USP10 promotes pancreatic ductal adenocarcinoma progression by attenuating FOXC1 protein degradation to activate the WNT signaling pathway.

Pancreatic ductal adenocarcinoma (PDAC) is a pathological subtype that accounts for more than 80% of pancreatic cancer (PC) cases, which is one of the most lethal cancers worldwide 1 . In contrast to the apparent progress in survival benefits for many other cancer types, the current 5-year survival rate for PC remains around 11% 2 . The early systemic spread and aggressive local growth of the disease lead to this poor prognosis 3 . More concerningly, a number of factors contribute to the lethality of PDAC, especially that the disease is not detected until an advanced stage, usually after distant metastasis 4 . Radical resection of the tumor is essential for the systemic treatment of PDAC, but this cannot be performed on many patients because of distant metastasis 5 . Therefore, exploring new sensitive metastasis-related biological targets at an early stage is crucial for increasing PDAC patient survival rates. Protein modifications, such as ubiquitination and phosphorylation, are regulated at the post-translational level by proteases and kinases, which participate in cell biological processes including tumor malignant progression 6 . Increasing recent evidence has suggested that deubiquitinating enzymes (DUBs) play an essential role in the ubiquitin-proteasome by hydrolyzing amide bonds between the single and polyubiquitin chains of substrate proteins 7 . Among the DUBs encoded in the human genome, five subtypes are characterized as cysteine peptidases, of which the ubiquitin-specific protease (USP) family is the largest group 8 . USP family members counteract ubiquitinase activity and affect protein functions involved in the regulation of protein stability, subcellular localization, and activity 9 . For example, USP25 stabilizes and decreases hypoxia-inducible factor (HIF)-1α degradation by deubiquitination, which promotes HIF1-α transcriptional activity and facilitates HIF-1α-mediated glycolysis 10 . Bhattacharya et al. found that silencing USP10 expression could induce endoplasmic reticulum stress-induced unfolded protein response and suppress PDAC cell viability 11 . Additionally, our previous study indicated that the kinesin KIF15 could serve as scaffolding protein to recruit USP10 and PGK1, which can promote PGK1 stability by USP10 mediated deubiquitination and increase aerobic glycolysis in PDAC 12 . However, the biological role and therapeutic potential of USP10 have not yet been fully explored in PDAC. The Forkhead box C1 (FOXC1) transcription factor participates in various biological processes in both normal and tumor cells 13 . The literature has indicated that abnormal FOXC1 expression patterns are associated with tumor progression and poor prognosis 14 . FOXC1 positively transcriptionally regulates IGF-1R and promotes cell proliferation, migration, invasion, and angiogenesis in PC 15 . Therefore, exploring the relationship between DUBs and transcription factors may help develop a new therapeutic strategy for PDAC. In this study, we found that USP10 can interact with FOXC1 and increase its protein stability by deubiquitination. USP10 can also be transcriptionally regulated by FOXC1, leading to a positive feedback loop that is involved in regulating PDAC cell proliferation and metastasis. These data suggest that USP10 is a promising therapeutic target in PDAC.

PC tissue samples were obtained from patients at Southwest Hospital by surgical resection with their consent and stored at -80°C. The experiments involving these samples were approved by the Ethics Committee of Southwest Hospital, Third Military Medical University (Army Medical University). The HPDE, PANC-1, MIA PaCa-2, SW-1990, AsPC-1, BxPC-3, and HEK-293T cell lines were obtained from the American Type Culture Collection (ATCC; Manassas, VA, USA). PANC-1, MIA PaCa-2, SW-1990, and HEK-293T cells were cultured in Dulbecco's Modified Eagle Medium (DMEM; Gibco, Grand Island, NY, USA), while HPDE, AsPC-1, and BxPC-3 cells were cultured in RPMI 1640 medium (Gibco). All media were supplemented with 10% fetal bovine serum (FBS; Invitrogen, Carlsbad, CA, USA). FH535 (a β-catenin pathway inhibitor), MG132 (a proteasome inhibitor), and chloroquine (an autophagy inhibitor) were purchased from MedChemExpress (Monmouth Junction, NJ, USA). Commercially available antibodies included the following: anti-USP10 (Proteintech, Rosemont, IL, USA; 19374-1-AP; 1:1,000), anti-FOXC1 (Thermo Fisher Scientific, Waltham, MA, USA; PA1-807; 1:1,000), anti-MYC (Proteintech; 60003-2-Ig; 1:1,000), anti-β-Tubulin (Proteintech; 10094-1-AP; 1:1,000), anti-Flag (Proteintech; 66008-4-Ig; 1:1,000), anti-GAPDH (Proteintech; 60004-1-Ig; 1:50,000), anti-MMP-9 (Proteintech; 