Deubiquitinating Enzyme USP10 Promotes Hepatocellular Carcinoma Metastasis through Deubiquitinating and Stabilizing Smad4 Protein

Abstract
Hepatocellular carcinoma (HCC) has emerged as one of the most prevalent life-threatening cancers, and the high mortality rate is largely due to the metastasis. The sustained activation of Smad4 and transforming growth factor-β (TGF-β) is closely associated with advanced HCC metastasis. However, the regulatory mechanism underlying the aberrant activation of Smad4 and TGF-β pathway remains elusive. In this study, using a functional screen of USPs siRNA library, we identified ubiquitin-specific proteases USP10 as a deubiquitinating enzyme (DUB) that sustains the protein level of Smad4 and activates TGF-β signaling. Further analysis showed that USP10 directly interacts with Smad4 and stabilizes it through the cleavage of its proteolytic ubiquitination, thus promoting HCC metastasis. The suppression of USP10 by either shRNAs or catalytic inhibitor Spautin-1 significantly inhibited the migration of HCC cells, whereas the reconstitution of Smad4 was able to efficiently rescue this defect. Overall, our study not only uncovers the regulatory effect of USP10 on the protein abundance of Smad4, but also indicates that USP10 could be regarded as a potential intervention target for the metastatic HCC in Smad4 positive patients.

1.Introduction
Hepatocellular carcinoma (HCC) has emerged as the fifth most frequently diagnosed cancer and the second leading cause of cancer-related deaths worldwide, especially in South Africa and East Asia (Kim et al., 2018, Wang et al., 2019, Haider et al., 2019). The key reasons for the high mortality rate of HCC are diagnosis at an advanced stage and intrahepatic/extrahepatic metastasis, most of patients often diagnosed at advanced, unresectable or metastatic HCC, and just a minority of patients is diagnosed at an early stage (L’Hermitte et al., 2019, Haider et al., 2019, Zhu et al., 2018, Yuan et al., 2014). Treatment strategies for HCC remain very limited and there are no effective therapeutic strategies for metastatic HCC so far (L’Hermitte et al., 2019, Kim et al., 2018). In consideration of the limitations of advanced HCC treatment, novel and effective therapeutic strategies are urgently needed for metastatic HCC.
Multiple studies demonstrated that the over-activation of the transforming growth factor-β (TGF-β) pathway are closely associated with advanced HCC metastasis (Reichl et al., 2015, Yuan et al., 2014, Haider et al., 2019, Bertran et al., 2013). TGF-β is a pleiotropic cytokine that regulates a vast array of biological processes, including proliferation, migration/invasion, differentiation, etc (Dupont et al., 2009, Coulouarn et al., 2008). TGF-β exerts its effects through binding type II serine/threonine kinase receptors (TGFBR II) and inducing phosphorylation and activation of type I serine/threonine kinase receptors (TGFBR I). The ligand-activated TGFBR I then phosphorylates the cytoplasmic transducers, Smad2 and Smad3, which in turn form an active nuclear transcriptional complex with Smad4, to move into nucleus and regulate the transcription of downstream target genes (Haider et al., 2019, Coulouarn et al., 2008, Dupont et al., 2009, Zhang et al., 2017).
Smad4 is a unique and central transducer of TGF-β responses and is important for most TGF-β-induced biological effects, including tumor metastasis/invasion, embryonic development (Dupont et al., 2009). To further define the specific role of Smad4 in HCC development and

