HPV16 E6 and E7 expressing cancer cells suppress the antitumor immune response by upregulating KLF2-mediated IL-23 expression in macrophages

Abstract

Background

Human papillomavirus type 16 (HPV16) positive cancers have a tumor environment that induces antigen-presenting cells to increase IL-23 expression. Unclear is if HPV16 E6/E7 oncoproteins expressed in these cancers play a role in upregulating interleukin (IL)-23 in the tumor microenvironment (TME), and how this cytokine impacts the antitumor cytotoxic T-cell response in HPV16+ cancer.

Methods

CD8 T-cells targeting HPV16+ cancer cells were isolated from C57BL/6 mice bearing HPV16+ C3.43 tumors that were therapeutically vaccinated against HPV16 E6/E7 and incubated with IL-23. These T-cells were then co-incubated with HPV16+ target cells in a cytotoxicity assay to assess their cytolytic capacity. Additionally, carboxyfluorescein succinimidyl ester (CFSE) labeled T-cells were used to track the effect of IL-23 on their proliferation. The effect of IL-23 neutralization on vaccine-induced antitumor immunity during tumor progression was studied in vivo to assess its potential as either a standalone treatment or combined with a vaccine targeting HPV16 E6/E7. HPV16− tumors were engineered to express E6/E7 to find out if these oncoproteins upregulate IL-23. To understand how HPV oncoproteins in the TME affect transcriptional regulation of IL-23 producing cells, we used single-cell Assay for Transposase-Accessible Chromatin (ATAC)+RNA sequencing.

Results

Inside macrophages residing in the HPV+ TME, transcription factor enrichment and linkage analysis identified KLF2 as a potential regulator of Il23a. Overexpression of KLF2 in macrophages upregulates IL-23 production. CD8 T-cells that recognize HPV16+ cells incubated with IL-23 are inhibited in both their killing and proliferative capacities. IL-23 neutralization increased the presence of HPV-specific cytotoxic CD8 T-cells inside the HPV16+TME in an IL-17 independent manner. Combination of IL-23 neutralization followed by HPV16 E6/E7 vaccination increases survival by amplifying the anti-tumor immune response.

Conclusion

This study finds that the presence of HPV oncoproteins in tumor cells increases KLF2 expression in tumor-associated macrophages in vivo. It also shows that KLF2 upregulates IL-23 production in M2 macrophages, resulting in increased IL-23 levels in the TME. In addition, it is shown that elevated levels of IL-23 suppress the antitumor immune response and that IL-23 neutralization synergizes with therapeutic vaccination against HPV oncoproteins.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Human papillomavirus (HPV)+ cancers cause antigen-presenting cells to increase their production of interleukin (IL)-23 in vitro.

• IL-23 can increase T-regulatory cell-mediated immune suppression.

WHAT THIS STUDY ADDS

• This study shows that tumor cells expressing HPV E6 and E7 oncoproteins are involved in upregulating KLF2 in tumor-associated macrophages, thereby stimulating these cells to increase IL-23 production.

• This study provides in vivo insight into the suppressive effects of IL-23 on the antitumor immune response.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• This research shows that therapeutic vaccination strategies for patients with HPV+ cancer could potentially be enhanced by neutralizing IL-23.

Methods and materials

The Cancer Genome Atlas

Using UCSCXenaTools in RStudio, RNA expression levels (log2(norm count+1)) of IL17A, IL1B, IL6, TGFB1, and IL23A were obtained from bulk RNA sequencing of cancers available in The Cancer Genome Atlas (TCGA). The head and neck squamous cell carcinoma (HNSCC) group consisted of 604 patients of whom the human papillomavirus (HPV) status was determined for 124 patients by HPV in situ hybridization (ISH), or in absence of ISH test result, p16 status. International Classification of Diseases 10th edition determined anatomical site. Results were obtained for 84 patients with HPV− and 40 patients with HPV+ HNSCC. Average expression and the SEM were calculated for each gene and each cancer.

Animals and tumor models

C57BL/6 pathogen-free female mice (Taconic Biosciences, Germantown, New York, USA) were treated according to the Institutional Animal Care and Use Committee protocol (20065) approved by the University of Southern California. C3.43,^{1 2} B16-HPV and B16Green Fluorescent Protein (B16-GFP) (1×10⁵ cells) were injected subcutaneously on the right flank. Only mice with tumors >1000 mm³ were included in the survival study if requiring euthanasia unrelated to treatment. Mouse neutralizing antibodies used: isotype (BE0088), IL-17A (BE0173), and IL-23p19 (BE0313) (Bio X Cell, Lebanon, New Hampshire, USA). Mice received antibodies for 3 days (200 μg/dose, intraperitoneally (i.p.)). Tumor challenge occurred on day 4, with mice receiving maintenance dosing of antibodies (50 μg/dose, i.p.) two times a week until euthanasia. Vaccinations were peritumorous (1×10⁷ infectious units of Venezuelan equine encephalitis replicon particle (VRP)-HPV or VRP-GFP (subcutaneously)³ on indicated days. Retro-orbital bleeds were obtained under isoflurane, and heparin was added to blood to prevent clotting. Red blood cell contamination was removed using ACK (Lonza, Walkersville, Maryland, USA).

T-cell assays

CD8 T-cells were obtained from C3.43 tumor-bearing mice that were vaccinated on days 14 and 21 with VRP-HPV or VRP-GFP. On day 23, tumors were dissociated using a tumor dissociation kit (Miltenyi, Gaithersburg, Maryland, USA). Positive Selection for CD8a (StemCell, Cambridge, Massachusetts, USA, #18953) was used to obtain CD8 T-cells according to manufacturer instructions. Cells were cultured with Cytotoxic T Lymphocyte (CTL) media containing Roswell Park Memorial Institute 1640 (RPMI1640) media supplemented with 10% Fetal Bovine Serum (FBS), 1 mM Sodium Pyruvate, 1x β-mercaptoethanol, and 1× non-essential amino acids, to which 50 U/mL of recombinant mouse IL-2 (rmIL-2) (BioLegend, San Diego, California, USA) was added for 5 days in the presence or absence of 100 ng/mL recombinant mouse IL-23 (rmIL-23) (BioLegend). Cells were subsequently co-incubated for the indicated time with C3.43-gluc (gluc lentivirus was used to transduce C3.43⁴) at varying effector to target (E:T) ratios. 15 µL of 100 µM coelenterazine substrate (Nanolight Technology, Norman, Oklahoma, USA, #303–1) was added and luminescence was measured on ClarioSTAR Plus Platereader. Relative luminescence units (RLU) were used as a measure of cytotoxicity. Background RLU was determined as the average luminescence from wells with only C3.43-gluc cells. The average RLU from the highest E:T ratio in media-treated CD8 T-cells was taken as max cytotoxicity. To calculate individual cytotoxicity values, we used the formula: (Observed RLU − Background RLU) / (Max RLU − Background RLU) × 100. For the carboxyfluorescein succinimidyl ester (CFSE) proliferation assay, cells were stained at 10⁷ cells/mL in 30 µL of 5 µM CFSE for 15 min at 37°C and quenched with 500 µL CTL media. Cells were spun down and washed with fresh media, prior to further culture at standard conditions for 3 days. After 24 hours, cells were used to set up compensations for flow cytometry on BD FACSCanto II.

