Skp1-Cul1-F-box Ubiquitin Ligase (SCFβTrCP)-mediated Destruction of the Ubiquitin-specific Protease USP37 during G2-phase Promotes Mitotic Entry*

Background: USP37 regulates S-phase progression and is degraded in late M/G1. Results: USP37 undergoes biphasic degradation during G2 and late M/G1. Conclusion: SCFβTrCP and APCCdh1 coordinately regulate USP37 during the cell cycle. Significance: Precise regulation of USP37 activity is required for cell cycle progression. Ubiquitin-mediated proteolysis is a key regulatory process in cell cycle progression. The Skp1-Cul1-F-box (SCF) and anaphase-promoting complex (APC) ubiquitin ligases target numerous components of the cell cycle machinery for destruction. Throughout the cell cycle, these ligases cooperate to maintain precise levels of key regulatory proteins, and indirectly, each other. Recently, we have identified the deubiquitinase USP37 as a regulator of the cell cycle. USP37 expression is cell cycle-regulated, being expressed in late G1 and ubiquitinated by APCCdh1 in early G1. Here we report that in addition to destruction at G1, a major fraction of USP37 is degraded at the G2/M transition, prior to APC substrates and similar to SCFβTrCP substrates. Consistent with this hypothesis, USP37 interacts with components of the SCF in a βTrCP-dependent manner. Interaction with βTrCP and subsequent degradation is phosphorylation-dependent and is mediated by the Polo-like kinase (Plk1). USP37 is stabilized in G2 by depletion of βTrCP as well as chemical or genetic manipulation of Plk1. Similarly, mutation of the phospho-sites abolishes βTrCP binding and renders USP37 resistant to Plk1 activity. Expression of this mutant hinders the G2/M transition. Our data demonstrate that tight regulation of USP37 levels is required for proper cell cycle progression.

The APC recognizes substrates (e.g. cyclins, securin, Geminin) containing one or more destruction-targeting motifs (degrons), primarily the destruction box (D-box) RXXL (1,2), and the KEN box (3). The ability of the APC to recognize these degrons is conferred, at least in part, by the adaptor/activator proteins Cdc20 and Cdh1 (4). Rigid control of the APC is achieved by a variety of mechanisms. The expression of Cdc20 and Cdh1 as well as their interactions with the APC are cell cycle-dependent with APC Cdh1 active primarily in late mitosis through G 1 and APC Cdc20 active during mitosis (5,6). The APC exhibits autonomous regulation by targeting its activators and its cognate E2s for destruction (7,8). In addition, there are also a number of direct inhibitors of the APC. The bulk of APC Cdh1 activity is kept in check from G 1 -M by the inhibitor Emi1.
Similar to the APC, SCF ligases are denoted by the substrateadapting F-box protein. SCF Skp2 and SCF ␤TrCP have prominent roles in the cell cycle (9). In contrast to the APC, SCF complex activity toward substrates is largely mediated by the cell cycle-or stimulus-dependent phosphorylation of substrates. SCF ␤TrCP , for example, recognizes phosphorylated serines in the DSGXXS motif in its substrates (e.g. Emi1, Wee1, Claspin, Cdc25A) (9 -14). Intriguingly, many of these substrates are phosphorylated by the APC substrate Plk1 (12, 14 -16). There is also significant cross-talk between the APC and SCF ligases. For example, the ligases act in tandem to regulate the levels of a number of critical cell cycle regulators, including Cdc25A and Claspin (9,10,12,(17)(18)(19). In addition, APC Cdh1 controls SCF activity by targeting Skp2, whereas SCF ␤TrCP regulates APC activity by targeting Emi1 for destruction (11,13,20,21).
Given the critical function of ubiquitin in control of the cell cycle, deubiquitinating enzymes are expected to play central roles as well. Indeed, several deubiquitinating enzymes have been implicated in the cell cycle. Recently, we have identified the deubiquitinating enzyme USP37 as a regulator of the G 1 /S transition (22). USP37 regulates S-phase entry at least in part by enhancing cyclin A stability and accumulation (22). Consistent with its regulation of this key cell cycle transition, USP37 is required for zebrafish development (23). USP37 is regulated by the oncogenic transcription factor E2F1, and its expression is increased in several cancers (24 -28). Increased USP37 expression is associated with poor prognosis in non-small cell lung cancer (29). The phenotypes associated with aberrant USP37 activity are unlikely to be explained by its effect on cyclin A and have prompted us to explore USP37 biology further.
