CDT1 acts with the replisome to remodel replication forks

Abstract

Replication forks encounter problems during genome duplication that trigger replication stress. Fork reversal is a key stress tolerance pathway that helps mitigate replication challenges to facilitate DNA synthesis. Here, we report the function of origin licensing factor CDT1 in fork remodeling by replication fork reversal. Characterization of replication dynamics during early stages of origin reactivation revealed that CDT1 blocks fork progression without inducing DNA breaks. Notably, CDT1 mediates replication fork reversal under conditions of re-replication and genotoxic stress exposure, and this function is dependent on the interaction with the CMG helicase. Although proliferating cell nuclear antigen (PCNA) sequesters CDT1 for proteolytic degradation under unperturbed conditions, ATAD5-mediated PCNA unloading at stressed forks releases CDT1 to interact with the CMG helicase and promote fork remodeling. Thus, contrary to the notion that CDT1 function must be inactivated in the S phase, our findings uncover a regulatory mechanism that facilitates fork remodeling function of CDT1 in response to replication stress.

INTRODUCTION

Replisomes are macromolecular complexes that orchestrate DNA replication with chromatin assembly to facilitate accurate and complete duplication of the genome. The foundations for DNA synthesis are established during the origin licensing step in the preceding G₁ phase wherein the origin recognition complex (ORC), coupled with accessory factors CDC6 and CDT1, load the MCM2-7 helicase as double hexamers around double-stranded DNA (dsDNA) (1). During the G₁-S transition, the heightened activities of cyclin-dependent kinase (CDK) and DBF4-dependent kinase (DDK) mediate the recruitment of CDC45 and GINS to each MCM2-7 inactive hexamer to generate active replicative CDC45-GINS-MCM2-7 (CMG) helicase encircling the single-stranded leading strand template (2, 3). The replisomes, comprising the CMG helicase, replicative polymerases, and chromatin assembly factors, execute the faithful duplication of the genome and are eventually disassembled when converging forks meet during replication termination (4).

Following entry into the S phase, relicensing of the genome is prevented to ensure replication occurs once and only once per cell cycle. Origin refiring in the S phase is primarily suppressed by inhibiting the actions of licensing factors CDC6 and CDT1 (5, 6). CDK-mediated phosphorylation of CDC6 regulates its intracellular localization (7–11) and triggers proteolytic degradation by the SCF^CyclinF ubiquitin ligase complex (12). CDT1 is suppressed by three mechanisms. By interacting with proliferating cell nuclear antigen (PCNA) at replication forks, CDT1 is recognized for proteolytic degradation by the CRL4^CDT2 ubiquitin ligase complex (13). CDK-dependent phosphorylation triggers CDT1 destruction by the SCF^SKP2 complex in the late S phase (14). In addition, geminin interacts with CDT1 to prevent minichromosome maintenance (MCM) loading during the S phase (15, 16). Inhibiting the function of CRL complexes or excessive production of licensing factors triggers re-replication, which is a highly genotoxic event (17). Extensive re-replication slows replication forks (18) and results in replication-dependent double-strand breaks (DSBs) that occur due to head-to-tail fork collision events (19) and when re-replication forks approach single-stranded DNA (ssDNA) gaps in template DNA (20). However, little is known about the impact on replication during early stages of re-replication. Furthermore, independently of licensing, CDT1 blocks DNA synthesis (21, 22) and inhibits the CMG helicase in the early S phase (23) but the consequences at the fork in this context remain less understood.

Replication forks are constantly challenged by various DNA obstacles. Replication fork reversal is a stress tolerance mechanism that involves annealing of nascent DNA strands and reannealing of exposed template DNA to generate a four-way junction resembling a Holliday junction (24). Although fork reversal slows fork progression and facilitates repair or bypass of DNA lesions, the nascent regressed arm must be protected to prevent aberrant resection by nucleases (25, 26). Fork reversal is catalyzed by several enzymes including translocases SMARCAL1, ZRANB3, and HLTF, helicase FBH1, and recombinase RAD51 (27).

In this study, we report CDT1 function in facilitating fork remodeling by replication fork reversal. The ability of CDT1 to engage fork reversal is dependent on the interaction with the CMG helicase and this interaction is facilitated by ATAD5-mediated PCNA unloading. Furthermore, stress conditions triggering fork reversal cause PCNA unloading underlining the need for PCNA removal to promote CDT1-mediated fork regression.

RESULTS

Acute re-replication by MLN4924 curtails fork progression without inducing DNA breaks

To examine the impact of acute origin activation on DNA replication, we used MLN4924, which is a small molecule inhibitor of the NEDD8-activating enzyme (28). MLN4294-mediated suppression of SCF^SKP2 and CRL4^CDT2 triggers origin refiring due to CDT1 stabilization (29, 30). Acute MLN4924 exposure for 2 hours resulted in the disappearance of the slower migrating neddylated species of CUL4A comparable to prolonged exposure for 24 hours (Fig. 1A). Acute MLN4924 treatment also decreased neddylation of chromatin-bound cullin paralogs CUL1, CUL2, CUL3, CUL4B, and CUL5 (fig. S1A). Short-term MLN4924 exposure increased CDT1 levels (Fig. 1A), which is predominantly abundant in the nucleus (Fig. 1B). Transient MLN4924 exposure also increased origin activity, as observed by elevated detergent-resistant intensities of pMCM2.S139, a marker of origin activation (31) specifically in 5-ethynyl-2′-deoxyuridine–positive (EdU+) nuclei (Fig. 1C) and quantification of single-labeled tracts by DNA fiber analyses (fig. S1B). Next, we performed DNA fiber analysis and observed diminished fork progression after treating cells with either 2 or 24 hours of MLN4924 (Fig. 1D). Acute MLN4924 treatment also curtailed elongation in hTERT-RPE1 cells and human embryonic kidney (HEK) 293T cells, indicating that this effect is not cell line specific (fig. S1C). MLN4924-induced reduction in fork speed is also independent of p53 status (fig. S1D). Because 2-hour MLN4924 exposure causes a small increase in the number of new origins, we conclude that fork progression is curtailed on both preexisting and newly fired replication forks during the early stages of re-replication by MLN4924.

Fig. 1. Acute CDT1 stabilization decreases fork speed.

(A) Whole-cell lysates were extracted from U2OS cells following the indicated treatments. Representative immunoblot is shown from n = 2 with KU70 as loading control. Arrow represents the neddylated cullin species. hr, hours. (B) Cellular fractions were extracted following the indicated treatments. Representative immunoblot is shown from n = 3 with histone H3 as loading control. (C) U2OS cells were pretreated with either DMSO or MLN4924 for 2 hours, pulsed with EdU for the last 20 min, and immunostained using the pMCM2 antibody. Bar graph depicts means ± SD from n = 4. Representative staining images are shown. Scale bars, 20 μm. (D) U2OS cells were subjected to fork elongation assay as indicated. IdU lengths measured from dual-labeled fibers are plotted. Fiber data pooled from n = 3 are shown. At least 100 fibers were quantified per condition. Scale bars, 20 μm. (E) U2OS cells treated with DMSO, MLN4924, or CPT for the indicated times were subjected to neutral comet assay, and tail moments were quantified. Representative experiment is shown from n = 2. (F) Whole-cell lysates were extracted from U2OS cells subjected to the indicated treatments. Representative immunoblot is shown from n = 3 with histone H3 as loading control.

