Brachyury-targeted immunotherapy combined with gemcitabine against head and neck cancer

Brachyury is a transcription factor belonging to the T-box gene family and is involved in the posterior formation of the mesoderm and differentiation of chordates. As the overexpression of Brachyury is a poor prognostic factor in a variety of cancers, the establishment of Brachyury-targeted therapy would be beneficial for the treatment of aggressive tumors. Because transcription factors are difficult to treat with a therapeutic antibody, peptide vaccines are a feasible approach for targeting Brachyury. In this study, we identified Brachyury-derived epitopes that elicit antigen-specific and tumor-reactive CD4+ T cells that directly kill tumors. T cells recognizing Brachyury epitopes were present in patients with head and neck squamous cell carcinoma. Next, we focused on gemcitabine (GEM) as an immunoadjuvant to augment the efficacy of antitumor responses by T cells. Interestingly, GEM upregulated HLA class I and HLA-DR expression in tumor, followed by the upregulation of anti-tumor T cell responses. As tumoral PD-L1 expression was also augmented by GEM, PD-1/PD-L1 blockade and GEM synergistically enhanced the tumor-reactivity of Brachyury-reactive T cells. The synergy between the PD-1/PD-L1 blockade and GEM was also confirmed in a mouse model of head and neck squamous cell carcinoma. These results suggest that the combined treatment of Brachyury peptide with GEM and immune checkpoint blockade could be a promising immunotherapy against head and neck cancer.


Introduction
The most common histological type of head and neck cancer is squamous cell carcinoma, which accounts for 85% of these cases [1]. In 2018, the estimated number of new patients with head and neck squamous cell carcinoma (HNSCC) was 890,000, and approximately 450,000 patients died of Hidekiyo Yamaki, Michihisa Kono and Risa Wakisaka have been contributed equally to the work. this disease [2,3]. In the early stages, HNSCC, including oral, pharyngeal, and laryngeal cancers, may be curable with conventional treatments, such as surgery, radiation, and chemotherapy. However, more than 65% of patients with advanced HNSCC develop recurrent or metastatic disease [4]. Immune checkpoint inhibitors (ICIs) have recently been approved for the treatment of patients with recurrent or metastatic HNSCC. Although ICIs have proven the efficiency of immunotherapy in HNSCC, only a small percentage (approximately 20%) of patients with HNSCC have acquired a long-term benefit from ICIs [5,6]. Therefore, the development of novel cancer immunotherapies for HNSCC is required.
Brachyury is a transcription factor belonging to the T-box gene family and is implicated in the posterior formation of the mesoderm and differentiation of the chordate. Brachyury is mainly expressed in the embryonic stages, and its expression is negative in most mature cells, except in the testis and thyroid [7,8]. Recent studies have shown that Brachyury is expressed in various cancers, including lung [8], breast [9], rectal [10], and oral cancers [11]. Therefore, Brachyury is considered an attractive cancer-testis antigen for peptide vaccines that may specifically induce anti-tumor T cells with a relatively low risk of autoimmunity. While Palena et al. [7] have identified the HLA-A0201-binding CD8 + cytotoxic T lymphocyte (CTL) epitope in Brachyury protein, no studies have examined whether Brachyury contains an epitope to stimulate CD4 + helper T lymphocytes (HTLs) and whether this antigen is expressed in HNSCC other than oral cancer.
Recent evidence has suggested that modification of peptide sequences [12,13], routes of administration [14,15], immunoadjuvants [14,16,17], and administration protocols [18,19] can augment the efficacy of peptide vaccines. To promptly translate in clinic, chemotherapy, which has an immunological activity, is an ideal adjuvant [20]. Gemcitabine (GEM) is a pyrimidine antimetabolite that inhibits DNA synthesis by converting it to triphosphate in cancer cells. In addition to pancreatic, lung, and breast cancers, HNSCC, including nasopharyngeal cancer, can be a target of GEM [21][22][23][24]. Interestingly, GEM has been shown to have immunomodulatory effects without harming cellular immunity [25,26] and could be a promising immunoadjuvant in peptide vaccines.
In the present study, we evaluated the expression of Brachyury in HNSCC, and established Brachyury-targeted immunotherapy. Brachyury was found to be expressed in HNSCC tissues and cell lines. We identified novel helper epitopes from the Brachyury protein that can induce peptidespecific HTLs, which could directly kill Brachyury-expressing HNSCC cell lines in an HLA-DR-restricted manner. The precursor of Brachyury-reactive HTLs exists in patients with HNSCC. Furthermore, we showed that GEM upregulated HLA class I and HLA-DR expression in HNSCC cell lines, followed by augmenting tumor recognition by Brachyuryreactive HTLs. As GEM also upregulated the expression of PD-L1 in tumor cells, PD-1 blockade further upregulated peptide-specific T cell responses to GEM. The antitumor activity of GEM combined with PD-1 blockade was confirmed in a mouse syngeneic HNSCC model. These results suggest that the Brachyury-targeted HTL vaccine may be a promising approach for treating HNSCC with immunotherapy, and GEM combined with ICIs can be a powerful immunoadjuvant in T cell-based immunotherapy.

