Targeting phosphorylated p53 to elicit tumor-reactive T helper responses against head and neck squamous cell carcinoma

ABSTRACT The human T cell receptor is capable of distinguishing between normal and post-translationally modified peptides. Because aberrant phosphorylation of cellular proteins is a hallmark of malignant transformation, the expression of the phosphorylated epitope could be an ideal antigen to combat cancer without damaging normal tissues. p53 activates transcription factors to suppress tumors by upregulating growth arrest and apoptosis-related genes. In response to DNA damage, p53 is phosphorylated at multiple sites including Ser33 and Ser37. Here, we identified phosphorylated peptide epitopes from p53 that could elicit effective T helper responses. These epitope peptides, p5322-41/Phospho-S33 and p5322-41/Phospho-S37, induced T helper responses against tumor cells expressing the phosphorylated p53 protein. Moreover, chemotherapeutic agents augmented the responses of such CD4 T cells via upregulation of phosphorylated p53. The upregulation of phosphorylated p53 expression by chemotherapy was confirmed in in vitro and xenograft models. We evaluated phosphorylated p53 expression in the clinical samples of oropharyngeal squamous cell carcinoma and revealed that 13/24 cases (54%) were positive for phosphorylated p53. Importantly, the lymphocytes specific for the phosphorylated p53 peptide epitopes were observed in the head and neck squamous cell cancer (HNSCC) patients. These results reveal that a combination of phosphorylated p53 peptides and chemotherapy could be a novel immunologic approach to treat HNSCC patients.


Introduction
Despite recent advances in treatment, the survival rate of head and neck squamous cell carcinoma (HNSCC) patients has not improved for decades. Because advanced stages are initially detected in most patients, their prognosis is poor. The clinical use of nivolumab, an immune checkpoint inhibitor, revealed that the immunotherapy could be a novel treatment for HNSCC patients. However, the biomarker for select responders has not yet been found, and the expensive cost of the reagents hinder the broad application of immune checkpoint inhibitors. [1][2][3] While some of the researchers have considered peptide vaccines as an old concept, optimal peptide conformation, immune adjuvants, and route of administration eliminated cancer in preclinical studies. 4 Moreover, peptide vaccines have several advantages over other therapies such as costeffectiveness, ease of synthesis, and a broad acceptance in clinical practice. 5 Because CD4 helper T lymphocytes (HTLs) play a key role in direct killing of tumor cells in addition to their helper function for CD8 cytotoxic T lymphocytes (CTLs), 6 the identification of a helper epitope from a tumor-associated antigen (TAA) is necessary to develop an effective peptide vaccine. Post-translational modifications, such as phosphorylation and acetylation, are unique targets in selecting a TAA. 7 Andersen et al has described that CTLs can recognize the phosphorylated peptide as a completely different antigen compared to the non-phosphorylated peptide. 8 In addition to CTLs, it has been reported that the T cell receptor (TCR) of HTLs has an ability to recognize a phosphorylated peptide derived from melanoma antigen (MART-1) recognized by T cells. 9 Because phosphorylation is a hallmark of malignant transformation, 10 phosphorylated peptide-targeted immunotherapy can be a novel approach to attack cancers without damaging normal tissues that weakly express TAA. 7 The normal tissues that only express nonphosphorylated TAA might be spared in the phosphorylated antigen-targeted immunotherapy. Because MART-1 is only expressed in melanoma, the design of a phosphorylated peptide using a ubiquitous TAA would be valuable for the treatment of many types of cancer. p53 is an activator of proapoptotic signaling and widely expressed in tumors. More than 50% of human malignancies have a p53 mutation, and most of them are phosphorylated and acetylated. [11][12][13] Accordingly, p53 is a candidate TAA to identify phosphorylated helper epitope peptides. Recently, we hypothesized that malignancy associated post-translated modifications could have a high immunogenicity and showed that acetylated p53 proteins served as the target of anti-tumor immunity. 7 Not only hyperacetylation but also hyperphosphorylation have been observed in malignancies, and they occur at various stages of oncogenesis.
In the present study, we show that peptides p53 22-41/ Phospho-S33 and p53 22-41/Phospho-S37 (referred as p-p53 S33 and p-p53 S37 ) were able to elicit antigen-specific, tumor-reactive HTL responses. The treatment of tumor cells with chemotherapeutic reagents enhanced these responses. These results suggest that the tumor-associated phosphorylated peptide vaccines with chemotherapy could be a promising approach in the HNSCC immunotherapy.

