MCM7 Ubiquitination Regulates CMG Helicase Disassembly in Human

The disassembly of the replisome plays an essential role in maintaining genome stability at the termination of DNA replication. However, the mechanism of replisome disassembly remains unknown in human. In this study, we screened E3 ligases and deubiquitinases (DUBs) for the ubiquitination of minichromosome maintenance protein (MCM) 7 and provided evidence of this process driving CMG helicase disassembly in human tumor cells. SILAC-MS/MS was analyzed to identify ubiquitinated proteins in HeLa cells. The ubiquitination/deubiquitylation assay in vitro and in vivo were detected by Western blot. Thymidine and HU were implied to synchronized cell cycle(cid:0)and detect the role of ubiquitinated MCM7 in cell cycle. Cell fractionation assay was used to detect the function of ubiquitination of MCM7 in chromatin and non-chromatin. Aphidicolin (cid:0) Etoposide (cid:0) ICRF-193 and IR were applied to cause replication fork stalling. MG-132 and NMS-873 were used to inhibit the proteasome degradation and p97 segregase. Flow cytometer and FlowJo ow cytometry software were used to cell cycle analysis. A. cells synchronized by HU (2 mM) for 18 hours. After 2.5 hours of release, HeLa cells were treated with aphidicolin (1 μM), Adriamycin HCl µg/mL) and ICRF-193 (5 µM), and subsequently harvested at the indicated increasing time intervals after block release. The DNA content was measured by propidium iodide staining and ow cytometry. B. Proteasome is required for MCM7 deubiquitylation and CMG disassembly. HeLa cells were co-transfected with plasmids encoding Flag-MCM7, HA-Ub and HA-RNF8, and synchronized by HU (2 mM). After 2.5 h of release, cells were treated with MG-132 (20 µM) and harvested at the indicated time after block release. M2 beads pull down assay of chromatin-bound proteins was performed to detect the polyubiquitylation of MCM7. Chromatin extraction lysed in RIPA buffer was probed with γ-H2AX, MCM2, MCM7, PCNA, and CDC45 antibodies. C. HeLa cells were treated with NMS-873 (5 µM) after 2.5 hours of release and were detected as described in Figure 6B. D. HeLa cells were synchronized and treated as described Figure 6B-6C. Cells were harvested at the indicated increasing time intervals after block release and analyzed for DNA content by propidium iodide staining and ow cytometry.


Introduction
Replication in eukaryotic cells is precisely regulated. All chromosomes need to be duplicated only once in each cell cycle. The loading of the MCM2-7 DNA helicase into chromatin origins occurs during late mitosis and in the G1 phase, and the helicase is recruited by the origin recognition complex (ORC), Cdc6, and Cdt1 to assemble the pre-replicative complex (pre-RC) [1,2] . During the S phase in eukaryotes, MCM2-7 recruits Cdc45 and GINS to form the active replication fork helicase CMG, which unwinds double-stranded DNA [3,4] and functions in initiation and elongation during replication [5] . The termination of replication forks occurs when replication forks converge from neighboring origins in opposite directions. The removal of the replisome from fully duplicated DNA is the last stage of replication fork termination [6] .
In mitosis, TRAIP catalyzes MCM7 ubiquitylation with K6-and K63-linked ubiquitin chains, which promote replisome disassembly to remove any replisome from chromatin before cell division in Xenopus laevis egg extract [11] . When two replisomes converge at an interstrand DNA crosslink (ICL), TRAIP ubiquitylates MCM7 of CMG to activate CMG unloading and entry into the Fanconi anemia (FA) pathway in Xenopus egg extracts [12] . A recent study in mouse embryonic stem cells found that ubiquitin ligases CUL2LRR1 and TRAIP control p97-dependent replisome disassembly during DNA replication termination and mitosis, respectively [13] . However, the mechanism of human chromosomal replication dissociation is incompletely understood.
The E3 ubiquitin ligase RNF8 is a member of the RING nger family, which maintains genome integrity through participation in DNA damage repair. RNF8 functions in histone H2A and H2AX ubiquitination, which mediates the recruitment of 53BP1 and BRCA1 at sites of DNA damage to promote the DNA damage repair pathway [14][15][16][17][18][19][20][21] . RNF8 physically interacts with TPP1 to generate K63-linked polyubiquitin chains that stabilize TPP1 at telomeres, which is essential for protection of telomere end integrity [22] . In the late S/G2 phase, RNF8-mediated K63polyubiquitylation of tankyrase 1 promotes its stability and association with telomeres and then facilitates the resolution of sister telomere cohesion [23] . RNF8 also regulates mitotic exit, and overexpression of RNF8 causes aberrant mitosis and unresolved cytokinesis [24][25][26] .
It was reported that human MCM7 can be polyubiquitylated in HEK293T cells, which is catalyzed by HPV-18E6 protein in vivo and in vitro [27] . INT6 overexpression was found to increase polyubiquitylation of MCM7 on chromatin [28] . We further explored the processes of MCM7 ubiquitination and the role of MCM7 ubiquitination in regulating CMG complex disassembly during replication.