10375-2-AP; 1:1,000), anti-MMP2 (Proteintech; 10373-2-AP; 1:1,000), ant-Snail (ABclonal, Woburn, MA, USA; A5243; 1:1,000), anti-Vimentin (Proteintech; 10366-1-AP; 1:1,000), Phospho-β-Catenin (Ser33/37/Thr41) (Cell Signaling Technology, Danvers, MA, USA; 9561; 1:1,000), and β-Catenin (Cell Signaling Technology; 8480; 1:1,000). Total RNA was purified by resuspending the lysed cells in RNA Isolation Reagent (Vazyme BioTech, Nanjing, China) following the corresponding protocol. RNA concentrations were determined using a NanoDrop 2000c instrument (Thermo Fisher Scientific). Total RNA was reverse transcribed into cDNA using the HiScript cDNA synthesis kit (Vazyme BioTech). Then, qRT-PCR was performed using the SYBR green PCR mix (Vazyme BioTech). The primer sequences used in our experiments are shown in Supplementary Table 1 . A short hairpin RNA (shRNA) targeting USP10 or the negative control (NC) shRNA was constructed into the lentiviral vector pENTR/H1/TO (Thermo Fisher Scientific). Lentiviral vectors encoding the USP10 or FOXC1 gene were also constructed. The packaged viruses were then transfected into PC cells and cultured for 48 hours. Because all the lentiviruses expressed the puromycin resistance gene, the transfected cells were treated with puromycin for selection. Gene expression was validated via qRT-PCR to confirm the transfection efficiency. The corresponding shRNA sequences are included in Supplementary Table 2 . After the cells were transfected, TUNEL assays were performed using the One Step TUNEL Apoptosis Assay Kit (Beyotime, Shanghai, China) following the manufacturer's protocols. Images of stained cells were captured by fluorescence microscopy, with quantification performed using Image J software (NIH, Bethesda, MD, USA). Cells from different groups were seeded in 6-well plates (500 cells/well) with at least three replicate wells and cultured with complete medium at 37°C for 2 to 3 weeks. For visualization, 4% paraformaldehyde (PFA) was used to fix the cultured cells for 15 minutes at room temperature, followed by staining with 2% crystal violet for 20 minutes. Images were photographed and the number of colonies was calculated. For wound healing assays to assess cell migration, PC cells were seeded in 6-well plates and cultured until they were 80% to 90% confluent. The cells were then scratched with a 200 μL pipette tip in the middle of each well, after which they were cultured with serum-free medium for 24 hours to monitor the cell migration patterns. Five distinct wound site fields were randomly chosen, with cell migration measured using a microscope and Image J software. The percentage of wound gap closure was calculated as the ratio of the residual wound area to the original wound area. For Transwell migration assays, 5×10 4 PC cells were seeded in the upper chamber of the plates (BD BioCoat; BD Biosciences, Franklin Lakes, NJ, USA) in 200 μL serum-free medium, while 10% FBS was added to the lower chamber. After culturing for 36 hours, the cells that migrated from the upper chamber to the lower chamber were fixed with 4% PFA and stained with 0.1% crystal violet, then counted using a microscope. For Transwell invasion assays, the upper chamber was coated with matrigel (BD Biosciences). Then, 5×10 4 PC cells were seeded in the upper chamber in 200 μL serum-free medium, while the lower chamber contained 700 μL medium with 10% FBS. The subsequent steps were the same as the Transwell migration assay. For both the migration and invasion assays, five random cell fields were counted and the experiments were performed in triplicate. For IF analysis, different groups of PC cells were seeded into multi-chamber slides and incubated at 37°C. After the cells were 80% confluent, the slides were fixed in 4% PFA for 20 minutes at room temperature and permeabilized with 0.1% Triton X-100. The samples were then incubated with primary antibodies, followed by incubation with secondary antibodies. After the cell samples were fixed and stained with DAPI, images were captured by fluorescence microscopy. Total protein was extracted from cell lysates using RIPA buffer with protease inhibitor cocktail and phosphatase inhibitors. Then, the protein samples were heated in SDS loading buffer for 5 to 10 minutes at 98°C. The protein levels of each group were analyzed by standard western blot analysis procedures. The GAPDH or tubulin protein levels were used as an internal reference to normalize gene expression in corresponding experiments. For Co-IP experiments, protein samples were extracted from PC cells using NP40 lysis buffer. The protein A/G magnetic beads (MedChemExpress) were first washed in the IP buffer at least four times in preparation for the following steps. The protein concentration of 10% of the lysate was determined using a BCA kit (Biosharp, Shanghai, China). The lysates were then incubated overnight at 4°C with the corresponding IP antibody or IgG isotype control followed by incubation with the magnetic beads. The immunoprecipitates were analyzed using western blots, as described above. For in vivo ubiquitination assays, the indicated plasmids were transfected into 293T cells for 24 hours, then the efficiency was verified using western blot analysis. Next, 20 μM MG132 was added into the medium to inhibit proteasome activity. The cells were then lysed using NP40 buffer and immunoprecipitated with an anti-Flag or anti-FOCX1 antibody. The immunoprecipitates were then subjected to western blot analysis. Liquid chromatography tandem mass spectrometry (LC-MS/MS) was used for mass spectrometry analyses in this study. PC cells were treated with NP40 buffer to extract the total protein. The lysates were then incubated with an anti-USP10 antibody or control IgG for IP. The immunoprecipitated proteins were then digested with modified sequencing grade trypsin, which purified the truncation of proteins for analysis. The fragmented peptides were then analyzed by LC-MS/MS to identify USP10-interacting proteins. For ChIP assays, the cells were lysed using ChIP lysis buffer and the crosslinked chromatin and protein complexes were then lysed by sonication. Appropriate amounts of samples were stored as input DNA samples. An anti-FOXC1 antibody or normal rabbit IgG was used for IP. Then, qPCR was used to analyze FOXC1 binding to the USP10 promoter, with the results normalized to the input samples. The primer sequences used for qPCR are shown in Supplementary Table 3 . The expression levels of the indicated proteins were examined in tumor and normal samples using IHC analysis of a tissue microarray. The samples were incubated with anti-USP10, anti-Ki-67, anti-PCNA, and anti-FOXC1 antibodies, then incubated with a horseradish peroxidase-conjugated secondary antibody. The samples were visualized with DAB and counterstained with hematoxylin, then the images were captured using a light microscope. To assess the regulatory effects of FOXC1 on USP10 mRNA, dual-luciferase reporter assays were performed as described previously 16 . The USP10 promoter region from 2000 bp upstream to 1 bp downstream of the transcription start site was cloned into the pGL3-Basic firefly luciferase reporter plasmid (Promega, Madison, WI, USA). The relative luciferase activity was calculated by normalizing the firefly luciferase activity to the Renilla luciferase activity. For the subcutaneous xenograft experiments, six-week-old female BALB/c-nude mice (specific-pathogen-free) were purchased from Vital River (Beijing Vital River Laboratory Animal Technology Co., Ltd., Beijing, China). The mice were randomly divided into three groups and subcutaneously inoculated with sh-control, sh-USP10-1, or sh-USP10-2 PANC-1 cells (5×10 6 cells/100 µL) to establish the xenograft PDAC tumor-bearing model. Tumor sizes and volumes were monitored every three days. The tumor volume was calculated using the following formula: V (cm 3 ) = 1/2 × length × width 2 . For the metastasis model, six-week-old female BALB/c-nude mice were randomly divided into three groups. Each mouse underwent surgery without pain under anesthesia. The abdominal cavity was opened successively to expose the spleen, then the cell suspension (5×10 6 cells/100 µL) was slowly injected into the spleen. The abdomen was then closed layer by layer and the mouse was woken up from anesthesia. After ten weeks of feeding, the mice were painlessly sacrificed. The liver and lung tissues were dissected and measured by section observation under a microscope. All the tumor, liver, and lung tissues were paraffin-embedded. Tissue sections were analyzed using IHC assays and hematoxylin and eosin (H&E) staining, with the results observed by microscopy. All animal experiments were approved by the Southwest Hospital, Third Military Medical University (Army Medical University) Animal Care and Use Committee. The datasets used in this study were obtained from The Cancer Genome Atlas (TCGA; https://portal.gdc.cancer.gov ) and Gene Expression Omnibus (GEO; https://www.ncbi.nlm.nih.gov/geo/ ) databases, including TCGA-PAAD, GSE62452 , GSE16515 , and GSE130221 . All data were analyzed using R software (version 4.0.1). SPSS 22.0 software (Chicago, IL, USA) and GraphPad Prism 9.0 (La Jolla, CA, USA) were used to analyze the data. The t-test or one-way ANOVA was used to analyze the significant differences between groups. The Cox proportional risk regression model was used to assess the prognostic variables of overall survival (OS) and recurrence-free survival. The data are presented as the mean ± standard deviation (SD) from three individual experiments. A P -value < 0.05 was considered statistically significant.

