p53-mediated regulation of LINE1 retrotransposon-derived R-loops

Abstract

Long interspersed nuclear element 1 (LINE1/L1) retrotransposons, which comprise 17% of the human genome, typically remain inactive in healthy somatic cells but are reactivated in several cancers. We previously demonstrated that p53 silences L1 transposons in human somatic cells, potentially acting as a tumor-suppressive mechanism. However, the precise molecular mechanisms underlying p53-mediated repression of L1 and its life cycle intermediates remain unclear. In this study, we used DNA-RNA immunoprecipitation-sequencing experiments to investigate RNA-DNA hybrids, which are key intermediates formed during L1 retrotransposition. Our findings reveal that L1 mRNA-genomic DNA (cis L1 R-loops) and L1 mRNA-complementary DNA (trans L1 R-loops) hybrids are upregulated in p53^−/− cells. This increase is synergistic with L1 activation by histone deacetylase (HDAC) inhibitors (HDACi). However, treatment with a reverse transcriptase inhibitor reduces this accumulation, indicating that retrotransposition activity plays a significant role in R-loop accumulation. Interestingly, in WT cells, hyperactivated L1 transposons are suppressed upon HDACi withdrawal. L1 suppression in WT cells coincided with the recruitment of repressive marks, specifically H3K9me3 and H3K27me3, simultaneously preventing the addition of activating marks like H3K4me3, and H3K9ac at the L1 5′UTR. Mechanistically, we demonstrate that p53 cooperates with histone methyltransferases SETDB1 and G9A to deposit H3K9me3 marks at the L1 promoter, thereby silencing transposons. This study is the first to reveal novel roles of p53 in preventing the formation of L1-derived RNA-DNA hybrids (R-loops) and suppression of hyperactivated L1 elements by cooperating with histone methyltransferases, underscoring its critical role in maintaining genomic stability.

A substantial portion of the human genome is composed of non-long terminal repeat (non-LTR) retroelements called long interspersed element-1 (LINE-1 or L1), which is the only autonomously active retroelement in the human genome. LINE-1 encodes for two open reading frames (ORFs): ORF1 and ORF2. L1-ORF1p has an RNA-binding property, and ORF2p contains two enzymatic properties with reverse transcriptase (RT) and endonuclease (EN) activities (1, 2). LINE-1 elements propagate within the genome through retrotransposition, a process in which L1 mRNA is reverse-transcribed into complementary DNA (cDNA) and subsequently integrated into the genomic DNA (gDNA). During this process, RNA: DNA hybrids (R-loops) may form, potentially influencing genomic stability. Although L1 transposons are silenced in healthy somatic cells, they become activated in various human cancers and have recently been recognized as one of the hallmarks of cancer. Studies have established a direct correlation between increased retroelement expression and p53 mutations in cancers (3, 4, 5, 6). Recently it has been shown that p53 represses mobile genetic elements, suggesting a new axis of tumor suppression (7, 8, 9, 10). p53-mediated retroelement repression is conserved across various species, including flies, fish, and human somatic cell lines (4, 5, 10). Employing CRISPR-mediated wild-type (WT) and p53 knock-out (p53^−/−) cells, we have experimentally verified p53 as a direct transcriptional repressor of LINE1 (4). This is evidenced by p53 binding to the L1 5′UTR region, corroborating bioinformatic predictions of p53-binding sites in Alu sequences (11, 12, 13). The promoter regions of repetitive sequences are globally hypomethylated in various human cancers (3, 6), making these transcripts a potential source of R-loops (14, 15). During transcription, the persistence of an L1 mRNA-gDNA hybrid intermediate may lead to the formation of “cis R-loops” at L1 promoter regions. These R-loops, consisting of a three-stranded nucleic acid structure where an RNA/DNA hybrid is paired with a displaced single-stranded DNA (ssDNA), can lead to either physiological or pathological outcomes (16). L1-encoded ORF2p contains two functional domains essential for integration: the EN domain creates an ssDNA nick for integration, and the RT domain primes complementary cDNA synthesis based on the L1mRNA template. In this context, the reverse transcription activity of ORF2p creates an L1mRNA-cDNA hybrid structure, referred to as a “trans L1 R-loop” in this study. However, it remains to be investigated whether this intermediate structure is composed of three or four strands—a free ssDNA strand, a nicked DNA, L1mRNA serving as the template for reverse transcription, and the newly synthesized cDNA. Consistent with these, the recent structural findings on ORF2p identified domains that show extensive interactions with RNA templates (17).

Ectopic accumulation of (cis) R-loops, particularly those arising from transcriptional events, is associated with various genomic and epigenetic instabilities (18, 19, 20, 21, 22). Consequently, the accumulation of cis and trans R-loops derived from L1 sequences in p53-deficient cells can potentially lead to DNA damage and inflammation, an aspect of the oncogenic potential of L1 that remains unexplored. Furthermore, non-autonomous retroelements Alus (23, 24) and SINE-VNTR-Alu retrotransposons (SVAs) (25, 26), hijack L1 machinery for their retrotransposition; hence, activation of these elements can also participate in the generation of unscheduled R-loops. The activation of these retroelements in cancer cells, often in the context of p53 deficiency or other disruptions in genomic safeguards, suggests a complex interplay between L1 activity and genomic instability. A key insight from recent studies is the effect of treatment with hypomethylating drugs leading to the global induction of LINE1 expression in both cancer patients and cell lines (27, 28, 29) suggesting in addition to p53, several lines of defenses operate to silence these repetitive sequences in healthy somatic cells (29). This induction of L1 can lead to increased retrotransposition events, contributing to genomic instability through the generation of R-loops. Epigenetic mechanisms play a critical role in silencing L1 elements to preserve genomic integrity. For example, the loss of histone methyltransferases SETDB1 and G9A leads to the derepression of retrotransposable elements, thereby compromising genome integrity (30, 31). Similarly, SUV39H1 methyltransferase deposits H3K9me3 at the chromatin of evolutionarily young LTR and non-LTR transposons in mouse embryonic stem cells, ensuring their silencing (32). Understanding how p53 cooperates with histone methyltransferases remains an area of ongoing research, highlighting multiple silencing mechanisms that regulate LINE1 and its intermediate components, including R-loops formed during retrotransposition.

Herein, we demonstrate that the most recently evolved full-length retrotransposition-competent L1 elements are actively involved in the generation of trans R-loops. Our findings reveal that the overexpression of the RT domain of L1-ORF2p led to a notable increase in trans R-loop accumulation, which was accompanied by heightened levels of DNA damage and inflammation. Interestingly, treatment with a RT inhibitor attenuated these elevations in DNA damage and inflammation markers, underscoring the oncogenic implications of L1-derived trans R-loops. Our results also indicate that p53 in cooperation with SETDB1 and G9A promotes deposition of histone repressive marks at the L1 5′UTR, providing insights into potential therapeutic strategies for diseases associated with L1 activation.

Results

p53 controls L1 retrotransposition intermediary components

Similar to our previous observation (4), we noted increased levels of L1 transcripts (Figs. 1A and S1A) and ORF1 protein (Fig. S1, B and C) in p53^−/− cell lines of A375 and U2OS confirmed with Western blotting using p53 antibody (Fig. S1, D and E). We asked whether the observed increase in RNA-DNA hybrid levels in p53^−/−, relative to the WT controls, is attributable to derepressed L1 elements as shown in immunofluorescence (IF) experiments (Fig. 1, B and C). S9.6 antibodies (33), a widely used antibody for the visualization and detection of RNA-DNA hybrids, were used to conduct IF in WT and p53^−/− A375 cells. The results demonstrated an increased signal of cytoplasmic RNA-DNA hybrids in p53-deficient cells compared to the parental WT cells, corroborating to L1 derepression. Next, DNA-RNA immunoprecipitation (DRIP) assays were conducted utilizing the S9.6 antibody to investigate the propensity of L1 sequences in generating R-loops in p53^−/− cells. In this experiment, S9.6 antibody was used to precipitate RNA-DNA hybrids followed by quantification using quantitative polymerase chain reaction (qPCR) with the primers that binds the DNA region of interest (34). DRIP-qPCR using primers spanning 5′UTR, ORF1, and ORF2 regions (Fig. 1D) revealed that the L1 5′UTR region, which harbors RNA pol II promoter, was predominantly enriched in R-loops derived RNA-DNA hybrids, whereas ORF1 and ORF2 regions exhibited a modest enrichment (Fig. 1E). The specificity of our assay for detecting RNA-DNA hybrids was validated by pretreating the samples with RNase H before the antibody pulldown, which resulted in the complete loss of our ability to amplify the region (Fig. S1F). Given that the majority of the nuclear R-loop–derived RNA-DNA hybrids were observed in the cytoplasm, we leveraged an alternative method to specifically probe R-loops. This involved the overexpression of RNase H1 with D210N catalytic dead mutant (RNH-GFP) with the nuclear localization signal (NLS) fused at the N terminus in A375 WT and p53^−/− cells (19, 34, 35). Stable cell lines of RNH-GFP showed similar GFP expression between these genotypes (Fig. 1, F and G). qPCR experiments from GFP immunoprecipitated samples indicated stronger R-loop signal at L1 5′UTR and modest signals at L1-ORF1 and ORF2 in RNH-GFP cells, similar to S9.6 pull-down data (Fig. 1H). We used RPL13A as a positive control for R-loop enrichment in both WT and p53^−/− cells in all these assays (Fig. S1G). Our DRIP-qPCR assays did not note any difference in the levels of R-loops at ribosomal locus in WT and p53^−/− cells (Fig. S1, I and J). A comparative analysis of global run-on sequencing (GRO-seq) and DRIP-seq datasets indicated a high peak at full-length L1 indicating probably an involvement of retrotransposition process in R-loop formation. GRO-seq can detect nascent RNA that might form transient R-loops, which are quickly resolved and not stable enough to be captured by DRIP-seq (36). We also identified a partial overlap between the peaks from the GRO-seq and DRIP-seq data at the L1 5′UTR region. However, no overlapping peaks were detected at the L1-ORF1 and ORF2 regions in these two sequencing datasets (Figs. 1I and S1H), suggesting that the 5′UTR region demonstrates a higher tendency to form cis R-loops, likely due to nascent RNA annealing during the transcription. Our results indicate that the two ORF regions potentially contribute to the formation of trans R-loops through the L1 cDNA derived from distant L1 mRNAs.

