Persistence of RNA transcription during DNA replication delays duplication of transcription start sites until G2/M

Abstract

As transcription and replication use DNA as substrate, conflicts between transcription and replication can occur, leading to genome instability with direct consequences for human health. To determine how the two processes are coordinated throughout S phase, we characterize both processes together at high resolution. We find that transcription occurs during DNA replication, with transcription start sites (TSSs) not fully replicated along with surrounding regions and remaining under-replicated until late in the cell cycle. TSSs undergo completion of DNA replication specifically when cells enter mitosis, when RNA polymerase II is removed. Intriguingly, G2/M DNA synthesis occurs at high frequency in unperturbed cell culture, but it is not associated with increased DNA damage and is fundamentally separated from mitotic DNA synthesis. TSSs duplicated in G2/M are characterized by a series of specific features, including high levels of antisense transcription, making them difficult to duplicate during S phase.

Keywords: DNA damage; DNA replication; G2/M DNA synthesis; RNA polymerase II transcription; Replication origins; transcription-associated genome instability.

Graphical abstract

Figure 1

A system for analyzing transcription-replication coordination (A) Experimental design schematic and FACS analysis of propidium iodide (PI) and BrdU to monitor S-phase progression, with quantification of cells number in each box (bottom left, G0/G1 phase; top, S phase; bottom right, G2 phase; n = 3). (B) Representative genomic view of BrdU-seq (in blue) and Chr-RNA-seq (in red) on the long arm of chromosome 7, with a 95-Mb view at all time points. (C) Single-gene analysis of BrdU incorporation levels as percentage of the input at each time point over the indicated genes; n = 3. (D) Nascent transcription levels as relative ratio to G1/S compared to the fold changes measured by Chr-RNA-seq; n = 3 for RT-PCR, n = 2 for Chr-RNA-seq. Data represent mean ± SEM. Student’s t test; ^∗p < 0.05; ^∗∗p < 0.01; ^∗∗∗p < 0.001; ^∗∗∗∗p < 0.0001.

Figure 2

Transient transcription shutdown when genes are replicated (A) Quantification of Chr-RNA-seq levels over introns at the indicated time points, as fold change to G1/S. Genes are separated based on directionality as in Figure S1J. (B) Nascent transcription levels in early S and early/mid-S compared to G1/S as in Figure 1D for genes replicated in early S; n = 3, data represent mean ± SEM. (C) As in (A), with genes clustered by length. (D) Average metagene profile of Chr-RNA-seq in G1/S and early S around the TSS −1 kb/+2 kb of genes replicated in early S. (E) Fold change in Chr-RNA-seq levels for genes with reduced nascent transcription when replicated compared to the time point before, with levels during and after replication. The analysis includes 514 genes in early/mid-S, 144 genes in mid-S, 178 genes in mid-/late S, and all 716 genes in late S/G2; paired t test analysis. (F) As in (D) but specifically over the genes highlighted in (E). Box whiskers plots with line at the median, Mann-Whitney t test; ns, not significant; ^∗p < 0.05; ^∗∗p < 0.01; ^∗∗∗p < 0.001; ^∗∗∗∗p < 0.0001.

Figure 3

TSSs of transcribed genes remain under-replicated throughout S phase (A) Average metagene profile of BrdU-seq levels normalized to input DNA in early S at TSSs ±2.5 kb for all genes, transcribed genes, or not transcribed genes. (B) As for (A) for genes replicated in early S according to Figure S1D, separated in four groups by gene length or transcription levels. (C) As for (A) for transcribed and not transcribed genes in all five fractions of the Repli-seq. (D) As for (A) for genes >100 kb replicated in early S in all time points. (E) BrdU incorporation at TSSs and in the gene body for the indicated genes in all time points. Schematic of gene structure with gene length, exons as orange boxes, and introns as red lines; primer positions are specified by the red boxes above; nascent transcription levels at TSSs and gene body are shown as fold change compared to G1/S for SMURF1 first time points shown in Figure 2B; n = 3, data represent mean ± SEM, Student’s t test. (F) Relative enrichment of BrdU-seq levels of all time points compared to the levels in early S for genes >100 kb replicated in early S. (G) As for (B), with genes separated based on their P³R² levels as from Figure S3F. ^∗p < 0.05; ^∗∗p < 0.01.

Figure 4

Persistence of RNAPII at TSSs prevents timely replication in S phase (A) Immunofluorescence for pS10-H3 (green) for G2/M-mitotic cells, EdU Click-iT for DNA synthesis (red), and DAPI (blue) for nuclei staining. G2/M-EdU, but not mitotic-EdU, double-positive cells quantified in CTR, NELFA, and SUPT5H siRNA cells; n = 4. (B) G2/M-EdU double-positive cells quantified in CTR DMSO and cells treated in early S for 1 h with DRB (100 μM); n = 3. (C) Distribution of G-MiDS-specific peaks in green in relation to replication timing on the long arm of chromosome 7, as in Figure S1A. (D) Average metagene profile for BrdU-seq levels normalized to input DNA in S-phase time points and G2/M at TSSs ±2.5 kb of transcribed genes. (E) Snapshots from IGV TDF (Integrative Genomics Viewer tiled data file) of G-MiDS-specific BrdU-seq (green) and BrdU-seq in S-phase time points (blue) around TSSs of the indicated genes. (F) Average metagene profile for the BrdU-seq levels normalized to input DNA at TSS ±2.5 kb of the 449 hotspot genes in cells transfected with the denoted siRNA. (G) As for (D) across all transcribed genes and only G-MiDS hotspots for asynchronous BJ cells treated with BrdU for 30 min and sorted in G2/M. (H) As for (G), with U2OS cells. Data represent mean ± SEM; Student’s t test; ^∗p < 0.05; ^∗∗p < 0.01.

