Timeless couples G-quadruplex detection with processing by DDX11 helicase during DNA replication

Abstract

Regions of the genome with the potential to form secondary DNA structures pose a frequent and significant impediment to DNA replication and must be actively managed in order to preserve genetic and epigenetic integrity. How the replisome detects and responds to secondary structures is poorly understood. Here, we show that a core component of the fork protection complex in the eukaryotic replisome, Timeless, harbours in its C-terminal region a previously unappreciated DNA-binding domain that exhibits specific binding to G-quadruplex (G4) DNA structures. We show that this domain contributes to maintaining processive replication through G4-forming sequences, and exhibits partial redundancy with an adjacent PARP-binding domain. Further, this function of Timeless requires interaction with and activity of the helicase DDX11. Loss of both Timeless and DDX11 causes epigenetic instability at G4-forming sequences and DNA damage. Our findings indicate that Timeless contributes to the ability of the replisome to sense replication-hindering G4 formation and ensures the prompt resolution of these structures by DDX11 to maintain processive DNA synthesis.

Keywords: DNA replication; G-quadruplex; fork protection complex; replisome; timeless.

Figure EV1. Sensitivity of wild type (WT), timeless, ddx11 and fancj DT40 mutants to cisplatin (CDDP)

Cell viability, assessed by MTS assay, of DT40 wild type, ddx11, timeless and fancj, after 72 h in presence of cisplatin at the indicated doses. The values represent the means (error bars indicate SD) of two independent experiments performed in triplicate. *P < 0.05, ***P < 0.001 and ****P < 0.0001; one‐way ANOVA compared to the wild type.

Figure 1. Timeless and Tipin are required to maintain processive replication past G4 structures in vivo

1. The BU‐1 locus as a model system to record G4‐dependent replication stalling. The leading strand of a replication fork entering the locus from the 3′ end stochastically stalls at the +3.5 G4, leading to the formation of a region of ssDNA, with interruption of parental histone recycling and of histone modifications necessary to maintain normal expression of the locus (Schiavone et al, 2014).

2. Instability of BU‐1 expression in timeless cells. FACS plots of wild‐type and timeless (clone 1) DT40 cells stained with anti‐Bu-1 conjugated with phycoerythrin. Each line represents the Bu‐1 expression profile of an individual clonal population. Unstained controls are shown in blue.

3. Fluctuation analysis for Bu‐1 loss in wild‐type DT40 cells and two independent timeless clones generated by CRISPR‐Cas9 targeting (clones 1 and 2; Appendix Fig S1), timeless (clone 1) complemented by expression of human Timeless cDNA and a timeless mutant on a background in which the endogenous +3.5 G4 has been deleted (ΔG4) (Schiavone et al, 2014).

4. Fluctuation analysis for Bu‐1 loss in DT40 wild‐type and tipin cells.

Data information: In (C) and (D), each symbol represents the percentage of cells in an individual clone expanded for 2–3 weeks that have lost Bu‐1^high expression. At least two independent fluctuation analyses were performed, with 24–36 individual clones each cell line per repeat. Bars and whiskers represent median and interquartile range, respectively. ****P < 0.0001; one‐way ANOVA.

Figure 2. Identification and characterisation of a DNA‐binding activity in Timeless

1. Schematic drawing of human Timeless and its known domain structure (NTD: N‐terminal domain; DBD: DNA‐binding domain; PBD: PARP‐binding domain). A multiple sequence alignment of vertebrate Timeless sequences is shown underneath, with amino acid conservation coloured according to the Clustal colour scheme. The alignment is annotated with the extent and secondary structure elements of the two helical domains (N‐term and C‐term) composing the DBD.

2. Ribbon drawing of the 1.15 Å crystal structure at of the DBD C‐term. Helices are in red and labelled H1–H4.

3. Ribbon drawings of the N‐term and C‐term domains of the DBD determined by NMR. The two domains are shown in the same orientation to highlight their high degree of three‐dimensional similarity. The superposition of the 20 lowest energy structures is shown for each domain.