Here we report that destruction of USP37 is biphasic. USP37 is destroyed in G 2 by the concerted actions of Plk1 and SCF ␤TrCP , whereas APC Cdh1 targets the remaining pool at mitotic exit. By expressing UPS37 mutants that are resistant to SCF ␤TrCP -mediated ubiquitination, we demonstrate that destruction of this pool is required for mitotic entry. Importantly, this destruction event highlights the existence of additional substrates whose destruction is required for the G 2 /M transition.

Cell Culture
HeLa, 293T, U2OS, and T98G cells were obtained from ATCC and maintained in DMEM supplemented with 10% FBS. HeLa and 293T cells were synchronized as described (30). RO-3306 or nocodazole were added 5 h after release from thymidine. T98G cells were synchronized by incubation in DMEM without FBS for 72 h and stimulated to re-enter the cell cycle by the addition of 20% FBS. Cells were transfected with TransIT-LT1 (Mirus Bio) or RNAiMAX (Invitrogen) per the manufacturer's instructions. Where indicated, cells were treated with 100 ng/ml nocodazole, 10 M RO-3306 (EMD Millipore), 200 nM BI2536, and 10 M MG132 (Boston Biochem).

Plasmids and Recombinant Proteins
USP37 was subcloned into pDEST-CS2-MYC 6 and pDEST-GEX-6P1 using Gateway technology (Invitrogen). Mutants were generated by the QuikChange mutagenesis strategy. Additional plasmids were described previously (13,16). His 6 -Ubiquitin was generated by PCR and cloned into pCDNA5/FRT/TO (Invitrogen). Recombinant and in vitro translated protein were produced as described (30) except that USP37 was produced in wheat germ rather than rabbit reticulocyte lysate.

Western Blotting and Immunoprecipitation
Cell extracts were generated in EBC buffer (50 mM Tris (pH 8.0), 120 mM NaCl, 1% Nonidet P-40, 1 mM DTT, 25 mM ␤-glycerophosphate, 5 mM NaF, 1 mM NaVO 4 , and leupeptin, pepstatin, and chymotrypsin, each at 10 g/ml. For immunoprecipitation, equal amounts of cell lysates were incubated with the indicated antibodies for 2-12 h and washed in EBC buffer including inhibitors. Immunoprecipitation samples or equal amounts of whole cell lysates were resolved by SDS-PAGE, transferred to PVDF membranes (Millipore) probed with the indicated antibodies, and visualized with the LI-COR Odyssey infrared imaging system.

Ubiquitination
In Vivo-293T cells were transfected with a 1:1:2 ratio of His 6 -Ub:MYC-USP37:HA-␤TrCP between thymidine blocks. 10 M MG132 was added during the last 10 -12 h of culture. Lysates were generated as above, except that 2 mM N-ethylmaleimide was added to inactivate deubiquitinating enzymes and DTT was omitted. Equal amounts of lysates were adjusted to 1.5% SDS and boiled for 10 min. Lysates were cooled to room temperature, and ubiquitinated proteins were purified with Ni 2ϩ -agarose and processed as above.
In Vitro-The in vitro procedure was essentially as described (32) except that FLAG-␤TrCP was used, UbcH3 was the sole E2, and USP37 substrates were translated in vitro.

