The deubiquitylase USP7 is a novel cyclin F-interacting protein and regulates cyclin F protein stability

Cyclin F associates with USP7

To discover novel cyclin F-interacting proteins, we had previously performed mass spectrometry and identified proteins present in anti-HA immunoprecipitates from extracts of the U2OS human osteosarcoma cell line transiently expressing Flag-HA-Cyclin F or empty vector (Supplementary File 1 and Supplementary Information). Such proteins included known cyclin F-interacting proteins (for example, RRM2, CDC6, and casein kinase II). They also included USP7 (ubiquitin-specific protease 7), whose association with cyclin F has not been previously reported. USP7 is a deubiquitylase, and a master regulator of pathways maintaining genomic integrity: one such well-characterized function is its regulation of the MDM2-p53 axis [22]. To validate the interaction between cyclin F and USP7, we first transiently expressed Flag-HA-Cyclin F in HEK-293T cells. Immunoprecipitation of the ectopic cyclin F with antibodies against the HA tag, resulted in coprecipitation of the endogenous USP7 (Figure 1A). Normal mouse IgG did not coprecipitate Flag-HA-Cyclin F or USP7 (Figure 1A). These results suggest that cyclin F specifically interacts with USP7. The cyclin-box of cyclin F has been shown to mediate interaction with a number of its substrates (Figure 1B). To examine whether the cyclin-box mediated its interaction with USP7, we made use of the truncation mutant, Flag-HA-Cyclin F^1-270 (lacking both the cyclin-box and PEST region; Figure 1B). Like the wild-type Flag-HA-Cyclin F, Flag-HA-Cyclin F^1-270 was still able to coprecipitate endogenous USP7, suggesting that the interaction of cyclin F and USP7 is independent of the cyclin-box or PEST region of cyclin F (Figure 1C). Further, in this experiment, while USP7 coprecipitated with transiently expressed Flag-HA-Cyclin F, UBR5 (an E3 ubiquitin ligase) did not (Figure 1C).

Figure 1. Cyclin F interacts with USP7. (A) HEK-293T cells transfected with Flag-HA-Cyclin F were lysed and immunoprecipitated with anti-HA antibody or non-specific mouse immunoglobulin (IgG) as loading control. Immunocomplexes were immunoblotted as indicated. (B) Schematic representation of cyclin F WT and cyclin F^1-270 truncated mutant, highlighting the F-box, cyclin-box and PEST regions. (C) HEK-293T cells transfected with Flag-HA-Cyclin F WT or Flag-HA-Cyclin F^1-270 were immunoprecipitated and immunoblotted as indicated. (D) Endogenous USP7 was immunoprecipitated from HEK-293T cell extracts, with anti-USP7 antibody or an unrelated, anti-p-TAK1 antibody as a loading control (denoted as control). Immunocomplexes were immunoblotted as indicated. Asterisk denotes non-specific band.

We next assessed whether endogenous cyclin F interacts with endogenous USP7. Since the commercially available antibodies against cyclin F were unsuitable for immunoprecipitation, we chose to instead immunoprecipitate endogenous USP7 and check for coprecipitation of cyclin F. In HEK-293T cells, cyclin F was detected in USP7 immunoprecipitates, suggesting that endogenous cyclin F interacts with endogenous USP7 (Figure 1D). Importantly, mock immunoprecipitation with an unrelated antibody (anti-p-TAK1) did not appreciably detect USP7 or cyclin F (Figure 1D).

SCF^{Cyclin F} does not regulate USP7 protein levels

SCF^{Cyclin F} is an ubiquitin ligase, while USP7 is a deubiquitylase. To test whether SCF^{Cyclin F} targets USP7 for ubiquitylation and subsequent degradation, we first inhibited all SCF ligases with MLN4924, an inhibitor of NAE (NEDD8-activating enzyme); the activity of the latter is required for the activity of SCF ligases [23]. We found that the cyclin F levels are increased by treatment with MNL4924 compared to DMSO-treated controls (Figure 2A), suggesting that cyclin F could be regulated by an autocatalytic mechanism or by a different F-box protein. Additionally, in MNL4924-treated cells we observed a slower-migrating band detected specifically by anti-cyclin F antibody (Figure 2A, lane 2); we speculate this to be phosphorylated cyclin F, accumulating along with the non-phosphorylated form in the absence of cyclin F ubiquitylation [16]. On the other hand, treatment with MLN4924 increased RRM2, a well-documented cyclin F substrate, but did not affect USP7 protein levels, suggesting that SCF ligases, including SCF^{Cyclin F}, do not regulate USP7 protein abundance (Figure 2A).

