NSD2 upregulation is driven by high-risk HPV E6/E7 and disrupts epithelial differentiation in HPV-associated head and neck cancer

Abstract

Background: Head and Neck Squamous Cell Carcinoma (HNSCC) are classified in two main subtypes: HPV-positive (HPV+), driven by human papillomavirus (HPV) infections, and HPV-negative (HPV-), associated with environmental risk factors. Despite molecular and clinicopathological differences, neither subtype has effective tailored therapies. Since high-risk HPV oncoproteins E6/E7 affect several epigenetic regulators, characterizing the epigenetic landscape of HPV+ and HPV- HNSCC may uncover novel subtype-specific biomarkers and therapeutic targets.

Methods: Histone post-translational modifications were profiled in HPV+ and HPV- HNSCC tissues and cell lines using super-SILAC mass spectrometry. The same analysis was performed and combined with RNA-sequencing on E6/E7-transduced human primary keratinocytes (HKs) to identify relevant histone modifiers affected by HPV oncoproteins. Candidate gene was validated via E6/E7-mediated-siRNA knockdown in HPV + cell lines. Western Blot, RT-qPCR and Immunohistochemistry assessed gene expression. NSD2 expression was examined in patients’ tissue samples, TCGA data and 14 HNSCC cell lines. shRNA-mediated NSD2 knockdown followed by RNA-seq, cell proliferation and migration assays evaluated its oncogenic role in HNSCC. CaCl₂ treatments were used to investigate NSD2’s role in epithelial differentiation, while ALDH-positive cells were quantified by flow-cytometry. NSD2 overexpression was used to confirm results.

Results: HPV+ HNSCC exhibited elevated H3K36me2 levels, compared to HPV-. This alteration is driven by E6/E7-induced NSD2 upregulation. NSD2, a histone methyltransferase specific for H3K36 di-methylation is overexpressed in HPV+ relative to HPV- HNSCC and in both subtypes compared to normal tissue, suggesting crucial implications in HNSCC. Functional assays revealed that NSD2 promotes cell proliferation and migration in both the subtypes. Notably, we identified a novel role for NSD2 in inhibiting epithelial cell differentiation, particularly in HPV+ HNSCC, where its upregulation mediates the E6/E7-induced differentiation blockade.

Conclusions: We identified a novel HPV-driven epigenetic signature in HNSCC marked by increased H3K36me2 and its writer, NSD2. Our study highlights H3K36me2 as a potential biomarker for patient stratification and positions NSD2 as a promising therapeutic target across HNSCC subtypes, modulating both common and subtype-specific oncogenic pathways. Specifically, NSD2 inhibition in HPV+ tumors restores epithelial differentiation, offering a potential strategy to arrest tumor progression.

Supplementary Information: The online version contains supplementary material available at 10.1186/s13046-025-03631-0.

Keywords: Epigenetics; Epithelial cell differentiation.; Head and neck cancer (HNC); Head and neck squamous cell carcinoma (HNSCC); Histone post-translational modifications (hPTMs); Human papillomavirus (HPV); NSD2.

Fig. 1

MS-based profiling of histone PTMs in HPV- and HPV + HNSCC samples. A-B) Heatmaps showing the levels of the indicated differentially modified histone peptides, which are expressed as L/H ratios (where L = sample and H = internal standard). The hPTMs were normalized to the average values across the samples. Crossed boxes represent non-assessed values. The panel on the right, with its related color legend, indicates the statistically significant differences in HPV + vs. HPV- samples, calculated through an unpaired Mann-Whitney test. hPTMs analyzed in 9 HPV- and 19 HPV + HNSCC FFPE patients’ tissue samples and 8 HPV- and 67 HPV + HNSCC cell lines are respectively shown in Figure A and B. C) hPTMs analysis was performed on Human Primary keratinocytes (HKs) derived from the healthy skin of 5 different donors and transduced with empty and HPV-16 E6/E7 encoding plasmid. Heatmap showing the L/H ratios for the indicated differentially modified histone peptides for E6/E7 overexpressing HKs normalized over the empty control. Crossed boxes represent non-assessed values. The panel on the right, with its related color legend, indicates significant or close to significant differences for each hPTM in E6/E7 overexpressing HKs vs. EMPTY control by paired t-test. D-F) Graphs showing the H3K36me2 levels detected by MS in HPV- HNSCC and HPV + HNSCC FFPE patients’ tissue samples; Unpaired Mann-Whitney test (D), in HPV- HNSCC and HPV + HNSCC cell lines; Unpaired t-test (E), and in HKs transduced with the empty or E6/E7 overexpressing plasmid; paired t-test (F). Values are shown as log2 of L/H ratios. * p < 0,05.

