DNA Vaccines Targeting Novel Cancer-Associated Antigens Frequently Expressed in Head and Neck Cancer Enhance the Efficacy of Checkpoint Inhibitor

HPV-independent head and neck squamous cell carcinoma (HNSCC) is a common cancer globally. The overall response rate to anti-PD1 checkpoint inhibitors (CPIs) in HNSCC is ~16%. One major factor influencing the effectiveness of CPI is the level of tumor infiltrating T cells (TILs). Converting TILlow tumors to TILhigh tumors is thus critical to improve clinical outcome. Here we describe a novel DNA vaccines to facilitate the T-cell infiltration and control tumor growth. We evaluated the expression of target antigens and their respective immunogenicity in HNSCC patients. The efficacy of DNA vaccines targeting two novel antigens were evaluated with or without CPI using a syngeneic model. Most HNSCC patients (43/44) co-expressed MAGED4B and FJX1 and their respective tetramer-specific T cells were in the range of 0.06-0.12%. In a preclinical model, antigen-specific T cells were induced by DNA vaccines and increased T cell infiltration into the tumor, but not MDSC or regulatory T cells. The vaccines inhibited tumor growth and improved the outcome alone and upon combination with anti-PD1 and resulted in tumor clearance in approximately 75% of mice. Pre-existence of MAGED4B and FJX1-reactive T cells in HNSCC patients suggests that these widely expressed antigens are highly immunogenic and could be further expanded by vaccination. The DNA vaccines targeting these antigens induced robust T cell responses and with the anti-PD1 antibody conferring excellent tumor control. This opens up an opportunity for combination immunotherapy that might benefit a wider population of HNSCC patients in an antigen-specific manner.

HPV-independent head and neck squamous cell carcinoma (HNSCC) is a common cancer globally. The overall response rate to anti-PD1 checkpoint inhibitors (CPIs) in HNSCC is~16%. One major factor influencing the effectiveness of CPI is the level of tumor infiltrating T cells (TILs). Converting TILlow tumors to TILhigh tumors is thus critical to improve clinical outcome. Here we describe a novel DNA vaccines to facilitate the T-cell infiltration and control tumor growth. We evaluated the expression of target antigens and their respective immunogenicity in HNSCC patients. The efficacy of DNA vaccines targeting two novel antigens were evaluated with or without CPI using a syngeneic model. Most HNSCC patients (43/44) co-expressed MAGED4B and FJX1 and their respective tetramer-specific T cells were in the range of 0.06-0.12%. In a preclinical model, antigen-specific T cells were induced by DNA vaccines and increased T cell infiltration into the tumor, but not MDSC or regulatory T cells. The vaccines inhibited tumor growth and improved the outcome alone and upon combination with anti-PD1 and resulted in tumor clearance in approximately 75% of mice. Pre-existence of MAGED4B and FJX1-reactive T cells in HNSCC patients suggests that these widely expressed antigens are highly immunogenic and could be further expanded by vaccination. The DNA vaccines targeting these antigens induced robust T cell responses and with the anti-PD1

