The Deubiquitinase USP9X Maintains DNA Replication Fork Stability and DNA Damage Checkpoint Responses by Regulating CLASPIN during S-Phase

The timely and precise duplication of DNA in S-phase of the cell cycle is critical to maintaining genome stability and preventing tumorigenesis (1, 2). Indeed, genome duplication is a tightly coordinated process where mechanisms that regulate replication origin activation lead to the assembly of new replication forks in a defined spatiotemporal program (3). During S-phase, cells are extremely vulnerable to exogenous and endogenous sources of DNA damage that impede the progression of replication forks thus causing replication stress. Importantly, cellular surveillance mechanisms that overlap with the DNA damage response pathways detect, stabilize, and resolve stalled replication forks to help preserve genome stability (2).

CLASPIN, critical for both DNA synthesis and for signaling the presence of replication stress, is a ring-shaped protein that, together with the TIPIN–TIM1 complex, physically links DNA polymerases and DNA helicase (4–6); this is important for stabilization of replication forks, both during normal replication and upon prolonged arrest (1, 2, 7). CLASPIN also plays an important role in the replication stress response pathway (3, 8, 9), as ATR activation of CHK1 is favored by the binding of CHK1 to CLASPIN (2, 10–13).

Several kinases have been reported to phosphorylate CLASPIN and promote CHK1 activation, including CHK1 itself (4–6, 14), casein kinase 1 gamma 1 (CK1γ1; ref. 15), and CDC7 (16, 17). Phosphorylation of CLASPIN at its N-terminus by PLK1 can also contribute to switching off CHK1 once the damage is repaired, thus allowing cells to progress in the cell cycle (18, 19). Mechanistically, PLK1 phosphorylation of CLASPIN creates a phospho-degron domain recognized by the SCF-βTrCP ubiquitin ligase, leading to proteasome degradation (18, 19).

Ubiquitinylation regulates CLASPIN levels not only during checkpoint recovery but also during other phases of the cell cycle. For example in G₁, CLASPIN degradation is promoted by the APC/C Cdh1-mediated K48-linked polyubiquitinylation (20). Several deubiquitinylating enzymes (DUB) have been shown to counteract proteosomal degradation of CLASPIN: USP28 was reported to reverse the APC/C Cdh1-dependent degradation (20, 21), whereas USP7 and more recently USP29 have been shown to stabilize CLASPIN degradation mediated by SCF-βTrCP (22, 23). Two concurrent reports have recently identified USP20 as a fourth DUB, which regulates CLASPIN stability with a particular relevance during replication stress (24, 25). USP20 itself is regulated by ATR-dependent phosphorylation, resulting in its disassociation from the E3 ubiquitin ligase HERC2 and in its stabilization (24, 25).

The regulation of the DNA damage response by ubiquitinylation and deubiquitinylation is well established, with many DUBs involved in genome stability (26) often regulating multiple proteins involved in the same pathway. For example, USP7 not only regulates CLASPIN levels, but also directly deubiquitinylates CHK1 and RNF168 (27, 28), whereas USP28 controls the levels of 53BP1 and CHK2 (20).

We have identified USP9X as a novel DUB that binds CLASPIN and regulates its levels. USP9X is one of the largest ubiquitin-specific proteases (USP) and it has been implicated in a number of essential cellular processes, and knockout of this gene is embryonic lethal in mice (29). USP9X is involved in the regulation of cell adhesion molecules such as E-cadherin, chromosome segregation, NOTCH, and TGFβ signaling as well as apoptosis (comprehensively reviewed in ref. 30).

In different cancers, USP9X can act as an oncogene or as a tumor suppressor gene in a context-dependent manner: USP9X is overexpressed in multiple carcinomas including lymphomas, non–small cell lung and breast cancer (30). The best described oncogenic function of USP9X derives from its role in promoting the deubiquitinylation and stabilization of antiapoptotic protein MCL1. Furthermore, elevated expression of MCL1 is found in numerous cancers and is associated with resistance to chemotherapy (31).

