INHBA is Enriched in HPV-negative Oropharyngeal Squamous Cell Carcinoma and Promotes Cancer Progression

Abstract Patients with oropharyngeal squamous cell carcinoma (OPSCC) caused by human papilloma virus (HPV) exhibit a better prognosis than those with HPV-negative OPSCC. This study investigated the distinct molecular pathways that delineate HPV-negative from HPV-positive OPSCC to identify biologically relevant therapeutic targets. Bulk mRNA from 23 HPV-negative and 39 HPV-positive OPSCC tumors (n = 62) was sequenced to uncover the transcriptomic profiles. Differential expression followed by gene set enrichment analysis was performed to outline the top enriched biological process in the HPV-negative compared with HPV-positive entity. INHBA, the highest overexpressed gene in the HPV-negative tumor, was knocked down. Functional assays (migration, proliferation, cell death, stemness) were conducted to confirm the target's oncogenic role. Correlation analyses to reveal its impact on the tumor microenvironment were performed. We revealed that epithelial-to-mesenchymal transition (EMT) is the most enriched process in HPV-negative compared with HPV-positive OPSCC, with INHBA (inhibin beta A subunit) being the top upregulated gene. INHBA knockdown downregulated the expression of EMT transcription factors and attenuated migration, proliferation, stemness, and cell death resistance of OPSCC cells. We uncovered that INHBA associates with a pro-tumor microenvironment by negatively correlating with antitumor CD8+ T and B cells while positively correlating with pro-tumor M1 macrophages. We identified three miRNAs that are putatively involved in repressing INHBA expression. Our results indicate that the upregulation of INHBA is tumor-promoting. We propose INHBA as an attractive therapeutic target for the treatment of INHBA-enriched tumors in patients with HPV-negative OPSCC to ameliorate prognosis. Significance: Patients with HPV-negative OPSCC have a poorer prognosis due to distinct molecular pathways. This study reveals significant transcriptomic differences between HPV-negative and HPV-positive OPSCC, identifying INHBA as a key upregulated gene in HPV-negative OPSCC's oncogenic pathways. INHBA is crucial in promoting EMT, cell proliferation, and an immunosuppressive tumor environment, suggesting its potential as a therapeutic target for HPV-negative OPSCC.


Introduction
Head and neck squamous cell carcinoma (HNSCC) is the sixth most common cancer worldwide, with a high incidence in developing countries.The majority of HNSCCs is caused by tobacco and alcohol abuse, as well as infection with high-risk human papillomavirus (HPV; ref. 1).Oropharyngeal squamous cell HPV-positive OPSCC tends to have a more indolent course with slower growth (7).In contrast, HPV-negative OPSCC is associated with a more aggressive clinical course, poorer prognosis, and higher resistance to therapy (7,8).There is a high medical need to improve outcomes of patients with HPV-negative OPSCC.
Epithelial-to-mesenchymal transition (EMT) is a biological process that allows epithelial cells to acquire a mesenchymal phenotype characterized by increased motility, invasiveness, and resistance to therapy.Studies have shown that the activation of EMT-associated signaling pathways, such as TGFβ, is linked to increased invasiveness and poor prognosis (9).INHBA (inhibin beta A subunit) is a protein belonging to the TGFβ family.Preliminary bioinformatic analysis of public datasets has shown that INHBA is overexpressed in various solid tumors, including head and neck cancer, compared with respective healthy tissues (10)(11)(12)(13)(14)(15)(16).
Despite the significant progress in understanding the molecular processes underlying OPSCC, the fundamental mechanisms contributing to the aggressiveness of HPV-negative OPSCC are poorly understood, and the available therapeutic targets for this cancer subtype are limited.Current adjuvant treatments for HPV-negative OPSCC, such as radiotherapy, chemotherapy, and immunotherapy, have moderate efficacy in mitigating recurrence and are sometimes associated with significant morbidity and toxicity (17).Therefore, there is a need to identify biologically relevant therapeutic targets for HPV-negative OPSCC, which could, in turn, improve the treatment of this cancer subtype and poor patient outcome.To explore the differentially regulated molecular mechanisms between HPV-negative and HPV-positive OPSCC primary tumors, this study systematically compared the bulk mRNA profiles of the two subentities to outline an appropriate therapeutic target and evaluate its functionally oncogenic role.