Experimental Procedures
Plasmids, Antibodies, and Cell Culture Full-length MCM7 ampli ed from human cells was subjected to RT-PCR and cloned into the Flag-pcDNA3.1 and myc-pRK5 vectors. Plasmids harboring the K28R and K145R mutations in MCM7 were generated by PCR-mediated mutagenesis. RNF8, deubiquitinating enzymes (DUBs) and shRNA of DUBs were cloned into plasmids as previously described [29][30][31] .

Silac-ms/ms To Identify Ubiquitinated Proteins In Hela Cells
HeLa cells were labeled with "heavy" or "light" isotopic lysine using a SILAC Protein Quantitation Kit (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. Brie y, two cell lines were grown in Dulbecco's modi ed Eagle's medium supplemented with 10% dialyzed fetal bovine serum and with either the "heavy" (H) form of Lys 8 Arg 10 or "light" (L) Lys 0 Arg 0 for eight passages before being used in the assay. Cells, either without or with IR treatment (10 Gy, 2 h), were harvested and lysed in RIPA buffer (40 mm Tris, pH 8.0, 200 mm NaCl, 2 mm EDTA, 1% Nonidet P-40, and 1% SDS) on ice for 20 min. Equal amounts of protein from cells in the no-IR group and IR group were mixed. The proteins were digested with trypsin. To enrich di-GG-ubiquitinated peptides, the tryptic peptides were incubated with anti-di-GG agarose beads (PTM Biolabs Inc., Chicago, IL, USA) at 4°C for 4 h with gentle shaking. The beads were washed four times and the bound peptides were eluted from the beads with 1% tri uoroacetic acid. HPLC/MS/MS analysis was performed. The ratio of H/L peptides was normalized by eliminating the impact of protein level change. The results are shown in Supplementary Table 1.

Cell Synchronization
Cells were synchronized at the G1/S boundary by incubation with thymidine or hydroxyurea (HU). For thymidine blockade, thymidine (2 mM) was added to the cells at 30% con uence, and the cells were incubated at 37°C for 14 h. Then, thymidine was removed by washing the cells with prewarmed PBS, prewarmed (37°C) fresh medium was added, and the cells were incubated for 10 h. Subsequently, a second round of thymidine (2 mM) was added, and the cells were incubated for another 14 h. At this point, the cells were synchronized at the G1/S boundary. Cells were released by washing with prewarmed PBS and incubating cells with prewarmed fresh medium. For HU blockade, HU (2 mM) was added to the cells at 40% con uence, and the cells were incubated at 37°C for 18 h. Then, the HU was removed by washing the cells with prewarmed PBS, and prewarmed fresh medium was added for release. Cells were harvested at the indicated time after block release for Western blotting analysis and ow cytometry.

Chromatin Fractionation Analysis
Cells were harvested and subsequently were lysed with CSK buffer (10 mM HEPES pH 7.9, 300 mM sucrose, 100 mM NaCl, 3 mM MgCl 2 , 0.5% Triton X-100) containing 300 µg/ml RNase A. Then, the lysates were centrifuged at 5 000 rpm for 5 minutes at 4℃. The supernatant was collected as the non-chromatin fraction. The pellet enriched with chromatin-bound proteins was resuspended in CSK buffer, and the above procedures were repeated three times. The supernatant and the pellet were lysed with the indicated buffer, sonicated and centrifuged at 15 000 rpm for 10 minutes at 4℃. The protein concentration was measured by the Bradford assay, and the sample was then used for the Ni-NTA pull down assay or M2 beads pull down assay.
The eluted proteins were analyzed by immunoblotting.

