The study included 41 participants of whom 33 patients were diagnosed with HNSCC comprising 19 oropharyngeal SCC (OPSCC), eight oral cavity SCC (OCSCC), five laryngeal SCC (LSCC) and one hypopharyngeal SCC (HPSCC). The remaining eight cases were diagnosed with head and neck dysplasia (HND) and were included in the study as control samples. All patients attended the Maxillofacial and ENT Surgery Unit at the Istituto Nazionale Tumori IRCCS Fondazione G. Pascale. Each biopsy was divided in two sections: one was used for histopathological examinations, while the other was stored at −80°C in RNAlater Stabilization Solution (Thermo Fisher Scientific, Waltham, Massachusetts) and subsequently used for the molecular analyses.

SCC-derived HPV-negative PCA5 and CAL27 cell lines as well as HPV16-positive SCC152 and SCC154 cell lines were cultured in Dulbecco’s modified Eagle’s medium (DMEM) with 10% Fetal bovine serum (FBS), 1% L-glutamine and 1% penicillin/streptomycin antibiotic in 5% CO₂.

The study was conducted in accordance with the Declaration of Helsinki, and approved by the Institutional Review Board and Ethics Committee of Istituto Nazionale Tumori IRCCS Fondazione G. Pascale (authorization number n. 30/22 and n.9/23). The patients provided written informed consent to participate to the study.

Genomic DNA was extracted according to previously published method (Dassi et al., 2020). Briefly, 10 mg of tissue samples were digested with proteinase K (150μg/mL at 37°C overnight) in 100μL of lysis buffer (10 mM Tris–HCl pH 7.6, 5 mM EDTA, 150 mM NaCl, 1% SDS), then the DNA isolation was performed with phenol-chloroform-isoamyl alcohol (25:24:1) extraction and ethanol precipitation in 0.3 M sodium acetate (pH 4.6). The DNA quality and quantity was assessed by Nanodrop 2000c (Thermo Fisher Scientific) by calculating the 260 nm/280 nm and 260 nm/230 nm absorbance ratio. All the sample with a 260 nm/280 nm ratio between 1.8 and 2.0 were considered of good quality and included in the analyses.

For total RNA extraction, about 30 mg of tissue samples were dissociated with gentle MACS Octo Dissociator (Miltenyi Biotec, Bergisch Gladbach, Germany). Then, RNA was isolated from all the samples by using RNeasy MiniKit (Qiagen, Hilden, Germany) according to manufacturer procedure. The RNA quality and quantity was assessed as described for DNA.

The integrity of extracted DNA was assessed by PCR amplification of a 134 bp fragment within TP53 gene exon 7, using primers reported in Supplementary Table 1 (Hensel et al., 1991). Then, broad spectrum nested PCR was performed, according to a validated WHO protocol for the detection of alpha HPVs (Eklund et al., 2010). Specifically, 300 ng of genomic DNA were amplified with MY09/MY11 primer pairs (Resnick et al., 1990) for the outer reaction and with MGP primer set for the inner reaction (Söderlund-Strand et al., 2009), containing 5μL of outer reaction, as previously described (Dassi et al., 2020). The amplification reactions were verified by electrophoresis on a 7% polyacrylamide gel, followed by ethidium bromide staining and image analysis using the Gel Doc imaging system (Bio-Rad, Hercules, California).

The HPV genotypes were identified by direct automated DNA sequencing analysis of the amplified products using the primer GP5+ (de Roda Husman et al., 1994) at Eurofins Genomics GmbH (Ebersberg, Germany) and alignments of HPV sequences with those present in the GenBank database using the BLASTn software.¹

