Human papillomavirus subtype 16 (HPV16) is the primary cause of an increasing number of head and neck squamous cell carcinomas (HNSCC), providing strong rationale for T-cell immune therapies against HPV+ HNSCC. Here we assess immunogenicity of HPV16-specific CD8+ T cells (CTL) and characterize HPV-specific mechanisms of T-cell dysfunction. We identified 16 strong and 29 moderately immunogenic CTL-epitopes from HPV16 E2, E6, and E7 antigens restricted by 12 common HLA class I alleles. E2-specific CTL-reactivity was higher in patients with HPV+ HNSCC than in healthy controls (>3-fold; P = 0.026). Patient-derived E2, E6, and E7 peripheral CTLs exhibited heterogeneity in dysfunctional phenotypes. Immunogenomic analyses of 119 HNSCC transcriptomes revealed high T-cell infiltration and dysfunction in HPV+ HNSCC and correlation of HPV antigen expression with T-cell exhaustion gene signatures. Indoleamine 2,3-dioxygenase (IDO-1) was strongly expressed in HPV+ HNSCC versus HPV HNSCC (P = 0.001) and correlated with E7 expression (R2 = 0.84; P = 0.033). Combination treatment with PD-1 blockade and IDO-1 inhibition overcame profound CTL-dysfunction, enhancing HPV+ HNSCC sensitivity to CTL-cytotoxicity in vitro (up to 10-fold in E7-CTLs, P = 0.011). Our findings implicate mechanisms of T-cell escape in HPV+ HNSCC, wherein high tumoral HPV-antigen load results in high expression of immune dysfunction genes on tumor cells (e.g., IDO-1), and dysfunction of HPV-specific CTLs (e.g., E7, E2-CTLs). The HPV16 CTL-epitopes identified in this study, in combination with blockade of HPV+ HNSCC-specific PD-1/IDO-1 checkpoints, may be useful for targeted immunotherapy.

Significance: This study evaluates the HPV antigen T-cell immunogenicity role of inhibitory receptors and other exhaustion markers in the cytotoxic function of HPV antigen-specific CTLs and identifies combined inhibition of PD-1/IDO-1 as a strategy to enhance CTL targeting of HPV+ HNSCC. Cancer Res; 78(21); 6159–70. ©2018 AACR.

Head and neck squamous cell carcinoma (HNSCC) is the sixth most common cancer worldwide with close to 600,000 cases diagnosed annually (1). A subset of HNSCCs in the oropharynx are caused by the human papillomavirus (HPV), (HPV+ HNSCCs; ref. 2), which are molecularly and clinically distinct from non-HPV-associated HNSCCs (HPV HNSCCs; ref. 3). In the United States, 70% to 80% of HPV+ HNSCCs are caused by the oncogenic HPV type 16 (HPV16; ref. 4). Incidence of HPV+ HNSCC increased 225% from 1984 to 2004 and has now surpassed the incidence of cervical cancer (5). Although HPV vaccines effectively prevent HPV-related cancers, the impact of vaccination on HNSCC incidence may not occur until 2060, likely due in part to slow vaccine uptake, and the decades between infection and clinical HPV+ HNSCC diagnosis (6). As a result, over 600,000 cases are predicted in the interim, providing a strong rationale for the development of novel therapeutic strategies against HPV+ HNSCC.

The recent development of clinically effective tumor immunotherapies, such as checkpoint blockade (CKB) using PD-1/PD-L1 inhibitors (7, 8), has led to FDA approval of nivolumab and pembrolizumab for HNSCC (9, 10). It is now established that the clinical response to CKB is correlated with tumor neo-epitope load. Tumor-specific neo-epitopes have been directly targeted using therapeutic vaccines and/or adoptive T-cell therapy (ACT), and have been shown to enhance cytotoxic T-cell targeting of multiple solid tumors (11–13). Thus, there is a renewed interest in defining the HLA-restricted antigenic repertoire of tumor-infiltrating lymphocytes (TIL) to develop targeted therapeutic vaccines (14), to identify T-cell receptors (TCR) for ACT (11), and to monitor clinical responses to complex cancer immunotherapies (15, 16).

HPV-associated cancers express multiple viral antigens. HPV integration into host genome in cervical cancer results in derepression of the oncogenic drivers E6 and E7 (17). Immune therapies targeting E6 and E7 have thus been developed, including peptides, DNA, and ACT therapies (14, 18, 19). However, in comparison to cervical cancer, HPV+ HNSCCs have both lower rates of genome integration, and less interruption of the viral transcriptional regulatory gene E2 (2, 20, 21). Thus, subsets of HPV+ HNSCCs also express E2 in addition to E6 and E7 (2, 22).

We and others have previously shown that high titers of serum antibodies against HPV16-E2, E6, and E7 are detectable in most patients with HPV+ HNSCC, indicating immunogenicity and persistence of these antigens (23, 24). We thus hypothesized that patients with HPV+ HNSCC would have preexisting HPV-specific CD8+ T cells (CTL), and that HPV-antigen expression levels would influence CTL-dysregulation in tumor microenvironment. We identified the T-cell antigenic landscape of globally frequent HLA class I alleles from HPV16 E2, E6, and E7. By phenotyping HPV-specific CTLs from patients with HPV+ HNSCC, and analyzing the immune transcriptomes of 119 HNSCCs, we demonstrate intratumoral and peripheral CTL-dysfunction in HPV+ HNSCC. We show that this CTL-dysfunction can be reversed using targeted HPV-specific T-cell expansion, and synergistic inhibition of IDO-1 and PD-1. These results have implications for the development of effective T-cell therapies for HPV+ HNSCC.

HPV16 candidate CTL-epitope prediction

HPV16-candidate CTL epitopes were predicted using previously described prediction strategies developed by us (25, 26), except for the incorporation of immunogenicity scores. For the 15 HLA-class I alleles, 9-mer and 10-mer candidate epitopes derived from the HPV16 proteins E2, E6, and E7 were predicted from five independent prediction algorithms and normalized. Top four to five candidate peptides/HLA-allele were used for in vitro experiments. Further details are provided in Supplementary Materials and Methods. Predicted candidate HPV16-peptides, individual normalized, and total binding percentile scores are listed in Supplementary Table S1. For full list of candidate peptides, see Supplementary Table S2.

Epitope mapping from HPV+ HNSCC peripheral blood mononuclear cells

Peripheral blood mononuclear cells (PBMC) were obtained from stage III or stage IV patients with HPV+ HNSCC (MSSM cohort). Informed consent was obtained from all patients; the studies were conducted in accordance to the U.S. common rule and was approved by MSSM and ASU institutional review board. Patient characteristics are described in ref. 27. All HPV16-peptides (>80% purity) were synthesized by Proimmune, UK. HPV+ HNSCC PBMCs were thawed, rested with 1 μg/mL of CKB antibodies anti-PD1 (eBioscience), anti-CTLA4 (eBioscience) for 1 hour at 37°C. HPV16-peptides in pool or individually were added subsequently in biological triplicates, along with recombinant human IL2 (20 U/mL), human IL7 (5 ng/mL). On day 5, half the media was removed and replaced with fresh IL2 and peptide pool. On day 8, half the media was removed and fresh media, IL2, and peptide was added to the cells and replated into a 96-well multiscreen elispot plate for Elispot detection. Elispot detection is described in Supplementary Materials and Methods.