Results
Expression of phosphorylated p53 in oropharyngeal squamous cell carcinoma specimens We first analyzed the expression of p53 and phosphorylated p53 in surgical specimens of oropharyngeal squamous cell carcinoma (OPSCC) by immunohistochemistry. The clinical parameters of 24 OPSCC patients are described in Supplemental Table 1. As shown in Fig. 1A, p53 was expressed in 18/24 cases (75%) of OPSCC. Phospho-p53 S33 and phospho-p53 S37 were expressed in 13/24 cases (54%) and 10/24 cases (42%), respectively ( Fig. 1B and C). Ten cases (42%) of OPSCC were double positive for phospho-p53 S33 and phospho-p53 S37 , indicating that the multiple serine residues of p53 are simultaneously modified in OPSCC. The expressions of HLA-DR and p16 (a surrogate marker of human papilloma virus infection) were found in 7/24 cases (29%) and 14/24 cases (58%), respectively (Supplemental Table 1). Phosphorylated p53 expression was not related to other parameters such as tumor stage, p16 expression, or HLA-DR expression (Supplemental Table 2). These results provide a rational basis for the development of immunotherapy to treat HNSCC by targeting phosphorylated p53 proteins.
p53 Ser33 and Ser37 are immunogenic phosphorylation sites that induce CD4 T cell responses To confirm the immunogenicity of phosphorylated p53 peptides, we used the mono-phosphorylated peptides (p-p53 S33 or p-p53 S37 ) to elicit peptide-specific HTLs. As shown in Fig. 3 A, we successfully isolated HTLs specific for phosphorylated peptide p53 Ser33 (K2 and J44 from donor 3 and donor1, respectively) and p53 Ser37 (H2 from donor 4) but not the non-phosphorylated wt p53 22-41 peptide. HLA-DR molecules restricted the reaction of these HTLs (Fig. 3B). To define the restricting HLA class II molecules, L-cells transfected with individual HLA-DRs were used as APCs. As shown in Fig. 3C, p-p53 S33 peptide-specific HTLs K2 and J44 recognized the phosphorylated p53 peptide in the context of HLA-DR1, whereas the response of p-p53 S37 peptide-specific HTLs H2 was restricted by HLA-DR9.

