Replication Stress Induces Global Chromosome Breakage in the Fragile X Genome

Abstract

Fragile X syndrome (FXS) is a neurodevelopmental disorder caused by mutations in the FMR1 gene and deficiency of a functional FMRP protein. FMRP is known as a translation repressor whose nuclear function is not understood. We investigated the global impact on genome stability due to FMRP loss. Using Break-seq, we map spontaneous and replication stress-induced DNA double-strand breaks (DSBs) in an FXS patient-derived cell line. We report that the genomes of FXS cells are inherently unstable and accumulate twice as many DSBs as those from an unaffected control. We demonstrate that replication stress-induced DSBs in FXS cells colocalize with R-loop forming sequences. Exogenously expressed FMRP in FXS fibroblasts ameliorates DSB formation. FMRP, not the I304N mutant, abates R-loop-induced DSBs during programmed replication-transcription conflict. These results suggest that FMRP is a genome maintenance protein that prevents R-loop accumulation. Our study provides insights into the etiological basis for FXS.

Keywords: DNA double-strand breaks; DNA replication stress; DSB; FMRP; FXS; I304N; R-loops; chromosome fragile sites; fragile X syndrome; genome instability.

Figure 1.. Fragile X (FX) Cells Show Elevated DNA Damage under Replication Stress and an Intact DDR

(A) Western blots confirming the absence of FMRP expression in FX cell lines. (B) Increased γH2A.X foci formation in FX lymphoblastoid cells in APH. Scale bar, 10 μm. See also Figure S2. (C and D) Analysis of γH2A.X signal under APH induction by flow cytometry. The solid red boxes indicate the gated live cells, with γH2A.X signals greater than baseline. Two-way ANOVA followed by Sidak’s multiple testing was performed. Error bars indicate standard deviation. *p < 0.05 and ***p < 0.001. (E) DSB formation measured by neutral Comet assay. Three independent experiments were performed (n > 50 in each) and all trended similarly. The tail lengths from one representative experiment are shown in the boxplot. Representative images of comets with short or long Comet tail lengths are shown. (F) Expression of phospho-ataxia telangiectasia mutated (p-ATM) in NM and FX lymphoblastoid cells indicates intact DDR in FX cells. Western blots probed with anti-p-ATM and total ATM are shown in black- and blue-bordered panels, respectively. Both blots were controlled for loading using anti-glyceraldehyde 3-phosphate dehydrogenase (GAPDH). The ratios of p-ATM:total ATM are shown below the blots. The error bars indicate standard errors of the mean.

Figure 2.. FMRP Expression Ameliorates APH-Induced DSB Formation in FX Cells

(A) γH2A.X immunofluorescence in FX fibroblast cells (GM05848) carrying pMSCVpuro-EGFP or pMSCVpuro-EGFP-FMRP. Scale bar, 10 μm. (B) Quantitation of the γH2A.X immunofluorescence signals in (A) by boxplot. (C) EGFP and EGFP-FMRP fusion protein expression in the respective cells in (A).

Figure 3.. DSB Mapping by Break-Seq in Lymphoblastoid Cells

(A) Break-seq workflow. (B) Break-seq sample read pile-up of a PacI site leading to the identification of a “peak” in the “proof-of-principle” experiment. The table denotes total PacI sites identified in DMSO-treated and (0.03 μM) APH-treated FX cells. (C) Schematic for Break-seq analysis. The sequence reads were aligned with Bowtie2, followed by model-based analysis of ChIP-seq 2 (MACS2) peak calling while normalizing for copy-number variation by whole-genome sequencing (“total DNA”) for NM and FX cells, respectively (shown for FX cells only). DSB peaks found in at least 2 replicate experiments for each strain/treatment combination were identified as “consensus DSBs” by DiffBind. Peaks from different APH treatments (0.03 and 0.3 μM) were then pooled into a single set of “FX_ APH” DSBs, in contrast to the control datasets. The consensus DSBs for each strain/treatment combination were compared with each other (e.g., between “FX_APH” and “FX_DMSO”) to identify overlaps and condition-specific DSBs, ready for further comparison with genomic features such as RLFSs. (D) Whole-genome distribution of DSBs in the indicated categories.