INTRODUCTION
HNSCC is the sixth most common cancer in the world and is more prevalent in South East Asia where it is the second most common malignancy (1). The incident in some cases is linked to HPV but for HPV independent (HPV neg ) subgroup which represents 60-80% of all HNSCC cases, the causes are unclear and often linked to consumption of alcohol, smoking, smokeless tobacco and bethel quid [reviewed in (2)]. The HPV neg HNSCC is often difficult to treat; surgery and radiotherapy are associated with significant morbidity and a relatively static 5-year survival rate of around 50-60% (3).
Immune checkpoint inhibitors prolonged survival of some patients with HNSCC and have now gained the approval for subsets of patients with advanced disease, however the responses have only been observed in 14-25% of patients (4)(5)(6). While the response to checkpoint inhibitors is associated with many factors including expression of PD-L1 (7,8), tumor mutational burden (9) and the interferon gene signature (10), one major factor is the presence of pre-existing tumor-infiltrating T cells (TIL) (11)(12)(13)(14)(15). High TIL levels correlate with significantly longer survival, and many patients who respond to checkpoint inhibitors fall into this category. TILlow tumors, including most HPV neg HNSCC, are commonly categorized as 'immune desert' (absent T cells) or 'immune excluded' (excluded T cells) (13,16,17). This emphasizes the urgent need of novel immunotherapies that can work in these patients. One such intervention is to use cancer vaccines to induce or expand specific T cell responses against cancer.
For HPV positive (HPV pos ) HNSCC, cancer vaccines targeting oncoprotein E6 and E7 from HPV have induced a durable anti-tumor response (18,19). For HPV neg HNSCC such viral antigens are not available. Targeting mutated antigens (neoantigens) is an emerging strategy and has demonstrated clinical benefits in late-stage melanoma and newly diagnosed glioblastoma in combination with checkpoint inhibitors (20)(21)(22)(23). A mutanome-targeting DNA vaccine has entered a phase 1/2 trial for patients with locally advanced or metastatic solid tumors who received checkpoint inhibitor treatment but did not reach a complete response (24). However, the mutanome-targeting approach is time consuming and costly as it requires the generation of personalized vaccines for each patient. We therefore focused on developing cancer vaccines targeting tumor associated antigens (TAAs) in order to generate off-theshelf treatment to a large proportion of HNSCC patients. Recently the feasibility of this approach has been demonstrated in patients with unresectable melanoma post-anti-PD1 treatment. Antigen-specific polyfunctional T-cells were induced in most patients, and some patients had a partial response or stable disease (25).
We have previously shown that MAGED4B and FJX1 are over-expressed in a small cohort of primary HNSCC patients in Malaysia (26,27). Here we sought to develop a feasible therapeutic strategy by targeting these antigens in HNSCC globally. To achieve this, we first investigated their expression pattern and immunogenicity in HNSCC patient cohorts from Malaysia and the UK. We then designed DNA vaccines targeting MAGED4B and FJX1. We demonstrated that our novel DNA vaccine monotherapy delayed tumor growth in a mouse model expressing the target antigens and increased T cell infiltration into the tumor bed. Promisingly, our vaccine acted synergistically when co-administered with anti-PD1 treatment resulting in tumor clearance in a large proportion of mice.

Patient Sample
Samples from two cohorts of patients were processed at two separate facilities following surgery. For the Malaysian cohort, blood samples from 28 HNSCC patients were collected from the University Malaya Medical Centre, Tengku Ampuan Rahimah Hospital and Hospital Kuala Lumpur. Since HPV positive HNSCC is rare in Malaysia (28), these patients were not tested for HPV. For the UK cohort, blood samples from 21 HPV neg HNSCC patients were collected from patients treated at Poole Hospital at the time of surgery. The samples were both primary and recurrent tumors. The demographic and clinico-pathological information of 49 HNSCC patients from Malaysia and UK are presented in Table 1. Peripheral blood mononuclear cells (PBMC) were isolated from whole blood. HLA-A status for the Malaysian cohort was determined by direct staining of PBMC with mouse anti-human HLA-A2-PE and HLA-ABC-PE (BD Biosciences, US). HLA status for the UK cohort was determined using the AllSet Gold HLA-A SSP Kit (One Lambda, US). Formalin-fixed paraffin-embedded (FFPE) tissues from HNSCC patients that are in excess of diagnosis were identified and sectioned into 4µm thick for hematoxylin and eosin (H&E), as well as immunohistochemistry (IHC) staining. For tissues from normal organs, 3-4 independent tissue microarrays generated in-house were analyzed for the expression of MAGED4B and FJX1.

Peptides
MAGED4B and FJX1 overlapping peptide pools (OPP) consisting of 183 and 107 15-mers peptides with 11 amino acids overlap respectively. CEFT peptide pool consists of MHC class I and II peptides from cytomegalovirus (CMV), Epstein-Barr Virus (EBV), Influenza A and Clostridium tetani was used as positive control in the in vitro stimulation (IVS) ELISpot assay. All peptides were produced by JPT Peptide Technologies (Germany) with >70% purity by high-performance liquid chromatography (HPLC).