Next, we asked whether reduction in fork elongation during acute MLN4924 exposure is due to DNA breaks. Both neutral comet assay (Fig. 1E) and alkaline comet assay (fig. S1E) reveal lack of observable tail moments after transient MLN4924 exposure. Replication forks acutely challenged with MLN4924 also exhibit intact fork recovery, suggesting that MLN4924 does not cause replication fork collapse (fig. S1F). Although MLN4924 exposure for 24 hours triggered ataxia telangiectasia and Rad3-related (ATR) checkpoint signaling (Fig. 1F) and accumulation of cells in the S phase (fig. S1G), as previously observed (28), short-term MLN4924 exposure minimally affected cell cycle progression without activating the ATR signaling response. Collectively, these results indicate that fork deceleration in short-term, MLN4924-treated cells occur in the absence of DNA breaks and suggests the presence of a distinct mechanism to curb fork progression during acute origin reusage.

CDT1 interacts with the CMG helicase to curb fork progression

During the S phase, CRL4^CDT2 targets PCNA-bound CDT1 for proteasomal degradation. Because MLN4924 overproduces CDT1, and CDT1 suppresses nascent strand synthesis (21, 22), we reasoned that CDT1 might exhibit increased propensity to associate with active replication forks and disrupt replication fork progression. Dose-response analyses indicate that the gradual stabilization of CDT1 strongly correlated with a progressive decline in replication rates (R² = 0.97) (Fig. 2A and fig. S2A). Next, we conducted transient transfections with small interfering RNAs (siRNAs) targeting CUL1 (subunit of SCF^SKP2) or CUL4A/4B (subunits of CRL4^CDT2) to ask whether down-regulating CDT1-targeting cullins also stabilize CDT1 and reduce fork elongation. Although depletion of either CUL4A or CUL4B slowed replication forks, CUL1 loss minimally affected fork progression (fig. S2B). Notably, despite 72 hours of siRNA transfection, depleting either CUL1, CUL4, or CDT2 did not trigger CDT1 accumulation to the extent observed with MLN4924 exposure (Fig. 2B), supporting the notion that heightened CDT1 levels interfere with replication fork progression. To further support this conclusion, CDT1 was overexpressed under the control of doxycycline-inducible promoter. Inducible expression of CDT1 suppressed fork progression (Fig. 2C) without a detectable increase in DNA breaks (fig. S2C). Cell cycle distribution profiles also remain unchanged, ruling out possible off-target effects on cell cycle progression due to CDT1 overexpression (fig. S2D).

Fig. 2. CDT1 interacts with the CMG helicase to curtail replication elongation.

(A) U2OS cells were subjected to fork elongation assay as indicated. Fiber data pooled from n = 3 are shown. At least 100 fibers were quantified per condition. Representative immunoblot is shown with GAPDH as loading control. Replication speed is indicated in parentheses as kb/min. (B) Whole-cell lysates were extracted following transfection with the indicated siRNAs or 2 hours of treatment with MLN4924. Representative immunoblot is shown from n = 2 with GAPDH as loading control. Arrows represent the neddylated cullin species. (C) U2OS cells transfected with the inducible CDT1-expressing construct were subjected to fork elongation assay. Fiber data pooled from n = 3 are shown. At least 100 fibers were quantified per condition. Representative immunoblot is shown with KU70 as loading control. Dox, doxycycline. (D) HCT116 CDT1 mAID-Clover cell lines were subjected to fork elongation assay as indicated. Representative experiment is shown from n = 2. At least 100 fibers were quantified. Representative immunoblot is shown with GAPDH as loading control. (E) U2OS cells were subjected to fork elongation assay. Fiber data pooled from n = 3 are shown. At least 100 fibers were quantified per condition. (F) U2OS cells transfected with either CDT1.WT or CDT1.ΔPIP inducible constructs were subjected to fork elongation assay as indicated. Representative experiment is shown from n = 2. At least 100 fibers were quantified. Representative immunoblot is shown with GAPDH as loading control. (G) pMCM2-CDT1 proximity pairs were measured in the indicated samples, and values normalized to pMCM2-EdU PLA foci were plotted. Bar graph depicts means ± SD from n = 3. (H) U2OS cells stably transfected with WT CDT1 (WT), R210C (RC), or E468K (EK) inducible constructs were subjected to fork elongation assay. Fiber data pooled from n = 3 are shown. At least 100 fibers were quantified per condition. Representative immunoblot is shown with KU70 as loading control.

If up-regulation of CDT1 is responsible for impeding DNA replication, then removing CDT1 should rescue the fork progression defect in MLN4924-treated cells. Concomitant CDT1 degradation with MLN4924 treatment in CDT1-degron cell lines completely abrogated the restraint on replication fork progression (Fig. 2D). Transient depletion of CDT1 also largely restored fork progression rates in MLN4924-treated cells (fig. S2E). To exclude the possibility that defective fork progression in MLN4924-treated cells is due to new origin firing, we analyzed replication tracts using CDK2i roscovitine to block initiation of new origins. As a control, we included cells acutely treated with ATRi that causes fork slowing due to unscheduled origin activation (32). Although CDK2i restored fork progression in ATRi-treated cells, fork progression remained defective in cells cotreated with MLN4924 and CDK2i (Fig. 2E). These data indicate that CDT1 curbs replication fork progression independent of new origin firing.

Before PCNA-mediated degradation at replication forks, CDT1 suppresses CMG activity at the onset of the S phase (23). We wished to determine whether the fork curtailment occurs due to increased retention of CDT1 on PCNA or enhanced association of CDT1 with the CMG helicase. First, we quantified CDT1-PCNA proximity pairs using conditions that block PCNA-mediated CDT1 degradation at replication forks: CDT2 depletion, proteasomal inhibition, and MLN4924 treatment. Although CDT1 interaction with PCNA was elevated under all conditions tested (fig. S2F), slower forks were only observed in cells acutely treated with MLN4924 (fig. S2G), suggesting that DNA synthesis is not affected by increasing CDT1 engagement with PCNA. To confirm these observations, we used an inducible CDT1 construct that lacks the PCNA-interacting protein (PIP) degron (CDT1.ΔPIP). The PCNA association of CDT1.ΔPIP was substantially reduced by proximity ligation assays (PLAs) (fig. S2H). However, like wild-type CDT1 (CDT1.WT) overexpression, cells expressing the CDT1.ΔPIP exhibit shortened nascent DNA tracts, indicating that CDT1-mediated fork slowing is not dependent on PCNA interaction (Fig. 2F). Next, we asked whether CDT1 limits DNA synthesis due to its association with the CMG helicase. We first conducted PLAs to quantify CDT1 association with CMG using antibodies recognizing CDT1 and pMCM2.S139. Increased number of CDT1-pMCM2 proximity pairs was observed in EdU+ cells overexpressing CDT1 or CDT1.ΔPIP (Fig. 2G) and cells treated with MLN4924 (fig. S2I), suggesting that CDT1 interaction with CMG is heightened when overexpressed. Two CDT1 partial loss of function mutants, Arg²¹⁰ → Cys (R210C) and Glu⁴⁶⁸ → Lys (E468K), were previously reported to have impaired binding to the MCM2-7 complex (33). We used these hypomorphs to address whether fork restraint by CDT1 is dependent on the interaction with CMG. PLAs indicate reduced association of both CDT1 variants with CMG (fig. S2J). We then used these variants to examine effects on fork progression when CDT1 interaction with MCM is weakened. Overexpression of either the R210C or E468K CDT1 variants alleviated fork progression defects (Fig. 2H). Overall, these results demonstrate that CDT1 interacts with the CMG helicase to block fork movement.