In Vitro Ubiquitination Assay
An in vitro ubiquitination assay was performed with 20 ng of recombinant human ubiquitin-activating enzyme (UBE1; Boston Biochem), 50 ng of recombinant human UbcH5c/UBE2D3 (Boston Biochem), 2 µg of recombinant human ubiquitin (Boston Biochem), puri ed E3 ubiquitin ligase and substrate in ubiquitylation buffer (40 mM Tris-HCl pH 7.4, 5 mM MgCl 2 , 2 mM ATP, 2 mM DTT). Reactions were carried out in a nal volume of 30 µl for 1 h at 37°C and terminated by the addition of 4× SDS sample loading buffer. The reaction products were analyzed by immunoblotting.

In Vitro Deubiquitylation Assay
An in vitro deubiquitylation assay was performed by DUBs and ubiquitin-MCM7 puri ed from HEK293T cells. Reactions were carried out in deubiquitylation buffer (50 mM Tris-HCl pH 8.0, 50 mM NaCl, 1 mM EDTA, 10 mM DTT, and 5% glycerol) for 1 hour at 37°C.

Flow Cytometry
Cells were digested with trypsin, resuspended in PBS, and xed in ethanol (75%) at -20℃ overnight. Then, the cells were washed three times with PBS, and RNase A (100 ng/µl) was added for 15 min. The nuclei were stained with propidium iodide for 10 min. The DNA content was determined on a ow cytometer and analyzed by FlowJo ow cytometry software.

MCM7 is ubiquitylated by RNF8 in vivo and in vitro
To detect protein ubiquitination in the ionizing radiation (IR) induced DNA damage response pathway, we applied stable isotope labeling with amino acids in cell culture (SILAC) and mass spectrometry (MS) analysis in HeLa cells, which showed that the ubiquitination of MCM7 was obviously decreased, typically at Lys 145 (Supplementary Table 1). Due to E3 ligase RNF8 playing an essential role in DSB repair, we hypothesized that whether RNF8 can catalyze MCM7 ubiquitination. To test this hypothesis, H1299 cells were transiently transfected with plasmids expressing His-ubiquitin (Ub), myc-MCM7, and HA-RNF8. Western blotting analysis showed that polyubiquitin-conjugated MCM7 was bound to Ni-NTA resin ( Figure 1A). Additionally, we observed the same result in HEK293T cells ( Figure 1B). To con rm that RNF8 ubiquitylates MCM7, we performed a FLAG-M2 agarose immunoprecipitation assay. HEK293T cells were transiently transfected with plasmids expressing Flag-MCM7, HA-Ub, and RNF8, which showed that MCM7 was ubiquitinated by antibodies against Flag and Ub ( Figure 1C). To further verify that RNF8 ubiquitylates MCM7 in vitro, we puri ed Flag-RNF8 from HEK293T cells as the E3 ligase, pET-Flag-MCM7 from the Escherichia coli strain Rosetta ( Figure 1D) and Flag-MCM7 from HEK293T cells ( Figure 1E) as the substrate. In vitro ubiquitination assay showed that RNF8 catalyzed the polyubiquitylation of MCM7 directly ( Figure 1D-1E), but not the RING-inactive mutant RNF8 C403S (Figure1F). According to the HPLC-MS/MS results, the ubiquitination of MCM7 occurred mainly at Lys 28 and Lys 145. Next, we generated plasmids harboring site-speci c mutations in MCM7, namely, MCM7-K28R, MCM7-K145R and MCM7-K28R+K145R. The results showed that RNF8 catalyzes the polyubiquitylation of MCM7 speci cally on Lys 145 in vivo ( Figure 1G) and in vitro ( Figure 2A). Correspondingly, knockdown of RNF8 suppressed the polyubiquitylation of MCM7 in U2OS cells ( Figure 2B). Together, we identi ed that the E3 ligase RNF8 contributes for the polyubiquitylation of MCM7. Interesting, we also detected that MCM7 is highly ubiquitinated in the absence of RNF8 overexpression, which may be catalyzed by endogenous RNF8 or other E3s.