In Vitro Stimulation of T Cells
PBMCs were thawed and rested at high density (1x10 7 cells per well of a 24-well plate) for two days before T cells were isolated using CD4+ and CD8+ microbeads (Miltenyi, UK). The remaining T-cell depleted PBMCs were transfected with mRNA encoding individual antigens. DNA fragments encoding each antigen (MAGED4B, FJX1 or CEFT) were synthesised through the Invitrogen GeneArt service (ThermoFisher, UK) and inserted into the pcDNA3. The DNA sequence synthesised for each antigen consisted of the antigen sequence, flanked at the N-terminus with an Ig signalling peptide and at the C terminus with an MHC-I trafficking signal (MITD) (29). mRNA for each antigen was produced using the HiScribe T7 ARCA mRNA kit with tailing from linearized plasmids (New England Biolabs, UK). Five micrograms of mRNA encoding CEFT or 10µg of MAGED4B or FJX1 mRNA was used to transfect T-cell depleted PBMC using the Human Dendritic Cell Nucleofector kit (Lonza, UK) to serve as antigenpresenting cells (APC). Purified T cells were incubated with transfected and irradiated (15 Gy) APC (ratio1:1) in OpTmizer ™ medium (Thermo, UK) supplemented with 20 U/ml IL-2 and 5 ng/ml IL-15 (both PeproTech, US) at 37°C for 14 days. Half of the medium was changed to fresh OpTmizer medium supplemented with 20 U/ml IL-2 every 2-3 days. Stimulated T cells were subjected to ELISpot.
Human IFNg ELISpot Assay 1 x 10 4 stimulated T cells were incubated with peptide-loaded dendritic cells (DCs) at 1:30 ratio on a 96 well ELISpot plates precoated with anti-human IFNg antibody (mAb 1-DIK, Mabtech, UK) overnight at 37°C. DC were differentiated from autologous CD14+ monocytes, according to Miltenyi manufacturer's protocol in ImmunoCult ™ DC differentiation media (Stemcell Technologies, UK) for two weeks. DCs were loaded with 1 µg/ml  Figure S1A. All stained cells were analyzed using LSR Fortessa flow cytometer or FACSCanto (both BD Biosciences, US) and gated against fluorescence minus one control (FMO) or unstained controls. The flow panels were designed to accommodate different lasers and FACS machine capacities in both institutions. Details of all antibodies used in the flow cytometry staining are provided in Table S1. Analysis was performed using FlowJo software (BD Biosciences, US).

Generation of pDom Fusion DNA Vaccines
The human wild type sequences of MAGED4B and FJX1 were cloned separately into the pcDNA3.1 (Invitrogen, UK) plasmid as a fusion with the Dom sequence from fragment C of tetanus toxin as source of CD4+ T cell help (30). The sequence encoding mus IgH signal peptide MGWSCIIFFLVATATGVHS was inserted at the N-terminus of each construct to enhance secretion. DOM fragment and the antigen of interest MAGED4B or FJX1 were separated with a seven amino acid linker (AAAGPGP).
DNA sequencing was performed to confirm that the plasmids contained the correct DNA construct. The generated plasmid DNA fusions referred as pDom-MAGED4B (pDom-M) and pDom-FJX1 (pDom-F) (plasmid constructs in Figure S2 pDom plasmid without the insertion of antigens was used as vector control. All plasmids were propagated in DH-5a and purified using QIAGEN Plasmid Plus Giga Kit (QIAGEN, US), according to manufacturer's instruction.

Animal Study
All procedures involving the use of animals were reviewed and approved by the Animal Ethics Committee of Universiti Immunogenicity of pDom-M/F In Vivo C57BL/6 and HLA-A2 tg mice (transgenic for the HLA-A2.1 allele, HHD) (31, 32) used in the immunogenicity study were bred in animal facility of the University Southampton. C57BL/6 or HLA-A2 tg mice aged 6 -10 weeks (n=5/group) received either 50mg pDom-M, pDom-F or pDom vector control vaccines intramuscularly (i.m.) into each quadriceps muscles of the hind legs. C57BL/6 mice received a booster injection with an equivalent amount of the vaccine after one week, while HLA-A2 tg mice received a booster injection after 3 weeks with electroporation (EP). EP was carried out on mice anaesthetized by isofluorane, using a custom-made pulse generator from Inovio Pharmaceuticals, as described previously (33).
At endpoint, spleens were harvested from experimental mice, and subsequently mashed though 70 µm strainer to obtain a single cell suspension. Lymphocytes were then isolated from splenocytes using Lymphoprep ™ (STEMCELL, UK) following manufacturer's protocol. To determine the presence of IFNg secreting lymphocytes, anti-mouse IFNg ELISpot kit (BD Bioscience, UK) was used according to manufacturer's protocol. Mouse lymphocytes (2.5x10 5 cells/well) were incubated in complete RPMI media with no peptide (background control), p30 control peptide (1µM), MAGED4B or FJX1 OPP (1µM each peptide) on pre-coated anti-mouse IFNg ELISpot plates for 40h at 37°C. Samples were plated in triplicate. Spots were detected with a biotin-conjugated mouse IFNg antibody (BD Bioscience, UK) followed by incubation with Streptavidin-ALP (Mabtech) and BCIP/NBT (Thermo, UK). Plates were scanned using ImmunoSpots reader (AID, UK).
The mean values were represented as spot-forming cells (SFCs) per 10 6 cells. Levels were considered positive if at least two times above the background control.

