Genomic and transcriptomic profiling of radioresistant prostate and head and neck cancers implicate a BAHD1-dependent modification of DNA damage at the heterochromatin

Abstract

Radiotherapy is an integral modality in treating human cancers, but radioresistance remains a clinical challenge due to the involvement of multiple intrinsic cellular and extrinsic tumour microenvironment factors that govern radiosensitivity. To study the intrinsic factors that are associated with cancer radioresistance, we established 4 radioresistant prostate (22Rv1 and DU145) and head and neck cancer (FaDu and HK1) models by irradiating their wild-type parentals to 90 Gy, mimicking the fractionated radiotherapy schema that is often using in the clinic, and performed whole exome and transcriptome sequencing of the radioresistant and wild-type models. Comparative genomic analyses detected the enrichment of mismatch repair mutational signatures (SBS6, 14, 15, 20) across all the cell lines and several non-synonymous single nucleotide variants involved in pro-survival pathways. Despite significant inter-cell type heterogeneity of their transcriptomic profiles, 18 common dysregulated genes (5 upregulated and 13 downregulated) were identified across the 4 models, including the overexpression of bromo-adjacent homology domain containing 1 (BAHD1) gene, which is involved in heterochromatin formation. Interestingly, this coincided with our observation of increased histone 3 lysine 9 trimethylation (H3K9me3) and histone 3 lysine 27 trimethylation (H3K27me3) expression post-irradiation in our radioresistant cells. The dependency between BAHD1 and heterochromatin formation was confirmed by siRNA knockdown of BAHD1, indicating preferential reduction of H3K9me3 and H3K27me3 expression in the radioresistant cells, but not the wild-type parentals, and confirmed by clonogenic assays showing reversal of radioresistance post-siBAHD1 treatment. We further showed that inhibition of the BAHD1-heterochromatin formation axis led to reduced DNA double-strand break repair. Finally, analyses of treatment outcomes in 4 prostate and head and neck cancer radiotherapy cohorts suggested an increased risk of failures in tumours of high heterochromatin activity. Taken together, our results support a new model implicating BAHD1-dependent modulation of the heterochromatin in acquired radioresistance of prostate and head and neck cancers.

Fig. 1. Genomic profiling of RR prostate and head and neck cancer cells.

A Schema of generation and characterisation of radioresistant cell lines with downstream experiment plan (created with Biorender.com). B Total counts of SNV for four RR cell lines included synonymous and non-synonymous SNV. C Distribution of COSMIC mutational signatures identified by all SNVs in each cell line. Only SBS14 is shown for HK1, as other mutational signatures were removed due to overfitting caused by the limited number of SNVs in HK1. D The annotated non-synonymous SNV grouped by the related function in DNA repair, chromatin remodelling, and stemness pathways curated from Gene Ontology libraries. RR radioresistant, SNV single nucleotide variant.

Fig. 2. Transcriptome profiling of RR prostate and head and neck cancer cells.

A Differentially expressed genes (DEGs) in 22Rv1 and DU145, and (B) in FaDu and DU145. The DEGs were defined if the adjusted P < 0.05. C Expression profile of 221 common DEGs with consistent dysregulation direction in both 22Rv1 and DU145. D Expression profile of 1420 common DEGs with consistent dysregulation direction in both FaDu and HK1. Samples were ordered by cell lines and phenotype in both figures. E 18 DEGs with consistent dysregulation direction across four cell lines. The colour of each dot indicates the dysregulation direction, with red indicating upregulation and blue indicating downregulation. The size of each dot varies based on its fold change. The intensity of colour within each box reflects the range of P-values. F Significant pathways involved by the 5 upregulated DEGs across the 4 cell lines, curated from over-representation analysis (false discovery rate <0.1). The colour intensity within each colour filled box represents the rich factor of the gene in the pathway.

Fig. 3. Characterisation of the DNA damage responses and heterochromatin status of 22Rv1- and FaDu-RR cells relative to the parental WT cells.

A The representative western blot showed the changes in the expression of DSB, DNA repair, cell cycle arrest and heterochromatin markers under different time points post-4 Gy IR (normalised against control), GAPDH was used as a loading control. B and C-top panel Representative images of H3K9me3 (green) and H3K27me3 (red) foci in 22Rv1 cells at 0 and 1 h treatment time points and FaDu cells at 0 and 6 h treatment time points. Scale bar: 5 μm. B and C-bottom panel H3K9me3 intensity was quantified in AUCs and represented as percentage frequencies, compared between RR and WT; H3K27me3 foci were quantified as mean foci per cell, bars represent mean±SD, n = 3 per group. DSB DNA double-strand break, AUC area under the curve, WT wild-type, SD standard deviation.

Fig. 4. The heterochromatin status with and without siBAHD1 treated-22Rv1-RR and FaDu-RR cells were relative to the parental WT cells, and their clonogenic survivability.

A and B-top left Representative images of H3K9me3 (green) and H3K27me3 (red) foci in 22Rv1 cells at 1 h and FaDu cells at 6 h post-siBAHD1 treatment. Scale bar: 5 μm; (top right) Representative western blot showed the changes in H3K9me3 and H3K27me3 protein levels in 22Rv1 cells at 1 h and FaDu cells at 6 h post-siBAHD1 treatment; (bottom panel) H3K9me3 intensity was quantified in AUCs and represented as percentage frequencies, compared between RR and WT; H3K27me3 foci were quantified as mean foci per cell, bars represent mean±SD, n = 3 per group. C Colony forming assay of 22Rv1 and FaDu cells under different IR dosages, with and without siBAHD1 treatment, n = 3 per group. Asterisk indicates significance between siBAHD1-treated and control groups. Asterisk indicates P < 0.05.

Fig. 5. Reduced BAHD1 expression improved the radiosensitivity of 22Rv1-RR and FaDu-RR cells relative to the parental WT cells.

A A hypothesised model showing BAHD1 contributes to the enhanced heterochromatin response in RR cancer cells, whereby increased repair efficiency following irradiation (IR)-induced damage leads to decreased radiosensitivity. The inhibition of BAHD1 leads to chromatin unpacking, decreased repair efficiency and improved radiosensitivity (created with Biorender.com). B and C, left Representative images of co-localised γH2AX (green) and p-53BP1 (red) foci in 22Rv1-RR and FaDu-RR cells at 1 and 6 h post-IR, with and without siBAHD1 treatment. Scale bar: 5 μm; (right) Quantification of co-localised γH2AX and p-53BP1 mean foci per cell, bars represent mean±SD, n = 3 per group. D, E Representative western blot showed the changes in the expression of DSB, DNA repair, cell cycle arrest and heterochromatin markers of 22Rv1-RR cells and FaDu-RR cells at 1 and 6 h post-IR, with and without siBAHD1 treatment (normalised against control).

Fig. 6. Increased heterochromatin formation associated with risk of relapse in prostate cancer and head and neck cancer patients after radiotherapy.

Kaplan-Meier curves of the patients’ stratified heterochromatin activity in (A) NCCS prostate cancer cohort (n = 151); (B) Berlin et al. prostate cancer cohort from GRID (n = 121); (C) NCCS head and neck cancer cohort (n = 158); (D) Zhang et al. head and neck cancer (GSE102349) cohort (n = 88). The hazard ratios and 95% confidence intervals were computed by the COX proportional hazard model, and survival curves were compared by the log-rank test.