Immunohistochemistry (IHC)
Pretreatment biopsy tissues were obtained from the primary site of 52 patients with HNSCC treated at Asahikawa Medical University. The clinical characteristics of the patients including hematological parameters such as Neutrophil-Lymphocyte ratio (NLR) are presented in Supplemental Table 1. TNM staging was based on the 8th edition of the International Union against Cancer staging system. Brachyury expression was analyzed in formalin-fixed and paraffinembedded tissues (FFPE) from patients with HNSCC. The expression levels of MHC Class I, Class II, and PD-L1 were examined in the mouse tumor samples. Anti-human Brachyury (A-4, 1:100; Santa Cruz) mouse monoclonal antibody (mAb) was used for human tumors, and anti-mouse MHC class I (anti-H-2 Kb Ab, AF6-88.5, 1:100, BioLegend), anti-mouse MHC class II (I-A/I-E Ab, M5/114.15.2, 1:100, BioLegend), and anti-PD-L1 (1:100, Proteintech) antibodies (Abs) were used as primary Ab for mouse tumors. FFPE specimens were stained with VENTANA Benchmark GX (Roche Diagnostics) using Cell Conditioning 1 buffer (Roche Diagnostics) as an antigen retrieval solution and VENTANA ultraView Universal DAB Detection Kit (Roche Diagnostics). Staining intensity scores for Brachyury in tumor cells were graded as follows: 0, no staining; 1, weak; 2, moderate; and 3, strong. Quantity scores for Brachyury consisted of the percentage of positively stained tumor cells and were graded as 0, < 5%; 1, 5-25%; 2, 26-50%; and 3, > 50%. The IHC score was calculated as the sum of the staining intensity score and quantity score with reference to previous reports [27,28]. IHC scores ≥ 4 were defined as high expression and scores < 4 as low expression. The collection and analysis of clinical data were approved by the Institutional Ethics Committee of Asahikawa Medical University (#16,217). The correlationship between Brachyury expression and the survival of HNSCC patients was analyzed using The Human Protein Atlas (https:// www. prote inatl as. org/) [29].

Western blotting
Tumor cell line proteins were extracted using a MiuteTM Total Protein Extraction Kit (Invent Biotechnologies, Inc.). These samples were subjected to electrophoresis on NuPAGE Bis-Tris gels (Invitrogen) and transferred to an Immobilon-P membrane (Merck Millipore). The membrane was incubated with mouse anti-human Brachyury (A-4, Santa Cruz) and mouse anti-human β-actin Abs (C4, Santa Cruz), and protein expression was detected via chemiluminescence using the Amersham ECL Prime Western Blotting Detection System (GE Healthcare Life Sciences) and Invitrogen iBright Imaging Systems 1500 (Invitrogen). Protein expression was analyzed using ImageJ ver. 1.54a.