To identify the key USP involved in regulating PDAC cell migration, we conducted a functional experiment using a USP-specific siRNA library. After transfecting the siRNAs targeting the USP family, the expression levels of USPs were successfully knocked down. We then performed an unbiased loss-of-function screening with Transwell assays. Here, the individual knockdown of six DUBs resulted in a more than two-fold decrease in the cell migration rate. Among them, USP10 knockdown showed the strongest inhibitory effect on PDAC cell migration (Figure 1 A/B). Further analysis of TCGA data indicated that USP10 was significantly overexpressed in PDAC tissues compared with paired normal pancreatic tissues (Figure 1 C). USP10 also showed higher mRNA expression levels in PDAC tissues compared with normal pancreatic tissues in GEO datasets ( GES62452 , GES16515 , GSE130221 ) (Figure 1 D-F). Subsequently, Kaplan-Meier survival analysis showed that the TCGA-PAAD patients with higher USP10 mRNA expression levels had shorter OS rates and an unfavorable prognosis (Figure 1 G). To further confirm USP10 expression patterns in PDAC, a tissue microarray was examined, which suggested that the relative USP10 expression levels were higher in PDAC tissues (Figure 1 H). In parallel experiments, qRT-PCR and western blot analyses were performed to respectively verify the USP10 mRNA and protein expression patterns in PDAC cell lines (BxPC-3, AsPC-1, SW1990, MIA PaCa-2, PANC-1) and a human pancreatic ductal cell line (HPDE). The results indicated that USP10 showed higher mRNA and protein expression levels in PDAC cell lines compared with HPDE cells (Figure 1 I/J). The subsequent analysis of human PDAC tissues collected in our hospital revealed that USP10 expression was much higher in the PDAC specimens than in the paired normal pancreatic tissues (Figure 1 K). These data demonstrated that USP10 is a potential oncogene and overexpressed in PDAC. We also expanded the research samples to explore the clinical relevance, finding that high USP10 expression levels are often associated with lymph node metastasis (Table 1 ). However, there was no significant correlation between USP10 expression and tumor size, differentiation degree, or TNM stage (Table 1 ). Because our data suggested that USP10 might be an oncogene in PDAC, we next assessed its biological functions in this disease. First, we successfully generated stable overexpression and knockdown cell lines by respectively using lentiviral vectors encoding USP10 or a USP10-targeting shRNA (Figure 2 A/B). TUNEL assays were performed to examine the levels of apoptosis in PDAC cells following USP10 overexpression or knockdown. The results suggested that USP10 overexpression significantly suppressed the levels of apoptosis, while USP10 knockdown had the opposite effect (Figure 2 C/D). Moreover, colony formation assays were performed to evaluate the effect of USP10 expression on cell proliferation. Interestingly, USP10 overexpression significantly promoted both cell proliferation and colony formation. However, USP10 knockdown resulted in a decreased number of colonies and inhibited the cell proliferation rate (Figure 2 E/F). Similar effects were observed in animal experiments, which showed smaller subcutaneous tumor sizes in the USP10 knockdown groups than in the USP10 negative control group (Figure 2 G). Subsequently, qRT-PCR analysis was performed to detect USP10 mRNA expression levels in subcutaneous tumors from the different groups. The USP10 knockdown groups displayed lower USP10 expression levels (Figure 2 H). Notably, the tumor volume curve indicated that USP10 knockdown significantly inhibited PDAC growth (Figure 2 I). IHC assays were performed to confirm the protein expression patterns of Ki67 and PCNA, which are two markers that can reflect the strength of cell proliferation. The results suggested that Ki67 and PCNA both showed higher expression levels in the negative control group compared with the USP10 knockdown groups. Collectively, these data provided evidence that USP10 can promote PDAC cell proliferation (Figure 2 J/K). We next aimed to assess if USP10 plays an essential role in PDAC cell migration and invasion. Wound healing assays were performed to examine the migration abilities of PDAC cells following USP10 expression manipulation. The results revealed that USP10 knockdown in PANC-1 and MIA PaCa-2 cells significantly inhibited their migration (Figure 3 A). Moreover, data from the Transwell assays corroborated the wound healing assay results. These findings also suggested that USP10 knockdown decreased the PDAC cell migration and invasion rates (Figure 3 B). Furthermore, western blot analysis was performed to evaluate the effect of USP10 expression on the protein expression levels of MMP2, MMP9, Vimentin, and Snail1. Of note, the expression patterns of these abovementioned proteins were consistent with USP10 expression (Figure 3 C). USP10 knockdown significantly decreased the expression levels of these