HPV-CTL stimulation for phenotyping

HPV-specific T cells were generated by stimulating autologous HPV+ HNSCC patient B-cell antigen-presenting cells (APC; Supplementary Materials and Methods). APCs were either peptide pulsed with HPV16-epitopes, or transfected with whole HPV-antigen encoded in mammalian expression plasmid pCDNA3.2 (Invitrogen). APCs were washed and incubated with thawed whole HPV+ HNSCC PBMCs at a ratio of 1:2 (200,000 APCs: 400,000 PBMCs) supplemented with 20 U/mL recombinant human IL2 (R&D Systems), 5 ng/mL IL7 (R&D Systems). No CKB antibodies were added for phenotyping experiments. On day 5, partial media exchange was performed. On day 10, expanded HPV-CTLs were restimulated with peptide-pulsed or transfected APCs like day 1. CTLs were used for cytolytic assays or immunophenotyped after day 14 or 20.

Tetramer staining, HPV-CTL, and HPV+ cell line phenotyping

HPV16-tetramers were obtained from NIH Tetramer Core Facility at Emory University. Positive control Flu-M1 and BMLF1 pentamers were obtained from Proimmune, UK. For multimer staining, cells were resuspended in 100 μL staining buffer with 5% human serum and 1 mmol/L dasatanib (ThermoFisher Scientific), and each multimer was added at concentration of 1:100 for 30 minutes at room temperature. Cells were washed twice and restained with anti-CD8-PC5, anti-CD4-FITC, anti-CD14-FITC, and anti-CD19-FITC for exclusion gates, and either a combination of anti-PD1-BV605 and anti-CD39-BV-421 or anti-PD1-BV-605 and anti-TIM3-BV-421 for 30 minutes on ice. For memory markers, CCR7-BV421, anti-CD45RO- BV605, and anti-CD45RA-FITC were stained for 30 minutes on ice after multimer staining. HPV+ cell line PD-L1 staining was done for 30 minutes on ice. Samples were then washed twice in 1× PBS, and analyzed by Attune flow cytometer (ThermoFisher Scientific). Multimer and antibody details are provided in Supplementary Materials and Methods.

HPV-CTL cytotoxicity assays

Cervical cancer cell lines SiHa and Caski were obtained from ATCC. HPV+ HNSCC cell lines were obtained from the following sources: UPCI:SCC90 (SCC90) was obtained from ATCC, UM-SCC-47 (SCC47) and UM-SCC-104 (SCC104) from Merck Millipore. 93-VU-147T (147T) cell line was a kind gift from Dr. Josephine Dorsman, VU Medical Center, the Netherlands. Cell lines were periodically tested for Mycoplasma by PCR and confirmed to be Mycoplasma free. HLA-A*02:01+ HPV+ HNSCC+ SCC-104 cells were prelabeled with 0.5 μmol/L CellTracker Green CMFDA (ThermoFisher Scientific) for 1 hour and washed. HPV-specific CTLs were pooled by HPV-antigen, washed and resuspended in media supplemented with 20 U/mL IL2, with 1 μg/mL isotype IgG or anti-PD1 antibody, DMSO, and 1 μmol/L IDO-1 inhibitor epacadostat (Selleck Chemicals) in various combinations as described. HPV-CTLs were added at ratio of 5:1 to SCC-104 cells and incubated for 48 hours at 37°C, 5% CO2. Cocultured cells were harvested, neutralized with media supernatant from each well containing dead cells and centrifuged for 850 × g, 10 minutes. Cell pellets were washed twice with sterile 1× PBS, resuspended with 1 mL 1× PBS, and 2 μL propidium iodide (ThermoFisher Scientific) and cell death was assessed by flow cytometry.

RNA-seq data alignment

RNA-seq reads for each sample were quality checked using FastQC (version 0.10.1; Babraham Bioinformatics) and aligned to the human genome build 38 (GRCh38) primary assembly and HPV16 genome (GCF_000863945.1) simultaneously using STAR (version 2.5.2B). After alignment, variants were discovered following GATK Best Practices workflow for RNA-seq. Raw RNA-seq reads were preprocessed by adding read groups, indexing, marking duplicates and sorting, Split'N'Trim, reassigning mapping quality and base recalibration.

ssGSEA analysis of HPV and immune gene signatures

Log-transformed transcripts per million (Log2 TPM+1) from each HNSCC sample, after subtraction of low expression genes was used for ssGSEA as previously described in Şenbabaoğlu and colleagues (28). Predefined immune signatures (Supplementary Table S3) have been extensively validated in Şenbabaoğlu and colleagues (28) and Mandal and colleagues (29). ssGSEA scores were computed for each tumor sample using the R package GSVA, and Z-transformed across the cohort prior to analysis. To assess impact of HPV-gene expression on immune signatures, a correlation matrix was built using the R-library Corrplot with the Z-transformed ssGSEA scores and were displayed by hierarchical clustering of correlations. Individual gene expression analysis was performed by unsupervised hierarchical clustering methods and were used for heatmap analysis.

Statistical analysis

Categorical variables, such as Elispot data, and flow cytometric data were summarized as SFUs, and percentages. Continuous variables (RNA-seq data) were presented with mean with SEM. Unpaired T test with Welch's correction was used for all categorical variable analyses, and for continuous variable analyses nonparametric Wilcoxon rank-sum (Mann–Whitney) test was used. For heatmaps of T-cell frequencies and ssGSEA RNA-seq analyses, Z-transformation was performed to normalize the data across the cohorts. R statistical software V3.4.0 and Prism software (GraphPad Software) were used for data managements and statistical analyses. Significance levels were set at 0.05 (*), and P-values of 0.01 (**) or 0.001 (***) for all tests are indicated.