4. The DNA‐binding affinity of DBD was measured by fluorescence anisotropy, titrating the DBD protein against Cy3 3′‐labelled ssDNA, dsDNA and G4 DNA (see Appendix Table S2 for sequence details). The top panel shows binding curves for ss‐ and dsDNA, and the bottom panel shows the binding curve for the G4 DNA substrate. The data points represent the mean of at least three independent experiments, and the error bars indicate one standard deviation (SD).

5. Ribbon diagram of the superposition of DBD C‐term with the highly similar DNA‐binding domains of telomeric protein TRF1 (PDB ID 1W0T) (Court et al, 2005) and the bacterial cell cycle regulator GcrA (PDB ID 5Z7I) (Wu et al, 2018) in complex with their DNA substrates. A similar DNA‐binding mode by DBD would cause a steric overlap of helix H4 with the phosphate backbone of dsDNA. DBD C‐term is in light blue, TRF1 and GcrA proteins in brown and their DNA substrates in khaki.

Figure 3. The Timeless–Tipin complex shows a preference for binding G4 DNA

Fluorescence anisotropy was used to measure the binding affinity of Timeless–Tipin for the indicated DNA sequences.

1. ssG4: G4 flanked by single‐stranded DNA; ssHP: hairpin flanked by single‐stranded DNA; ss: single‐stranded DNA (Appendix Table S2 for sequence details).

2. Binding affinity of Timeless–Tipin for a range of G4 DNA sequences (see Appendix Table S2 for sequence details and references). Single‐stranded (ss20: 5′‐6FAM-ATAAGAGTGGTTAGAGTGTA) and double‐stranded (ds20: ss20 annealed to complementary sequence) DNA were also tested as controls.

Data information: Each data point is the mean of at least 3 independent experiments and the error bars indicate one SD.

Figure EV2. The Timeless–Tipin complex shows a preference for binding G‐quadruplex DNA structures

1. Coomassie‐stained SDS–PAGE gel of purified Timeless–Tipin complex.

2. Electrophoretic mobility shift assay (EMSA) showing the binding of Timeless–Tipin to G‐quadruplex sequence BU1A + 3.5. Mutation of the G‐quadruplex sequence (BU1A + 3.5 mut) disrupts Timeless–Tipin binding (see Appendix Table S2 for sequence details). Timeless–Tipin and DNA are both present at a final concentration of 5 μM.

3. EMSA showing the binding of Timeless–Tipin to G‐quadruplex sequences (ssG4, dsG4) but not single‐stranded DNA (ss), double‐stranded DNA (ds) or hairpin‐containing sequences (ssHP, dsHP). Timeless–Tipin and DNA are both present at a final concentration of 5 μM.

Figure 4. The C‐terminus of Timeless is required for processive G4 replication

Fluctuation analysis for the generation of Bu‐1 loss variants. Top to bottom: wild type, timeless (clone 1), a timeless mutant (timeless ∆C) generated by CRISPR‐Cas9 targeting exon 16 which truncates the protein removing the CTD containing both the DBD and the PARP‐binding domains. Then, complementation of timeless#1 with human Timeless (hTim), hTim∆816–1,208 (lacking both the DBD and PBD), hTim∆816–965 (lacking the DBD) and hTim[1:1,000], lacking the PBD. At least two independent fluctuation analyses were performed with 24–36 individual clones each cell line per repeat. Bars and whiskers represent median and interquartile range, respectively. *P < 0.05 and ****P < 0.0001; one‐way ANOVA for comparison with the wild‐type cells.

