USP37 prevents premature disassembly of stressed replisomes by TRAIP

Abstract

The E3 ubiquitin ligase TRAIP associates with the replisome and helps this molecular machine deal with replication stress. Thus, TRAIP promotes DNA inter-strand crosslink repair by triggering the disassembly of CDC45-MCM2-7-GINS (CMG) helicases that have converged on these lesions. However, disassembly of single CMGs that have stalled temporarily would be deleterious, suggesting that TRAIP must be carefully regulated. Here, we demonstrate that human cells lacking the de-ubiquitylating enzyme USP37 are hypersensitive to topoisomerase poisons and other replication stress-inducing agents. We further show that TRAIP loss rescues the hypersensitivity of USP37 knockout cells to topoisomerase inhibitors. In Xenopus egg extracts depleted of USP37, TRAIP promotes premature CMG ubiquitylation and disassembly when converging replisomes stall. Finally, guided by AlphaFold-Multimer, we discovered that binding to CDC45 mediates USP37's response to topological stress. In conclusion, we propose that USP37 protects genome stability by preventing TRAIP-dependent CMG unloading when replication stress impedes timely termination.

Extended Data Fig. 1:. Model for TRAIP function in S and M phases.

a, Using its trans-ubiquitylation mode in S phase, TRAIP can ubiquitylate CMGs that have converged on an ICL (i) or DPCs and other obstacles encountered by the replisome (ii), but not the replisome it travels with (which we call cis ubiquitylation) (iii). b, In mitosis, TRAIP undergoes a conformational change that allows it to cis ubiquitylate the CMG it travels with.

Extended Data Fig. 2:. USP37 knock-out hypersensitises to topoisomerase inhibitors and replication stress inducing agents.

a, Schematic of the CRISPR screen in U2OS cells. b-c, Western blot validation of USP37 KO in RPE-1 TP53 KO cells (b) and U2OS cells (c). d-h, Clonogenic survival assays of WT and USP37 KO U2OS cells upon treatment with (d) camptothecin, (e) etoposide, (f) hydroxyurea, (g) aphidicolin or (h) talazoparib. n ≥2 independent experiments. Bars represent means ± SEM. i, Western blot validation of mCherry-USP37^WT, mCherry-USP37^C350A (catalytically inactive), or mCherry-USP37^8A expression in USP37 knockout cells.

Extended Data Fig. 3:. Validation of TRAIP KO in TP53 KO and TP53/USP37 KO RPE-1 cells.

a, Scheme of the tracking indels by decomposition (TIDE)-based cell competition assay shown in Fig. 2b. Scheme generated with BioRender. b, TIDE-validation of TRAIP KO in TP53 KO and TP53/USP37 KO RPE-1 cells.

Extended Data Fig. 4:. TRAIP loss improves the viability of USP37 KO cells upon treatment with topoisomerase inhibitors but not aphidicolin.

a-c, Clonogenic survival assays as in Fig. 2c–e of CTRL or a different USP37 KO clone transduced with a control sgRNA (LacZ) or with a sgRNA targeting TRAIP upon treatment with camptothecin (a), ICRF-193 (b) or aphidicolin (c). The CTRL data are the same as in Fig. 2c–e. n=3 independent experiments. Bars represent means ± SEM. Half plot points in c indicate zero percent viability. d, RT-qPCR experiment to monitor TRAIP knock-down efficiency. n=2 independent experiments. Bars represent means.