RESULTS
USP37 Is Targeted by APC Cdh1 in G 1 -Previously, we demonstrated that APC Cdh1 targets USP37 for destruction (22). USP37 is modified by Lys11-linked polyubiquitin chains in late mitosis/G 1 and is destroyed with similar kinetics to other APC Cdh1 substrates (22) (supplemental Fig. S1, A and B). However, the requirement for Cdh1 for destruction of USP37 in G 1 was not determined. We therefore depleted Cdh1 or Cdc20 in thymidine-nocodazole synchronized HeLa cells and followed the kinetics of USP37 destruction (supplemental Fig. S1C). As expected, USP37 levels remained stable in cells depleted of Cdh1, but not Cdc20. To confirm that this was due to direct activity of Cdh1 toward USP37 and not a cell cycle defect, we performed knockdown experiments in T98G cells synchronized in G 0 by serum starvation. Cells were transfected with siRNAs at serum stimulation to prevent additional Cdh1 expression. Consistent with our HeLa cell data, depletion of Cdh1 in T98G cells resulted in premature accumulation of USP37 and cyclin A (supplemental Fig. S1D).
USP37 Is Unstable in G 2 -During the course of the above experiments, we observed that the levels of USP37 in nocodazole-arrested cells were lower than those in thymidine-arrested cells. This observation was surprising in light of the results above and our previous study (22) as APC Cdh1 is thought to be inactive from G 1 /S through anaphase. We therefore examined USP37 levels in HeLa cells synchronized by a double-thymidine block as they progressed from S-phase through early G 1 (Fig.  1A). Cell cycle progression was monitored by flow cytometry (supplemental Fig. S2A). USP37 levels steadily declined as cells progressed through G 2 (6 -8 h) (Fig. 1, A and B, supplemental Fig. S2A). The rate of degradation slowed as cells progressed through mitosis and into G 1 (8 -12 h) (Fig. 1, A and B, supplemental Fig. S2A). We compared the decline of USP37 levels with substrates of APC Cdc20 and APC Cdh1 as well as the SCF ␤TrCP substrate Emi1. Because APC substrates decline ␤TrCP-mediated Destruction of USP37 in two activator-dependent waves, we analyzed USP37 destruction in two phases and set the level of all proteins to 1 at the first time point of each phase (i.e. 6 and 8 h) (Fig. 1, C and D). As cells transited G 2 through the early stages of mitosis, USP37 levels declined by 50% similar to Emi1, whereas the APC Cdc20 substrates remained stable (Fig. 1C, supplemental Fig. S2A). USP37 degradation slowed as APC Cdc20 substrates were degraded and then paralleled the degradation of APC Cdh1 substrates ( Fig. 1, C and D, supplemental Fig. S2A). The apparent stabilization of USP37 during the period of APC Cdc20 substrate destruction is in agreement with the existence of a pool of USP37, which remains stable in nocodazole (Fig. 1G, supplemental Fig. S1, A-C). The timing of the second wave of destruction is consist-ent with the Cdh1-dependent destruction of this mitosis-stable pool of USP37 (supplemental Fig. S1C). We confirmed that this was not an artifact of thymidine synchronization by examining USP37 levels in quiescent T98G cells stimulated to enter the cell cycle by serum addition (Fig. 1E). Flow cytometry analysis indicated that destruction of USP37 begins in G 2 (supplemental Fig. S2B). Similar results were obtained in HeLa cells released from a nocodazole arrest (Fig. 1F). USP37 levels began to decline prior to APC substrates in both of these populations as well (Fig. 1, E and F). Together these data indicate that a pool of USP37 is degraded in G 2 . To test this hypothesis, we examined USP37 levels in T98G cells treated in late S/G 2 with the Cdk1 inhibitor (RO-3306) to prevent mitotic entry (34). Indeed, FIGURE 1. Biphasic destruction of USP37. A, the protein levels of USP37, substrates of APC and SCF ligases, and mitotic markers (phospho-histone H3 Ser-10 (pH3); phospho-pRb (pRb)) were monitored throughout the cell cycle in HeLa cells synchronized by a double thymidine block. B-D, quantitation of protein levels in A (normalized to actin) for the indicated cell cycle phases and hours after release from the thymidine block. To facilitate comparison of USP37 degradation with substrates of both APC Cdc20 and APC Cdh1 , protein levels in C are determined relative to the 6-h time point, whereas in D, they are recalculated and compared relative to the 8-h time point. A. U., arbitrary units. E, T98G cells were synchronized in G 0 by serum starvation and stimulated to re-enter the cell cycle by serum addition. Protein levels were analyzed as in A. F, HeLa cells were arrested in mitosis with a thymidine-nocodazole block. Protein levels were analyzed as in A. G, T98G cells, treated as in E, were blocked in G 2 with RO-3306 (Cdk1i) at 20 h after stimulation and analyzed as in A. H, HeLa cells were synchronized by double thymidine, thymidine-RO-3306, or thymidine-nocodazole blocks. Protein levels were analyzed as in A. Quantification of USP37 protein levels is presented below the immunoblots.