Patient Samples and External Cohort
A total of 62 patients with OPSCC were included in our study cohort (Ulm).A total of 51 representative tissue samples from primary tumor were obtained during surgery at the University Hospital Ulm, Germany, and 11 were provided by C. Ottensmeier from Southampton University, Southampton, UK.The designated ethical committee has given their approval for this study (Ulm University: approval number 222/13; 90/15; UK Medical Research and Ethics Committee: approval number MREC 09/H0501/90).Informed written consent was obtained before patient samples were taken between July 2013 and August 2022.All procedures involving patient sample collection were conducted in full compliance with the ethical standards outlined in the Declaration of Helsinki.HPV status was determined by RNA sequencing (RNA-seq), PCR for HPV DNA, and p16 IHC.The Cancer Genome Atlas (TCGA) gene and miRNA expression values along with clinical parameters of 70 OPSCC samples were downloaded using the R package TCGAbiolinks (2.23.2).TCGA patient with barcode TCGA-BB-A6UM lacks miRNA data.Hence, only 69 samples were included in the miRNA analysis.HPV status of TCGA samples was determined utilizing the metdata provided on p16 IHC and HPV-DNA ISH.Missing HPV status was determined by collecting HPV oncogenes normalized RNA reads per million and applying the cutoff recommended in a previous study (18).

RNA-seq
The AllPrep DNA/RNA Mini Kit (Qiagen, 80204) was used to extract total RNA from snap-frozen tumor tissues.Quality of total RNA was evaluated using an Agilent 2100 Bioanalyzer Instrument (Agilent RNA 6000 Pico).The next-generation sequencing method from Illumina was used to sequence RNA samples, following the manufacturer's recommendations.Briefly, libraries were made from 500 ng of input material using TruSeq Stranded mRNA.Libraries of 51 samples were pooled and single-end sequenced in one lane of the HiSeq 2500/NovaSeq 6000 System.A total of 11 samples libraries were paired-end sequenced on HiSeq2500 platform.Sequence data were converted to FASTQ format using bcl2fastq (2.20.0.422).After being mapped to the human genome (hg38) using STAR (2.0.9) and multimapping reads being eliminated, highquality readings were converted to gene-specific read counts for identified genes using featureCounts (2.0.0).As a fraction of all reads, almost 75% of the total reads were uniquely mapped.Unmapped reads were matched to HPV highrisk type genomes using a viGen bioinformatic approach (19).If samples had at least 500 reads for either the HPV E6 or E7 RNA or at least 500 reads for all HPV oncogenes combined (E1, E2, E4, E5, E6, E7, L1, L2), they were considered HPV-positive.

Transfection of UDSCC1 Cells
A total of 250,000 UDSCC1 cells in 2.5 mL complete medium were seeded per well of a 6-well plate (Sarstedt, 83.3920.300)and cultured for 24 hours.

Western Blot Analysis
Cell lysis was performed using RIPA buffer supplemented with protease and phosphatase inhibitors (1 tablet/

Scratch Assay
The cell monolayer was scratched at the day of transfection using a pipette tip and documented every 24 hours using an Olympus CK30 microscope (Zeiss) at 10-fold magnification.Image processing was performed using python (3.8.5).
scikit-image (0.18.3) was used to read images and the entropy function was used to calculate entropy at each pixel position.Images were plotted using matplotlib (3.5.0) and Otsu method was utilized to compute the optimal threshold for image binarization.Area in pixels (A) with entropy lower than the threshold was calculated using numpy (1.21.2) and used as a proxy for the open wound area.Open wound percentage was calculated as follows, where t denotes any timepoint and 0 denotes the starting timepoint:

Proliferation Assay
Immediately before transfection, cells were stained with 2.5 μmol/L CellTrace CFSE (Thermo Fisher Scientific, C34554) at 37°C for 20 minutes in the dark, according to manufacturer's instruction for adherent cells.Cells were harvested using Trypsin/Ethylenediaminetetracetic acid (EDTA) (PAN-Biotech, P10-023100) and proliferation was measured on a Gallios flow cytometer (Beckman Coulter).

Cell Death Assay
Apoptosis and necrosis were measured using FITC Annexin V Apoptosis Detection Kit with 7-Aminoactinomycin D (7-AAD) (BioLegend, 640922) according to manufacturer's instruction.Briefly, cells were washed with annexin V binding buffer, provided with the kit, and resuspend in 100 μL annexin V binding buffer.Cells were stained with 5 μL FITC annexin V and 5 μL 7-AAD for 15 minutes in the dark and measured on a Gallios flow cytometer (Beckman Coulter).

