Meiotic DNA break resection and recombination rely on chromatin remodeler Fun30

Abstract

DNA double-strand breaks (DSBs) are nucleolytically processed to generate single-stranded DNA for homologous recombination. In Saccharomyces cerevisiae meiosis, this resection involves nicking by the Mre11-Rad50-Xrs2 complex (MRX), then exonucleolytic digestion by Exo1. Chromatin remodeling at meiotic DSBs is thought necessary for resection, but the remodeling enzyme was unknown. Here we show that the SWI/SNF-like ATPase Fun30 plays a major, nonredundant role in meiotic resection. A fun30 mutation shortened resection tracts almost as severely as an exo1-nd (nuclease-dead) mutation, and resection was further shortened in a fun30 exo1-nd double mutant. Fun30 associates with chromatin in response to DSBs, and the constitutive positioning of nucleosomes governs resection endpoint locations in the absence of Fun30. We infer that Fun30 promotes both the MRX- and Exo1-dependent steps in resection, possibly by removing nucleosomes from broken chromatids. Moreover, the extremely short resection in fun30 exo1-nd double mutants is accompanied by compromised interhomolog recombination bias, leading to defects in recombination and chromosome segregation. Thus, this study also provides insight about the minimal resection lengths needed for robust recombination.

Keywords: Exo1; Fun30; Meiosis; Recombination; Resection.

Figure 1. Testing chromatin remodeler mutants for shortened meiotic resection.

(A) Left, an overview of meiotic DSB formation and resection within the context of local chromatin structure. Right, schematic of S1-Southern blotting and S1-seq methods. RE restriction enzyme. (B) Resection endpoint distributions detected by S1-Southern blotting at the GAT1 (top) and CCT6 (bottom) hotspots in the indicated mutants. All samples were collected at 4 h in meiosis. Vertical black lines to the left of the gene maps indicate probe positions. P parental length restriction fragments. Blot images are representative of n = 3 biological replicates except for sae2∆ and sae2∆ fun30∆ (n = 2 biological replicates). Source data are available online for this figure.

Figure 2. The meiotic resection landscape in fun30∆ mutants.

(A) S1-seq signals around the CCT6 hotspot in reads per million mapped (RPM). Reads mapping to the top strand are shown in blue; bottom-strand reads are in red. All S1-seq data were the average of two biological replicates collected at 4 h in meiosis. Spo11-oligo data are from (Pan et al, 2011). (B) Genome average of S1-seq signal around 3908 hotspots. Bottom-strand reads were reoriented and combined with the top strand to calculate the average. Data were smoothed with a 100-bp Hann window. (C) Histograms of resection tract lengths calculated for “loner” hotspots that had no other hotspot within 3 kb (n = 405). Lighter colored bars indicate tracks that were omitted to calculate the censored median estimates shown in parentheses. Censoring had little effect, indicating that the measurements are not strongly influenced by outliers. (D) Schematic comparing FUN30 and fun30∆ for the distance to the most distal MRX/Sae2 nicking positions (as measured in exo1-nd mutants) and the inferred lengths of Exo1 digestion.

Figure 3. DSB-dependent recruitment of Fun30 to chromatin.

(A–C) Fun30-myc ChIP-seq data were the average of 2 biological replicates collected at 4 h in meiosis. The DSB map (S1-seq in a sae2∆ mutant) is the same dataset shown in Appendix Fig. S1A,B as sae2∆ #2. The Rec114-myc ChIP-seq sample collected at 4 h in meiosis is from a previous study (Murakami and Keeney, 2014). Fun30 ChIP-seq datasets were normalized using a spike-in control. All data were smoothed using a 1 kb Parzen (triangular) sliding window. (A) Fun30 ChIP-seq signals across a representative region of chromosome III. The upper graph shows the normalized ChIP-seq coverage profiles for the wild type, the DSB-defective spo11-yf mutant, or a wild-type strain carrying untagged Fun30. The lower graph shows the DSB-dependent component of the Fun30 ChIP-seq signal, obtained by subtracting the normalized values from the spo11-yf strain from the values from the tagged wild-type strain. DSBs and Rec114 ChIP-seq profiles are shown for comparison. (B) Average Fun30 ChIP-seq signals around previously defined DSB hotspots (Mohibullah and Keeney, 2017) and Rec114 peaks (Murakami and Keeney, 2014). Profiles around a set of random genomic positions are shown for comparison. (C) Correlations (Pearson’s r) between the DSB-dependent Fun30 ChIP-seq signals (summed in 1-kb windows) and both DSBs (left graph) or Rec114 ChIP-seq (right graph).

