USP37 prevents unscheduled replisome unloading through MCM complex deubiquitination

Abstract

The CMG helicase (CDC45-MCM2-7-GINS) unwinds DNA as a component of eukaryotic replisomes. Replisome (dis)assembly is tightly coordinated with cell cycle progression to ensure genome stability. However, factors that prevent premature CMG unloading and replisome disassembly are poorly described. Since disassembly is catalyzed by ubiquitination, deubiquitinases (DUBs) represent attractive candidates for safeguarding against untimely and deleterious CMG unloading. We combined a targeted loss-of-function screen with quantitative, single-cell analysis to identify human USP37 as a key DUB preventing replisome disassembly. We demonstrate that USP37 maintains active replisomes on S phase chromatin and promotes normal cell cycle progression. Proteomics and biochemical assays revealed USP37 interacts with the CMG complex to deubiquitinate MCM7, antagonizing replisome disassembly. Significantly, USP37 protects normal epithelial cells from oncoprotein-induced replication stress. Our findings reveal USP37 to be critical to the maintenance of replisomes in S phase and suggest USP37-targeting as a potential strategy for treating malignancies with defective DNA replication control.

Fig. 1. A targeted siRNA screen identifies USP37 as the antagonizing DUB for replisome disassembly.

a Model displaying the molecular players involved in replication termination. During replication termination, the CMG replicative helicase (CDC45-MCM2-7-GINS) is poly-ubiquitinated with lysine 48 (K48) ubiquitin linkages, and the replisome is disassembled through p97. A deubiquitinase (DUB) could antagonize the ubiquitination-dependent disassembly to prevent premature replisome disassembly. b Workflow for chromatin flow cytometry assays to study replication and bound MCM. RPE1 cells were treated for 24 h with either p97i or a panel of siRNAs to knock down selected DUBs individually (siDUB). Cells were labeled with EdU (thymidine analog) 30 min prior to harvesting, then soluble proteins were pre-extracted to retain only chromatin-bound proteins such as MCM2 (one of the replisome components). Cells were then fixed and stained for EdU (for active DNA synthesis), MCM2 (as a representative subunit for the MCM2-7 complex), and DAPI (for total DNA content) for flow cytometric analysis. c Chromatin flow cytometry for RPE1 cells treated with 20 nM siControl or 1.25 μM of CB-5083 (p97 inhibitor) for 24 h, and pulsed with EdU for 30 min before harvesting. Cells were stained for bound MCM2, and DAPI (for DNA content). In the late S/G2/M gate, control cells are divided into high ( > 10³) versus low ( < 10³) bound MCM. Representative of two biological replicates. d Histograms of the late S/G2/M-MCM^DNA-positive cells from (C). e RPE1 cells were treated with siControl or siDUB at 20 nM as indicated. Box and whisker plots for EdU intensity per cell in S phase. Box represents 25^th−75^th percentile with line at median. Cells in each sample were randomly down-sampled to 2400 cells per sample. Data is combined from two independent biological replicates. Relative fold-change of the means of EdU intensity from the two replicates was computed: siControl versus siUSP37, unpaired two tailed t test, p = 0.0115. Source data are provided as a Source Data file. f Bound MCM in late S/G2/M from cells treated as in (e). Left: Histograms of normalized counts of the late S/G2/M-MCM^DNA-positive cells. Representative of one biological replicate. Right: Relative percentage of high MCM, late S/G2/M-MCM^DNA-positive cells computed from at least two independent biological replicates; mean with error bars ± SEM. Unpaired two tailed t test for the means of the three replicates for siControl versus siUSP37, p < 0.0001. Source data are provided as a Source Data file.

Fig. 2. USP37 retains active CMG on S phase chromatin.

a Illustration of CMG at active replisomes versus MCM loaded at inactive (dormant) origins that lack CDC45. b Workflow. Cells were treated with siControl or siUSP37; doxycycline was added concurrently with the siRNA treatment to express the siRNA-resistant ^RUSP37^WT, or ^RUSP37^CS, or ^RUSP37^ΔPH. Cells were EdU-labeled and harvested after 24 h and analyzed by flow cytometry for endogenous bound CDC45 and DNA synthesis. c, e, g Immunoblotting for endogenous USP37 or ectopic ^RUSP37^WT, ^RUSP37^CS, or ^RUSP37^ΔPH in RPE1 cells treated with siControl or siUSP37 at 5 nM ± doxycycline at 5, 5, or 2.5 ng/mL, respectively. Representative of > three biological replicates. Asterisk denotes a cross-reacting band. d, f, h Box and whisker plots for chromatin-bound CDC45 per cell in S phase from the same samples in (c, e, and g). Box represents 25^th−75^th percentile with line at median. Cells in each sample were randomly down-sampled to 4500 cells. Each plot is a representative of at least > three biological replicates. Relative fold-change of the means of bound CDC45 intensity from at least three biological replicates was computed. For D: siUSP37 versus siUSP37+ectopic ^RUSP37^WT, unpaired two tailed t test, p = 0.0672. For F: siUSP37 versus siUSP37 + ectopic ^RUSP37^CS, unpaired two tailed t test, p = 0.6038. For H: siUSP37 versus siUSP37 + ectopic ^RUSP37^ΔPH, unpaired two-tailed t test, p = 0.2957. Source data are provided as a Source Data file.