Extended Data Fig. 5:. CMG ubiquitylation by TRAIP during topological stress requires trapping of TOP2α on DNA.

a, Plasmid DNA was replicated in the presence or absence of 200 μM ICRF-193 in extracts containing [α−³²P]dATP. Replication intermediates were then separated on a native agarose gel and visualized by autoradiography. At early time points, replication of plasmid DNA generates a “theta” structure (θ), a late replication intermediate formed when replisomes converge, that is subsequently converted to fully replicated closed and open circular products (“CC” and “OC”, respectively). ICRF-193 delays conversion of theta structures to CC and OC, indicative of impaired replisome convergence, and causes accumulation of a smear that probably represents highly catenated daughter molecules (Ref., see Extended Data Fig. 2a). The images are separate agarose gels, which were processed and imaged in parallel. b, Western blot analysis of mock, USP37, and TOP2α depletions. Related to Extended data Fig. 5c. c, Plasmid DNA was incubated in the indicated egg extracts in the presence or absence of 200 μM ICRF-193. At the specified times, chromatin was recovered and blotted for the indicated proteins. d, Plasmid DNA was incubated in the indicated egg extracts in the presence or absence of 200 μM ICRF-193 and 50 ng/μl aphidicolin. At specified times, chromatin was recovered and immunoblotted for the indicated proteins. Aphidicolin inhibits DNA synthesis and thus prevents formation of pre-catenanes while still allowing DNA unwinding and accumulation of supercoils^,,, which are presumably located ahead of the fork. As expected, aphidicolin increased RPA binding and decreased TOP2α recruitment to chromatin (lane 7 vs 1). However, in the presence of aphidicolin, ICRF-193 still caused accumulation of TOP2α on DNA, albeit to a lesser extent (lanes 10–12 vs 4–6) suggesting that TOP2α can be trapped on supercoils ahead of the replisome. e, Western blot analysis of mock and USP37 depletions. f, Western blot analysis of mock and USP37 depletions supplemented with recombinant USP37 expressed in wheat germ extract. Abbreviations as in Fig. 3b. g, Western blot analysis of mock, USP37, and TRAIP depletions supplemented with recombinant TRAIP expressed in wheat germ extract. Abbreviations as in Fig. 3c. Blue arrows, endogenous TRAIP.

Extended data Fig. 6:. TRAIP promotes premature disassembly of CMGs stalled at distances of up to 1 kb.

a, Western blot analysis of mock, USP37, and TRAIP depletions. b, Top, schematic of the meDPC substrates used, including the variable distance between distal leading strand meDPCs. The indicated meDPC substrates were incubated in egg extracts in the presence or absence of 200 μM p97-i. At the specified times, chromatin was recovered and blotted for the indicated proteins. USP37 depletion efficiency is shown in Extended data Fig. 6c. c, Western blot analysis of mock and USP37 depletions. d, Western blot analysis of mock, USP37, and TRAIP depletions supplemented with recombinant TRAIP expressed in wheat germ extract. Abbreviations as in Fig. 4b. e, Western blot analysis of mock, USP37, and TRAIP depletions. f, Plasmid DNA was pre-incubated with LacR and then replicated in the presence or absence of RTEL1 in extracts containing [α−³²P]dATP. Replication intermediates were separated on a native agarose gel and visualized by autoradiography. RTEL1 depletion delayes progression of replication forks through the LacR array, as seen by stabilization of the theta structure (θ) and absence of fully replicated closed and open circular products (“CC” and “OC”, respectively). The images are part of the same agarose gel, which was cropped to remove irrelevant information. RTEL1 depletion efficiency is shown in Extended data Fig. 6g. g, Western blot analysis of mock and RTEL1 depletions. h, Western blot analysis of USP37 and RTEL1 depletions. i, Western blot analysis of USP37 and RTEL1 depletions.

Extended data Fig. 7:. USP37’s PH domain is predicted to interact with CDC45.