RNF8 dependent polyubiquitylation of MCM7 occurs on chromatin in the late S phase of the cell cycle
In the late M and G1 phases, the MCM2-7 DNA helicase is loaded at chromatin origins and gradually dissociates from chromatin during DNA replication in the S phase, although the total amount of MCM7 protein in the nucleus remains relatively constant [32] . To identify the distribution of ubiquitinated MCM7, we performed a chromatin fractionation assay with CSK buffer. We found that polyubiquitylation of MCM7 catalyzed by RNF8 occurred mainly in chromatin ( Figure 2C-2E). Given that the association of MCMs with chromatin is cell cycle regulated, we synchronized HeLa cells at the G1-S boundary with double thymidine block, which showed that cells entered S phase during 2-6 h and the G2-M phase during 8-10 h after block release respectively ( Figure 2F). Then, we isolated chromatin extraction from synchronized HeLa cells by double thymidine block and performed M2 beads pull down analysis, which showed that polyubiquitylation of MCM7 increased greatly but transiently after 6 h of release ( Figure 3A), while the total amount of chromatin-bound MCM7 gradually diminished from the G1 phase to late M phase of the cell cycle ( Figure 3A). In synchronized HeLa cells blocked by HU, polyubiquitylation of MCM7 was transiently increased after 5 h of release ( Figure 3B), suggesting that ubiquitylation of MCM7 also occurs in the late S phase. This is slightly different to thymidine double block, which showed polyubiquitylation of MCM7 was transiently increased after 6 h of release. This may be caused by the difference of cell cycle progression after block and release by thymidine and HU. In conclusion, we showed that polyubiquitylation of MCM7 mediated by RNF8 occurs only transiently in vivo and is mainly involved in the nal stages of DNA replication. This is consistent with previous reports in S. cerevisiae and Xenopus [7,8] . Moreover, the polyubiquitylation of MCM7 also occurred at the G1-S boundary and in the late M phase, suggesting an unknown function in the cell cycle, which remains to be further investigated.
3. MCM7 is polyubiquitylated by RNF8 with K63-linked ubiquitin chains Differently linked ubiquitin chains have distinct topologies and cellular functions. In S. cerevisiae (by SCF Dia2 ) and X. laevis (by Cul2 LRR1 ), MCM7 is modi ed with K48-linked ubiquitin chains, which leads to replisome disassembly in the late S phase [7,8] . MCM7 is catalyzed by TRAIP to form K6-and K63-linked ubiquitin chains, which promote the mitotic disassembly pathway in X. laevis [11] . To uncover the novel function of ubiquitinated MCM7 in human tumor cells, we detected the formation of the ubiquitin chains of MCM7 mediated by RNF8. To determine the half-life of MCM7, U2OS cells were treated with cycloheximide (CHX), an inhibitor of eukaryotic protein synthesis, while the total amount of MCM7 remained stable ( Figure 3C). To determine whether proteasome activity in uences the lifespan of MCM7, U2OS cells were treated with MG-132, while we did not detect obvious degradation ( Figure 3D). OTU DUBs (ovarian tumor-associated proteases domain-containing proteins) recognize and hydrolyze speci c ubiquitin chain types and can be used to identify the linkage types on a ubiquitinated substrate [33] . To investigate linkage-speci c polyubiquitin conjugation of MCM7 catalyzed by RNF8, we performed OTU DUBs to cleave the ubiquitin chain of MCM7 in vitro. The results showed that OTUD1 greatly cleaved the polyubiquitylation chains of MCM7 ( Figure 3E). OTUD1 is highly active and speci cally cleaves the K63linked ubiquitin chain [33] , suggesting that MCM7 ubiquitination catalyzed by RNF8 forms the K63-linked ubiquitin chain. Consistent with this notion, the K63-linked ubiquitin chain of MCM7 was enriched by K63-UIM in U2OS cells ( Figure 3F). In HEK293T cells, the overexpression of His-Ub mutant K63R plasmid decreased polyubiquitylation of MCM7mediated by RNF8 ( Figure 3G). In conclusion, MCM7 is polyubiquitylated by RNF8 with K63-linked ubiquitin chains.