Figure 2. SCF^{Cyclin F} does not regulate USP7 protein levels. (A) HCT116 cells treated with either DMSO or MLN4924 (3 μM) for 5 h were lysed, and immunoblotted as indicated. β-Actin was the loading control. Asterisk denotes non-specific band. (B) HEK-293T cells were transfected with an empty vector or Flag-HA-Cyclin F for 48 h, lysed, and immunoblotted as indicated. (C) HeLa cells treated with DMSO, etoposide (10 μM) or MG132 (10 μM) for 4 h were lysed, and immunoblotted as indicated. Asterisk denotes non-specific band.

In a variety of cell lines, cyclin F overexpression has been shown to downregulate a number of its substrates, including CP110 and RRM2, by ubiquitin-mediated proteasomal degradation [6, 8]. In line with these reported findings, overexpression of cyclin F in HEK-293T cells led to a decrease in RRM2 levels, but did not affect USP7 levels (Figure 2B), suggesting that SCF^{Cyclin F} does not regulate USP7.

Additionally, we asked whether other treatments known to decrease or increase cyclin F levels would perturb USP7 protein levels. In agreement with previous reports, we found that cyclin F is downregulated in response to treatment with DNA-damaging agent, etoposide (Figure 2C). This effect was specific to cyclin F, since cyclin A2, with which cyclin F shares the highest amino acid sequence similarity, remained unperturbed (Figure 2C). In contrast, treatment of cells with a proteasomal inhibitor, MG132, resulted in buildup of a slower-migrating cyclin F band (Figure 2C). This slower-migrating species could be interpreted as ubiquitylated cyclin F; alternatively, it might represent phosphorylated form of cyclin F. Of note, SCF^β-TrCP has been shown to target cyclin F for ubiquitin-mediated degradation following its phosphorylation by casein kinase II [18]. Thus, the slower-migrating cyclin F band might be phosphorylated form of cyclin F, accumulating due to inhibition of its degradation post MG132 treatment. p21, a short-lived protein whose stability is also regulated by the ubiquitylation pathway, is also upregulated in response to MG132 treatment, indicating that an effective concentration of MG132 was used in our experiment (Figure 2C). Importantly, treatment with either etoposide or MG132 did not affect USP7 levels (Figure 2C). Collectively, these data suggest that SCF^{Cyclin F} does not regulate USP7 protein levels.

Cyclin F protein and mRNA are downregulated in response to P22077 treatment

We next sought to evaluate whether the deubiquitylase activity of USP7 contributed to regulation of cyclin F protein levels. We utilized the small-molecule inhibitor, P22077 which predominantly inhibits USP7 in vivo [24]. P22077 irreversibly inhibits USP7 by selectively forming a covalent bond with the critical Cys²²³ within the active site in the catalytic domain [25]. In unstressed cells, USP7 is required for the stability of MDM2, an E3 ubiquitin ligase responsible for ubiquitylating p53 and targeting it for proteasomal degradation [22, 26]. In USP7-depleted cells, auto-ubiquitylated-MDM2 becomes unstable, resulting in the stabilization of p53 [27]. Therefore, we sought to first confirm the inhibition of USP7 in our experiments by evaluating p53 levels in P22077-treated cells. HCT116 cells treated with P22077 showed an upregulation in p53 levels, indicative of USP7 inhibition (Figure 3B). We did not evaluate p53 levels in P22077-treated HeLa cells, where p53 is preferentially degraded by viral E6-dependent ubiquitylation [22]. Treatment of both HeLa and HCT116 cells with P22077 led to an approx. 70% reduction in cyclin F levels, as early as 2 h, compared to DMSO-treated controls (Figure 3A, 3B). On the other hand, cyclin A2, remained unperturbed by P22077 treatment (Figure 3A, 3B). These latter observations suggest that P22077 does not induce gross changes in cell cycle within the 4 h treatment timeframe, and that the reduction in cyclin F observed are likely a consequence of USP7 inhibition. Likewise, USP7 levels remained unaffected by P22077 treatment (Figure 3A, 3B). These data suggest the deubiquitylase activity of USP7 specifically plays a role in regulating cyclin F levels. Similarly, treatment of HCT116 cells with FT671, a highly specific inhibitor of USP7 and structurally unrelated to P22077, also led to approx. 70% reduction in cyclin F levels, compared to DMSO-treated control (Supplementary Figure 1) [28].