a, Side-by side comparison of USP37 recovery in sperm chromatin spindown (SCS) and plasmid pulldown (PPD) procedures. Xenopus sperm chromatin or plasmid DNA were incubated in egg extracts supplemented with indicated inhibitors for 15 min, then recovered and blotted for the indicated proteins. p97-i, NMS-873. The images are all part of the same Western blot, which was cropped to remove irrelevant information. b, Western blot analysis of mock-, USP37-, SMC3-, and USP37/SMC3-depletions. Related to Extended data Fig. 7c. c, The meDPCs substrate (pDPC4-1033) was incubated in indicated egg extracts. At specified times, chromatin was recovered and immunoblotted for the indicated proteins. USP37 and TRAIP depletion efficiencies are shown in Extended data Fig. 7b. d, Predicted alignment error (PAE) plots generated by the 3 AF-M models for the complex of human USP37 and CDC45. e, AlphaFold-Multimer (AF-M) prediction of human CDC45 and USP37 interaction colored by pLDDT value, a measure of the confidence of local amino acid positioning. For simplicity, only amino acid residues (aa) 1–120 corresponding to the USP37 PH domain are shown. f, Space filling representation of the same model shown in Extended data Fig. 7b colored by chain. g, Close-up views of key Human USP37 and CDC45 residues that were predicted to interact by AF-M. h, Close-up views of key Xenopus USP37 and CDC45 residues predicted to interact. i, Alignment of human (residues 1–120) and frog (residues 1–117) USP37. Black boxes indicate residues that were substituted to alanines in the USP37^8A mutant. Although the residues D85 was not predicted to interact by AF-M v2.3 (panels d and e), they were predicted to interact by AF-M v2.2 (not shown) and were therefore also mutated. “*”, conserved residues; “:”, conservative substitution; “.”, semi-conservative substitution; “ “, non-conservative substitution; “-“, gap. j, A representative in vitro deubiquitylation assay with recombinant Xenopus HA-USP37 variants expressed in wheat germ extract. HA-USP37 was immobilized on anti-HA magnetic beads and incubated with K48-linked tetraubiquitin (Ub₄). At the indicated times, reactions were stopped with 2x Laemmli sample buffer and blotted for the indicated proteins. k, Western blot analysis of mock and USP37 depletions supplemented with recombinant USP37 expressed in wheat germ extract. Abbreviations as in Fig. 5c. l, Western blot analysis of mock- and USP37-depletions supplemented with recombinant USP37 expressed in wheat germ extract.

Extended data Fig. 8:. USP37 interaction with CDC45 is required for its protective function towards topoisomerase inhibitors.

a-b, Clonogenic survival assays of control (CTRL) cells or USP37 KO19 cells complemented with vectors expressing mCherry (EV), mCherry-USP37^WT, mCherry-USP37^C350A (catalytically inactive), or mCherry-USP37^8A (defective for CDC45 interaction) upon treatment with camptothecin (a) or etoposide (b); n=3 independent experiments. Bars represent means ± SEM.

Fig. 1:. USP37 loss protects cells from topoisomerase poisons- and replication stress-associated DNA damage

a, Rank-plot showing gene enrichment scores (normZ) of CRISPR screen hits upon camptothecin treatment in human U2OS cells. Negative scores represent dropouts; that is genes whose loss is predicted to promote drug sensitivity. Blue dots indicate well characterised factors affecting cellular sensitivity to camptothecin; the red dot indicates USP37. b-e, Clonogenic survival assays of control (CTRL) and USP37 knockout (KO; results from two independent clones plotted) RPE-1 TP53 KO cells upon treatment with (b) camptothecin, (c) etoposide, (d) ICRF-193, or (e) aphidicolin. n=3 independent experiments. f-g, Quantification of RPA (f) or γH2AX (g) foci in S-phase cells (EdU positive). Cells were treated with the indicated doses of the drugs for 4h. n ≥ 4 independent experiments. Bars represent median. CPT is camptothecin. h-j, Clonogenic survival assays of control (CTRL) cells or USP37 KO (clone 10) cells complemented with vectors expressing mCherry (EV), mCherry-USP37^WT or mCherry-USP37^C350A (catalytic inactive) upon treatment with camptothecin (h), etoposide (i), or aphidicolin (j); n=3 independent experiments. Bars represent means ± SEM.