Figure EV3. The C‐terminus of Timeless is not required for its interaction with DDX11

HEK293T cells were transiently transfected with a plasmid encoding Flag‐hTimeless, or co‐transfected with plasmids encoding hDDX11 and Flag‐Timeless or with Timeless mutated to delete the DNA‐binding domain (ΔDBD: deletion of region S816–S965) or PARP‐binding domain (PARP*: truncation at V1000). Twenty‐four h after transfection, whole‐cell extracts were subjected to immunoprecipitation with anti‐Flag magnetic beads. Western Blot analyses were performed to detect overexpressed DDX11 protein in the pulled down samples using a specific antibody. Upper panel: Input and pulled down samples transfected with different Timeless constructs detected with an anti‐Flag antibody. Bottom panel: Input and pulled down samples transfected with different Timeless constructs detected with an anti‐DDX11 antibody. Tubulin was used as a loading control for the input samples.

Figure 5. DDX11 is required for processive replication in collaboration with Timeless

1. Enhanced recruitment of DDX11 to chromatin associated PCNA following exposure to 4 μM PDS for 24 h. PCNA was precipitated from cross‐linked chromatin and the immunoprecipitate blotted for FLAG‐DDX11.

2. Fluctuation analysis for Bu‐1a loss in wild‐type DT40 cells, two independent ddx11 clones generated by CRISPR‐Cas9 targeting (clones 1 and 2) and one ddx11 clone generated by conventional homologous recombination gene targeting (clone 3). Each symbol represents the percentage of cells in an individual clone expanded for 2–3 weeks that have lost Bu‐1a^high expression.

3. Fluctuation analysis for Bu‐1a loss variant generation in wild‐type cells, ddx11 (clone 1) cells, ddx11 (clone 1) complemented by expression of chicken DDX11 WT cDNA, ddx11 (clone 1) complemented by expression of helicase‐dead form of chicken DDX11 (K87A) cDNA, and a ddx11 clone generated in cells in which the endogenous +3.5 G4 has been deleted (ΔG4).

4. Fluctuation analysis for Bu‐1 loss in two independent timeless/ddx11 double‐mutant clones (#1 and #2), ddx11 expressing DDX11^KAK (see Appendix Fig S3), and fancj and fancj/ddx11 double mutants. Fluctuation analyses for wild type, timeless #1 (Fig 44) and ddx11 #1 (Fig 55A) are shown for comparison.

Data information: In all cases, at least two independent fluctuation analyses were performed, with 24–36 individual clones each cell line per repeat. Bars and whiskers represent median and interquartile range, respectively. ****P < 0.0001; one‐way ANOVA for comparison with wild‐type cells.

Figure EV4. The catalytic activity of FANCJ is required for its role in suppressing G4‐induced instability of BU‐1 expression

Fluctuation analysis for the generation of Bu‐1 loss variants in an inducible system to study FANCJ function (see Materials and Methods for full details). Briefly, FANCJ‐deficient DT40 cells are rescued with two transgenes, one encoding the wild‐type protein and the other the mutant in question. The wild‐type transgene is flanked by loxP sites and can be deleted by expression of Cre recombinase, induced by treatment of cells with tamoxifen. The K52R and Q25A mutants of FANCJ both disrupt the helicase activity of the enzyme (Cantor et al, 2001; Wu et al, 2012b). S990A disrupts the interaction of FANCJ with BRCA1, which is important for the role of FANCJ in homologous recombination (Xie et al, 2010). K141/142A disrupts the interaction of FANCJ with MutLα, which is needed for efficient interstrand crosslink repair (Peng et al, 2007). The data for each cell line represent the pooled results of at least two independent fluctuation analyses with a minimum of 32 data points per condition. Bars and whiskers represent median and interquartile range, respectively. ****P < 0.0001; one‐way ANOVA for comparison between uninduced and tamoxifen‐induced lines.