Tumour Models
The mouse B16F10 cell line expressing the human HLA-A2 gene (B16F10-HLA-A2) was provided by Professor Eric Tartour (Sorbonne Paris Cite, University). B16 is one of the most frequently used model for evaluation of the efficacy of anti-PD1 antibody and its combinations hence it has been chosen for the currently study in favor of less well characterized HNSCC models which has recently emerged (34). It was cultured in RPMI 1640 (Gibco, UK) supplemented with 10% heat inactivated-fetal bovine serum (FBS, Gibco UK), penicillin/ streptomycin (100 U/ml) and 1 mg/ml G418 (EMD Millipore, US) at 37°C in a 5% CO 2 humidified atmosphere. We generated two different tumor models expressing our target antigens; B16F10-HLA-A2-MAGED4B (BAM) and B16F10-HLA-A2-FJX1 (BAF). The expression of HLA-A2 in BAM and BAF cell lines was confirmed by flow cytometry using human HLA-A2-PE-conjugated antibody (clone BB7.2, BD Biosciences, US).
MAGED4B and FJX1 expression levels were confirmed by western blotting using custom made anti-MAGED4B (Dundee Cell Products Ltd, UK) and anti-FJX1 (HPA059220, Sigma Aldrich, US) antibodies respectively.

Efficacy of pDom-M/F In Vivo
AAD mice (transgenic for HLA-A2.1/H2-Dd allele) used in the efficacy study were bred in the animal facility of National University Malaysia. AAD mice (6-10 weeks) were inoculated with 1x10 6 BAM cell line (n=27 mice) at the right flank at day 0. These mice were randomized to receive vaccination of 100 µg pDom-M/F DNA vaccines (50 µg of each vaccine) or 50 µg pDom DNA vaccine i.m into each quadriceps muscles of the hind legs on day 5 after palpable tumors were observed (~25cm 3 ). A booster vaccination at the same dose was administered at day 12 (the vaccination schedule is depicted in Fig. 3A). Tumour sizes were evaluated every 3-4 days and volumes were calculated using the formula: volume = ½ (length X width 2 ). Tumour growth inhibition (TGI) was calculated using the formula: TGI=(Vc-Vt)/Vc X 100%; where Vc and Vt are the average tumor volume of control and treatment group respectively at endpoint. On day 26, mice were sacrificed and tumors were harvested for analysis of T cell infiltration by IHC and flow cytometry as described below.
To evaluate the efficacy of combination treatment of pDom-M/F vaccine with anti-PD-1 antibody, AAD mice were inoculated with 1x10 5 BAM (n=32 mice) or BAF cells (n=39 mice). Subsequently, mice were randomized into four treatment groups; pDom-M/F, anti-PD-1 antibody, pDom-M/F + anti-PD-1 antibody, or pDom + IgG isotype control. pDom-M/F vaccinations were given at day 5 and day 26 post-cell inoculation. One hundred microgram of anti-PD-1 monoclonal antibody (CD279, Bioxcell, US) or rat IgG2a isotype control (Bioxcell, US) was given intraperitoneally every 3 days from day 3 as indicated in Figure 4A. Tumour sizes were measured every 3-4 days and tumor volume were calculated using the formula described above.