proteins, suggesting that USP10 can potentially promote the epithelial-mesenchymal transition (EMT) and extracellular matrix degradation, further supporting that USP10 can promote PDAC migration and invasion. Next, we developed metastasis models in nude mice using stable low expression of USP10 and negative controls. The results revealed that USP10 knockdown successfully decreased the number of liver metastases (Figure 3 D). Moreover, Kaplan-Meier survival analysis demonstrated that the negative control group mice showed a shorter OS than those in the USP10 knockdown groups (Figure 3 E). H&E staining showed that USP10 knockdown resulted in fewer metastases from the microcosmic perspective (Figure 3 F/G). These results illustrated that USP10 can promote PDAC migration and invasion in vitro and in vivo . The multiple biological functions of DUBs are mainly dependent on their interacting target proteins 17 . To investigate the potential target proteins, we used a lentivirus encoding USP10 to increase USP10 expression levels in cells, then performed Co-IP experiments to pull down the proteins interacting with USP10 and analyzed them by LC-MS. The following proteins ranked highest: ITGB1, RAF1, SIRT6, FOXC1, TP53, and SMAD4 (Figure 4 A/B). Bioinformatics analysis using the ubibrower database ( http://ubibrowser.bio-it.cn/ubibrowser_v3/ ) allowed us to narrow down the putative substrate proteins of USP10. We excluded the proteins previously known to bind to USP10. We also found FOXC1 to be a potential USP10 substrate protein by LC-MS (Figure 4 C), so we subsequently focused our experiments on this most promising candidate. IF assays were performed to further confirm the interaction between USP10 and FOXC1 and examine their localization. The results illustrated that USP10 co-localized with FOXC1, mainly accumulating in the nucleus (Figure 4 D). Next, we used endogenous and exogenous Co-IP experiments to detect if these proteins directly interact with each other. Notably, the results confirmed that FOXC1 could interact with USP10 via endogenous and exogenous Co-IPs (Figure 4 E/F). To further assess if USP10 could affect FOXC1 expression patterns, PCR and western blot analyses were performed. Interestingly, the results indicated that knocking down USP10 expression had a more significant effect on FOXC1 protein levels compared with its mRNA levels (Figure 4 G/H), with USP10 knockdown leading to decreased FOXC1 protein expression (Figure 4 H). Additionally, USP10 overexpression significantly increased the protein levels, but not the mRNA levels, of FOXC1 ( Supplementary Figure 1 A/B). Furthermore, we found that FOXC1 was overexpressed in TCGA-PAAD tissues compared with normal tissues, with these higher FOXC1 expression levels being associated with shorter patient OS and disease-free survival rates ( Supplementary Figure 2 A-C). PCR and western blot analyses both confirmed that FOXC1 was highly expressed in PDAC cell lines and tissues ( Supplementary Figure 2 D-F). Next, we successfully transfected a FOXC1-targeting shRNA into PANC-1 and MIA PaCa-2 cell lines ( Supplementary Figure 2 G). Colony formation and TUNEL assays suggested that FOXC1 knockdown led to inhibited proliferation and increased apoptosis in PDAC cells ( Supplementary Figure 2 H/I). Overall, this evidence suggested that USP10 can potentially exert carcinogenic effects by interacting with and stabilizing FOXC1 protein. Our data suggested that decreased USP10 expression levels led to reduced protein expression, but not mRNA expression, of FOXC1. To investigate how USP10 can affect FOXC1 protein expression patterns, we examined the effects of USP10 on FOXC1 in the autophagy and proteasome pathways. Chloroquine is an inhibitor that can induce cell autophagy in PDAC 18 . We therefore aimed to use chloroquine to reverse the reduced FOXC1 expression levels in the sh-USP10 group. However, chloroquine treatment had little effect on FOXC1 expression in this group (Figure 5 A). In addition, we found that USP10 knockdown diminished FOXC1 expression, which could be blocked by using MG132, a proteasome inhibitor (Figure 5 A). These results suggested that USP10 potentially affects FOXC1 protein expression through the ubiquitination pathway, which may also be related to the deubiquitination function of USP10. Subsequently, we found that increased USP10 expression could promote higher FOXC1 expression levels following transfection of wild-type USP10 (Figure 5 B). However, after transfecting mutant USP10 (USP10 C424A), which lost its deubiquitinating enzyme activity, no effects on FOXC1 expression were observed (Figure 5 C). We then examined if USP10 is involved in regulating FOXC1 protein stability. PANC1 and MIA PaCa-2 cells infected with a USP10-encoding or empty lentivirus were treated with the protein synthesis inhibitor cycloheximide (CHX, 40 μM) for the indicated times, then the FOXC1 protein expression levels were analyzed. The results revealed that USP10 overexpression markedly enhanced the protein