On the contrary, in pancreatic ductal adenocarcinoma, USP9X acts as tumor suppressor with low expression associated with poor patient outcome and widespread metastasis (32). However, little is known at the molecular level about USP9X tumor suppression function. The ubiquitin ligase ITCH is a USP9X substrate and may be involved in this role, as ITCH was partially responsible for the ability of USP9X to promote anoikis and suppress colony formation (32). Also in ERα-positive breast cancers, USP9X downregulation can promote resistance to tamoxifen, as ERα is stabilized on chromatin, driving the activation of ERα-responsive genes (33).

In this work, we uncover a novel role of USP9X in the maintenance of genome stability. We show that USP9X interacts with CLASPIN and that its downregulation destabilizes CLASPIN in an S-phase–specific manner. We also demonstrate that USP9X depletion impacts CLASPIN function in both replication fork dynamics and the DNA damage response, resulting in the accumulation of DNA damage.

U2OS osteosarcoma cells were obtained from Noel Lowndes laboratory (National University of Ireland Galway, Galway, Ireland) in 2011 and were first authenticated by short tandem repeat analysis and subsequently certified in 2015 by transposon profiling. Flp-In T-REx 293 cells (purchased from Life Technologies and validated by ATCC in 2011) conditionally expressing CLASPIN tagged with a dual C-terminal Flag/OneSTrEP tag were generated in this study (Supplementary Materials and Methods). Cells were cultured at 37°C, 5% CO₂ in DMEM supplemented with 1% penicillin–streptomycin and heat-inactivated 10% FBS (Sigma-Aldrich).

Plasmids expressing full-length USP9X or a catalytically inactive mutant (C1566A) were described previously (34) and obtained from MRC PPU Reagents - University of Dundee (https://mrcppureagents.dundee.ac.uk). Transfections were performed using jetPRIME (Polyplus) according to the manufacturer's instructions.

Doxycycline (Sigma-Aldrich) was used at 1 μg/mL to induce the expression of CLASPIN, hydroxyurea (HU; Sigma-Aldrich) was used at 2 mmol/L, MG132 (Sigma-Aldrich) at 10 μmol/L, nocodazole (Noc; Sigma-Aldrich) at 1 mmol/L, cyclohexamide (Chx; Sigma-Aldrich) at 10 1 mmol/L, and WP1130 (SelleckChem) at 5 μmol/L. For experiments using UV, cells were irradiated with 20 J/m² using a UV lamp (UVItec).

siRNAs were previously described and purchased from Sigma-Aldrich: siUSP9X(i) - AGAAAUCGCUGGUAUAAAUUU (35), siUSP9X(ii) - GCAGUGAGUGGCUGGAAGUTT (36), siClaspin - GCACAUACAUGAUAAAGAA (37), siHERC2 - GCGGAAGCCUCAUUAGAAA (25), siUSP20 - CCAUAGGAGAGGUGACCAA (25), siUSP7 - ACCCUUGGACAAUAUUCCU (38). As control, the Ambion negative control #1 (Ambion) was used. Cells were transfected with 100 nmol/L of all siRNAs for 48 hours using jetPRIME (Polyplus).

Whole-cell extracts were prepared in buffer A (50 mmol/L Tris-HCl, pH 7.4, 300 mmol/L NaCl, 1 mmol/L EDTA, 1% Triton-X) containing protease and phosphatase inhibitors (Sigma-Aldrich) For immunoblotting, proteins were resolved by SDS-PAGE and transferred onto nitrocellulose membranes prior to overnight incubation with primary antibodies and infrared-labeled secondary antibodies. Immunoreactive bands were visualized and quantified using Odyssey Infrared Imaging Systems (Li-COR Biosciences).

For immunoprecipitation experiments, precleared protein extracts were incubated with antibodies that were prebound to protein A/G beads (Santa Cruz Biotechnology). For Strep-affinity purifications, precleared extracts were incubated with Strep-Tactin Sepharose resin (IBA). Following extensive washes in buffer A, bound proteins were recovered by eluting in buffer A containing 10 mmol/L biotin (Sigma-Aldrich) or by boiling the beads in Laemmli buffer.

