The interest in SSX proteins as immunotherapeutic targets began with the identification of SSX2 as the tumor antigen HOM-Mel-40 [22
], which had been shown to be the target of humoral immune responses in as many as 10% of patients with melanoma [22
]. In other studies humoral SSX2 immune responses were observed in 2 out of 74 colon cancer patient sera samples [99
], 1 out of 100 prostate cancer patient sera [74
], and antibody responses to SSX4, but not SSX2, have been identified in 2 out of 131 patients with gynecological cancers but not in healthy controls [100
]. SSX common antigen antibody responses have also been detected in sera from pancreatic, lung, breast, colon, and ovarian cancer patients but not in healthy control individuals [101
]. Additionally, antibodies specific for SSX1, 2, 3, and 4 have been found in the sera of cancer patients with melanoma, colon cancer, and/or breast cancer [27
]. In another study by Valmori et al. analysis of SSX4 IgG immune responses in epithelial ovarian cancer (EOC) patients showed that two patients out of 109 had both SSX2 and SSX4 IgG antibody responses [103
]. These responses were confirmed by ELISA, and both patients with dual antibody responses were alive and showed antibody responses >5 years after initial therapy.
Observing that cancer patients can have preexisting SSX2 IgG-specific immune responses, other investigators sought to identify if cancer patients can have cell-mediated immune responses to SSX proteins. Ayyoub et al. carried out the first investigation to identify SSX2 T-cell epitopes by utilizing an altered reverse immunology strategy [104
]. Purified standard proteasome complexes from human erythrocytes were incubated in vitro
with a library of overlapping SSX2 peptides covering the entire 188 aa protein sequence. The digested products were analyzed by mass spectrometry, which identified 12 peptides that were processed from the proteasome: SSX2p5–13, p7–15, p15–24, p16–24, p40–49, p41–49, p50–59, p53–61, p57–65, p58–67, p59–67, and p103–111. These peptides were subsequently cultured with cells from the tumor-infiltrating lymph node (TILN) of an SSX2-positive metastatic melanoma lesion (LAU 50), and an IFNγ
-ELISPOT was performed to detect SSX2 peptide-specific T-cell immune responses. SSX2 peptide p41–49 elicited the greatest T-cell immune response from the TILN, which led the group to synthesize HLA-A2 multimers containing SSX2 peptide p41–49 (). These multimers were used to stain the TILN cells and showed clearly positive SSX2 p41–49-specific CD8+ T cells. It was also shown that SSX2p41–49 CD8+ T cells (CTL clone LAU 50 E2.4) could lyse peptide-pulsed T2 cells and HLA-A2+ melanoma cell lines. Interestingly, when this clone was incubated with COS7 cells transfected with plasmids encoding HLA-A2 and SSX2 or SSX4, high levels of IFNγ
secretion and lysis were observed by the clone with SSX2-transfected cells but not SSX4-transfected cells. Since SSX2 and SSX4 only differ by 2 aa within the antigenic epitope region, this result suggested high specificity of the clone for SSX2 p41–49. None of the other predicted SSX2 epitopes were recognized by TILN from this patient.
Table 3 SSX epitopes: sequences, haplotype restrictions, and recognition. The known SSX immunogenic class I and class II MHC peptides or epitopes are shown. Amino acid sequence, haplotype restriction, tumor presentation, family member recognition, and references (more ...)