Colony Formation Assay
A total of 1,000 cells per well were seeded in 6-well plates, transfected the next day and cultured for 2 weeks without medium change.Then, cells were washed with ice-cold PBS (Gibco, 14190-094) and fixed with prechilled methanol (Thermo Fisher Scientific, 10365710) for 15 minutes before being rinsed once more with PBS.Colonies were stained with 0.5% crystal violet dye (Carl Roth, T123.1).Images were taken using Keyence BZ-X microscope (Keyence) at 5-fold magnification to scan the entire well.Image processing was performed using BZ-II Analyzer 1.0 (Keyence).

Aldehyde Dehydrogenase Activity Assay
Aldehyde dehydrogenase (ALDH) activity assay (Abcam, ab155893) and data analysis were performed following the manufacturer's instructions.Absorption was measured using TECAN spectrophotometer at 450 nm.

Data Analysis and Statistical Analysis
Data analysis was performed in R (4.1.1).Data wrangling was carried out utilizing tidyverse (1.3.1),data.table(1.14.2), and dplyr (1.0.9).Differential expression analysis was performed using deseq2 (1.34.0), and the shrinkage estimator used was "apeglm" (20).The threshold for differentially expressed genes was determined with an absolute log fold change (LFC) >1 and a FDR < 0.05 with a correction for batch effect.The RNA expression transcript per million (TPM) values were used for correlation and survival analysis binarization.Correlation matrices were generated using corrplot (0.92).The Spearman correlation was calculated using R base stats package.Correlations with P value < 0.05 were considered significant.Hallmark gene sets were loaded using msigdbr (7.5.1).Gene set enrichment analysis (GSEA) with 1,000 permutations was performed and plotted using fgsea (1.20.0);gene sets with FDR < 0.05 were considered significantly regulated.Binarization of patients based on expression cutoffs was done using matrixStats package (0.61.0).Survival analysis was conducted by fitting survival curves using the survival package (3.3-1) and generating Kaplan-Meier plots utilizing the survminer package (0.4.9).The restricted mean survival time (RMST) was computed and plotted using the survRM2 package (1.0-4).The immunedeconv (2.0.4)R wrapper package was used to perform immune deconvolution on RNA-seq data using the quan-TIseq program as method to predict M1 and M2 macrophages, monocytes, CD4 + regulatory and non-regulatory T cells, CD8 + T cells, B cells, neutrophils, natural killer cells, and myeloid dendritic cells infiltration ratios.Venn diagrams were generated using ggvenn (0.1.9).Assessment of complementarity between miRNA and 3 -UTR (untranslated region) of INHBA and targeted prediction was done utilizing miRDB (21) and TargetScan 8.0 (22).Visualization was done using ggplot2 (3.3.6).Clinical data summary was generated using summarytools (1.0.1).

Data Availability Statement
The data generated in this study are publicly available in Sequence Read Archive database at PRJNA967751.

Study Population
The clinicopathologic characteristics of the Ulm and TCGA OPSCC cases (n = 62; n = 70) are described in Table 1.There were no significant statistical differences between the cohorts.

GSEA Identified INHBA as an EMT driver in HPV-negative Tumors
To investigate the differential regulation of biological processes underlying the differences between HPV-negative and HPV-positive OPSCC, we performed differential expression analysis of 23 HPV-negative and 39 HPV-positive tumors followed by GSEA.The analysis revealed significant (FDR < 0.05) enrichment for gene sets related to EMT [normalized enrichment score (NES) = 2.36], myogenesis (NES = 2.24), angiogenesis (NES = 1.79), components of the apical junction complex (NES = 1.65),TNFα signaling via NFκB (NES = 1.55), and downregulated genes by KRAS activation (NES = 1.52), and attenuation of upregulated genes during transplant rejection (NES = −2.14) in HPV-negative versus HPV-positive tumors (Fig. 1A; Supplementary table S1).EMT was the top enriched biological process in the HPV-negative tumor (Fig. 1B).The EMT gene set included 200 genes, with 127 identified as the leading edge of enrichment (p-EMT).Fifty-five of the p-EMT were shown to be significantly upregulated in the HPV-negative tumor (FDR < 0.05).INHBA exhibited the highest upregulation with an LFC of 2.9 (Fig. 1C; Supplementary Table S2).

