Extended Data Fig. 6: dDSB fragments reveal preferential gap formation at sites of Msh2-independent full gene conversions.
From: Spo11 generates gaps through concerted cuts at sites of topological stress
a, Raw data of publicly available S288C/SK1 hybrid octads were re-aligned to the parental reference genomes and re-analysed to create a high-precision and high-confidence recombination event map, with focus on full gene conversions (‘6:2’ events) (Methods, Supplementary Table 4). Average number of crossover (CO), non-crossover (NCO) and double crossover (dCO) events of msh2∆ (four octads with on average 211.8 ± 32.8 events), msh2∆ mutLγ∆ (four octads, 189.0 ± 8.8 events), and msh2∆ exo1∆ (two octads, 148.5 ± 29.0) octads are depicted by doughnut plots, in which the outer ring represents all events, and the inner circle denotes the 6:2-containing events (with a radius proportional to their fraction). Among 6:2 events, crossovers are markedly overrepresented in msh2∆ mutants (P = 4.3 × 10−12), but not in msh2∆ mutLγ∆ double mutants (P = 0.54 Fisher’s exact test) as the MutLγ-independent full gene conversion events show an only slightly enhanced tendency for crossovers (29.1% of 6:2 versus 25.5% of total events). Because the preferential association of 6:2 events with crossovers disappears when MutLγ is defective, the crossover-associated 6:2 events seem to be generated by the MutLγ pathway, not by dDSBs. Deletion of another MutLγ component, Exo1, in conjunction with Msh2, yields even fewer full gene conversion events (14.8%), because it also affects the early stages in both DSB and dDSB processing, including the interhomologue bias44, and thus affects 6:2 formation from all pathways. b, c, As in Fig. 4f, g, but with msh2∆ octad data. Msh2-independent 6:2 events (n = 349 from four octads) have significantly higher dDSB (gap) coverage than randomly distributed events. In b, the cumulative fraction of Msh2-independent 6:2 events is plotted against the number of identified gaps that completely overlap them. Plots based on gap distributions from (left to right) wild-type, rad50S and spo11Y135F strains are shown. Results from distributing 6:2 events 1,000× randomly are included (light blue dots). Values above random indicate a positive correlation between the incidences of 6:2 events and gap probabilities. P values were determined by Wilcoxon rank test and Bonferroni multiple-test correction. In c, the dDSB fragment coverage of wild-type (top) and rad50S (bottom) strains per observed or 1,000× randomly distributed (light blue) 6:2 events are plotted. Median values of wild-type and rad50S were significantly different from random (Mood’s median test, P = 7.0 × 10−14 and 4.6 × 10−30, respectively). The central lines indicate the median, the boxes indicate the interquartile range, and whiskers span maximally 1.5 times the interquartile range.