Empty vector and Strep-CLASPIN expressing cells were grown for 5 passages in medium (R6K4) or light (R0K0) SILAC medium (Dundee Cell Products, catalog no. LM016 and LM014), respectively. Protein extracts were prepared as above and loaded onto individually preequilibrated Strep-Tactin Sepharose columns (IBA). Proteins were recovered by biotin competition, mixed at 1:1 ratio, concentrated on an Amicon centrifugal filter (10 kDa cutoff), and resolved on a 4% to 12% precast gel (Bio-Rad). Gels were stained by Coomassie Colloidal Blue (Sigma-Aldrich) and lanes were cut into 10 slices, each of which was subject to reduction, alkylation, and trypsin digestion as described previously (39). Peptides were separated and analyzed on an UHPLC (Easy-nLC 1000, Proxeon) connected to a QExactive mass spectrometer (Thermo Fisher Scientific). Detailed protocol is available in Supplementary Materials and Methods.

For the analysis of DNA damage, cells were fixed with 1% formaldehyde/PBS and permeabilized with PBS containing 0.05% saponin and 1% BSA. Cells were then stained with primary rabbit anti-pSer139 Histone H2A.X antibody (Cell Signaling Technology, catalog no. 9718s) and mouse anti-Strep antibody (Qiagen, catalog no. 34850) followed by incubation with secondary Alexa Fluor 647 donkey anti-rabbit antibody (Invitrogen) and Alexa Fluor 488 goat anti-mouse antibody (Invitrogen), respectively. DNA was stained with 1 μg/mL DAPI (Sigma-Aldrich). Data were acquired on a BD FACSCanto II cytometer (BD Biosciences) and analyzed with FlowJo software v10. Statistical analysis of three independent experiments was performed using Prism (GraphPad Software).

Cells were fixed with 4% paraformaldehyde and then permeabilized with PBS-TX (PBS 0.1% Triton X-100). Primary rabbit anti-pSer139 Histone H2A.X antibody (Cell Signaling Technology, catalog no. 9718s) was used together with mouse anti-Strep antibody (Qiagen, catalog no. 34850). Secondary antibodies were Alexa Fluor 546 goat anti-rabbit antibody and Alexa Fluor 488 goat anti-mouse antibody (Invitrogen). Cells were examined using an IX71 Olympus microscope with a 40× oil immersion objective.

To obtain further insights into CLASPIN function and regulation, we aimed to identify CLASPIN-interacting proteins. To this end, we generated a Flp-In T-Rex HEK-293 cell line in which a tagged-CLASPIN (Strep-CLASPIN) fusion protein is conditionally expressed from a TET-controlled promoter upon the addition of doxycycline to the culture medium. The levels of CLASPIN in these cells increased by approximately 5-fold following induction with doxycycline and the fusion protein was easily detected with either anti-Strep or anti-CLASPIN antibodies (Fig. 1A). Strep-CLASPIN did not affect cell proliferation for at least 3 days, nor did it affect the distribution of cells within the different phases of the cell cycle. Furthermore, Strep-CLASPIN protein levels fluctuated during the cell cycle with maximum expression during the S-phase (Supplementary Fig. S1).

By exploiting the Strep–tag at the C-terminus of the protein, we devised an affinity purification protocol to enrich for Strep-CLASPIN and potential interacting proteins. After 24 hours of induction with doxycycline, empty vector or Strep-CLASPIN cells were lysed in a 300 mmol/L NaCl containing buffer, as this is sufficient to extract most of the protein from chromatin (Supplementary Fig. S2). Protein extracts were loaded onto Strep-Tactin columns and, after extensive washing, bound proteins were eluted by addition of biotin. By immunoblotting, we confirmed that Strep-CLASPIN protein was highly enriched in the partially purified fractions, and that a known CLASPIN-interacting protein, CDC45, was also specifically found only in the fractions obtained from the Strep-CLASPIN cells (Fig. 1B).

Notably, by staining these fractions with silver, we observed that many more proteins were present in the eluate derived from Strep-CLASPIN cells compared with the control, suggesting that several different proteins could potentially bind to CLASPIN (Fig. 1C). To determine the identity of these proteins, we used a SILAC proteomic approach, where empty vector and Strep-CLASPIN cells were grown on media containing unlabeled or labeled amino acids. Protein extracts were prepared as above and loaded onto individually preequilibrated Strep-Tactin columns prior to washing and biotin-mediated elution. Eluates from empty vector and Strep-CLASPIN affinity purifications were mixed 1:1, concentrated, and proteins were resolved using SDS-PAGE prior to in-gel digestion and quantitative mass spectrometry analysis. We were able to identify and quantify 334 proteins, excluding potential contaminants and reverse hits; these show a distribution markedly skewed toward the bait, indicating that the negative control is relatively clean and that most of the proteins are potential CLASPIN interactors. Indeed, together with the bait Strep-CLASPIN, we found that several known CLASPIN interactors were detected, including MCM2-3-6 subunits of the MCM complex. Notably, among the enriched proteins, we observed a large number of proteasome subunits and enzymes involved in ubiquitin-mediated proteolysis, including the deubiquitinase USP9X (Supplementary Table S1).