Figure 1.

Elevation of L1 encoded intermediate components in p53^−/−cells.A, qPCR using the primers that detects consensus sequences of L1 5′UTR, L1ORF1, and L1ORF2 transcripts in WT and p53^−/− of A375 cell lines. The transcript levels were normalized using β-actin and relative fold change was calculated with respect to the parental WT. The individual data points indicate biological replicates. The data shown are the averages of three biological replicates. ∗p < 0.05 and ∗∗p < 0.01, (Mann–Whitney U test). B, immunofluorescence of RNA-DNA hybrids in A375 WT and p53^−/− cells stained with S9.6 antibody (Alexa Fluor 488, green) and 4′,6-diamidino-2-phenylindole (blue). The scale bar represents 5 μm. The confocal images shown are representative of three biological replicates. C, quantification of S9.6 whole cell signal intensities from A375 WT and p53^−/− were computed using Fiji. Minimum 150 cells were analyzed for RNA-DNA hybrid quantification from three biological replicates. ∗∗p < 0.01, ns = non-significant (Mann–Whitney U test). For the quantification of S9.6 signal, the total S9.6 fluorescence intensity was measured across entire cells. Background fluorescence intensity was established at seven arbitrary units (a.u.), and only S9.6 intensities surpassing this threshold were included in the analysis. D, schematic showing the methodology opted for RNA-DNA precipitation experiment where the RNA-DNA hybrids were immunoprecipitated. Half arrows at the schematic indicate primers amplifying different regions of L1 sequence at 5′UTR, ORF1, and ORF2. RPL13A was used as a positive control for quantification of the R-loop locus and ZNF544 as negative locus. E, immunoprecipitation of RNA-DNA hybrids by DRIP-qPCR was performed using S9.6 antibody to quantify accumulation of L1-sequence derived R-loops in A375 WT and p53^−/− cells. Enrichment of RNA-DNA hybrids in each region was normalized to input values. The individual data points indicate three biological replicates. Data represent mean ± SD from three biological experiments. ∗p < 0.05, ns = non-significant (Mann–Whitney U test). F and G, the assessment of RNase H1 D210N-GFP, a catalytic mutant of RNase H1 (RNH-GFP) in WT and p53^−/− A375 cells. F, the representative immunofluorescent images show A375 WT and p53^−/− cells stably expressing RNH-GFP construct. The nuclear localization of RNH-GFP (green), nucleus (4′,6-diamidino-2-phenylindole, blue) and (G) the immunoblot indicate the expression of RNH-GFP in WT and p53^−/− A375 cell lines. H, DRIP-qPCR analysis was performed on WT and p53^−/− cells using GFP antibodies on RNH-GFP stable cell lines of A375 WT and p53^−/− cells. Primers targeting the L1 5′UTR, ORF1, and ORF2 loci were used in both assays. RPL13A used as a positive control for the R-loop enrichment and ZNF544 as negative locus. Signal values of RNA-DNA hybrids immunoprecipitated in each region, normalized to input values. The individual data points indicate three biological replicates. Data represent mean ± SD from the three independent experiments. ∗p < 0.05, ns = non-significant (Mann–Whitney U test). I, this figure compares GRO-seq and DRIP-seq signals for a specific L1 genomic region in HEK293T cells. The dark blue peak represents GRO-seq data, indicating nascent RNA synthesis at the L1 region, while the black peak corresponds to the input signal for DRIP-seq, serving as a control for R-loop interactions. The brown peaks illustrate DRIP-seq data from S9.6 antibody immunoprecipitation, which enriches R-loops (RNA-DNA hybrids). The gene locations associated with this L1 region, including RPRD1A, FHOD3, KIAA1328, and CELF4 are shown for reference. The data also provides insights into the genomic locations of L1 in chromosome 18 of the hg38 assembly. DRIP, DNA-RNA immunoprecipitation; GRO-seq, global run-on sequencing; LINE1/L1, long interspersed nuclear element 1; qPCR, quantitative polymerase chain reaction.

L1cDNA contributes to R-loops formation in p53-deficient cells

We next investigated if L1cDNA formed during the retrotransposition life cycle participated in trans R-loop formation. The objective was to explore the role of trans R-loops formed by L1 elements—an aspect that has not been investigated previously—in genomic instability. The full-length retrotransposition competent L1Hs generates an identical DNA copy of itself, known as L1cDNA, through the reverse transcription of L1mRNA. Hence, within the retrotransposition life cycle, there is potential for the formation of two types of RNA-DNA hybrids that can serve as R-loop; cis R-loops (L1mRNA-gDNA) occurring at the L1 5′UTR promoter during transcription, and trans R-loops (L1mRNA-cDNA) formed during reverse transcription. Genome-wide DRIP-seq studies have established that cis R-loops are predominantly localized in GC-rich regions and promoter regions during transcription (37, 38). Recent evidence further suggests that p53 plays a crucial role in the global regulation of cis R-loops, and viral proteins have been shown to induce R-loops in p53-inhibited cells (39, 40). To determine whether L1-derived cDNA contributes to the formation of trans R-loops, we performed a DRIP-sequencing experiment on RNH-GFP–expressing A375. Fisher's exact test revealed significantly elevated R-loop peaks at L1 in p53^−/− compared to WT cells (Fig. 2A). The heat map analysis of R-loop read counts across various transposon families in A375 WT and p53^−/− cells (Fig. 2B), as well as in HEK293T and IMR90 cell lines (Fig. S2A); both harboring p53 mutations, suggests that younger and active L1 subfamilies are major contributors to R-loops. This observation is further supported by fold change calculations based on Z-scores (Fig. S2B). Violin plot analysis using Wilcoxon tests reveals total R-loops across all transposon families. LINEs and LTRs mostly contribute to R-loops, whereas short interspersed nuclear elements (SINEs) and endogenous retroviruses (ERVs) exhibit minimal or no significant R-loop enrichment (Fig. 2C). Similar to A375, heat map from K562 DRIP-seq datasets shows elevated R-loop levels specifically associated with LINEs (Fig. 2D). The snapshots of the genomic loci of L1 from the RNH-GFP–expressing WT and p53^−/− A375 cells (Fig. 2E), IMR90 and HEK293T (Fig. S2, C and D), and K562 (Fig. S2E) using genome browser tracks further validated our analyses that indeed the full-length L1 form R-loops. Next, to investigate whether L1 hyperactivation leads to increased accumulation of R-loop formation, we treated A375 WT and p53^−/− cells with an HDACi Sodium Butyrate (NaB) (Fig. S2F) that are known to induce repetitive elements, including L1 transposons, on WT and p53^−/− cells (29, 41, 42). The qPCR (Fig. S2G) on NaB-treated WT and p53^−/− A375 cells and RNA-seq analysis of NaB (GSM8006366) and Trichostatin A (TSA) (43) exposed cells (Fig. S2H) showed that L1 expression was elevated upon HDAC inhibition. These results indicate that in addition to p53, several other host defense mechanisms are active in the genome to suppress L1 transposons. Consequently, IF assays on WT and p53^−/− cells treated with NaB revealed an increased signal of S9.6 detecting the RNA-DNA hybrids. Rescue experiments employing 3TC, a well-known RT inhibitor that inhibits L1cDNA formation, led to decreased S9.6 signals in both NaB-treated WT and p53^−/− cells (Fig. 2, F and G). Thus, L1cDNA participates in generating L1 trans R-loops (L1mRNA-cDNA) derived RNA-DNA hybrids. Next, we performed DRIP-qPCR to assess RNA-DNA hybrid levels precipitated by S9.6 antibody in A375 WT and p53^−/− cells following 3TC treatment. The results showed a significant reduction of R-loops at the L1-5′UTR, L1-ORF1, and ORF2 regions (Fig. S3A). The observed reduction in R-loop levels at ORFs following 3TC treatment suggests that these sequences are particularly prone to forming trans R-loop structures. Consistent findings were obtained through immunoprecipitation using a GFP antibody in both A375 WT and p53^−/− cells–expressing RNH-GFP (Fig. 2H). These findings underscore that although majority of the L1 5′UTR region contributes to cis R-loop formation, L1-ORFs likely participates in trans R-loop formation involving L1cDNA. This data highlights the involvement of L1 in both types of R-loop signals, arising from L1mRNA-gDNA as well as L1mRNA-cDNA hybrids.