progression, we analyzed The Cancer Genome Atlas (TCGA) data sets for patients with HCC in order to gain insight on the relevance between Smad4 and HCC pathogenesis. We found that the survival rate significantly decreases in HCC patients with high levels of Smad4 and TGF-β compared to those with low expression levels (Supplementary Fig. S1A). Moreover, we found that the mRNA levels of Smad4 closely correlate with the high risk of HCC patients (Supplementary Fig. S1B) and similar observation was also achieved from that of SERPINE1/JUNB/CDKN2B, three well-known TGF-β signaling target genes (Supplementary Fig. S1C). Taken together, these data indicated a close association of Smad4 with HCC progression, thus the suppression of Smad4 may represent an effective strategy to restrain the metastasis potential of HCC cells.
Given that it is ubiquitination, rather than phosphorylation, which tightly controls the protein abundance and function of Smad4 (Dupont et al., 2009, Zhou et al., 2017, Wan et al., 2005, Liang et al., 2004), we conducted a functional RNA interference screen to globally profile the contribution of ubiquitin-specific proteases (USPs), the major subfamily of deubiquitinating enzymes (DUBs), to the transcriptional activity of Smad4 in HCC models. Among the several top “hits” which could reinforce this pathway, USP10 stood out since depletion of USP10 using specific shRNA minimizes smad4 protein levels. Further study demonstrated that USP10 promotes TGF-β signaling and HCC metastasis through the regulation on the protein abundance of Smad4, but imposing minimal effect on Smad2 and Smad3. Mechanistically, USP10 directly binds to Smad4 and reverts the proteolytic Lys-48-linked poly-ubiquitin chain on Smad4, thus leading to its stabilization. Moreover, ablation of USP10 by either shRNAs or USP10 inhibitor Spautin-1 significantly suppresses the metastasis of HCC cells in vitro and/or in vivo, whereas the reconstitution of Smad4 was able to efficiently rescue this defect, indicating that targeting USP10 could be a novel therapeutic strategy of metastatic HCC displaying high level of Smad4.

2.Materials and Methods
The HCC cell lines Bel-7402 and HepG2 were obtained from Cell Bank of Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences. All the cell lines were authenticated using DNA fingerprinting (variable number of tandem repeats), confirming that no cross contamination occurred during this study. Bel-7402 were maintained in RPMI-1640 medium (Gibco, #31800) with 10% fetal bovine serum (FBS; Gemini, #900-108) and HepG2 were maintained in Dulbecco’s modified Eagle’s medium (DMEM) (Gibco, #12800) with 10% FBS (ExCellBio, #FSP500). All cells were grown at 37°C in a humidified atmosphere containing 5% CO2.The antibodies against USP10 (#8501), Smad4 (#38454), Smad2/3 (#3102), β-TRCP (#4394), Vimentin (#3932), Slug (#3471) were purchased from Cell Signaling Technology, Danvers, MA, USA. N-cadherin (#610921) was purchased from BD Biosciences, Franklin Lakes, NJ, USA. Anti-HA-tag (db2603) and Anti-GAPDH (db106) were purchased from diagbio Biosciences, Hangzhou, China. Anti-DYKDDDDK-tag (#A00187), Anti-c-Myc-tag (#A00704-100) and anti-DYKDDDDK IP resin (#L00425) were purchased from Gen-Script, Nanjing, China. The anti-HA affinity gel (#B23302) was purchased from BioTools, Switzerland. The immunobloting analysis of proteins in cell lysates was performed as previously described (Zhu et al., 2018). The USP10 inhibitor Spautin-1 was purchased from TargetMol, Shanghai, China. The Recombinant Human TGF-β Protein (TGF-β, 240-B-002) was purchased from R&D Systems, Minneapolis, MN, USA.Plasmids and siRNAs were transfected using Jet PRIME (Polyplus, #114-15), according to the manufacturer’s instructions, and real-time PCR was performed as previously described (Zhu et al., 2018).

USPs siRNA sequences were listed in Supplementary Table S1, the sequences of the primer pairs were listed in Supplementary Table S2 and the relevant shRNAs sequences were listed in Supplementary Table S3.Bel-7402 cells were infected with lentivirus encoding the indicated shRNAs for 16 h and were supplemented with fresh culture medium. After 3 days, these cells were seeded in six-well dishes at a density of 1*106 cells per well and grown to be confluent. The denuded area of constant width was scrapped by using a sterile yellow micropipette tip, and the cell debris were washed with 1*PBS. The wounding closure was monitored and photographed after 24 h.Transwell migration assays were performed using uncoated poly carbonate membranes with 8 μm pores in 24-well microchemotaxis chambers. Bel-7402 cells were infected with lentivirus encoding the indicated shRNAs for 16 h and supplemented with fresh culture medium. After 3 days, 2*104 Bel-7402 cells were seeded into the top of a 24-well migration chamber in 200 μL serum-free RPMI 1640, and the lower chamber was filled with 600 μL RPMI 1640 medium containing 10% FBS and 5 ng/mL TGF-β. After incubation for 24 h, the cells migrated to the lower chamber were fixed with 100% Ethanol and stained with purple crystal. Then cells were photographed and counted in three random fields per insert.BALB/c nude mice were maintained in a pathogen-free animal facility and all animal experiments were performed in accordance with the regulations of Institutional Animal Care and Use Committee (IACUC). And the protocols for the animal study were approved by the Animal Research Committee at Zhejiang University, with ethical approval number IACUC-s19-006.