Rnf168 And Brcai Promote The Polyubiquitylation Of Mcm7
At double-strand breaks (DSBs) induced by IR, RNF8 is responsible for the initiation of K63-linked ubiquitylation in the DNA damage response, which is subsequently ampli ed by RNF168 for the further recruitment of BRCA1, which is the regulator of the DSB HR response [14][15][16][17][18][19][20][21] . During ICL repair-mediated unloading in Xenopus, BRCA1 acts upstream of MCM7 polyubiquitylation and recruits p97 to promote CMG unloading [34,35] . Therefore, we determined whether RNF168 and BRCA1 promote the ubiquitylation of MCM7. We used HeLa cells for overexpression of plasmids containing myc-MCM7, His-Ub and Flag-RNF168 or Flag-BRCA1 and found that both promoted the ubiquitylation of MCM7 in vivo ( Figure 4A and 4C). However, we did not detect RNF168 mediated polyubiquitylation in vitro ( Figure 4B), suggesting the function of RNF168 may be not directly regulated. In chromatin fractionation, the results showed that polyubiquitylation of MCM7 mediated by RNF168 and BRCA1 occurred on chromatin ( Figure 4D). In synchronized HeLa cells blocked by HU, polyubiquitylation of MCM7 promoted by RNF168 was at the G1/S boundary and slightly increase after 6-7h of block release ( Figure 4E). In addition, polyubiquitylation of MCM7 catalyzed by BRCA1 was transiently increased after 7h of release ( Figure 4F), suggesting that ubiquitylated MCM7 occurs mainly in the late S phase or at the S/G2 boundary. Together, these results show that RNF168 and BRCA1 play a role in promoting polyubiquitylation of MCM7 in chromatin during termination of DNA replication. Consistent with the result for RNF8, the polyubiquitylation of MCM7 mediated by RNF168 and BRCA1 also occurs at the G1-S boundary and in the late M phase. The molecular basis of this regulation remains to be determined.