INHBA Regulates the Expression of the EMT Transcription Factors and Enhances Migration in an HPV-negative OPSCC Cell Line
A strong positive correlation was detected between the expression of INHBA and EMT transcription factors SNAI (Rho = 0.62), SNAI (Rho = 0.82), and TWIST (Rho = 0.69) in the primary tumor (Fig. 3A-C).INHBA was found to be differentially expressed under native conditions in UDSCC1 (HPV-negative) and UDSCC2 (HPV-positive) cell lines, with significantly higher mRNA and protein levels in UDSCC1 (Fig. 3D and E).To verify the role of INHBA in regulating the expression of EMT transcription factors in HPV-negative tumors, two siRNAs against INHBA (KD1 and KD2) were transfected into UDSCC1, and UDSCC2 was used as a positive control to evaluate whether the knockdown of INHBA relatively shifts the oncogenic features of UDSCC1 to a phenotype comparable to that of UDSCC2.INHBA knockdown efficiencies were evaluated on RNA level (Fig. 3F, KD1: 47% after 24 hours, 80% after 48 hours, KD2: 49% after 24 hours, 76% after 48 hours) and validated on protein level (Fig. 3G, KD1: 24% after 24 hours, 45% after 48 hours, KD2: 17% after 24 hours, 45% after 48 hours).INHBA knockdown significantly downregulated the expression level of SNAI 24 and 48 hours after transfection and SNAI 48 hours after transfection, while TWIST showed a nonsignificant decrease (Fig. 3F).
INHBA knockdown showed a modest upregulation of E-Cadherin protein levels (Supplementary Fig. S1).As EMT enhances the migratory ability of cells, we examined the INHBA knockdown effect on the migratory potential of UDSCC1.The wound healing assay revealed that INHBA knockdown decreased the wound closure capability of UDSCC1 cells and rendered them less

INHBA Potentiates Proliferation and Attenuates Cell Death of UDSCC1
To further elucidate the role of INHBA in promoting the aggressiveness of HPVnegative OPSCC cells, we examined its knockdown effect on the proliferation capacity and cell death resistance in UDSCC1.INHBA knockdown notably reduced the proliferative ability of UDSCC1 cells rendering it comparable to that of the less proliferative UDSCC2 (Fig. 5A and B).In addition, the knockdown of INHBA significantly promoted the death of UDSCC1 cells, yielding a higher level of primarily late apoptotic but also early apoptotic and necrotic cell percentages (Fig. 6A and B).These results indicate that INHBA overexpression in HPV-negative tumors promotes the aggressiveness of cancer cells by enhancing proliferation while attenuating cell death simultaneously.

INHBA Associates with the Expression of HNSCC Stemness-related Markers and Potentiates Clonal Expansion
As undifferentiated cancer cells, most likely cancer stem cells, are the driving force of clonal expansion, a colony formation assay was performed to investigate whether INHBA knockdown influences the ability of single cells, particularly cancer stem cells, to resist the absence of cell-to-cell contact and develop into a colony via clonal expansion.Because this assay aimed not to assess the proliferative potential but rather the ability of single cells to survive and expand, and because INHBA was proven to influence proliferation, small colonies were also counted while keeping the time variable fixed for all conditions.Colonies ranging in size between 10 and 50 cells and those of 50 or more were counted (Fig. 7A and B).Small colonies were relatively more prevalent in UDSCC1 with INHBA knockdown.The number of colonies recorded for UD-SCC2 and UDSCC1 cells upon INHBA knockdown was significantly lower than the negative control cells (Fig. 7A and C).Consequently, expression correlation analysis was done between INHBA and previously described HNSCC stemnessrelated markers in primary tumors.The analysis revealed a significant positive correlation between INHBA expression level and nine markers.A positive correlation (Rho > 0.4) was recorded between INHBA and CD, ALDHL, HIFA, and ZFP, simultaneously (Fig. 8A).To functionally validate the association, we performed ALDH enzyme activity assay, as increased ALDH activity serves as a cancer stem cell marker.ALDH activity was significantly reduced in UDSCC1 cells upon INHBA knockdown compared with the negative control (Fig. 8B).Overall, these results suggest that INHBA potentially promotes OPSCC stemness.