Fig. 2:. Genetic screen unveils functional connections between USP37 and TRAIP.

a, Biplot showing gene enrichment scores (normZ) reflecting viability in untreated conditions of WT (y-axis) and USP37 KO (x-axis) U2OS cells. b, U2OS USP37 WT cells (left histogram) and USP37 KO cells (right histogram) were transfected with a CRISPR sgRNA targeting TRAIP and samples were collected 4, 7, 11 and 15 days afterwards and subjected to TRAIP sequence analysis. Percentages of unedited or edited cells at the TRAIP locus at the indicated timepoints are plotted. c-e, Clonogenic survival assays of CTRL or USP37 KO RPE-1 TP53 KO cells transduced with a LacZ control sgRNA or with a sgRNA targeting TRAIP upon treatment with camptothecin (c), ICRF-193 (d), and aphidicolin (e). n=3 independent experiments. Bars represent means ± SEM. Half plot points indicate zero percent viability. f-g, Quantification of RPA (f) or γH2AX (g) foci in S-phase cells (EdU positive). Cells were treated with 20nM camptothecin for 4h. n = 3 independent experiments. Bars represent median.

Fig. 3:. USP37 depletion induces premature CMG unloading by TRAIP in egg extracts treated with ICRF-193.

a, Plasmid DNA was incubated in the indicated egg extracts in the presence or absence of 200 μM ICRF-193. At specified times, chromatin was recovered and immunoblotted for the indicated proteins. USP37 depletion efficiency is shown in Extended Data Fig. 5e. Ub-MCM7, ubiquitylated MCM7; Ub-MCM6, ubiquitylated MCM6. b, Plasmid DNA was incubated in the indicated egg extracts in the presence of 200 μM ICRF-193 and, where indicated, 200 μM NMS-873 (p97-i). Extracts were supplemented with recombinant WT or catalytically inactive (C347S) USP37 expressed in wheat germ extract, or wheat germ extract with empty vector (EV). At specified times, chromatin was recovered and immunoblotted for the indicated proteins. USP37 depletion efficiency and levels of recombinant proteins are shown in Extended Data Fig. 5f. ΔM, ΔMOCK; Ub-MCM4, ubiquitylated MCM4. c, As in b, but where indicated, TRAIP was co-depleted with USP37. Egg extracts were supplemented with recombinant WT TRAIP expressed in wheat germ extract, or wheat germ extract with EV. Depletion efficiencies and levels of recombinant proteins are shown in Extended Data Fig. 5g. d, Dot plot indicating the number of PLA foci between CDC45 and EdU (replicating DNA) in untreated or camptothecin treated cells. Bar represents median. n=3 independent experiments.

Fig. 4:. USP37 depletion induces TRAIP-dependent disassembly of CMGs converged at tandem DPCs, but not at terminated forks or ones separated by a LacR array.

a, Top, a schematic of DNA protein cross-link substrate. meDPCs, methylated DNA protein cross-links. The square bracket indicates the distance between distal leading strand DPCs. Bottom, meDPCs substrate was incubated in indicated egg extracts in the presence or absence of 200 μM p97-i. At specified times, chromatin was recovered and immunoblotted for the indicated proteins. USP37 and TRAIP depletion efficiencies are shown in Extended Data Fig. 6a. b, Same as a, top, but the distance between distal leading strand DPCs was 1033 bp. Bottom, meDPCs substrate was incubated in the indicated egg extracts in the presence or absence of 200 μM p97-i. Extracts were supplemented with recombinant WT TRAIP expressed in wheat germ extract or wheat germ extract with empty vector (EV). Depletion efficiencies and levels of recombinant proteins are shown in Extended Data Fig. 6d. Dashed blue line indicates probable CRL2^Lrr1-dependent ubiquitylation resulting from termination events. Dashed red line indicates TRAIP-dependent ubiquitylation. c, Top, a schematic showing terminated CMGs and inhibited CRL2^Lrr1 by Cul-i. Bottom, plasmid DNA was replicated in the indicated egg extracts in the presence of p97-i (200 μM) and in the presence or absence of Cul-i (400 μM). At 90 min after replication initiation, chromatin was recovered and immunoblotted for the indicated proteins. USP37 and TRAIP depletion efficiencies are shown in Extended Data Fig. 6e. d, Top, a schematic of CMGs converged at the DNA protein cross-link substrate versus the LacR-bound lacO array. Bottom, plasmid DNA containing 32x lacO array was preincubated with LacR and then replicated in the indicated egg extracts. At 30 min after replication initiation, reactions were supplemented with either buffer or Cyclin B/CDK1 (50ng/ul). At the specified times, chromatin was recovered and immunoblotted for the indicated proteins. USP37 and RTEL1 depletion efficiencies are shown in Extended Data Fig. 6h. e, Same as d, but at 40 min after replication initiation, reactions were supplemented with p97-i (200 μM). USP37 and RTEL1 depletion efficiencies are shown in Extended Data Fig. 6i.