Establishing B16 HPV tumor model

The HPV16 E6/E7 transfection plasmid was obtained by cloning the mouse mouse phosphoglycerate kinase (mPGK) promoter, HPV16 E6, P2A, HPV16 E7, IRES, and GFP into pRosa26-1 (Plasmid #21714, AddGene, Watertown, Massachusetts, USA) linearized with XbaI (#FD0684, Thermo Fisher Scientific, Waltham, Massachusetts, USA). Using primers (online supplemental STable 1) to PCR amplify the gene sequences (Integrated DNA Technologies, Newark, New Jersey, USA), the In-Fusion Snap Assembly Starter Bundle (Takara-Bio, San Jose, California, USA) was used to ligate six insert sequences into pRosa26-1. The plasmid sequence was confirmed by sequencing (Azenta, Burlington, Massachusetts, USA). The same process was used for the GFP control plasmid by cloning mPGK and GFP into pRosa26-1 instead. B16 cells were incubated with ribonuclear protein (0.5 uL Cas9 (Thermo Fisher Scientific) + 10 uL guide RNA (gRNA) ACUCCAGUCUUUCUAGAAGA (Synthego, Redwood City, California, USA), prior to electroporation using Cell Line Nucleofector Kit V running Amaxa Nucleofector 2 on the B16 setting. These cells were subsequently incubated with 2 μg of SacI (#FD1133, Thermo Fisher Scientific) linearized plasmid prior to repeating electroporation. Cells were sorted for GFP expression using the BD FACSAria II after 5 days. The messenger RNA (mRNA) of single-cell B16 clones was isolated (RNEasy Plus Kit, Qiagen, Germantown, Maryland, USA) and checked for expression of HPV16 E6/E7 expression by reverse transcription-polymerase chain reaction (#95 087–200, Quantabio, Beverly, Massachusetts, USA) using E6 and E7 primer pairs (online supplemental STable 1) and compared with mRNA from B6MEC2, C3.43, and TC-1. C57BL/6 mouse embryonic cells (B6MEC2), used to generate C3.43,^{1 2} were used to normalize expression among the different cell lines, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as internal control.

Protein lysates were obtained using Pierce IP lysis buffer supplemented with Halt Protease Inhibitor Cocktail (Thermo Fisher Scientific). Protein concentration was determined using Bradford protein dye (BioRad, Hercules, California, USA). 100 µg of protein was run using Invitrogen reagents (online supplemental STable 1). Briefly, samples were boiled, run on a 10% bis-tris gel and transferred to a nitrocellulose membrane. Membranes were blocked for 1 hour in 5% (weight/volume) non-fat dry milk (NFDM) in Tris-buffered saline with 0.1% Tween-20 (TBS-T) at room temperature (RT), and stained with HPV16 E6 (GTX132686, Genetex, Irvine, California, USA), HPV16 E7 (GTX133411, Genetex), or β-actin (Cell Signaling Technologies, Danvers, Massachusetts, USA) overnight at 4C in 4% NFDM in TBS-T. Blots were washed three times the next day with TBS-T for 5 min each and incubated with secondary antibody (IRDye, LI-COR, Lincoln, Nebraska, USA) in 4% NFDM in TBS-T for 1 hour RT. Blots were imaged and analyzed using the LI-COR Odyssey DLx Imager and its Image Studio software (LI-COR).

IL-23 quantitation and KLF2 overexpression

To establish the protein concentration of interleukin (IL)-23 in B16 HPV+ and HPV− tumor models, tumors were harvested after reaching a minimum size of 500 mm³ and sent to Raybiotech (Peachtree Corners, Georgia, USA) for analysis using Single Molecule Array (Simoa) protein analysis. Each tumor model was grown until the average size was>1000 mm³ and resected. These tumors were subsequently frozen in liquid nitrogen and sent out for IL-23 quantitation.

A lentiviral vector expressing mKLF2 under an EF1 promoter (VectorBuilder, Chicago, Illinois, USA) was used to overexpress KLF2 in RAW264.7 macrophages. Overexpression was confirmed using mCherry overexpression. To confirm KLF2 overexpression and measure its effects on IL-23 production and secretion, 1×10⁶ RAW^WT or RAW^KLF2 cells/well were cultured in Dulbecco’s Modified Eagle Medium (DMEM) with 10% FBS and incubated for 72 hours. RNA was collected from the cells and analyzed for KLF-2 and IL-23 expression by quantitative PCR (qPCR) (primers online supplemental STable 1) and conditioned media was analyzed for IL-23 content through LEGENDplex mouse macrophage 13-plex (BioLegend) according to manufacturer protocol.

Flow cytometry/FACS sorting

Lympholyte-M (Cedarlane Laboratories, Burlington, Canada) was used to remove dead cells, tumor cells, and tumor debris from single cell tumor suspensions. For antigen-presenting cell analysis, these cells were then incubated with 100 ng/mL of lipopolysaccharide (LPS) for 8 hours at which point Brefeldin A was added and incubated for an additional 16 hours at 37°C. Cells were then harvested, blocked with FcX, stained for 25 min on ice (stain panels, online supplemental STable 1) and washed twice in fluorescence-activated cell sorting (FACS) buffer (phosphate-buffered saline (PBS) with 2% FBS and 0.01% NaN3). Intracellular epitopes of cells were made accessible (BD, #554714) and were blocked with rat serum (StemCell) prior to stain for 30 min on ice (stain panels online supplemental STable 1). Cells were analyzed on BD Canto after two washes with perm buffer and one wash with FACS buffer. For T-cell analysis, cells were stimulated with 20 ng/mL phorbol 12-myristate 13-acetate (PMA) and 500 ng/mL of ionomycin for 1 hour and brefeldin A for an additional 5 hours before going through the same staining process as macrophages. Biotinylated H-2DB monomers containing HPV16 E7 (49–57, RAHYNIVTF), obtained from the National Institute of Allergy and Infectious Diseases Tetramer Facility (Atlanta, Georgia, USA), were tetramerized using streptavidin-phycoerythrin. Gating was established through fluorescence minus one using FlowJo.

For multiome (Assay for Transposase-Accessible Chromatin (ATAC)+ RNA) single-cell sequencing, macrophages (CD11b+, F4/80+) and dendritic cells (F4/80−, CD11c+) were harvested through cell sorting on the BD Aria II. For each tumor model, tumors were grown in five mice until an average size of >1000 mm³. Tumors were resected, dissociated (Miltenyi), and combined prior to immunostaining. On ice, cells were blocked using mouse FcX for 10 min, and stained (staining panel in online supplemental STable 1) for 20 min with fluorochrome antibodies in the dark. Macrophages (CD11b+, F4/80+) and dendritic cells (DCs)(F4/80−, CD11c+) were sorted using BD FACSAria II. Approximately 1×10⁵ live macrophages and 0.5×10⁵ live DCs were obtained and combined for each tumor model.