Inhibition of replication by DNA damage obviously reduces polyubiquitylation of MCM7 mediated by RNF8
DNA damage caused by physical genotoxic agents and chemical agents can induce genome instability, which causes fork replication stalling or collapse, disturbs replication fork progression, and triggers cell cycle arrest. Aphidicolin is an inhibitor of DNA polymerase α that blocks DNA replication in the S phase [36] . IR can induce single-strand breaks (SSBs), DSBs, and base damage [37] . Doxorubicin (Adriamycin) HCl blocks DNA synthesis by inserting itself into DNA and inhibiting DNA topology isomerase II [38] . Actinomycin D inhibits the initiation of DNA replication in mammalian cells [39] . DNA topoisomerase II (Top2) modulates the topological state of double-stranded DNA and allows the completion of DNA replication [40] . Both etoposide and ICRF-193 inhibit the activity of Top2, blocking replication fork termination between replisomes and the accumulation of blocked forks on chromatin in the late S phase [41,42] .
To determine whether the polyubiquitylation of MCM7 occurs in the nal stage of DNA synthesis, we treated HeLa cells with HU to synchronize the cell cycle. After 2.5 h of release, HeLa cells were treated with various physical genotoxic agents and chemical agents, causing early-S phase and late-S phase damage in DNA replication. Because CMG unloading involves polyubiquitylation of CMG's MCM7 subunit, we detected the ubiquitinated MCM7 and CMG disassembly on chromatin after 4-7 h of block release. The results showed that DNA damage signi cantly reduces the polyubiquitylation of MCM7 but increases the amount of MCM2 and MCM7 in chromatin extraction ( Figure 5A-5F), suggesting prolonged association of replicative CMG helicase with chromatin. At the same time, ow cytometric analysis also showed that DNA damage caused S phase blockade ( Figure 6A). In particular, the Top2 inhibitor ICRF-193 arrested cells at the late S phase ( Figure 6A), consistent with the decreased ubiquitination of MCM7 ( Figure 5F), which further veri ed that the ubiquitination of MCM7 occurred at the end of DNA replication. These results suggest that DNA damage can signi cantly reduce the polyubiquitylation of MCM7 in late S phase, suggesting that polyubiquitylation of MCM7 occurs only when DNA replication can be completely duplicated. 6. Proteasome and Cdc48/p97 segregase are required for MCM7 deubiquitylation and CMG disassembly In X. laevis and S. cerevisiae, MCM7 deubiquitylation during replication termination depends on p97/Cdc48/VCP segregase but not proteasomal degradation [7,8] . MCM7 polyubiquitylation acts as the signal for p97-mediated extraction and unloading of the CMG complex from chromatin [7,8,10,13] . To investigate the mechanism of MCM7 deubiquitylation in human tumor cells, we inhibited proteasomal degradation with MG-132 treatment, which showed increased ubiquitination of MCM7 and prolonged CMG association on chromatin in the late S phase ( Figure 6B). Blocking p97-mediated polyubiquitylation segregation by NMS-873 also resulted in increased polyubiquitylation of MCM7 and prolonged association of the helicase components MCM2, CDC45 and PCNA with chromatin ( Figure 6C). Moreover, ow cytometry analysis showed that cells treated with MG-132 or NMS-873 were arrested in the S phase ( Figure 6D). Together, these results suggest that deubiquitylation of MCM7 mediated by p97 and proteasomal degradation could drive the disassembly of the replicative helicase during termination.
Consistent with the results for S. cerevisiae, X. laevis, C. elegans embryos and mouse embryonic stem cells [7,8,13] , we identi ed that polyubiquitylation of MCM7 plays a conserved role in regulating replisome disassembly from chromatin in humans.
7. USP29 and ATXN3 promote MCM7 deubiquitylation and CMG disassembly Substrate release from the p97 complex requires the cooperation of a DUB, which trims polyubiquitin to an oligoubiquitin chain that is then translocated through the pore [43] . To further explore the mechanism of CMG disassembly, we systematically screened the DUBs that are responsible for deubiquitylation of MCM7 in vitro and in vivo. We puri ed ubiquitinated MCM7 and 77 DUBs for the in vitro deubiquitylation assay, which showed that USP7, USP8, USP12, USP15, USP18, USP19, USP20, USP28, USP29, USP30, USP31, USP33, USP36, USP37, DUB3, USP45, OTUD1, ATXN3, JOSD2, and OTUD6A can deubiquitylate MCM7 ( Figure S1A). Therefore, we designed shRNAs targeting these DUBs for knockdown experiments in vivo. Due to the lack of partial DUB plasmids, we also designed shRNAs targeting MINDY1, MINDY2, MINDY3, MINDY4, USP50, USP53 and ZRANB1. Then, HEK293T cells were used to package lentivirus, and U2OS cells were infected. The results showed that USP29, USP50, USP53, MINDY2 and ATXN3 could remove the polyubiquitylation of MCM7 in vivo ( Figure S1B). To further identify the DUBs of MCM7, we knocked down USP29, USP50, USP53, ATXN3 and MINDY2 in U2OS cells with the indicated siRNAs and found that USP29 and ATXN3 were more e cient at catalyzing the deubiquitylation of MCM7 ( Figure 7A).
In U2OS cells synchronized by HU, knockdown of USP29 and ATXN3 increased the polyubiquitylation of MCM7 and prolonged its accumulation on chromatin. Moreover, we detected the inhibition of replisome disassembly ( Figure 7B). Correspondingly, knocking down RNF8, USP29 and ATXN3 perturbed the cell cycle and arrested S phase progression ( Figure 7C-7D). Therefore, USP29 and ATXN3 promote MCM7 deubiquitylation and CMG disassembly ( Figure 7E).