The identification of USP9X was particularly intriguing as several deubiquitinases have recently been shown to be involved in regulating CLASPIN including USP28, USP7, USP20, and USP29 (20, 22–25).

To assess whether USP9X interacts with CLASPIN, we first repeated the affinity purification of Strep-CLASPIN protein on a smaller scale, and analyzed the eluted proteins by immunoblotting. Indeed, USP9X was specifically detected in the fractions derived from Strep-CLASPIN–expressing cells and not in the control (Fig. 2A). By performing reciprocal immunoprecipitation experiments using USP9X antibodies or IgG control and extracts from Strep-CLASPIN–expressing cells, we detected an anti-Strep immunoreactive band in the USP9X immunoprecipitated material (Fig. 2B). More importantly, by immunoprecipitating USP9X from extracts prepared from U2OS cells, we recovered endogenous CLASPIN (Fig. 2C). Altogether, these data indicate that USP9X, either directly or indirectly, associates with both endogenous and ectopically expressed CLASPIN.

To begin exploring the relationship between USP9X and CLASPIN, we performed siRNA-mediated depletion of USP9X in HEK-293 cells using two different siRNAs targeting USP9X mRNA, that have been extensively validated in several studies (35, 36). We observed that both USP9X siRNAs efficiently reduce USP9X protein levels when measured 48 hours posttransfection (Fig. 3A). Interestingly, depletion of USP9X also correlated with an obvious decrease of CLASPIN, although the depletion was not as pronounced as by a specific siRNA that directly targets CLASPIN mRNA (Fig. 3A). The effect of USP9X depletion on CLASPIN levels was not only found in HEK-293 cells but also observed in U2OS cells (Fig. 3B). Growth and distribution of cells within the cell cycle was not grossly affected by USP9X depletion in either cell line (Supplementary Fig. S3).

These observations suggested the possibility that USP9X, being an ubiquitin-specific peptidase, may regulate CLASPIN levels by affecting its turnover, possibly through the proteasome. If this is the case, then proteasome inhibition would be expected to rescue CLASPIN levels and indeed upon treatment with MG132, CLASPIN levels were stable in the presence of USP9X depletion, in a manner similar to p21, a CDK inhibitor and well-known substrate of the proteasome (Fig. 3C; ref. 40). The level of USP9X itself was increased by MG132 treatment in cells transfected with control siRNA but not in cells transfected with USP9X-targeting siRNA. Thus, the decrease of CLASPIN observed upon USP9X depletion is most likely caused by accelerated proteasome-mediated degradation. To test this hypothesis, U2OS cells were transfected with either control or USP9X-targeting siRNA. Cycloheximide was added to prevent further protein synthesis, samples were harvested at different time points, and protein levels analyzed by semiquantitative immunoblotting. Although in USP9X-depleted cells, CLASPIN levels were already reduced at the beginning of the experiments, its level rapidly decreased to become almost undetectable after 6 hours, whereas in control cells, CLASPIN levels decline with a slower kinetics (Fig. 3D and Supplementary Fig. S4).