An increase in phosphorylated p53 was induced by treatment with chemotherapeutic reagents in vitro and in murine xenograft models
To test the ability of phosphorylated p53 peptide-reactive HTLs to directly recognize tumors, we first examined the expression of phosphorylated p53 in tumor cells. Western blot analysis showed that phosphorylated p53 was expressed in several tumor cell lines, including HNSCC ( Fig. 4A-C). EBC1 (lung SCC) and SAS (HNSCC) cells expressed neither wt nor phosphorylated p53. Because DNA damage induces the phosphorylation of p53, tumor cells were treated with chemotherapeutic reagents. As previously reported, [16][17][18] phosphorylated p53 expression was augmented by the addition of doxorubicin and cisplatin, a key drug for HNSCC treatment (Fig. 4C).   In clinical settings, it is important to have an increase in phosphorylated p53 in tumors in vivo. Consistent with western blot findings, we observed that the positive staining of tumor cells for phosphorylated p53 (Fig. 5A-D), and total p53 levels (Supplemental Fig. 1A-D) were increased in murine xenograft tumors from cisplatin or doxorubicin treated mice. It is noteworthy that the chemotherapy induced a diffuse expression of phosphorylated p53 in the tumor. These findings suggest that such anti-cancer drugs enable not only an induction of immunogenic cell death (ICD) 19,20 but also enhancement of tumor antigens.
Direct recognition of phospho-p53 positive tumor cells by phosphorylated peptide-reactive CD4 T cells Next, we evaluated whether phosphorylated p53-reactive HTLs could directly recognize phosphorylated p53 expressing tumor cells. The peptide-reactive HTLs and HLA-DRmatched phosphorylated p53-positive tumor cells were cocultured, and IFN- production in the culture supernatant was measured. As shown in Fig. 6A, all phosphorylated p53-specific HTLs (K2, J44, and H2) could directly recognize the HLA-DR-matched phosphorylated p53-positive tumor cells but not HLA-DR-matched phosphorylated p53-negative tumor cells in an HLA-DR dependent manner. These results indicate that the phosphorylation of p53 is maintained through the antigen processing machinery and the phosphorylated peptides in tumor MHC class II molecules are naturally presented to HTLs.
Because the Western blot analysis presented in Fig. 4 showed that p53 phosphorylated at Ser33 or Ser37 and total p53 protein expression levels in tumor cells were enhanced by the treatment with doxorubicin or cisplatin, tumor cells were pretreated with these reagents before coculturing with HTLs. As shown in Fig. 6B, the responses of p-p53 S33 or p-p53 S37 -reactive HTLs against tumor cells were augmented by treatment of the tumor cells with doxorubicin or cisplatin. Most importantly, granzyme B production from HTL clone J44 against tumor cells was observed with cisplatin treatment, indicating that the HTL had a cytotoxic effect on tumor cells (Fig. 6C). These results suggest that the accumulation of DNA damage by chemotherapy can act as an adjuvant to amplify expression of phosphorylated p53 and augment the efficacy of phosphorylated p53-targeted peptide vaccines.

Recognition of phosphorylated p53 peptides by PBMCs from cancer patients
The existence of precursor T cells specific for the phosphorylated peptides in the PBMCs of HNSCC patients is a prerequisite for the application of a phosphorylated p53 peptide vaccine in the clinic. To evaluate the responses of T cells from HNSCC patients to p-p53 S33 and p-p53 S37 , we carried out a short-term culture using peptide-stimulated PBMC from five HNSCC patients. Since the volumes of blood obtained from these patients were small, we could not establish T cell lines to perform more detailed studies or HLA restriction analyses. A tetanus toxoid peptide (TT 830-843 ) was used as a positive control because this peptide generates robust HTL responses regardless of the HLA-DR alleles. 21 As shown in Table 1, substantial T cell responses to phosphorylated p53 peptides were observed in HNSCC patients but not in healthy donors, suggesting that the precursor of phosphorylated p53-reactive HTLs exists in HNSCC patients. Moreover, HTL responses to p-p53 S33 was  higher than that to p-p53 S37 . This result indicates that phosphorylated Ser33 might be more immunogenic than phosphorylated Ser37, which is consistent with the results in Fig. 2A.

