Deep structure-function analysis of the endonuclease Mus81 with dominant mutational scanning

Abstract

Protein structure-function relationships are critical for understanding molecular mechanisms and the impacts of genetic variation. Mutational scanning approaches can deliver scalable analysis, usually through the study of loss-of-function variants. Rarer dominant negative and gain-of-function variants can be more information rich, as they retain a stable proteoform and can be used to dissect molecular function while retaining biological context. Dominant variant proteoforms can still engage substrates and interact with binding partners. Here, we probe the structure-function relationships of the Mus81 endonuclease by ectopic expression of deep mutational scanning libraries to find amino acid variants that confer dominant sensitivity to genotoxic stress and dominant synthetic lethality. Screening more than 2,200 MUS81 variants at 100 positions identified 13 amino acids that can be altered to elicit a dominant phenotype. The dominant phenotype of these variants required the presence of the obligate Mus81 binding protein, Mms4. The dominant variants affect amino acids in a contiguous surface on Mus81 and fall into two distinct classes: residues that bind the catalytic magnesium atoms and residues that form the hydrophobic wedge. Most of the variant amino acids were conserved across species and cognate variants expressed in human cell lines resulted in dominant sensitivity to replication stress and synthetic growth defects in cells lacking BLM helicase. The dominant variants in both yeast and human MUS81 resulted in phenotypes distinct from a MUS81 knockout. These data demonstrate the utility of dominant genetics using ectopic expression of amino acid site saturation variant libraries to link function to protein structure providing insight into molecular mechanisms.

Keywords: deep mutational scanning; dominant genetics; genetic interactions; structure function analysis; variant analysis.

Fig. 1.

Site-saturated variant library screening results of the MUS81 ERCC4 domain. (A) Scheme for constructing and screening a 100 amino acid site saturated heteroallelic collection for dominant sensitivity to MMS. (B) Dominant variants from the dominant MMS sensitivity screen are plotted for each amino acid position. Specific amino acid variants are represented by color. Spot assays of representative E358, K411, or S419 dominant MUS81 variants retransformed into a (C) MUS81⁺ background or D) mus81Δ background (for K411 mutants only). Mutants were assayed for their sensitivity to low (0.005%) and high (0.015%) MMS concentrations.

Fig. 2.

Dominant MUS81 ERCC4 residues are highly conserved across species. (A) Aligned MUS81 ERCC4 domain from four species, S. cerevisiae, Drosophila melanogaster, Arabidopsis thaliana, and Homo sapiens. Red arrows denote variant sites from the dominant screen. Green arrows denote variant sites from the failure-to-complement screen. (B) Human crystal structure of Mus81 (Blue & Yellow) in complex with EME1 (Pink) and a 5′ flap DNA substrate (PDB ID: 4P0P). The Mus81 SSVL region is denoted in yellow with dominant residues in red. INSET) Ribbon model structure of the MUS81 hydrophobic wedge with dominant variant residues labeled (red). Corresponding yeast residues in brackets. (C) Heatmaps of the fitness defects exhibited by each variant isolated at the indicated position relative to the empty vector control in a MUS81+ background- green no effect, yellow sensitive to 0.015% MMS, red sensitive to 0.005% MMS (Upper), and in a mus81Δ background – Green is equivalent to WT, yellow is equivalent to mus81Δ, red is more sensitive than mus81Δ (Lower).

Fig. 3.

The Mus81 heterodimer is necessary for dominant effects. A) BY4741 and mus81Δ cells were transformed with either an empty vector or a vector containing the indicated MUS81 allele and grown in selective media to log phase and diluted to OD₆₀₀ = 0.1 in SC ± MMS or CPT. Shown are area-under-the-curve (AUC) graphs for the average growth curves of three 16-h technical replicates for the indicated strains. Fitness scores for MUS81⁺ strains are calculated relative to the empty vector control, and fitness scores for the mus81Δ strains were calculated relative to wild-type MUS81 control. (B) BY4741 strains with the indicated MUS81 allele integrated at the endogenous MUS81 locus were grown to log phase and diluted back to OD₆₀₀ = 0.1 in SC ± MMS or CPT. Shown are area-under-the curve (AUC) graphs for the average growth curves of 3 16-h technical replicates for the indicated strains. All fitness scores are calculated relative to the wild-type MUS81 control. (C) Wild-type (BY4741) and mms4Δ strains were transformed with the URA+ constitutive expression vector pAG416-GPD containing the indicated insert and spot assayed at 0.005% of methyl methanesulfonate (MMS) or 7.2 μM camptothecin (CPT). Strains were spotted at 10-fold dilutions and plated on synthetic complete media lacking uracil (SC–URA) ± MMS or CPT for 3 d at 30 °C before being imaged. (D) BY4741 cells were cotransformed with pAG415-GPD::MMS4 and pAG416-GPD containing the indicated MUS81 allele to co-overexpress both subunits and transformants were selected for on SC media lacking uracil and leucine (SC–URA–LEU). 18-h area under the curve (AUC) graphs for strains grown in SC–URA–LEU containing the no DNA damaging agent (DDA), MMS, or CPT relative to the pAG416-GPD vector control.

Fig. 4.