Figure 3. USP7 regulates cyclin F protein levels. (A) HeLa cells treated with DMSO, MLN4924 (10 μM), or P22077 (25 μM) for the indicated hours were lysed, and immunoblotted as indicated. β-Actin was the loading control. Asterisks denote non-specific bands. Fold changes were calculated with densitometric values for cyclin F and cyclin A2 blots using β-Actin as loading control. (B) HCT116 cells were treated with DMSO or P22077 (25 μM) for 5 h. Where indicated, cells were treated with MG132 (10 μM) for 4 h prior to harvest. Cells were then lysed and immunoblotted as indicated. Asterisks denote non-specific bands. Fold changes were calculated with densitometric values for cyclin F and cyclin A2 blots using β-Actin as loading control. (C) HCT116 cells were treated with DMSO or cycloheximide (CHX; 50 μg/mL) for 75 min. Where indicated, bafilomycin A1 (BafA1; 100 nM) and/or MG132 (12.5 μM) were added along with CHX. Cells were lysed, and immunoblotted as indicated. Asterisks denote non-specific bands.

We next asked whether inhibition of the proteasomes by MG132 treatment would rescue the P22077-mediated downregulation of cyclin F. Treatment with MG132 alone led to an accumulation of diffuse slower-migrating bands of cyclin F, compared to that in DMSO-treated control, suggestive of accumulation of ubiquitylated cyclin F as a consequence of proteasomal inhibition (Figure 3B). Treatment with MG132 could not rescue P22077-mediated reduction in cyclin F (Figure 3B). This observation prompted us to examine two possible scenarios: first, we ascertained whether cyclin F is indeed degraded by the ubiquitin-proteasome system as reported by others. We examined the stability of cyclin F using cycloheximide in the presence or absence of a pair of inhibitors, namely, MG132 or bafilomycin A1 (an inhibitor of vacuolar-type H(+)-ATPase and consequently an inhibitor of protein degradation in the lysosomes) [29]. Cyclin F levels robustly reduced to less than 50% compared to control, when de novo protein synthesis was inhibited by cycloheximide (Figure 3C). This reduction was rescued only when cells were co-treated with MG132, but not bafilomycin A1, suggesting cyclin F is indeed targeted for degradation by the proteasomal system and not by lysosomal degradation (Figure 3C). The accumulation of LC3B-II, which is degraded in the lysosomes, indicates that lysosomal enzymes were effectively inhibited at the concentration of bafilomycin A1 used (Figure 3C) [30]. Second, we examined whether the P22077-induced reduction in cyclin F protein levels was due to a reduction in cyclin F mRNA levels. We performed reverse transcriptase-PCR analysis on HeLa cells treated with DMSO or P22077, for 1 and 4 h. While there was no appreciable difference in cyclin F mRNA levels at 1 h between control- and P22077-treated cells, there was nearly 50% reduction at 4 h upon P22077 treatment (Figure 4A). These latter results suggest USP7 regulates cyclin F mRNA levels and that this regulation is dependent on its deubiquitylase activity. Similar effects were observed in HCT116 cells as well (Figure 4B).

Figure 4. USP7 regulates cyclin F mRNA levels. (A) HeLa cells treated with DMSO or P22077 (25 μM) for the indicated hours were collected and total RNA was extracted. cDNAs prepared from each sample were amplified using primer pairs specific to each gene as indicated. The PCR products were analyzed by agarose gel electrophoresis and stained using ethidium bromide. RPL35A was the loading control. Fold change in cyclin F:RPL35A ratios were determined using densitometry values of the amplicons. The values represent an average from two independent experiments. (B) HCT116 cells treated with DMSO or P22077 (25 μM) for 4 h were analyzed as in (A).