CDT1 mediates replication fork remodeling by replication fork reversal

To delineate the mechanism of CDT1-mediated fork deceleration, we examined the presence of either parental or nascent ssDNA at replication forks by immunofluorescence (IF) under nondenaturing conditions (34). Using nucleoside analog 5′-iododeoxyuridine (IdU), cells were either labeled overnight or for a short duration to detect exposed parental ssDNA or nascent ssDNA respectively followed by immunostaining with IdU antibody under native conditions. Transient MLN4924 exposure failed to induce accumulation of parental ssDNA (fig. S3A). Exposed template DNA was detectable in cells treated with hydroxyurea (HU) and ATR inhibitor (ATRi) VE-821, confirming that the assay was functional. In contrast, MLN4924 treatment increased nascent DNA exposure (Fig. 3A). This exposure was due to CDT1 stabilization as CDT1 depletion abrogated the accumulation (fig. S3B). As DNA breaks were not observable in MLN4924-treated cells, thereby eliminating break-induced resection as a source of ssDNA, we hypothesized that the exposed nascent ssDNA is indicative of fork reversal events. Consistent with our prediction, depletion of fork reversal enzyme SMARCAL1 reduced the exposure of nascent ssDNA in MLN4924-treated cells (Fig. 3B). Because fork reversal slows fork progression, we asked whether the fork restraint is SMARCAL1-mediated in MLN4924-treated cells. DNA fiber analyses indicate restoration of fork progression upon removal of SMARCAL1 (Fig. 3C). Induction of CDT1 overexpression alone was sufficient to trigger SMARCAL1-dependent fork slowing (fig. S3C), suggesting that CDT1 accumulation causes fork remodeling by replication fork reversal. Slowed fork progression in MLN4924-treated cells was rescued upon removal of SMARCAL1 or ZRANB3 but not HLTF (fig. S3D).

Fig. 3. CDT1 mediates replication fork remodeling.

(A) U2OS cells treated with either DMSO or MLN4924 or cotreated with HU and ATRi were immunostained for exposed nascent IdU under native conditions. Representative experiment is shown from n = 3. Representative images are shown. Scale bars, 20 μm. (B) siNT- or siSMARCAL1-transfected U2OS cells were treated with either DMSO or MLN4924 and immunostained for exposed nascent IdU under native conditions. Representative experiment is shown from n = 3. (C) siNT- or siSMARCAL1-transfected U2OS cells subjected to fork elongation assay as indicated. Fiber data pooled from n = 3 are shown. At least 100 fibers were quantified per condition. Representative immunoblot is shown with GAPDH as loading control. (D) U2OS cells transfected with the indicated siRNAs for 48 hours were subjected to nascent strand degradation assay as shown. Fiber data pooled from n = 3 are shown. At least 100 fibers were quantified per condition. Representative immunoblot is indicated with GAPDH as loading control. (E) mAID-Clover-CDT1 HCT116 cells transfected with the indicated siRNAs for 48 hours were subjected to nascent strand degradation assay as shown. Representative experiment is shown from n = 2. At least 120 fibers were quantified. Representative immunoblot is shown with GAPDH as loading control. (F) siNT- or siBRCA2-transfected U2OS cells stably transfected with WT, R210C, or E468K inducible CDT1 constructs were subjected to nascent strand degradation assays as shown following pretreatment with doxycycline for 48 hours. Representative experiment is shown from n = 2. At least 100 fibers were quantified. Representative immunoblot is shown with GAPDH as loading control. (G) U2OS cells transfected with the indicated siRNAs were subjected to fork elongation assay as indicated. Representative experiment is shown from n = 2. At least 60 fibers were quantified. Representative immunoblot is shown with GAPDH as loading control.

We considered two possible scenarios to explain why fork reversal is engaged when CDT1 is overexpressed. First, CDT1 accumulation could generate replication obstacles to stall ongoing forks. Lack of ATR activation during acute MLN4924 exposure suggests that fork stalling events are largely absent. This led us to postulate that CDT1 may act as a fork remodeling factor, thereby necessitating the need to be degraded during the S phase to ensure unperturbed fork progression. To test this idea, we performed fork protection assays to examine CDT1 function under conditions that trigger fork reversal. In HU-treated, BRCA2-deficient cells, nucleases digest nascent DNA at reversed replication forks and inhibiting fork reversal restores fork protection (35). We used these conditions to examine whether CDT1 alters nascent DNA protection. Stability of nascent DNA is largely intact when CDT1 is removed (Fig. 3D). Notably, CDT1 codepletion in either siBRCA2-transfected U2OS or hTERT-RPE1 cells restored replication fork protection (Fig. 3D and fig. S3E). This is not due to indirect effects of CDT1 depletion as siCDT1-transfected cells exhibit normal fork progression (fig. S3F) with largely unchanged S-phase distribution (fig. S3G). To further corroborate these observations, we performed fork protection assays in CDT1-degron HCT116 cells. Pretreatment with the ligand abolished nascent strand degradation when BRCA2 is absent (Fig. 3E). To rule out the possibility that the forks in CDT1-depleted cells are predisposed to being resistant to nascent DNA digestion, we added the ligand during the HU treatment under BRCA2-deficient conditions (fig. S3H). Concomitant addition of HU and ligand also restored fork protection, demonstrating that CDT1 is likely involved in generating reversed forks for nascent DNA digestion. Measurement of chromatin-bound RAD51 by quantitative imaging showed that reduced RAD51 intensity in BRCA2-depleted cells persists when CDT1 was codepleted (fig. S3I), indicating that restoration of fork protection is not due to increased RAD51 accumulation at forks. CDT1 steady-state levels also remain unchanged upon HU exposure (fig. S3J). Notably, under HU-treated, BRCA2-deficient conditions, nascent DNA digestion persisted upon inducible expression of WT CDT1. In comparison, expression of both CDT1 hypomorphs restored fork protection (Fig. 3F). These results are consistent with idea that CDT1 interaction with CMG generates reversed forks.

To further test that CDT1 promotes replication fork reversal in the presence of replication stress, we measured replication elongation rates under low-dose camptothecin (CPT) conditions that induce fork reversal (36). If CDT1 mediates fork reversal, removing CDT1 should increase the length of replication tracts in the presence of low-dose CPT. Replication tracts were restored to control levels when CDT1 was inactivated (Fig. 3G). In the absence of fork reversal, primase and DNA directed polymerase (PRIMPOL) mediates repriming downstream of the lesion to reinitiate DNA synthesis (37). Thus, we codepleted PRIMPOL in CDT1-deficient cells and observe that removing PRIMPOL abrogated the rescue in CPT-treated, CDT1-depleted cells, indicating that PRIMPOL is engaged as an alternative pathway when fork remodeling by CDT1 is abolished (Fig. 3G). Furthermore, acute CDT1 degradation also restores fork progression across multiple replication stress conditions that trigger fork reversal (36) (fig. S3K). Overall, these results imply CDT1 as a facilitator of replication fork reversal.