Flow cytometry
After

In vitro generation of Brachyury peptide-reactive CD4+ T cells
The procedure used to elicit peptide-reactive CD4 + T cell (HTL) lines from healthy donor peripheral blood mononuclear cells (PBMCs) has been described in detail previously [32]. In brief, dendritic cells (DCs) were induced by stimulating CD14 + cells and isolated using the EasySepTM Human CD14 + Positive Selection Kit II (STEMCELL technology), with GM-CSF (50 ng/ml, PeproTech, Rocky Hill) and IL-4 (1000 IU/ml, PeproTech, Rocky Hill, NJ). HTLs isolated using the EasySepTM Human CD4 + T Cell Isolation Kit (STEMCELL Technology) were stimulated by peptide-pulsed autologous DCs for one cycle and γ-irradiated autologous PBMCs for two cycles. HTLs were assessed for IFN-γ production with or without Brachyury peptide stimulation using ELISA kits (BD Pharmingen), according to the manufacturer's instructions. Microcultures with significant IFN-γ production after Brachyury peptide stimulation were subsequently proliferated, and HTL lines were isolated by limiting dilution.
(ATP) production was measured using a human HMGB1 ELISA kit (Arigobio) and an ATP Assay kit (Abcam). AIM-V medium (Invitrogen) supplemented with 3% human male AB serum (Innovative Research) was used as the complete culture medium for T cell experiments. All blood samples were collected after obtaining written informed consent.

Cytotoxicity assay
Supernatants of Brachyury-reactive HTL lines co-cultured with tumor cell lines were assessed using Human Granzyme B DuoSet ELISA kits (R&D Systems). To evaluate direct killing activity, target tumor cell lines were labeled using the CellTraceTM CFSE Cell Proliferation Kit (Invitrogen). After 6 h of co-culturing with various effector/target cell (E:T) ratios of Brachyury-reactive HTL lines and tumors, the number of dead tumor cells, labeled using 7-AAD viability staining solution (BioLegend), was quantified by flow cytometry.
The viability of HNSCC cells was assessed using Cell Titer 96 ® Queous One Solution Reagent (Promega), according to the manufacturer's instructions.

Brachyury-reactive T cell responses in patients with HNSCC
PBMCs from eight HNSCC patients were co-cultured with Brachyury peptides or PADRE peptide in 96-well plates, as described previously [33]. PBMCs (1 × 10 5 ) were stimulated with peptides (10 µg/mL) for two cycles every week, and IFN-γ production in the supernatants was measured using ELISA. All investigations were approved by the Institutional Ethics Committee of Asahikawa Medical University (#16,217), and written informed consent was obtained from all participants.

In vivo assessment of combination therapy with GEM and ICIs
C57BL/6 mice (female, 6-8 weeks old) were purchased from Charles River Laboratories Japan Inc. (Yokohama, Japan). All mice were bred in a pathogen-free facility at the Asahikawa Medical University. The experimental protocol was approved by the Institutional Animal Care and Use Committee of Asahikawa Medical University (#20,001). C57BL/6 mice were subcutaneously injected with 1 × 10 6 MOC1 cells per mouse. The mice were intraperitoneally treated with 30 mg/kg GEM weekly and/or 200 µg anti-PD-1 Ab (RMP1-14; Bio X cell) three times per week after the tumor diameter reached approximately 7-8 mm. Tumor volume was monitored twice a week by measuring two opposing diameters using an electronic caliper. Tumor volume was calculated using the following formula: tumor volume (mm 3 ) = 0.5 × length (mm) × width 2 (mm). Results are presented as mean tumor volume (mm 3 ) ± standard deviation (SD). Tumors were harvested on day 49 to assess the tumorinfiltrating lymphocytes (TILs). The surface markers of the TILs were assessed using flow cytometry.

Statistical analysis
All results are presented as the mean ± SD. Statistical analysis was performed using Student's t-test and one-or two-way ANOVA. Survival curves were analyzed using the log-rank (Mantel-Cox) test. Significance levels are defined as ns (not significant, p > 0.05), *p < 0.05, **p < 0.01, and ***p < 0.001. P values were calculated using GraphPad Prism 8 (GraphPad Software Inc.).