Flow Cytometry Analysis of Mouse Tumour Infiltrating Lymphocytes
To determine the expression of T cell markers after pDom-M/F treatment in mouse models, tumor infiltrating lymphocytes (TILs) were isolated from mouse tumor samples. Briefly, 500mm 3 tumors were harvested and minced into smaller pieces (<3mm) and digested in 5 ml digestion buffer [RPMI 1640 containing collagenase type I (200 U/ml; Gibco, US) and DNase I (10 U/ml; Invitrogen, US)] and incubated at 37°C for 20 min with agitation. After incubation, tumor suspension was pressed through a 40 µm cell strainer, and rinsed with cold MACS buffer (0.5% BSA+2mM EDTA in PBS). Cells were spun at 500g for 5 min and finally resuspended in PBS with purified rat anti-mouse CD16/CD32 (mouse Fc blocker) (BD Pharmingen US), and proceed to stain with viability dye (FVS780), anti-CD4, CD8 and PD-1 antibodies (all from BD Biosciences, US) for 30 min at 4°C. To determine the presence of MDSC and T-regs, additional tubes were stained with viability dye (FVS780), anti-CD4, CD45.2, CD11b, and GR1. Cells were subjected to fixation and permeabilization with Mouse FoxP3 Buffer Set (BD Biosciences, US), and stained with anti-FOXP3 (eBioscience, US). Details of all antibodies used in the flow cytometry staining is provided in Table S1. PD1-positive CD4+ or CD8+ T cells were quantitated based on the relative expression of PD1, gating strategy is detailed in Figure S3.

Immunohistochemistry
To assess the expression level of MAGED4B and FJX1 in HNSCC samples from the UK cohort, anti-MAGED4B polyclonal antibody (1:200, Novus-Bio) and anti-FJX1 polyclonal antibody (1:200, Novus-Bio) were used. Deparaffinization, rehydration, antigen retrieval, and IHC staining were performed using a Dako PT Link Autostainer using EnVision FLEX Target Retrieval Solution, High pH (Agilent Dako, UK) and DAKO Auto-stainer Link48 ™ in the Cellular Pathology Department of the University Hospital of Southampton NHS Trust. Images were captured using ZEISS Axio scanner.
For the Malaysian cohort, IHC was performed on patient FFPE samples using anti-MAGED4B polyclonal antibody (1:100, Sigma Aldrich, US) and anti-FJX1 polyclonal antibody (1:200, Sigma Aldrich, US). Antigen retrieval was performed in citrate buffer pH 6 and Tris-EDTA pH 9 for MAGED4B and FJX1 respectively using microwave heating method. IHC staining were performed using Dako Cytomation Envision+ Dual Link System-HRP (DAB+) kit (Dako,US) following protocol recommended by manufacturer. Direct comparisons of antibodies to each antigen used for the UK and the Malaysian cohorts were performed using 5 independent HNSCC samples and the same data were obtained. The Novus antibodies were chosen in Southampton because they generated less background with the automated system.
For mouse IHC staining, FFPE sections were stained with rabbit anti-mouse CD8a (1:400; clone D4W2Z; Cell Signaling Technologies, US). All sections from mice were processed after antibody staining using Dako Animal Research Kit (Dako, US) according to the recommendation by manufacturer (27). The field containing the highest density of CD8-positive cells within the tissue were identified. CD8-positive cells were counted by 3 individuals including a board-certified pathologist and graded as "less than 5 cells", "5 to 10 cells" and "more than 10 cells" (35,36).

Statistics
Data are shown as the mean ± SD or median ± interquartile as described in the figure legends. Statistical differences in the expression of MAGED4B and FJX1 mRNA levels between tumor and normal samples in TCGA were determined using Kruskal Wallis test. ELISpot and FACS results analysis were performed using the Mann-Whitney test for non-parametric data. Tumour volume data was analyzed using the two-way repeated measures ANOVA by time and treatment using GraphPad PRISM as indicated in the figure legends.