USP7 regulates cyclin F protein stability

We next investigated whether USP7 regulates cyclin F protein stability. HCT116 cells were transfected with Flag-HA-cyclin F in the presence or absence of Flag-USP7-WT. As shown in Figure 5A, half-life of cyclin F was increased by USP7 overexpression (~ 60 min), whereas half-life of cyclin F in mock-transfected cells was about 30 min. We next evaluated whether this stabilization was dependent on the deubiquitylase activity of USP7. We employed two approaches: the first involved pharmacological inhibition of USP7 activity by P22077, and the second involved the expression of a catalytically-deficient mutant of USP7. Ectopic cyclin F levels robustly reduced to less than 50% of control levels, when de novo protein synthesis was inhibited by cycloheximide. This reduction was rescued in cells overexpressing wild-type Flag-USP7 (Figure 5B, 5C), and regardless of USP7 inhibition by P22077 (Figure 5B; last lane). These data suggest that the USP7-induced stability of cyclin F was independent of the deubiquitylase activity of USP7. This finding was further corroborated in experiments with the catalytic mutant of USP7, Flag-USP7-CS, that harbors a mutation of the critical Cys²²³ to Ser within the catalytic domain. As shown in Figure 5C, the cycloheximide-induced reduction in cyclin F levels was efficiently rescued in cells overexpressing Flag-USP7-CS, similar to that in cells overexpressing Flag-USP7-WT.

Figure 5. USP7 regulates cyclin F protein stability. (A) HCT116 cells were co-transfected with Flag-HA-Cyclin F WT and Flag-USP7 WT or empty vector. Twenty-four hours post transfection, cells were treated with DMSO or cycloheximide (CHX; 50 μg/mL) for the indicated minutes, lysed, and immunoblotted as indicated. α-Tubulin was the loading control. (B) HeLa cells were co-transfected with Flag-HA-Cyclin F WT and Flag-USP7 WT or empty vector. Forty-eight hours post transfection, cells were treated with CHX (50 μg/mL) for 75 min. Where indicated, P22077 (25 μM) was added to cells 2 h prior to the addition of CHX. Cells were then lysed, and immunoblotted as indicated. (C) HeLa cells were co-transfected with Flag-HA-Cyclin F WT and Flag-USP7 WT, Flag-USP7 CS, or empty vector. Forty-eight hours post transfection, cells were treated with CHX (50 μg/mL) for 75 min, lysed, and immunoblotted as indicated.

Association between USP7 and cyclin F under genotoxic stress

Consistent with previous reports, we observed that cyclin F protein levels are downregulated in cells treated with DNA-damaging agents, such as etoposide (Figure 2C). This downregulation has been attributed, in part, to increased proteasomal degradation of cyclin F following its ATR-dependent ubiquitylation [8]. We, therefore, hypothesized that under conditions of DNA damage, the interaction between cyclin F and USP7 might be reduced in order to afford a robust downregulation of cyclin F levels. To assess how the interaction between endogenous cyclin F and USP7 was influenced under unstressed and DNA-damaging conditions, we performed immunoprecipitation of endogenous USP7 from extracts of HeLa cells treated with or without etoposide. Western blot analysis showed that cyclin F was present in the anti-USP7 immunoprecipitates, but not in control immunoprecipitates with an unrelated, anti-p-TAK1 antibody (Figure 6). Notably, this association was weakly detected in cells subjected to genotoxic stress with etoposide, compared to that in untreated or MG132-treated cells (Figure 6). While the levels of cyclin F detected in the anti-USP7 immunoprecipitates only mirrors the cyclin F level present in the corresponding total extract (input lanes for Figure 6), it is also plausible that these results suggest that the association between USP7 and cyclin F might be disrupted in cells in response to DNA damage.

Figure 6. Interaction between USP7 and cyclin F under DNA-damaging condition. HeLa cells were treated with DMSO (control), etoposide (10 μM), or MG132 (10 μM) for 4 h. Endogenous USP7 was immunoprecipitated (IP) with anti-USP7 antibody or an unrelated, anti-p-TAK1 antibody (mock IP) as a loading control. Immunocomplexes were immunoblotted as indicated; β-actin was the loading control. Asterisk denotes non-specific band.

Expression vectors

Flag-HA-tagged Cyclin F construct was a kind gift from Dr. Michele Pagano (NYU Grossman School of Medicine, USA) [15]. Flag-HA-Cyclin F^1-270 (truncation mutant) was created by site-directed mutagenesis using Flag-HA-Cyclin F wild-type construct as the template. Presence of the desired mutation was verified by sequencing. pQFlag-USP7-WT and pQFlag-USP7-CS were purchased from Addgene (plasmids #46751 and #46752, respectively, deposited by Drs. Goedele Maertens and Gordon Peters).