Figure 2.

Increased accumulation of L1 trans R-loops derived RNA-DNA hybrids in p53^−/−. A, the bar graph using Fisher’s exact test shows that LINE 1 mostly contributes to the R-loop. The x-axis shows the R-loop peak distribution across different genomic regions. The y-axis shows the total number of R-loop peaks that encapsulates LINEs, UTRs, exons, introns, and intergenic regions from DRIP-seq performed on WT and p53^−/− A375-expressing RNH-GFP. Both the LINE1 peaks (p = 1.574 × 10⁻²) and the overall peaks (p = 1.546 × 10⁻¹²) showed significant enrichment. Other genomic regions with p-values that indicate high relevance include UTRs (p = 1.282 × 10⁻³⁴), exons (p = 2.816 × 10⁻⁷²), introns (p = 1.430 × 10⁻¹⁷), and intergenic regions (p = 3.765 × 10⁻¹⁰⁹). B, the DRIP-sequencing on WT and p53^−/− A375 cell lines–expressing RNH-GFP. The violin plot represents the proportion of L1-derived R-loops relative to the total read counts per 100 loci. Heat map analysis from WT and p53^−/− A375-expressing RNH-GFP displaying the number of R-loop reads of transposable element (TE) families. Each row represents a TE family, and columns correspond to different sample types. The red represents a high number of R-loops, and blue represents a comparative lower number of R-loops, with a color gradient from red to blue. This gradient conveys the number of R-loop peaks across transposon families. L2C, L2b, L2a, L1PB1, L1PA9, L1PA8, L1P1, L1PA3, L1PA2, represent the L1 families classified based on their ages. SINE-VNTR-Alus (SVA) are nonautonomous, young, and active retrotransposons considered to be mobilized by the LINE1 reverse transcriptase in trans. C, using two-sided Wilcoxon tests, the distribution of various TE subfamilies associated with R-loop formation are shown. The Log2 fold changes and p values for TE families are SINE = 3.7 (p value = 0.09), LTR = 1.6 (∗p value = 0.018), Alu = 1.6 (p value = 0.67), DNA = 1.5 (p value = 0.88), ERV = 1.7 (p value = 0.863), and LINE = 4.08 (∗∗∗p value = 1.17 × 10⁻⁸). The findings reveal that LTR elements exhibit mild enrichment (∗p 0.05) in R-loop levels, whereas LINE elements are considerably enriched (∗∗∗p 0.001). R-loop levels are not significantly observed with SINE, Alu, DNA, or ERV sequences. D, DRIP sequencing analysis from K562 cell lines demonstrate that mostly the retrotransposition competent full-length L1 transposon families (L1Hs) contribute to R-loops. The violin plot represents the proportion of L1-derived R-loops relative to the total read counts per 100 loci. The heat map displays the number R-loop reads of TE families. Each row represents a TE family, and columns correspond to different sample types. E, UCSC Genome Browser custom track displays R-loop peaks identified from DRIP-seq data in WT and p53^−/− A375 cells–expressing RNH-GFP indicating the distribution of R-loops across these loci. This track also includes repetitive elements identified by RepeatMasker, allowing comparison of their genomic distribution relative to R-loops. Gene locations across the R-loop peaks are highlighted, with mapped genes including NAT10, ELF5, EHF, PDHX, LDLRAD3, and TRAF6 in the A375 datasets. F, representative confocal microscopy images of three biological replicates showing RNA-DNA hybrid accumulation in WT and p53^−/− A375 cells after treatment with either vehicle (NFW) or NaB for 24 h, followed by 3TC treatment for 72 h. For the illustration of strategies involving vehicle or NaB drug (25 μM) treatment on A375 WT and p53^−/− refer to the (Fig. S2F). The immunofluorescence images are representative of three biological replicates. The scale bar represents 5 μm. G, quantification of S9.6 intensities from the whole cells was done using Fiji software (49). For the quantification of S9.6 signal, the total S9.6 fluorescence intensity was measured across entire cells. Background fluorescence intensity was established at 7 a.u. and only S9.6 intensities surpassing this threshold were included in the analysis. A total of 150 cells, derived from 50 cells per replicate, were assessed for quantification. Violin plots exhibit combined data across three biological replicates. ∗p < 0.05, (Mann–Whitney U test). H, DRIP-quantitative polymerase chain reaction on RNH-GFP stably transfected cells of A375 p53 WT and p53^−/− cells before and after 3TC treatment. Immunoprecipitation was carried out using the GFP antibody. Enrichment of RNA-DNA hybrids in each region, normalized to input values.The individual data points indicate three biological replicates. Data represent mean ± SD from three independent experiments. ∗p < 0.05, ns = non-significant (Mann–Whitney U test). DRIP, DNA-RNA immunoprecipitation; LINE1/L1, long interspersed nuclear element 1; NaB, sodium butyrate; NFW, nuclease-free water.

L1cDNA-derived R-loops induce DNA damage and inflammatory responses in p53-deficient cells

To investigate the deleterious effects of L1-derived trans R-loops, we utilized an L1 ORF2p overexpression system, as ORF2p plays a critical role in the formation of L1 cDNA. To specifically exclude the contribution of the EN domain to DNA damage, we overexpressed the two domains of L1-ORF2p—EN and RT—individually. Here, we overexpressed EN, RT (C terminally fused with GFP) in WT, and p53^−/− A375 cells. An empty vector was used as a negative control. The RT domain maintains RT activity autonomously within the cytoplasm (17). Overexpression of the RT domain resulted in elevated levels of cytoplasmic RNA-DNA hybrids as evidenced by increased S9.6 signals, consistent with recent findings. Upon 3TC treatment, the augmented RNA-DNA hybrids in RT-GFP were completely mitigated but at a modest level in EN-GFP in p53^−/−–overexpressing cells (Fig. 3, A–D), confirming the involvement of L1cDNA-derived RNA-DNA hybrid formation. We propose that the increase in R-loops observed in EN-overexpressing cells is likely an indirect result of DNA damage caused by the EN domain. R-loops tend to form at sites of DNA double-strand breaks (44), indicating that DNA damage itself is a source of R-loop formation.

Figure 3.

Effect of reverse transcriptase inhibitor on RT domain overexpressing p53^−/−cells.A and B, examining the RNA-DNA hybrid levels before and after the treatment of RT inhibitor (3TC) caused due to ectopic overexpression of different ORF2p domains endonuclease (EN) and RT in A375 WT (A) and p53^−/−. B, the overexpressed proteins are C terminally GFP-tagged (green, Alexa Fluor 488) and S9.6 (magenta, Alexa Fluor 568) that visualize RNA-DNA hybrids are mostly localized at the cytoplasm, the nucleus is stained with 4′,6-diamidino-2-phenylindole (blue). The scale bar represents 5 μm. C and D, quantification of S9.6 intensity across whole cell was carried out by considering the number of GFP-positive cells across10 fields, from which 8 GFP-positive cells from each field were subjected to further calculations from the EV, EN, and RT domains of L1-ORF2p in WT (A) and p53^−/− (B) followed by 3TC treatment for 3 days post transfection of 24 h. The background intensity was again set at 7 a.u., and only S9.6 signals exceeding this threshold were considered. Statistical significance was evaluated using the Mann–Whitney U test, with significance levels represented as follows: p∗<0.05, p∗∗<0.01, and ns = non-significant. LINE1/L1, long interspersed nuclear element 1; EV, empty vector.