Frequency and specificity of HPV16 E2, E6, and E7-specific CTLs in HPV+ HNSCC

We performed a systematic analysis of potential CTL-epitopes from HPV16 E2, E6, and E7 antigens restricted by 15 globally frequent HLA class I alleles representative of major HLA supertypes (Fig. 1A; Supplementary Fig. S1; ref. 30). We used a comprehensive CTL-epitope prediction strategy we previously developed by incorporating stringent selection criteria (Materials and Methods) to control for inter-algorithmic variations (25, 26). Fifty-nine candidate peptides (24 from E2, 20 from E6, and 15 from E7) were selected covering 13 of the 15 common HLA class-I alleles as candidate HPV-CTL peptides based on predicted HLA-affinity and antigen-processing percentile scores (Fig. 1A; Supplementary Fig. S1). Several previously described HLA-A*02:01-restricted HPV16 E6 and E7 epitopes were predicted with high scores (e.g., E6-KLP epitope, total percentile 94.6; Supplementary Table S1) confirming our prediction strategy. Within the 59 candidate HPV16-peptides, E2 had the lowest number of previously defined CTL-epitopes (3/24, 12%), whereas E6 and E7 had higher number of previously described CTL-epitopes (35% and 46%, respectively). The number of predicted HPV16-peptides ranged from 15 peptides (A*02:01), to 0 peptides (B*40:01, B*44:02) among the selected HLA-alleles (Fig. 1A). To determine if lack of HLA-binding motifs in the three HPV-antigens can poise specific HLA-alleles as risk-factors for HPV+ HNSCCs, we calculated the odds-ratio of HLA-allele frequencies in HPV+ HNSCCs (N = 77), compared with HPV HNSCCs (N = 64; Supplementary Materials and Methods). HLA B*40:01, which had no predicted HPV16-peptides for E2, E6, and E7 had an odds-ratio of 7.48 compared with HPV HNSCCs (Fig. 1A and B; P = 0.059), and had poor-binding peptides for all HPV16-antigens (bottom 20th percentile compared with other HLAs; Supplementary Fig. S2; Supplementary Table S2). HLA-alleles A*24:02, B*07:02, and B*51:01 were also overrepresented (OR ≥ 2) in HPV+ HNSCCs, although they were not statistically significant. Of note, HLA-B*07:02 (OR = 2; Fig. 1B) has been previously reported to be associated with poor clinical outcome in cervical cancer and escape HPV-specific T-cell (HPV-CTL) recognition (31). These results point to the importance of CTL-mediated control of HPV16 malignancies.

Because HPV-CTLs in PBMCs are not abundant (32, 33), we used PBMCs stimulated for 10 days with candidate peptides and CKB antibodies αCTLA4 and αPD-1 to enhance HPV-CTL reactivity (Supplementary Fig. S3). We compared HPV-CTL frequency in PBMCs between patients with HPV+ HNSCC (N = 18) and healthy controls (N = 14) by IFNγ Elispots using antigen-specific peptide pools (Fig. 1C and D; Supplementary Table S1; ref. 27). The HLA-frequency distribution of this cohort largely mirrored median HLA-frequency distribution in the United States (Supplementary Fig. S1B). IFNγ responses against HPV16-E2 were substantially more common (>3-fold higher) in HPV+ HNSCC PBMCs compared with healthy control PBMCs (Wilcoxon rank-sum test, P = 0.0261; Fig. 1C). Moderate-to-high E6-reactivity was observed in patients with HPV+ HNSCC (1.5-fold higher in HPV+ HNSCCs), whereas E7-reactivity was generally low (Fig. 1C and D). To determine if PBMC T-cell reactivity correlates with B-cell immunity, we measured IgG serologic responses to the E2, E6, and E7 antigens in all 18 patient samples in which we tested CTL reactivity. E2- and E7-specific serum IgG titers were higher relative to E6 (>2-fold, P < 0.05) in the patients (Fig. 1E). The majority of patients who had IgG to E2, E6, and E7 also had a measurable CTL response (E2 = 72%, E6 = 60%, E7 = 70%, respectively; Fig. 1F). There was strong concordance between seroreactivity and T-cell reactivity within same antigens (Chi-squared independence test, P = 0.03). Finally, we observed a modest trend towards decreased HPV CTL-response with advanced age and tumor stage of the patient but these were not statistically significant (Supplementary Fig. S4). Thus, our results indicate that E2 and E6 antigens are more CTL-reactive than E7 in patients with HPV+ HNSCC, and HPV-CTL response can be enhanced by CKB antibodies.

Mapping immunodominant epitopes of HPV16 E2, E6, and E7 in HPV+ HNSCCs

To identify novel CTL epitopes from E2, E6, and E7, we performed a second IFNγ Elispot analysis using individual predicted HPV16-peptides against patient-specific HLA-alleles (Fig. 2A). Fifty-one of 59 predicted peptides elicited a T-cell response in at least one patient, indicating a high degree of success (86%) of our prediction-validation strategy (Fig. 2A; Supplementary Table S1). Consistent with pooled-antigen Elispot results (Fig. 1C and D), we observed sub-dominant E7-specific CTL-reactivity relative to E2, E6 epitope-specific CTL-responses (Fig. 2A). Sixteen epitopes had an average response frequency of ≥100 mean spot forming units (SFU)/106 PBMCs and were classified as strongly immunogenic (representative e.g., Fig. 2B–E; Supplementary Table S1). Twenty-nine CTL-epitopes had an average response frequency between 10 and 100 SFU/106 PBMCs (moderately immunogenic), whereas six epitopes had an average response frequency <10 SFU/106 PBMCs (low immunogenic). The majority of moderate to highly immunogenic epitopes (77%) were novel, or had not been described with the observed HLA restriction (Fig. 2B–E; Supplementary Table S1). We also observed 16 unique epitopes that elicited a cross-reactive response to other alleles within the same supertype supporting the strategy for HLA-supertype based epitope prediction (Supplementary Table S1). Figure 2E shows a representative example, where an HLA-A*11:01 restricted E2-peptide had strong predicted binding affinity and elicited strong CTL-reactivity to HLA-A*68:01 (A3 supertype). Within the E2 antigen, most CTL-epitopes (52%) were clustered within the trans-activating DNA-binding domain, 23% in the hinge region, and 24% in the DNA-binding domain (Supplementary Fig. S5A). Within E6, the immunodominant regions (70% of epitopes) encompassed amino acids 37–109 with 40% of epitopes arising in the first zinc finger domain (Supplementary Fig. S5B). Interestingly, the zinc finger domain of E7 also had 42% of the CTL-epitopes (Supplementary Fig. S5C). Thus, we have enhanced the landscape of E2, E6, and E7 CTL-epitopes, and mapped the immunodominant regions of E2, E6, and E7 for future studies.

HPV-specific CTLs exhibit dysfunctional phenotype in patients with HPV+ HNSCC

We first interrogated whether responding ex vivo stimulated HPV-specific peripheral CTLs in patients with HPV+ HNSCC were in naive or memory T-cell compartments. After one round of ex vivo stimulation by autologous APC presenting cognate HPV16-antigen, the majority (>79%) of HPV-specific CTLs detectable by antigen-specific multimers in HLA-A*02:01+ patients exhibited memory phenotype distributed between effector memory (TEM, CD45ROhiCCR7lo) and central memory (TCM, CD45ROhiCCR7hi) compartments (representative examples in Supplementary Fig. S6A and S6B).

E2-CTLs had a slightly higher frequency in the TEM compartment, compared with E6, E7 CTLs. Although CTL dysfunction in chronic viral infections and cancers has been extensively described (34), few studies have focused on the extent of T-cell exhaustion in patients with HPV+ HNSCC because of the difficulties in studying low-frequency HPV-CTLs (33). Although the majority of HPV-CTLs were in the memory compartment, we then evaluated the prevalence of dysfunctional HPV-CTLs from patients with HPV+ HNSCC after stimulation in the absence of CKB antibodies. We reasoned that ex vivo activated HPV-specific PD1+ CTLs that are poised towards the exhaustion spectrum will become further dysfunctional after APC-stimulation and acquire additional inhibitory markers characteristic of profound dysfunction, defined by antigen-specific CTLs exhibiting the CD39+PD-1+ or TIM-3+PD-1+ phenotype (DPEx-phenotype; refs. 35–38).