Every BALB/c nude mice was injected with 1*106 cells in 200 μL medium via tail vein. All of the BALB/c nude mice were euthanized 12 weeks post-injection and liver surface metastatic nodules were enumerated (Hao et al., 2014).Bel-7402 cells were seeded in 96-well dishes at a density of 1*104 cells per well in triplicate and cultured in RPMI 1640 medium containing the indicated concentrations of Spautin-1 (0 μM, 2.5 μM, 5 μM, 10 μM) for 24 h and cell proliferation was assessed by the Sulforhodamine B (SRB) assay.Flag-tagged Smad4 and HA-tagged ubiquitin mutant Lys-K48 only (K48) were transfected into 293T cells. After 36 h, 293T cells were lysed with 4% SDS and then incubated with anti-DYKDDDDK IP resin overnight at 4°C. The poly-ubiquitinated Smad4 from the cell lysate was pulled down by anti-DYKDDDDK IP resin and incubated with bacterial-expressed recombinant human USP10 (rhUSP10) protein for 2 h at 37°C in vitro. Subsequently, the ubiquitination levels of Smad4 were analyzed by western blotting.The plasmid of TGF-β transcriptional promoter reporter system (PGL4.14-SBE4-luc) containing 4-repeated sequences of “GTCTAGAC” (a sequence of Smad3 and Smad4 specifically recognize and bind to) (Zawel et al., 1998)) and a firefly luciferase was utilized to detect the activation of the TGF-β signaling, and a plasmid containing renilla luciferase was utilized as internal control.

These plasmids were transfected into 293T cells that infected with lentivirus encoding the indicated shRNAs. Cells were starved without serum overnight and then incubated with 10 ng/mL TGF-β for 6 h. Subsequently, the luminescence reading obtained with PGL4.14-SBE4-luc was monitored and divided by the one obtained with renilla luciferase.293T cells were transfected with pooled USPs siRNAs (a mix of oligos for each gene, 50 nM), PGL4.14-SBE4-luc and the plasmid containing renilla luciferase and supplemented with fresh culture medium after 5 h. Cells were starved without serum overnight and then incubated with 10 ng/mL TGF-β for 6 h. Subsequently, the luminescence reading obtained with PGL4.14-SBE4-luc was monitored and divided by the one obtained with renilla luciferase.All experiments were repeated at least 3 times in our study, the statistical data were presented as the mean SD. Student t-test was used to determine statistical significance of the differences and P-values of 0.05 were considered statistically significant.

3.Results
Ubiquitin-specific proteases (USPs) are the largest subfamily of DUBs and most of them are abnormally activated or expressed in a variety of malignant tumor, making them as ideal anticancer targets (Yuan et al., 2018). Since Smad4, the only co-Smad of TGF-β signaling, plays a critical role in the migration and the transcriptional activity of Smads complexes (Fig. 1A-1C), we sought to investigate the contribution of USPs involved in Smad4 regulation.To this end, the cell-based luciferase-screening model (Fig. 1C) was utilized to explore the potential USPs regulating TGF-β signaling by using siRNAs to inhibit the expression of 54 knownor predicted human USPs (Supplementary Table S1). The pooled USPs siRNAs (a mix of oligos for each gene, 50 nM) were transfected in 293T cells. Subsequently, these cells were treated with TGF-β for 6 h and followed by a TGF-β transcriptional activity assay. The results showed that siRNA pools for eight USPs including USP7, USP10, USP11, USP21, USP26, USP29, USP33 and USPL1 could remarkably inhibited TGF-β responses (Fig. 1D, relative luciferase activity < 0.5).However, among these 8 DUBs with potential regulatory effect on TGF-β transcriptional activity, only the depletion of USP10 could significantly minimize Smad4 protein levels (Fig. 1E and 1F).