Fig. 2

HR-HPV E6/E7 regulate NSD2 expression levels in HKs and HNSCC cell lines. A-C) HKs were transduced with empty or HPV-16 E6/E7 encoding vectors and harvested after selection. A) Total RNA was extracted and RNA-seq was performed on HKs derived from 4 donors and transduced with HPV16-E6/E7. The Volcano Plot shows the log2FC values (over the empty control) of the major known histone modifiers. The legends on the right indicate the main known targets (colors) and the function (shape) of the plotted histone modifiers. For the most strongly regulated genes, the corresponding names are reported in the figure. B) Total RNA was extracted and analyzed by RT-qPCR. The histogram shows the mRNA levels of NSD2 normalized on the housekeeping gene RPLP0. Values of 5 independent experiments, normalized relative to the empty group average, are expressed as means ± SD. Unpaired t-test with Welch’s correction. C) Western blot showing NSD2 levels in E6/E7-transduced HKs. TP53 was used as a surrogate marker of E6 overexpression. Vinculin and β-actin were used as loading controls. D-E) HPV + HNSCC cell lines were transfected with siE6/E7 or siLuc control and harvested after 72 h. D) Lysates were processed and analyzed by immunoblot with the indicated antibodies. Vinculin was used as a loading control. E Total RNA was extracted and analysed by RT-qPCR. For each cell line NSD2 mRNA levels were normalized on the RPLP0 housekeeping gene and expressed as means (± SD) of fold changes of at least 2 independent experiments. One sample t-test. F-G) HKs were transduced with the recombinant retroviral vector pLXSN encoding for the E6/E7 of low risk (HPV-6, HPV-10) and high-risk (HPV-16, HPV-18) HPV genotypes or with the empty vector as a control. F) Histogram showing the mRNA levels of NSD2 normalized on the housekeeping gene RPLP0. Values of 3 independent experiments, normalized relative to the empty control group average, are expressed as means ± SD. Unpaired t-test. G) Lysates were processed and NSD2 protein levels were analyzed by immunoblot. β-actin was used as a loading control. * p < 0,05; ** p < 0,01; ns, not significant

Fig. 3

NSD2 expression levels in HNSCC cell lines and patients’ tissue samples. A) Immunoblot showing the NSD2 protein levels in 7 HPV- and 7 HPV+ HNSCC cell lines. Vinculin was used as a loading control. The histogram on the right shows the optical densitometric quantification of NSD2 bands normalized to Vinculin. Plotted results are the average of three independent experiments (± SD). Unpaired t-test with Welch’s correction. B) Total mRNA was extracted from 8 HPV- (the same of Fig. 3A plus UM-SCC-18) and 7 HPV+ HNSCC cell lines.The histogram shows the NSD2 mRNA levels of each group, analyzed by RT-qPCR and normalized on the housekeeping gene RPLP0. Each dot represent the average value of three independent replicates of each cell line. Values are expressed as means ± SD. Unpaired t-test. C) Correlation between the H3K36me2 levels, measured by MS, and NSD2 levels, measured by densitometric quantification of immunoblot bands, of HPV- and HPV+ HNSCC cell lines: Spearman correlation coefficient r = 0,62; p-value = 0,047. D) Representative images of NSD2 immunohistochemistry performed on 9 HPV- and 13 HPV+ HNSCC patients’ specimens. The histogram on the right shows the quantification of NSD2 signal, measured as DAB mean intensity. Values are expressed as means ± SD. Unpaired t-test. E) Total mRNA was extracted from 11 normal, 7 HPV- and 4 HPV+ HNSCC cryopreserved specimens. The histogram shows the NSD2 mRNA levels of each group, analyzed by RT-qPCR and normalized on the housekeeping gene RPLP0. Values are expressed as means ± SD. one-way ANOVA, multiple comparison test. F) NSD2 expression levels of 44 normal, 415 HPV- and 72 HPV+ HNSCC specimens from the TCGA Pancancer 2018 dataset. Data were downloaded from the “GDC Data Portal”. Values are expressed as means ± SD. one-way ANOVA multiple comparison test. G) NSD2 levels of HNSCC specimens from TCGA Pancancer 2018 dataset, clusterized according to the histological grade (G1, G2, G3, G4). Values are expressed as means ± SD. one-way ANOVA multiple comparison test. H) NSD2 levels of HPV- and HPV+ HNSCC specimens from TCGA Pancancer 2018 dataset, clusterized according to the histological grade (G2, on the left; G3, on the right). Values are expressed as means ± SD. Unpaired t-test. *, p ≤ 0,05; ** p < 0,01; *** p < 0,001; ****, p ≤ 0,0001; ns, not significant