Multiome single-cell RNA + ATAC sequencing

Antigen-presenting cells (APCs) from HPV+ and HPV− tumors were obtained as described under FACS sorting. Their nuclei were isolated according to 10x Genomics (Pleasanton, California, USA) nuclei isolation protocol. Optimal lysis conditions were determined as 0.5X Lysis Buffer for 2 min. Nuclei were visually quality controlled under a microscope and on a Countess II. HPV− (10,000) and HPV+ (9,000) tumor-derived nuclei were processed on a Chromium Next GEM Chip J 4rxn using a Chromium X Series Controller. Libraries QC was done on an Agilent D1000 TapeStation. Gene expression (GEX) libraries (RNA) were run on SP 200 cycles NovaSeq 6000 and ATAC libraries were run on P3 200 cycles NextSeq 2000 (Illumina). Cellranger-arc-2.0.1 was used to align and process fastq files. Both ATAC and GEX data were loaded into Seurat V.4 objects. Gene expression was normalized using SCTransform (SCT) and ATAC data was processed using Signac (reference genome mm10). Transcription factor (TF) binding motifs were annotated in accessible chromatin regions using AddMotifs, prior to running ChromVAR to compute motif deviation scores per motif per cell. Motif position frequency matrices were obtained from JASPAR2024 (vertebrates). Seurat objects were integrated using SCT-based integration, and dimensionality reduction was performed using principal component analysis (PCA) for RNA and Latent Semantic Indexing (LSI) for ATAC. A multimodal weighted nearest neighbor graph was constructed by incorporating PCA and LSI reductions, and clustering on this graph was performed using the Leiden algorithm. FindAllMarkers was used to identify what genes define each cluster. Cells were grouped by Il23a expression status. More narrowly defined open chromatin peaks were called using standard parameters of Signac CallPeaks and MACS2 grouped by HPV status. FindMarkers was used on the peaks assay, grouped by IL-23 status to identify genomic regions associated with IL-23 expression, and FindMotifs was used to identify TFs able to bind there with significant confidence. Candidate TFs were filtered to only include those of which motif enrichment was more than twofold compared with background and with a p value <0.05, expressed in >60% of Il23a+ cells, and had a presto area under the curve (AUC) >0.5. LinkPeaks was used to identify enhancer regions correlated to Il23a expression on chr 10 where the Il23a gene body is located. TFBSTools was used to analyze the genomic region of interest for TF binding sites which were again filtered for >60% of expression in Il23a+ cells and with an AUC score >0.5. ChromVAR was used to compare motif activity scores between HPV+ and HPV− APCs.

Statistics

Prism was used for statistical analysis and graphical representation of data. The mRNA expression of cytokines in HNSCC tumors was compared between HPV+ and HPV− tumors using a one-sided t-test, testing the null hypothesis that HPV does not increase indicated cytokine expression. For comparison of IL-23R expression and CFSE mean fluorescence intensity (MFI) on CD8 T-cells, a two-sided t-test was used. Cytotoxicity between media and IL-23 treated cells was compared using two-way analysis of variance (ANOVA). Comparing tumor growth curves was done through Holm-Sidak’s Multiple Comparisons analysis, and survival using Log-rank (Mantel-Cox). Comparison between cell populations obtained through flow cytometry on tumors from multiple treatment groups was done by two-way ANOVA. Simoa protein concentrations were analyzed using a one-tailed unpaired t-test. Mean cell frequencies and average expression (MFI) between APCs derived from HPV± tumor models were analyzed using one-sided t-test. RNA and Motif analysis in multiome data used Mann-Whitney U test. Macrophage RNA expression of Klf2 and Il23a expression was done using a two-way ANOVA. IL-23 protein expression in macrophages was analyzed using a two-tailed unpaired t-test.

Background

HPV infects tissues found in the cervix and oropharynx, where integration of its oncogenes E6 and E7 into the host genome can lead to cancer.^{5 6} It is the cause of cancer-related death in over 300,000 people globally every year.7,9 Over 90% of cervical cancer is caused by HPV,¹⁰ while the population of HNSCC is comprised of either HPV− or HPV+ tumors.

High-risk HPVs (hr-HPVs), the subtypes able to cause cancer, express the early (E)6 and E7 proteins to ensure that the cell cycle proteins needed for viral genome replication continue to be expressed in cells which no longer divide in the supra-basal layer of the stratified epithelium under healthy conditions.⁵ The E6 and E7 oncoproteins primarily achieve this through a combination of E6 binding the ubiquitin ligase E6 associated protein (E6-AP), which targets P53 for ubiquitin-mediated degradation, and E7 manipulating the retinoblastoma protein and others in the centrosome cycle to ensure continued cell cycle progression.^{5 11 12} This combination causes unchecked proliferation of the host cells and allows for accumulation of mutations throughout the host genome in the absence of immune-mediated clearance. This manifests itself clinically as neoplasia and is referred to as squamous intraepithelial lesions (SIL), classified as either low or high grade (LSIL and HSIL), which can progress to cancer if left untreated.^{13 14} Progression of chronic HPV infections to LSIL, HSIL, and ultimately cancer is accompanied by a progressively higher frequency of HPV integration into the host genome.15,18 In the majority of HPV+ cancers, HPV loses its replicative ability on incorporation into the DNA, but expression of oncogenic HPV E6/E7 proteins persists.^{5 18}

HPV type 16 (HPV16) is the subtype found in most HPV+ cancers^{19 20} and its E6/E7 proteins were found to have functions beyond increasing unchecked proliferation and survival. HPV E6 is able to bind the PSD-95, Dlg, and ZO-1 (PDZ) domain and α-helix domain of proteins,21,23 while there is a wide variety of proteins found to be affected by the presence of HPV E7, including DNA methyltransferases, histone acyl-transferases, histone deacetylases, and histone methyl transferases.^{23 24} Through altering the host transcriptome, these oncoproteins can also dampen the host immune response against HPV.25,28 Among the many impacts of E6/E7 on the host cell during the infectious lifecycle of HPV are their ability to modulate uclear factor kappa B cells (NF-κB) signaling, downregulate toll-like receptor 9 (TLR-9) expression, and reduce major histocompatibility complex class I (MHC I) expression.25,28 Progression of HPV-driven disease occurs in patients failing to mount a successful HPV-specific adaptive immune response, but how HPV oncoproteins affect the adaptive immune response continues to be an area of active investigation.

The immune modulatory mechanisms for evading the host immune response through HPV16 E6/E7 during initial viral infection remain present in the tumors that rely on continued expression of these oncoproteins.29,31 Therapeutic vaccination strategies targeting HPV16 E6/E7 in HPV16+ murine tumor models are highly effective in treating early-stage tumors.^{3 32} However, this therapeutic effect is gone or diminished in late-stage tumors.³³ Understanding which immune modulatory pathways continue to suppress the antitumor immune response through HPV16 E6/E7 can be uncovered through identifying enriched signaling pathways in HPV16+ cancers. Pathway analysis of the environment of HPV+ cervical cancer tissue shows that the IL-17 signaling pathway is among the most upregulated pathways in cervical cancer tissue as compared with healthy control tissue.³⁴ This pathway is triggered through the IL-17 family of cytokines known for signaling through IL-17A (IL-17) and IL-17F.35,37 How native IL-17 signaling impacts survival in patients with HPV+ cancer and the role that HPV oncoproteins play in its expression remains unclear. High levels of IL-17+ cells are generally correlated to poor survival,³⁸ while conflicting results exist for the effect of increased IL-17+ cells in the tumor microenvironment (TME) of patients with HPV+ cervical and HNSCC.39,41 Overexpression of IL-17 suppresses the HPV-specific immune response,⁴² but even though the IL-17-producing T-helper cells (Th17) become the dominant circulating CD4 T-cell subset in the TME of HPV+ cancer, the increase of the IL-17 cytokine inside the serum and the HPV+ tumor environment is modest.^{43 44}