First, chronic EBV-BMLF1 antigen-specific HLA-A*02:01-restricted CTLs compared with acute Flu-M1 antigen specific CTLs, displayed a substantially higher DPEx-phenotype (CD39+PD-1+ 7.5-fold, TIM-3+PD-1+ 67-fold), indicating the validity of this approach (Supplementary Fig. S6C).

In four HLA-A*02:01+ patients with HPV-specific CTLs detectable by antigen-specific tetramers (Fig. 3A–C), and one HLA-A*68:01+ patient where CD137-positivity defined HPV-CTLs, we phenotyped total and E2, E6, and E7-specific DPEx CTLs indicating substantial exhaustion. As shown in the representative example Pt.7002, E2-CTLs were higher in frequency than E6 and E7-CTLs (Fig. 3A and B). Within the HPV16-antigens in Pt.7002, CTLs stimulated with E7-transfected APCs exhibited the highest levels of CD8+DPEx fold-change relative to mock-antigen transfected (2- to 4-fold; Fig. 3B) followed by E2 (1- to 3-fold; Fig. 3B) and E6, respectively. In three of five patients with HPV+ HNSCC, total CD8+ DPEx and HPV-specific CD8+DPEx cells were higher in CTLs stimulated with E7-antigen (between 2- and 10-fold) relative to E2/E6 antigen-stimulated CTLs (Pts. 7002, 7007, 7012; Fig. 3B and C, unpaired Welch t test, P < 0.05), independent of HLA-status (e.g., HLA-A*68:01+ Pt. 7007; Fig. 3C). In the other two patients, E2-CD8+DPEx was higher than E7-CD8+DPEx (2- to 3-fold, Pts. 7035, 7050; Fig. 3C, unpaired Welch t test, P < 0.05), indicating heterogeneity in HPV-specific CTL-dysfunction in HPV+ HNSCC. Interestingly, compared with E2 and E7-CTLs, E6-CD8+DPEx remained relatively low in most patients, and there was an inverse relationship within patients between E7-CD8+DPEx and E2/E6-CD8+DPEx (Fig. 3C). Unsupervised hierarchical clustering of DPEx-frequencies of total CD8+ and HPV-specific CD8+ T cells revealed this trend where high E2-CD8+DPEx and E6-CD8+DPEx co-occurred in patients who had relatively lower E7-CD8+DPEx and vice versa (Tukey's multiple comparisons test, E2 vs. E7, P = 0.014, E6 vs. E7, P = 0.084, E2 vs. E6, P = NS; Fig. 3D). These results suggest that in patients with HPV+ HNSCC, E7/E2 CTL relative dysfunction levels may be higher compared with E6-CTLs, although larger cohort experiments are necessary to validate these observations.

HPV16-antigen load correlates with T-cell exhaustion

To provide a broader analysis of immune dysfunction from the tumor side in HPV+ HNSCCs, we performed an immune signature analysis of two large publicly available HNSCC transcriptomes (TCGA, UM-cohorts, N = 119, 51 HPV+, 68 HPV; refs. 3, 22). We used previously validated immune signatures representing tumor-infiltrating immune cell subsets and performed single-sample gene set enrichment analysis (ssGSEA; Supplementary Materials and Methods; Supplementary Table S3 lists the gene signatures) to score the HPV+ and HPV subsets (28, 29, 39). Patients with HPV+ HNSCC in general had higher immune infiltration scores compared with HPV HNSCCs (Supplementary Fig. S7), with 36/51 (70%) of HPV+ HNSCC samples represented in the T-cell-high gene cluster, and few HPV+ HNSCC samples (17%) with very low immune cell infiltration scores, confirming and expanding the findings in previous studies (29). To assess the impact of HPV gene expression on immune cell infiltration, we calculated the Spearman correlation coefficients among ssGSEA scores for the entire gene sets across all patients, including HPV16 genes and performed an unsupervised clustering on the correlation matrix (Fig. 4A). We observed that HPV16-gene signatures formed a distinct module, correlating best with cytotoxic T-cell infiltration (CYT, cytotoxic), B-cells, regulatory T-cell signatures (TIL.Treg), and dysfunctional T-cell signatures (Exhaustion, TIGIT signatures; Fig. 4A; Supplementary Table S3). Interestingly, HPV-gene signatures also negatively correlated with neutrophils and other myeloid gene signatures, indicating that lymphocytes dominate the immune landscape of HPV+ HNSCCs (Fig. 4A; Supplementary Fig. S7). We also observed a previously described exhaustion gene set correlating with HPV-gene sets in this module (Spearman ρ = 0.33; Fig. 4A). Unsupervised hierarchical clustering on expression levels of constituent genes within the exhaustion gene set revealed three main groups with low (L), moderate (M), and high (H) expression of immune regulatory gene expression within patients with HPV16+ HNSCC (Fig. 4B). We then analyzed individual HPV16-gene expression in the HPV16+ HNSCC tumors (N = 40) stratified into Exhaustion-high (EX-H) and Exhaustion moderate/low tumors (EX-ML; Fig. 4C). EX-H HPV16+ HNSCC tumors had higher gene expression of E1, E2, E4, E6, and L2 genes compared with the EX-ML subset (Fig. 4C, unpaired Welch t test; E1, E2, P < 0.01; E4, E6, L2, P < 0.1). E7-expression was comparably high in both the subsets, whereas E5 known to downregulate MHC-class I expression (40), remained low in the EX-H subset (Fig. 4C). These computational analyses along with previous HPV-CTL-dysfunction experiments (Fig. 3) suggest that HPV-specific CTLs have T-cell exhaustion at tumor sites, driven by intratumoral HPV-antigen expression.

IDO-1 represents a novel HPV+ HNSCC-specific immune target

We analyzed differential expression of constituent genes within the exhaustion gene set between HPV+ HNSCCs and HPV HNSCCs (Fig. 5A; Supplementary Fig. S8). We observed several well-known T-cell regulatory genes such as LAG-3, GAL-9, CEACAM-1, and CTLA-4 overexpressed in HPV+ HNSCCs compared with HPV HNSCCs, consistent with high T-cell infiltration and dysfunction as observed in our results (Figs. 3 and 4; Supplementary Fig. S7), and other studies (29). Interestingly, we also observed exhaustion genes NRP1 and CD73 that were selectively upregulated in HPV HNSCCs compared with HPV+ HNSCCs (Supplementary Fig. S8), indicating distinct types of T-cell dysfunction between the two HNSCC subtypes. Indoleamine 2,3-dioxygenase (IDO-1), an L-tryptophan catabolizing enzyme was one of the highest differentially expressed gene (based on ranked P-value) from the exhaustion set HPV+ HNSCCs compared with HPV HNSCCs (Fig. 5A; Mann–Whitney test, P = 0.0012). A cross-cancer (N = 30 types, 45,708 total tumors) gene expression analysis from cBioPortal (41), revealed that IDO-1 is also highly expressed in the HPV-malignancy cervical cancer (Supplementary Fig. S9A). Within the HPV+ HNSCCs, IDO-1 mRNA levels correlated with HPV16-E7 expression (Fig. 5B; Pearson r = 0.447, P = 0.001), and with tumor PD-L1 expression (Fig. 5B; Pearson r = 0.71, P < 0.0001). IDO-1 expression further correlated with M1-Macrophage signature (Pearson r = 0.517, P < 0.0001), and as expected with cytotoxic signature (Pearson r = 0.61, P < 0.0001), but only modestly with M2-Macrophage signature (Pearson r = 0.24, Supplementary Fig. S9B). To experimentally validate these in silico immune signature analyses indicating that HPV-antigen expression can impact immune regulatory gene expression such as IDO-1 (Fig. 4), we performed immunoblotting for IDO-1 expression in a panel of HPV16+ cancer cell lines (2 cervical, 4 HPV+ HNSCC). These showed variability in HPV16-E7 protein expression (Fig. 5C). IDO-1 expression followed a striking correlation with E7-protein expression in the same cell lines (R2 = 0.84, P = 0.033; Fig. 5C). Transfection of the 3-HPV16-antigens into a non-HPV cell line (HEK-293-T) did not alter IDO-1 protein expression (Supplementary Fig. S9C). Of note, PD-L1 protein expression on the same cell lines did not correlate with E7-antigen expression (Spearman ρ = 0.17; P = NS; Supplementary Fig. S9D).