We also observed an elevated transcripts of DNA damage indicators (RAD50, BRCA1, and Rb) and inflammatory gene markers (IL6, STING, IRF-7, MCP-1, CXCL-10, IFN-β, and ISG-15) in EN and RT domain expressing cells using qPCR experiments. Upon the treatment with 3TC, decrease in accumulation of RAD50, BRCA1, RB, and inflammatory gene markers was observed (Fig. 4A), suggesting aberrant formation of L1mRNA-cDNA with deleterious effects on genome stability. Moreover, to confirm if the overexpression of RT-GFP leads to the generation of L1 trans R-loops, we conducted sequential DRIP-qPCR experiments (Fig. 4B). These experiments involved immunoprecipitation using GFP from in vivo RNH-GFP–expressing p53^−/−, followed by IP with S9.6 antibodies to capture specifically L1 derived R-loops. This sequential immunoprecipitation assay validated that RT domain alone can bind to the RNA-DNA hybrids in p53^−/−. Subsequent IF assays using the γH2AX antibody (Fig. 4, C–F) and Western blotting using 53BP-1 antibody (Fig. S3, B and C) revealed elevated DNA damage in cells-overexpressing EN domain, whereas a moderate level of DNA damage was also observed in cells expressing RT domain in A375 cells. Treatment with 3TC did not mitigate EN-induced DNA damage; however, a partial rescue was observed for the RT domain, suggesting that the RT domain generated L1 trans R-loop (L1 mRNA-cDNA) mediates genotoxic insults.

Figure 4.

Overexpression of EN and RT domains corelates with increased DNA damage and inflammatory markers.A, the heat map shows a consolidated qPCR of DNA repair genes (RAD50, BRCA1, and Rb) and inflammatory markers (STING, IRF7, MCP, CXCL10, IFN-β, and ISG15) from the exogenously expressed EV, EN, and RT domain in A375 p53^−/− cells followed by 3TC treatment for 3 days. The subsequent legend shows the level of expression following overexpression. The samples were harvested after 3TC treatment for 3 days on 24 h transfected cells. The data from three biological replicates were included for heat map generation. The normalized p value of the data is less than 0.05. B, sequential DRIP-qPCR was performed on RT-GFP exogenously overexpressing A375 p53^−/− followed by vehicle or 3TC treatment for 3 days. The immunoprecipitation was performed using GFP antibody followed by another round of immunoprecipitation by S9.6 on the GFP immunoprecipitated samples. Enrichment of RNA-DNA hybrids in each sample, normalized to the respective input values. The individual data points indicate two biological replicates. Data represent mean ± SD from the two independent experiments. C and D, representative fluorescence images depict the transient overexpression of EV, EN, and RT domains of L1-ORF2p in WT and p53^−/− A375 cells followed by 3TC treatment for 3 days post transfection of 24 h. The overexpressed proteins are C terminally GFP-tagged (green, Alexa Fluor 488), and cells are stained with anti γH2AX antibodies (magenta, Alexa Fluor 568), nucleus is stained with 4′,6-diamidino-2-phenylindole (blue). The scale bar represents 5 μm. E and F, box plot shows the quantitation of γH2AX foci in GFP-positive cells of WT (C) and p53^−/− (D) A375 cells transfected with EV, EN-GFP, and RT-GFP constructs. The microscopy images of GFP-positive cells were utilized to quantify the total number of γH2AX foci per nucleus. Total 10 fields were taken of which 8 GFP-positive nuclei per field were analyzed, and the average foci count was calculated across three biological replicates (in total 80 GFP-positive cells). Image analysis was performed using ImageJ (Fiji) software, where a filter was applied to select foci with a size range between 1.0 μm and 1.5 μm, excluding those outside this range from further analysis. A total of 80 nuclei, derived from three biological replicates, were used for the quantification of γH2AX foci. Statistical significance was evaluated using the Mann–Whitney U test, with significance levels represented as follows: p∗<0.05, p∗∗<0.01, and ns = non-significant; DRIP, DNA-RNA immunoprecipitation; LINE1/L1, long interspersed nuclear element 1; EV, empty vector; EN, endonuclease; qPCR, quantitative polymerase chain reaction.

p53 promotes resilencing of L1 transposons and their intermediates

Given our findings that delineate trans-repressive functions of p53 in maintaining repression of L1 and thus its intermediary components (R-loops), we sought to investigate the role of p53 in mediating the re-establishment of L1 silencing; a process that mirrors the epigenetic reprogramming observed in early embryos and primordial germ cells in vivo (45). To assess this, we employed HDACi drugs; TSA and NaB to induce L1 activation in a reporter system where L1 5′UTR-eGFP is inserted at the huAAVS1 site in both WT and p53^−/− cells of A375 (Fig. 5, A–D) and U2OS (Fig. S4, A–D). Following a 24-h treatment with HDACi drug (Figs. 5A and S4A), activation of L1 reporters were assessed by quantifying the GFP-positive cells by comparing the vehicle and 24 h NaB panels using the confocal microscopy. The efficacy of both WT and p53^−/− cells to suppress L1 5′UTR-eGFP upon drug withdrawal of 3 and 6 days elucidated almost a complete silencing of GFP in WT but not in p53^−/− cells, this can be observed by comparing the panels of cells treated with HDACi drugs and those observed at 4 and 7 days (Figs. 5B, and S4B). Similarly, the expression levels of eGFP transcripts correlated with increased GFP intensities in L1 5′UTR reporter cells in response to HDACi (Fig. 5C). This activation of GFP closely corresponded to activation of endogenous L1-ORF1, as demonstrated through quantification using qPCR (Fig. 5D). Four days after the withdrawal of TSA, the activity of the L1 5′UTR reporter returned to uninduced levels in p53 WT cells, but not in p53^−/−. In line with previous research on HDACi treated embryonic stem cells (29), our experimental model demonstrated that inhibiting HDAC with TSA led to hyperacetylation of histones H3 and H4. This was evident from the increased signal observed in western blotting when probed with antibodies targeting H3ac and H4ac in U2OS L1 5′UTR reporter cell lines (Fig. S4, C and D).

Figure 5.

The dynamics of the L1 5′UTR-eGFP reporter upon exposure to HDAC inhibitors and subsequent drug withdrawal.A, illustrates the schematic of NaB drug treatment and withdrawal in L1 5′UTR-eGFP WT and p53^−/− A375 cells. The samples were collected for qPCR, Western blotting, and microscopy at 24 h post drug (HDACi) treatment, as well as at fourth day and seventh day of drug withdrawal. NaB denotes drug treatment for 24 h, NaB-4D indicates drug withdrawal of 3 days, and NaB-7D signifies withdrawal after 6 days post 24 h treatment (GFP 488 and 4′,6-diamidino-2-phenylindole channels). B, confocal imaging of WT and p53^−/− A375 cells harboring L1 5′UTR-eGFP construct. The reporter cells were treated with either vehicle or NaB for 24 h, followed by withdrawal for 3 and 6 days of post drug treatment. Samples were collected on the first day, fourth day, and seventh day for confocal imaging post NaB treatment for GFP activation and silencing. Presented here are representative confocal images from three independent immunofluorescence experiments. The scale bar represents 5 μm. C and D, assessment of LINE1 activation and resilencing upon HDACi drug exposure. qPCR analysis was performed to quantify (C) eGFP and (D) L1 ORF1p transcripts in WT and p53^−/− A375 L1 5′UTR cells treated with vehicle or NaB at the first day, fourth day, and seventh day. The individual data points indicate three biological replicates. The data represent mean ± SD across three independent biological replicates. Statistical significance was evaluated using the Mann–Whitney U test, with significance levels represented as follows: p∗<0.05, p∗∗<0.01, and ns = non-significant. The primers used for qPCR analysis of ORF1 and eGFP transcripts are provided in Table S1. E, illustration indicates the nature of LRE3 plasmid which is a surrogate marker of retrotransposition. Cells that undergo retrotransposition are GFP-positive (green). F, the representative confocal images demonstrate the expression of GFP (green) in WT and p53^−/− A375 cells following exposure to NaB (D) drugs for 24 h, followed by drug withdrawal on third and sixth day. The scale bar represents 5 μm. The data represent mean ± SD across three independent biological replicates. Statistical significance was evaluated using the Mann–Whitney U test, with significance levels represented as follows: p∗<0.05, p∗∗<0.01, and ns = non-significant. G, percentage of GFP-positive cells in WT and p53^−/− A375 at 24 h of NaB treatment, followed by drug withdrawal at third and sixth day. GFP-positive cells were counted from a minimum of five fields from the three independent experiments. The individual data points indicate three biological replicates. The data represent mean ± SD across three independent biological replicates. Statistical significance was evaluated using the Mann–Whitney U test, with significance levels represented as follows: p∗<0.05, p∗∗<0.01, and ns = non-significant. H, the representative confocal images demonstrate S9.6 (green) and 4′,6-diamidino-2-phenylindole (blue) stained in WT and p53^−/− A375 cells following exposure to NaB (D) drugs for 24 h, followed by drug withdrawal of 3 or 6 days. The scale bar represents 5 μm. The data represent mean ± SD across three independent biological replicates. Statistical significance was evaluated using the Mann–Whitney U test, with significance levels represented as follows: p∗<0.05, p∗∗<0.01, and ns = nonsignificant. I, the S9.6 fluorescence intensity of RNA-DNA hybrids was quantified in whole cells of A375 WT and p53^−/− cell lines under vehicle or HDACi treatment conditions, as shown in panel (A). A total of 75 cells were analyzed for RNA-DNA hybrid quantification across three biological replicates. For the quantification of S9.6 signal, the total S9.6 fluorescence intensity was measured across entire cells. The background fluorescence intensity was considered at 7 a.u. and only S9.6 intensities surpassing this threshold were included in the analysis. The data represent mean ± SD across three independent biological replicates. Statistical significance was evaluated using the Mann–Whitney U test, with significance levels represented as follows: p∗<0.05, p∗∗<0.01, and ns = non-significant. LINE1/L1, long interspersed nuclear element 1; HDACi, histone deacetylase inhibitor; NaB, sodium butyrate; qPCR, quantitative polymerase chain reaction.