What stands out in patients with HPV+ cancer is that IL-23, the primary maturation and expansion factor of IL-17 producing cells, is the most substantially upregulated cytokine in the blood.⁴⁴ IL-23, made up out of an IL-12p40 and IL-23p19 subunit, is secreted by APCs.^{45 46} Early in vitro work investigating the regulation of IL-23 in the HPV+ tumor microenvironment suggests this happens in an IL-6 and IL-1β dependent manner, and that HPV+ tumors induce IL-23 in APCs through fibroblasts.⁴⁷ Little is known about the effects of IL-23 and downstream signaling on the naturally occurring antitumor immune response inside HPV+ tumors. In murine skin cancer models, inhibition of IL-23 signaling has been found to increase T regulatory cell-mediated immune suppression and reduce CD8 T-cell presence in the tumor immune microenvironment.48,50

The reliance of HPV+ cancer cells on the E6/E7 oncoproteins for survival has hampered the ability to study their role in modulating IL-23 producing cells in vivo. This highlights the need for in vivo modeling to elucidate the effect of IL-23 on the HPV-specific antitumor immune response and to investigate if HPV oncoproteins drive APCs to produce IL-23. Because IL-23 expression is able to interfere with CD8 T-cell functionality and is upregulated in HPV16 E6/E7 expressed cancers, we hypothesized that these oncoproteins induce IL-23 expression to suppress the antitumor immune response. The murine C3.43 tumor model was previously established by our group through transforming B6 mouse embryonic fibroblast (B6MEF) with the full-length HPV16 viral genome under its native promoters.^{1 2} In this study, it was used along with E6/E7 expressing VRPs as a therapeutic vaccine to further dissect the role of IL-23 in suppressing the HPV-specific immune response and its effect on HPV16+ tumor growth.^{3 32} In addition, new tumor models were established to study the involvement of HPV16 E6/E7 in driving IL-23 production.

Results

To assess if IL-17 expression in HPV+ tumors is not only associated with IL-23, but also drives their respective expression, the publicly available transcriptomic TCGA data on HPV+ and HPV− HNSCC tumor samples was analyzed. Of the IL-17-polarizing factors IL-1β, IL-6, Transforming Growth Factor beta (TGF-β) and the expansion factor IL-23, only RNA for IL-23p19 (IL23A) and downstream IL-17 (IL17A) were significantly elevated in the HPV+ HNSCC tumors as compared with HPV- HNSCC tumors (figure 1A). On stratifying the patients with HNSCC by anatomical origin of the cancer, this same trend could be observed in the oropharyngeal subset of patients with HNSCC (online supplemental SFigure 1). The significance of these findings is highlighted when analyzing the expression of these cytokines among the 36 most common cancers in the USA. Both HNSCC and cervical cancer are among the cancers that contain the highest levels of IL17A and IL23A RNA (online supplemental SFigure 2). But when separating patients with HNSCC by their HPV status, HPV+ HNSCC rank as the highest expressors of both IL17A and IL23A, correlating HPV status to increased IL-23 and IL-17 expression in human cancers (figure 1B). These data suggest that the presence of HPV during virus-induced transformation could play a role in modulating the tumor environment to promote IL-23 production and therefore enhance expansion of IL-17-producing cells.

Figure 1. IL-23, a cytokine highly expressed in HPV+ cancer, reduces HPV16 E7 specific CD8 T-cell expansion and cytotoxicity. Patients with HNSCC from TCGA were stratified by known HPV status and their mRNA expression of (A) IL17A and the IL-17 polarizing factors IL6, IL1B, TGFB1, and IL23A were compared between HPV+ and HPV− HNSCC. (B) Mean expression of IL23A and IL17A mRNA of all TCGA cancers were ranked by their IL17A expression. (C) IL-23R expression of HPV-specific CD8 T-cells obtained from VRP-HPV (HPV+) and VRP-GFP (HPV−) vaccinated C3.43 tumor-bearing mice. (D) HPV16 E7-specific CD8 T-cells were incubated with 100 ng/mL of IL-23 for 5 days and characterized for HPV specificity and IFN-γ/IL-17 expression. (E) Effects of IL-23 on proliferation of HPV specific CD8 T-cells were measured through CFSE analysis, where MFI is inversely correlated to proliferation rate. (F) A cytotoxicity assay was performed for 6 hours at various effector (CD8 T-cells) to target (C3.43-gluc) ratios (E:T) and luminescence was compared between media control and IL-23 (100 ng/mL) treated CD8 T-cells. Shown are either boxplot or mean (±SEM). Functional T-cell based experiments were repeated three times with similar results. *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001. HNSCC, head and neck squamous cell carcinoma; HPV, human papillomavirus; IL, interleukin; TGF, Transforming Growth Factor; CFSE, Carboxyfluorescein Succinimidyl Ester; GFP, Green Fluorescent Protein; IFN, Interferon; MFI, mean fluorescence intensity; mRNA, messenger RNA; TCGA, The Cancer Genome Atlas; VRP, Venezuelan equine encephalitis replicon particle.

The HPV16+C3.43 tumor model was used to assess if IL-23 can directly interfere with the HPV-specific antitumor immune response. CD8 T-cells with specificity for HPV16 E7 were obtained by therapeutically vaccinating mice bearing C3.43 tumors with HPV16 E6/E7 VRPs (VRP-HPV).^{3 26 32} CD8 T-cells from VRP-HPV vaccinated mice were harvested from tumor-bearing mice and confirmed for HPV16 E7 specificity by flow cytometry using Major Histocompatibility Complex (MHC) tetramer staining. In the process of confirming these cells to be IL-23 receptor (IL-23R) positive, we observed that the majority of CD8 T-cells from VRP-HPV vaccinated mice express IL-23R and that its expression was more frequent (>60%) when compared with CD8 T-cells obtained from VRP-GFP vaccinated mice (<40%, figure 1C). HPV16-specific T-cells were incubated with IL-23 for 5 days in complete culture medium, followed by functional assessment for cytokine secretion to determine the effect of IL-23 signaling on their functionality. While the E7-tetramer positive cell population did not change between exposure to media alone or IL-23 (figure 1D, top panels), the percentage of tetramer-positive T-cells able to produce interferon gamma (IFN-γ) was significantly reduced when exposed to IL-23 (figure 1D, bottom panels), suggesting that IL-23 in the TME could suppress antitumor activity of HPV-specific T-cells.

Next, the HPV-specific T-cell proliferation was assessed using CD8 T-cells isolated from vaccinated mice. CFSE-labeled CD8 T-cells were incubated for 3 days in the presence of either media or media+IL-23 before analysis of CFSE MFI throughflow cytometry. This resulted in a significantly higher CFSE MFI signal in IL-23 treated CD8 T-cells, indicating a reduction in cell division, compared with control treated cells that exhibited a baseline level of proliferation (figure 1E). This suggests IL-23 suppresses proliferation of CD8 T-cells that target HPV+ cells.