Fig. 4

Silencing NSD2 reduces cell proliferation in HPV + and HPV- HNSCC cell lines. 4 HPV- and 4 HPV + HNSCC cell lines were transduced with shNSD2 (_A) and scrambled control. A) Immunoblots showing NSD2 and H3K36me2 protein levels in HNSCC cell lines upon NSD2 silencing. β-actin and H4 total were respectively used as housekeeping controls. B) Graph showing NSD2 mRNA levels of NSD2-silenced HNSCC cell lines, analyzed by RT-qPCR and normalized on the RPLP0 housekeeping gene. Values are represented as fold changes on the scrambled control and expressed as means ± SD. One-sample t-test. C) Cell proliferation was assessed on 4 HPV- (in blue) and 4 HPV+ (in red) NSD2-silenced HNSCC cell lines. Cells were counted at day 0, 2, 4, 6 and 8. Results of at least two replicates are shown. Unpaired t-test. D) Colony Formation Assay performed on 4 HPV- (in blue) and 4 HPV+ (in red) NSD2-silenced HNSCC cell lines. Representative images are reported for each condition and the respective histograms on the right show the quantified number of colonies/well. Results are the average of at least two replicates. Unpaired t-test. *, p ≤ 0,05; ** p < 0,01; *** p < 0,001; ****, p ≤ 0,0001; ns, not significant.

Fig. 5

Silencing NSD2 reduces cell migration in HPV + and HPV- HNSCC cell lines. 4 HPV- and 4 HPV + HNSCC cell lines were transduced with shNSD2 (_A) and scrambled control. A)Histograms show the mRNA levels of mesenchymal markers (Vimentin, N-cadherin, Fibronectin and Snai2), analyzed by RT-qPCR, in 4 HPV- (in blue) and 4 HPV+ (in red) HNSCC cell lines. Data were normalized on RPLP0. Values of at least two independent experiments are represented as fold changes on the scrambled control and expressed as means ± SD. One-sample t-test. B) Immunoblots showing Vimentin levels in 4 HPV- and 4 HPV+ NSD2-silenced HNSCC cell lines. β-actin was used as housekeeping. C) Representative phase contrast pictures of wound healing assays at time 0 and 24 or 48 h, on two HPV- (in blue) and two HPV+ (in red) HNSCC upon NSD2 silencing. Graphs represent the percentage of wound closure expressed as means ± SD. For each cell line values were obtained from at least three replicates and normalized relative to the scrambled group average. Unpaired t-test. D) Representative phase contrast pictures of transwell migration assays performed on an HPV- and an HPV + HNSCC cell line upon NSD2-silencing. Cells were fixed and stained respectively after 24–48 h. Graphs represent the quantified migrated cells: for each cell line values were obtained from at least two replicates and normalized relative to the scrambled group average. Values are expressed as means ± SD. Unpaired t-test. *, p ≤ 0,05; **, p ≤ 0,01; ***, p ≤ 0,001; ****, p ≤ 0,0001; ns, not significant (t-test).