Discussion
CMG helicase disassembly is conserved across evolution from yeasts to humans. Here, we identi ed that MCM7 of the MCMs protein family is controlled by ubiquitination and deubiquitylation. We proposed a model in which at the end of DNA replication, RNF8 is recruited to catalyze the polyubiquitylation of MCM7, which is the signal for the p97 segregase, proteasome, USP29 and ATXN3 to promote the disassembly of terminated replisomes by removing the ubiquitination of MCM7. However, whether the effect of CMG disassembly is due to direct regulation of ubiquitinated MCM7 is unclear.
Given that the RNF8-mediatedpolyubiquitylation of MCM7 occurs mainly at Lys 145, we attempted to construct cell lines with mutation of MCM7 at Lys 145 to directly detect the role of polyubiquitylation of MCM7 in the disassembly of CMG. We applied adenine base editing by xCas9-ABE for A⋅T to G⋅C conversion, which can theoretically mutate the codon AAG of MCM7 at Lys 145 to the codon GAG/GGG/AGG. After clonal cell culture and gene sequencing of the PCR products, we screened nearly 100 cell lines, but no mutant clone was established. One possible explanation is that mutation of MCM7 at Lys 145 severely disrupted cell survival or that the mutation e ciency of xCas9-ABE base editing at Lys 145 was not su cient. Therefore, we lack direct evidence to prove that polyubiquitylation of MCM7 drives the disassembly of CMG.
In the ubiquitin-proteasome system, p97 induces the initial unfolding of ubiquitylated substrates to prepare them for subsequent proteasomal degradation [44] . p97 extracts and segregates multimeric complexes that are well folded or located in cell membranes or chromatin, such as the ring-shaped protein complex Ku70/80 (Ku) and CMG complex [7,8,[45][46][47] . In S. cerevisiae/Xenopus and C. elegans embryos, the p97-Ufd1-Npl4 complex disassembles the entire MCM complex by extracting the ubiquitin-Mcm7 subunit, while the proteasome plays a minor role [47,7,8] . We found that both the proteasome and p97 segregase modulate the deubiquitylation of MCM7 and the CMG disassembly pathway. MCM7 is catalytically attached to a long polyubiquitin chain in human tumor cells, suggesting that p97 may play a role in ubiquitin chain unfolding to facilitate the subsequent deubiquitylation of MCM7 and proteasome degradation, which remains to be veri ed. In our study, we did not directly prove that MCM7 is a substrate of proteases and p97 in the late S phase. Proteases and p97 usually have many substrates. Therefore, the process of proteases and p97 regulating the deubiquitinating of MCM7 is not clear. It was shown that the inhibition of proteasome and p97 also led to increased total levels of RNF8 and ubiquitinated RNF8, which may cause the increase of ubiquitinated MCM7 [48] . Therefore, it will be important in subsequent studies to characterize the coordination of p97 segregase, proteasome and DUB in promoting MCM7 deubiquitylation and CMG helicase disassembly.
We identi ed that RNF8, RNF168 and BRCA1 are required for the polyubiquitylation of MCM7. The difference of catalyzation process mediated by RNF8, RNF168 and BRCA1 remains to be determined. These E3s are necessary for the DSB-induced ubiquitination cascade. Moreover, TRAIP ubiquitin ligasemediated ubiquitin-MCM7 promotes the mitotic pathway of replisome disassembly and certain forms of DNA damage, such as ICL [11,12] . Therefore, we hypothesize that replication termination may utilize a similar mechanism to displace the CMG complex from chromatin, such as DNA damage repair. In our study, we found that the polyubiquitylation of MCM7 mediated by RNF8 and BRCA1 also occurred in the late M phase ( Figure 3A and Figure 4F), indicating that human tumor cells may have a mitotic pathway for CMG helicase disassembly, analogous to the process of mitotic CMG disassembly that was originally observed in C. elegans embryos and X. laevis egg extracts [10,11] .However, the polyubiquitinated MCM7 at the G1-S boundary ( Figure 3A and 4F) have not been reported, which remains to be further investigated. In mouse embryonic stem cells, study found that ubiquitin ligases CUL2LRR1 and TRAIP control p97dependent replisome disassembly during DNA replication termination and mitosis, respectively [13] , which may also regulate the ubiquitination of MCM7 in Human, and is worth for further investigation.
In conclusion, in the late S phage of cell cycle, RNF8 catalyzes the poly-ubiquitination of MCM7, and then initiates the disassembly of CMG helicase from chromatin, which is mediated by p97, proteasome, USP29 and ATXN3. Altogether, our founding provides insight into the mechanism of replisome disassembly during DNA replication termination in human. We reveal the novel function of the poly-ubiquitylation of MCM7, which is a regulatory signal to control CMG complex unloading at replication termination sites. Availability of data and material: All data and material are available on reasonable request. Competing interests: The authors declare that they have no con ict of interest.  H1299 cells were transfected with plasmids encoding His-Ub, myc-MCM7 and HA-RNF8 at a ratio of 1:1:2, and were subject to Ni-NTA pull down assay. Western blotting assays were performed with MCM7 antibody. The cell lysate extracted by RIPA buffer was probed with RNF8, myc and Actin antibodies. B. HEK293T cells were transfected and analyzed as described in Figure 1A. C. HEK293T cells were transfected with plasmids encoding HA-Ub, Flag-MCM7 and RNF8 at a ratio of 1:1:2. After 24-48 hours of transfection, cells were harvested and lysed. Flag tagged proteins were puri ed with the M2 beads pull down assay from the lysates. Western blotting assays were performed with MCM7 and Ub antibodies.