INHBA Gene Expression Correlates with the Ratio of OPSCC Infiltrating Immune Cells
The HNSCC tumor microenvironment is known to be immunosuppressive, particularly in aggressive primary tumors.Hence, we investigated the correlation between INHBA expression level and the computed infiltrating ratio of different types of immune cells.The expression level of INHBA in primary tumors negatively correlated with the infiltration ratio of CD8 + cytotoxic T cells (Rho = −0.42),indispensable in orchestrating the antitumor immune response.Furthermore, a negative correlation (Rho = −0.54)with the level of B-cell infiltrates was revealed.Incidentally, a weak negative correlation (Rho = −0.32)and a strong positive correlation (Rho = 0.64) were disclosed between INHBA expression level and M2 and M1 macrophage infiltration level, respectively (Fig. 9A).M1 macrophages are typically known to be critical players in initiating antitumor immunity.Recently, however, a pro-tumor M1 macrophage subtype was described in melanoma (23) and characterized by increasing the formation of a dense matrix around cancer cells via upregulating HA synthase 2 (HAS) due to the activation of the TNFα signaling via NFκB, which in turn, was shown in the current study also to be augmented in HPVnegative OPSCC tumors (Fig. 1A).Hence, we correlated the expression level of the tumor-upregulated gene signature, comprising HAS, MMP, CXCL, CXCL, ILB, and IL, with the ratio of infiltrating M1 macrophages.The genes' RNA levels positively correlated with M1 ratios infiltrating the OPSCC primary tumor (Fig. 9B).These results suggest that INHBA is a potential immune response modulator by associating with a pro-tumor microenvironment in OPSCC tumors.

Three miRNAs are Outlined as Putative Regulators of INHBA Gene Expression
It is well known that interaction between miRNA and mRNA is a vital mode of gene expression regulation in cancer.We applied a three-step approach to evaluate a possible regulation of INHBA expression by miRNAs in the primary tumor.First, we performed an expression correlation utilizing the paired TCGA samples to outline miRNAs that negatively correlate with the INHBA mRNA level.Then, we utilized complementarity and interaction prediction of two online databases to delineate the putative interacting candidates.Finally, we assessed the prognostic impact of the outlined miRNAs.A total of 12 miRNAs were found to negatively correlate with INHBA expression (Rho < −0.4; P < 0.05; Fig. 9C; Supplementary Table S3).A total of 71 and 26 miRNAs were predicted, by miRDB and TargetScan, respectively, to interact with INHBA mRNA 3 UTR.Only three of the latter intersected with the negatively correlating miRNA; miR-106a, miR-20b, and miR-9-1/2/3 were the candidates of interest, and their respective binding sites in the 3 UTR were defined (Fig. 9D).After binarizing patients into high and low expression categories based on the median of raw reads as a cutoff, survival analysis showed that most to almost all HPV-negative patients were in the low expression category of the three different miRNAs, thus yielding an inverse prognostic signature to that of INHBA.Patients with the lower expression level of all three miRNAs had a worse prognosis (Fig. 9E-G).These findings suggest that miR-106a, miR-20b, and miR-9-1/2/3 likely downregulate the expression of INHBA, hindering its oncogenic role in the primary tumor.