As multiple mechanisms have been shown to control CLASPIN degradation in different phases of the cell cycle and to obtain better insights into the role of USP9X in CLASPIN stability, we used the Strep-CLASPIN cells. These were grown in doxycycline-containing medium, arrested in either S-phase with HU or in mitosis with nocodazole, and the rate of degradation of Strep-CLASPIN protein analyzed as above. First, we observed that the Strep-CLASPIN protein, similarly to endogenous CLASPIN is stabilized in S-phase compared with mitosis (Fig. 3E) and second, we found that the treatment with the small-molecule WP1130, a deubiquitinase inhibitor that shows some specificity for USP9X (41), destabilizes the protein only in S-phase cells. We then looked at ubiquitinylation levels of the Strep-CLASPIN protein in S-phase. For these experiments, Strep-CLASPIN–expressing cells were transfected with a construct expressing HA-tagged ubiquitin, arrested in S-phase by HU, and treated with WP1130 in the presence of MG132; ubiquitinylation was then assessed by Strep-affinity purification and anti-HA immunoblotting. Figure 3F shows that in untreated cells, the levels of CLASPIN ubiquitinylation are hardly detectable over background (Fig. 3F, lane 1–2). As expected, MG132 caused an increase in ubiquitinylated CLASPIN and addition of WP1130 further enhanced accumulation of ubiquitinylated forms (Fig. 3F, lanes 4 and 5), suggesting that USP9X could contribute to the removal of ubiquitin chains from CLASPIN. The role of USP9X in protecting CLASPIN from proteasome-dependent degradation was then further assessed by overexpressing either wild type or a catalytically dead version of USP9X carrying a single amino acid change in its active site (C1566A; ref. 34) in U2OS cells. Twenty-four hours after transfection, cycloheximide was added for 6 hours to the culture and the residual levels of CLASPIN assessed by Western blotting. Figure 3G and Supplementary Fig. S5 show that overexpression of USP9X indeed prevents the CLASPIN degradation, whereas the C1566A-mutant proteins fails to stabilize CLASPIN, indicating that the catalytic activity of USP9X is required.

To gain insight into the mechanism by which USP9X protects CLASPIN, we first explored the relationship between USP9X and other DUBs that contribute to CLASPIN stabilization during S-phase, namely, USP29, USP7, and USP20. U2OS cells were transfected with siRNA targeting these DUBs and protein levels assessed by immunoblotting 48 hours later. As USP20 itself is stabilized by the HERC2 ubiquitin ligase, a HERC2-targeting siRNA was also used in these experiments. First, we found that depletion of USP9X did not affect the levels of any of these DUBs (Fig. 4). We then asked the reciprocal question of whether depletion of USP7, USP20, and USP29 regulates USP9X abundance. We observed that neither USP7 nor USP20 depletion affected the levels of USP9X (Fig. 4, lanes 3 and 5). Similarly HERC2 depletion did not change USP9X levels although, as reported previously (25), promoted USP20 stabilization (Fig. 4). In these experiments, we also noticed that USP20 depletion correlated with a decrease in USP29, suggesting a functional cross-talk between the two DUBs (Fig. 4). Our repeated attempts to downregulate USP29 by siRNAs did not result in a convincing depletion of the protein, thus we could not assess whether USP29 directly affects USP9X levels.

Altogether these data suggest that USP9X may act in an independent manner from the previously reported USPs that control CLASPIN stability in S-phase.

The finding that USP9X controls CLASPIN levels in S-phase prompted us to test the hypothesis that DNA replication dynamics could be affected. In particular, we investigated whether USP9X depletion would affect ongoing replication forks and origin firing using a DNA fiber labeling technique (42–44). U2OS cells were transfected with control, USP9X, or CLASPIN siRNAs, and after 48 hours nascent DNA was sequentially labeled for 30 minutes with the nucleotide analogue iododeoxyuridine (IdUrd) followed by washing for 30 minutes with chloro-deoxyuridine (CldU). Cells were then lysed, DNA fibers prepared, and nascent DNA analyzed by fluorescence microscopy, with IdUrd-containing DNA visualized with TRITC-conjugated antibodies (red) and CldU-containing DNA with FITC-conjugated antibodies (green). Strikingly, we observed that in the USP9X-depleted cells, the percentage of termination events and stalled forks, scored as red only tracks, was greatly increased compared with siRNA control–transfected cells and was at a similar level to that seen by CLASPIN depletion (Fig. 5A). Interestingly, and similar to what has already been reported for CLASPIN depletion (7), the amount of DNA synthesized during the second labeling period was severely reduced as observed by short green tracks in the USP9X-depleted cells. In our experiments, neither USP9X nor CLASPIN depletion appeared to have a significant impact on the origin firing (green only tracks).