IDO-1 inhibition enhances T-cell targeting of HPV+ HNSCCs

IDO-1 inhibitors are being evaluated in preclinical and clinical settings to enhance tumor immunity (42, 43). We therefore explored the possibility of validating our immunogenomic analyses by exploiting IDO-1 inhibition to overcome HPV-CTL exhaustion, especially using the more dysfunctional E7-CTLs (Fig. 3). We fluorescently labeled the HLA-A*02:01+ HPV+ HNSCC cell line UM-SCC-104 (SCC-104), which has high IDO-1, E7 and PD-L1 expression (Fig. 5C; Supplementary Fig. S9D). SCC-104 cells were previously reported to have a distinct hierarchy of HPV16-antigen expression where E7 > E6 > E2 (44). We assessed HPV-CTL–mediated cytotoxicity on SCC-104 cells, after co-incubation with ex vivo expanded HPV-CTLs from an HLA-A*02:01+ HPV+ HNSCC patient in the presence of either anti-PD-1 antibody (αPD-1+DMSO) or IDO-1 inhibitor epacadostat (Ig+IDO-1i), or both (αPD-1+IDO-1i; Fig. 5D and E). Within E7-CTLs, single-agent treatment with either αPD-1 or IDO-1i individually resulted in a three- to five-fold increase in sensitivity of SCC-104 to E7-CTL–mediated cytotoxicity compared with mock (Ig+DMSO) treatment (Fig. 5D and E, αPD-1 vs. mock, P = 0.024; IDO-1i vs. mock, P = 0.064). In contrast, synergistic combination blockade with both αPD-1+IDO-1i resulted in a 10-fold increase in tumor cytotoxicity compared with mock treatment (Fig. 5C and D, P = 0.011), and a two- to three-fold increase in tumor cytotoxicity compared with the single-agent treatments (Fig. 5D and E, P = 0.04 compared with IDO-1i, P = 0.013 compared with αPD-1). Similar results were obtained with αPD-1+IDO-1i combination therapy on E2 and E6 CTL cytotoxicity although to a lesser extent (three-fold increase for E2, and five-fold increase for E6 compared with mock treatment, P < 0.01), likely reflecting the lower expression of these antigens in SCC-104 cell line (44). These results demonstrate that IDO-1 is a novel HPV+ HNSCC-specific checkpoint correlating with HPV-antigen expression, and combination inhibition of PD-1 and IDO-1 can sensitize HPV+ HNSCCs to HPV-CTL–mediated cytotoxicity.

HPV-driven malignancies remain an ideal model system for cancer immunotherapy, due to (i) a long lead time from infection to malignancy, (ii) emerging immune and viral biomarkers for early detection, (iii) the persistent tissue expression of viral oncogenes, and (iv) epidemiologic evidence of the central role of T-cell control of viral persistence. However, the dynamics of viral persistence within immunocompetent individuals and the mechanisms of tumor immune escape remain largely unknown, in particular for HPV+ HNSCC. The emerging epidemic of HPV+ HNSCC and lack of screening modalities represents a major clinical challenge and opportunity for targeted T-cell immunotherapy.

In this study, we have expanded the spectrum of HPV+ HNSCC-specific immune therapeutic targets at the CTL-epitope level and at the target tumor cell-modulatory level. We chose E2, E6, and E7-antigens, as they induce strong B-cell immunity, have been detected in preinvasive and/or invasive cervical cancer, and we confirmed viral antigen expression in HPV16+ HNSCC transcriptomes. Most studies that have attempted to define CTL-immunogenicity from HPV16 have primarily focused on a limited number of HLA-alleles (e.g., A*02:01) and peptides from E6 and E7 (Supplementary Table S1), with limited data on immunogenic targets in in HPV+ HNSCC (32, 45, 46). The 15 HLA alleles chosen for this study are predicted to include 10/12 of HLA supertypes and over 95% of the global population (30, 32). Of the 15 HLA alleles, we failed to identify peptides for HLA-B*40:01 and HLA-B*4402. HLA B*40:01 is significantly overrepresented in the HPV+ HNSCC cohort compared with HPV HNSCC (Fig. 1B), but these data remain to be confirmed in larger datasets and association studies. Viral immune escape by altering HLA-binding CTL epitopes has been documented in HIV-1 and HCV infections (47, 48), but not as well for DNA viruses such as HPV, where the mutation rates are markedly lower.

We identified several immunogenic CTL-epitopes from the three HPV16-antigens (Fig. 2; Supplementary Table S1). In our experiments, addition of PD-1 and CTLA-4 CKB antibodies aided our ability amplify and detect low-frequency HPV-CTL-response in both healthy and HPV+ PBMCs ex vivo (Fig. 1C and D; Supplementary Fig. S3). A potential caveat of this approach is that we did not specifically assess tumor-infiltrating HPV-TILs in these patients; however, the presence of HPV-antigen-specific CTLs in the memory compartments, along with the expression of tumor-specific dysfunction markers such as PD-1, CD39, and TIM3 lends support to their tumor specificity (37, 38). Our results indicate that HPV16-E2 and E6 induce more CTL responses than HPV16-E7 (Figs. 1 and 2). HPV16-E6 and E7 have been the dominant targets for T-cell–based immune therapies against HPV thus far (14, 18). In contrast, E2-specific CTL-reactivity has been unexplored as an immunotherapeutic target in HPV+ HNSCCs due to the assumption that E2-locus is interrupted by viral integration, similar to that observed in cervical cancer (17). However, several recent whole genome studies in HNSCCs have indicated that viral breakpoints in HPV+ HNSCCs are distributed throughout the genome, with preferential integration in the E1 region (2, 21). E2 is also a larger antigen more than three times the size of E6, E7, possibly explaining the bigger spectrum of CTL-epitopes from the protein. These results, taken along with the high proportion of episomal full-length HPV DNA in HNSCC lesions (49), and our data demonstrating strong E2-specific T-cell and B-cell reactivity (Figs. 1 and 2) warrant further investigation of E2 as a T-cell therapeutic target in addition to E6 and E7 in HPV+ HNSCCs.