Abbreviations
The cell lysate was probed via Western blotting using anti-RNF8 and anti-Flag antibodies. D. RNF8 catalyzed MCM7 ubiquitination in vitro. Ubiquitylation assays were performed at 37°C for 1 h in the presence of UBE1, E2 (UbcH5c), Ub, Flag-RNF8, and pET-Flag-MCM7 puri ed from the E. coli strain Rosetta. Reaction mixtures were analyzed by Western blotting with MCM7, Ub and RNF8 antibodies. E.
Flag-MCM7 puri ed from HEK293T cells was used as the substrate in vitro ubiquitination assay, and analyzed with MCM7 antibody. F. Reactions were performed as described in Figure1D  ubiquitylation. U2OS cells were transfected with His-Ub and a control or RNF8 siRNA. The proteins were puri ed and enriched by Ni-NTA agarose and probed with antibodies against MCM7. C. RNF8 dependent polyubiquitylation of MCM7 occurs on chromatin. HeLa cells were co-transfected with plasmids encoding Flag-MCM7, RNF8 and HA-Ub and lysed in CSK buffer. The Flag-tagged proteins of the non-chromatin and chromatin fractions were enriched by M2 beads pull down assay, and detected with Ub and MCM7 antibodies. D. HeLa cells were co-transfected with plasmids encoding myc-MCM7, HA-RNF8 and His-Ub, and fractioned in CSK buffer. The His-tagged proteins of the non-chromatin and chromatin fractions were enriched by Ni-NTA pull down assay, and detected with MCM7 and myc antibodies. E. HEK293T cells were co-transfected with myc-MCM7, HA-RNF8 and His-Ub plasmids and lysed in CSK buffer. Samples were puri ed and blotted as shown in Figure 2D. F. HeLa cells were synchronized by double thymidine block and harvested at the indicated increasing time intervals after block release. The DNA content was measured by propidium iodide staining and ow cytometry.  Reaction mixtures were analyzed by western blot with MCM7 and Ub antibody. C. BRCA1 increased the polyubiquitylation of MCM7.HeLa cells were transfected with plasmids encoding myc-MCM7, His-Ub and Flag-BRCA1. Samples were puri ed and blotted as shown in Figure 4A. D. The polyubiquitylation of MCM7 promoted by RNF168 and BRCA1 occurred on chromatin. HeLa cells were transfected with Flag-MCM7, His-Ub, RNF8, RNF168 and BRCA1 and lysed in CSK buffer. The non-chromatin and chromatin fractions were extracted and analyzed by a Ni-NTA pull down assay and probed with MCM7 antibody. E.
The polyubiquitylation of MCM7 mediated by RNF168 was cell cycle related. HeLa cells were cotransfected with plasmids encoding myc-MCM7, Flag-RNF168 and His-Ub, and were treated with HU for synchronization. Ni-NTA pull down analysis of chromatin-bound proteins was performed and probed with MCM7 antibody. F. The polyubiquitylation of MCM7 mediated by BRCA1 was cell cycle related. HeLa cells were transfected with myc-MCM7, Flag-BRCA1 and His-Ub, synchronized and blotted as described in Figure 4E.  Proteasome and Cdc48/p97 segregase are required for MCM7 deubiquitylation and CMG disassembly. A. HeLa cells were synchronized by HU (2 mM) for 18 hours. After 2.5 hours of release, HeLa cells were treated with aphidicolin (1 μM), Adriamycin HCl (0.5 µg/mL) and ICRF-193 (5 µM), and subsequently harvested at the indicated increasing time intervals after block release. The DNA content was measured by propidium iodide staining and ow cytometry. B. Proteasome is required for MCM7 deubiquitylation and CMG disassembly. HeLa cells were co-transfected with plasmids encoding Flag-MCM7, HA-Ub and HA-RNF8, and synchronized by HU (2 mM). After 2.5 h of release, cells were treated with MG-132 (20 µM) and harvested at the indicated time after block release. M2 beads pull down assay of chromatin-bound proteins was performed to detect the polyubiquitylation of MCM7. Chromatin extraction lysed in RIPA buffer was probed with γ-H2AX, MCM2, MCM7, PCNA, and CDC45 antibodies. C. HeLa cells were treated with NMS-873 (5 µM) after 2.5 hours of release and were detected as described in Figure 6B. D. HeLa cells were synchronized and treated as described in Figure 6B-6C. Cells were harvested at the indicated increasing time intervals after block release and analyzed for DNA content by propidium iodide staining and ow cytometry.