HPV16 viral load quantification was performed by droplet digital PCR (ddPCR). The Minimum Information for Publication of Quantitative Digital PCR Experiments for 2020 (dMIQE2020) checklist was followed for reactions setup (Supplementary Table 2). Briefly, the amplification of HPV16 E6 gene was performed in a 20μL reaction mixture including 10μL of 1X QX200 ddPCR EVAGreen Supermix (Bio-Rad), 300 nM each of E6 specific forward and reverse primers (Supplementary Table 1), 2 μL of DNA and nuclease-free water. The 20 μL reaction mixtures were loaded into DG8 Cartridges (Bio-Rad) with 70 μL of QX200 Droplet Generation Oil for EvaGreen (Bio-Rad) for automatic droplet generation using the QX200 Droplet Generator (Bio-Rad). Droplets were transferred into 96-well plates, which were heat sealed using the PX1 PCR Plate Sealer (Bio-Rad). Amplification was performed in duplicate in a CFX96 thermal cycler (Bio-Rad) and the droplets were read by a QX200 Droplet Reader (Bio-Rad). Finally, data were analyzed with QuantaSoft software version 1.7 (Bio-Rad). TP53 exon 7 was also amplified with specific primers (Supplementary Table 1). The viral copy number was calculated by normalizing the HPV16 E6 gene copy number against the amount of cellular DNA (TP53 gene) with the following formula: viral copy number/GE = number of E6 copies/(number of TP53 copies/2). Limit of blank (LOB) was assessed by amplification of HPV-negative NTERA2, HT3, and PCA23 cell lines DNA (Supplementary Figure 1). Limit of detection (LOD) of the reactions was determined by amplification of 1:10 serial dilutions of SiHa cell line DNA (Supplementary Figure 2).

A total of 250 ng of RNA for each sample was reverse transcribed by using the iScript cDNA Synthesis Kit (Bio-Rad) in a 20 μL volume reaction containing 1 μL of iScript reverse transcriptase, 4 μL of 5X iScript reaction mix and nuclease-free water. The reaction was incubated in a Mastercycler X50s (Eppendorf, Hamburg, Germany) thermal cycler at 25°C for 5 min and 46°C for 20 min, then the enzyme was inactivated at 95°C for 1 min.

The HNRNPA1, HNRNPA2B1, SRSF1, SRSF2, SRSF3, BRM, and SAM68 splicing factors transcripts were amplified in all samples and cell lines, while the E6 and E6*I transcripts were amplified in HPV16-positive samples as well as in cell lines by qPCR using specific primer pairs reported in Supplementary Table 1. The amplification mixture included 10 μL of 1X Sso Advanced Universal SYBR Green Supermix (Bio-Rad), 10 pmol of each primer, 2 μL of cDNA and nuclease-free water in a final volume of 20μL. The reactions were performed in duplicate with the CFX96 real time PCR Detection System (Bio-Rad).

Gene expression levels were normalized with the 2^–ΔCt method using GAPDH and ACTB as reference genes. Fold changes were calculated with the ΔΔCt method. All Ct values were corrected for primer pairs efficiency, which was calculated generating standard curves of SiHa cDNA serial dilutions.

The amplification of HPV16 E6 and E6*I mRNAs was performed in a 20 μL reaction mixture including 10 μL of 1X QX200 ddPCR EVAGreen Supermix (Bio-Rad), 200 nM each of E6 and E6*I specific forward and revers primers, 2 μL of cDNA and nuclease-free water (Supplementary Tables 1, 2). The annealing temperature of E6 and E6*I primers was optimized by amplifying SiHa cell line cDNA with a temperature gradient ranging from 52 to 62.8°C for E6 and from 46 to 56°C for E6*I. The optimal annealing temperature was 52°C for both primer pairs. LOB was assessed by amplification of HPV-negative NTERA2, HT3, and PCA23 cell lines cDNA (Supplementary Figure 1). LOD was determined by amplification of 1:2 serial dilutions of SiHa cell line cDNA into HPV-negative NTERA2 cell line cDNA (Supplementary Figure 2).

The PCA5 cell line was stably transduced with empty LXSN retroviral vector, LXSN carrying the HPV16 E6 ORF (LXSN_E6) and LXSN carrying HPV16 E6/E7 ORFs (LXSN_E6E7) using Lipofectamine 2000 reagent (Thermo Fisher Scientific), following the manufacturer’s instruction. Briefly, the Psi2 retrovirus packaging cell line was transduced with 10 μg of pLXSN empty vector, pLXSN_E6 and pLXSN_E6E7 DNA following the manufacturer’s instructions. Then, the medium containing the virions produced by Psi2 cells was collected, filtered with 0.22 μm filters and used for the transduction of PA317 packaging cell line with 4 μg/mL polybrene (Sigma-Aldrich, St. Louis, Missouri, United States). After selection of transduced PA317 cells with 600 μg/mL G418 (Thermo Fisher Scientific), the medium containing the viral particles was collected, filtered with 0.22 μm filters and used for the stable transduction of PCA5 cell line.