Discussion
Many studies have shown that patients with HPV-positive OPSCC have a better prognosis than HPV-negative patients, mainly due to higher immunogenicity, increased responsiveness to radiation and chemotherapy resulting in lower recurrence rates and distant metastasis (3)(4)(5)(6)(7)(8).However, research utilizing unbiased data-driven comparative analysis to delineate relevant therapeutic targets potentially promoting OPSCC aggressiveness in HPV-negative patients is still lacking.In the current study, we applied an exploratory data analysis approach to examine the differentially regulated molecular processes between HPV-negative and HPV-positive OPSCC tumors and to uncover key players contributing to the differential prognosis.We identified EMT as the top enriched process in HPV-negative tumors, with INHBA being the leading upregulated gene.The knockdown of INHBA in an HPV-negative OPSCC cell line confirmed its functional role in invasion, proliferation, apoptosis resistance, and stemness.We found that INHBA is likely involved in accompanying a pro-tumor microenvironment.Moreover, we outlined three miRNAs potentially involved in repressing INHBA expression that are downregulated in HPV-negative compared with HPV-positive tumors.These findings reveal the functional role of INHBA in OPSCC progression and aggressiveness, suggesting it to be used as potential therapeutic target for INHBA-enriched tumors, principally in favor of patients with HPV-negative OPSCC.
In OPSCC, our study uncovered that the prognostic impact of INHBA expression is HPV dependent, with most patients harboring INHBA-enriched tumors being HPV-negative.Incidentally, RUNX2 was shown to upregulate the expression of INHBA in HNSCC (24), and recent reports confirmed an HPV-integration site upstream of RUNX in close proximity to its promoter (25,26).Consequently, a disruption in the expression of RUNX by HPVgenome integration could influence the RUNX2/INHBA axis in HPV-positive OPSCC, leading to INHBA downregulation.Moreover, patients with HPVnegative OPSCC are known for high nicotine consumption throughout their lifetime.Ashour and colleagues showed that nicotine upregulates INHBA in chicken embryos (27).Hence, INHBA overexpression could be attributed to nicotine consumption.
INHBA affects cancer progression by regulating a plethora of oncogenic hallmarks.In breast cancer, INHBA has been shown to induce EMT and enhance the motility of malignant cells (28).The latter was also shown by Kang and colleagues to result from INHBA overexpression in prostate cancer cells (10).In addition, INHBA was found to promote cancer progression in lung adenocarcinoma by accentuating the proliferation of tumor cells (12).Li    Another possibility is the upregulating of SMAD/ by INHBA, as described in nasopharyngeal carcinoma (32) and pancreatic ductal adenocarcinoma (33).SMAD2/3, in turn, can interact with FOXO1/3/4, which are responsible for the upregulation of CDKN1A (34).Regarding apoptosis resistance, an interaction between SMAD2/3 and TP53 activates the transcription of SERPINE, which in turn was shown to activate the anti-apoptotic PI3K/Akt pathway (35).
In OPSCC primary tumors, we could show a significant positive correlation between the expression of INHBA and nine previously described HNSCC stemness-related markers, with CD, ALDHL, HIFA, and ZFP exhibiting the strongest association.We additionally confirmed the essentiality of INHBA for clonal expansion as its knockdown in vitro significantly hindered the capability of single UDSCC1 cells to survive, expand, and form colonies.Furthermore, we could show that the knockdown of INHBA attenuated the activity of the ALDH enzyme, a known marker of cancer stem cells (36).In line with our findings, Lonardo and colleagues showed INHBA to be vital for the ability of pancreatic cancer stem cells to self-renew and maintain their stemness (14).One possible mechanism is via CD, which, in our study, was shown to have a strong positive correlation with INHBA expression.In head and neck cancer, CD44 was shown to inhibit the phosphorylation of GSK3β, which is required to maintain stem cell self-renewal (37).
The relationship between INHBA and tumor immune infiltrates was studied in various solid cancers to understand its impact on the tumor microenvironment.(43).As CCL19 is known to be a B-cell chemoattractant (44), this could possibly predate the low levels of B-cell tumor infiltration.
To our knowledge, the association between INHBA and M1 macrophage infiltration has not been reported previously.However, a positive correlation was disclosed between INHBA expression and macrophage infiltration level in breast (40) and cervical cancer (45).Incidentally, we showed a positive association between INHBA expression and M1 macrophage infiltration in OPSCC tumors.M1 macrophages are typically considered to be antitumor.However, Kainulainen and colleagues recently described a protumor M1 macrophage subtype in melanoma, accentuating invasive melanoma cells by upregulating the TNFα signaling via NFκB in tumor cells (23).Our study revealed that the resulting tumor-upregulated molecular signature is also associated with the level of M1 macrophages infiltrating OPSCC tumors.
Various miRNAs were identified as oncogenes and tumor suppressors, and their aberrant expression was implicated in the development and progression of various types of cancers (46).miRNAs were extensively researched in HNSCC and shown to hold a prognostic value (47).We previously showed that miRNA as exosomal cargo correlates with clinical parameters in HNSCC (48,49) and can alter biological processes like EMT (50).In the current study, three putative miRNAs, miR-106a, miR-20b, and miR-9, were outlined as repressors of INHBA expression, prompting their HPV-dependent upregulation in OPSCC prognostically favorable.The role of miRNAs in cancer is controversial and highly dependent on cancer context and entity.Still, miR-106a was hitherto reported to be upregulated by HPV E7 and to enhance radiation sensitivity by downregulating the expression of the RUNX axis in HNSCC (51).miR-20b, which belongs to the miR-106a-363 cluster, was associated with a better prognosis in HNSCC (52).miR-9 also, was shown to function as a tumor-suppressive miRNA in oral squamous cell carcinoma (OSCC), OPSCC, and nasopharyngeal carcinoma (NPC; refs.[53][54][55]. Recently, Xiao and colleagues revealed that 10 mmol/L of metformin, an oral antihyperglycemic used to treat type 2 diabetes mellitus, attenuates the expression of INHBA and inhibits the proliferation of colorectal cancer cells (56).An epidemiologic analysis based on three multicentric studies showed that metformin given for the treatment of diabetes is associated with longer overall survival of patients with HNSCC despite the negative prognostic impact of diabetes itself (57).In addition, it was demonstrated that the combination of metformin and cisplatin significantly improved the therapeutic effect of cisplatin in OSCC (58) and NPC (59).In contrast, another study found no significant prognostic impact of metformin in patients with HNSCC (60).The conflicted finding regarding the impact of metformin on HNSCC prognosis may be attributed to discrepancies in the HPV status of patients; INHBA enrichment in HPV-negative tumors, as shown in the current study, possibly renders the impact of metformin HPV-dependent.Interestingly, an analysis of samples from a metformin clinical trial in HNSCC showed that, following metformin treatment (dose of 1000 mg), cancer cell apoptosis was substantially higher in HPV-negative mucosal HNSCC tumor samples than in HPV-positive OPSCC tumor samples.Also, metformin-treated specimens displayed a significantly higher CD8 + effector T-cell infiltrate than specimens not treated (61).Hence, metformin in the neoadjuvant and adjuvant treatment of patients with HPV-negative OPSCC with INHBA-enriched tumors should be considered.

