), but abolished the response to an epitope on another viral antigen (hepatitis B virus core protein),49 suggesting that lysosomal targeting does not necessarily enhance T-cell immunity. We propose that presentation of classical epitopes would be selectively enhanced by lysosomal targeting, whereas responses to recycling epitopes generated independent of lysosomal processing for loading mature MHC class molecules (reviewed by Robinson and Delvig30) would be induced poorly or not at all. However, we demonstrate here that immunodominant epitopes from the three major immunogenic regions of the V antigen of Y. pestis, are uniformly presented to helper T cells by the classical pathway following lysosomal processing. This is consistent with the reported robust protection of mice against challenge with Y. pestis by vaccination with rV,12–14 suggesting that responses to the full range of immunodominant epitopes were induced.
In conclusion, an understanding of the mechanisms of antigen processing of multiple epitopes of candidate vaccine antigens will contribute to the development of defined adjuvants to target subunit vaccines for the induction of optimal helper T-cell responses.