PCNA unloading by ATAD5 promotes fork restraint by CDT1

We wished to define the mechanism that enables CDT1 to participate in replication fork reversal despite being subjected to proteolysis in the S phase. Exposure to replication stress triggers rapid unloading of PCNA from stressed forks (38, 39), and this unloading also accompanies CDT2 removal as observed in previously published replication fork proteomic datasets (38) (fig. S4A). We reasoned that the reduced presence of PCNA at stalled forks would favor CDT1 to associate with CMG and promote fork reversal. To test this idea, we first examined the abundance of PCNA at MLN4924-treated replication forks by iPOND (isolation of proteins on nascent DNA) (40). Cells exposed to MLN4924 for 2 hours exhibit reduced levels of PCNA at replication forks compared to dimethyl sulfoxide (DMSO)–treated cells (Fig. 4A), and this observation was further corroborated by IF analyses (fig. S4B). Inducible CDT1 overexpression also suppressed chromatin-bound PCNA levels in replicating cells (fig. S4C). Despite increasing new origin firing, CDT1 stabilization reduced PCNA levels at forks, suggesting that PCNA is actively removed from sites of DNA synthesis. The ATAD5-RLC complex is primarily responsible for PCNA unloading during fork stalling events (41, 42). Transient depletion of ATAD5 using two different siRNAs restored PCNA intensities in MLN4924-treated cells (Fig. 4B and fig. S4D), indicating that the ATAD5-RLC complex functions in PCNA unloading from replication forks when CDT1 is overproduced. If PCNA unloading enhances CDT1 association with CMG, we predicted that increasing PCNA levels at MLN4924-treated forks will disfavor CDT1-CMG association and restore fork speeds. ATAD5 depletion reduced CDT1-CMG proximity pairs (fig. S4E) and rescued elongation rates (Fig. 4C) in MLN4924-treated cells. ATAD5 depletion alone had no impact on cell cycle progression (fig. S4F), although we cannot rule out other secondary effects due to ATAD5 removal. These results suggest that PCNA unloading by ATAD5 promotes CDT1 function to limit re-replication.

Fig. 4. ATAD5-dependent PCNA unloading mediates CDT1 function in fork remodeling.

(A) HEK293T cells were left unlabeled or labeled with EdU for either 10 min in the presence of DMSO or 13.5 min in the presence of MLN4924 before performing iPOND. EdU-labeled cells chased into EdU-free media for 2 hours were also included. Representative immunoblot of input and captures is shown from n = 2 with histone H3 as loading control. (B) U2OS cells transfected with either siNT or two different ATAD5-targeting siRNAs were treated as indicated and immunostained for chromatin-bound PCNA, and mean intensities in EdU+ nuclei were quantified. Representative experiment is shown from n = 2. (C) siNT- or siATAD5-transfected U2OS cells were subjected to fork elongation assay as indicated. Representative experiment is shown from n = 2. At least 100 fibers were quantified. (D) U2OS cells transfected with the indicated siRNAs were subjected to nascent strand degradation assays. Representative experiment is shown from n = 2. At least 100 fibers were quantified. Representative immunoblot is indicated with GAPDH as loading control. (E) siNT- or siATAD5 #2-transfected U2OS cells were subjected to fork elongation assay as indicated. Fiber data pooled from n = 3 are shown. At least 100 fibers were quantified per condition. Unt, untreated. (F) U2OS cells transfected with either CDT1.WT or CDT1.ΔPIP inducible constructs were cotransfected with siNT, siBRCA2, or siATAD5 with siBRCA2 and subjected to nascent strand degradation assay. Fiber data pooled from n = 3 are shown. At least 100 fibers were quantified per condition. Representative immunoblot is indicated with GAPDH as loading control. (G) Model depicting regulation of CDT1 at active and stalled forks is shown.

To test whether PCNA unloading also operates at HU-stalled forks to facilitate CDT1-mediated fork remodeling, we first performed PLAs to test whether CDT1 associates with nascent DNA under conditions of HU exposure. Although the proximity of PCNA to newly synthesized DNA is reduced under HU conditions, CDT1 association with nascent DNA is enhanced following HU exposure (fig. S4G). This result suggests that PCNA unloading by ATAD5 facilitates CDT1 engagement with stalled forks. Next, we tested whether ATAD5 contributes to the fork degradation phenotype in BRCA2-deficient cells. For this purpose, we performed fork protection assays in cells codepleted of BRCA2 and ATAD5. Although nascent DNA tracts underwent digestion in BRCA2-depleted cells, ATAD5 removal restored nascent DNA stability (Fig. 4D). ATAD5 depletion also restored fork protection in BRCA2-depleted hTERT-RPE1 cells (fig. S4H), indicating that PCNA unloading by ATAD5 facilitates the generation of reversed forks, as observed previously (42), to cause nascent strand degradation in BRCA2-deficient cells. To examine whether ATAD5 also promotes PCNA removal under conditions other than high-dose HU, we exposed cells to additional stress conditions that induce replication fork reversal (36). Under all conditions tested, we observe reduced chromatin-bound PCNA levels in EdU+ nuclei (fig. S4I) and suppressed fork speeds (Fig. 4E). Furthermore, depletion of ATAD5 restored both PCNA levels (fig. S4I) and elongation rates (Fig. 4E) under replication stress conditions, suggesting that PCNA eviction facilitates replication fork reversal by CDT1.

Next, to establish that restoration of fork protection in ATAD5-depleted, BRCA2-deficient cells is linked to the ability of CDT1 to interact with CMG, we used CDT1.PIPΔ inducible vector to disrupt PCNA engagement under ATAD5-deficient conditions. We predicted that PCNA retention at forks by ATAD5 removal disfavors CDT1-CMG interaction. Under these conditions, blocking CDT1-PCNA interaction would help CDT1 reassociate with CMG and invoke nascent strand degradation. To test this prediction, we first asked whether the reduced fork speeds by CDT1.PIPΔ overexpression (Fig. 2F) is also dependent on fork reversal. DNA fiber assays indicate that SMARCAL1 depletion restored elongation rates when CDT1.PIPΔ is overexpressed (fig. S4, J and K), consistent with the observation that CDT1.PIPΔ engages with the CMG helicase (Fig. 2G). Next, we performed fork protection assays in cells codepleted of BRCA2 and ATAD5 and asked whether overexpressing CDT1.PIPΔ elicits nascent strand degradation. BRCA2-deficient cells expressing either CDT1.WT or CDT1.PIPΔ continued to exhibit nascent strand degradation (Fig. 4F), indicating that fork reversal is not affected by either CDT1 variant. Although IdU/5-chloro-2′-deoxyuridine (CldU) ratios remain unchanged in BRCA2 and ATAD5 codepleted cells overexpressing CDT1.WT, CDT1.PIPΔ triggered nascent strand degradation likely due to preferential association of CDT1 with CMG. Collectively, these results suggest that PCNA unloading at forks facilitates CDT1 interaction with CMG to promote fork reversal.

DISCUSSION

Replication fork reversal is a key stress tolerance pathway commonly used by cells in response to replication stress (27, 43). Although much is known about fork reversal proteins that remodel DNA to convert three-way junctions into four-way structures, less is understood about factors that associate with the CMG helicase to facilitate this process. This study reveals an unexpected role of licensing factor CDT1 in mediating fork remodeling by interacting with the CMG helicase.

We find that acute stabilization of CDT1 by neddylation inhibitor MLN4924 suppressed fork speeds. Although the decrease in fork progression due to persistent origin reusage is primarily linked to DSBs, the restricted fork movement during acute re-replication is independent of break induction. Rather, CDT1 associates with the CMG helicase at replication forks to curtail DNA synthesis independently of new origin firing. In addition, we observe that CDT1 triggers replication fork reversal not only during re-replication but also under conditions where cells are exposed to replication stress inducing agents. The ability of CDT1 to participate in fork reversal is mediated by ATAD5-dependent unloading of PCNA. PCNA unloading occurs in response to MLN4924 and inducible expression of CDT1, suggesting that the function of the PCNA unloading apparatus is tightly coupled with replication fork progression. As PCNA interaction stimulates CDT1 proteolysis at forks, PCNA removal permits CDT1 to engage with the CMG helicase. Last, we find that PCNA unloading by ATAD5 occurred in response to several conditions that trigger fork reversal and caused nascent strand degradation in BRCA2-deficient cells, highlighting the need to disengage PCNA at replication forks to facilitate fork regression by CDT1.