SCC152 cells were seeded in 6-well plates at 3 × 10⁵ cells/well for 24 h and then transfected with 10 μM of siRNAs targeting SAM68 (siSAM68) (Tichon et al., 2018), using Lipofectamine RNAiMAX Reagent (Thermo Fisher Scientific) following the manufacturer’s instructions. The siRNAs used were the following: siSAM68_1 (5′-CAUAAGAACAUGAAACUGA-3′), siSAM68_2 (5′-GCACCCAUAUGGACGUUAU-3′), siSAM68_3 (5′-UAUGAUGGAUGAUAUCUGU-3′) and siSAM68_4 (5′-ACAAGGGAAUACAAUCAAA-3′). In addition, cells were transfected with non-targeting siRNA (siNC, 5′-UUCUCCGAACGUGUCACGU-3′) (Yan et al., 2019). Untransfected cells (blank) were used as control. In addition, cell viability was evaluated at the time of silencing as well as 24 and 48 h after transfection by counting live cells stained with 0.4% Trypan blue (Thermo Fisher Scientific) using the Luna-II automated cell counter (Logos Biosystems, Anyang-si, Gyeonggi-do, South Korea). All the experiments were performed in triplicate.

Whole cell protein extracts were obtained from PCA5 cell line transduced with LXSN, LXSN_E6 and LXSN_E6E7 vectors as well as from SCC152 cell line transfected with siSAM68 by using RIPA Lysis Buffer System (Santa Cruz Biotechnology, Texas, United States) and quantified by Bio-Rad Protein Assay (Bio-Rad). Sam68 protein expression was analyzed by western blot. Briefly, 60 μg proteins were separated on 4–15% Mini-PROTEAN TGX precast protein gels (Bio-Rad) and transferred to Amersham™ Hybond P 0.45 PVDF blotting membrane (GE Healthcare, Illinois, United States), then mouse anti-Sam68 primary antibody (Santa Cruz Biotechnology, RRID: not available, sc-514468, 1:3,000 dilution) was incubated overnight at 4°C. Secondary anti-mouse IgG conjugated to horseradish peroxidase (Bio-Rad, #1706516, RRID:AB_2921252, 1:1,000 dilution) was incubated with the membrane for 1 hour at room temperature and protein bands were detected by chemi-luminescent method by Pierce™ ECL Western Blotting Substrate (Thermo Fisher Scientific). The α-Actin-1 was used for normalization (mouse anti-α-Actin-1, MAB1501, RRID:AB_2223041, 1:1,000 dilution).

PCA5 cells transduced with LXSN and LXSN_E6 vector were seeded on glass bottom in 8 well chamber slide (Ibidi, Gräfelfing, Germany) for 48 h at 2.5 × 10⁴/well. Cells were fixed in 4% paraformaldehyde and permeabilized for 10 min in 1X PBS containing 0.1% TritonX-100, and incubated overnight with anti-Sam68 (Santa Cruz Biotechnology, RRID: not available, sc-514468, 1:100 dilution) or anti-HPV16 E6/18 E6 (C1P5) (Santa Cruz Biotechnology, RRID:AB_675656, sc-460, 1:50 dilution) antibodies in blocking solution (1X PBS, 0.05% TritonX-100, 5% FBS). The Anti-phalloidin Alexa Fluor 647 (Invitrogen, A22287), and 4 μM Hoechst 33342 (Thermo Fisher Scientific) were used for cytoskeleton and nuclei counterstaining, respectively. Images were acquired with a Stellaris 5 DMI8 confocal fluorescence microscope (Leica, Wetzlar, Germany) by using 63X oil immersion objective.

HNSCC gene expression datasets were downloaded from The Cancer Genome Atlas (TCGA) and from the Broad Institute GDAC Firehose database. Normalized RSEM (RNA-Seq by Expectation-Maximization) data from Illumina HiSeq 2000 (IlluminaHiSeq_RNASeqV2) were collected for gene expression analysis. Information about HPV infection positivity and HPV genotype were also obtained, based on TCGA RNA-Seq data analyzed by Nulton et al. (2017). Splicing factors transcripts levels were stratified according to the HPV positivity, tumor anatomical sub-site and tumor grade. Moreover, Kaplan-Meier curves were generated using KmPlot online software² to investigate the prognostic value of splicing factors expression (Győrffy, 2024). Patients were assigned to low and high expression groups on the base of median gene expression value.