We detected low levels of E7-CTLs compared with E2 and E6-CTLs in this study. This can be due to (i) inaccurate prediction of CTL-epitopes, (ii) inherently low immunogenicity of E7-antigen, (iii) low antigen load in patients, or (iv) higher levels of dysfunctional E7-specific CTLs. Our ability to accurately predict previously described epitopes from E7 and the successful identification of novel CTL-epitopes from E2 and E6 across various HLA-alleles (Fig. 2A), argues against a suboptimal prediction strategy. The presence of high levels of serum titers against E7 in patients with HPV+ HNSCC indicates that the antigen is presented and is immunogenic at least in context of B-cell immunity (Fig. 1E and F). Gene expression analysis of HPV+ HNSCC tissue and cell lines showed that E7-antigen load is high in patient tumors, consistent with several other studies (22, 49). In our exploratory experiments, E7-CTLs tended to exhibit higher levels of PD1+CD39+ or PD1+TIM3+ DPEx-phenotypes compared with E2 and E6-CTLs after ex vivo stimulation in three/five independent patients with HPV+ HNSCC (Fig. 3). In particular, E7-CTL relative dysfunction levels occurred distinct from E2/E6 CTL dysfunction within the same patient (Fig. 3B–D). Because antigen persistence and the subsequent magnitude of CTL-response are major factors in chronic viral T-cell exhaustion (50), we speculate that in each of these patients, the variably dysfunctional CTLs might reflect temporal tumor HPV-load or tumor heterogeneity. Although the effect of low sample size in these experiments is not precluded, future studies with larger cohorts that compare and correlate tumor antigen load in vivo with the dynamics of E2, E6, and E7-CTL phenotypes in peripheral blood and tumor will be needed to verify our findings.

Although other studies have shown high levels of immune infiltration in HPV+ HNSCCs (22, 29), we show here that HPV-antigen load likely drives high CTL-infiltration and CTL-dysfunction (Fig. 4A), arguing for better response to immune CKB therapies. Indeed, preliminary data from ongoing HNSCC clinical trials targeting the PD-1/PD-L1 axis indicate more benefit for HPV+ HNSCCs compared with HPV HNSCCs (9, 10). Our data thus provide mechanistic insights into this clinical response, wherein, high HPV-antigen load likely drives T-cell infiltration into HPV+ HNSCCs causing immune selection pressure for HPV-CTL dysfunction (in particular E7-CTLs), and that immune CKB can at least partially reverse this effect in HPV+ HNSCCs (∼32% ORR in ref. 10).

In addition to the well-described PD-1 checkpoint, we demonstrate by computational analysis and subsequent experimental in vitro validation that IDO-1 is highly expressed in HPV+ HNSCC and other HPV-driven malignancies (Fig. 5A; Supplementary S9A). In preclinical murine melanoma models, TILs increase tumor IDO-1 expression (51) and IDO-1 inhibitors are showing promise in clinical trials in particular with PD-1/PD-L1 blockade (52). HPV+ HNSCCs expressing IDO-1 might similarly be driven by HPV-specific-CTL infiltration in response to high tumoral HPV antigen load (Figs. 4 and 5). In vitro, this resistance to CTL-targeting by HPV+ HNSCCs is apparent in the absence of PD-1/IDO-1 inhibition, where only ∼5% of SCC-104 cells were sensitive to CTL-cytotoxicity regardless of the antigen-specificity of the HPV-CTLs (Fig. 5E). In contrast, inhibition of IDO-1 alone or in combination with PD-1 blockade significantly enhances tumor cell cytotoxicity of E7-CTL (and, to a lesser extent, E2 and E6-CTL) derived from patients with HPV+ HNSCCs (Fig. 5E). These data suggest that IDO-1/PD-1 blockade may have a significant effect to activate preexisting HPV-specific CTLs in HPV+ HNSCCs.

Sixty years after the discovery of HPV and 10 years after FDA approval of the first HPV vaccine, HPV-associated malignancies remain a major public threat, with an estimated 14 million new HPV-infections occurring every year (53). The presence of highly-expressed viral antigens makes HPV+ HNSCCs a promising setting for targeted immunotherapies. We propose that vaccination or ACT to HPV16-specific CTL epitopes from E2, E6, and E7, and targeted immune modulation with a combination of PD-1 and IDO-1 inhibition warrant further evaluation in HPV+ HNSCC and other HPV-associated malignancies.

A. Sikora reports receiving a commercial research grant from Tessa Therapeutics. No potential conflicts of interest were disclosed by the other authors.

Conception and design: S. Krishna, A.K. Read, K.S. Anderson

Development of methodology: S. Krishna, P. Ulrich, A.K. Read, S. Kim-Schulze, M.A. Wilson Sayres, K.S. Anderson

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Krishna, P. Ulrich, E. Wilson, F. Parikh, A.K. Read, S. Kim-Schulze, M. Posner, A. Sikora, K.S. Anderson

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Krishna, E. Wilson, P. Narang, S. Yang, S. Kim-Schulze, J.G. Park, M. Posner, M.A. Wilson Sayres, A. Sikora, K.S. Anderson

Writing, review, and/or revision of the manuscript: S. Krishna, P. Ulrich, S. Kim-Schulze, J.G. Park, M. Posner, M.A. Wilson Sayres, A. Sikora, K.S. Anderson

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. Krishna, A.K. Read, K.S. Anderson