␤TrCP-mediated Destruction of USP37 NOVEMBER 9, 2012 • VOLUME 287 • NUMBER 46 USP37 levels declined dramatically in G 2 cells (Fig. 1G) as confirmed by flow cytometry (supplemental Fig. S2C). We confirmed that USP37 is degraded in G 2 by analyzing USP37 protein levels in cells synchronized at G 1 /S, G 2 , or M. USP37 levels were significantly lower in G 2 and mitotic HeLa cells than at G 1 /S (Fig. 1H). Similar results were obtained with MCF-7 and HCT116 cells (supplemental Fig. S2D). Together these data indicate that USP37 is degraded in a biphasic manner during the G 2 /M and M/G 1 transitions in APC-independent and APC Cdh1 -dependent events, respectively.
USP37 Interacts with the SCF ␤TrCP -The SCF ␤TrCP ligase targets a number of cell cycle regulators, including Emi1, Wee1, Bora, and, Claspin, to promote progression from G 2 to mitosis (11)(12)(13)(14)35). The timing of the decline in USP37 levels mirrored decline in these substrates and suggests that USP37 may be a substrate of SCF ␤TrCP . USP37 has been reported to interact with SCF ␤TrCP (36). We first confirmed that USP37 interacts with SCF ␤TrCP . Examination of HA-tagged ␤TrCP immunocomplexes readily revealed the presence of coexpressed USP37 (see supplemental Fig. S3C). We next determined the ability of epitope-tagged proteins to coprecipitate endogenous interacting proteins. USP37-FLAG immunocomplexes contained endogenous Cul1, ␤TrCP, and Skp1 ( Fig. 2A). Similarly, endogenous USP37 was present in immunocomplexes of tagged Cul1 and ␤TrCP (data not shown). Finally, we determined that USP37 was present in immunoprecipitates of endogenous ␤TrCP (Fig. 2B). From these results, we conclude that USP37 is an SCF ␤TrCP -interacting protein.
USP37 Is Targeted for Destruction by SCF ␤TrCP in G 2 -We next determined whether SCF ␤TrCP participates in the degradation of USP37. We first tested whether perturbing the interaction of ␤TrCP with the SCF altered USP37 stability. Coexpression of USP37 with the dominant-negative ␤TrCP ⌬-F-box mutant, which cannot interact with Skp1, resulted in increased USP37 levels (Fig. 3A). We then tested the ability of ␤TrCP to induce destruction of USP37. Expression of ␤TrCP reduced levels of exogenous USP37 (Fig. 3B). To confirm that ␤TrCP was inducing ubiquitination, we performed an in vivo ubiquitination experiment. The addition of MG132 stabilized USP37 in the presence of ␤TrCP and induced the accumulation of ubiquitinated forms (Fig. 3C, left panels). We confirmed that these were ubiquitin-USP37 conjugates by purifying ubiquitinated proteins from the denatured lysates and probing for MYC-USP37 (Fig. 3C, right panel). Finally, we tested the requirement for ␤TrCP in the degradation of USP37 in G 2 . Serum-starved T98G cells were depleted of ␤TrCP via siRNA oligonucleotides known to target both ␤TrCP1 and ␤TrCP2 (supplemental Fig.  S3A), and USP37 levels were monitored after serum stimulation (35,37). In control siRNA transfected cells, USP37 levels peaked at 20 h and began to drop at 24 h. In contrast, USP37 levels in si␤TrCP transfected cells remained stable through the end of the experiment (32 h) (Fig. 3D). As previously reported, ␤TrCP depletion resulted in the accumulation of cyclins A and B (supplemental Fig. S3B) (11). Together these results indicate that USP37 is a ␤TrCP substrate.
