Csm2-Psy3 DNA binding is important for Shu complex function. a Surface view of S. cerevisiae Csm2 (light gray; K189, R190, R191, R192) and Psy3 (dark gray; K199, R200, K201) with the predicted DNA-binding residues highlighted in magenta and predicted DNA-binding loops in light and dark blue, respectively (ref. ; model structure derived from PDB 3VU9). b In vitro analysis of Csm2-Psy3 binding to a DNA fork substrate compared to Csm2-Psy3 DNA-binding mutants (Csm2-K189A/R190A/R191A/R192A and/or Psy3-K199A/R200A/K201A) by fluorescence anisotropy. Increasing concentrations of Csm2-Psy3 or the indicated mutants were added to 25 nM 3′-fluorescein-labeled double-flap substrate and DNA binding was assessed. Dissociation constants (K_d) and associated standard deviations from triplicate experiments were determined by non-linear curve fitting to a one-site binding model. c Cells expressing the csm2-KRRR psy3-KRK double mutant exhibit increased MMS sensitivity relative to csm2-KRRR or psy3-KRK cells. The DNA-binding residues shown in a were mutated to alanines and integrated into the genomic CSM2 and PSY3 loci. Fivefold serial dilution of WT, csm2∆, psy3∆, csm2-KRRR, psy3-KRK, and csm2-KRRR psy3-KRK cells onto rich YPD medium or YPD medium containing 0.02% MMS were incubated for 2 days at 30 °C prior to being photographed. d Spontaneous and MMS-induced mutation rate at the CAN1 locus were measured in WT, csm2∆, psy3∆, csm2-KRRR, psy3-KRK, and csm2-KRRR psy3-KRK cells. Error bars indicate 95% confidence intervals

Csm2 is recruited to chromatin when abasic sites accumulate. a Csm2-Psy3 DNA binding is critical for survival when abasic sites accumulate. Fivefold serial dilutions of the indicated yeast strains on rich medium (YPD) or rich medium containing 0.002% MMS. Abasic sites accumulate by combined disruption of the AP endonucleases (APN1, APN2) and AP lyases (NTG1, NTG2) in the presence of MMS. Csm2-Psy3 double DNA-binding mutant (csm2-KRRR psy3-KRK) exhibits similar MMS sensitivity to csm2∆ cells when abasic sites accumulate. b Csm2 is enriched at the chromatin when abasic sites accumulate in a DNA-binding-dependent manner. Csm2-6HA-expressing cells were synchronized in G1 with alpha factor and released into YPD medium or YPD medium containing 0.02% MMS for 1 h before cellular fractionation. Csm2 protein levels from whole-cell extract (W), supernatant (S), and chromatin (C) fractions from the indicated strains were determined by western blot using HA antibody. Kar2 and histone H2B were used as fractionation controls (S and C, respectively). The results from three to five experiments were plotted with standard deviations, as fold enrichment relative to the untreated WT (Csm2-6HA). The p value between Csm2-6HA apn1∆ apn2∆ ntg1∆ ntg2∆ and WT (treated and untreated) or csm2-KRRR-6HA apn1∆ apn2∆ ntg1∆ ntg2∆ was calculated using an unpaired two-tailed Student’s t-test between experimental samples and in each case was p ≤ 0.05. c Csm2 chromatin association increases in an MMS dose-dependent manner. Same as b except that Csm2-6HA or Csm2-6HA apn1∆ apn2∆ ntg1∆ ntg2∆ were treated with 0%, 0.01%, 0.02%, or 0.03% MMS and results were quantified as described in b

The Shu complex promotes bypass of APOBEC3B-induced lesions. a APOBEC3B-induced mutation rates on the lagging strand of a replication fork measured using a CAN1 reporter integrated 16 kb from ARS216 on chromosome II. Expression of the cytidine deaminase APOBEC3B induces primarily lagging strand mutations caused by dU templated replication (G to A transition). The uracil glycosylase Ung1 removes U resulting in abasic site formation (AP) in the lagging strand, which can be bypassed by TLS (G to A transition, G to C transversion). b CSM2 and PSY3 are in the same pathway as UNG1 and their deletion results in similar mutation rates individually or in combination with each other. Mutation rates of the indicated genotypes were measured in CAN1 reporter strains transformed with either an empty or APOBEC3B-expressing vector. Error bars indicate 95% confidence intervals. c In the absence of CSM2 or PSY3, abasic sites accumulate and TLS predominates resulting in primarily G to A transitions (red) or G to C transversions (green) within the CAN1 locus. APOBEC3B expression in WT, ung1∆, or csm2∆ ung1∆ cells primarily result in G to A transitions. The rate reported represents the proportion of the CanR mutants observed from sequencing multiplied by the mutation rate determined in b. G to T substitutions are indicated in blue. Other mutations consisting of rare substitutions at A:T base pairs, insertions, deletions, and complex events composed of multiple mutations are depicted in purple. d The strand bias of CAN1 mutations from APOBEC3B expression was evaluated by Sanger sequencing. APOBEC3B expression results in a mutation bias in the lagging strand from templated replication of C deaminations. CSM2, PSY3, or UNG1 deleted cells exhibit more G mutations (green) than C mutations (red). Other mutations as defined in b are indicated in purple. Individual mutation rates were calculated as in c. Statistical significance of strand bias in APOBEC3B-expressing strains was determined by a two-tailed G-test with p < 0.05 for all genotypes