Study supervision: K.S. Anderson

We thank Marika Hopper, Padhmavathy Yuvaraj from Anderson lab for help with cell culture and R-ELISA analysis, and Samantha Cotsmire (Biodesign CIVV, ASU) for helpful advice. We thank Yasin Şenbabaoğlu (Genentech) for input on ssGSEA analysis. S. Krishna acknowledges funding support from ASU School of Graduate Education Dissertation Completion Fellowship, and School of Biological and Health Systems Engineering, ASU. We also thank the Human Immune Monitoring Center, Ichan School of Medicine for patient sample processing and management. This work was supported by institutional funds from Arizona State University.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Torre
LA
,
Bray
F
,
Siegel
RL
,
Ferlay
J
,
Lortet-Tieulent
J
,
Jemal
A
. 
Global cancer statistics, 2012
.
CA Cancer J Clin
2015
;
65
:
87
108
.
2.
Parfenov
M
,
Pedamallu
CS
,
Gehlenborg
N
,
Freeman
SS
,
Danilova
L
,
Bristow
CA
, et al
Characterization of HPV and host genome interactions in primary head and neck cancers
.
Proc Natl Acad Sci U S A
2014
;
111
:
15544
9
.
3.
The Cancer Genome Atlas Network
. 
Comprehensive genomic characterization of head and neck squamous cell carcinomas
.
Nature
2015
;
517
:
576
82
.
4.
Chaturvedi
AK
,
Anderson
WF
,
Lortet-Tieulent
J
,
Curado
MP
,
Ferlay
J
,
Franceschi
S
, et al
Worldwide trends in incidence rates for oral cavity and oropharyngeal cancers
.
J Clin Oncol
2013
;
31
:
4550
9
.
5.
Gillison
ML
,
Broutian
T
,
Pickard
RKL
,
Tong
Z-Y
,
Xiao
W
,
Kahle
L
, et al
Prevalence of oral HPV infection in the United States, 2009–2010
.
JAMA
2012
;
307
:
693
703
.
6.
Gillison
ML
,
Chaturvedi
AK
,
Anderson
WF
,
Fakhry
C
. 
Epidemiology of human papillomavirus-positive head and neck squamous cell carcinoma
.
J Clin Oncol
2015
;
33
:
3235
42
.
7.
Pardoll
DM
. 
The blockade of immune checkpoints in cancer immunotherapy
.
Nat Rev Cancer
2012
;
12
:
252
64
.
8.
Robert
C
,
Schachter
J
,
Long
GV
,
Arance
A
,
Grob
JJ
,
Mortier
L
, et al
Pembrolizumab versus ipilimumab in advanced melanoma
.
N Engl J Med
2015
;
372
:
2521
32
.
9.
Ferris
RL
,
Blumenschein
G
 Jr
,
Fayette
J
,
Guigay
J
,
Colevas
AD
,
Licitra
L
, et al
Nivolumab for recurrent squamous-cell carcinoma of the head and neck
.
N Engl J Med
2016
;
375
:
1856
67
.
10.
Chow
LQM
,
Haddad
R
,
Gupta
S
,
Mahipal
A
,
Mehra
R
,
Tahara
M
, et al
Antitumor activity of pembrolizumab in biomarker-unselected patients with recurrent and/or metastatic head and neck squamous cell carcinoma: results from the phase Ib KEYNOTE-012 Expansion Cohort
.
J Clin Oncol
2016
;
34
:
3838
45
.
11.
Schumacher
TN
,
Schreiber
RD
. 
Neoantigens in cancer immunotherapy
.
Science
2015
;
348
:
69
74
.
12.
Ott
PA
,
Hu
Z
,
Keskin
DB
,
Shukla
SA
,
Sun
J
,
Bozym
DJ
, et al
An immunogenic personal neoantigen vaccine for patients with melanoma
.
Nature
2017
;
547
:
217
21
.
13.
Tran
E
,
Robbins
PF
,
Lu
Y-C
,
Prickett
TD
,
Gartner
JJ
,
Jia
L
, et al
T-cell transfer therapy targeting mutant KRAS in cancer
.
N Engl J Med
2016
;
375
:
2255
62
.
14.
Kenter
GG
,
Welters
MJP
,
Valentijn
ARPM
,
Lowik
MJG
,
Berends-van der Meer
DMA
,
Vloon
APG
, et al
Vaccination against HPV-16 oncoproteins for vulvar intraepithelial neoplasia
.
N Engl J Med
2009
;
361
:
1838
47
.
15.
Rizvi
NA
,
Hellmann
MD
,
Snyder
A
,
Kvistborg
P
,
Makarov
V
,
Havel
JJ
, et al
Cancer immunology. Mutational landscape determines sensitivity to PD-1 blockade in non-small cell lung cancer
.
Science
2015
;
348
:
124
8
.
16.
Rosenblatt
J
,
Stone
RM
,
Uhl
L
,
Neuberg
D
,
Joyce
R
,
Levine
JD
, et al
Individualized vaccination of AML patients in remission is associated with induction of antileukemia immunity and prolonged remissions
.
Sci Transl Med
2016
;
8
:
368ra171
.
17.
Woodman
CBJ
,
Collins
SI
,
Young
LS
. 
The natural history of cervical HPV infection: unresolved issues
.
Nat Rev Cancer
2007
;
7
:
11
22
.
18.
Draper
LM
,
Kwong
MLM
,
Gros
A
,
Stevanović
S
,
Tran
E
,
Kerkar
S
, et al
Targeting of HPV-16+ epithelial cancer cells by TCR gene engineered T cells directed against E6
.
Clin Cancer Res
2015
;
21
:
4431
9
.
19.
Trimble
CL
,
Morrow
MP
,
Kraynyak
KA
,
Shen
X
,
Dallas
M
,
Yan
J
, et al
Safety, efficacy, and immunogenicity of VGX-3100, a therapeutic synthetic DNA vaccine targeting human papillomavirus 16 and 18 E6 and E7 proteins for cervical intraepithelial neoplasia 2/3: a randomised, double-blind, placebo-controlled phase 2b trial
.
Lancet
2015
;
386
:
2078
88
.
20.
Rusan
M
,
Li
YY
,
Hammerman
PS
. 
Genomic landscape of human papillomavirus-associated cancers
.
Clin Cancer Res
2015
;
21
:
2009
19
.
21.
Akagi
K
,
Li
J
,
Broutian
TR
,
Padilla-Nash
H
,
Xiao
W
,
Jiang
B
, et al
Genome-wide analysis of HPV integration in human cancers reveals recurrent, focal genomic instability
.
Genome Res
2014
;
24
:
185
99
.
22.
Zhang
Y
,
Koneva
LA
,
Virani
S
,
Arthur
AE
,
Virani
A
,
Hall
PB
, et al
Subtypes of HPV-positive head and neck cancers are associated with HPV characteristics, copy number alterations, PIK3CA mutation, and pathway signatures
.
Clin Cancer Res
2016
;
22
:
4735
45
.
23.
Anderson
KS
,
Dahlstrom
KR
,
Cheng
JN
,
Alam
R
,
Li
G
,
Wei
Q
, et al
HPV16 antibodies as risk factors for oropharyngeal cancer and their association with tumor HPV and smoking status
.
Oral Oncol
2015
;
51
:
662
7
.
24.
D'Souza
G
,
Kreimer
AR
,
Viscidi
R
,
Pawlita
M
,
Fakhry
C
,
Koch
WM
, et al
Case–control study of human papillomavirus and oropharyngeal cancer
.
N Engl J Med
2007
;
356
:
1944
56
.
25.
Chowell
D
,
Krishna
S
,
Becker
PD
,
Cocita
C
,
Shu
J
,
Tan
X
, et al
TCR contact residue hydrophobicity is a hallmark of immunogenic CD8+ T cell epitopes
.
Proc Natl Acad Sci U S A
2015
;
112
:
E1754
62
.
26.
Krishna
S
,
Anderson
KS
. 