MAGED4B and FJX1 Are Expressed in HNSCC Tumour Samples
Using TCGA-HNSC project data, we demonstrated that both MAGED4B and FJX1 gene expression levels in the HPV negative (n=415) and HPV positive (n=72) tumour tissue samples were significantly higher than levels in adjacent normal tissues (n=44, Figure 1A). HPV status of TCGA samples was available from cBioportal for the PanCancer and was the result of complete analysis of the cohort for HPV transcripts (37). Importantly, either one or both antigens were also found expressed in other cancer types including lung, breast, oesophageal, and stomach cancer at the mRNA level ( Figure S4). We subsequently examined the protein expression of MAGED4B and FJX1 in HNSCC tumour tissues from both Malaysian and UK cohorts by IHC. All tested samples from Malaysian cohort (n=28) expressed both antigens, except one patient who did not express FJX1 ( Table 1). All samples from the UK cohort (n=16) expressed both MAGED4B and FJX1 ( Table 1). In total 43/44 samples were positive for both antigens (representative images are depicted in Figure 1B; testes used as positive control). The evaluation of expression of MAGED4B and FJX1 in five major organs and healthy oral epithelia revealed very low levels in basal keratinocytes of stratified squamous epithelium with no expression above the background elsewhere ( Figure S5).

MAGED4B and FJX1 Specific T Cells Were Detected in HNSCC Patients at High Frequency
T cells specific for neo-antigens were detected in patients with HPV neg HNSCC (38), indicating that the patients' immune system is capable of recognising the tumour antigens in this disease. To determine whether HNSCC patients inherently harbour MAGED4B-and FJX1-specific T cells, we generated MAGED4B 501-509 /HLA-A2 tetramer using the HLA-A2 epitope we previously identified (39). MAGED4B-specific CD8+T cells were identified in 4/4 HLA-A2 pos patients from the Malaysian cohort and 8/11 HLA-A2 pos patients from the UK cohort ( Figure 1C and Table 2) at a frequency observed at a similar level to neoepitopes in melanoma (range 0.06-0.12%) (40). PBMC samples of HLA-A2 neg patients were used to confirm the specificity ( Figure 1C, Figure S1B and Table 2). Since no FJX1 epitope was available for a tetramer, the T cell responses were analysed by IFNg ELISpot using OPP for the entire antigenic sequence. In parallel we also analysed the responses to MAGED4B OPP to extend the data beyond the HLA-A2 positive cohort. We initially performed ex vivo ELISpot using PBMCs from two patients ( Figure S1C), which suggested specific T cell responses could be captured without in vitro restimulation (IVS) occasionally but not reliably. We therefore opted for IVS ELISpot; 3/7 and 5/7 patients responded to FJX1 OPP or MAGED4B OPP respectively with 3/7 demonstrating responses against both ( Figure 1D; representative examples, Table 2). Collectively, we have confirmed both antigens are immunogenic in HPV neg HNSCC patients.

DNA Vaccines Targeting MAGED4B and FJX1 Induce T Cell Responses in Mice
To induce broad CD4/CD8+ anti-tumor responses irrespective of HLA subtypes, the full-length MAGED4B or FJX1 sequences were linked to 3' Dom sequence and inserted into pcDNA3 plasmid to give rise to pDom-MAGED4B (pDom-M, Figure S2A) and pDom-FJX1 (pDom-F, Figure S2B). In vivo immunogenicity was tested using the HLA-A2 transgenic (tg) and wildtype C57BL/6J mice using our previous protocols (31). Splenocytes from immunized mice were evaluated by IFNg ELISpot using MAGED4B or FJX1 OPP to stimulate antigenspecific immune responses. In wildtype mice, pDom-M showed significantly higher antigen specific T cell responses to MAGED4B OPP than control pDom immunized mice (p=0.012, Figure 2A). Similarly, T cell response against FJX1 OPP was also significantly elevated in mice received pDom-F as compared to the pDom control group where no responses were detected (p=0.012). In HLA-A2 tg mice, vaccination with pDom-M and pDom-F resulted in a significant T cell response against MAGED4B (p=0.016, Figure 2B) or FJX1 (p=0.037, Figure 2B) OPP respectively as compared to the pDom control. Both vaccines induced robust T-cell responses with lower levels observed in HLA-A2 tg mice because of lower number of T-cells and lower MHC I expression (41). The responses to pDom vector control were negative in both strains as expected.