Figure 6. Timeless and DDX11‐deficient cells have impaired growth and increased H2AX phosphorylation in the presence of a G4 ligand

1. Growth curves for DT40 wild type, ddx11, timeless and ddx11/timeless cells, with and without 4 μM pyridostatin (PDS). Cells were seeded at 5 × 10⁴ cells/ml on day 0 and the viable cells were counted each 24 h for 4 days. Bars represent SD of two independent experiments performed in duplicate. Doubling times (DMSO): WT 13 h, timeless 18 h, ddx11 16 h, ddx11/timeless 24 h. Doubling times (PDS): WT 13.6 h, timeless 27 h, ddx11 25.7 h, ddx11/timeless 47.5 h.

2. DDR signalling detected by phosphorylation of histone H2AX (γ‐H2AX) by flow cytometry in untreated cells or cells exposed to 4 μM PDS for 3 days. Pale histogram, untreated; dark histogram, treated; black dotted line, positive control cells treated with 0.1 μM cisplatin, also for 3 days.

3. Quantification of γ‐H2AX in DT40 wild type, ddx11, timeless and ddx11/timeless cells treated with 4 μM PDS for 3 days. The central band represents the median, the box the 25^th–75^th centile and whiskers the minimum to maximum range of three independent experiments performed in duplicate. *P < 0.05, **P < 0.01, ***P < 0.001 unpaired, two‐tailed t‐test for each pairwise comparison ± PDS. See also Appendix Fig S4 for immunofluorescence images of the γ‐H2AX signal.

Figure EV5. Gene expression dysregulation in timeless and ddx11 DT40 cells

1. Dysregulated genes in timeless (left panel), ddx11 (centre panel) and fancj (right panel) mutants relative to wild type. All genes with > 1 transcript per million in both conditions are plotted. Significantly (P ≥ 0.95) upregulated genes shown in red; downregulated in blue.

2. Venn diagram showing the overlap in genes deregulated in timeless, ddx11 and fancj relative to wild type. P < 2.2 × 10⁻¹⁶ for each pairwise comparison (Fisher hypergeometric distribution).

3. GC content around the TSS in genes dysregulated in timeless (blue), ddx11 (red) and in both mutants (“overlap”, orange) compared with all genes (black).

4. Correlation of magnitude and direction of change of genes dysregulated (relative to wild type) in fancj vs. timeless (left panel) and ddx11 (right panel) DT40 cells. r _s (Spearman rho) is shown for each correlation.

Figure 7. Loss of DDX11 and Timeless leads to genome wide expression dysregulation of genes with G4s around the transcription start site (TSS)

1. Correlation of magnitude and direction of change of genes dysregulated (relative to wild type) in timeless vs. ddx11 DT40 cells. r _s (Spearman rho) is shown for each correlation.

2. Genes dysregulated in timeless, ddx11 and fancj mutants have a higher density of G4s around their TSS. Cumulative fraction of the genes dysregulated in timeless (red), ddx11 (blue) and fancj (green) containing n (x‐axis) G4 motifs within 1.5 kb of the TSS compared with all genes (black). P values calculated with the Kolmogorov–Smirnov test.

3. Metagene analysis showing G4 frequency around the TSS of genes dysregulated (up or down) in timeless (left panel), ddx11 (centre panel) and fancj (right panel) compared with all genes (black line). G4 frequency is calculated separately for coding (above the x‐axis) and template strands (below the x‐axis).

4. A model for recognition and processing of replisome associated G4s by Timeless/DDX11. Current evidence suggests that Timeless is a constitutive component of the replication fork. We suggest that the C‐terminus of Timeless may help detect G4s in the vicinity of the replisome by a combination of direct recognition through the DNA‐binding domain and indirectly through the PARP‐binding domain. It is not currently possible to distinguish whether this mechanism would operate ahead or behind the fork itself (Lerner & Sale, 2019), although recent structural evidence placing Tof1, the yeast homologue of Timeless, ahead of the fork (Baretić et al, 2020) would appear to make the first possibility more probable. However, for failure to resolve G4s ahead of the fork to result in uncoupling would require the CMG helicase to traverse the structure, as has been suggested for interstrand crosslinks (Sparks et al, 2019).