Fig. 3. USP37 interacts with the replisome components.

a Triplicate samples of HEK-293T cells expressing FLAG-tagged USP37 or an empty vector (EV) control were subjected was FLAG immunoprecipitation. Immunoprecipitates were washed, eluted, and subjected to proteomic analysis. Log2 fold change (FC) ratios were calculated using the averaged Log2 LFQ intensities of ^FlagUSP37 IP compared to control IP, and students t test performed for each pairwise comparison, with p-values calculated. Source data are provided as a Source Data file. b Gene Ontology analysis was performed on significantly enriched proteins to reveal the majority of USP37 interactors are involved in DNA replication and cell cycle progression. P-values were calculated by ref. . c All components of the CMG complex were enriched in the FLAG-USP37 IP-MS. The fold enrichment over control, p-value, and number of peptides identified are shown. Data from (a). d HEK-293T cells were transfected for 48 h with FLAG-tagged USP37 or an empty vector as a control. FLAG-USP37 was subjected to FLAG immunoprecipitation and analyzed by immunoblot. The indicated endogenous components of the CMG complex were co-precipitated by USP37. Representative of >three biological replicates. e USP37 schematic. FL corresponds to full-length USP37, Δ corresponds to USP37 lacking the PH domain, and PH corresponds to a USP37 fragment containing the PH domain only. The USP37 catalytic domain contains its active site at C350 as well as an insertion containing three ubiquitin interacting motifs (UIMs). The interaction between USP37 FL, Δ, or PH was assessed as described in (d). USP37 interacts with the CMG complex partially through its PH domain. Representative of > three biological replicates.

Fig. 4. USP37 regulates the CMG complex by deubiquitinating MCM7.

a USP37 was depleted using siRNA for 48 h in RPE1 cells stably expressing a 6xHis-FLAG-tagged ubiquitin construct. Ubiquitinated proteins were pulled down using Ni-NTA, revealing that USP37 siRNA increases endogenous MCM7 ubiquitination, as observed by immunoblotting. Representative of two biological replicates. b MCM7 ubiquitination was analyzed as described in (a), except that cells were treated with 5 µM of the p97i CB-5083 for the last 4 h before harvesting. Inhibition of p97 strongly increases MCM7 ubiquitination, and this is even more pronounced after USP37 depletion. Representative of three biological replicates. c HeLa cells stably expressing CDC45^GFP were depleted of USP37 before being stabilized prior to S phase using thymidine. After release, cells were treated with DMSO or 5 µM of p97i before CMG complexes were isolated using biotinylated anti-GFP Nanobodies (Nb) conjugated to Strep-tacin resin. Inhibition of p97 combined with depletion of USP37 significantly enhanced ubiquitination of endogenous MCM7. Representative of two biological replicates. d Ubiquitinated MCM7 isolated from HEK-293T cells was mixed with 100 nM of recombinant USP37 WT or a catalytically inactive mutant (C350S). The in vitro deubiquitination assay shows that USP37 WT, but not C350S, deubiquitinates Ub-MCM7. Representative of > three biological replicates. e FLAG-tagged USP37 was ectopically expressed for 48 hours and subsequently purified from HEK-293T cells by FLAG immunoprecipitation. USP37 immunoprecipitates were mixed with 1 µM of K11, K48, or K63 tetra-ubiquitin chains, revealing that USP37 cleaves Tetra- and Tri-Ub more efficiently than Di-Ub. Representative of two biological replicates.