CDT1 licenses the genome during the G₁ phase by loading double MCM hexamers (1). CDT1 inhibition in the ensuing S phase is necessary to prevent re-replication (44–47). Our studies underline an additional need for CDT1 proteolysis to promote unperturbed replication fork progression. Although geminin blocks CDT1-MCM interaction in S and G₂ phases of the cell cycle (15, 16, 48), CRL4^CDT2 degrades CDT1 bound to PCNA at replication forks (45). We propose two spatially separated modes of restricting CDT1 function wherein geminin acts globally to prevent excessive origin firing and PCNA acts locally, specifically at forks, to prevent perturbations to fork progression. We also show that CDT1 interaction with CMG is required to curtail fork speeds as CDT1 hypomorphs that display reduced association with the CMG helicase are not detrimental to fork progression when overexpressed. These results are consistent with the role of CDT1 in inhibiting the CMG helicase in the early S phase (23) and inhibiting nascent strand synthesis when overexpressed during DNA replication (21, 22).

Our findings also demonstrate that CDT1 facilitates replication fork reversal. In MLN4924-treated cells, fork slowing is SMARCAL1 and ZRANB3 dependent but HLTF independent. The observed lack of HLTF involvement was also observed during re-replication in geminin-depleted cells (49). Notably, when CDT1 is overproduced, fork movement is restricted by fork reversal as depletion of DNA translocase SMARCAL1 restores elongation rates. Because the SMARCAL1 dependency was observed under conditions that did not involve addition of agents that block DNA replication, we postulated that CDT1 was stimulating fork reversal. In support of this prediction, CDT1 loss restored fork protection at HU stalled forks in BRCA2-deficient cells wherein reversed forks are substrates for nascent strand degradation. In addition, in the presence of low-dose CPT, CDT1 depletion activated PRIMPOL-dependent DNA synthesis, further demonstrating that CDT1 promotes fork reversal to counteract replication stress. On the basis of these observations, CDT1-mediated fork remodeling could also block CMG movement in the early S phase (23), provided that sufficient lengths of nascent leading and lagging strands are generated to engage fork reversal enzymes.

Our data support a model where CDT1 interacts with the CMG replicative helicase to promote fork reversal. CDT1 could stimulate replication fork reversal by inhibiting the CMG helicase activity. Alternately, the CMG helicase configuration could be altered by CDT1 to facilitate replication fork reversal. Single molecule experiments have shown that the CMG helicase switches between ssDNA and dsDNA binding modes when uncoupled from replicative polymerases (50). This mechanism involves MCM10 and is independent of the MCM2-5 gate used by CDT1 to encircle dsDNA during the origin licensing step (51, 52). A recent study also shows the CMG helicase is trapped in an ssDNA bubble ahead of the replication fork, allowing RAD51 and fork reversal enzymes to remodel DNA behind the CMG helicase (53). Considering these observations, we propose that CDT1 is an accessory factor that would enable the CMG helicase to navigate replication obstacles in a versatile manner. CDT1 binding could reconfigure MCM2-7 interaction interfaces of the CMG complex to help encircle either the nascent-parental DNA duplex on the leading strand or the parental DNA duplex ahead of the fork junction, which would entail mechanisms preventing CMG disassembly from dsDNA. Alternately, CDT1 could trap the CMG helicase ahead of the fork, without transitioning onto dsDNA. In this configuration, unrestrained helix unwinding would be prohibited due to inhibition of the CMG helicase activity by CDT1. Whether CDT1 modulates CMG function by direct engagement or by altering replisome composition remains to be investigated. Our data are also consistent with the observations that the CMG helicase is not displaced during fork reversal events (53, 54) because CDT1-induced replication fork remodeling during MLN4924 treatment had no impact on replication fork recovery.

Our results indicate that PCNA removal from replication forks is a near ubiquitous response to conditions that induce replication fork reversal and illustrate the need to displace PCNA to promote fork regression. Consistent with this model, depletion of PCNA unloader ATAD5 reverses nascent strand degradation in BRCA2-deficient cells treated with high-dose HU likely due to the absence of generating a reversed fork substrate. In addition, ATAD5 depletion mitigates fork slowdown across several mild replication perturbations that trigger replication fork reversal, indicating that PCNA unloading favors fork regression. We speculate that PCNA mobilization enables fork reversal enzymes to translocate along the dsDNA in an unobstructed manner and catalyze branch migration. For instance, freeing the 3′ nascent DNA end on the leading strand could stimulate HLTF recognition of the 3′ DNA end to catalyze fork reversal (55). In addition, our results illustrate a model wherein PCNA unloading promotes CDT1-mediated fork reversal. In this model (Fig. 4G), PCNA at active replication forks sequesters CDT1 away from the CMG helicase to ensure forks proceed without any interruptions. However, in response to replication stress, reduced PCNA, and thereby CDT2, partially relieves CDT1 inhibition, allowing CDT1 to redistribute toward the CMG helicase. This model implies that several molecules of PCNA must be removed to promote CDT1 function at stalled replication forks. This is observed in prior studies depicting PCNA unloading from lagging strands under conditions of replication stress (38, 39). Both replication fork reversal and PCNA removal occur within minutes of exposure to replication stress. Rapid removal of PCNA would thus promote substrate recognition by fork reversal enzymes and CDT1 relocalization to ensure that the CMG location or function is tightly coordinated with fork reversal events.

Although our genetic studies are consistent with the role of CDT1 in promoting replication fork reversal, additional validation by electron microscopy will be important to determine the structure of the replication intermediates generated under CDT1-proficient and CDT1-deficient conditions. Our studies did not examine the consequences of altering CDT1 function on replication fork stability, genome integrity, and cell survival to genotoxic stress. Separation-of-function mutants that dissect the fork remodeling function from licensing will be critical to elucidate the role of CDT1 under conditions of replication stress.

Together, this study identifies CDT1 as a replication stress tolerance factor and provides an avenue to further investigate how CDT1 interaction with the CMG helicase facilitates fork reversal. Furthermore, MLN4924 is currently investigated in clinical trials as a chemotherapeutic agent and targeting CDT1 function at forks could help potentiate MLN4924 toxicity in cancers.

MATERIALS AND METHODS

Cell lines

U2OS cells (RRID:CVCL_0042) were cultured in Dulbecco’s modified Eagle’s medium (DMEM) (Gibco, 11965-092) supplemented with 7.5% fetal bovine serum (FBS) (GeminiBio, S11150). hTERT-RPE1 cells (RRID:CVCL_4388) were cultured in DMEM/F-12 media (Gibco, 11320-033) supplemented with 7.5% FBS. HEK293T cells (RRID:CVCL_0063) were cultured in high-glucose with l-glutamine media [American Type Culture Collection (ATCC), 30-2002] supplemented with 10% FBS and 2 mM l-glutamine (Gibco, 25030-081). HCT116 CDT1 mAID-Clover cell lines were cultured in McCoy’s 5A modified media (Gibco, 16600-082) containing 10% FBS and 2 mM l-glutamine. HCT116 p53 +/+ and HCT116 p53−/− cells were cultured in McCoy’s 5A media containing 7.5% FBS and were a gift from R. Ohi. Stable Flp-in Trex U2OS cell lines with inducible expression of CDT1 WT and variants were a gift from J. Cook. All cell cultures were grown at 37°C with 5% CO₂. U2OS, hTERT-RPE1, and HEK293T cells are derived from human females. HCT116 cells are derived from human males. U2OS, hTERT-RPE1, and HEK293T cells were purchased from the ATCC and authenticated using short tandem repeat profiling. HCT116 cell lines were not verified. HCT116 AID2-CDT1 cell lines were generated by CRISPR-Cas9–mediated genome editing as previously described (56, 57). Briefly, we designed a CRISPR-Cas9 plasmid for targeting the CDT1 locus [5′-GACGCGGCGCTGCTCCATGG(CGG)-3′]. A donor plasmid harboring mAID-Clover and a selection marker (HygroR) with two homology arms (∼500 base pairs each) was constructed. HCT116 cells stably expressing OsTIR1(F74G) from the AAVS1 locus were transfected with the CRISPR and donor plasmids. Clones surviving hygromycin selection were screened for biallelic insertion of the tagging construct by genomic polymerase chain reaction, and expression of the fusion protein was confirmed by immunoblotting.

