TOP1 and R-loops facilitate transcriptional DSBs at hypertranscribed cancer driver genes

Abstract

DNA double-stranded breaks (DSBs) pose a significant threat to genomic integrity, and their generation during essential cellular processes like transcription remains poorly understood. In this study, we employ several techniques to map DSBs, R-loops, and topoisomerase 1 cleavage complex (TOP1cc) to comprehensively investigate the interplay between transcription, DSBs, topoisomerase 1 (TOP1), and R-loops. Our findings reveal the presence of DSBs at highly expressed genes enriched with TOP1 and R-loops. Remarkably, transcription-associated DSBs at these loci are significantly reduced upon depletion of R-loops and TOP1, uncovering the pivotal roles of TOP1 and R-loops in transcriptional DSB formation. By elucidating the intricate interplay between TOP1cc trapping, R-loops, and DSBs, our study provides insights into the mechanisms underlying transcription-associated genomic instability. Moreover, we establish a link between transcriptional DSBs and early molecular changes driving cancer development, highlighting the distinct etiology and molecular characteristics of driver mutations compared to passenger mutations.

Keywords: Biological sciences; Cancer; Molecular mechanism of gene regulation; Transcriptomics.

Graphical abstract

Figure 1

Breaks are enriched at highly expressed genes (A) The distribution of DSBs in breast cancer MCF7 cells along ChromHMM-defined chromatin states of human mammary epithelial cells (HMEC). (B) A zoomed log scaled scatterplot showing a positive correlation between break density and expression levels; each dot represents a gene, and for each gene break density and expression were measured and normalized to gene size. (C) Mean expression (TPM) positively correlates with break density, genes were grouped into 22 groups with increasing break enrichment, the 22nd group being the group with the highest mean break enrichment. (D) Ratio of observed breaks vs. expected breaks at highly and low expressed genes. (E) Plot showing break density at gene body of the 1,000 most transcribed genes (continuous line) and 1,000 least transcribed genes (dashed line). sBLISS experiments were done in duplicates, and CEL-seq was done in quadruplicates. P-values were calculated by chi-square test and correlation coefficients by Spearman’s correlation test. ∗∗∗p < 0.001. The expected number of breaks in a set of regions is calculated based on the assumption that breaks are distributed randomly across the genome.

Figure 2

Transcriptional DSBs are associated with TOP1 (A) A log scaled scatterplot showing positive correlation between expression and TOP1 levels for the top 50% expressed genes; each dot represents a gene, and each color represents a gene group used in B. (B) Median of expression (TPM) correlates with TOP1 levels, for the top 50% expressed genes. (C) Heatmap of break density at TOP1-binding regions across the genome, the plot shows DSBs at the center of TOP1 peaks. (D) Ratio of observed breaks to expected breaks for low TOP1 and high TOP1 genes. (E) Representative images of immunofluorescent staining of MCF-7 ctrl (siSc) and TOP1-depleted (siTOP1) MCF-7 cells using γ-H2AX and 53BP1 antibodies. (F) Quantification of the number of γ-H2AX and 53BP1 foci per nucleus in MCF-7 ctrl and MCF-7 siTOP1 (E). sBLISS and immunofluorescence experiments were done in duplicates. P-values for high TOP1 vs. low TOP1 were calculated by chi-square test, P-value for immunofluorescence staining was calculated using Student’s t test, and correlation coefficients by Spearman’s correlation test. ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001.