To determine if the cytotoxic activity of HPV-specific T-cells is impacted by the presence of IL-23 in the media, Gaussia luciferase (gluc) expressing C3.43 cells (gluc-C3.43) were created.⁴ Positive selection was used to enrich CD8 T-cells, which did not inhibit their cytotoxicity as shown by the successful release of luciferase on killing gluc-C3.43 (online supplemental SFigure 3). The cytotoxicity of IL-23-polarized CD8 T-cells was significantly reduced as compared with their non-polarized counterparts (figure 1F), suggesting that in addition to reducing proliferative potential and cytokine production from virus-specific T-cells, IL-23 also reduced their cytotoxic potential.

The above data suggests that inhibition of IL-23 signaling could be a therapeutic strategy to enhance natural and/or vaccine-induced immunity against HPV+ cancers. Therefore, the significance of neutralizing IL-23 or IL-17 on HPV+ tumor progression was determined using C3.43 tumors to test their effect on tumor progression. C57BL/6 mice were treated with IL-23 or IL-17 neutralizing antibodies and subsequently implanted with C3.43 tumor cells (figure 2A). We found that both cytokines can modulate HPV+ tumor growth, as blocking of both IL-17 or IL-23 signaling showed tumor growth inhibition (figure 2B, online supplemental SFigure 4) leading to a significantly improved median survival (figure 2C). To further understand what effect neutralizing IL-17 or IL-23 treatments had on the TME, intratumoral T-cell and natural killer (NK) cell populations were analyzed. Only IL-23 depletion caused more HPV-specific cytotoxic T cells to be present in the tumor (figure 2D), while the circulating population of these cells remained unaffected (figure 2E). IL-23 depletion successfully reduced IL-17 expressing cell populations (figure 2F), but did not significantly reduce neutrophils inside the tumor (online supplemental SFigure 5). NK, CD4 regulatory T-cells, and overall CD4 T-cell and CD8 T-cell levels remained unaffected (figure 2G,H,I online supplemental SFigure 5). Because no significant changes to the immune environment were observed when IL-17 is neutralized, C3.43 cancer cells were analyzed and shown to express IL-17RA suggesting that they can sense IL-17 in their environment (online supplemental SFigure 6).

Figure 2. Neutralizing IL-23 in HPV+ cancers reduces C3.43 tumor growth and increases HPV16 specific CD8 T-cells. (A) Experimental outline of prophylactic depletion of IL-17 and IL-23 in C57BL/6 mice bearing C3.43 HPV16+ tumors. (B) Tumor growth and (C) survival of C3.43 tumor bearing mice that continued to receive treatment with vehicle control (n=10) or either anti-IL-17 (n=10) or anti-IL-23 (n=10) antibodies dosed two times a week (50 μg/dose). (D) TILs were harvested on the day of euthanasia and analyzed by flow cytometry for HPV16 E7 tetramer+ CD8 T-cell frequency. (E) Retro-orbital bleeds were taken 14 days post-tumor challenge and analyzed for HPV16 E7 tetramer+ CD8 T-cells. (F) Other TIL frequencies measured by flow cytometry were IL-17A positive cells, (G) CD8a+ CD3+ cells, (H) CD4+ cells, and (I) T-reg cell populations. Mice received treatment until euthanasia. For flow cytometry of TILs, n ranged between 7 and 10 mice/treatment/stain. Error bars are SEM. *p<0.05; **p<0.01; ****p<0.0005; n.s., not significant. Ctrl, control; HPV, human papillomavirus; IL, interleukin; i.p., intraperitoneally; TIL, tumor-infiltrating leukocyte; T-reg, regulatory T-cells; Foxp3, forkhead box P3.

Because treatment with IL-23 neutralizing antibodies increased the presence of HPV specific T-cells inside the HPV+ tumor, while IL-23 signaling was shown to contribute to a decrease in both their killing and proliferation capacity, the ability of IL-23 neutralizing antibodies to enhance VRP-HPV therapeutic vaccine efficacy was tested on late-stage HPV+ tumors. C3.43 tumor-bearing mice were treated, starting on day 14 post-tumor challenge, with IL-23 neutralizing antibodies for three consecutive days. On the third day of monoclonal antibody treatment, these mice received their first of two VRP vaccinations given 1 week apart (figure 3A). Isotype antibodies and VRP-GFP were used to control for effects on the immune system unrelated to the neutralization of IL-23 and assess if only the HPV-specific immune response is affected. Standalone IL-23 neutralization slightly improved survival, similar to what is observed with VRP-HPV vaccination alone, but it is only when these two treatments are combined that we observe synergistic effects in reducing the rate of tumor growth (figure 3B) and improved survival (figure 3C). VRP-HPV successfully enhanced HPV-specific T-cell infiltration with HPV VRPs alone (isotype+VRP HPV group), but infiltration was boosted when the VRP-HPV were combined with IL-23 neutralizing antibodies (figure 3D). Although no statistical significance was observed in therapeutic effects of IL-23 neutralization alone on tumor growth, 3/10 mice were still able to mount a successful HPV-specific T-cell response inside the tumor. Cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) expression on CD4 and CD8 T-cells was evaluated because it is a known inhibitor of T-cell expansion. Although IL-23 neutralization alone did not reduce CTLA-4 expression on CD4 and CD8 T-cells, a significant downregulation of CTLA-4+ CD4 and CD8 T-cells in the combination therapy group was observed (figure 3E,F).

Figure 3. Combination of IL-23 neutralization and HPV16 E6/E7 vaccination leads to synergistic improvement of survival of C3.43 tumor-bearing mice, and improved HPV16 E7-specific CD8 T-cell tumor infiltration. (A) Schematic overview of B6 mice bearing C3.43 tumors (Day 0, n=10/group) that were treated with IL-23 neutralizing antibodies or isotype controls (Day 14–16) in combination with VRP-GFP or VRP-HPV (Day 16+23). The impact of treatment on (B) tumor growth and on (C) survival of tumor-bearing mice. TILs were analyzed throughflow cytometry to determine CD8 T-cell positive for (D) HPV E7 tetramer+ and (E) CTLA-4. (F) TIL-derived frequency of CD4 T-cells positive for CTLA-4. Error bars represent SEM. For flow cytometry, n ranged between 4 and 10 mice/treatment/stain. The experiment was repeated twice with similar results. $/#/*=p<0.05; $$/##/**=p<0.01; $$$/###/***=p<0.001; $$$$/####/****=p<0.0001. GFP, green fluorescent protein; CTLA-4, cytotoxic T-lymphocyte-associated protein 4; Ab, antibody; Ctrl, control; HPV, human papillomavirus; IL, interleukin; i.p., intraperitoneally; TIL, tumor-infiltrating leukocyte; VRP, Venezuelan equine encephalitis replicon particle.