To confirm that USP37 is directly regulated by SCF ␤TrCP , we sought to identify the degron mediating its targeting by ␤TrCP. The consensus ␤TrCP recognition sequence is DSGXXS, where both serine residues are phosphorylated (9). Several variants of this phospho-degron have been identified (Fig. 4B) (9). Examination of the USP37 sequence revealed no consensus degrons, but instead identified multiple sequences resembling noncanonical degrons (Fig. 4, A and B). We generated a series of Nand C-terminal deletion mutants to identify sequences that direct USP37 degradation and tested their ability to bind ␤TrCP (Fig. 4A, supplemental Fig. S3C). The deletion panel was designed to maintain individual structural motifs, termed Boxes 1-6, within the catalytic domain of USP family members that have been identified in structural models of USP37 (38). The pattern of ␤TrCP interaction with these fragments highlighted a unique insertion between Boxes 4 and 5 of the USP37 catalytic domain that contains three ubiquitin-interacting motifs as well as the APC Cdh1 -targeting KEN box (Fig. 4A) (22). The ubiquitin-interacting motif insertion contains three potential ␤TrCP binding sites, (Fig. 4B). An additional deletion mutant that bisects the ubiquitin-interacting motif insertion further highlighted residues 756 -888, which contain two potential binding motifs. We created S 3 A mutants of the likely phospho-sites within these two motifs (Fig. 4B). USP37 S858A, but not the S790A mutant, exhibits weakened binding to ␤TrCP (Fig. 4C). Consistent with direct targeting of USP37 by SCF ␤TrCP , USP37 S858A is resistant to ␤TrCP-driven degradation (Fig. 4D).
Plk1 Triggers the SCF ␤TrCP -mediated Destruction of USP37-The Plk1 kinase is a key regulator of the G 2 /M transition and functions in part by phosphorylating ␤TrCP recognition sites in proteins that must be destroyed for mitotic entry/progression (e.g. Emi1, Wee1) (14,16). We therefore hypothesized that Plk1 might regulate the destruction of USP37 as well. Indeed, Ser-858 is highly conserved and lies within a consensus Plk1 phosphorylation motif, (D/N/E/Y)X(S/T) (supplemental Fig.  S3D) (39). We found that USP37 and Plk1 were able to interact in cells, confirming the likelihood that Plk1 plays a role in UPS37 degradation (Fig. 5A). We then asked whether manipulating Plk1 activity would affect USP37 levels in cells. Expression of Plk1 reduced levels of coexpressed USP37, whereas the dominant-negative Plk1 K82R caused an increase in USP37 lev- ␤TrCP-mediated Destruction of USP37 els relative to control cells (Fig. 5B). We then asked whether Plk1 could stimulate ␤TrCP-mediated destruction of USP37. Indeed, expression of both Plk1 and ␤TrCP resulted in a strong reduction in USP37 levels that was rescued by proteasome inhibition (Fig. 5C). We then asked whether Plk1 activity is required for the destruction of USP37. We transfected serum-starved T98G cells with siRNA targeting Plk1, which prevented the expression of the kinase upon cell cycle entry. In control populations, USP37 exhibited a steady decline after 24 h, as Plk1 levels increased. USP37 levels remained high in the Plk1-depleted populations through 32 h (Fig. 5D). We further confirmed the involvement of Plk1 by inhibiting its function in thymidine-synchronized U2OS cells with the small molecule BI2536 (supplemental Fig. S3E). Consistent with the siRNA results, inhibition of Plk1 prevented the degradation of USP37 as cells approached mitosis.