Fig. 6

FNSD2 overexpression increases cell proliferation and cell migration in HPV + and HPV- HNSCC cell lines. NSD2 was overexpressed in 2 HPV- (UM-SCC-4, UM-SCC-18) and 2 HPV+ (UM-SCC-47, 93-VU147-T). A) Western Blots showing the NSD2 and H3K36me2 protein levels in 2 HPV- and 2 HPV + HNSCC cell lines upon NSD2 overexpression. Vinculin and H4 total were used as loading controls. B) Cell proliferation has been assessed in 2 HPV- (in blue) and 2 HPV+ (in red) HNSCC cell lines upon NSD2 overexpression. Cells were counted every 2 days for five time points (day 0, 2, 4, 6, 8). Values of two replicates are shown. Unpaired t-test. C) Colony Formation Assay performed on 2 HPV- (in blue) and 2 HPV+ (in red) NSD2-overexpressing HNSCC cell lines. Representative images are reported for each condition and the respective histograms on the right, show the quantified number of colonies/well. Results are the average of at least two replicates. Unpaired t-test. D) Western Blots showing the Vimentin protein levels in 2 HPV- and 2 HPV + HNSCC cell lines, upon NSD2 overexpression. Vinculin was use as a loading control. E) Total mRNA was extracted from an HPV- (UM-SCC-4) -in blue- and an HPV+ (93-VU147-T) -in red- HNSCC cell line upon NSD2 overexpression. Histogram show the mRNA expression levels of Vimentin, N-cadherin, Fibronectin and Snai2 analyzed by RT-qPCR and normalized on the RPLP0 housekeeping gene. Values of at least three independent experiments are represented as fold changes on the empty vector control and expressed as means ± SD. One-sample t-test. F) Phase contrast representative pictures of wound healing assays performed on an HPV- (UM-SCC-4) and an HPV+ (93-VU147T) HNSCC cell line transduced with the empty or the NSD2-overexpressing vector. Images were acquired at time 0 and 24 h. Graphs on the right represent the percentage of wound closure. For each cell line values from at least three replicates were normalized relative to the empty group average and are expressed as means ± SD. Unpaired t-test. *, p ≤ 0,05; **, p ≤ 0,01; ***, p ≤ 0,001; ****, p ≤ 0,0001 (t-test).

Fig. 7

NSD2 silencing affects common and distinct pathways in HPV+ and HPV- HNSCC cell lines. A) RNA-seq was performed on 4 HPV- (UM-SCC-4, UM-SCC-6, UM-SCC-18, UM-SCC-19) and 3 HPV+ (UD-SCC-2, UM-SCC-104, UPCI: SCC-152) HNSCC cell lines transduced with shNSD2_A or scr control. Heatmap showing the DEGs in HPV- and HP+ HSNCC cell lines upon NSD2 silencing. K-means clustering has been applied with a number of clusters k = 9. The red-green scale represents the log2FC values, while the blue and white boxes indicate if the gene is statistically significant or not, respectively, for each of the two subgroups. Color legend indicates the number of each cluster. B) Gene Ontology (GO) analysis was performed on identified clusters. Bar plots show significant gene ontologies enriched in Cluster 2, 5 and 9. C) Total mRNA was extracted from the HPV+ HNSCC cell lines, UM-SCC-104 and UPCI:SCC-152, upon NSD2 silencing. Histograms showing the mRNA expression levels of a panel of differentiation markers analyzed by RT-qPCR and normalized on RPLP0. Values of at least three independent experiments are represented as fold changes on the scrambled control and expressed as means ± SD. One-sample t-test. D) Graph showing the ΔNp63α mRNA expression levels, analyzed by RT-qPCR and normalized on RPLP0. Values of at least two independent experiments, represented as fold changes on the scrambled control, are expressed as means ± SD. One-sample t-test. E) Immunoblot showing ΔNp63α protein levels in 4 HPV + HNSCC cell lines, upon NSD2 silencing. To optimize the visualization, both low exposure and high exposure acquisition are shown. Vinculin was used as loading control. The histogram on the right, shows the optical densitometric quantification of ΔNp63α bands normalized on Vinculin. F) Immunoblot showing NSD2 and ΔNp63α protein levels in 7 HPV- and 7 HPV+ HNSCC cell lines. Vinculin was used as loading control. G) Correlation between ΔNp63α and NSD2 protein levels measured by densitometric quantification of Western blot bands in HPV- (Spearman correlation coefficient r=−0,714; p-value = 0,08) and HPV+ HNSCC cell lines (Spearman correlation coefficient r = 0,924; p-value = 0,006). H) ALDH + cells were detected and quantified through the ALDEFLUOR assay in UPCI:SCC-152 cell line, upon NSD2-silencing, and acquired by FACS analysis. Treatment with DEAB inhibitor was used as negative control. The histogram shows the mean of four independent experiments ± SD. Paired t-test. I) Sphere formation assay was performed in UPC:SCC-152 cell line upon NSD2-silencing. Images were acquired 15 days post-plating; spheres > 70 μm in diameter were counted. The histogram shows the mean of two independent experiments ± SD. Unpaired t-test. *,p ≤ 0,05; **,p ≤ 0,01; ns, not significant (t-test).