Discussion
In this study, we demonstrated that the phosphorylated peptides from p53 tumor suppressor proteins are candidates for a helper peptide vaccine to treat HNSCC. Since MDM2 tightly regulates the amount of p53 under normal conditions, the amount of p53 is maintained at a very low level. On the other hand, tumors frequently express an aberrant amount of p53. 22 Consequently, it appears that self p53-reactive T cells, which have a low affinity for TCR, might only react to tumor (high burden of p53) but not to normal tissues (low burden of p53). Because only lymphocytes from HNSCC patients but not from healthy donors reacted to phosphorylated p53 peptides in a short-term culture (Table 1), the increased amount of p53 in tumors may spontaneously increase the number of p53reactive HTLs. In response to DNA damaging reagents or ultraviolet irradiation, the phosphorylation of p53 at multiple sites including transactivation domain (Ser33 and Ser37) increases the stability and activity of p53 by decreasing its ability to bind MDM2 in tumors. 13,23,24 Because T cells have been reported to precisely recognize phosphorylated epitopes and post-translationally modified epitopes on MHC molecules, 25-27 the phosphorylated epitope peptides from p53 would be an ideal antitumor vaccine to combine with conventional therapies. Phosphorylated p53 peptides bound to multiple HLA-DR alleles (Fig. 2C), indicating that these peptides can be applied to a broad population. There are multiple phosphorylated serine residues in p53, including Ser6, Ser9, Ser15, Ser18, Ser20, Ser33, Ser37, Ser46, Ser372, and Ser392. 11, 28 We targeted Ser33 and Ser37 because they are located close to the sequence previously reported for eliciting HTL responses against wt p53 peptides. 14 There is an overlap between the peptide sequence that we selected and the p53 [25][26][27][28][29][30][31][32][33][34][35] peptide previously defined as a CTL epitope, indicating that this peptide has a capacity to activate p53-reactive CTLs in addition to p53-reactive HTLs. 29 Although the phosphorylation of p53 might induce phosphorylated peptide-specific CTL responses, 8 it is also possible that the phosphorylation of the p53 25-35 peptide might attenuate the stability of the MHC-non-phosphorylated peptide-TCR complex to stimulate CTLs. Further research is required to examine whether our peptide has an ability to elicit phosphorylated p53-reactive CTLs. The stability of phosphorylated peptides in vivo also remains to be elucidated.
While T-cell based immunotherapy mainly focuses on CTLs, HTLs have cytotoxic activity against tumor cells in addition to their helper function. 6 In addition to the pMART-1-specific HTL response, 9 this report opens the avenue to utilize phosphorylated serine bearing peptides as helper epitope peptides. To induce robust responses to peptide vaccines in vivo, the appropriate selection of adjuvants is indispensable. Toll-like receptor (TLR) 3 or 7 ligand is one of the desirable adjuvants for an antitumor peptide vaccine. TLR3 and TLR7 ligands induce the expression of MHC Class II on DCs, 30 which plays an indispensable role in activating HTLs through the presentation of peptide. Interestingly, imiquimod, a TLR7 ligand, induced apoptosis through p53 activation. 31 However, p53 signaling increases the expression of TLR3, 32,33 indicating that the TLR3 or 7 ligand elicits p53 expression in a positive feedback loop. Thus, these adjuvants may work in two different ways: 1) activation of APCs or 2) the upregulation of antigens (p53). Because we previously reported that the EGFR inhibition augments the expression of MHC Class II on HNSCC cells, 34 an EGFR inhibitor that has been currently available as a treatment option in the patients with HNSCC could be an interesting adjuvant to combine with the phosphorylated p53 peptide vaccine.
During the malignant transformation, post-translational modifications play a pivotal role. 35 A number of tumor cell types tend to accumulate the p53 protein, which mainly exhibits phosphorylation and acetylation. 11 Because the DNA-damaging reagents such as cisplatin and doxorubicin augmented the expression of phosphorylated p53, the recognition of the tumor was augmented by phosphorylated p53specific HTLs (Fig. 4, 6B, and 6C). Having been first-line treatments for HNSCC patients, 36 platinum-based drugs are acceptable to use as an adjuvant. Cisplatin has been reported to upregulate the expression of antigen processing machinery components. 37 Since the induction of ICD has been appreciated in doxorubicin, the effect of cisplatin in inducing ICD might be inferior to its derivative oxaliplatin. 20 However, it is still possible that cisplatin stimulates APCs by inducing HMGB1 from dead tumor cells. 19 Interestingly, p53 activation might also play a role in stimulating innate immunity via the induction of HMGB1 and ATP from tumor cells. 38,39 Thus, chemotherapeutic reagents activate APCs and increase the amount as well as the presentation of antigens in tumors. Moreover, the production of IFN- from T cells can counteract platinum-resistance by altering the metabolism in cancerassociated fibroblasts, indicating that the cisplatin-induced infiltration of T cells renders tumors sensitive to chemotherapy. Collectively, non-lymphoablative chemotherapy and T-cell-targeted immunotherapy could attack tumors by complementing each other.
In conclusion, we have identified novel helper epitopes from phosphorylated tumor protein p53 (p-p53 S33 and p-p53 S37 ) that can induce antigen-specific HTL responses. The variant of amino acid sequences of these peptides has not been reported, 40 suggesting that these peptides are wellconserved. Furthermore, cisplatin and doxorubicin may work as adjuvants for immunotherapy by upregulating phosphorylated p53. This approach could be a desirable therapy to treat HNSCC with chemotherapy, which expresses a high level of phosphorylated p53. The p53 staining patterns were characterized into three categories: (a) ++: overexpression (at least 50% of the tumor cells express nuclear staining); (b) +: normal expression (no more than 49% of the tumor cells express nuclear staining); and (c) -: negative when there was a complete absence of staining in the tumor. The p53 expression in neighboring normal tissues served as an internal positive control. In order to quantify the number of DAB staining cells, we used a BZ-X710 microscope (Keyence, Tokyo, Japan) with the BZ-H3░C hybrid cell count software (Keyence).