Expression of Brachyury in HNSCC
To evaluate whether Brachyury is expressed in HNSCC, IHC analysis was performed on tissue samples from 52 patients with oropharyngeal cancer. Brachyury was mainly localized in the nucleus of the tumor cells (Fig. 1a). Staining intensity scores for Brachyury in tumor cells were graded as follows: no staining, weak, moderate, and strong staining. Brachyury overexpression was observed in 67% of the HNSCC cases (Fig. 1b). No correlation was found between brachyury expression and sex, age, tumor stage, smoker, or alcohol consumption (Supplemental Table 1 SCC, nor standardized uptake value measured on fluorodeoxyglucose positron emission tomography (FDG-PET) was related to the Brachyury expression. Although more than 50% of p16-negative patients expressed Brachyury, this protein was significantly expressed in p16-positive patients (82%). We next evaluated the expression of Brachyury protein in HNSCC cell lines (SAS, HSC2, HSC3, HSC4, Sa-3, CA9-22, and HPC 92Y) by western blot analysis. Brachyury was expressed in all HPV-negative (Fig. 1c) and HPVpositive HNSCC tumor cell lines (Supplemental Fig. 1).
These results indicate that HNSCC cell lines might express Brachyury, regardless of HPV status. We next analyzed the correlationship between the expression of Brachyury and the patient survival using the public database. As shown in Fig. 1d, the survival of patients with advanced HNSCC was significantly poor in the group with high Brachyury expression. Collectively, Brachyury could be a promising target for treating advanced HNSCC with poor prognosis.

In vitro generation of Brachyury-specific CD4 helper T cells
The generation of antitumor T cells is a feasible therapeutic approach for targeting transcription factors, including Brachyury. We next investigated whether Brachyury-specific CD4 HTLs could be induced in human PBMCs. According to two computer-based algorithms that predict the ability of a peptide sequence to bind to HLA-DR molecules, two Brachyury-derived peptides (Brachyury 189-203 : ETQFIAVTAYQNEEI, Brachyury 258-272 : ANPHPQFGGALSLPS) were selected. CD4 + T cells isolated from the PBMCs of healthy volunteers were stimulated with autologous DCs pulsed with Brachyury peptides. Brachyury 189-203 -reactive HTLs and Brachyury 258-272 -reactive HTLs were derived from healthy volunteers K (HLA-DR9/53) and Y (HLA-DR4/8/53), respectively. These Brachyury-reactive HTLs produced IFN-γ in response to Brachyury peptides (Fig. 2a). To confirm HLA-DR restriction, Brachyury-reactive HTLs were incubated with peptide-pulsed autologous PBMCs in the presence of anti-HLA-DR mAb. An anti-HLA class I mAb was used as the negative control. As shown in Fig. 2b, the response of HTLs to Brachyury-derived peptides was significantly inhibited by anti-HLA-DR mAb but not by anti-HLA class I mAb, indicating that T cells recognize peptides present on HLA-DR. To clarify which HLA-DR allele is involved in antigen presentation, several L cells expressing an individual HLA-DR molecule were used as APCs to evaluate the reactivity of Brachyury-reactive HTLs. Brachyury 189-203 -reactive HTLs or Brachyury 258-272 -reactive HTLs recognized peptides on HLA-DR53 and HLA-DR4, respectively (Fig. 2c).