Following on this work, Ayyoub et al. utilized fluorescent HLA-A2/SSX2p41–49 multimers to determine the relative frequencies of SSX2 p41–49 CD8+ T cells in melanoma patients with SSX2-positive or negative tumors and in healthy donors [105
]. They found that SSX2 p41–49-specific CTLs were identifiable in both melanoma patients and healthy donors, although at lower frequency in healthy donors or patients with SSX2-negative tumors. They showed that p41–49-specific CTLs from the TILN and peripheral blood mononuclear cells (PBMCs) of patients with SSX2-expressing tumors were better able to lyse tumor cells compared to CTLs from the TILN and PBMCs from patients with SSX2-negative tumors or PBMCs from healthy donors. It was also shown that SSX2-specific CTL frequency increased with disease progression in at least one melanoma patient. SSX2 p41–49 T cells have also been identified in patients with hepatocellular carcinoma (HCC) [116
]. These CD8+ T cells were identified using SSX2 p41–49/HLA-A2 multimers in one out of six patients with HCC, both in the TILN and PBMCs. A polyclonal SSX2 p41–49 T-cell line generated from this patient lysed p41–49-pulsed T2 cells, melanoma cell line MEL 275, and a p41–49 peptide-pulsed HCC cell line. This was the second solid human malignancy with evidence of recognition by SSX2 p41–49-specific CTL.
The immunogenicity of SSX2 p41–49 was later highlighted when Rubio-Godoy and colleagues identified this peptide through an approach called positional scanning of synthetic combinatorial peptide library analysis [117
]. In this report, 3.1 × 1011
nonamer peptides, arranged in a positional scanning format, were screened for melanoma-reactive CTL of unknown specificity and assessed for their ability to elicit peptide-specific CTL. The identified optimal peptide sequence (AAAPKIFYA) was very similar to SSX2 peptide p41–49 (KASEKIFYV). Recognition of this epitope by melanoma-reactive CTL clone LAU 50/4D7 was confirmed by cytotoxicity assay, and interestingly, the native SSX2 peptide p41–49 had recognition that was in the same range as the optimal peptide. Attempts to modify the anchor residues of the optimally identified peptide or the native p41–49 epitope did not result in enhanced peptide recognition. Additionally, no cross-reactivity of p41 from SSX1, SSX3, SSX4, or SSX5 was observed. Further, a biometrical analysis of the screening data was used to generate a scoring matrix of all predicted peptides in public protein databases that could potentially be recognized by clone LAU 50/4D7, and, amazingly, this approach ranked SSX2 p41–49 27th out of approximately 16 million nonamer peptides [118
]. If this analysis was restricted to known tumor antigens in humans, this rank was 2nd out of 400,271 peptides. Ranked scores for other SSX peptides were much lower, however. It was noted that this method of scoring does not omit other SSX peptides as potentially immunogenic, but it does emphasize the immunogenicity of SSX2 p41–49.
In an effort to identify additional class I MHC-restricted SSX2 epitopes, Wagner et al. used a reverse immunology approach to identify SSX2 epitopes that were predicted to have affinity for the human HLA-A2 MHC class I complex using the SYFPEITHI algorithm [108
]. PBMCs from seven breast cancer patients and eleven healthy donors were evaluated for reactivity to peptides p41–49 and p103–111, and p167–175, which had all been predicted from the peptide-binding algorithm to have affinity for HLA-A2. They found that 5/7 (71%) of the HLA-A2+ breast cancer patients and 6/11 (55%) of the HLAA2+ healthy controls had T cells that were reactive to SSX2 peptide p103–111 by IFNγ
-ELISPOT assay. HLA-A2 restriction of the responses was verified using specific HLA blocking antibodies. SSX2 p103–111 was also shown to be a naturally presented SSX2 epitope by the recognition of SSX2+ SK-MEL-37 melanoma cell line and transfected COS7 cells by p103–11 peptide-specific CD8+ T cells (). Also, SSX2 p103–111-specific T cells could lyse peptide-pulsed target cells in cytotoxicity assays. Interestingly, no association was found between SSX2 antibody titer in breast cancer patient serum and p103–111-specific T-cell responses, indicating that the humoral and cell-mediated responses to this antigen are regulated independently. Rentzsch and colleagues found that PBMCs from one out of ten primary breast cancer patients exposed to SSX2-p103–111 significantly increased their mRNA expression of IFNγ
by QT-RTPCR analysis [109
]. This immune response was also detected in one of eleven healthy control individuals. We have also recently identified SSX2 p103–111 as an SSX2 epitope using HLA-A2 transgenic mice immunized with a genetic vaccine encoding SSX2, and further identified that p103–111-specific CTL can lyse HLA-A2-expressing prostate tumor cells (Smith, manuscript submitted).