Measurement of antigen-specific responses with CD4 T cells
An evaluation of HTL responses to diverse antigens was performed as previously described. 34 The tumor cells were treated with IFN-γ (500 IU/ml, PeproTech, Rocky Hill, NJ) for 48 hours to induce the expression of HLA-DR before culturing with HTLs. To enhance phosphorylation of p53 in tumor cells, the tumor cells were treated with or without 1 μM doxorubicin (CST, Denver, CO), or cisplatin (5 or 10 μM; R&D Systems, Minneapolis, MN) during the last 24 hours of IFN-γ treatment.

Granzyme B ELISA
To evaluate the cytotoxic activity of HTL clones, the supernatant from antigen-specific responses by CD4 T cells were collected and assessed for Granzyme B by ELISA (MABTECH, Stockholm, Sweden) according to manufacturer's instructions.
Analysis of wild-type p53 and phosphorylated p53 expression in a xenograft model Six-week-old female BALB/c-nu/nu mice were obtained from Charles River Japan (Yokohama, Japan) and were maintained in accordance with the animal facility at the Asahikawa Medical University. Five mice were injected subcutaneously with 2 × 10 6 HSC4 cells in a flank. When the tumor reached 5 mm, PBS (control), cisplatin (7.5 mg/kg), or doxorubicin (5 mg/kg) was injected into the peritoneal cavity. Twenty-four hours later, mice were sacrificed, and tumor samples were collected for evaluating the p53 expression by immunohistochemistry. All animal experiments were approved by the Institutional Animal Care and Use Committee of Asahikawa Medical University.

Measurement of peptide-specific responses in cancer patients
Peripheral blood lymphocytes were isolated from fresh heparinized whole blood of HNSCC patients by Lymphoprep TM (Axis-Shield PoC AS, Oslo, Norway)-centrifugation, and cultured with peptides (10 μg/mL) in 96-well plates as described previously. 7,43 The lymphocytes were stimulated for 2 cycles (7 days/cycle). Seven days after the second peptide stimulation, production of IFN-γ or GM-CSF was assessed by ELISA (BD PharMingen, #555142). The institutional ethics committee approved this study protocol (approval number #16040).

Statistical analysis
The data were analyzed using Student's t-test and Chi-squared test. p < 0.05 indicates statistical significance. Error bars in the data indicate the standard error of the mean (SEM). GraphPad Prism 6 was used for the analyses.

Disclosure of Potential Conflicts of Interest
Esteban Celis has filed patent applications based on the use of synthetic peptides and poly-IC combinatorial vaccines. The rights of the patent applications have been transferred to the Moffitt Cancer Center (Tampa, FL). The other authors declare no conflict of interest.