Statistical analysis was performed using GraphPad version 6 (Prism). The Pearson’s (r) and the Spearman’s (ρ) correlation coefficients were calculated for correlation analyses. The U Mann–Whitney test was used to evaluate differences in mRNA levels among sample groups. Variables with p ≤ 0.05 were considered statistically significant.

The study included a cohort of 33 patients diagnosed with HNSCC and eight with HND lesions. Demographic and clinic-pathologic data for all HNSCC patients included in this study are summarized in Table 1. In particular, 72.7% (n = 24/33) of patients were males and 27.3% (n = 9/33) females, with a median age of 60 years (interquartile range, IQR: 56–67 years) and 55 years (IQR: 54–77) at diagnosis, respectively (Table 1). Most of tumors were OPSCC (57.6%, n = 19/33) and were poorly differentiated (G3-G4) (54.5%, n = 18/33) (Table 1). The search for HPV DNA was performed in all samples by broad spectrum nested PCR using MY09/11 and MGPs primer pairs followed by direct sequencing analysis of amplified products. HPV DNA was detected in 48.5% (n = 16/33) of HNSCC (Table 1). Stratification by anatomical sub-site showed that 47.4% (n = 9/19) of OPSCC, 75% (n = 6/8) of OCSCC and 20% (n = 1/5) of LSCC were positive for HPV DNA. Search for HPV DNA sequences in the GenBank database allowed to identify HPV16 as the most common viral genotype, being detected in 88.9% (n = 8/9), 83.3% (n = 5/6), and 100% (n = 1/1) of HPV-positive OPSCC, OCSCC, and LSCC, respectively. The HPV33 was the second most frequent genotype being detected in 11.1% (n = 1/9) and 16.7% (n = 1/6) of HPV-positive OPSCC and OCSCC, respectively.

The eight patients diagnosed with HND included six men and two women, with a median age of 50 (IQR: 33–55) and 51 (IQR: 42–60) years, respectively. HPV DNA was detected in 37.5% (n = 3/8) of the samples, which were all positive for HPV16.

Viral load was measured in HPV16-positive HNSCC and HND as well as in SCC152 and SCC154 cell lines, through the evaluation of HPV16 E6 DNA and TP53 copy number by ddPCR, with the formula E6/(TP53/2) to obtain the viral copy number/GE. The HPV16 load ranged from < 1 copy/GE to 27 copies/GE in HNSCC, while it was < 1 copy/GE in HND, 1 copy/GE in SCC154 and 320 copies/GE in SCC152 cell line (Supplementary Table 3).

The expression of E6 full-length and E6*I mRNAs were analyzed in 13 HPV16-positive HNSCC, including 7 OPSCC, 5 OCSCC, 1 LSCC, in three HPV-16 positive HND as well as in SCC152 and SCC154 cell lines either by qPCR or by ddPCR. Amplicons were visualized by polyacrilammide gel electrophoresis and subjected to direct Sanger sequencing to verify the specificity of primer pairs used for E6*I isoform amplification (Supplementary Figure 3). The E6 and E6*I transcripts were both identified in five HNSCC, with E6*I being significantly more expressed than E6 in all samples (p < 0.01), while only the E6*I isoform was detected in two HNSCC cases both by qPCR and ddPCR (Figure 1A). The E6 and E6*I mRNAs were both detected also in SCC152 and SCC154 cell lines, with higher levels of the E6*I isoform (Figure 2B).

In particular, the E6*I expression was observed in 71.4% (n = 5/7) OPSCC, 20% (n = 1/5) of OCSCC, and 100% (n = 1/1) of LSCC. Both E6 and E6*I were expressed in SCC152 and SCC154 cell lines, while no viral transcripts were detected in HND. Concordant results were obtained by analyzing E6 (ρ = 0.8, p = 0.001) and E6*I (ρ = 0.9, p < 0.001) expression with qPCR and ddPCR (Supplementary Table 4). Moreover, a statistically significant correlation was found between E6*I levels and the viral load (ρ = 0.8, p < 0.001) (Supplementary Table 3).

The expression levels of HNRNPA1, HNRNPA2B1, SRSF1, SRSF2, SRSF3, BRM, and SAM68 genes were analyzed in HNSCC (Figure 2) as well as in CAL27, SCC152, and SCC154 cell lines (Figure 3) by qPCR. HNSCC were stratified by HPV status, considering HPV-positive those expressing the E6*I isoform. The analysis showed a statistically significant over-expression of SRSF3, BRM and SAM68 transcripts in E6*I-positive versus E6*I-negative HNSCC (p < 0.5) (Figure 2).