Tumor recognition and cytotoxic activity of Brachyury-specific CD4+ helper T cells
We then evaluated the direct response of Brachyuryreactive HTLs to Brachyury-expressing tumor cell lines. As shown in Fig. 3a, Brachyury 189-203 -specific CD4 + T cells recognized HLA-DR53-positive HPC 92Y and HSC4 cells. This recognition was significantly prevented by anti-HLA-DR mAb. Similarly, the Brachyury 258-272 -specific CD4 + cell line reacted with HLA-DR4-positive HPC 92Y and HSC4 cells. These HTLs did not react with brachyurypositive but with HLA-DR-unmatched HSC3 cells. These results indicate that Brachyury-derived helper epitopes were naturally present on tumoral HLA-DRs. In addition to IFN-γ, Brachyury-reactive HTLs produced granzyme B, an essential protease to kill tumors (Fig. 3b). To confirm whether Brachyury-reactive HTLs could directly kill tumors, we evaluated the cytotoxicity of these T cells. Brachyury 189-203 -specific HTLs showed cytotoxicity against HLA-DR-matched but not against HLA-DRunmatched tumor cells (Fig. 3c). Similarly, HLA-DRmatched cells, but not HLA-DR-unmatched cells, were killed by Brachyury 258-272 -specific HTLs, suggesting that these HTLs were directly cytotoxic to HNSCC cells.
Next, we evaluated whether Brachyury peptides could be processed from tumor-derived proteins and presented on HLA-DR in professional APCs. As shown in Fig. 3d, Brachyury-reactive HTLs could react with DCs pulsed with lysates from HLA-DR-unmatched HNSCC cells, indicating that Brachyury-derived helper epitopes were conserved during the exogenous antigen-processing pathway.

Brachyury-reactive T cells in patients with HNSCC
The presence of peptide-specific precursor T cells is crucial for inducing antitumor responses in peptide vaccines.
To confirm the presence of Brachyury-reactive T cells in patients with HNSCC, PBMCs from these patients were stimulated with Brachyury-derived peptides. The clinical characteristics of the eight patients with HNSCC are summarized in Table 1. Interestingly, five out of eight HNSCC patients responded to both Brachyury 189-203 and Brachyury 258-272 peptides, whose responses were blocked by the addition of anti-HLA-DR Ab (Fig. 4). Thus, Brachyury-derived peptides can be considered promising HTL epitopes for the development of peptide vaccines in patients with HNSCC.

GEM augmented Brachyury-reactive T cell responses through the upregulation of MHC expression
Because tumor cells often downregulate the expression of surface MHC molecules to escape from immune cells [34], immune adjuvants that can upregulate the expression of tumoral MHCs are essential for effective peptide vaccines.
To promptly translate the findings in the clinic, we focused on the immunological aspects of chemotherapy. First, we evaluated the effects of cisplatin (CDDP) and radiotherapy on MHC expression, both of which are common therapeutics for the treatment of HNSCC. As shown in Supplemental Fig. 2, CDDP but not radiotherapy upregulated HLA-DR expression. However, CDDP downregulated the expression of HLA class I on tumor. GEM has been used to treat HNSCC, including nasopharyngeal cancer [21][22][23][24]. As GEM modulates antitumor immunity [25,35], we next focused on the activity of GEM in tumoral MHC expression. First, we evaluated the direct cytotoxicity of GEM in HNSCC cells. As shown in Supplemental Fig. 3, GEM killed HNSCC cells with an IC 50 of approximately 40 nM. Interestingly, treatment with a relatively low dose of GEM (25 nM) augmented HLA class I, HLA-DR, and PD-L1 expression levels in HNSCC cells (Fig. 5a, b). In addition, GEM induced immunogenic cell death, as confirmed by the release of HMGB1 and ATP (Fig. 5c, d). Brachyury expression was also upregulated following GEM treatment (Supplemental Fig. 4).
Next, we investigated the reactivity of Brachyury-reactive HTLs with the HNSCC cells treated with GEM. The production of both IFN-γ and Granzyme B from Brachyury-reactive HTLs was higher in GEM-treated HNSCC cells than that in naïve HNSCC cells (Fig. 5e, f). As GEM augmented the expression of PD-L1 in tumors in addition to MHCs, anti-PD-1 Ab further upregulated the Brachyury-reactive T cell responses against HNSCC cells treated with GEM, suggesting that the immune-enhancing activity of GEM might be increased by ICIs.