T-cell epitope discovery for therapeutic cancer vaccines
.
Methods Mol Biol
2016
;
1403
:
779
96
.
27.
Parikh
F
,
Duluc
D
,
Imai
N
,
Clark
A
,
Misiukiewicz
K
,
Bonomi
M
, et al
Chemoradiotherapy-induced upregulation of PD-1 antagonizes immunity to HPV-related oropharyngeal cancer
.
Cancer Res
2014
;
74
:
7205
16
.
28.
Şenbabaoğlu
Y
,
Gejman
RS
,
Winer
AG
,
Liu
M
,
Van Allen
EM
,
de Velasco
G
, et al
Tumor immune microenvironment characterization in clear cell renal cell carcinoma identifies prognostic and immunotherapeutically relevant messenger RNA signatures
.
Genome Biol
2016
;
17
:
231
.
29.
Mandal
R
,
Şenbabaoğlu
Y
,
Desrichard
A
,
Havel
JJ
,
Dalin
MG
,
Riaz
N
, et al
The head and neck cancer immune landscape and its immunotherapeutic implications
.
JCI Insight
2016
;
1
:
e89829
.
30.
Lund
O
,
Nielsen
M
,
Kesmir
C
,
Petersen
AG
,
Lundegaard
C
,
Worning
P
, et al
Definition of supertypes for HLA molecules using clustering of specificity matrices
.
Immunogenetics
2004
;
55
:
797
810
.
31.
Ellis
JR
,
Keating
PJ
,
Baird
J
,
Hounsell
EF
,
Renouf
DV
,
Rowe
M
, et al
The association of an HPV16 oncogene variant with HLA-B7 has implications for vaccine design in cervical cancer
.
Nat Med
1995
;
1
:
464
70
.
32.
Riemer
AB
,
Keskin
DB
,
Zhang
G
,
Handley
M
,
Anderson
KS
,
Brusic
V
, et al
A conserved E7-derived cytotoxic T lymphocyte epitope expressed on human papillomavirus 16-transformed HLA-A2+ epithelial cancers
.
J Biol Chem
2010
;
285
:
29608
22
.
33.
Badoual
C
,
Hans
S
,
Merillon
N
,
Van Ryswick
C
,
Ravel
P
,
Benhamouda
N
, et al
PD-1-expressing tumor-infiltrating T cells are a favorable prognostic biomarker in HPV-associated head and neck cancer
.
Cancer Res
2013
;
73
:
128
38
.
34.
Barber
DL
,
Wherry
EJ
,
Masopust
D
,
Zhu
B
,
Allison
JP
,
Sharpe
AH
, et al
Restoring function in exhausted CD8 T cells during chronic viral infection
.
Nature
2006
;
439
:
682
7
.
35.
Wherry
EJ
,
John Wherry
E
,
Kurachi
M
. 
Molecular and cellular insights into T cell exhaustion
.
Nat Rev Immunol
2015
;
15
:
486
99
.
36.
Gupta
PK
,
Godec
J
,
Wolski
D
,
Adland
E
,
Yates
K
,
Pauken
KE
, et al
CD39 expression identifies terminally exhausted CD8+ T cells
.
PLoS Pathog
2015
;
11
:
e1005177
.
37.
Canale
FP
,
Ramello
MC
,
Núñez
N
,
Furlan
CLA
,
Bossio
SN
,
Serrán
MG
, et al
CD39 expression defines cell exhaustion in tumor-infiltrating CD8+ T cells
.
Cancer Res
2018
;
78
:
115
28
.
38.
Simoni
Y
,
Becht
E
,
Fehlings
M
,
Loh
CY
,
Koo
S-L
,
Teng
KWW
, et al
Bystander CD8 T cells are abundant and phenotypically distinct in human tumour infiltrates
.
Nature
2018
;
557
:
575
9
.
39.
De Simone
M
,
De Simone
M
,
Arrigoni
A
,
Rossetti
G
,
Gruarin
P
,
Ranzani
V
, et al
Transcriptional landscape of human tissue lymphocytes unveils uniqueness of tumor-infiltrating T regulatory cells
.
Immunity
2016
;
45
:
1135
47
.
40.
Campo
MS
,
Graham
SV
,
Cortese
MS
,
Ashrafi
GH
,
Araibi
EH
,
Dornan
ES
, et al
HPV-16 E5 down-regulates expression of surface HLA class I and reduces recognition by CD8 T cells
.
Virology
2010
;
407
:
137
42
.
41.
Cerami
E
,
Gao
J
,
Dogrusoz
U
,
Gross
BE
,
Sumer
SO
,
Aksoy
BA
, et al
The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data: figure 1
.
Cancer Discov
2012
;
2
:
401
4
.
42.
Gangadhar
TC
,
Hamid
O
,
Smith
DC
,
Bauer
TM
,
Wasser
JS
,
Luke
JJ
, et al
Preliminary results from a Phase I/II study of epacadostat (incb024360) in combination with pembrolizumab in patients with selected advanced cancers
.
J ImmunoTherapy Cancer
2015
;
3
:
O7
.
43.
Prendergast
GC
,
Malachowski
WP
,
DuHadaway
JB
,
Muller
AJ
. 
Discovery of IDO1 inhibitors: from bench to bedside
.
Cancer Res
2017
;
77
:
6795
811
.
44.
Olthof
NC
,
Huebbers
CU
,
Kolligs
J
,
Henfling
M
,
Ramaekers
FCS
,
Cornet
I
, et al
Viral load, gene expression and mapping of viral integration sites in HPV16-associated HNSCC cell lines
.
Int J Cancer
2015
;
136
:
E207
18
.
45.
Rudolf
MP
,
Man
S
,
Melief
CJ
,
Sette
A
,
Kast
WM
. 
Human T-cell responses to HLA-A-restricted high binding affinity peptides of human papillomavirus type 18 proteins E6 and E7
.
Clin Cancer Res
2001
;
7
:
788s
795s
.
46.
Ressing
ME
,
Sette
A
,
Brandt
RM
,
Ruppert
J
,
Wentworth
PA
,
Hartman
M
, et al
Human CTL epitopes encoded by human papillomavirus type 16 E6 and E7 identified through in vivo and in vitro immunogenicity studies of HLA-A*0201-binding peptides
.
J Immunol
1995
;
154
:
5934
43
.
47.
Price
DA
,
Goulder
PJ
,
Klenerman
P
,
Sewell
AK
,
Easterbrook
PJ
,
Troop
M
, et al
Positive selection of HIV-1 cytotoxic T lymphocyte escape variants during primary infection
.
Proc Natl Acad Sci U S A
1997
;
94
:
1890
5
.
48.
Petrovic
D
,
Dempsey
E
,
Doherty
DG
,
Kelleher
D
,
Long
A
. 
Hepatitis C virus–T-cell responses and viral escape mutations
.
Eur J Immunol
2012
;
42
:
17
26
.
49.
Olthof
NC
,
Speel
E-JM
,
Kolligs
J
,
Haesevoets
A
,
Henfling
M
,
Ramaekers
FCS
, et al
Comprehensive analysis of HPV16 integration in OSCC reveals no significant impact of physical status on viral oncogene and virally disrupted human gene expression
.
PLoS One
2014
;
9
:
e88718
.
50.
Mueller
SN
,
Ahmed
R
. 
High antigen levels are the cause of T cell exhaustion during chronic viral infection
.
Proc Natl Acad Sci U S A
2009
;
106
:
8623
8
.
51.
Spranger
S
,
Spaapen
RM
,
Zha
Y
,
Williams
J
,
Meng
Y
,
Ha
TT
, et al
Up-regulation of PD-L1, IDO, and Tregs in the melanoma tumor microenvironment is driven by CD8 T cells
.
Sci Transl Med
2013
;
5
:
200ra116
.
52.
Sheridan
C
. 
IDO inhibitors move center stage in immuno-oncology
.
Nat Biotechnol
2015
;
33
:
321
2
.
53.
Dunne
EF
,
Markowitz
LE
,
Saraiya
M
,
Stokley
S
,
Middleman
A
,
Unger
ER
, et al
CDC grand rounds: reducing the burden of HPV-associated cancer and disease
.
MMWR Morb Mortal Wkly Rep
2014
;
63
:
69
72
.