To investigate the role of HPV E6/E7 oncoproteins in modulating immune suppressive IL-23, a tumor model was established to study the effects of the viral oncoproteins on IL-17 and IL-23 production. Since knocking down HPV16 oncogenes in established HPV+ tumor models leads to a P53 induced senescence or cell death, an alternative tumor model for immune-competent mice was needed. To guide what cancer model was best suited to study the role of HPV16 oncoproteins in driving IL-23, TCGA tumors were ranked by their IL-23 RNA expression (online supplemental SFigure 7). The best candidate for our proposed model was skin cutaneous melanoma (SCM) because it expressed lower levels of IL-23 in comparison to HPV-driven cancers. This cancer is physiologically relevant because, like HPV+ tumors, it is found in stratified epithelium and would therefore grow under a similar immune microenvironment. In addition, the murine B16 melanoma tumor model used to study SCM has been used to investigate the effect of IL-17 producing T-cells.4248 50,53 An overexpression vector containing the HPV16 E6/E7 proteins separated by a purine 2A cleavage site was knocked into the rosa26 safe harbor site of B16 cells to create the HPV+ tumor B16-HPV, while an HPV− expression vector expressing only GFP was knocked in to serve as negative control to account for immunogenic effects of GFP termed B16-GFP (figure 4A). HPV16 E6/E7 expression in the knock-in model was lower in B16-HPV than C3.43 or TC-1 (figure 4B) while still expressing detectable protein by western blot (figure 4C). Comparing the RNA expression of B16-HPV to C3.43 and TC-1 shows HPV oncogene expression is not excessive and would therefore be physiologically relevant. Both B16-HPV and B16-GFP tumor models were able to grow as subcutaneous tumors in immunocompetent C57BL/6 mice, although HPV+ tumors demonstrated a significantly faster growth rate compared with HPV− B16 tumors (figure 4D). Detecting IL-23 in these cancer tissues confirmed that IL-23 was indeed upregulated in an HPV oncoprotein-dependent manner (figure 4E).

Figure 4. Establishing an HPV16 E6/E7 expression tumor model. (A) Expression cassettes used to knock in HPV16 E6/E7 oncogenes or a mock control under a mouse PGK promoter and with a GFP marker gene. The mRNA expression of oncoproteins was confirmed and compared with B6MEC2, C3.43, and TC-1 by (B) qPCR normalized for GAPDH. Protein expression was confirmed by (C) western blot normalized for β-actin. (D) B6 mice were challenged with either B16-GFP (n=3) or B16-HPV (n=3) to confirm tumor take and (E) IL-23 protein was detected using Simoa, normalized for protein content per gram of tissue. *p<0.05. mRNA, messenger RNA; qPCR, quantitative PCR; GFP, green fluorescent protein; mPGK, mouse phosphoglycerate kinase; TC-1, transformed C57BL/6 1 cells; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; HPV, human papillomavirus; IL, interleukin; Simoa, single molecule array;.

To further characterize the TME of B16 tumors expressing the E6/E7 proteins, tumor-infiltrating leukocytes (TILs) were isolated from established tumors and assessed for IL-23 expression. Flow cytometry revealed that the primary source of IL-23 is CD45+ immune cells, irrespective of HPV status (figure 5A). Further classification of these IL-23+ cells shows that the majority expressed the macrophage marker F4/80 and not CD11c, representative of DCs (figure 5B). This indicates that irrespective of tumor cell HPV status, macrophages are the primary IL-23 producing cells. In an effort to understand if the increased level of IL-23 was driven by an increase in overall F4/80+ macrophages present inside the TME, overall frequencies of macrophages in TIL were measured (figure 5C), but no significant increase in their frequency was observed in HPV+ cancers (figure 5D). Instead, our results show an increase in the frequency of IL-23+ macrophages inside HPV+ tumors (figure 5E), an effect not detected for CD11c+ DCs (online supplemental SFigure 8). Comparing IL-23 expression between HPV+ and HPV− tumor-derived macrophages shows that macrophages express significantly more IL-23 in HPV+ tumors as compared with HPV− tumors (figure 5F,G). These data suggest that HPV oncogenes can cause both significantly more IL-23 expression and an increase in the frequency of IL-23 expressing APCs inside the tumor. Staining CD45+ immune cells with anti-IL-17 showed that the presence of HPV oncoproteins also resulted in a significant increase in the frequency of IL-17 producing cells (figure 5H) and a larger percentage of Th17 (figure 5I). In confirming that the new HPV tumor model can receive IL-17 signals, a significant increase in IL-17RA expression was detected in the presence of HPV16. This could amplify signaling through the IL-17 pathway in HPV+ cancer cells (online supplemental SFigure 9).

Figure 5. HPV oncoproteins drive increased IL-23 expression in macrophages. TILs isolated from B16-HPV (HPV+) and B16-GFP (HPV−) were analyzed by flow cytometry for IL-23 expression. IL-23+ cells were stratified by (A) CD45+ expression. (B) CD45+, IL-23+ immune cells were stained with (B) macrophage marker F4/80 and DC marker CD11c. (C) Gating strategy for TILs using (D) macrophage marker F4/80 and (E) the frequency of their IL-23 expression. (F) IL-23 MFI for HPV=red, GFP control=green, representing expression of IL-23 in (G) F4/80+ macrophages from either tumor model. The effect of oncoproteins expression on IL-17 expression in (H) CD45+ immune cells and (I) CD4+T-cells (Th17). Shown are mean frequencies (±SEM). Data on APCs is aggregated from three independent tumor challenge experiments with B16-GFP n=11 and B16-HPV n=12 mice. *p<0.05; **p<0.01; ****p<0.0001. APCs, antigen-presenting cells; HPV, human papillomavirus; GFP, green fluorescent protein; DC, dendritic cell; IL, interleukin; MFI, mean fluorescence intensity; TIL, tumor-infiltrating leukocyte.

To investigate potential HPV-driven transcriptional regulation of Il23a in the antigen-presenting cells of HPV+ cancers, tumor-derived macrophages (CD11b+, F4/80+) and dendritic cells (F4/80−, CD11c+) were analyzed by multiome (RNA+ATAC) sequencing. The data from RNA and ATAC sequencing were combined into a single Uniform Manifold Approximation and Projection (UMAP) using a weighted nearest neighbor analysis (figure 6A). Mapping of IL-23 expression shows most of the identified cell clusters have some cells expressing Il23a mRNA, with most IL23a positive cells mapping to cluster 0 (figure 6B). Adgre1 (F4/80) macrophages, rather than Itgax (CD11c) DCs, contribute to IL-23 expression inside the tumor (figure 6C). There was no apparent difference observed in Il23a promoter occupation of Il23a+ versus Il23a− tumors (online supplemental SFigure 10), suggesting the gene is likely regulated through enhancers and/or repressors. Cluster 0 was found to be highly enriched for the canonical M2 macrophage marker Mrc1 (online supplemental SFigure 11). Analyzing Il23a+ cells for M1 and M2 markers further shows that these cells are indeed enriched and highly expressing M2 markers (figure 6D).

Figure 6. HPV+ tumors express IL-23 in M2 macrophages. Macrophages and DCs were harvested from HPV− and HPV+ tumors and processed for single cell RNA+ATAC sequencing. (A) Visualization of how RNA and ATAC sequencing results are reduced into a weighted nearest neighbor analysis (WNN). (B) Cells that are Il23a RNA positive (Il23a+) are shown in yellow, cluster 0 contour is outlined in white. (C) Of the Il23a+ cells, cells that are Itgax+ (CD11c) and/or Adgre1+ (F4/80) are shown in yellow. (D) Dot plot comparing macrophage M1 (Nfkb1, Tnf, Il6, STAT1) and M2 (Mrc1, Igf1, Stab1, F13a1, Dab2) between Il23a+ and Il23a− cells. HPV, human papillomavirus; IL, interleukin; ATAC, assay for transposase-accessible chromatin; UMAP, uniform manifold approximation and projection; DCs, dendritic cells.