Not only has the endogenous processing and presentation of SSX2 peptide p103–111 been demonstrated by reactivity of p103–111-specific CTL for SSX2-expressing tumor cells, it has also recently been shown that this peptide epitope is directly presented on the surface of cancer cells [73
]. By utilizing a human phage library screening technique Held et al. generated SSX2 p103–111-specific Fab antibodies that specifically recognized and bound to this peptide in the context of HLA-A2. These antibodies were used to stain melanoma cell lines in fluorescence microscopy studies, and it was observed that a majority of SK-Mel-37 cells expressed p103–111/HLA-A2, whereas Me275 cells expressed very little (<1% of cells). These cells were also used to stimulate SSX1 p103–111-specific T cell clones, and it was determined that the expression of p103–111/HLA-A2 on the cell surface, and not the total SSX2 protein expression levels of the cells, was correlated with enhanced CTL recognition and activation.
Ayyoub et al. also identified the first CD4 class II MHC SSX2 epitope [110
]. This epitope was mapped to aa 19–34 of SSX2 using truncated peptide assays and was recognized by CD4+ T cells from an SSX2-expressing melanoma patient (). No responses to this epitope were found in healthy controls, and class II blocking antibody experiments verified that this epitope is recognized in the context of HLA-DPB1*0101. Many studies have documented the atypical expression of class II MHC molecules on the surface of colon carcinomas, breast cancer cells, sarcomas, and melanoma cells [120
]. This expression prompted the investigators in the current study to evaluate whether SSX2 p19–34 is processed endogenously and presented on the surface of tumor cells. Interestingly, SSX2 CD4+ T cells failed to recognize IFNγ
-treated melanoma cells expressing SSX2 and class II MHC molecules (HLA-DP); however, they were stimulated by antigen-loaded dendritic cells. This result suggests that p19–34 may not be endogenously presented on tumor cells but can
be loaded exogenously and presented by APCs. Ayyoub et al. also identified an HLA-DRB*1101-restricted SSX2 class II peptide p45–59 [112
]. CD4+ T cells specific for this epitope were detected among PBMCs and in the TILN from melanoma patients but not in healthy control individuals. Using the SYFPEITHI algorithm, p45–59 was ranked as the 2nd highest predicted HLA-DR-binding peptide for SSX2, whereas this peptide ranked 3rd, 13th, and 2nd for SSX1, SSX4, and SSX5, respectively. Titrated concentrations of peptides incubated with p45–59-specific CD4+ T cells showed cross-reactivity to SSX5 p45–59 and low-level reactivity to SSX1 p45–59 in IFNγ
ELISA assays. Eleven of nineteen melanoma patients had CD4+ T cells responses to this epitope region (i.e., aa 37–58). Contrary to what was found for the class II peptide p19–34, T cells recognized epitope p45–59 on both antigen-loaded DCs as well as HLA-DR+ tumor cells. This peptide was also described by Neumann and colleagues in which p45–59 was found to be restricted to HLA-DRB1 subtypes *0701, *1101, *1302 and B3*0301, and it was demonstrated that p45–59 CD4+ T cell responses could be induced in 3/6 of breast cancer patients and 1/5 of healthy controls [113
]. However, no correlation was found between SSX2 antibody titer and CD4+ T cell responses.
After the identification of the first two class II SSX2 epitopes Ayyoub and colleagues identified yet another CD4 peptide that overlapped with these epitopes [111
]. This epitope region was narrowed down to amino acids p37–51 of SSX2 by truncation experiments and was shown to be restricted to HLA-DRB*0301 by class II MHC blocking antibody assays. Since this was the third immunogenic peptide identified in the KRAB domain region of SSX2, it was theorized that this region may be a “hot spot” for T-cell recognition. Again, wild type melanoma cell lines expressing MHC class II molecules or tumor cells transfected with plasmid encoding SSX2 were not recognized by epitope-specific CD4+ T cells directly, however, p37–51 peptide-pulsed tumor cells were
recognized by both SSX2 p37–51-specific CD4+ T cells and SSX2 p41–49 peptide-specific CD8+ T cells. Of note, the transfected cells were recognized by the p41–49-specific T cells but not the p37–51-specific CD4+ T cells, confirming that the p41–49 peptide can be endogenously processed and presented by tumor cells. Cross reactivity of SSX2 p37–51-specific CD4+ T cells was observed for SSX4 and SSX5 but not SSX1 or SSX3 p37–51 peptides in an IFNγ
peptide titration ELISA assay.