Csm2-Psy3 binds an abasic site analog in double-flap DNA. a Schematic of double-flap substrates containing abasic site analogs (tetrahydrofuran; indicated by an arrow and highlighted text) on the 3′ (AP1–4) or 5′ (AP5–8) DNA strand. Note that the complementing oligonucleotide, which is unmutated, is FITC labeled on the ssDNA end of the double-flap substrate. b Csm2-Psy3 binds to double-flap substrates with abasic site analogs along the 3′ DNA with equal affinity. DNA binding by Csm2-Psy3 was measured by fluorescence anisotropy, with increasing concentrations of the Csm2-Psy3 heterodimer added to 25 nM fluorescein-labeled double-flap substrate (AP1–4). Error bars indicate standard deviations measured in triplicate experiments. Data were fit to a one-site binding model and average dissociation constants (K_d) were calculated. Note the improved binding compared to Fig.  due to increased sample purity (described in detail in Supplementary Methods). c Csm2-Psy3 binds more tightly to a double-flap substrate containing an abasic site analog on the 5′ oligonucleotide at the junction (AP6, highlighted in red). In vitro-binding assays as described in b except with double-flap substrates containing 5′ abasic site analogs (AP5–8). d DNA binding by the Csm2-Psy3-Shu1-Shu2 was measured by fluorescence anisotropy, with increasing concentrations of the indicated proteins added to 5 nM AP6 (highlighted in red) or 2.5 nM double-flap DNA substrate. The data were fit to a quadratic equation

APE1 activity is inhibited by Csm2-Psy3 DNA binding. a Increasing Csm2-Psy3 protein concentrations decrease APE1 activity on a double-flap substrate containing a lagging strand abasic site analog (AP6). Representative gel and bar graph showing dose-dependent protection of the double-flap abasic site analog at the junction (AP6, 100 nM) by Csm2-Psy3 (0–5 µM) from APE1 AP endonuclease activity (100 nM). AP6 was incubated with Csm2-Psy3 or buffer for 5 min before APE1 addition. Reactions were stopped after 1min to limit APE1 catalytic cycles. Error bars indicate standard deviations from three experiments. b Csm2-Psy3 protects a double-flap substrate containing a 3′ abasic site analog (AP6) over time from AP endonuclease cleavage. Representative gel and bar graph showing Csm2-Psy3 (5 µM) protection of AP6 (100 nM) from APE1 endonuclease activity (100 nM) over an extended time course (0–30 min). More than 60% of AP6 bound by Csm2-Psy3 persists even after 30min. Error bars represent standard deviations from three experiments. c Rad51 inhibits APE1 cleavage of AP6 in a concentration-dependent manner. Shown is a representative gel from three experiments. Results from three experiments are quantified and error bars represent standard deviations. d Shu1-Shu2 does not significantly enhance Csm2-Psy3 protection of AP6 against APE1 endonuclease. Representative gels and bar graph showing protection of AP6 (100 nM) against APE1 (100 nM) by Shu1-Shu2-Csm2-Psy3 (0–5 µM) with (bottom gel) or without (top gel) added Rad51 (2 µM). Error bars represent standard deviations from three experiments

Model of Shu-mediated DNA strand-specific damage tolerance. The Shu complex DNA-binding components, the Rad51 paralogs Csm2-Psy3, bind to abasic sites at a double-flap junction to promote Rad51-mediated template switching while preventing AP endonuclease cleavage. MMS-induced DNA damage is primarily repaired by the BER pathway. However, if a replication fork encounters DNA damage such as an abasic site (yellow star), then the fork can stall or collapse. The Shu complex (blue ovals) binds abasic sites on the lagging strand template proximal to the dsDNA fork stem. Shu complex DNA binding (1) promotes Rad51 filament formation (green ovals) and (2) likely prevents AP endonuclease cleavage (orange Pac-man) and DSB formation. Thus, the Shu complex mediates a DNA strand-specific damage tolerance pathway enabling error-free lesion bypass through a template switch using the newly synthesized sister chromatid. This strand-specific lesion bypass pathway allows replication to continue efficiently in an error-free manner and the abasic site to be repaired by BER after the fork progresses