Using a different labeling strategy, we asked whether USP9X depletion could affect replication fork restart after forks have been blocked by treating cells with ribonucleotide inhibitor HU. In these experiments, cells were first labeled for 30 minutes with IdUrd, then IdUrd was removed and HU was added, and after two further hours, the drug was washed away and CldU was added to label replication products. Again, compared with controls, in both USP9X- and CLASPIN-depleted cells, we observed an increase in the number of red only tracks indicative of replication forks that collapsed or were unable to restart after HU removal (Fig. 5B).

The role of CLASPIN in mediating ATR-dependent phosphorylation of CHK1 in the response to DNA replication blockade or DNA damage has been widely documented, thus we predicted that lower CLASPIN levels may also impact on ATR signaling. To verify this hypothesis, U2OS cells transfected with either control or USP9X-specific siRNAs were treated with HU, samples were collected throughout a 2-hour time course experiment, and protein analyzed. Figure 6A and Supplementary Fig. S6A show that CHK1 phosphorylation at Ser317 is severely reduced in USP9X-depleted cells compared with the control. Low CLASPIN levels and defective phosphorylation of CHK1 upon HU treatment were also observed in USP9X-depleted hTERT-immortalized retinal pigment epithelial RPE1 cells (Supplementary Fig. S6B). To assess whether defective checkpoint signaling in USP9X-depleted cells was not limited to HU, but a more general response to replication stress, U2OS cells were irradiated with 20 J/m² of UV light and samples collected 30 minutes and 2 hours after irradiation. Figure 6B shows that again the levels of CHK1 phosphorylation at Ser317 were reduced. We then assessed the effects of prolonged replication stress in USP9X-depleted cells using a colony-forming assay. In these experiments after transfection with control and USP9X siRNAs, U2OS cells were incubated for 18 hours with increasing amounts of HU, plated and the number of colonies scored after 10 days. We noticed that in this cell line, USP9X depletion caused a substantial decrease in plating efficiency without any added replication stress; however, upon HU treatment, the decrease in colony formation is further enhanced in USP9X-depleted cells when compared with control cells at all doses tested (Fig. 6B). The loss of clonogenic potential upon depletion of USP9X could indicate the inherent problems with ongoing replication upon depletion of USP9X in a normal cell cycle, but clearly the decrease in survival observed upon HU treatment reveals that USP9X plays an important role in dealing with replication stress.

The observations that USP9X depletion affects replication fork stability and that USP9X contributes to efficient induction of the DNA replication checkpoint, suggested that depletion of USP9X would lead to accumulation of DNA damage. As expected, a large number of nuclei of USP9X-depleted cells stained positive for phosphorylated H2AX at serine 139 (γ-H2AX). The γ-H2AX staining pattern was very heterogeneous, with some cells showing only punctuate staining and others with very bright pan nuclear staining, these features are reminiscent of the phenotype we observe upon CLASPIN depletion (Fig. 7A and Supplementary Fig. S6C; ref. 45). DNA damage upon USP9X depletion was observed in a different cell line and with two independent USP9X-targeting siRNAs (Supplementary Fig. S6C).

Importantly, the addition of doxycycline, causing ectopic expression of Strep-CLASPIN protein, significantly reduced the percentage of cells with damaged DNA (Fig. 7A and B).

Altogether, these data show that reduced levels of CLASPIN, as a result of USP9X depletion, are among the primary contributing factors causing DNA damage.

In this work, we provide evidence that USP9X is critical for efficient DNA replication. USP9X depletion causes spontaneous replication fork stalling and increased sensitivity to HU. We also show that USP9X controls CLASPIN levels, most likely through counteracting its ubiquitinylation in S-phase. Strikingly, the effect of USP9X depletion on DNA replication is almost indistinguishable from the direct downregulation of CLASPIN itself, suggesting that CLASPIN is the main target of USP9X in the process of DNA replication. This idea is further reinforced by the fact that the amount of DNA damage in USP9X-depleted cells, likely accumulating because of fork instability, is partially rescued by ectopic expression of CLASPIN.