FIGURE 1
FIGURE 1 GSEA identifying INHBA as a leading driver of EMT in HPV-negative tumors.A, Dot plot of pathways enriched and depleted in HPV-negative compared with HPV-positive OPSCC tumor as determined by GSEA of transcriptomes from n = 62 primary tumor samples (Ulm cohort).Results were ranked according to the NES, dots were colored according to FDR and size was determined by the number of genes.B, GSEA plot of the top pathway (EMT) showing enrichment score (green curve), position of genes (black lines) with the leading edge (p-EMT), NES, and FDR.C, Volcano plot showing the log 2 fold change of genes upregulated in HPV-positive (red) and HPV-negative OPSCC (blue), with INHBA (red circle) as the highest upregulated leading edge in HPV-negative tumors.

FIGURE 2
FIGURE 2 Kaplan-Meier survival analysis and RMST in patients with OPSCC based on INHBA expression levels.A and B, Kaplan-Meier plots with log-rank test for overall survival of patients with OPSCC (A: Ulm cohort, B: TCGA cohort) based on tumor INHBA expression.Blue indicates low and red indicates high INHBA expression, as determined using Q3 as cutoff.(C-F) RMST defined as the area under the survival curve (pink) based on INHBA expression status (C: INHBA-high in all tumors, D: INHBA-low in all tumors, E: INHBA-high in HPV-negative tumors, F: INHBA-low in HPV-negative tumors).

FIGURE 3 INHBA
FIGURE 3 INHBA expression in OPSCC tumor and cell lines and its correlation with EMT transcription factors.Correlation of INHBA expression with the expression of SNAI1 (A), SNAI2 (B), and TWIST1 (C) in OPSCC tumors (Ulm cohort).TPM = transcripts per million.D, Relative expression of INHBA mRNA in native UDSCC1 (HPV-negative) and UDSCC2 (HPV-positive) cells as determined by qRT-PCR and normalized to RPL30 and UDSCC1.Results are shown as mean ± SD of n = 6 replicates.INHBA expression was not detected in n = 3 UDSCC2 samples.Expression level between UDSCC1 and UDSCC2 was compared by unpaired t test with **, P ≤ 0.01.E, Representative Western blot analysis (of n = 3 replicates) showing INHBA protein expression in native UDSCC1 and UDSCC2 cells.Numbers below blots indicate relative INHBA intensity normalized to GAPDH.F, Relative expression of INHBA, SNAI1, SNAI2, and TWIST1 mRNA in UDSCC1 cells upon transfection with Mock siRNA [Mock knockdown (KD)] and two INHBA siRNAs (INHBA KD1 and KD2) after 24 hours (t24) and 48 hours (t48).Target gene expression was determined by qRT-PCR and normalized to RPL30 and Mock KD. Results are shown as mean ± SD of n = 3 replicates.Expression level between Mock KD and INHBA KD1/2 was compared by unpaired t test with *, P ≤ 0.05; **, P ≤ 0.01; and ns, P > 0.05.G, Representative Western blot analysis (of n = 3 replicates) showing INHBA protein expression in Mock KD and INHBA KD1/KD2 at t24 and t48.Numbers below blots indicate relative INHBA intensity normalized to GAPDH.