Transfections

siRNA transfections in U2OS cells were performed using Dharmafect1 (Dharmacon, T-2001-03). siRNA transfections in HCT116 and hTERT-RPE1 cells were performed using (Lipofectamine RNAiMAX, 13778030). Plasmid transfections were performed using FuGENE HD (Promega, E2311). Plasmid transfections were analyzed 72 hours posttransfection. siRNA transfections, except for siCDT1, were analyzed 72 hours posttransfection unless otherwise indicated. siCDT1-transfected cells were analyzed either 24 or 48 hours posttransfection as indicated. For inducible expression of CDT1 using the pInducer20-Blast-CDT1-HA vector, fresh media containing doxycycline (100 ng/ml) were added 24 hours posttransfection and incubated for additional 48 hours before experimentation. For inducible expression of CDT1 using pCW-Cdt1-mCherry-Puro and pCW-Cdt1ΔPIP-mCherry-Puro, fresh media containing doxycycline (500 ng/ml) were added 24 hours posttransfection and incubated for additional 48 hours before experimentation. For inducible expression of CDT1 in stable U2OS cell lines, cells were treated with media containing doxycycline (500 ng/ml) for 48 hours before experimentation. Cells without doxycycline were used as negative controls.

DNA fiber analysis

A total of 250,000 cells were plated the day before experimentation. On the day of the experiment, cells were pulse labeled with IdU and CldU using experimental designs as indicated below.

For fork elongation assays, cells were pulsed with 20 μM CldU for 30 min, washed with 1x Hanks’ balanced salt solution (HBSS), and subsequently pulsed with 100 μM IdU for 30 min. MLN4924 (1 μM) or DMSO was added 1 hour prior and kept throughout CldU and IdU labeling for a total of 2-hour treatment. For experiments using CDT1-mAID cell lines, 5 μM 5-Ph-IAA was added for 2 hours as indicated in the respective experiments. For experiments using low-dose replication perturbations, the following reagents were added concomitantly with IdU for 30 min: 0.5 mM HU, 25 nM CPT, 100 nM aphidicolin (APH), and 1 μM cisplatin (CIS). The IdU lengths of dual-labeled replication tracts were quantified and plotted.

For fork protection assays, cells were sequentially pulsed with 20 μM CldU and 100 μM IdU for 30 min each followed by release into 4 mM HU for 4 hours. MLN4924 (1 μM) was added with HU treatment in indicated samples. For experiments using CDT1-mAID cell lines, 5 μM 5-phenyl-indole-3-acetic acid (5-Ph-IAA) was added as indicated in the respective experiments. IdU and CldU lengths of dual-labeled replication tracts were measured, and IdU/CldU ratios were quantified to examine defects in replication fork protection.

For fork recovery assays, cells were treated with 20 μM CldU for 30 min, rinsed with 1x HBSS, and treated with either 2 hours of 1 μM MLN4924, 4 hours of 4 mM HU and 3 μM ATRi (VE-821), or 30 min of DMSO, rinsed again with 1x HBSS, and pulse labeled with 100 μM IdU for 30 min. To quantify, the percentage of CldU-only fibers compared to dual-labeled fibers was plotted.

For new origin firing analysis, cells were pulsed with 20 μM CldU for 30 min, rinsed with 1x HBSS, and then pulse labeled with 100 μM IdU for 120 min with or without MLN4924. Following pulse labeling, cells were harvested and resuspended in cold 1x Dulbecco’s phosphate-buffered saline (DPBS). Two microliters of DNA was spread on microscope slides, allowed to dry for 5 min, and lysed with 10 μl of lysis buffer [50 mM EDTA, 200 mM tris-HCl (pH 7.4), and 0.5% SDS] for 6 min. DNA was spread by tilting the microscope slides at a 15° angle to stretch the DNA down the slide and air dried. Slides were then fixed in 3:1 methanol:acetic acid for 2 min, air dried, and stored at −20°C overnight. The next day, slides were rehydrated in phosphate-buffered saline (PBS) for 5 min and then denatured in 2.5 M HCl for 38 min at room temperature. Slides were then blocked in PBS containing 0.1% Triton X-100 and 10% goat serum for 1 hour and stained with primary antibodies for 2 hours to recognize CldU (1:100, anti-BrdU, Abcam) and IdU (1:10, anti-BrdU, BD Biosciences), followed by staining with secondary antibodies (1:350, Goat anti-rat Alexa Fluor 594, and 1:175, Goat anti-mouse Alexa Fluor 488) for 1 hour. Fibers were imaged on Keyence BZ-X810 at 60x oil immersion, and fiber lengths were analyzed on ImageJ.

Immunofluorescence

For IF experiments, 250,000 cells were seeded on coverslips the day prior. For experiments with 1 μM MLN4924 or 10 μM MG132, cells were treated for 2 hours before experiment, with a 10 μM EdU pulse in the final 20 min of treatment to identify replicating cells. For experiments with HU, cells were pulsed for 10 min with 10 μM EdU before a 4-hour 4 mM HU treatment. Cells were then preextracted with Triton X-100 permeabilization buffer (20 mM Hepes, 50 mM NaCl, 3 mM MgCl₂, 300 mM sucrose, and 0.5% Triton X-100) for 10 min at 4°C before fixation with 3% paraformaldehyde/2% sucrose for 10 min at room temperature. Before antibody staining, coverslips were permeabilized a second time using Triton X-100 permeabilization buffer and blocked in 5% bovine serum albumin (BSA) in PBS solution for 15 min each. Replicating cells were identified via EdU click chemistry using Alexa Fluor 488 azide before incubation with primary and secondary antibodies. Samples were incubated in primary antibody for one hour at 37C and secondary antibody for 40 min at room temperature. Coverslips were mounted using ProLong Gold containing 4′,6-diamidino-2-phenylindole (DAPI) (Invitrogen). For PLAs, the Duolink In Situ Mouse/Rabbit Kit (Sigma-Aldrich, DUO92013) was used according to the manufacturer’s instructions. For in situ protein interaction with nascent DNA replication forks (SiRF) assay (58), cells were labeled with 10 μM EdU for 20 min and left either untreated or treated with HU for 4 hours concomitantly with EdU before fixation and conjugated to biotin-azide using click chemistry before PLA analysis. Image acquisition was performed on a Keyence BZ-X810 fluorescence microscope using a 40X objective (numerical aperture: 0.75). A BZ-X800 analyzer was used to measure mean nuclei intensity. Macro cell count on the Keyence BZ-X800 analysis software was used to quantify PLA foci.

ssDNA detection assays

To detect nascent ssDNA, cells were pulsed with 10 μM IdU for 15 min. Under conditions with MLN4924, cells were pretreated for 2 hours with IdU added in the last 15 min of treatment. Under conditions with HU + ATRi, cells were treated for 4 hours after the 15-min IdU pulse. Cells were preextracted with Triton X-100 permeabilization buffer (20 mM Hepes, 50 mM NaCl, 3 mM MgCl₂, 300 mM sucrose, and 0.5% Triton X-100) for 10 min on ice before fixation with 3% paraformaldehyde/2% sucrose for 10 min at room temperature. Coverslips were then blocked with 5% BSA/PBS for 15 min at room temperature before incubation with primary antibodies (1:50, anti-BrdU, BD Biosciences) for 1 hour at 37°C. After rinsing with 1x PBS, coverslips were incubated with secondary antibody (1:500, Goat anti-mouse Alexa Fluor 488) for 40 min at room temperature before mounting with ProLong Gold containing DAPI. IdU foci were quantified using macro cell count on the Keyence BZ-X800 analysis software.