Fig. 5. USP37 preserves inactive CMG on S phase chromatin and ensures proper cell cycle progression under stress conditions.

a Expression-corrected CERES correlation scores were downloaded from the DepMap database for genes similar to USP37 knockout. Proteins involved in DNA replication and the DNA damage response are significantly enriched. Proteins are color coded similarly to Fig. 3A (MCMs = blue, CDC45 = purple, polymerases = brown). Insert consists of ~ 750 of the > 17000 proteins analyzed, or ~ 4%. Source data are provided as a Source Data file. b Workflow. Cells were treated with siControl or siUSP37; hydroxyurea (HU) was added concurrently with the siRNA treatment to induce replication stress at 150 μM for 16 h. HU-treated cells were harvested at 16 h or released into fresh media without HU to recover for an additional 8 h, then harvested at 24 h. Cells were analyzed by flow cytometry for endogenous bound γH2AX and DNA content to distinguish G1 from G2 cells. c Bar graph for % γH2AX positive cells, acquired by chromatin flow cytometry from three biological replicates; mean with error bars ± SEM, unpaired two-tailed t test: siControl versus siUSP37 in HU or siControl versus siUSP37 8 h after release from HU; p = 0.1290, 0.6141, respectively. Source data are provided as a Source Data file. d Bar graph for % G2 cells from the same samples in (c). Mean with error bars ± SEM, unpaired two-tailed t test: siControl versus siUSP37 8 hours after release from HU; p = 0.0273. Source data are provided as a Source Data file. e Immunoblotting for USP37 or p21 from the same samples in (c). Representative of three biological replicates. f Box and whisker plots for chromatin-bound CDC45 per cell in S phase from the same samples in (c). Box represents 25^th−75^th percentile with line at median. The aggregate of three biological replicates was randomly down-sampled to 11,000 cells per sample. Relative fold-change of the means of bound CDC45 intensity from the three replicates was computed: siControl versus siUSP37 or siControl + HU versus siUSP37 + HU, unpaired two-tailed t test, p < 0.0001, p = 0.0422, respectively. Active CMG = ongoing forks under normal conditions; stalled CMG = stalled forks under stress conditions. Source data are provided as a Source Data file.

Fig. 6. USP37 ensures replication progression during oncogene-driven stress.

a RPE1 cells engineered for either doxycycline-inducible cyclin E1 or c-MYC were treated to induce expression simultaneously with USP37 depletion to examine effects on DNA replication and total metabolic activity. b Immunoblotting for USP37 or cyclin E1 in RPE1 cells as outlined in (a). siControl or siUSP37 were used at 5 nM. Doxycycline was added simultaneously with the siRNA at 100 ng/mL to overproduce cyclin E1 for 24 h as indicated. Representative of three biological replicates. Asterisk denotes a cross-reacting band. c Box and whisker plots for EdU intensity per cell for the same samples in (b). Box represents 25^th−75^th percentile with line at median. The aggregate of three biological replicates was randomly down-sampled to 5300 cells per sample. Relative fold-change of the means of EdU intensity from the three replicates was computed: siControl versus siUSP37 + cyclin E1 or siUSP37 versus siUSP37 + cyclin E1, unpaired two tailed t test, p < 0.0001, p < 0.0001, respectively. Source data are provided as a Source Data file. d Box and whisker plots for chromatin-bound CDC45 per cell in S phase. Box represents 25^th−75^th percentile with line at median. The aggregate of three biological replicates was randomly down-sampled to 5300 cells per sample. Relative fold-change of the means of bound CDC45 intensity from the three replicates was computed: siControl versus siUSP37 + cyclin E1 or siUSP37 versus siUSP37 + cyclin E1, unpaired two-tailed t test, p = 0.0065, p = 0.0663, respectively. Source data are provided as a Source Data file. e Bar graph for % γH2AX positive cells, acquired by chromatin flow cytometry from three biological replicates; mean with error bars ± SEM, unpaired two-tailed t test: siControl versus siUSP37+ cyclin E1 or siUSP37 versus siUSP37 + cyclin E1; p = 0.0216, p = 0.0398, respectively. Source data are provided as a Source Data file. f Normalized fluorescence intensity of the metabolic reagent resazurin (see “Methods”) was measured for RPE1 Cyclin E1 cells treated with siControl or siUSP37 with or without cyclin E1 overproduction to induce replication stress for five days total. n = 3 replicates, data represent mean ± SEM. p-values were computed by one-way ANOVA. Source data are provided as a Source Data file. g Immunoblotting for USP37, cyclin E1, or p53 in RPE1 cells. siControl (10 nM), siUSP37 (5 nM), or siTP53 (10 nM) were used. Doxycycline was added for 24 h prior to siRNA at 100 ng/mL to overproduce cyclin E1 as indicated. Representative of four replicates. h Normalized fluorescence intensity of the metabolic reagent resazurin was measured for RPE1 Cyclin E1 cells treated with siControl, siUSP37, or siTP53 with cyclin E1 overproduction to induce replication stress for five days total. n = 4 replicates, data represent mean ± SEM. p-values were computed by one-way ANOVA compared to siControl + cyclin E1 overexpression. Source data are provided as a Source Data file.