The potent inhibitory effect on Smad4 was further verified by the suppression on TGF-β transcriptional activity, by 3 independent shRNAs targeting different coding regions of USP10 (Fig. 1G).Meanwhile, we analyzed The Cancer Genome Atlas (TCGA) data sets for patients with HCC in order to gain insight on the relevance between USP10 and HCC pathogenesis, as shown in Supplementary Fig. S2A, we found mRNA levels of USP10 closely correlate with the high risk of HCC patients. In line with these data, USP10 influenced the prognosis of HCC patients, as the Progression-Free Survival (PFS) significantly decreases in hepatitis virus-related HCC patients harboring higher levels of USP10 (Menyhart O et al., 2018 ) (http://kmplot.com/analysis/index.php?p=service&cancer=liver_rnaseq) (Supplementary Fig.S2B). Moreover, we analyzed Gene Expression Profiling Interactive Analysis (GEPIA) data sets for the gene expression correlation analysis of USP10 and TGF-β signaling target genes in patients with HCC, we found a positive correlation between USP10 and CDKN2B and the value of correlation coefficient is 0.49 (Supplementary Fig. S2C). Taken together, these data show that USP10 is involved in the malignancy of HCC and functions as a novel element in TGF-β signal transduction.The inhibitory effects of USP10 shRNAs on TGF-β responses led us to examine a potential role of USP10 in the regulation of Smad4 turnover and function. We found that depletion of USP10 using USP10 specific shRNA in HepG2 and Bel-7402 cells significantly decreased Smad4protein levels, but had little effect on Smad2/3 protein levels (Fig. 2A, 2B and Supplementary Fig. S3A). However, these inhibitory effect on Smad4 can be restored by the proteasome inhibitor MG132, indicating that USP10 influences Smad4 protein levels through the ubiquitin-proteasome system (Supplementary Fig. S3B).

Conversely, overexpression of USP10-WT (the wild type of USP10) but not USP10-C424A (the catalytically deficient mutant of USP10) in HepG2 and Bel-7402 cells dramatically increased the protein levels of Smad4, but not that of Smad2/3 (Fig. 2C, 2D and Supplementary Fig. S3C). To further validate the regulatory effect of USP10 on Smad4, we introduced a small molecule inhibitor Spautin-1 (Liu et al., 2011), which can specifically suppress deubiquitinase activity of USP10. As shown in Fig. 2E and Supplementary Fig. S3D, Spautin-1 significantly decreased Smad4 protein levels. To demonstrate that the effect of Spautin-1 on Smad4 is dependent on its catalytic inhibition on USP10, we examined the effect of Spautin-1 on Smad4 protein levels when depletion of USP10 in HCC cell. As shown in Supplementary Fig. S3E, when we knocked down USP10 in HCC cell, Spautin-1 imposed minimal effect on Smad4 protein levels, indicating the effect of Spautin-1 on Smad4 relies on the existence of USP10. To exclude the possibility that USP10-induced Smad4 levels was owing to the transcriptional regulation, we performed qRT-PCR assay to examine mRNA levels of Smad4 in HepG2 and Bel-7402 cells. The results showed that Smad4 mRNA levels in USP10-depleted/overexpressed cells remained almost unchanged compared to that of control cells (Fig. 2F and Supplementary Fig. S3F), suggesting that the effect of USP10 on Smad4 is not mediated through the regulation at the transcriptional levels, but dependent on its deubiquitinase activity.We then tested whether USP10 can regulate the protein stability of Smad4.