To detect parental ssDNA, cells were pulsed with 10 μM IdU for 48 hours before rinsing once with warm PBS and released into fresh prewarmed media for 2 hours. Following, cells were treated with either 2 hours MLN4924 or 4 hours HU + ATRi before treating with Triton X-100 permeabilization buffer for 10 min at 4°C. Fixation with 3% paraformaldehyde/2% sucrose for 10 min at room temperature was performed before blocking with 5%BSA/PBS for 15 min at room temperature. Coverslips were incubated with primary antibodies (1:50, anti-BrdU, BD Biosciences) for 1 hour at 37°C, rinsed with 1x PBS, and incubated with secondary antibody (1:500, Goat anti-mouse Alexa Fluor 488) for 40 min at room temperature before mounting with ProLong Gold containing DAPI. IdU foci were quantified using macro cell count on the Keyence BZ-X800 analysis software

Protein extraction and immunoblotting

To obtain whole-cell lysates, cells were lysed on ice for 30 min in radioimmunoprecipitation assay lysis buffer [50 mM tris-Cl (pH 7.4), 1% NP-40, 150 mM NaCl, 0.1% SDS, 1 mM dithiothreitol (DTT), 0.5% sodium deoxycholate, 1 mM aprotinin, 1 mM leupeptin, 1 mM phenylmethylsulfonyl fluoride (PMSF), 20 mM β-glycerolphosphate, 1 mM MgCl₂, 1 mM sodium orthovanadate, and Pierce Universal Nuclease (1 μg/ml)]. After centrifugation at 13,200 rpm for 15 min at 4°C, supernatants were collected and lysates were quantified before immunoblotting on SDS–polyacrylamide gel electrophoresis gels.

For chromatin fractionation experiments, roughly 5 million cells were harvested for each sample and the procedure was performed as described previously (59). Cells were resuspended in buffer A containing [10 mM Hepes (pH 7), 10 mM KCl, 1.5 mM MgCl₂, 0.34 M sucrose, 10% glycerol, 0.1% TritonX-100, 1 mM DTT, 1 mM aprotinin, 1 mM leupeptin, 0.1 mM PMSF, and 1 mM Na₃VO₄] and left on ice for 5 min to extract cytoplasmic proteins. Cells were spun at 1300g for 5 min, and the supernatant was collected and spun again at 20,000g for 15 min to isolate the cytoplasmic fraction. The remaining cell pellet was washed once with buffer A to remove excess cytoplasmic proteins and spun for 5 min at 1300g, and the supernatant was discarded. The remaining cell pellet was resuspended in buffer B containing (3 mM EDTA, 0.2 mM EGTA, 1 mM DTT, 1 mM aprotinin, 1 mM leupeptin, 0.1 mM PMSF, and 1 mM Na₃VO₄) and lysed on ice for 30 min. Cells were spun at 1700g for 5 min, and the supernatant was isolated as the nuclear fraction. The remaining cell pellet was washed once with buffer B and spun for 5 min at 1700g, and the supernatant was discarded. The cell pellet was resuspended in 20 mM tris-HCl (pH 8), 0.2 mM MgCl₂, 2 mM NaCl, 1 mM aprotinin, 1 mM leupeptin, and Pierce Universal Nuclease (250 U/ml) and sonicated at 22% amplitude for 15 s to extract chromatin-bound proteins.

Comet assay

DNA breaks were detected using the R&D Systems CometAssay Electrophoresis System II comet assay kit (R&D Systems, 4250-050-ES) according to the manufacturer’s protocol. Briefly, cells were harvested and resuspended in low-melt agarose (Bio-Rad, #1613111), before spreading on a comet slide. After solidifying at 4°C, slides were immersed in buffer containing [0.5% SDS, 200 mM tris-HCl (pH 7.4), and 50 mM EDTA] for 1 hour at room temperature to lyse the cells. For neutral comet assay, slides were immersed in cold tris-acetate-EDTA (TAE) buffer for 30 min and placed in the comet apparatus for electrophoresis at 21 V for 40 min. Following incubation in DNA precipitation solution, slides were dried and stained the following day with SYBR Green solution. For alkaline comet assay, slides were immersed and electrophoresed in Alkaline Unwinding Solution (pH > 13, 200 mM NaOH, and 1 mM EDTA) in place of TAE, immersed in 70% ethanol for 5 min, and then dried and stained with SYBR Green the following day. Tail moments were scored using the Comet Score software (TriTek), and data were presented as box and whisker plots.

Flow cytometry

For cell cycle analysis, cells were harvested and processed in accordance with the manufacturer’s instructions (Abcam, ab287852). Cells were fixed overnight in 70% ethanol at −20°C and analyzed using flow cytometry.

iPOND Western blotting

The iPOND Western blotting (WB) procedure was performed as described previously (38). The day before the experiment, four dishes of 25 million HEK293T cells were seeded to obtain 50 million cells per dish the next day. On the day of the experiment, cells were pretreated for 2 hours with 1 μM MLN4924 or DMSO. The DMSO sample were labeled with 10 μM EdU for 10 min. MLN4924 samples were labeled with 10 μM EdU for 13.5 min to account for slowed fork progression. For the thymidine chase sample, following EdU labeling for 10 min, the media were discarded and replaced with equilibrated media containing 10 μM thymidine for 1 hour. EdU was not added to the “No EdU” sample as a negative control. Following treatments, media were discarded and replaced with 10 ml of 1% formaldehyde in PBS for 10 min at room temperature. Cross-linking was quenched by adding 1 ml of 1.25 M glycine. Cells were scraped with a cell lifter and decanted into 50-ml conical tubes. All plates for each sample were combined at this step. Cells were centrifuged for 5 min at 900g (4°C) and washed three times with cold 1x PBS. To permeabilize cells, the cell pellet was resuspended in 0.25% Triton X-100 in PBS at a concentration of 1 × 10⁷ cells/ml and incubated at room temperature for 30 min. Cells were spun down for 5 min at 900g (4°C), washed once with cold 0.5% BSA/PBS, and washed once more with cold PBS. From here, click chemistry reactions were performed to conjugate EdU labeled DNA to biotin. Cells were resuspended in buffer containing final volumes (10 μM biotin azide, 10 mM sodium ascorbate, and 2 mM copper sulfate in PBS) and rotated for 2 hours. Cells were spun down for 5 min at 900g (4°C), washed once with cold 0.5% BSA/PBS, and washed once more with cold PBS. Next, cells were lysed in [1% SDS in 50 mM Tris (pH 8.0), aprotinin (1 μg/ml), and leupeptin (1 μg/ml)] and sonicated at 33% amplitude for 30-s on and 30-s off, for a total of 2 min. The lysate was spun down for 10 min at 16,100g, and the supernatant was isolated. A 0.25% volume of each sample was set aside as “input” samples. To capture the biotin-tagged DNA, the lysate was rotated with streptavidin MyOne C1 beads at a concentration of 10-μl beads for every 1 × 10⁷ cells for 1 hour at room temperature. From here, beads were washed with a series of salt washes beginning with low salt [1% Triton X-100, 20 mM Tris (pH 8.0), 2 mM EDTA, and 150 mM NaCl], followed by high salt [1% Triton X-100, 20 mM Tris (pH 8.0), 2 mM EDTA, and 500 mM NaCl], and then lithium chloride salt buffer [100 mM Tris (pH 8.0), 500 mM LiCl, and 1% NP-40]. To elute proteins, beads were resuspended in 2x Laemmli buffer and boiled for 30 min. The supernatant is the eluted iPOND capture and was processed using immunoblotting.