Figure 4.. DSB Association with Genomic Features

(A) Plot of total number of DSBs in each of the indicated categories. (B) DSB density (per megabase of DNA) across each chromosome for the indicated categories. (C) Venn diagrams compare concordance between NM and FX cells for every treatment. (D) Distribution of DSBs in early-versus late-replicating regions of the genome, as defined by Hansen et al. (2010) in the indicated samples (see Method Details). (E) Distribution of DSB peaks relative to genes in the indicated samples. Genic features include introns, exons, 5′- and 3′ UTRs, promoters, and the immediately downstream (<1 kb from the 3′ UTR) regions (ImmediateDownstream). Note the break in the x axis to show all of the genic features. (F) Correlation between DSBs associated with genes and with early-replication timing sequences. (G and H) Gene Ontology (GO) terms for genes associated with DSBs in FX-untreated (G) and APH-treated (H) samples. Plotted are the BH (Benjamini-Hochberg) adjusted p values for the GO terms. (I) Examples of genes containing drug-induced DSBs specifically in the FX cells. The top tier of the plots annotate the DSB positions in genes labeled red or blue for Watson- or Crick-strand-encoded, respectively. The bottom tier show sequence read distribution in each data track, with increasing numbers following a blue to red color scale.

Figure 5.. DSB Correlation with RLFSs and Increased RNA:DNA Hybrids in FX Cells In Vivo

(A and B) Aggregated DSBs from the indicated samples around the start (A) or the end (B) of RLFS in a 4,000-bp window centering on the RLFS. (C) Confocal images of immunofluorescence staining with the S9.6 antibody. The nuclear boundary is traced by Lamin staining (inset 1). The RNA:DNA hybrid signals are enriched in the areas of the nucleus lacking DAPI staining, which are presumed the nucleoli (arrowhead in inset 2). Scale bar, 20 μm. (D) Quantification of nuclear S9.6 signals in two independent experiments (n ≥ 33 in each) using one-way ANOVA with multiple comparisons. *p < 0.0001. (E) Validation of the co-localization of RNA:DNA hybrids with nucleolus by co-staining with S9.6 and anti-nucleolin. Shown is an example of DMSO-treated FX cells. Similar observations were made with APH-treated cells (data not shown). (F) Nuclear S9.6 signals were reduced by the ectopic expression of EGFP-FMRP, compared to the EGFP control, in FX cells. Quantification was done similarly as in (D). *p < 0.05 and ****p < 0.0001.

Figure 6.. FMRP Expression Suppresses RLFS-Induced DSB Formation

(A) A non-functional LEU2 marker containing 2 inserted direct repeats and driven by a galactose-inducible GAL1 promoter was placed next to an origin of replication (ARSH4), such that the direction of transcription is convergent or co-directional with respect to the direction of the proximal replication fork. Upon galactose induction, convergent replication and transcription would induce DSBs and homologous recombination repair to generate a functional LEU2, resulting in leucine prototrophy. Two RLFSs from the human genome (RLFS1–1 from the promoter of FMR1 and RLFS-2 from intron 5 of the fragile histidine triad) were inserted between the direct repeats to test for enhanced DSB and recombination. A non-RLFS sequence without predicted R-loop forming propensity and with similar G-richness in both strands served as the control. All of the sequences were similar in size (~500 bp). The RLFSs were inserted in the sense or anti-sense orientation with respect to LEU2 transcription (i.e., G-rich strand on the non-template or template strand, respectively), with the sense orientation expected to preferentially induce R-loop formation. The control sequence was also inserted in two orientations, and no difference in RF was observed between them (see Figure S5C). RF is calculated based on the percentage of leucine prototrophs after plating. (B) The effect of ectopic expression of indicated genes on the pRS313 plasmid, under the cytomegalovirus (CMV) promoter, on RLFS-induced RF. Error bars indicate standard deviation. **p < 0.01, ***p < 0.001, and ****p < 0.0001. See Figure S7 for additional control experiments.

Figure 7.. Proposed Model for FMRP R-Loop Regulation and DSB Prevention

(A) Illustration of a normal cell without any treatment, showing FMRP in the cytoplasm and in the nuclear periphery, possibly engaged in mRNA transport. (B) Under replication stress induced by DMSO and APH, FMRP increases its presence in the nucleus. At the junction of replication and transcription collision, FMRP, in conjunction with R-loop processing factors such as the THO-TREX complex, is involved in R-loop removal and avoidance of a deleterious collision (inset). dsDNA, double-stranded DNA. (C) In FX cells, increased protein synthesis rate demands high-level rRNA production on the rDNA-bearing chromosomes, which in turn causes increased levels of (RNA Pol II) transcription elsewhere on these chromosomes, represented by the chromosome loops tethered to the nuclear pores for active transcription. The absence of FMRP permits stable R-loop formation and DSBs upon the collision of replication and transcription (inset). ssDNA, single-strand DNA.