stability of FOXC1 (Figure 5 D/E). From these results, we speculated that USP10 can promote FOXC1 protein stability to increase its expression levels via the ubiquitination pathway. To test this hypothesis, the ubiquitination levels were detected after infection with a lentivirus encoding USP10 or shUSP10 from both endogenous and exogenous perspectives. The exogenous experimental results indicated that after co-transfection of HA tagged-FOXC1 with sh-USP10 or Flag tagged-USP10, the USP10 knockdown groups displayed significantly increased FOXC1 ubiquitination, while the USP10 overexpression groups showed more FOXC1 deubiquitination (Figure 5 F/G). Consistently, the endogenous results also suggested that USP10 knockdown facilitated FOXC1 ubiquitination, while USP10 overexpression increased FOXC1 deubiquitination (Figure 5 H/I). We subsequently detected the FOXC1 ubiquitination levels following co-transfection of wild-type USP10 (Flag-USP10) or mutant USP10 (Flag-USP10 C424A) with HA-FOXC1 in HEK 293T cells. Notably, with MG132 treatment, USP10 C424A promoted FOXC1 ubiquitination compared with wild-type USP10 (Figure 5 J). To determine the key domain for the interaction between FOXC1 and USP10, we constructed full-length USP10 and truncated bodies with or without the USP domain (Figure 5 K). The Co-IP results indicated that FOXC1 could interact with the full-length USP10 and USP10 (337-798), which contained the USP domain (Figure 5 L). Previous studies have reported that serine 272 is a critical residue for maintaining proper FOXC1 protein stability 19 , 20 . In addition, our current data demonstrated that USP10 could inhibit FOXC1 protein degradation and promote its stability by deubiquitination. From this, we aimed to determine if USP10 can affect FOXC1 phosphorylation levels. Several reports have shown that ERK1/2 and P38 are the main kinases that phosphorylate FOXC1 19 - 21 . Here, we treated cells with epidermal growth factor (EGF, 100 ng/mL) for 5 minutes, then examined the effect of USP10 on FOXC1 phosphorylation levels. The results suggested that overexpressing wild-type USP10 led to significantly decreased FOXC1 phosphorylation levels compared with overexpressing the USP10 C424A mutant (Figure 6 A). In addition, cells transfected with mutant FOXC1 (S272A) resulted in downregulated FOXC1 phosphorylation levels, with USP10 overexpression partially rescuing this effect (Figure 6 B). To ensure the reliability of the observed phosphorylation differences, we redesigned the experiment and found that the phosphorylation levels increased with HA-FOXC1 expression in the wild-type group, but not in the mutant HA-FOXC1 (S272A) group (Figure 6 C). Additionally, we found that FOXC1 S272A promoted FOXC1 protein retention in the nucleus, while wild-type FOXC1 partially caused translocation to the cytoplasm (Figure 6 D). Interestingly, we also observed that USP10 C424A could increase FOXC1 nuclear translocation compared with wild-type USP10 (Figure 6 F). To investigate the mechanism by which USP10 promotes PDAC progression, we analyzed the signaling pathways that the downstream FOXC1 protein is potentially involved in. Using RNA sequencing (RNA-seq) analysis with three paired sh-NC and sh-FOXC1 cells, differentially expressed genes were identified and analyzed via Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO) analyses. We found that the WNT, mTOR, HIF-1, Notch, and FoxO signaling pathways were associated with FOXC1 (Figure 7 A-C). Additionally, gene set enrichment analysis (GSEA) of TCGA datasets showed that the WNT pathway was closely correlated with FOXC1 expression patterns in PDAC (Figure 7 D). Therefore, we determined the WNT pathway to be a possible candidate signaling pathway for FOXC1 activation. We also found that FOXC1 downregulation resulted in decreased USP10 mRNA expression levels by RNA-seq (Figure 7 B), which was consistent with TCGA data showing that FOXC1 mRNA expression levels were positively associated with those of USP10 (Figure 7 E). PCR analysis then revealed that upregulating FOXC1 remarkably elevated USP10 mRNA expression levels in PANC-1 cells (Figure 7 F). Next, western blot analysis suggested that knocking down FOXC1 expression led to lower protein expression levels of both FOXC1 and USP10 (Figure 7 G). Western blot analysis also demonstrated that FOXC1 upregulated phosphorylated β-catenin in the cytoplasm and β-catenin in nucleus, as well as increased USP10 expression in both the cytoplasm and nucleus, suggesting that FOXC1 activated the WNT signaling pathway (Figure 7 H). However, sh-USP10 and the WNT signaling inhibitor FH535 could partially reverse the activity of FOXC1 and weaken WNT signaling pathway activation (Figure 7 H). These results suggested that USP10 could activate WNT signaling by FOXC1, with FOXC1 and USP10 potentially being part of a positive feedback loop. To investigate the underlying mechanism of how USP10 mRNA is upregulated by FOXC1 overexpression, we first screened the USP10 gene promoter