HepG2 and Bel-7402 cells infected with lentivirus encoding USP10 shRNAs were treated with or without the protein synthesis inhibitor cycloheximide (CHX, 40 μM) for the indicated times, and the Smad4 protein levels were analyzed. We found that Smad4 protein was less stable and the half-life was reduced from > 12 h to 6-9 h upon USP10 depletion (Fig. 2G and 2H), further denoting that USP10 regulates Smad4 protein levels at the post-translational level. However, depletion of USP10 imposed little effect on the half-life of Smad2/3 protein (Fig. 2I and Supplementary Fig.S3G). Taken together, these results demonstrate that USP10 activates TGF-β pathway by specifically stabilizing Smad4 protein in HCC models.We then explored the mechanism by which USP10 modulates Smad4 stability, firstly by examining whether USP10 is a bona fide Smad4 binding protein. The plasmids encoding HA-tagged Smad4 and non-tagged USP10 were co-expressed in 293T cells, and cells were subsequently harvested for co-immunoprecipitation (Co-IP) with the anti-HA antibody. As shown in Fig. 3A, USP10 could be indeed co-precipitated from cell lysates together with HA-tagged Smad4 by anti-HA antibody, suggesting an exogenous interaction between USP10 and Smad4. In addition, 293T cells were co-transfected with plasmids encoding Flag-tagged Smad4 and non-tagged USP10 (WT and C424A mutant, respectively) to examine whether the catalytic activity of USP10 is required for Smad4 binding. As shown in Fig. 3B, USP10-C424A also efficiently interacted with Smad4 similar as USP10-WT, suggesting the catalytic activity of USP10 is not required for Smad4 binding. In addition, we also observed an interaction between the Flag-tagged Smad4 and endogenous USP10 (Fig. 3C).

More importantly, the interaction between the endogenous USP10 and endogenous Smad4 was further demonstrated in our Co-IP analyses (Fig. 3D). Therefore, these results collectively revealed the specific interaction between USP10 and Smad4 both at the endogenous and exogenous levels.As USP10 directly binds to Smad4 and regulate its protein stability, we hypothesized that USP10 exerts such regulation through the deubiquitination of Smad4. To better illustrate the poly-ubiquitin chain of Smad4, the proteasome inhibitor MG132 was utilized in the subsequent analyses. Firstly, 293T cells were transfected with expression plasmids encoding for HA-tagged ubiquitin, Flag-tagged Smad4, either alone or in combination with His-tagged β-TRCP (a known E3 Ligase of Smad4 (Wan et al., 2005)) as well as Myc-tagged Cullin-1, then Flag-tagged Smad4 was immunoprecipitated and its ubiquitination pattern was visualized by immunoblotting againstHA-tagged ubiquitin. As shown in Fig. 4A and Supplementary Fig. S4A, β-TRCP dramatically increases the levels of Smad4 ubiquitination. Next, we performed a deubiquitination assay by co-transfecting cells with USP10-WT or USP10-C424A, and a significant decrease in poly-ubiquitylated Smad4 protein observed in cells transfected with USP10-WT, whereas co-expression of USP10-C424A failed to decrease Smad4 ubiquitination (Fig. 4A and Supplementary Fig. S4A).

In the contrast, depletion of USP10 markedly increased Smad4 ubiquitination (Fig. 4B and Supplementary Fig. S4B), indicating that USP10 is an essential DUB for Smad4.Ubiquitin contains seven lysine (Lys)-residues, including Lys-6, Lys-11, Lys-27, Lys-29, Lys-33, Lys-48 and Lys-63, and every ubiquitin monomer could exert a specific function by forming different poly-ubiquitin chains (Harrigan et al., 2018, Komander et al., 2009, Yuan et al., 2018). The formation of poly-ubiquitin chains ultimately determines the destiny of the substrate proteins (Yuan et al., 2018). To further explore which type of poly-ubiquitin chain of Smad4 was eliminated by USP10, we next examine the types of poly-ubiquitination occurring on Smad4. We co-transfected expression plasmids encoding for Flag-tagged Smad4 in combination with HA-tagged ubiquitin or Lys-6 only (K6), Lys-11 only (K11), Lys-27 only (K27) or Lys-29 only (K29), Lys-33 only (K33), Lys-48 only (K48) or Lys-63 only (K63) ubiquitin mutant. Flag-tagged Smad4 was immunoprecipitated and its ubiquitination pattern was visualized by immunoblotting against HA-tagged ubiquitin. We found that Smad4 had multiple types of poly-ubiquitination, including the proteolytic Lys-48-linked poly-ubiquitin chains (Supplementary Fig. S4C).A fore mentioned data suggested that USP10 regulated the protein stability of Smad4, thus raising the possibility that the proteolytic ubiquitination, namely, Lys-48-linked poly-ubiquitin chains (Yuan et al., 2018, Komander et al., 2009), on Smad4, was modulated by USP10.