Figure 3

Transcriptional DSBs are associated with R-loops (A) A log scaled scatterplot showing positive correlation between expression and R-loops levels for the top 50% expressed genes, each dot represents a gene. DRIP data were extracted from GEO: GSE81851. (B) Mean expression (TPM) correlates with R-loops levels for the top 50% expressed genes. (C) Heatmap of break density at R-loops regions across the genome, the plot shows DSBs at the center of R-loops peaks. (D) Ratio of observed breaks to expected breaks for low R-loops and high R-loops genes. (E) Representative images of immunofluorescent staining of MCF-7 EV and MCF-7 overexpressing RNase H1 for γ-H2AX and 53BP1. (F) Number of γ-H2AX and 53BP1 foci per nucleus, for MCF-7 EV and MCF-7 RNase H1. sBLISS and immunofluorescence experiments were done in duplicates. P-values for high TOP1 vs. low TOP1 were calculated by chi-square test, P-value for immunofluorescence staining was calculated using Student’s t test, and correlation coefficients by Spearman’s correlation test. ∗p < 0.05, and ∗∗∗p < 0.001.

Figure 4

RNase H1 overexpression and TOP1 knockdown decrease break enrichment at gene body of highly expressed genes (A) The distribution of DSBs in breast cancer MCF7 cells along ChromHMM-defined chromatin states of HMEC in control cells (transfected with scramble RNA and infected with EV-GFP), cells knocked down for TOP1 (also infected with EV-GFP), cells overexpressing RNase H1 (also transfected with scramble RNA), and cells knocked down for TOP1 and overexpressing RNase H1. Bar height is break enrichment relative to other chromatin states. (B) Breakome percentage in 1000 most transcribed genes (top) and 1,000 least transcribed genes (bottom) with the different manipulations. Expression data are taken from our CEL-seq. (C) Breakome percentage in 100 top R-loop-enriched genes (top) and 100 top R-loop-deprived genes (bottom) with the different manipulations. (D) Breakome percentage in 100 top TOP1-enriched genes (top) and 100 top TOP1-deprived genes (bottom) with the different manipulations. (E) Color-coded heatmap representing the change in break density for the top 1,000 expressed genes. The break density values have been standardized using Z scores for rows to enhance the visualization of relative differences. (F) Color-coded heatmap representing the change in break density for the top 2,000 break-prone genes. (G) Plot showing break density at gene body of highly transcribed genes shown in B with the different manipulations. sBLISS experiments were done in duplicates. P-value was calculated by chi-square test. n.s. p > 0.5, and ∗∗∗p < 0.001.

Figure 5

Estradiol-associated DSBs are mediated by R-loops and TOP1 (A) Representative images of immunofluorescent staining of MCF-7 EV non-treated, MCF-7 EV treated with estradiol (E2), and MCF-7 cells overexpressing RNase H1 and treated with E2, for γ-H2AX and 53BP1. (B) Quantification of number of γ-H2AX and p53BP1 foci per nucleus, for MCF-7 EV NT and MCF-7 EV treated with E2, and MCF-7 cells overexpressing RNase H1 treated with E2. (C) Representative images of immunofluorescent staining of MCF-7 siSc non-treated, MCF-7 siSc treated with estradiol (E2), and MCF-7 cells knocked down for TOP1 and treated with E2, for γ-H2AX and 53BP1. (D) Quantification of number of γ-H2AX and p53BP1 foci per nucleus, for MCF-7 siSc NT and MCF-7 siSc treated with E2, and MCF-7 cells knocked down for TOP1 and treated with E2. (E) Color-coded heatmap showing the change in break density between control cells (vehicle and EV-GFP), cells incubated with estradiol and infected with EV-GFP, cells incubated with estradiol and overexpressing RNase H1, for estrogen-responsive genes with the highest positive differential break density. (F) Color-coded heatmap showing the change in break density between control cells (transfected with scramble RNA and treated with vehicle), cells incubated with estradiol and transfected with scramble RNA, cells knocked down for TOP1 and treated with estradiol, for estrogen-responsive genes with the highest positive differential break density. sBLISS and immunofluorescence experiments were done in duplicates. P-value was calculated using Student’s t test ∗p < 0.05, and ∗∗p < 0.01.