Figure 4. S1-seq distribution relative to nucleosomes.

(A) Average S1-seq resection signal (colored lines; 41-bp smoothed) and MNase-seq (gray filled; 0 h in meiosis in wild-type (Pan et al, 2011)) centered on midpoints of +3 or +1 nucleosomes with no other hotspots ≤3 kb downstream (n = 1815 for +3 nucleosomes and 1832 for +1 nucleosomes). The color coding for the genotype for the S1-seq profiles is preserved throughout this figure. (B) Tighter view of the fun30∆ exo1-nd resection profile from panel A. The boxplot below shows that substantial S1-seq signals overlap with the MNase signals within the +1 nucleosome. The total S1-seq reads in the +1 nucleosome (−73 to +73 bp relative to the midpoint) and NDR (-219 to -74) regions were taken as 100%, and the percentages of S1-seq in the four regions are shown in light blue numerals. Boxplot elements are as described in Fig. EV1C. (C, D) Resection tract lengths in the absence of Fun30 correlate with basal nucleosome occupancy. S1-seq profiles from fun30∆ (blue) or fun30∆ exo1-nd (magenta) are shown. Hotspots were grouped according to resection patterns using k-means clustering applied to spatial distributions of S1-seq signals in fun30∆ (C) or fun30∆ exo1-nd (D). The fun30∆ curves differ between (C, D) because the clusters contain different subgroups of hotspots in each panel. We used hotspots that were separated from their nearest neighboring hotspot by >1 kb and considered each side of each hotspot separately (2778 hotspots and 5556 hotspot sides). The S1-seq and premeiotic (t = 0 h) MNase-seq maps were averaged within each cluster and plotted as a function of distance from the hotspot midpoint. The number of hotspot sides in each cluster (n) and median resection lengths are indicated.

Figure 5. Decreased spore viability and recombination defects in fun30∆ exo1-nd.

(A) Spore viability of the indicated strains (n = 110 tetrads generated from mass mating followed by sporulation for each genotype (see methods), number of replicates = 1). (B) Spore viability patterns for the tetrads shown in (A). (C) Estimates of contributions from random spore death and MI nondisjunction to total spore death. Tetrad dissection data from (B) were subjected to the TetFit algorithm (Chu and Burgess, ; Chu et al, 2017). Data from iml3∆ and msh5∆ (Chu and Burgess, 2016) are shown for comparison. (D) MI nondisjunction (NDJ) frequency measured using spore-autonomous fluorescent markers (Thacker et al, 2011). The numbers of scored tetrads were 1241 (wild-type), 1644 (exo1-nd), 1913 (fun30∆), and 2053 (fun30∆ exo1-nd). Error bars indicate 95% confidence intervals calculated by binomial distribution. Numbers above bar graphs indicate P values calculated by a two-sided exact binomial test (MI-NDJ). n.s. not significant (P > 0.05). Source data are available online for this figure.

Figure 6. Diminished interhomolog bias at HIS4LEU2 in fun30∆ exo1-nd mutants.

(A) One-dimensional (1D) gel analysis of DSBs and crossovers (COs) at the HIS4LEU2 hotspot. (B) Quantification of DSBs and crossovers from 1D gel analysis. DSB and crossover levels are shown as a percentage of the total hybridization signal per lane. (C) Meiotic division time courses. The graph shows the percentage of cells that have completed one or both divisions. (D) Representative two-dimensional (2D) gels of SEIs and dHJs at HIS4LEU2. Arrows indicate interhomolog joint molecules (green) or Mom–Mom (red) and Dad–Dad (blue) intersister molecules (color coding as in Fig. EV3A). (E) Quantification of SEIs and dHJs from 2D gel analyses. In all graphs, the data are the mean ± SD for three independent meiotic cultures. (F) Representative gel images of crossovers and noncrossovers. (G) Quantification of crossovers and noncrossovers (mean ± SD for three independent meiotic cultures). Source data are available online for this figure.

Figure 7. Interhomolog bias at the ERG1 hotspot.