The combination of ATAC and RNA sequencing allows for genome-wide regions of open chromatin to be identified and compared between HPV− and HPV+ tumor-derived APCs, while transcription factors known to bind the sequences enriched within these regions can be verified for their expression through RNA. To find chromatin accessible regions associated with IL23a expression, a differential accessibility analysis between IL23a+ and IL23a− cells was performed. TF motif enrichment analysis was applied to these regions to identify which TF motifs are significantly enriched more than twofold as compared with background open chromatin (figure 7A). The list of candidate TFs was narrowed down further by cross-referencing them with RNA level expression data to retain only those that are expressed in the majority of cells (>60%). An AUC analysis was then performed comparing IL23a+ and IL23a− cells to identify TFs whose RNA expression is also positively correlated with IL23a expression (AUC >0.5, figure 7B). The only candidate TFs with an AUC value >0.5 and expressed in over 60% of Il23a+ cells were KLF2 and KLF7. Comparing the expression and motif activity score of KLF2 and KLF7 between HPV+ and HPV− tumor-derived macrophages shows that both have a similarly significant motif activity score, while KLF2 RNA is more significantly upregulated in HPV+ tumors (figure 7C/D). Therefore, it is most likely that KLF2 functions as a transcriptional regulator of Il23a expression in an HPV-dependent manner.

Figure 7. KLF2 and KLF7 are upregulated in TAM in an HPV-dependent manner. (A) TFs that are significantly enriched in open chromatin regions associated with Il23a expression. Candidate TFs that were more than twofold enriched compared with background open chromatin are highlighted in green. (B) Candidate TFs positively enriched in RNA level expression of Il23a+ cells (AUC) and further narrowed down by those expressed in>60% of Il23a+ cells are indicated in green. (C/D) Violin plot comparing KLF2 and KLF7 expression and motif activity score in APCs derived from HPV− and HPV+ tumors. (E) Linkage plot showing an open chromatin peak associated with Il23a expression. (F) TFs able to bind potential distal enhancer were plotted by AUC and percent expressed in Il23a+ cells (green=AUC >0.5, expression in Il23a+ cells >60%. (G) Candidate TF binding sites in the linked region were identified by scanning the sequence with their position weight matrices, using a minimum match score threshold of 95%, with relative scores>0.97 marked in green. APCs, antigen-presenting cells; AUC, area under the curve; HPV, human papillomavirus; TF, transcription factor; TAM, tumor associated macrophage.

Most gene enhancers and silencers act in cis, functioning on the same chromosome as the gene they regulate. To identify potential Il23a regulatory regions, a chromatin linkage analysis was performed on chromosome 10, where Il23a is located. This approach evaluates whether chromatin accessibility at surrounding genomic regions correlates with Il23a expression across single cells. The analysis identified a single open chromatin region 16.6 Mb upstream of the Il23a promoter (chr10:111,684,990–111,685,244; figure 7E). Motif analysis of this peak cross-referenced with RNA levels expression data again shows KLF2 and KLF7 as candidate TFs (figure 7F), with KLF2 showing the highest predicted motif match score across all tested transcription factors (figure 7G).

To verify if KLF2 is able to upregulate IL-23 expression in macrophages, it was overexpressed in RAW264.7 macrophages in vitro. Klf2 overexpression in RAW macrophages (RAW^KLF2) was confirmed using qPCR (figure 8A), and comparison of IL-23 expression between RAW^WT and RAW^KLF clearly indicates that KLF2 upregulates IL-23 expression on both the RNA and protein level in macrophages (figure 8B).

Figure 8. KLF2 upregulated IL-23 in macrophages. (A) RAW264.7 mouse macrophages overexpressing KLF2 (RAW^KLF2) were compared with wild-type RAW (RAW^WT) for expression of KLF2 and IL-23 through qPCR. (B) Cells were grown for 72 hours, and media was analyzed for IL-23 protein concentration. Shown are mean expression values (±SEM). Data is from three biological repeats. *p<0.05; **p<0.01; ****p<0.0001. IL, interleukin; qPCR, quantitative PCR.

Discussion

HPV16 E6/E7 expression is often upregulated during the disease progression from infected tissue to cancer tissue.^{30 31} The effect of E6/E7 on immune evasion occurs both through modulating inflammatory cytokines such as IFN and IL-6,25,28 in addition to pushing unique mRNA and protein content into the extracellular vesicles to communicate with its environment.54,56 This shows that these viral oncoproteins in HPV+ cancer can shape the TME and its immune environment. In vitro results from HPV+ tumor models indicate that local mature APCs downregulate inflammatory IL-12 and instead produce IL-23.⁴⁷ IL-12 functions as an adjuvant to vaccination by expanding vaccine-induced T-cells and reducing their tolerization.^{57 58} While supplemental IL-12 is explored in clinical trials for treatment of HPV+ cancers,⁵⁹ the mechanism behind increased levels of IL-23 and the therapeutic value of targeting it in HPV+ tumors remains an understudied area of research.

The IL-17 signaling pathway is among the most upregulated pathways in HPV+ cervical cancer tissue as compared with healthy controls.³⁴ This is illustrated by HPV+ cervical tissue showing significantly elevated levels of IL-17+ cells as disease progresses from pre-cancerous cervical intraepithelial neoplasia to cervical cancer^{38 39} and Th17 cells becoming the dominant T-helper cell subset.⁴⁴ IL-17 production in human T-cells is guided by the polarizing factors IL-6 and TGF-β in vitro, but additional signals are required in vivo from IL-1β and IL-23 to mature and expand Th17 cells and increase their production of IL-17.⁶⁰ Therapeutic targeting of TGF-β has been explored in a phase II clinical trial to treat HPV-associated malignancies due to its pleiotropic immune suppressive effects beyond just the induction of IL-17 and is a promising treatment which induces higher levels of HPV-specific T-cells,⁶¹ which warrants its own investigation into potential combination with HPV therapeutic vaccines. More recent work has connected the finding that HPV16 E6 increased production of IL-6 in cancer cells⁶² to IL-23 production in APCs in vitro.⁴⁷ The unique content of extracellular vesicles produced by HPV E6E7+ cells could be the driving force behind this. It was therefore surprising to observe higher IL-6 expression in patients with HPV− HNSCC, while this did not translate into higher IL-23 expression in these patients (figure 1A). Our analysis of tumor-derived APCs uncovered another regulatory mechanism of IL-23 expression in HPV+ tumors, where presence of HPV oncoproteins in the cancer cells leads to increased expression of the transcription factor KLF2 in M2 macrophages, leading them to increase production of IL-23. Literature has shown that long-term P53 expression suppresses KLF2 expression.⁶³ With HPV+ cell-derived exosomes containing minor levels of HPV E6,⁵⁶ it is potentially the long-term continuous exposure to these vesicles in vivo that helps E6 to suppress P53 in tumor-associated macrophages as they do in endothelial cells.⁶³ This might induce the tumor-associated macrophages to express IL-23 in a KLF-dependent manner.