Since it was previously observed that SSX4 was expressed in ~20% of malignant melanomas in a report by Türeci et al. [76
], Ayyoub and colleagues also evaluated the presence of CD4+ T cells specific to this antigen in melanoma patients by stimulating CD4+ T cells isolated from the PBMCs of four melanoma patients with overlapping peptides derived from the amino acid sequence of SSX4 in the presence of APCs [115
]. Intracellular IFNγ
cytokine staining revealed that all four patients had CD4+ T cells that were activated in the presence of at least one of the SSX4 peptides (). Again, as was found for the SSX2 class II peptides, CD4+ T cells could not recognize melanoma cells directly but were activated by APCs loaded with SSX4 antigen, which suggested that this peptide may not be presented through the endogenous pathway in tumor cells. Interestingly, five out of the seven peptides identified in this study mapped to the KRAB domain of SSX4. Four previously identified SSX2 epitopes also mapped to this region, further suggesting that this may be a “hot spot” for T-cell recognition. Valmori et al. found that epithelial ovarian cancer (EOC) patients also had SSX4 CD4+ T cells that recognized some of these epitopes [103
Several of the SSX4 class II peptides that were identified previously were later reported by Godefroy et al. to also be epitopes encoded from the SSX1 protein [114
]. In this study an overlapping pool of peptides spanning the entire SSX1 protein sequence was incubated with PBMCs from cancer-free individuals; SSX1-peptide-specific CD4+ T cells were identified in 5/5 of the donors analyzed. These T-cell populations almost exclusively recognized epitopes from three defined regions of the protein sequence, including the KRAB domain and two C-terminal regions of the protein that are retained in the SS18-SSX fusion construct () [115
]. Evaluating the cross-reactivity of these peptide-specific clones for other SSX family members, however, revealed only limited recognition of p41–60 for SSX5. This lack of cross-reactivity was explained by the presence of several amino acid differences in this epitope region between family members. As before, no endogenous processing and presentation of SSX1 epitopes by tumor cells was demonstrated.
Since it had been shown that tumors can express more than one SSX family member at the same time and that T-cell responses can be found to multiple SSX family members in cancer patients, He et al. assessed the utility of an altered peptide ligand strategy to identify class I epitopes that could be used to target multiple SSX family members [107
]. Using peptide prediction algorithms He and colleagues identified SSX p57–65 and p99–107 as shared epitopes between family members SSX1–9. Since SSX p57–65 had the higher binding score for all nine family members, this peptide was selected for further analysis. Using T2 binding assays with four altered p57–65 peptides specific to SSX1–9, it was shown that altered peptide P4 (AMTKLGFNV), which is encoded by SSX6 and SSX8, had the greatest in vitro
binding affinity for HLA-A2, and this affinity was stable at low peptide concentrations. All four altered p57–65 peptides were shown to elicit peptide-specific CTL from the PBMCs of healthy HLA-A*0201 individuals, however P4-specific CTL showed the greatest lysis and IFNγ
secretion when incubated with peptide-pulsed target cells. Additionally, P4 showed the greatest cross-reactivity to other altered p57–65 peptides presented by target cells in cytotoxicity assays. HLA-A2.1/Kb
mice immunized with P4 generated CTL that could be cultured ex vivo
, and were shown to also lyse T2 cells displaying the immunizing peptide or the three other altered p57–65 peptides, which suggests that this altered peptide strategy might be successfully utilized to target multiple SSX family members ().