Antibodies

The following antibodies were used for WB or IF as indicated: CDT1 (WB 1:500, IF 1:200, ab202067; Abcam; RRID:AB_2651122), CDT1 (PLA 1:200, sc-365305; Santa Cruz Biotechnology; RRID:AB_1084780), CUL1 (WB 1:500, sc-17775; Santa Cruz Biotechnology; RRID:AB_627325), CUL2 (WB 1:500, sc-166506; Santa Cruz Biotechnology; RRID:AB_2230072), CUL3 (WB 1:500, catalog no. 2759; Cell Signaling Technologies; RRID:AB_2086432), CUL4A (WB 1:500, catalog no. ab72548; Abcam; RRID:AB_1268363), CUL4B (WB 1:500, catalog no. VMA00360; Bio-Rad), CUL5 (WB 1:500, catalog no. A302-173A; Bethyl Laboratories; RRID:AB_1659801), p53 (WB 1:500, catalog no. 9282T; Cell Signaling Technology; RRID:AB_331476), CDT2 (WB 1:500, catalog no. ABS500; Sigma-Aldrich), pMCM2 (Ser¹³⁹) (PLA 1:200, IF 1:200, catalog no. 12958; Cell Signaling Technology; RRID:AB_2798069), PCNA (IF 1:200, PLA 1:200, WB 1:1000, catalog no. sc-56; Santa Cruz Biotechnology; RRID:AB_628110), phospho-Chk1 (Ser³¹⁷) (WB 1:500, catalog no. 2344; Santa Cruz Biotechnology; RRID:AB_331488), CHK1 (WB 1:500, catalog no. sc-8408; Santa Cruz Biotechnology; RRID:AB_627257), phospho RPA32 (S4/S8) (WB 1:250, catalog no. A300-245A; Bethyl Laboratories; RRID:AB_210547), RPA32 (WB 1:1000, IF 1:200, catalog no. ab2175; Abcam; RRID:AB_302873), glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (WB 1:8000, catalog no. MAB374; Millipore; RRID:AB_2107445), KU70 (WB 1:2000, catalog no. ab92450; Abcam; RRID:AB_10562280), histone H3 (WB 1:6000, catalog no. ab1791; Abcam; RRID:AB_302613), 5-bromo-2′-deoxyuridine (BrdU) (IF 1:100, catalog no. 347580; BD Biosciences for IdU immunostaining; RRID:AB_1001521), BrdU (IF 1:10, catalog no. ab6326; Abcam for CldU immunostaining; RRID:AB_305426 ), RAD51 (IF 1:200, catalog no. sc-398587; Santa Cruz Biotechnology; RRID:AB_2756353), SMARCAL1 (WB 1:500, catalog no. sc376377; Santa Cruz Biotechnology; RRID:AB_10987841), ZRANB3 (WB 1:1000, catalog no. A303-033A; Bethyl Laboratories; RRID:AB_10773114), HLTF (WB 1:1000, catalog no. ab183042; RRID:AB_3095312), BRCA2 (WB 1:250, catalog no. OP95; Millipore Sigma; RRID:AB_2067762), CCDC111/PRIMPOL (WB 1:1000, 28924-1-AP; Proteintech; RRID:AB_2918349), RAD51 (IF 1:200, catalog no. 398587; Santa Cruz Biotechnology; RRID:AB_2756353), ATAD5 (WB 1:500, catalog no. A301-057A; Bethyl Laboratories; RRID:AB_2258720), Rabbit Biotin (IF 1:200 catalog no. A150-109A; Bethyl Laboratories; RRID:AB_67327), and Mouse Biotin (IF 1:200, catalog no. 200-002-211; Jackson ImmunoResearch Labs; RRID:AB_2339006).

Chemicals and compounds

The following chemicals and compounds were used: MLN4924 (MedChemExpress, HY-70062), CPT (Sigma-Aldrich, 7689-03-4), HU (Sigma-Aldrich, 127-07-1), ATRi VE-821 (Selleck Chemicals, S8007), APH (Sigma-Aldrich A0781-5MG), CIS (Sigma-Aldrich, P4394), MG132 (Selleck Chemicals, S2619), 5-Ph-IAA (Sigma-Aldrich, SML3574), CldU (Sigma-Aldrich, C6891), EdU (Sigma-Aldrich, 61135-33-9), IdU (Sigma-Aldrich, I7125), 97% doxycycline hyclate (Thermo Fisher Scientific, 446060250), and roscovitine (Selleck Chemicals, S1153).

siRNAs and plasmids

The siRNAs used in this study include the following: CUL1 (Dharmacon, L-004086-00-0005), CUL4A (Dharmacon, L-012610), CUL4B (Dharmacon, L-017965), CDT2 (Dharmacon, L-020543-00-0005), nontargeting (NT) (Dharmacon, D-001810-01-20), SMARCAL1 (Dharmacon, J-013058-06-0005), ZRANB3 (Dharmacon, D-010025-03-0005), HLTF (Dharmacon, L-006448-00-0005), BRCA2 (QIAGEN, SI02653434), CCDC71/PRIMPOL (Dharmacon, L-019360-02-0005), CDT1 (Dharmacon, L-003248-00-0005), ATAD5 #1 (Dharmacon, custom 5′-GUAUAUUUCUCGAUGUACA-3′), and ATAD5 #2 [pool of J-004738-05-0002, J-004738-06-0002, J-004738-07-0002, and J-004738-08-0002 (all from Dharmacon]. ATAD5#1 was used unless otherwise indicated. The plasmids used in this study include pInducer20-Blast-CDT1-HA (Addgene, plasmid no. 109335), pCW-Cdt1-mCherry-Puro (Addgene, plasmid no. 193763), and pCW-Cdt1ΔPIP-mCherry-Puro (Addgene, plasmid no. 193765).

Quantification and statistical analyses

All statistical analyses were completed using GraphPad Prism 10. For pairwise comparisons within the representative experiment, including DNA fiber assays, the Kruskal-Wallis one-way analysis of variance (ANOVA) test was used followed by Dunn’s multiple comparisons test to calculate the P value. For pairwise comparisons using multiple biological repeats, including DNA fiber assays, the repeated-measures one-way ANOVA test was used with Holm-Sidak correction for multiple comparisons to calculate the P value. For DNA fiber assay data with pooled repeats, individual data points (gray), mean values of independent experiments (black symbols) and mean of the means (red bar) are shown. Horizontal lines in scatter dot plots represent the mean ± SEM in representative experiments and mean ± SD in experiments with pooled repeats. Box and whisker plots represent the 10th to 90th percentiles with horizontal line depicting the median value. Bar graphs represent the mean ± SD from at least three repeats. Significance threshold for all experiments uses a P value of 0.05 as the cutoff. A P value of n.s. is nonsignificant. The number of biological replicates is represented by “n” in the figure legends. Details regarding the sample size and the number of biological replicates for DNA fiber assays are included in the respective figure legends. At least 150 nuclei were used for measuring nuclear intensities, and at least 100 cells were used to quantify tail moments in comet assays. Data points in IF experiments were randomly chosen to avoid selection bias. Sample details were blinded for at least one repeat.

Supplementary Materials

This PDF file includes:

Figs. S1 to S4