Synergistic effect of GEM with an immune checkpoint inhibitor in an HNSCC mouse model
To further evaluate the antitumor effect of GEM combined with ICIs, we compared the antitumor activity of GEM with and without PD-1 blockade in a mouse syngeneic HNSCC model (MOC1). First, the expression in murine MHCs after GEM treatment was assessed. Similar to that in humans, GEM enhanced MHC class I, MHC class II, and PD-L1 expression in MOC1 cells (Fig. 6a, b). Tumor-bearing mice were intraperitoneally injected with GEM and/or anti-PD-1 Abs (Fig. 6c). As shown in Fig. 6d, the addition of anti-PD-1 Abs significantly augmented the antitumor activity of GEM in vivo. In accordance with the in vitro experiments (Fig. 6a, b), MHC Class I and Class II expression levels in tumors were upregulated in tumors treated with Fig. 4 The existence of Brachyury-reactive precursor T cells in patients with HNSCC. PBMCs from patients with HNSCC were cocultured with Brachyury peptides for two cycles every week. The T cell response to Brachyury peptide was assessed by measuring IFNγ production in the supernatants using ELISA. Anti-HLA-DR mAb was used to assess HLA restriction in T-cells. The PADRE peptide was used as a positive control. The data are representative of triplicate experiments. Bars and error bars represent mean and SD, respectively. (*p < .05, **p < .01, ***p < .001, Student's t test) GEM in vivo (Fig. 6e). GEM alone increased the infiltration of lymphocytes, which was further enhanced by PD-1 blockade.
To clarify the effect of GEM with ICIs on immune cells, the T cell profiles of the spleen and TILs were examined after three cycles of treatment (day 49). As shown in Figs. 6f, g, effector memory (CD44 + CD62L − ) and central memory (CD44 + CD62L + ) CD4 + T cells were significantly increased by the combination of GEM and ICIs in TILs but not in the spleen. As the percentage of central memory CD8a + T cells was comparable, effector memory (CD44 + CD62L − ) CD8a + T cells were also upregulated by the combination of GEM and ICIs in TILs. These results suggest that dual GEM/ICI treatment increased infiltration of tumor-reactive memory HTLs and might be an ideal adjuvant to combine with the HTL-targeted peptide vaccine.