Fig. 5:. Predicted CDC45-USP37 interaction surface is important for USP37 function at the replisome.

a, Xenopus sperm chromatin was incubated in egg extracts supplemented with indicated inhibitors for 10 min, then recovered and blotted for the indicated proteins. p97-i, NMS-873; CDK-i, p27Kip; DDK-i, PHA-767491. b, Right, the predicted structure in Extended Data Fig. 7c was aligned to the Cryo-EM structure of human CMG (PDB: 7pfo). Left, the same model rotated by 90°. Orange dashed line indicates a flexible region connecting the PH domain of USP37 and the catalytic domain. c, Xenopus sperm chromatin was incubated in mock- or USP37-depleted egg extracts supplemented with recombinant WT or putative CDC45-binding USP37 mutant (8A) expressed in wheat germ extract or wheat germ extract containing empty vector (−), then recovered and blotted for the indicated proteins. DNA-replication inhibitor, Geminin was added where indicated to monitor non-specific binding. See also Extended Data Fig. 7k for depletion efficiencies and recombinant protein levels. d, Plasmid DNA was incubated in the indicated egg extracts in the presence of 200 μM ICRF-193 and, where indicated, 200 μM NMS-873 (p97-i). Extracts were supplemented with the indicated recombinant USP37 variants expressed in wheat germ extract or wheat germ extract containing empty vector (EV). At specified times, chromatin was recovered and blotted for the indicated proteins. USP37 depletion efficiencies and levels of recombinant proteins are shown in Extended Data Fig. 7l. e, Extracts of HEK293T cells expressing mCherry (empty vector, EV) or mCherry-USP37^WT or mCherry-USP37^8A were subjected to mCherry immunoprecipitation (IP) followed by western blotting for the indicated proteins. This experiment was repeated three times with similar results. f, Dot plot indicating the number of PLA foci between mCherry-tagged E.V, USP37^WT, USP37^C350, or USP37^8A and EdU (replicating DNA) in untreated or camptothecin treated cells. Bar represents median. n=3 independent experiments. g-h, Clonogenic survival assays of control (CTRL) cells or USP37 KO (clone 10) cells complemented with vectors expressing mCherry (EV), mCherry-USP37^WT or mCherry-USP37^8A (defective for CDC45 interaction) upon treatment with camptothecin (g) or etoposide (h). Note that CTRL+EV and USP37 KO+ mCherry-USP37WT samples are the same as in Fig. 1h and i. n=3 independent experiments. Bars represent means ± SEM.

Fig. 6:. Model of USP37’s role in protecting replisomes from premature disassembly

In wild-type cells (WT), USP37 counteracts TRAIP ubiquitylation when CMGs are stalled at sites of DPCs/topological stress. Thus, USP37 activity allows replisomes to eventually complete DNA synthesis and be unloaded by a termination-specific CRL2^Lrr1-dependent pathway. In the absence of USP37, TRAIP hyper-ubiquitylates CMG at sites of topological stress, causing premature CMG disassembly, DNA damage, and cell death. When both USP37 and TRAIP are absent converging CMGs cannot be prematurely ubiquitylated and, thus, may ultimately terminate normally. For simplicity, the region in between two stalled replisomes is shown as plectonemes.