region for possible FOXC1 binding sites using the JASPAR database ( http://jaspardev.genereg.net/ ) (Figure 8 A). We found three potential FOXC1 binding sites in the USP10 promoter, namely region 1 (203-213), region 2 (448-458), and region 3 (1088-1098) (Figure 8 B). Dual-luciferase reporter assays were performed to further examine this, with the results suggesting that region 2 exhibited high FOXC1 binding. However, regions 1 and 3 had little promoter inducibility (Figure 8 C/D). We then conducted ChIP assays, which also demonstrated that region 2 was the binding site for FOXC1 in the USP10 promoter at the chromatin level (Figure 8 E/F). Taken together, FOXC1 can regulate USP10 at the transcriptional level to form a positive feedback control loop. Our earlier results suggested that USP10 knockdown can significantly suppress PDAC cell proliferation and metastasis. To validate that the USP10-mediated tumor-promoting effects were dependent on FOXC1, we conducted a series of functional experiments. TUNEL assay data indicated that FOXC1 overexpression could partially abolish the USP10 knockdown-mediated pro-apoptotic effect (Figure 9 A/B). In addition, we found that FOXC1 overexpression could partially reverse the inhibition of PDAC cell proliferation by USP10 knockdown (Figure 9 C/D). Consistent with the colony formation assay results, our Transwell assay data also illustrated that FOXC1 overexpression could abolish the inhibition of PDAC cell migration and invasion by USP10 knockdown (Figure 9 E/F). Western blot analysis suggested that the USP10 knockdown-mediated low expression patterns of MMP-2, MMP-9, Vimentin and Snail1 were abolished by FOXC1 overexpression (Figure 9 G). Additionally, we performed animal experiments to verify if FOXC1 overexpression alone could restore cell proliferation and migration inhibition from USP10 knockdown. The results demonstrated that FOXC1 overexpression could restore both USP10-mediated cell proliferation and migration in vivo (Figure 9 H/I). Overall, these results confirmed that the USP10-FOXC1 axis can promote PDAC cell proliferation and metastasis.

Chemotherapy and surgery are the main PDAC treatment options. However, only 15% to 20% of patients are eligible for surgery at the time of diagnosis. Most PDAC patients have distant metastases at diagnosis, and removal of the primary lesion by major surgery is unlikely to affect prognosis 22 . Unfortunately, even in cases where surgical removal is possible, nearly 75% of patients relapse over two years, suggesting that micrometastatic disease was also present in these patients 23 . Consequently, there is an urgent need to uncover the metastatic mechanism of PDAC. Evidence has suggested that dysfunctional protein expression is associated with PDAC malignant progression. Proteins, especially those in eukaryotic cells, maintain normal cellular function under homeostatic conditions, 80% of which are mediated by the ubiquitin-proteasome system 6 . The dynamic regulation of ubiquitination and deubiquitination maintains the stability, activity, and localization of key proteins 24 . Thus, when certain key metastasis-related proteins are modified by ubiquitination or deubiquitination, the carcinogenic effect may be amplified, resulting in the spread of PDAC. In recent years, mounting data have indicated that DUBs play crucial roles in tumorigenesis 25 . In the current study, we initiated a functional screening of a USP-targeting siRNA library for PDAC cell migration, finding that USP10 serves as a key regulator of FOCX1 expression to support WNT signaling pathway signaling and thereby ultimately increasing PDAC cell proliferation and metastasis. In addition, a WNT pathway inhibitor could partially inhibit the effects of FOXC1 on PDAC cells. Mechanistically, USP10 could directly interact with FOXC1 and promote its protein stability by deubiquitination, leading to FOXC1 overexpression. FOXC1 could activate the WNT signaling pathway to promote PDAC progression, while upregulating USP10 mRNA expression levels and forming a positive feedback loop. These results demonstrated that targeting USP10 is a potential novel therapeutic strategy for preventing PDAC metastasis. Previous studies have suggested that USP10 is involved in various biological processes, including DNA repair, cell cycle regulation, autophagy, and tumorigenesis 26 , 27 . Li et al. reported that USP10 can participate in modulating the stability and expression levels of NLRP7 protein, which promoted the polarization of pro-tumor M2-like macrophages by inducing the secretion of C-C motif chemokine ligand 2. Downstream, NLPR7 could promote NF-κB signaling activation and facilitate colorectal cancer progression 28 . Shi et al. suggested that USP10 could promoted IGF2BP1 protein stabilization by deubiquitination, resulting in high IGF2BP1 expression levels in breast cancer. IGF2BP1 could recognize and bind to the m6A sites on CPT1A mRNA, promoting the N6-methyladenosine modification and breast cancer metastasis 