The ability of Plk1 to regulate USP37 stability suggests that it should modulate the interaction of USP37 with ␤TrCP. We tested this model with an in vitro pulldown assay. Recombinant GST-USP37 was utilized to capture in vitro translated ␤TrCP. As expected, in the absence of phosphorylation, USP37 was unable to interact with ␤TrCP (Fig. 5E, lane 2). Surprisingly, the addition of recombinant Plk1 had little, if any, effect on the ability of USP37 to bind ␤TrCP (Fig. 5E, lanes 3 and 4). However, efficient interaction of Plk1 with its substrates requires a priming phosphorylation event, which is frequently mediated by cyclin-dependent kinases (40,41). Cdk2 in complex with either cyclin A or cyclin E is able to phosphorylate USP37 (22). We reasoned that phosphorylation by these kinases may promote phosphorylation by Plk1 and ␤TrCP binding. Neither cyclin A-Cdk2 nor cyclin E-Cdk2 induced a strong interaction between USP37 and ␤TrCP (Fig. 5E, lanes 5-8). However, Plk1  ␤TrCP-mediated Destruction of USP37 NOVEMBER 9, 2012 • VOLUME 287 • NUMBER 46 in the presence of either cyclin-Cdk2 complex induced a strong interaction between USP37 and ␤TrCP (Fig. 5E, lanes 9 and 10). Importantly, the S858A mutation dramatically reduced the ability of Plk1 to induce ␤TrCP binding (Fig. 5F). We then confirmed that Plk1-mediated, Ser-858-dependent binding to ␤TrCP was required for ubiquitination by the SCF by performing in vitro ubiquitination assays (Fig. 5G). Indeed, ubiquitination of USP37 S858A by SCF ␤TrCP was dramatically reduced in comparison with wild type. Taken together, these results identify Plk1 as a novel USP37-interacting and regulatory protein.
Destruction of USP37 by SCF ␤TrCP in G 2 Is Required for Mitotic Entry-We next sought to confirm that SCF ␤TrCP and Plk1 were mediating destruction of USP37 in G 2 . We asked whether ␤TrCP or Plk1 was required for USP37 destruction in T98G cells stimulated to enter the cell cycle and arrested in G 2 with RO-3306. In line with results from unperturbed cells (Figs. 3D and 5D), depletion of ␤TrCP or Plk1 or treatment with BI2536 stabilized USP37 in G 2 -arrested T98G cells (Fig. 6A). We next determined that phosphorylation of Ser-858 is required for destruction in G 2 . USP37 or the S858A mutant was expressed during the second thymidine block of HeLa Tet-On cells. Expression was shut off by releasing cells into doxycycline-free media, and protein stability was monitored in cells arrested in G 2 as above. Consistent with our in vitro data, the S858A mutant remained stable, whereas the wild type protein was destroyed. Together we interpret these results to confirm that Plk1 and SCF ␤TrCP cooperate to trigger destruction of USP37 in G 2 .
To determine the physiological requirement for destruction of this pool of USP37, we determined the cell cycle profile of asynchronous HeLa cells that were transiently expressing USP37 or USP37-S858A by flow cytometry. Expression of USP37 caused a modest increase in the G 2 /M population, whereas expression of ␤TrCP-resistant USP37 caused a 2-fold increase of the G 2 /M population (Fig. 6, C and D). Immunofluorescence and flow cytometry analyses revealed increased cyclin B1-positive, nonmitotic cells in USP37 S858A-expressing cells, but not the mitotic markers MPM-2 and phosphohistone H3 (Ser-10), consistent with an accumulation in G 2 rather than mitosis (Fig. 6, E-H). Taken together, we conclude that failure to degrade USP37 during G 2 prevents mitotic entry.

DISCUSSION
In this study, we have demonstrated that the deubiquitinase USP37 undergoes biphasic degradation during late G 2 and mitotic exit/early G 1 . We confirmed that USP37 is targeted for destruction at mitotic exit and in G 1 by the APC Cdh1 ligase as indicated by our previous study (22). We have presented several lines of evidence implicating SCF ␤TrCP and Plk1 as the ligase and triggering kinase responsible for USP37 destruction in G 2 : (i) USP37 interacts with Plk1, ␤TrCP, and components of the SCF in vitro and in vivo; (ii) expression of ␤TrCP or Plk1 downregulates USP37 in a proteasome-dependent manner, whereas dominant-negative proteins increase USP37 levels; (iii) ␤TrCP induces USP37 ubiquitination in vivo and in vitro; (iv) phosphorylation of USP37 by Plk1 promotes binding to ␤TrCP; (v) loss of ␤TrCP or Plk1 activity by siRNA or chemical inhibition stabilizes USP37; and (vi) mutation of the ␤TrCP degron stabilizes USP37. Together these data indicate that phosphorylation by Plk1 leads to SCF ␤TrCP -mediated ubiquitination and subsequent destruction of a pool of USP37 during G 2 .