Next, the combined ability of HPV16 E6/E7 to affect the frequency and expression of IL-23 in APCs inside HPV+ tumors were assessed. APCs, in particular macrophages, have previously been described as the primary source of IL-23 in both murine B16 melanomas⁴⁸ and human colorectal cancer.⁵⁰ Our results are in line with these findings and show that macrophages are the primary IL-23 producing cell type. Our data now show that IL-23 from these tumor-associated macrophages can suppress the antitumor immune response and that HPV oncoproteins are helping to drive this (figure 5).

Literature shows splenic CD8 T-cells increase their expression of IL-17 in response to stimulation of IL-23 with no effects observed on IFN-γ expression.⁶⁴ No increase in IL-17 expression was observed in CD8 T-cells derived from vaccinated C3.43 and incubated with IL-23. Instead, a decrease in IFN-γ expression was detected (figure 1D), highlighting the importance of studying these cells in their physiological environment. Our results indicate that HPV-specific CD8 T cells induced by therapeutic vaccination upregulate IL-23R during their active killing stage (figure 1C). This upregulation is also found for negative regulators of T-cell activation such as programmed cell death protein 1 (PD-1) and CTLA-4.^{65 66} When stimulating these vaccine-induced T-cells with IL-23, they were reduced in their cytotoxic capacity, their ability to proliferate, and capacity to produce IFN-γ (figure 1F,G). This suggests IL-23R signaling functions as a checkpoint inhibitor.

Signaling in human CD8 downstream of IL-23R activates the transcription factor STAT3.^{46 67} Effects of STAT3 signaling in tumor resident CD8 T-cells are dependent on the source of the signaling, with IL-10 and IL-21 helping to terminally differentiate and enhancing the cytotoxic potential of CD8 T-cells. This is not observed for STAT3 signaling through the IL-6 receptor (IL-6R) or PD-1.68,70 Instead, IL-6R-activated signal transducer and activator of transcription 3 (STAT3) is reported to inhibit the ability of PD-1 to mediate suppression of T-cell effector functions.^{68 71} These observations show that the effects of STAT3 signaling are environment dependent and raise the importance of studying IL-23R signaling in TME-derived CD8 T-cells.

IL-17 overexpression in HPV16 E6/E7 overexpressing TC-1 mouse tumors has been shown to reduce the therapeutic effect of vaccine-induced HPV16 E7 specific CD8 T-cells by recruitment of neutrophils.⁴² Although neutralizing IL-17 in the HPV16 E6/E7+ C3.43 cancer model had a therapeutic benefit (figure 2B,C), no increased presence of CD8 T-cells (figure 2G) was observed. This could be due to the significantly higher expression of HPV oncoproteins in transformed C57BL/6 1 cells (TC-1) (figure 4B), leading to increased HPV-specific CD8 T-cell receptor (TCR) engagement. Unlike IL-17 neutralization, only when targeting IL-23 with antibodies was an increase in survival and an increase in HPV-specific CD8 T-cells observed (figure 2B,C). With overall CD8 T-cell numbers remaining unaffected by IL-23 neutralization (figure 2G), the increase in HPV-recognizing CD8 T-cells indicates that a limited HPV-specific immune response could be mounted (figure 2D). These data suggest that IL-23 suppresses the HPV-specific immune response in an IL-17-independent manner, even though IL-17-producing cells are reduced. IL-17 signaling may therefore have a more direct effect on tumor cells. In addition to effects on the immune system, IL-17 affects vascularization and can directly impact the rate of cell division.⁷² We observed HPV16 to increase IL-17RA expression (online supplemental SFigure 8). As the IL-17 receptor pathway is enriched in HPV+ cancer tissue, this provides an alternate explanation of how IL-17 neutralization has an impact on tumor progression even though this cytokine is not found to be substantially upregulated in the serum and TME of patients with HPV+ cancer.^{43 44} Lineage-committed Th17 cells have been described to produce IL-10 in the presence of IL-27, a cytokine related to IL-12.^{73 74} The presence and potential role of IL-27 in HPV+ cancers is unknown and could provide a better understanding of how Th17 impacts clinical outcome in patients with HPV+ cancer.

The combination of increased presence of HPV-specific CD8 T-cells when using IL-23 neutralizing antibodies in HPV+ cancer (figure 2D) and decreased cytotoxicity and proliferation of these T-cells in the presence of IL-23 (figure 1E,F) raised the question of whether these antibodies could be used therapeutically. HPV16 E6/E7 presents an attractive therapeutic vaccination target because HPV+ cancer cells rely on the continued expression of these foreign antigens for survival.²⁹ VRP-HPV are virus particles able to mount an immune response against the HPV16 E6/E7 proteins through induction of cytotoxic CD8 T-cells. These cells are effective in clearing early-stage HPV16+ tumors but unable to overcome the late-stage suppressive HPV+ TME.^{3 32 33} When IL-23 neutralization is combined with a therapeutic HPV16 E6/E7 vaccine, a significantly delayed tumor growth and increased survival of HPV+ tumor-bearing mice is observed (figure 3B,C). This large reduction in growth rate combined with an elevated presence of HPV-specific CD8 T-cells inside the tumor (figure 3D) and a reduction of CTLA-4-expressing T-cells (figure 3E,F) indicates that IL-23 neutralization synergizes with VRP-HPV through enhancing HPV16-specific CD8 T-cell proliferation and their cytotoxic ability. The mechanism through which this occurs requires further studies. Our results suggest IL-23 enhances CD8 T-cell exhaustion similar to IL-6R.⁶⁹ Alternatively, the possibility exists that chronic IL-23 signaling leads to terminal exhaustion of the CD8 T-cells, similar to what is observed in chronic IFNγ signaling.⁵²

Beyond IL-23R signaling in E7-specific CD8 T-cells, there are indirect pathways through which IL-23 neutralization can suppress the antitumor response. One such mechanism would be through disrupting the IL-23/IL-22 axis, which helps to maintain a pro-tumorigenic environment. IL-23 promotes IL-22 production in Th17 cells,^{75 76} which supports cancer progression by promoting cancer cell proliferation, survival, and angiogenesis.^{77 78} Additionally, IL-22 has been described to reduce the CD8 T-cell responses in chronic viral lymphocytic choriomeningitis virus infection. IL-22 expression is elevated in HPV+ cervical cancer,⁷⁹ and reducing IL-22 expression through neutralizing IL-23 could therefore also diminish immune evasion mechanisms and enhance antitumor immunity by shifting the balance towards more effective cytotoxic T-cell responses.

Collectively, our data indicate HPV16 E6/E7 oncoproteins are actively stimulating macrophages in the HPV+ TME to produce IL-23 by upregulating KLF2. This leads to suppression of the HPV-specific adaptive immune response through inhibition of HPV-specific CD8 T-cells. Further research into the clinical impact of US Food and Drug Administration-approved IL-23 neutralizing antibodies to enhance therapeutic HPV16 E6/E7 vaccination in humans with HPV+ cancers is warranted.

Data availability statement

Data are available upon reasonable request.