Next, to determine whether p53 also enables resilencing of retrotransposition activation, we used a synthetic L1 designated 99-gfp-LRE3 retrotransposition indicator (46, 47), using HDACi treatment, similar experiment was conducted on stably transfected parental WT and p53^−/− A375 and U2OS cells, schematic shown (Fig. 5E). This indicator mimics the retrotransposition cycle generating all L1 intermediate components allowing direct measurements of de novo retrotransposition rates by visualizing GFP-positive cells (green). Consistent with L1 5′UTR reporter, we noted re-silencing of L1-retrotransposition in WT cells upon withdrawal experiments after TSA and NaB treatments (Figs. 5F, S4, E and G). Whereas no re-silencing was observed in p53^−/− cells as quantified in (Figs. 5G, S4, F and H).

This observation prompted us to investigate whether the R-loops (RNA-DNA hybrids) induced by HDACi treatment undergo re-silencing in WT and p53^−/− cells. To assess this, we used S9.6 antibody to detect RNA-DNA hybrids in the HDACi (NaB) drugs and withdrawal experiments. We observed that after 4 and 6 days of drug withdrawal, there was almost complete suppression of R-loops in WT cells, but not in p53^−/− cells (Fig. 5, H and I). These results suggest p53 plays a pivotal role in resilencing of L1 and their derived R-loops.

p53 promotes resilencing of L1 transposons and their intermediates by restoring histone repressive marks at L1 5′UTR

To mechanistically delineate p53-mediated resilencing of activated L1, we chemically induced L1 transposons in both A375 WT and p53^−/− L1 5′UTR-eGFP reporters, as demonstrated in (Fig. 5) and quantified histone repressive marks H3K9me3 and H3K27me3 by performing chromatin immunoprecipitation (ChIP)-qPCR studies using primer sets targeting the L1 5′UTR (Fig. 6, A and B). When cells were treated with NaB, L1 activation in both WT and p53^−/− cells was accompanied by a loss of repressive marks (H3K9me3 and H3K27me3) and an increase in activating marks (H3K9ac and H3K4me3) at the L1 5′UTR within 24 h (Fig. 6, C and D). However, in WT cells, these repressive marks were restored to levels similar to untreated controls after a 4-day drug withdrawal, while activating marks (H3K9ac and H3K4me3) decreased. Notably, this restoration of chromatin marks was absent in p53^−/− cells.

Figure 6.

Histone modification dynamics in L1 5′UTR-eGFP reporter cells upon HDACi exposure and withdrawal.A–D, chromatin immunoprecipitation followed by qPCR analyses was conducted on the 5′UTR region of A375 WT and p53^−/− L1 5′UTR-eGFP stable cells. Primers targeting the p53-binding region were utilized, with immunoprecipitation carried out using antibodies against histone modifications H3K9me3, H3K27me3, H3K9ac, and H3K4me3. The results illustrate levels of H3K9me3 (A) and H3K27me3 (B), representative of repressive histone marks, normalized to input values, as well as levels of activation marks H3K9ac (C) and H3K4me3 (D). Reporter cells were treated with HDACi (NaB drug) for 24 h, followed by withdrawal for 3 and 6 days. Samples for analyses were collected at first day, fourth day, and seventh day (for drug withdrawal), as depicted (refer to Fig. 5). It is noteworthy that alterations in H3K9me3, H3K27me3, H3K9ac, and H3K4me3 enrichment are specific to the p53 status (refer to Fig. S5, A and B). The individual data points indicate three biological replicates. The data represent mean ± SD across three independent biological replicates. Statistical significance was evaluated using the Mann–Whitney U test, with significance levels represented as follows: p∗<0.05, p∗∗<0.01, and ns = non-significant. E, qPCR was performed to assess the knockdown of SETDB1 and G9A from the siRNA transfected WT and p53^−/− A375 L1 5′UTR-eGFP cell lines. The dark gray bar plot refers to the untransfected cells and the light gray being nontargeting control (NTC). The individual data points indicate three biological replicates. The data represent mean ± SD across three independent biological replicates. Statistical significance was evaluated using the Mann–Whitney U test, with significance levels represented as follows: p∗<0.05, and ns = non-significant. F and G, Western blotting was performed to assess the reduction in the expression levels of SETDB1 and G9A in the siRNA transfected samples of WT and p53^−/− A375 L1 5′UTR-eGFP cells. Please refer (Fig. S5, C and D) for their quantification using Fiji software. H, qPCR to measure the transcript levels of L1 ORF1 in SETDB1 and G9A knocked down A375 WT and p53^−/− cell lines. The individual data points indicate three biological replicates. The data represent mean ± SD across three independent biological replicates. Statistical significance was evaluated using the Mann–Whitney U test, with significance levels represented as follows: p∗<0.05 and ns = nonsignificant. I, ChIP-qPCR to quantify the deposition of H3K9me3 marks at L1 5′UTR in SETDB1 knocked down cells of A375 WT and p53^−/− L1 5′UTR-eGFP cells. The individual data points indicate three biological replicates. The data represent mean ± SD across three independent biological replicates. Statistical significance was evaluated using the Mann–Whitney U test, with significance levels represented as follows: p∗<0.05 and ns = nonsignificant. J, ChIP-qPCR to quantify the deposition of H3K9me3 marks at L1 5′UTR in G9A knocked down cells of A375 WT and p53^−/− L1 5′UTR-eGFP cells.The individual data points indicate three biological replicates. The data represent mean ± SD across three independent biological replicates. Statistical significance was evaluated using the Mann–Whitney U test, with significance levels represented as follows: p∗<0.05 and ns = nonsignificant; LINE1/L1, long interspersed nuclear element 1; HDACi, histone deacetylase inhibitor; NaB, sodium butyrate; ChIP, chromatin immunoprecipitation; qPCR, quantitative polymerase chain reaction.

The reversal of L1 activation in WT cells during drug withdrawal closely aligned with the re-establishment of repressive chromatin marks, suggesting that p53 plays a key role in resilencing activated L1 transposons by promoting repressive marks and removing activating marks (Figs. 5, S4, and, S6, A–D). These epigenetic changes were specific to the L1 5′UTR and not observed at p21, a canonical p53-binding site, as confirmed by ChIP-qPCR (Fig. S5, A and B).

Next, we aimed to investigate whether p53 facilitates the deposition of epigenetic repressive marks, specifically H3K9me3, at the L1 5′UTR in collaboration with histone methyltransferases, exploring underlying mechanisms of retroelement repression. SETDB1 is required for ERV retrotransposon silencing in differentiated somatic cells (31), the G9a loss leads to retrotransposon activation (30). SUV39H1 enzyme catalyses H3K9me3 at constitutive heterochromatin (48). We employed specific siRNAs (see Table S1 for sequences) to knock down SETDB1, G9a, and SUV39H1 in A375 L1 5′UTR-eGFP reporter cells. The decreased expression levels of SETDB1 and G9A was quantified using qPCR in the WT and p53^−/− A375 L1 5′UTR-eGFP cells (Fig. 6E) and further validated using the Western blotting (Fig. 6, F and G) and its quantification using Fiji (50) (Fig. S5, C and D). The knockdown of SETDB1 and G9A led to elevation in the levels of L1-ORF1 in both WT and p53^−/− A375 L1 5′UTR-eGFP reporter cell lines (Fig. 6H). We observed a decrease in H3K9me3 deposition in L1 5′UTR-eGFP WT and p53^−/− A375 cells following the knockdown of SETDB1 and G9A (Fig. 6, I and J). However, we were unable to detect SUV39H1 at either the transcript or protein levels in A375 L1 5′UTR-eGFP cells. Taken together, these results show that p53 may recruit SETDB1 and G9A dimeric complexes to play a significant role in depositing H3K9me3 marks to mediate L1 transposon silencing. The findings depicted in (Figs. 5 and 6), and (Figs. S4 and S5) demonstrate that p53 establishes resilencing of L1 transposons upon HDACi withdrawal by mediating recruitment of epigenetic repressive and inhibiting activating marks at L1 5′UTR.