Fig. 8

NSD2 regulates epithelial cell differentiation and mediates the E6/E7-induced differentiation blockade. A-C) HPV16-E6/E7 oncoviral proteins were overexpressed in Human Primary Keratinocytes, that were also transduced with shNSD2_A or scrambled control and treated with 1,2 mM CaCl₂. A) Representative phase contrast images showing the morphological changes of HKs upon the indicated treatments. B) Immunoblot showing changes in NSD2 and IVL protein expression levels. P53 was used as a surrogate marker of E6 overexpression. GAPDH was used as loading control. C) Histograms showing the mRNA expression levels of the differentiation markers IVL, TGM1, S100A8 and S100A9, analyzed by RT-qPCR and normalized on RPLP0. Values of two technical replicates are expressed as means ± SD. D) Total mRNA was extracted from HKs overexpressing NSD2. The histogram shows the mRNA levels of ΔNp63α, IVL, TGM1, S100A8, S100A9, SPRR2A, SPRR2E, SPRR3 and FLG, analyzed by RT-qPCR and normalized on RPLP0. Values of at least three independent experiments are represented as fold changes on the empty vector control and expressed as means ± SD. One sample t-test. E) Representative confocal images of NSD2-overexpressing HKs. Phalloidin (red) was stained to show cell size and morphology, Nuclei were stained with DAPI (blue) and NSD2 with Alexa 488 (green). The graph shows the difference in cell size between NSD2-overexpressing HKs and control. Cells’ area was quantified using phalloidin and nuclei staining was used for cellular segmentation. Values are expressed as mean ± SD. Unpaired t-test. F-G) HKs transduced with the NSD2-encoding vector or with the empty control were treated with 1,2 mM CaCl₂. F) Representative phase contrast pictures showing the morphological changes of NSD2-overexpressing HKs upon 1,2 mM CaCl₂ treatment. G) Immunoblot showing changes in IVL, ΔNp63α, H3K36me2 protein levels. β-actin was used as loading control. H) Panel showing representative images of IHC performed on normal and HNSCC patients’ FFPE tissue samples and acquired with the Nanozoomer brightfield microscopy. Specimens were stained for NSD2, Ki67, ΔNp63α, IVL and TGM1. Both HPV- and HPV+ HNSCC cases are tumors of Grade3 (G3). I) RNA-seq data from TCGA Dataset were downloaded and analysed through the SRPlot software. Graphs showing the correlation between NSD2 and IVL or TGM1. Calculated Spearman correlation is −0,39 (p < 0,0001) for NSD2 and IVL and − 0,25 (p < 0,0001) for NSD2 and TGM1. *, p ≤ 0,05; **, p ≤ 0,01; ***, p ≤ 0,001; ****, p ≤ 0,0001 (t-test)