To verify this hypothesis, we asked whether USP10 could cleave the Lys-K48 (K48)-only-ubiquitin chain on Smad4. As shown in Fig. 4C, 4D, Supplementary Fig. S4D and Fig. S4E, USP10-WT significantly decreased Lys-K48-linked poly-ubiquitination on Smad4, but imposed little effect on Lys-K48R-linked poly-ubiquitin chains, indicating that only the Lys-K48-linked ubiquitin chains would be removed by USP10. This effect was dependent on the deubiquitination activity of USP10, since the catalytic activity loss mutant USP10-C424A was incapable to cleave thepoly-ubiquitin chains.To further confirm the deubiquitination activity of USP10 toward Smad4, we utilized a cell free system and performed an in vitro deubiquitination assay using bacterial-expressed recombinant human USP10 (rhUSP10). Flag-tagged Smad4 and HA-tagged Lys-K48-linked ubiquitin mutant were transfected into 293T cells. Subsequently, poly-ubiquitinated Smad4 from the cell lysate pulled down by anti-DYKDDDDK (anti-Flag) IP resin and incubated with rhUSP10 protein for 2 h at 37°C in vitro. The results showed that rhUSP10 effectively removed the Lys-K48-linked poly-ubiquitination from Smad4 (Fig. 4E and Supplementary Fig. S4F).These data collectively demonstrate that USP10 deubiquitinates and stabilizes Smad4 through the cleavage of Lys-K48-linked poly-ubiquitin chain from Smad4, thus preventing its proteasomal degradation.Canonical TGF-β/Smad4 signaling has pleiotropic functions, among which, growth-arrest is the main response induced in normal epithelia or early carcinomas, whereas promotion of invasion and metastasis behaviors prevails in advanced tumors (Coulouarn et al., 2008, Dupont et al., 2009, Reichl et al., 2015).

In this context, we examined the function of USP10 in the metastasis of HCC cells. Epithelial-mesenchymal transition (EMT) plays fundamental roles in the early metastatic tumor by endowing tumor cells with a more motile and invasive potential (Yuan et al., 2014), so we firstly examined the effect imposed by USP10 on EMT relevant proteins. As shown in Fig. 5A, depletion of USP10 in Bel-7402 cells significantly decreased the expression of EMT marker proteins (N-cadherin, Vimentin and Slug). Next, we depicted the roles of USP10 on HCC metastasis using wounding healing assays and migration assays. Fig. 5B displayed that depletion of endogenous USP10 with shRNAs markedly decreased cell migration compared to that of the control cells, as indicated by the inhibition rate on wound healing as 56.15%, 59.19% and 76.42%, respectively (P < 0.001). Notably, depletion of USP10 imposed minimal effect on the cell proliferation, at least under the experimental conditions (Supplementary Fig. S5A). To extend our in vitro observations concerning USP10’s role in promoting HCC metastasis, we evaluated whether USP10 could promote HCC metastasis invivo. Firstly, we stably knocked down USP10 in Bel-7402 cells. Subsequently, these cells were intravenously injected into BALB/c nude mice. As shown in Fig. 5C, compared with the parental Bel-7402 cells, depletion of USP10 in Bel-7402 cells significantly decreased liver surface metastatic nodules of BALB/c nude mice. Taken together, these data suggested that depletion of USP10 significantly inhibit HCC metastasis both in vitro and in vivo.