Discussion
This study reveals the potential of Brachyury as a target antigen for peptide vaccines and the combination of GEM and ICIs as promising immunoadjuvants. To the best of our knowledge, only a few reports have elucidated the immunogenicity of Brachyury and its corresponding epitopes. Brachyury amplification is associated with poor prognosis in various cancers, including lung [36], breast [9], rectal [10], and oral cancers [11]. Brachyury-positive cells express cancer stem-like markers [37] and epithelial-mesenchymal transition [8], which cause tumor metastasis [9,11] and resistance to chemotherapy and radiation therapy [8,38]. As we showed in this study, the expression of Brachyury could be augmented by chemotherapy to confer drug resistance. In addition, the survival of patients with advanced HNSCC was significantly poor in the group with high Brachyury expression. Therefore, the development of novel therapies targeting Brachyury could be beneficial for patients with aggressive or treatment-resistant tumors. Because Brachyury was expressed in the majority of patients with HNSCC and cell lines, Brachyury-targeted immunotherapy can be applied to a large number of patients. Since we previously found that FGFR was overexpressed in HNSCC [39], FGFR pathway would be an upstream of Brachyury expression [40].
HTLs play an important role in the immune response to tumors, either by maintaining CTLs or by exerting direct antitumor effects [41,42]. The identification of epitopes . c-e Experimental schema. C57BL/6 mice were subcutaneously injected with 1 × 10 6 MOC1 cells. The mice were intraperitoneally treated with 30 mg/kg GEM weekly or/and anti-PD-1 Ab thrice per week from day 18 when the tumor size was approximately 7-8 mm. d Tumor growth curves. Control (Red), gemcitabine monotherapy (Green), anti-PD-1 Ab monotherapy (Blue), and combination therapy with GEM and anti-PD-1 Ab (Yellow) (n = 3/group). Bars and error bars represent the mean and SD, respectively (*p < .05, **p < .01, ***p < .001, twoway ANOVA). e Representative hematoxylin and eosin (HE) staining and immunohistochemical (IHC) images of MHC Class I, Class II, and PD-L1 of tumor on day 49. Scale bar = 100 μm. f, g The mice were sacrificed on day 49, and the percentages of T cells in spleens and TILs were evaluated using flow cytometry. The data shown are representative of the triplicate experiments. e The percentages of CD44 + CD62L − CD4 + T cells and CD44 + CD62L + CD4 + T cells in spleens and TILs, f The percentages of CD44 + CD62L − CD8a + T cells and CD44 + CD62L + CD8a + T cells spleens and TILs. Bars and error bars represent the mean and SD, respectively. (*p < .05, **p < .01, ***p < .001, Student's t test) that can bind to MHC Class II is necessary to establish HTL-targeted immunotherapy. In silico sequence analysis suggests that the Brachyury peptide binds to several common HLA-DR alleles (DRB1 * 0101, DRB1 * 0401, DRB1 * 0701, DRB1 * 1101, and DRB1 * 1501), and we clarified that the Brachyury peptides can bind to HLA-DR4 and DR53. Although we could not evaluate all the predicted alleles due to the limited allele availability from healthy donors, the epitopes we found in this study have the potential to be presented on the common HLA-DR alleles. Antitumor responses to immunotherapy require both antigen-responsive CTLs and HTLs, even in tumors that have lost MHC class II molecules [43]. Brachyury 189-203 (ETQ-FIAVTAYQNEEI) contains a potential HLA-A2601 binding epitope, Brachyury 190-198 (TQFIAVTAY). Although peptide-reactive activation of CTL is beyond the scope of this study, Brachyury 190-198 is a novel candidate of CTL peptide vaccine. As peptides containing both HTL and CTL epitopes have shown high efficacy in clinical applications [44], the CTL-activating activity of Brachyury 189-203 should be explored in the future.
Because ICIs are effective against immunologically "hot" tumors characterized by the infiltration of T cells, it is rational to combine ICIs with another approach that can robustly convert "cold" to "hot" tumors. T cells recognize antigens on tumoral MHCs to infiltrate tumor tissues in which MHC expression is often downregulated. In this study, we showed that the increase in MHC expression induced by GEM may have a synergistic effect with ICIs.
The induction of PD-L1 in tumors by GEM, as confirmed by this study and others [45] further rationalizes the combination of GEM and ICIs. The combination of anti-PD-1/ PD-L1 antibody and gemcitabine has been used in several phase 2 studies with acceptable toxicities [46,47]. Because the dose of gemcitabine we used in this study was comparable to the clinical dose (30 mg/kg), GEM would be tolerable to use as an immune adjuvant. Notably, the central memory HTLs were upregulated in TILs from tumors treated with GEM and ICIs, suggesting that this combination might be suitable for combination with HTL-targeted immunotherapy, such as Brachyury-targeted peptide vaccine. Because no potent HTL epitope has been identified in MOC1, further studies are warranted to prove that GEM and ICIs could stimulate tumor-specific HTLs. Only a few studies have examined the immunomodulatory effects of GEM on HNSCC. In other types of cancer, GEM has been suggested to activate T cells [25], increase antigen presentation in DCs [48], and suppress T regulatory cells and bone marrow-derived suppressor cells [49]. In addition, GEM induces damage-associated molecular patterns (DAMPs) in tumors, which promote the migration and activation of antigen-presenting cells [50]. In this study, we showed that GEM induced immunogenic cell death-related DAMPs (ATP and HMGB1) in HNSCC cells. Because MHC expression, Brachyury expression, and DAMPs production were upregulated by GEM, further studies are required to determine which factor is necessary to stimulate antitumor HTLs.

Conclusion
We identified novel helper epitopes from Brachyury that can elicit direct antitumor activity via antigen-specific HTLs. Being expressed in the majority of HNSCC patient samples and cell lines, Brachyury may be a promising target as an HNSCC antigen. Moreover, the expression levels of MHC classes I and II were upregulated by GEM, followed by increased tumor recognition by T cells. The addition of anti-PD-1 Abs further increased antitumor T cell activity by GEM, which was also confirmed in a mouse syngeneic HNSCC model. GEM along with ICIs would be favorable adjuvants to combine with Brachyury-targeted peptide vaccines.