29 . In PDAC studies, silencing USP10 expression led to an unfolded protein response and upregulation of PERK and IRE1α, resulting in increased endoplasmic reticulum stress that thus ultimately decreased the metastatic potential 11 . However, a more detailed metastatic mechanism of PDAC was unclear. In this study, we focused on a new binding protein, FOXC1, which is also a reported oncogene in many tumor types. Our mass spectrometry analysis results indicated that USP10 potentially interacts with SIRT6, p53, and RAF1. Previous studies have suggested that USP10 can promote Raf-1 protein expression and activate the Raf-1/MEK/ERK pathway in endometriosis and glioblastoma. USP10 inhibits hepatic steatosis, insulin resistance, and inflammation through Sirt6. Additionally, USP10 functions as a Sirt6 deubiquitinase to promote tumorigenesis and cardiac myocyte hypertrophy. Furthermore, p53 protein stability and localization can be regulated by USP10, with USP10 suppressing tumor cell growth in cells with wild-type p53. The aforementioned studies demonstrate the protein interactions between USP10 and SIRT6, p53, and RAF1. However, direct evidence supporting their involvement in PDAC is lacking. Nevertheless, it is undeniable that USP10 may contribute to PDAC progression through its interactions with these proteins. Considering these findings, the current study focused on investigating FOXC1, which has not been previously studied, to elucidate its role in conjunction with USP10 in PDAC. Therefore, we found that knocking down USP10 expression could mediate anti-tumor effects, which were partially blocked by FOXC1. Additionally, FOXC1 overexpression was regulated by USP10 via deubiquitination. These finding suggested that USP10 could function as an oncogene to precisely control PDAC progression by stabilizing FOXC1 protein. FOXC1 is a key regulator of lineage norms and its expression is highly regulated throughout development. Phosphorylation of S272 is key for regulating FOXC1 stability and transcriptional activity 19 , 20 . ERK1/2-mediated phosphorylation of S272 inhibits rapid FOXC1 degradation by the 26S proteasome. Interestingly, we found that phosphorylation of S272 depends on the deubiquitination modification of FOXC1 by USP10, which stabilized FOXC1 localization in the nucleus and decreased its degradation. Numerous studies have confirmed that FOXC1 overexpression is frequently observed in cancers, with reports in more than 16 types of cancer, and is related to unfavorable prognosis. Because FOXC1 is a transcription factor and central hub gene that controls hundreds of gene networks, upregulated FOXC1 expression in cancer has widespread effects on many key biological processes related to tumor growth. During cancer progression and metastasis, FOXC1 mediates the cellular plasticity, partial EMT, treatment resistance, invasion, and migration of tumor stem cells 30 . In addition, FOXC1 is interconnected with multiple signaling pathway, such as EGFR, NF-κB, ERK, and PI3K/AKT 31 - 33 . Here, we found that FOCX1 knockdown inhibited PDAC cell proliferation, migration, and invasion. By integrating sequencing, bioinformatics predictions, and other experimental data, we determined the underlying mechanism to be that USP10 can induce FOXC1 overexpression and activate the WNT signaling pathway. A previous study in gastric cancer confirmed that FOXC1 binds to the promoter region of the β-catenin gene and transactivates its expression, which activates WNT signaling 34 . Another study found that FOXC1 can also form complexes with the unphosphorylated β-catenin protein in the cytoplasm, thereby facilitating β-catenin entry into the nucleus. Once in the nucleus, FOXC1 separates from β-catenin, thereby regulating c-MYC transcription and promoting gastric cancer cell proliferation 35 . Luciferase reporter and ChIP assay data from Cao et al. indicated that β-catenin can be a direct transcriptional target of FOXC1. FOXC1 promotes cancer stem cell-like properties by upregulating β-catenin and activating WNT signaling 36 . Taken together, our findings suggested that FOXC1 can activate WNT signaling by regulating β-catenin transcription. Thus, USP10 can upregulate FOXC1 expression and activate WNT signaling, which promotes PDAC cell proliferation and metastasis. Interestingly, we also found that FOXC1 could bind to the USP10 gene promoter, supporting USP10 transcriptional activation. These data indicate the presence of a positive feedback loop between USP10 and FOXC1 that can further support PDAC malignant progression.

In summary, our study demonstrated that USP10 is a key mediator of PDAC progression through supporting FOXC1 protein stability via deubiquitination. FOXC1 can promote USP10 transcriptional activity, suggesting that USP10 and FOXC1 form a positive feedback loop to activate WNT signaling and accelerate PDAC progression. USP10 represents an exciting therapeutic target that is a potential new strategy for treating PDAC.

Supplementary figures and tables.