Our experiments demonstrate that USP9X depletion decreases levels of endogenous CLASPIN in a proteasome-dependent manner and increases the rate of CLASPIN degradation. Furthermore, we find that the two proteins can be coimmunoprecipitated, and that CLASPIN stabilization requires the catalytic activity of USP9X, suggesting that CLASPIN is very likely a bona fide in vivo substrate of USP9X. USP9X is mostly a cytoplasmic and membrane-associated protein (30); however, a small fraction has been reported to be nuclear (46). We reconfirmed these findings by immunofluorescence experiments and we found that some USP9X can accumulate in the nucleus of U2OS cells when nuclear export is inhibited (Supplementary Figs. S7 and S8). It is tempting to speculate that some USP9X activity may play a direct role at replication forks and in agreement with this idea a proteomic study reported that USP9X, such as CLASPIN, is preferentially found associated with nascent chromatin compared with mature postreplicative chromatin (47).

It is intriguing that four different deubiquitinases, USP7, USP29, USP20, and USP9X target the same protein in a nonredundant manner. As depletion of USP9X does not alter the protein levels of any of these DUBs and vice versa, it is possible that each DUB may be involved in limiting CLASPIN ubiquitinylation at different specific sites on the protein thus allowing multiple layers of regulation to be independently imposed on CLASPIN.

Consistent with a dual role for CLASPIN at replication forks and as a mediator of checkpoint signaling, USP9X depletion decreases, but does not abolish CHK1 activation in response to HU and UV and leads to hypersensitivity to drug-induced replication stress. Increased DNA damage and a diminished ability to promote efficient checkpoint signaling may also be the reason for the hypersensitivity to other genotoxic agents such 5-fluorouracil in USP9X-deficient colorectal cancer cells (48).

Inhibition of USP9X either as single agent or in combination with other drugs has been suggested as therapeutic strategy for several leukemias and solid tumors, however, most of the studies focused on the role of USP9X in the stabilization of the antiapoptotic protein MCL1 (35) and on the increased rate of apoptosis. In our study, we did not observe an obvious increase in apoptotic cell death by siRNA-mediated USP9X depletion and the additive effects of USP9X depletion with HU were only revealed with clonogenic assays that have long-term endpoints. This highlights the possibility that excessive genome instability may be the main reason of the loss of the proliferative capability in our cellular models.

Interestingly, in pancreatic cancer USP9X acts as a tumor suppressor by limiting K-RAS induced cellular transformation and by suppressing tumor aggressiveness by a mechanism that still requires further understanding (32). It is tempting to speculate that the protection of replication forks and protection from DNA damage could be an important factor in the tumor suppressor role of USP9X.

In conclusion, we propose that USP9X, by regulating CLASPIN, is a novel player in the replication of the genome, in the DNA damage/replication checkpoint and its function is important for the maintenance of genomic stability. Further studies in systems that more closely recapitulate human cancers, both in vitro and in vivo, will be required to understand how this novel function of USP9X contributes to either cancer development or to the outcome of treatments targeting the DNA replication process and/or the integrity of DNA.

No potential conflicts of interest were disclosed.

Conception and design: E. McGarry, D. Gaboriau, M.D. Rainey, A. Bachi, C. Santocanale

Development of methodology: E. McGarry, D. Gaboriau, M.D. Rainey, A. Bachi

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): E. McGarry, D. Gaboriau, U. Restuccia

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): E. McGarry, D. Gaboriau, U. Restuccia, A. Bachi

Writing, review, and/or revision of the manuscript: E. McGarry, D. Gaboriau, U. Restuccia, A. Bachi, C. Santocanale

Study supervision: C. Santocanale

The authors thank Sandra Healy, Bob Lahue, and Brian McStay for critical reading of manuscript, Kevin Wu, Karolina Kujawa, Aisling O'Connor, and Grainne Donnellan for technical assistance and all the members of the Santocanale laboratory for discussion and support. The authors also thank Simona Polo for discussion and reagents, Ciaran Morrison for help with microscopy, and the NCBES flow cytometry facility that is funded by NUIG, the Irish Government's PRTLI4-5 and NDP 2007–2013. Confocal microscopy was performed in the Facility for Imaging by Light Microscopy (FILM) at Imperial College London.

This work was supported by Science Foundation Ireland (SFI) grant 12/IP/1508 to C. Santocanale and by Breast Cancer Now grant (2012 May PR089). E. McGarry was supported by Molecular Medicine Ireland with a CTRSP Scholarship.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.