FIGURE 6
FIGURE 6 Cell death upon INHBA knockdown.A, Cell death of native UDSCC1, Mock KD, INHBA KD1/KD2 and native UDSCC2 cells as assessed by Annexin V/7-AAD staining and flow cytometry.Results are shown as mean ± SD of the percentages of dead cells normalized to UDSCC1 (n = 3).Conditions were compared with native UDSCC1 using unpaired t test with *, P ≤ 0.05; **, P ≤ 0.01; ****, P ≤ 0.0001; and ns, P > 0.05.B, Representative flow cytometry plots for each condition.Gates show early apoptotic in the lower right quadrant, late apoptotic in the upper right quadrant and necrotic cell populations in the upper left quadrant.t0-plots represent the baseline timepoint right before transfection.

FIGURE 7
FIGURE 7 Colony formation ability upon INHBA knockdown.A, Ability for clonal expansion of native UDSCC1 cells, Mock KD, INHBA KD1/KD2, and UDSCC2 cells as assessed by colony formation assay.Colonies between 10 and 50 cells were considered small (red) and colonies greater 50 cells were considered big (blue).Results are shown as mean ± SEM of the number of total colonies (n = 3).Conditions were compared with native UDSCC1 using unpaired t test with **, P ≤ 0.01; ***, P ≤ 0.001; and ns, P > 0.05.B, Representative images of small and big colonies in UDSCC1 and UDSCC2 cells stained with crystal violet.C, Representative images of colony formation for each condition.

FIGURE 9
FIGURE 9 Correlation analyses between INHBA expression, immune infiltrates and miRNAs.A, Waterfall plot showing Spearman correlation between INHBA expression level and infiltrating ratios (IR) of different immune cells, as computed by immune cell deconvolution.Rho > 0.4 was considered relevant (dashed line).*, P ≤ 0.05; ***, P ≤ 0.001; ns, not significant.B, Correlation matrix between INHBA and the pro-tumor M1 signature showing Spearman rho and P-value symbols (*, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001) in the upper triangular matrix and scatter plots in the lower triangular matrix.TPM = transcripts per million.C, Waterfall plot showing Spearman correlation between (Continued on the following page.)

TABLE 1
Clinicopathologic features of patients with OPSCC in Ulm and The impact of INHBA expression on the clinical outcome of patients with OPSCC was evaluated (Ulm cohort) and validated (TCGA cohort) by survival analysis with the log-rank test.Patients were binarized into high and low expression categories based on the third quantile (Q3) of TPM as a cutoff.Most patients in the high expres- sion category were HPV-negative (Ulm: 81%; TCGA: 83%), which mirrors the enrichment analysis results.Relative expression levels of INHBA significantly impacted the overall survival of patients.Patients with higher expression levels had a worse prognosis than those with low levels of INHBA (Fig.2A and B).When evaluated by the INHBA expression status of the two cohorts, the RMST recorded was lower by 31.99 months in INHBA-high versus INHBA-low OP-SCC (60.56 vs. 92.55,respectively;Fig.2C and D).In HPV-negative patients, the RMST recorded was lower by 25.83 months in INHBA-high versus INHBA-low OPSCC (52.25 vs. 78.08,respectively,Fig.2E and F) but did not reach statistical significance (P = 0.09).Our data outlines INHBA as an HPV-dependent prognostic biomarker in OPSCC by delineating a subpopulation of patients with HPV-negative OPSCC, overexpressing INHBA in the primary tumor.
(29)colleagues showed that targeting INHBA in ovarian cancer cells is implicated in hindering the activation of stromal fibroblasts(29).Nonetheless, in OPSCC, the functional role of INHBA still needed to be elucidated.The enrichment of INHBA in tumors is shown in the current study to be implicated in the upregulation of EMT in HPV-negative OPSCC.We demonstrated that INHBA was expressed in UDSCC1, an HPV-negative OPSCC cell line.We could also show that the knockdown of INHBA made UDSCC1 cells less migratory and proliferative while simultaneously potentiated cell death.Its knockdown (40)ould show that INHBA negatively correlates with CD8 + T cells in OPSCC primary tumors, which concurs with the findings by Pinjusic and colleagues demonstrating the indirect ability of INHBA to reduce CD8 + T cells tumor infiltration in melanoma in vivo models by downregulating the production of CXCL9/10 (38).In contrast, INHBA expression correlated positively with infiltration of CD8 + T cells in colorectal(39)and breast cancer(40).Furthermore,