Cell culture and transfections

HEK-293T, HeLa, and HCT116 cell lines were cultured in DMEM (Sigma-Aldrich, MO, USA) containing 10% fetal bovine serum (Thermo Fisher Scientific, MA, USA) supplemented with 100 IU/mL penicillin and 100 mg/mL streptomycin (Thermo Fisher Scientific). All cell lines were maintained in a humidified incubator at 37° C with 5% CO₂. Where indicated, cells were transfected with expression vectors using Lipofectamine 3000 (Thermo Fisher Scientific) according to manufacturer’s instructions.

Chemical inhibitors

Protease inhibitor cocktail set III and phosphatase inhibitor cocktail set V were purchased from Sigma-Aldrich. Where indicated, cells were treated with 10 or 12.5 μM MG132 (Sigma-Aldrich), 3 μM MLN4924 (Sigma-Aldrich), 25 μM P22077 (Sigma-Aldrich), 10 μM FT671 (MedChemExpress, NJ, USA), and 100 nM Bafilomycin A1 (Sigma-Aldrich).

RNA isolation and reverse transcriptase-PCR

Total RNA was purified from HeLa and HCT116 cells using TRI reagent (Sigma-Aldrich) according to manufacturer’s instructions. Single-stranded cDNA was reverse transcribed from 1 μg of total RNA using high-capacity cDNA reverse transcription kit (Thermo Fisher Scientific) according to manufacturer’s instruction. One microliter of the RT reaction product was mixed with 1x DreamTaq PCR master mix (Thermo Fisher Scientific) and 0.5 μM of each primer in a total reaction volume of 20 μL. The primers used were as follows: 1) human cyclin F: forward – ^5’CCAGGGAACCTGAAGCTCTTT^3’ and reverse – ^5’TCGCTTTCCCAGAGGAGGTA^3’: 2) human RPL35A: forward – ^5'GGGTACAGCATCACTCGGA^3' and reverse – ^5'ACGCCCGAGATGAAACAG^3'. PCR cycles were carried out using the Eppendorf Mastercycler Nexus GX2. The PCR products were analyzed by electrophoresis in 1 % agarose gels and visualized by staining with ethidium bromide. RPL35A expression was used for normalization.

Immunoprecipitation and immunoblotting

For immunoprecipitations, extracts were prepared by cell lysis in immunoprecipitation buffer (50 mM HEPES, pH 7.2; 150 mM NaCl; 1 % NP-40; 1 mM EDTA; 1:100 dilution of protease inhibitor cocktail set III; 1:50 dilution of phosphatase inhibitor cocktail set V; and 0.5 mM dithiothreitol). Cleared lysates were incubated with antibody for 16 h, and subsequently with protein A/G-agarose beads (Thermo Fisher Scientific) for 2 h, at 4° C, on a rotator. Immunocomplexes were washed 3 times with immunoprecipitation buffer and eluted bound-proteins were resolved in 8 or 12 % SDS-polyacrylamide gels. For other experiments, extracts were prepared by cell lysis in RIPA buffer (50 mM HEPES, pH 7.2; 150 mM NaCl; 1 % NP-40; 0.5 % sodium deoxycholate; 0.1 % SDS; 1 mM EDTA; 1:100 dilution of protease inhibitor cocktail set III, 1 mM orthovanadate; 5 mM NaF; and 1 mM dithiothreitol). Following electrophoresis, proteins were transferred onto a PVDF membrane. Blocking and incubation with primary antibodies were performed in Tris-buffered saline containing 0.1 % Tween-20 and supplemented with 5 % skimmed milk powder. Proteins were visualized by ChemiDoc XRC+ gel imaging system (Bio-Rad, CA, USA), using secondary horseradish peroxidase-conjugated antibodies and enhanced chemiluminescence reagent (femtoLucent plus-HRP, G-Biosciences, MO, USA).

Antibody sources

Antibodies to cyclin F (D9K2U), USP7 (D17C6), UBR5 (D6O8z), LC3 (4775S), and β-Actin (13E5) were from Cell Signaling Technology (MA, USA). Anti-HA-probe (F7) was from Santa Cruz (TX, USA), anti-cyclin A2 (BF683) from Bethyl Laboratories (TX, USA), anti-p53 (DO7) from Abcam (Cambridge, UK), anti-rabbit and anti-mouse IgG-HRP from Genei Laboratories (Bangalore, India), and anti-rabbit IgG-HRP from Cloud-Clone Corp (TX, USA).

Data availability statement

The authors confirm that the data supporting the findings of this study are available within the article.