(A) Physical map of the ERG1 hotspot on chromosome VII showing diagnostic SacII restriction enzyme sites and the position of the Southern blot probe. Parental chromosomes were distinguished by SacII restriction enzyme site polymorphisms. (B) Two-dimensional (2D) gel Southern blot analysis of SEIs and dHJs at ERG1. Arrows indicate interhomolog dHJs (green) and either Mom–Mom (red) or Dad–Dad (blue) intersister dHJs. (C) Quantification of dHJs at ERG1 from 2D gel analysis. The data are the mean ± SD for three independent meiotic cell cultures. (D) Progression of interhomolog bias at HIS4LEU2 throughout meiotic prophase. The strength of interhomolog bias was estimated by calculating the log₂-transformed ratio of IS-dHJ to IH-dHJ from the dataset in Fig. 6E (mean ± SD of three biological replicates). Smaller values indicate stronger interhomolog bias. (E) Different impacts of shortening resection tract length on interhomolog bias and crossover formation at the HIS4LEU2 locus. Blue points show the log₂-transformed ratio of IS-dHJs to IH-dHJs calculated at the time point with maximal total dHJ signal for each individual replicate (wild-type: 4, 5, 4 h; fun30∆: 5, 6, 5 h; exo1-nd: 4, 5, 5 h; fun30∆ exo1-nd: 7, 7, 7 h). Gray points show the average of the maximum crossover frequency for each time course. Values were calculated from the datasets in Fig. 6B,E. Error bars indicate mean ± SD. Source data are available online for this figure.

Figure EV1. DSB-dependent Fun30 enrichment.

(A) Total Fun30 ChIP-seq coverage normalized to the spike-in control. Bars are the means from two biological replicates; open circles show the individual values for each replicate. (B) Average Fun30 ChIP-seq signals around ARS, tRNA, and centromere. The random sites here and in (C) are the same as in Fig. 3B. (C) DSB-dependent Fun30 enrichment. Box plots summarize the distributions across all of the indicated elements from (B) and Fig. 3B for Fun30 ChIP-seq signal summed in 1-kb windows. Note the different y-axis scales for left and right parts of the plot. In all box plots, thick horizontal bars denote medians, box edges mark the upper and lower quartiles, and whiskers indicate values within 1.5-fold of the interquartile range. Outliers are not shown. Here and in (D), numbers above brackets indicate p values of two-sided Wilcoxon tests. (D) ARS and tRNAs that are closer to Rec114 binding sites tend to exhibit higher DSB-dependent Fun30 ChIP-seq signals. The ARS and tRNA regions from (B, C) were subdivided into two groups based on the distance to the nearest Rec114 peak or hotspot center: “close” indicates elements less than the median distance away and “far” indicates the rest. Proximity to Rec114 peaks was associated with a significantly higher DSB-dependent Fun30 ChIP-seq signal, whereas proximity to hotspots showed no such pattern. These results suggest that at least some of the DSB-dependent recruitment of Fun30 to ARS or tRNA genes is a consequence of fortuitous proximity or overlap of (some of) these elements with Rec114 ChIP peaks.

Figure EV2. MRX/Sae2 nicks within the + 1 nucleosome.

Examples of resection endpoint distributions in exo1-nd fun30∆ at three representative loci that contributed to the average shown in Fig. 4B. S1-seq signals (41-bp smoothed) from the top (blue) or bottom (red) strand are shown, dependent on the orientation of the gene where the +1 nucleosome is located. Numbers in light blue indicate the percentages of S1-seq signal in the four windows spanning the +1 nucleosome and NDR (see legend to Fig. 4B).

Figure EV3. Physical assay detecting recombination intermediates at the HIS4LEU2 hotspot.

(A) Physical map of the HIS4LEU2 locus showing diagnostic XhoI restriction enzyme sites and the position of Southern blot probe A. “Mom” and “Dad” indicate the two parental versions of the locus; COs, crossovers; NCOs, noncrossovers; MM’ IS-dHJ, intersister double-Holliday junction; MD IH-dHJ, interhomolog double-Holliday junction; DD’ IS-dHJ, intersister double-Holliday junction; SEIs, single-end invasions. Positions of XhoI sites are indicated as circled Xs. (B) Example one-dimensional gel analysis showing parental signals, DSBs, COs, NCOs, and joint molecules (JMs). The Southern blot images are reproduced from Fig. 6A,F. (C) Example two-dimensional gel displaying parental signals and recombination intermediates. Green arrow or text indicate interhomolog species; red and blue arrows and text indicate intersister species. (D) Key steps in crossover and noncrossover formation during meiosis.

Figure EV4. DSBs formation at various natural hotspots.

(A–H) Representative one-dimensional gel analyses of DSBs and corresponding quantification at the ERG1 (A, B), CYS3 (C, D), BUD23 (E, F), and ARG4 (G, H) hotspots. Error bars indicate mean ± SD for three independent cultures.

Figure EV5. Schematic representation of DSB-dependent Fun30 recruitment in the tethered loop-axis complex.

A model proposed based on findings in this study. In response to Spo11 cleaving DNA within the TLAC, Fun30 is recruited to the DSB ends and remodels the nucleosomes.