Our data identify USP37 as a node in the complex circuitry connecting the APC and SCF ligases throughout the cell cycle. USP37 joins Cdc25A and Claspin as an S-phase regulator that is coordinately regulated by SCF ␤TrCP and APC Cdh1 (10,12,18,19,37). A significant portion of USP37 remains throughout mitosis. These observations suggest an improved model for the biological role of USP37. Inhibition of APC Cdh1 promotes FIGURE 5. Plk1 mediates targeting of USP37 by SCF ␤TrCP . A, MYC-USP37 and HA-Plk1 K82R were transfected and analyzed as in Fig. 2. IP, immunoprecipitates. B, MYC-USP37 was transfected with HA-Plk1 constructs as in Fig. 3A. Vec, vector. C, 293T cells were transfected with the indicated constructs and treated as in Fig. 3, B and C. D, serum-starved T98G cells were stimulated to enter the cell cycle and transfected with the indicated siRNAs. Protein levels were analyzed throughout the cell cycle. siCTRL, control siRNA; siPlk1, siRNA targeting Plk1. E, recombinant GST-USP37 was incubated with the indicated kinases, captured on GSH-agarose, and tested for the ability to bind in vitro translated HA-␤TrCP. Cyc A, cyclin E; Cyc A, cyclin A. F, in vitro translated MYC-USP37 proteins were treated as in E, immunoprecipitated, and analyzed for interaction with HA-␤TrCP. G, MYC-USP37 proteins were treated as in E and mixed with E1, E2, ubiquitin, ubiquitin-aldehyde, and an energy-regenerating system in the presence (ϩ) or absence (Ϫ) of SCF ␤TrCP purified from 293T cells.
S-phase entry. USP37 then bifurcates into two pools, one of which controls substrates that prevent mitotic entry. Destruction of this pool by SCF ␤TrCP promotes the G 2 /M transition. We postulate that the remaining mitotic-stable pool promotes mitotic progression and must be degraded to promote the G 1 state, similar to many APC Cdh1 substrates. Further studies will be required to define the roles of these pools and how the mitotic pool remains stable. Certainly, identification of additional USP37 substrates will be a major step in elucidating the function of these pools of USP37.
In contrast to USP37, ␤TrCP-resistant mutants of the APC inhibitor Emi1 cause accumulation of APC substrates, includ-ing cyclin A, which must also be degraded for progression past prometaphase (13,42,43). Although USP37 is implicated in its stability, cyclin A is unlikely to mediate the cell cycle effects we have observed (22). First, failure to degrade cyclin A results in delay in mitosis rather than G 2 (42,43). Second, cyclin A levels are not altered when USP37 levels drop in G 2 -arrested cells. Although we cannot exclude that cyclin A stability is mediated by the remaining pool of USP37, this is unlikely to be the case as cyclin A is degraded in mitosis despite the presence of this fraction of USP37. This is also consistent with the specificity of USP37 for APC Cdh1 and the dependence of cyclin A destruction in G 2 and M upon APC Cdc20 (22,44). These data suggest that ␤TrCP-mediated Destruction of USP37 NOVEMBER 9, 2012 • VOLUME 287 • NUMBER 46 destruction of USP37 in G 2 is not prerequisite for destruction of cyclin A and indicate additional substrates. Intriguingly, the distinct cell cycle arrests caused by ␤TrCP-resistant USP37 and Emi1 also suggest that that these inhibitors regulate specific population of APC Cdh1 .
In summary, the results of our study further confirm the role of USP37 as a potent cell cycle regulator and underscore the need for identifying additional substrates of this enzyme to gain a better understanding of its critical functions.