The analysis of cell lines showed a statistically significant up-regulation of splicing factors HNRNPA1, HNRNPA2B1, SRSF1, SRSF3, BRM, and SAM68 in both SCC152 and SCC154 compared to CAL27 (Figure 3).

Notably, a statistically significant correlation was found between SAM68 and E6*I levels in HNSCC and cell lines (r = 0.72, p = 0.03) (Figure 4).

The PCA5 cell line was transduced with LXSN empty vector as well as with LXSN carrying HPV16 E6 ORF (LXSN_E6) or HPV16 E6 and E7 ORF (LXSN_E6E7). Gene expression analysis by qPCR showed higher expression of E6*I compared to E6 in cells transduced with LXSN_E6 compared to those transduced with LXSN_E6E7 (Figure 5A). Moreover, SAM68 was found significantly up-regulated in LXSN_E6-transduced PCA5 cell line at protein level and at lesser extent at mRNA level (Figures 5B,C).

The subcellular localization analysis of Sam68 and HPV16 E6 proteins was performed by confocal fluorescence microscopy in LXSN and LXSN_E6 transduced PCA5 cell line. The analysis showed that Sam68 localization was mainly nuclear, with increased levels in cells transduced with LXSN_E6 vector compared to those transduced with LXSN empty vector (Figure 6A). In addition, the presence of HPV16 E6 oncoprotein was confirmed in the nucleus of cells transduced with LXSN_E6 vector while no signal was observed in the nucleus of cells transduced with the LXSN empty vector (Figure 6B).

The SCC152 cell line was transfected with four siRNAs targeting SAM68 (siSAM68) or with a scrambled control siRNA (siNC). The efficacy of SAM68 silencing at 48 h post-transfection was analyzed by qPCR, showing a statistically significant reduction of SAM68 mRNA levels in cells transfected with siSAM68 compared to control and untreated cells (blank) (p < 0.05) (Figure 7A). A statistically significant down-regulation of HPV16 E6*I expression was also observed in cells transfected with siSAM68 (p < 0.05) (Figure 7B). Down-regulation of Sam68 at protein level was confirmed by western blot analysis (Figure 7C). In addition, the growth curve analysis of SCC152 cell transfected with siSAM68, with siNC and of blank cells showed a statistically significant reduction of cell number at 48 h after transfection (p < 0.05) (Figure 7D).

The clinical and RNA-Seq datasets of 520 HNSCC and 44 head and neck normal tissues was downloaded from TCGA and the Broad Institute GDAC Firehose databases. Normalized mRNA expression levels of HNRNPA1, HNRNPA2B1, SRSF1, SRSF2, SRSF3, BRM, and SAM68 genes were extracted and stratified by HPV status in OPSCC and OCSCC.

The analysis of OPSCC showed a statistically significant up-regulation of HNRNPA2B1, SRSF1, SRSF2, SRSF3 and SAM68 in HPV-positive cases compared to normal tissues (p < 0.001). Moreover, the expression of SRSF1, SRSF2, SRSF3, BRM and SAM68 was statistically significant higher in HPV-positive versus HPV-negative cases (p < 0.05) (Figure 8).

The analysis of OCSCC showed a statistically significant up-regulation of HNRNPA2B1, SRSF1, SRSF2, and SAM68 in HPV-positive cases compared to normal tissues (p < 0.01). In addition, the analysis showed that the expression of HNRNPA2B1, SRSF2, SRSF3, BRM, and SAM68 was statistically significant higher in HPV-positive versus HPV-negative OCSCC (p < 0.05) (Figure 9). Overall, these data confirmed the results obtained in HNSCC by qPCR.

The stratification of expression data by tumor grade revealed that HNRNPA1, HNRNPA2B1, SRSF2, SRSF3, BRM, and SAM68 levels were significantly higher in poorly differentiated/undifferentiated (G3-G4) HNSCC compared to well differentiated/moderately differentiated (G1-G2) HNSCC (p ≤ 0.02) (Supplementary Figure 4). Finally, the Kaplan-Meier overall survival analysis of HNSCC by KMPlot showed no prognostic values of splicing factors expression (Supplementary Figure 5).