In order to demonstrate that Smad4 indeed involved in the regulation of USP10 promoting HCC metastasis, we investigated whether Smad4 could rescue the metastasis defect caused by USP10 depletion. As shown in Fig. 5D of wounding healing assays, when over-expressed Smad4, the cell motile potential were reversed compared to USP10 depletion. Similar observation was also achieved from the transwell migration assays (Fig. 5E and Supplementary Fig. S5B).To further verify the function of USP10 on HCC metastasis, the small molecule inhibitor Spautin-1 was introduced. We found that Spautin-1 inhibited HCC metastasis in a dose-dependent manner and the inhibition rate is 46.18% (P < 0.001), 58.72% (P < 0.01), 67.95% (P < 0.01), for 2.5 μM, 5 μM, 10 μM, respectively (Fig. 5F). Interesting, Spautin-1 imposed minimal effect on the cell proliferation (Supplementary Fig. S5C), suggesting that under our experimental conditions, USP10 mainly modulated cell migration rather than cell proliferation, through the regulatory effect on Smad4. We then examined the protein levels of USP10 and Smad4 in Diethylnitrosamine (DEN)-induced HCC mice. As shown in Supplementary Fig. S5D, the protein levels of USP10 was more abundant and highly correlated with Smad4 expression in tumorous regions compared with that in non-tumorous tissues. These results demonstrate that USP10 reinforces the TGF-β signaling and promotes the metastasis of HCC cells. Moreover, depriving of USP10 catalytic activity by small molecule inhibitor exerts a significant repression on HCC metastasis.

4.Discussion
Metastasis remains an unsolved problem accounts for the high mortality of HCC patients, and the lack of intervention targets against metastasis greatly hinders the improvement of HCC treatment. The current study initiated a functional screening of USPs siRNA library on the TGF-β signaling, and identified that USP10 acts as a new regulator of Smad4 to reinforce the TGF-β/Smad4 pathway, thus ultimately leads to the enhanced metastasis of HCC cells. Mechanically, this pro-metastatic effect of USP10 was exerted through its deubiquitination and stabilization of Smad4 through the cleavage of K48-linked poly-ubiquitin chains from Smad4. Furthermore, targeting USP10 by either shRNAs or small molecule inhibitor Spautin-1 evidently inhibit HCC metastasis, revealing that targeting USP10 could be a novel therapeutic strategy against HCC metastasis and the specific inhibitors of USP10 could be potential anti-metastatic agents. Smad4 functions as a vital transcriptional factor of TGF-β signaling and is indispensable for TGF-β responses as well as relevant biological effects (Dupont et al., 2009, Ogawa et al., 2019). In view of the pleiotropic functions of canonical TGF-β signaling, Smad4 is extensively involved in tumor progression, playing drastically different roles in varied tumor models. Several lines of evidence showed that Smad4 may act as a tumor suppressor in colorectal cancer (Ogawa et al., 2019, Wasserman et al., 2018, Yan et al., 2016) and pancreatic cancer (Blackford et al., 2009, Xu et al., 2010). Wasserman and the colleagues demonstrated that Smad4 is closely associated with the high risk of recurrence or death of colorectal cancer patients, and the loss of Smad4 predicts worse clinical outcome, decreased immune infiltrate and resistance to chemotherapy (Wasserman et al., 2018). In pancreatic cancers, the inactivation of Smad4 significantly correlates with poor prognosis in patients with surgically resected adenocarcinoma of the pancreas (Blackford et al., 2009).

In contrast, Smad4 was also found to play important oncogenic roles in a vast diversity of cancer types, including breast cancer (Taylor et al., 2013, Xue et al., 2014), ovarian cancer (Chan et al., 2017), glioblastoma (Eichhorn et al., 2012) and HCC (Moon et al., 2017), (Huang et al., 2018). And metastasis is the most common outcome in TGF-β-hyperactivated cancer cells, generally owing to the promotion of epithelial-mesenchymal transition (Taylor et al., 2013, Bertran et al., 2013). Particularly, in HCC models, TGF-β/Smad4 promotes invasion/metastasis by promoting ERK pathway-mediated FGFR4 expression (Huang et al., 2018); or through lncRNA-activated by TGF-β (lncRNA-ATB) by competitively binding the miR-200 family and upregulating ZEB1/2 (Yuan et al., 2014). In line with these studies, we found that the activation of TGF-β/Smad4 is closely correlated with the poor prognosis of HCC patients. These clues collectively implicate that TGF-β/Smad4 signaling is fundamental for the malignancy of HCC; consequently, the interruption of TGF-β/Smad4 signaling is a potential anti-metastatic therapy for advanced HCC.