Figure 5

Uncoupling of replication forks efficiency at origins of replication near TSSs (A) Average metagene profile for the denotated strand of strand-specific Ok-seq from Chen et al. (2019) TSSs ±50 kb of transcribed genes >100 kb in BJ-hTERT cells. (B) As for (A) but for TSSs ±10 kb, with orange and black arrows indicating the start positions of the Okazaki fragments on “+” or “−” strands. (C) Average metagene profile for Ok-seq from Chen et al. (2019) TSSs ±10 kb of transcribed genes >100 kb or G-MiDS hotspot genes in BJ-hTERT without strand specificity. (D) Average metagene profile for Ok-seq transcribed/not-transcribed strand from Petryk et al. (2016) TSSs ± 50 kb of transcribed genes >100 kb in HeLa cells on + or − strands. (E) Average metagene profile of MCM7 (Sugimoto et al., 2018), RPA2 (Zhang et al., 2017), and ORC1 (Dellino et al., 2013) ChIP-seq in HeLa cells at TSSs of transcribed genes >100 kb.

Figure 6

G-MiDS is not dependent on the DNA damage response and is distinct from MiDAS (A) Average metagene profile and heatmap for γH2AX/H2AX at G-MiDS-specific peaks ±10 kb. (B) Quantification of γH2AX/H2AX levels around TSSs ±1 kb of all transcribed genes and G-MiDS hotspots. (C) Quantification of γH2AX/H2AX levels around TSSs ±1 kb of all transcribed genes and G-MiDS hotspots in CTR cells and after KD of NELFA. (D) G2/M-EdU double-positive cells quantified in CTR DMSO cells or with the inhibitors (Rad51i = 25 μM, Rad52i = 20 μM, ATMi = 10 μM, ATRi = 4 μM, CD437 = 5 μM, aphidicolin [APH] = 10 μM, camptothecin [Campto] = 1 μM, and etoposide [Eto] = 10 μM) for 30 min once released from G2 arrest; DRB (100 μM) treated for 1 h before and then 30 min once released from G2 arrest; n ≥ 3. (E) Immunofluorescence for pS10-H3 (green) to label mitotic cells, EdU Click-iT for DNA synthesis (red), and DAPI (blue) for nuclei staining. Only mitotic-EdU double-positive cells from prometaphase on were quantified; cells were treated with DMSO or DRB as in (D); n = 4. (F) Immunofluorescence for Bloom (BLM; green) and DAPI (blue) for nuclei staining in cells released after the Ro3306 G2 arrest for 80 min in DMSO or DRB (100 μM), quantifying ultrafine bridges (UFB), anaphase bridges (AB), and lagging chromosomes (Lagg Chr), with examples highlighted by white arrows; n ≥ 3. Data represent mean ± SEM; Student’s t test; ns, not significant; ^∗p < 0.05; ^∗∗p < 0.01; ^∗∗∗∗p < 0.0001.

Figure 7

G-MiDS-hotspot-gene-specific features (A) As in Figures 3A and 3B for all transcribed genes and transcripts separated by gene length in cells transfected with the indicated siRNA. (B) Gene length and transcription levels at the late-S/G2 time point for G-MiDS hotspot genes toward all other transcribed genes. (C) Directionality analysis of G-MiDS hotspot genes compared with directionality of all transcribed genes. (D) Replication timing of G-MiDS hotspot genes compared with all transcribed genes. (E) Quantification of the ratio between antisense and sense transcription at TSSs ±1 kb of all transcribed genes and the 449 G-MiDS hotspots. (F) Heatmap analysis of the levels of antisense and sense transcription at TSSs ±2.5 kb for all the transcribed genes and G-MiDS hotspots on the + and − strands. (G) Model describing how the TSS is occupied by RNAPII and general transcription factors (GTF) throughout the cell cycle, with the RNAPII moving along genes. When DNA replication approaches the TSS during S phase, it may encounter GTF/RNAPII, skipping the TSS and restarting downstream of it. This may be mediated by the activation of origins of replication near TSSs. Later during S phase, DNA replication may fill the resulting gaps. However, in cases of genes with high steady expression levels of sense and TSS-associated antisense transcription, completion of the duplication of TSSs will occur in G2/M, when RNAPII and GTF are removed from TSSs. Box and whisker plots with the line at the median; Mann-Whitney t test; ^∗∗∗∗p < 0.0001.