Vaccination With pDom-M/F Vaccine Increased T Cell Infiltration and Delayed Tumour Growth
We subsequently evaluated efficacy of these two vaccines using the BAM tumor model. This model expresses HLA-A2 (42) ( Figure S6A). It expresses FJX1 endogenously and was engineered to express MAGED4B because the antigen is not expressed in mice ( Figure S6B). The HLA-A2 mice were challenged with 10 6 BAM tumor cells and once the tumors were palpable, mice were immunized with either combined pDom-M and pDom-F (referred as pDom-M/F vaccine) or pDom control as indicated ( Figure 3A). Mice immunized with pDom-M/F showed a delay in their tumor growth as compared to mice received pDom (average tumor sizes of pDom and pDom-M/F at endpoint (day 26): 1641.22 mm 3 and 887.24 mm 3 , TGI: 45.94%; Figure 3B).
Subsequently, we determined whether vaccination could increase infiltration of T cells into the tumor, as demonstrated in a DNA vaccine clinical trial targeting E6/E7 in HPV pos HNSCC (18). Tumours from the pDom-treated group predominantly displayed an immune deserted phenotype, where immune cells were absent or scarce throughout the tissue ( Figure 3C). This phenotype was expected as the parental tumor were previously reported to have low TILs (43). By contrast, tumors from mice vaccinated with pDom-M/F showed increased levels of T cells infiltrated into the tumor ( Figure 3C). Enumeration of CD8+ T cells in the tumors further indicated that the infiltration of CD8+ T cells was higher in the pDom-M/F-vaccinated tumors as compared to pDomvaccinated tumors ( Figure 3D). As tumor-specific CD4 effector T cells can also contribute to anti-tumor protective mechanisms (44), we further determined whether they also increased. Encouragingly, pDom-M/F vaccine also induced CD4+ T cells in TILs as demonstrated by flow cytometry analysis but not MDSC (CD45.2+ CD11b+GR1+) and regulatory T cells (CD4+ FoxP3+) ( Figure 3E). In addition, we detected a significant increase in PD-1 expressing CD4+ and CD8+ T cell population in the pDom-M/F group as compared to  pDom (p=0.018 and p=0.029 respectively; Figure 3F). The increase in PD-1 expression level amongst CD4+ and CD8+ T cells indicated a combination with an anti-PD-1 antibody would likely to boost the T cell response.
Further, there was no significant difference in body weight between pDom and pDom-M/F treated animals and histopathology analysis of 5 major organs (kidney, lung, heart, spleen and liver) revealed no pDOM-M/F induced toxicity (data not shown).

Combination Therapy of pDom-M/F and Anti-PD-1 Inhibited Tumour Growth
We next investigated if vaccination with pDom-M/F in combination with anti-PD1 antibody would lead to an improved tumor control. Anti-PD1 group served as a comparator while isotype control IgG plus pDom served as a negative control. When compared to IgG + pDom control group, growth inhibition was clearly demonstrated in pDom-M/F, anti-PD-1 and the combined treatment group (at day 35,  throughout the experiment. We therefore were able to confirm that pDom-M/F DNA vaccines were able to significantly reduce the tumor burden and this was enhanced by combination with anti-PD1 antibody achieving a complete clearance of tumor in 50%-75% of mice in BAF and BAM model respectively.