It has been well-established that phosphorylation of Smad2 and Smad3 dictates their nuclear-cytoplasmic shuttling, thus plays an essential role in the regulation of these R-smads (Haider et al., 2019, Dupont et al., 2009, Zhang et al., 2017). On the contrary, for the only known co-smad, namely, Smad4, ubiquitination/deubiquitination may represent a more critical mechanism to regulate Smad4, since ubiquitin ligases β-TRCP and Skp2 has been found to control the abundance of this only known co-smad (Izzi and Attisano, 2006, Liang et al., 2004, Wan et al., 2005), whereas minimal impact are imposed by phosphorylation. Several lines of evidence have revealed that USP9X and USP4 removed the mono-ubiquitination of Smad4 in lysine 519, promoting TGF-β responses (Dupont et al., 2009, Zhou et al., 2017). However, it still remained to known which DUB(s) could control the protein stability. In the current study, we found that USP10 imposed an important reinforcing effect on Smad4, by cleaving its proteolytic poly-ubiquitin chain and the subsequently stabilization.

Several lines of evidence have unraveled the diverse roles of USP10 in tumorigenesis as well as malignancy in different cancer types (Lin et al., 2013, Yuan et al., 2010, Weisberg et al., 2017, Guturi et al., 2016, Yang et al., 2014, Sun et al., 2018). Lin et al. found that USP10 antagonizes the transcriptional activity of c-Myc by deubiquitinating and stabilizing SIRT6, to inhibit cell growth and tumor formation in colon cancers (Lin et al., 2013). Yuan et al. showed that USP10 reversed MDM2-mediated p53 nuclear export and degradation by deubiquitinating and stabilizing cytoplasmic p53 (Yuan et al., 2010). Under DNA damage, USP10 becomes stabilization and translocates to the nucleus to activate and stabilize nuclear p53. Furthermore, USP10 inhibits tumor cell proliferation by increasing p53 levels in cells with wild-type p53, but exacerbates tumorigenesis in tumor cells with mutant p53 background. However, in acute myeloid leukemia (AML) patients with activating mutations in FMS-like tyrosine kinase 3 (FLT3), USP10 deubiquitinates and stabilizes FLT3 by removing the proteolytic poly-ubiquitin chains and preventing FLT3 degradation, to aggravate tumor progress (Weisberg et al., 2017). These studies indicated that the diverse functions of USP10 are largely dictated by its substrates in the specific cell context of the cancer models. Nevertheless, the roles of USP10 in cancer metastasis remain to be clarified. Ouchida, et al. identified that USP10 functions as a deubiquitinase and regulator of the EMT-transcription factor Slug to promote cell migration in multiple tumor cells, including non-small cell lung carcinoma, ovarian cancer, fibrosarcoma and breast cancer (Ouchida et al., 2018). However, the functions of USP10 on HCC metastasis are still unclear. The current study identified the pro-metastatic roles of USP10 in HCC, and found that USP10 promotes TGF-β signaling and HCC metastasis by stabilizing Smad4 through the cleavage of proteolytic K48-linked ubiquitination.

5.Conclusions
In summary, our data provide insights into the function of USP10 on HCC metastasis. We found that USP10 plays a key role in the metastasis of advanced HCC by deubiquitinating and stabilizing Smad4, a vital transcriptional factor of TGF-β signaling. Furthermore, we demonstrate that depletion of USP10 or depriving of its catalytic activity with small molecule inhibitor Spautin-1 significantly represses the metastasis of HCC cells in vitro and/or in vivo, whereas the reconstitution of Smad4 was able to efficiently rescue this defect. Taken together, our study not only identifies USP10 as a new pro-metastatic factor in HCC, but also unravels the mechanistic framework by which USP10 reinforces the TGF-β signaling through the stabilization of Smad4, thus provides potential therapeutic targets for the treatment against metastatic HCC in Smad4 positive patients.