Dominant genetic interactions between MUS81 variants and SRS2 and SGS1. (A) Wild-type (BY4741, SRS2+) and srs2Δ strains transformed with the URA+ constitutive expression vector pAG416-GPD containing the indicated insert were spot assayed at various concentrations of DNA damaging agents (DDA). Strains were spotted at 10-fold dilutions and plated on SC media lacking uracil (SC–URA) containing no DDA or the indicated concentration of MMS and CPT and incubated for 3 d at 30 °C before being imaged. (B) AUC graph for the average growth curves of three 18-h technical replicates for BY4741 and srs2Δ cells containing the pAG416-GPD expression vector with the following inserts: empty (EV), wild-type MUS81 (WT), mus81-DD (DD), and mus81-K411A (K411A). All fitness scores are calculated relative to the wild-type background with an EV control under the same conditions. MMS = methyl methanesulfonate, CPT = camptothecin. (C) Wild-type (BY4741) and strains with the indicated genotypes for SRS2, and RAD51 were transformed with the URA+ constitutive expression vector pAG416-GPD containing the indicated insert and spot assayed at various concentrations of methyl methanesulfonate (MMS). Strains were spotted at 10-fold dilutions and plated on synthetic complete media lacking uracil (SC –URA) containing the indicated concentrations of MMS and incubated for 3 d at 30 °C before being imaged. (D) sgs1Δ cells transformed with the URA+ constitutive expression vector pAG416-GPD containing no insert (empty), wild-type MUS81, or mus81-K411A were plated on SC media lacking uracil (SC–URA) and grown for 3 d at 30 °C. (E) Wild-type (BY4741), sgs1Δ, and top3Δ strains were engineered to express wild-type MUS81, mus81-D414A-D415A (mus81-DD), or mus81-K411A from the galactose-inducible GAL4 promoter at the ura3Δ0 genomic locus. Cultures were spotted in 10-fold serial dilutions on SC media with either dextrose or galactose and incubated for 3 d at 30 °C.

Fig. 5.

The dominant mus81-K411A mutant exhibits distinct cellular phenotypes compared to mus81Δ. (A and B) DNA content and FACS profiles for exponentially growing cells. (A) Wild-type (BY4741) and mus81Δ strains were grown to log phase in the absence of DNA damaging agents (DDA)s MMS and CPT and sampled for flow cytometry and microscopy. Exponentially growing cultures were then split in two, and the indicated DDA was added as indicated to each. Cultures were incubated overnight in the DDA-containing media. For cultures in log phase, a sample was taken for flow cytometry and microscopy. For strains that had reached saturation, their cultures were restarted in the same media and grown to log phase (~4 h) before sampling for flow cytometry and microscopy. (B) Same as A but for haploid cells containing the indicated MUS81 allele at the endogenous MUS81 locus. (C) Representative DIC/DAPI microscopy images of cells from (A and B). (D) Representative DIC/DAPI microscopy images of cells from B). (Scale bar, 5 µm.) (E) TOP2-GFP bridges were monitored in exponentially growing live cells 3 h after 0.005% MMS treatment. Top2-GFP cells carrying Mus81 variants were treated with 0.005% MMS for 3 h to induce replication stress. A quantification of total bridge count is shown for both untreated and MMS treated cells. A representative image for bridges is shown on the right. All error bars represent means ± SEM, n = 3 biologically independent replicates, >75 cells per replicate, >300 cells per data point. ***P = 0.0003, **P < 0.002, ns, P > 0.1. Fisher’s test.

Fig. 6.

Human cells overexpressing MUS81 mutations are sensitive to DNA damaging agents. (A) MUS81 levels as determined by Western blot of HT29 lines made using pLX302 (Vector, WT, DD, E277G, D307G, K335A) or pLentiCRISPR v2 (sgLUC, sgMUS81). (B) Sensitivity of HT29 cell lines shown in A to MMS and CPT. Cell growth was determined by clonogenic assay. The figure legend applies to both graphs. (C) Clonogenic growth of MUS81-overexpressing mutations and MUS81 CRISPR-Cas9 population knockout in HCT116 WT and BLM homozygous knockout cell lines. *P < 0.05, **P < 0.005 as determined by Welch’s unpaired t test. (D) Propidium iodide-stained cell cycle profiles of select flow cytometry samples treated with either DMSO or CPT. (E) Percentage of cells staining beyond the G2/M peak in cell cycle analysis of HT29 MUS81 overexpressing and CRISPR-Cas9 cell lines. (F) Percentage of multipolar anaphases (representative images LEFT) in MUS81 overexpressing and CRISPR-Cas9 knockout lines. (G) Examples of anaphases stained by immunofluorescence with ERCC6L (Left) and DAPI (Middle). The number of DAPI-only, or DAPI and ERCC6L bridges per anaphase of HT29 MUS81 overexpressing and CRISPR-Cas9 cell lines. Flow cytometry and immunofluorescent results are the average of three independent experiments and were analyzed using a two-way ANOVA plus Tukey. ****P < 0.0001 compared to vector or sgLUC control; +P < 0.05, ++P < 0.005 compared to K335A; #P < 0.05, ##P < 0.005, ###P < 0.0005 compared to sgMUS81.