DISCUSSION
Therapeutic vaccine targeting tumor antigens is a promising strategy to activate the immune system to eradicate cancer. This strategy confers advantage over other non-specific therapies, as it does not only induce tumor-specific immune responses, but also promotes establishment of immunological memory. Early trials in HNSCC had focused on targeting cancer driving mutations such as p53 and K-Ras, however no significant clinical outcomes were reported (45). Vaccines targeting MUC1 (NCT02544880) and CEA (NCT00924092; NCT00027534) which are in phase I/II clinical trials have yet to report results while the phase I study targeting Survivin-2B demonstrated low efficacy (46). Several recent trials in solid tumors demonstrated significant immunogenicity (20)(21)(22)25) with achieving partial or clinical remission. Remarkably, the antigens which have been targeted are not only mutated antigens for which the central tolerance is not expected but widely expressed antigens, tissue-specific antigens as well as cancer testis antigens (CTA) (25,47,48). The latter has long been thought of as "good" antigens due to their safety and immunogenicity; targeting NY-ESO1 and MAGE-A3 recently have demonstrated clinical benefits (25). However, most well-characterized CTAs are not expressed in a large proportion of head and neck cancer (49). Previously, the expression of MAGED4B and FJX1 CTAs has been described in HNSCC patients in Malaysia (26,50,51). Here we were able to confirm their expression at the RNA levels using 522 HNSCC samples deposited at TCGA. Through the collaborative effort of our Malaysian and UK teams, these findings have been further validated in two independent cohorts of patients in Malaysia (n=28) and the UK (n=16), confirming 43/44 patients are coexpressed FJX1 and MAGED4B in both primary and recurrent tumors. This is a remarkably high frequency not observed for other CTAs in HNSCC [reviewed in (52)]; our data also demonstrate consistency between several ethnic groups from the UK and Malaysia. Alcohol consumption and smoking are common etiological factors for HNSCC in both Malaysia and UK, while betel quid chewing is specific to the Malaysia cohort (53,54). Since the causes of the disease are different, the discovery of antigens that are expressed in over 90% of patients is unexpected and confers a rare chance for the development of 'off-the shelf' cancer vaccine applicable worldwide. These antigens are relatively unexplored and therefore a very few defined epitopes are available (39,50). For MAGED4B, we were able to detect the tetramer specific response in HLA-A2 patients (approx. 80%) with the levels that are similar to those reported for mutated antigens reassuring that the antigen is remarkably immunogenic in this patient cohort. We were not able to generate a tetramer using the only HLA-A2 epitope (11mer) from FJX1 described previously (50).
We probed the immunogenicity of FJX1 using IVS ELISpot designed to expand memory T responses in PBMCs. Three out of eight patients generated responses to FJX1, confirming preexisting memory T cell responses to FJX1. Notably, the same patients also demonstrated responses to MAGED4B.
T cell infiltration into tumor is an important criterion for a successful immunotherapy and is linked to better patient prognosis (13,14). Several approaches including targeted therapy, radiotherapy and chemotherapy have also shown the ability to convert immunologically cold tumors into hot to increase response to checkpoint inhibitors (55)(56)(57), however these approaches are associated with treatment-associated side effects. We demonstrated in this study that the pDom-M/F vaccination were able to inhibit the growth of MAGED4B and FJX1-expressing tumors in HLA-A2 transgenic mice by approximately 50%. Importantly, pDom-M/F vaccination were able to convert cold B16 tumors into hot tumors with an increase of CD8+ T cells. This is encouraging as an effective vaccination in cancer therapy is frequently associated with high degree of cytotoxic T cells infiltration into the tumor. We were not able to determine if these are antigen specific CD8+ T cells as our HLA-A2 tetramer or the corresponding peptide was not presented in the HLA-A2 transgenic mice. The work is currently underway to define T-cell epitopes to allow further characterization of antigen-specific responses in murine models.
High expression of PD-1 on TILs has been shown to impair the anti-tumour immune responses in humans by engaging the PD-L1 and to inhibit TCR-mediated proliferation and cytokine production (58). The upregulation of PD-1 levels in vaccinated mice supports our strategy to combine the pDom-M/F vaccine with an anti-PD-1 antibody and this can potentially be applied in the clinic as anti-PD-1 antibodies have now been approved for the treatment of recurrent/metastatic HNSCC. Promisingly, we observed a complete tumour elimination or static tumour growth in the majority of mice that received both vaccine and anti-PD-1 antibody treatment. Our results are one of the few TAA-based DNA vaccines that show near complete tumour clearance in preclinical models when combined with the anti-PD-1 antibody. Overall, our study provides novel findings in which a DNA vaccine targeting TAAs frequently expressed in HNSCC is able to enhance the efficacy of anti-PD-1 therapy in preclinical settings. Clinical trials using selected epitopes including our own has focused on patients' cohort expressing a particular HLA allele (most frequently A2 and A24) (47,48,59). Although we tested our vaccines in the HLA-A2 model, for clinical translation a full-length antigen vaccine is preferable to provide a population-wide coverage irrespective of HLA genotype (31).
This vaccine was designed with HPV neg patients in mind, HPV pos HNSCC patients may also benefit from this vaccine as the antigens are also expressed in these cancers. Both antigens are additionally co-expressed in non-small cell lung cancer ( Figure S4), suggesting relevance of our approach for this common cancer. A phase I/II clinical trial testing our DNA vaccine delivered as doggybone DNA vaccine (33) in combination with anti-PD-1 antibody is due to begin recruitment of patients with recurrent HNSCC in the first half of 2022.