Discussion

Our comprehensive analyses, utilizing both experimental and bioinformatic approaches, elucidate the critical role of p53 in repressing RNA-DNA hybrids that arise from L1 retrotransposition-derived R-loops (Fig. 7). Previous studies have predominantly focused on the mutagenic potential of L1 de novo insertions, yet the implications of nucleic acid intermediates, such as L1 RNA-DNA hybrids, in driving genomic instability and inflammation have remained largely unexplored. While new L1 insertions are relatively rare in cancer genomes, L1 activation is a common occurrence in various sporadic cancers, as demonstrated by recent pan-cancer analyses (8, 50). Our finding regarding the prominent involvement of L1 promoter in cis R-loop formation aligns with previous genome-wide studies, highlighting the enrichment of R-loops in promoter regions globally (14, 15). This is corroborated by the observation of overlapping peaks of DRIP-seq at the L1 promoter with those identified from the GRO-seq, suggesting the occurrence of transcriptional pausing at L1 5′UTR that leads to the formation of cis RNA-DNA hybrids. Conversely, the participation of L1-ORF1 and ORF2 genic regions in trans R-loop formation is validated by our observation of diminished R-loop signals upon 3TC treatment in RNH-GFP–expressing cells (Fig. 2). The dual nature of R-loops in cellular physiology—where they can be both deleterious and advantageous—led us to hypothesize that retrotransposon-derived R-loops could act as a barrier to the completion of retrotransposition, thereby preventing excessive amplification of L1 copies in the genome. However, our data reveal that the reduction in DNA damage markers and inflammation following RT overexpression coupled with decreased cytosolic RNA-DNA hybrids upon 3TC treatment (Figs 3 and 4), points to the role of L1 RNA-DNA hybrids in promoting genomic instability and inflammation (51)—hallmarks of cancer (5, 52, 53). These findings suggest that, beyond the canonical retrotransposition insertion events, intermediate components like RNA-DNA hybrids may possess oncogenic potential, positioning L1 R-loops as double-edged swords in the context of genomic integrity.

Figure 7.

The proposed model depicts roles of the two plausible R-loops derived from LINE1 retrotransposons in p53-deficient cells, that is, cis R-loop which includes the L1 mRNA-genomic DNA and trans R-loop that includes the L1 mRNA-complementary DNA hybrids. We also show that p53 collaborate with Histone methyltransferases SETDB1 and G9A for resilencing of active LINE1 transposons. LINE1/L1, long interspersed nuclear element 1.

Furthermore, our study demonstrates that inhibition of HDACs lead to the loss of methylation and the gain of acetylation marks at the L1 5′UTR, resulting in the induction of L1 and its derived R-loops. In p53-deficient cells, the withdrawal of HDACi fails to restore L1 silencing, leading to sustained L1 derepression, DNA damage, and inflammation, all of which are associated with elevated levels of L1 R-loop–derived RNA-DNA hybrids. In contrast, WT cells rapidly re-establish L1 silencing by promoting H3K9me3 and H3K27me3 deposition; reorganizing both repressive and activating chromatin modifications, illustrating the robust epigenetic response that is compromised in p53-deficient contexts (refer to Figs. 5 and 6).

The inability to resilence L1 sequences in p53-deficient cells may underlie resistance to cancer therapies involving HDACi drugs in p53-driven cancers. Notably, treatment with 3TC, a potent inhibitor of L1 RT, effectively prevented the further activation of L1 R-loops (RNA-DNA hybrids), providing the first evidence of trans L1 R-loops formation in unstressed p53-deficient cells as well as HDACi drug-treated WT and p53^−/− cells. Building on our earlier work, which established the role of p53 in repressing L1 activity in human somatic cell lines like A375, U2OS, and HBEC3kt—but not in HCT116—this study suggests that the absence of p53 may lead to L1 behavior that is modulated by other factors or specific cellular contexts (54, 55), in particular, the absence or inactivation of p53 in commonly used cell lines like HeLa (56) and K562 (57).

Additionally, our study revealed that the L1-ORF2p RT domain plays a critical role in regulating RNA-DNA hybrids triggering acute inflammatory transcriptional programs (58, 59, 60), particularly in the absence of p53. This suggests an unexplored oncogenic risk associated with L1 activity. The cytosolic accumulation of RNA-DNA hybrids and trans L1 R-loops, as revealed by sequential DRIP-qPCR assays, further supports this model. The use of 3TC not only mitigated nearly all inflammatory gene expression but also prevented DNA damage in both WT and p53-deficient cells, highlighting the potential therapeutic value of targeting L1 retrotransposition intermediates. These insights point to the potential of incorporating RT inhibitors into personalized cancer therapies to enhance the efficacy of existing treatment protocols by targeting retroelement activity and reducing genomic instability (61).

Experimental procedures

Cell culturing

A375 (melanoma), U2OS (osteosarcoma) were a kind gift from Jerry Shay (UTSW, originally obtained from American Type Culture Collection) to Abrams lab (UTSW). CRISPR-mediated p53 WT and p53 knockout (p53^−/−) cells of A375 and U2OS were generated in Abram’s lab, UT Southwestern Medical Center and were obtained as a kind gift. A375 and U2OS cells were cultured in Dulbecco’s Modified Eagle’s Medium (Gibco, #12100046). All cell lines were grown at 37 °C, 5% CO₂ and passaged using Trypsin (Gibco, #15400054) in the presence of an anti-antimycotic antibiotic (Gibco, #15240062).

Generation and screening of p53 KO cells

CRISPR cas9 editing (pX458) was used to generate p53^−/− cell lines in the laboratory of Prof. John Abrams at UTSW. p53^−/− clones were identified by Western blot using p53 monoclonal DO-1 (Santa Cruz, #sc-126) and L1-ORF1p expression levels were examined using anti L1-ORF1p antibody employing IF and Western Blotting. Individual clones were grown until 100% confluent in 6-well plates, and whole-cell lysates were prepared in radio-immunoprecipitation assay (RIPA) lysis buffer (NaCl 140 mM, EDTA 1uM, Tris–HCl 10 mM, 1% Triton X-100, 0.1% SDS, and 0.1% sodium deoxycholate) containing protease inhibitors (Roche). Protein was quantified using Bradford assay (Protein Assay Kit, Bio-Rad) and an equal amount of total protein was loaded. Antibodies used were as follows: p53 monoclonal DO-1 (Santa Cruz, #sc-126), anti-L1-ORF1p, clone4H1 (Millipore, #MABC1152), and β-actin (Sigma, #A-2228). Fiji software was used to measure the band intensities of L1ORF1p and β-actin from Western blot experiments.

Generation of L1-5′UTR-eGFP reporter cells

The 5′UTR of L1 was cloned at the FseI and AfeI sites of the huAAVS1-eGFP-Hyg donor plasmid and was transfected into A375 WT and p53^−/− cells to generate stable cell lines in the laboratory of Prof. John Abrams (UT Southwestern Medical Center). Post transfection, cells were subjected to selection with 100 μg/ml hygromycin for 2 weeks, where following trypsinization, hygromycin was added every 2 days. Stable clones were subsequently validated via qPCR.

3TC experiments

2.5 × 10⁵ WT and p53^−/− A375 and U2OS cells were plated into 6-well plates. After overnight seeding media was changed with fresh media-containing vehicle, HDACi. After 24 h of HDACi treatment media was replaced with fresh containing 3TC (10 μM) (Sigma, #L1295). After 4 days and 7 days of 3TC treatment, cells were harvested for RNA isolation or fixed for IF. qPCR was conducted using QuantStudio 5 (Applied Biosciences). Gene expression was normalized to β-actin, and fold induction was calculated relative to the WT. Primer sequences utilized in this study are available in Table S1.

HDACi treatment

TSA (Sigma Aldrich, #T8552) and NaB (Sigma-Aldrich, #567430) were used as HDACis. All drug treatments were performed in Dulbecco's Modified Eagle Medium (Gibco, #12100046) supplemented with 10% (v/v) fetal bovine serum (Gibco, #A5256701) and 1X antibiotic-antimycotic solution (Gibco, #15240062). Cells were treated with TSA at a concentration of 100 nM for 24 h. NaB was administered at a concentration of 25 μM, as described. For TSA treatment, DMSO was used as vehicle control, whereas nuclease-free water (NFW) was used for NaB treatment.

Effect of HDACi in WT and p53^−/− cells expressing LRE3 retrotransposition

A total of 1 × 10⁶ A375 WT and p53^–/– cells were transfected with equal amounts of the 99-GFP-LRE3 (LRE3) plasmid by using Effectene transfection reagent (QIAGEN, #301425) using the manufacturer's protocol. After 72 h, cells were selected for puromycin resistance. Following 3 days of selection, cells were treated with NaB and fixed at intervals of 24 h, 4 days, and 7 days. GFP-positive cells were counted from five images per sample. Three biological replicates were performed.