Figure 6

TOP1 KD and/or RNase H OE decrease TOP1cc at highly expressed genes (A) Scatterplot showing the correlation between TOP1 and TOP1cc density at the top 50% expressed genes. (B) Boxplot showing TOP1 density mean at high TOP1cc genes vs. low TOP1cc genes. (C) Boxplot showing break density at high TOP1cc genes vs. low TOP1cc genes. (D) Distribution of TOP1cc along the gene bodies of highly expressed genes. (E) Representative images of immunofluorescence staining of TOP1cc after TOP1 KD. (F) Quantification of TOP1cc intensity. (G) Color-coded heatmap showing the change in TOP1cc density with the different manipulations at highly expressed genes. (H) Stacked bar plot showing the percentage of genes exhibited a decrease/no change/increase in TOP1cc density for highly expressed genes that decreased in break density upon the different manipulations. (I) Boxplot showing R-loops density mean at high TOP1cc genes vs. low TOP1cc genes. (J) TOP1cc enrichment over background at R-loop voids, random R-loop peaks, and the top 1000 R-loop peaks. (K) Graphical illustration of the proposed mechanism of transcriptional DSBs, optimally, TOP1 participates in efficient cleavage and re-ligation reactions maintaining the native structure of DNA and preventing genomic instability. When co-transcriptional R-loops are formed, the engaged TOP1 have higher chance of trapping in its TOP1cc form, leaving a single-stranded break (SSB) behind. Along with the SSB formed as a result of R-loops processing, a DSB is formed during transcription. Correlation coefficient was calculated by Pearson’s test. P-value was calculated by Mann-Whitney test. P-value of the IF was calculated by Student’s t test. ∗∗p < 0.01, and ∗∗∗p < 0.001. The horizontal bars represent the median and whiskers extend from –1.5 × IQR to +1.5 × IQR from the closest quartile, where IQR is the inter-quartile range.

Figure 7

Cancer genes are enriched with transcriptional DSBs (A) Boxplot of mutation frequency at driver genes and frequently mutated passenger genes. (B) Boxplot of break density at driver genes and frequently mutated passenger genes. (C) Boxplot of normalized expression at driver genes and frequently mutated passenger genes. (D) Boxplot of TOP1 density at driver genes and frequently mutated passenger genes. (E) Boxplot of R-loops density at driver genes and frequently mutated passenger genes. (F) Log-scaled scatterplot demonstrating the correlation between break density and expression levels at driver genes. Red circles are identifying highly expressed drivers used in the heatmap in H, blue circles are identifying lowly expressed drivers used in the heatmap in I. (G) Color-coded heatmap of break density at driver genes with the different manipulations. (H) Color-coded heatmap of break density at highly expressed driver genes with the different manipulations. (I) Color-coded heatmap of break density at lowly expressed driver genes with the different manipulations. (J) Log-scaled scatterplot demonstrating the correlation between break density and expression levels at passenger genes. (K) Color-coded heatmap of break density at passenger genes with the different manipulations. (L) Color-coded heatmap demonstrating the direction of change of R-loops enrichment at driver genes after RNase H OE. (M) Color-coded heatmap demonstrating the direction of change of R-loops enrichment at passenger genes after RNase H OE. (N) Color-coded heatmap demonstrating the direction of change of TOP1cc enrichment at driver genes after TOP1 KD. (O) Color-coded heatmap demonstrating the direction of change of TOP1cc enrichment at passenger genes after TOP1 KD. (P) Stacked bar graph demonstrating percentages of genes that decreased, didn’t change, or increased in the heatmaps mentioned earlier (L‒O). (Q) Genes that overlap the top lists of TOP1, breaks, TOP1cc, and R-loops with their available classification obtained from the literature. Gene enrichments are statistically significant and as follows (hypergeometric test), oncogenes PV = 1.103765e-27, tumor suppressors + dual function = 6.100708e-05, Actins = 2.772229e-05, Histones = 3.672962e-06. ∗∗∗p < 0.001. Correlation coefficient was calculated by Pearson’s test. P-value was calculated by Mann-Whitney test. The horizontal bars represent the median and whiskers extend from –1.5 × IQR to +1.5 × IQR from the closest quartile, where IQR is the inter-quartile range.