Cloning of NLS-tagged RNase H1 with D210N

To generate the NLS-tagged RNase H1 D210N construct, cDNA fragments encoding the human RNase H1 catalytic dead mutant (D210N) were synthesized (procured from IDT) and subsequently amplified via PCR for cloning into the pEGFP-N3 vector using EcoRI (NEB, #R3101S) and KpnI (NEB, #R3142S) restriction sites. A single SV40 NLS (sequence: CCCAAAAAGAAACGCAAAGTG) was incorporated at the 5′ end via the forward primer during PCR amplification. The primer sequences used for PCR amplification are provided in Table S1.

Generation of stable cell lines expressing RNase H1 D210N (RNH-GFP)

To study R-loop in WT and p53^−/− cells, nucleus localization sequences containing GFP-tagged, catalytically inactive RNase H1 D210N was cloned in pEGFP-N3. The cloned sequence was verified by Sanger sequencing and A375 WT and p53^−/− cells were transfected for the stable cell line generation using G418 (Gibco, #11811064).

Real-time PCR experiments (qPCR)

Total RNA was isolated using TRIzol (Invitrogen, #15596026), then treated with Turbo DNase (Thermo Fisher Scientific, #AM1907) to remove DNA. qPCR was conducted utilizing cDNA synthesis kit Verso (Thermo Fisher Scientific, #AB1453A). Applied biosciences QuantStudio 5 was used for real-time PCR and normalization of gene expression was done against β-actin (for standard gene expression analysis fold change = 2-ΔΔCT, ΔΔCT = Δ target sample – Δ reference sample), RPL134 for DRIP-qPCR experiments, and fold induction was determined relative to the WT. Refer to Table S1 for the primer sequences employed in qPCR assays.

Chromatin immunoprecipitation-qPCR

1 × 10⁷ A375 WT and p53^−/− cells were grown in a 150 cm² dish and were fixed with 0.75% formaldehyde in flasks at room temperature for 10 min and subsequently quenched by 125 mM glycine solution was added to stop the fixation. The cells were scraped from the flask and were resuspended in ice cold ChIP lysis buffer (50 mM Hepes-KOH pH 7.5, 140 mM NaCl, 1 mM EDTA pH8, 1% Triton X-100, and 0.1% sodium deoxycholate, 0.1% SDS, Protease inhibitor, Sigma #539131). The cell lysate (i.e., chromatin) was sonicated using an M220 Focused-ultrasonicator (Covaris) to release chromatin lengths to 400 bp, peak power 75.0, duty factor = 10, cycles/burst = 200, for 240 s. For each immunoprecipitation, 25 μg chromatin was diluted ten times with RIPA buffer and incubated with the following antibodies: H3K9me3 (Cell Signaling Technologies #13969), H3K27me3 (Cell Signaling Technologies, #9733), H3K9ac (Cell Signaling Technologies, #9649), and H3K4Me3 (Cell Signaling Technologies, #9751). Agarose A/G beads (EMD Millipore, #IP0515MLl) were charged and incubated with the antibodies overnight, following high salt, low salt, and LiCl washes the next day and eluted. Eluted chromatin was treated with RNAse A (Thermo Fisher Scientific, #12091021) and proteinase K (NEB, #P8107S), crosslinks were reversed, and samples were purified using a PCR purification kit (QIAGEN). ChIP enrichment was quantified by qPCR with QuantStudio 5 (Applied Biosciences).

DNA-RNA immunoprecipitation-qPCR

1 × 10⁷ A375 p53 WT and p53^−/− cells or cells expressing RNH D210N were plated into a 150 cm² dish for overnight (Tarsons) and were fixed by 1% formaldehyde, followed by quenching via glycine to a final concentration of 125 mM. Cells were rinsed twice with 10 ml cold PBS. Five milliters of cold PBS was added and dishes were scraped thoroughly with a cell scraper and transferred. Cells were lysed in 750 ul lysis buffer composed of 140 mM NaCl, 1 mM EDTA (pH8),1% Triton X-100, 0.1% sodium deoxycholate, 0.1% SDS, and protease inhibitors (added fresh each time, Sigma #539131) per 10 million cells. Cell lysis was performed at room temperature. The chromatin was sonicated to shear DNA to an average fragment size of 400 bp by the M220 Ultrasonicator (Covaris) as previously mentioned for ChIP. To check for abolition for DNA-RNA hybrid, RNase H (5 U/μl) was added prior to immunoprecipitation to the chromatin and was incubated overnight at 37 °C (this step is only applicable to check whether or not RNase H abolishes R-loops). Approximately 25 μg of RNA-DNA hybrid per immunoprecipitation (IP) was used. Each sample was diluted 1:10 with RIPA buffer (50 mM Tris–HCl, pH 8150 mM NaCl, 2 mM EDTA pH 8, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS) and protease inhibitors (added fresh each time, Sigma #539131). Primary antibody RNA-DNA hybrid (S9.6) antibody (Merck Millipore, #MABE1095) or GFP (Thermo Fisher Scientific, A-11122) was added to all samples and incubated at 4 °C for 2 h. 60 μl of Pierce A/G Agarose beads were taken for 25 ug of RNA-DNA hybrid. Beads were charged and incubated with the antibody incubated sample at 4° overnight in shaking. The immunoprecipitated samples were washed by low salt, high salt, and LiCl wash buffers. RNA-DNA hybrids were eluted by slowly adding 120 μl of elution buffer (1% SDS and 100 mM NaHCO3) to the protein A/G beads and vortex for 15 min at 30 °C. For reverse crosslinking reaction, the samples were treated with 4.8 μl of 5 M NaCl and 2 μl RNase A (10 mg/ml) at 65 °C overnight. 2 μl proteinase K (20 mg/ml) was added and incubated while shaking at 60 °C for 1 hour. The RNA-DNA hybrids were purified using a PCR purification kit (QIAGEN). Analysis of RNA-DNA hybrid level was performed by qPCR.

Sequential DRIP-qPCR

A375 p53^−/− cells were stably transfected with LINE 1 RT domain and was harvested. Cross-linked chromatin from was immunoprecipitated with antibody against RT-GFP (anti-GFP, Thermo Fischer, #A-11122) following all steps similar to (DRIP methodology in the above section), chromatin was eluted in nuclease-free water. Eluted chromatin was diluted 10-fold, subjected to a second immunoprecipitation with S9.6 (Merck Millipore, #MABE1095) antibody, and then eluted with nuclease-free water. The chromatin was purified and then analyzed via qPCR.

Quantification of RNA-DNA hybrids using qPCR

RNA-DNA Hybrid levels were measured by qPCR using 2X POWER UP SYBR Green (Invitrogen, #A25741) and analyzed via the QuantStudio 5 Real-Time PCR System (Applied Biosciences). Primer sequences are listed in Table S1. The qPCR results were analyzed using the percentage input method. The RNA-DNA hybrid enrichment was calculated based on the IP/input ratio. Positive locus was taken for RPL13A gene and negative locus for ZNF544. Three biological replicates were performed.

DRIP sequencing

1 × 10⁷ A375 p53 WT and p53^−/− cells or cells expressing RNH mutant (D210N) were plated into a 150 cm² dish and the standard protocol for DNA preparation was followed like DRIP-qPCR methodology. The immunoprecipitated samples along with their inputs were used for library preparation; A375 WT input and RNH-GFP were immunoprecipitated, A375 p53 ko and RNH-GFP were immunoprecipitated. DNA was sheared into 400 bp fragments using M220 Focused-ultrasonicator (Covaris) to release chromatin lengths to 400 bp and indexed NGS libraries were prepared by end repairing, A-tailing, adaptor ligation, and amplification procedures using NEBNext Ultra II DNA Prep Kit (Cat. #E7645, NEB). Upon library preparation, libraries were analyzed using Agilent Technologies 4200 Tapestation then sequenced on Illumina NovaSeq 6000 with 2 × 150 bp pair-end reads. Initial quality assessment of the sequencing reads was conducted via FastQC (V0.12.1; http://www.bioinformatics.babraham.ac.uk/projects/fastqc/) (62) (fastqc < reads.fastq > -o < output_directory>), followed by adapter trimming and removal of low-quality bases using Cutadapt (3.5) (10.14806/ej.17.1.200) (cutadapt -a <adapter_sequence > -o < trimmed_reads.fastq> <reads.fastq>) with specific parameters optimized for stringent cleaning. Cleaned reads were aligned to the human reference genome (hg38) using Bowtie2 (v2.5.4; http://bowtie-bio.sourceforge.net/bowtie2/index.shtml) (bowtie2 --local --sensitive-local --no-unal --no-mixed --no-discordant --no-overlap --no-dovetail -x <index> -q < reads.fastq > -S <output.sam>) with parameters tailored for efficient short-read mapping. Subsequently, MACS3 (3.0.0) (10.1186/gb-2008-9-9-r137) (macs3 callpeak -t <reads.bam> -c < control_reads.bam > -f BAM -g hs -n <output> --outdir <directory> --qvalue 0.05 --broad --broad-cutoff 0.1) was employed for peak calling to identify significant enrichment regions indicative of R-loops, thereby generating genomic coordinates of potential peaks. Visualization of identified peaks was performed using the UCSC Genome Browser, where BAM files containing aligned reads were uploaded alongside the hg38 reference genome for comprehensive genomic distribution analysis.

Immunofluorescence

A375 WT and p53^−/− cells were cultured on poly-L-lysine (Sigma, #P4707)-coated coverslips, washed with PBS, and fixed with 4% paraformaldehyde. The cells were treated with 0.1% Triton X-100 in PBS and 10% fetal bovine serum and were subsequently incubated with primary antibodies at room temperature for 1 h or 4° overnight. After washing, secondary antibodies were added and incubated in the dark at room temperature for 1 h. Cells were washed with PBS and counterstained with 4′,6-diamidino-2-phenylindole. All images were taken using the same microscope parameters, and fluorescence quantification was performed using Fiji software. Primary antibodies used were LINE1 ORF1p (Merck Milipore, #MABC1152), Anti RNA-DNA Hybrid antibody (Merck Milipore, #MABE1095), γH2AX (Invitrogen, #MA1-2022), at a dilution of 1:50. Secondary antibodies used were: anti-mouse 568 (Thermo Fisher Scientific, #A-11031), anti-mouse 488 (Thermo Fisher Scientific, #A-10680), anti-rabbit 488 (Thermo Fisher Scientific, #A-11008), and anti-rabbit 568 (Thermo Fisher Scientific, #A-11011), at a dilution of 1:1000. For all IF experiments secondary only control antibodies for mouse and rabbit were used in each cell line.

Microscopy

Images were captured using 60x oil objective Leica Sellaris 8 Falcon (IISER Berhampur), Leica sp8 (ILS Bhubaneswar BT/INF/22/SP28293/2018). All images were processed via Fiji software package.

Microscopy quantification

The digital image data of GFP-positive cells were utilized to quantify the total number of γH2AX foci per nucleus. Total 10 fields were taken of which 8 GFP-positive nuclei per field were analyzed, and the average foci count was calculated across three biological replicates (In total 80 GFP-positive cells). Image analysis was performed using ImageJ (Fiji) (49) software, where a filter was applied to select foci with a size range between 1 μm and 1.5 μm, excluding those outside this range from further analysis. A total of 80 nuclei, derived from three biological replicates, were used for the quantification of γH2AX foci.

For the quantification of S9.6 signal (limited to Figs. 1, 2, and 5), the total S9.6 fluorescence intensity was measured across entire cells. Background fluorescence intensity was established at seven arbitrary units, and only S9.6 intensities surpassing this threshold were included in the analysis. A total of 150 cells, derived from 50 cells per replicate, were assessed for quantification. In the case of (Figs. 3), 10 fields were selected, from which eight GFP-positive cells were subjected to further calculations. The background intensity was again set at seven arbitrary units, and only S9.6 signals exceeding this threshold were considered. Statistical significance was evaluated using the Mann–Whitney U test, with significance levels represented as follows: p∗ < 0.05, p∗∗ < 0.01, and p∗∗∗ < 0.001.

Gene knockdown by siRNA

A375 WT and p53^−/− cells were seeded at a density of 0.1 × 10⁶ cells per well in 12-well plates. siRNA transfections targeting SETDB1, G9A, and a nontargeting control were performed in both A375 WT and p53^−/− cells using TransMessenger reagent (QIAGEN), following the manufacturer’s instructions. We used 0.8 μg of siRNA in each well. Cells were incubated for 72 h posttransfection, after which they were harvested for Western blot analysis. Gene knockdown efficiency was confirmed by Western blot.

The siRNA sequences utilized for SETDB1 (63) and G9A (64) as described by along with the nontargeting siRNA control sequence (GenScript) are mentioned in Table S1.

Western blot

Protein extraction from A375 cells as using ice-cold RIPA buffer and a protease inhibitor cocktail (Sigma #539131). Bicinchoninic acid assay (Thermo Fisher Scientific, #23227) was used to assess protein concentration in the extracts. Proteins were resolved by SDS–PAGE and transferred to a nitrocellulose membrane (Bio-Rad, #1620112). Membranes were blocked for 1 h at room temperature in PBS Tween-20 with 5% nonfat milk. Subsequently, membranes were incubated overnight at 4 °C with the primary antibodies: anti-SETDB1 (Abclonal, #A6145), anti-G9A (Abclonal, #A19288), anti-β-actin (Cell Signaling Technologies, #A2228), H3 total (Abcam,#ab1791), H4 total (Abcam,#ab10158), H4ac (Abcam,#ab45166), H3ac (Abcam,#ab4441), and 53BP1 (Cell Signaling Technologies, #4937). Antibodies against β-actin, SETDB1,G9A, 53BP1, H3, H4, H3ac, and H4ac were diluted 1:1000 PBS containing 3% bovine serum albumin. The nitrocellulose membranes were then incubated with the horseradish peroxidase-conjugated secondary antibodies for 1 h at room temperature. Goat anti-rabbit IgG-horseradish peroxidase (Cell Signaling Technologies, #7074) was used against SETDB1, G9A, 53BP1, H3, H4, H3ac, and H4ac antibodies diluted 1:3000 in PBS Tween-20 containing 1% nonfat milk, whereas anti-mouse IgG (H  + L) (DyLight 800 4X PEG Conjugate) was used against actin antibody at a dilution of 1:15,000 in the same buffer. The detection of the immunoreactive bands was performed with the Clarity Western ECL Substrate (Bio-Rad, #170-5061) at 10 min exposure. All experiments have been performed 3 times and representative results of one experiment are shown. We used Fiji software to quantify the Western blot signals. First, we selected the lanes corresponding to each sample using the rectangle tool and then used the "Plot Lanes" function to generate intensity profiles of the bands. The area under each peak, representing the band intensity, was measured. To ensure accurate comparisons, we calculated the percentage contribution of each band to the total signal and normalized the target protein bands to the total protein loaded. Finally, the values were expressed relative to the control sample, which was set to 1. For PTMs, that is, H3ac and H4ac, total H3 and H4 were used as loading control.

DRIP-seq methodology

DRIP-seq data were downloaded in sequence read archive format from the NCBI sequence read archive [IMR-90 cell line (SRR2028291, SRR2028292), HEK293T cell line (SRR2028293, SRR2028294, SRR5379771, SRR5379772, SRR5379773, SRR5379774, SRR5379775, and SRR5379776), K562 cell line (SRR5379780, SRR5379781, and SRR5379782)]. The analysis aimed to quantify R-loops and identify retrotransposable elements in a sample. Raw sequencing reads underwent quality control using FastQC (V0.12.1) (fastqc < reads.fastq > -o < output_directory>), processed reads were aligned to the human reference genome (hg38) using Bowtie2 (v2.5.4) (10.1038/nmeth.1923) (bowtie2 --local --sensitive-local --no-unal --no-mixed --no-discordant --phred33 -I 10 -X 800 --no-overlap --no-dovetail -x <index_base> -q < reads.fastq > -S <output.sam>), annotated with a RepeatMasker-generated a retroelement annotation file in gene transfer file format, containing genomic coordinates, strand, conservation scores relative to consensus sequence, and relational information for repetitive elements by element, family, and class based on the December 2013 RepeatMasker open-4.0.5 Dfam 2.0 repeat masking of the hg38 human genome assembly.

Postalignment, R-loop peaks were identified using peak-calling algorithms like MACS3 (3.0.0) (10.1186/gb-2008-9-9-r137) (macs3 callpeak -t <reads.bam> -c < control_reads.bam > -f BAM -g hs -n <output> --outdir <directory> --qvalue 0.05 --broad --broad-cutoff 0.1). Subsequently, 100 loci with significant R-loop enrichment were selected for further analysis. Statistical assessments were conducted to evaluate R-loop enrichment, correlation with retrotransposable elements, and differential expression across multiple cell lines.

Heat maps depicting the expression patterns of retrotransposable elements were generated using the R programming language and the ggplots (4.3.3) (10.1007/978-3-319-24277-4_9) package. These visualizations provided insights into the spatial distribution and differential expression of retrotransposable elements associated with R-loop formation across various genomic loci and cell lines. Additionally, violin plots were created to visualize the distribution of R-loop counts per 100 loci within the same samples.

GRO-seq methodology

GRO-seq data were obtained from the NCBI GEO dataset [HEK293T cell line (SRR5379790 and SRR5379791)]. The methodology for analysis is similar to that of DRIP-seq, which has been explained in detail in the DRIP-seq section.

Data availability

Datasets supporting the current study is deposited NCBI Geo SRR BioProject: PRJNA1161127.

Original code will be shared by the lead contact upon request.

Supporting information

This article contains supporting information.

Conflict of interests

The authors declare that they have no conflicts of interest with the contents of this article.

Reviewed by members of the JBC Editorial Board. Edited by Karin Musier-Forsyth