Introduction

Ovarian cancer, a prevalent gynecological malignancy, often eludes early detection due to nonspecific symptoms and the absence of reliable early diagnostic biomarkers. Consequently, approximately 70% of cases are diagnosed at an advanced stage, characterized by extensive metastasis in the peritoneal cavity [1, 2]. Despite advances in therapeutic strategies, the effectiveness of conventional treatments for metastatic ovarian cancer remains limited, resulting in a dismal 5-year survival rate of approximately 30% for these patients [3]. Therefore, unraveling the molecular mechanisms driving ovarian cancer metastasis and identifying novel therapeutic targets are imperative.

Posttranslational modifications (PTMs) play a pivotal role in cellular development and maintenance of tissue-specific protein expression. They influence protein activity, function, and degradation at the posttranslational level [4]. Currently, over 600 types of PTMs, such as phosphorylation, ubiquitination, acetylation, glycosylation, and methylation, have been identified in mammalian proteins [5]. Undoubtedly, most studies have demonstrated that abnormal PTMs are known to alter protein properties and disrupt biological functions, contributing to the development and progression of various cancers [4, 6]. For instance, Zhao et al. indicated that PSMD14 directly interacted with LRPPRC and inhibited its ubiquitination, thereby promoting cancer metastasis [7]. Additionally, recent research has focused on the interplay between different types of PTMs in cancer progression, illustrating the complexity of these regulatory mechanisms, such as acetylation and ubiquitination and phosphorylation and ubiquitination [8]. For instance, Ren et al. demonstrated that p300 acetylation of p65 blocked FBXW2-induced p65 ubiquitination to enhance breast tumor growth [9]. In addition, Xiao et al. reported that ERK-mediated phosphorylation of PD-1 at Thr234 was essential for PD-1 deubiquitination and subsequent tumor immunotherapy [10]. These findings suggest that PTMs appear to play a critical role in tumorigenesis and cancer progression, and targeted PTMs might offer a potential new strategy for cancer therapy [11].

Receptor-interacting protein kinases (RIPKs), comprising RIPK1 through RIPK5, are serine/threonine kinases implicated in the genesis and progression of multiple tumor types [12, 13]. Of particular interest is RIPK4, which was originally identified in the context of protein kinase C regulation [14], and now is recognized for its overexpression in several cancers [15], including ovarian cancer [16]. Recently, the regulation of RIPK4 at the transcriptional and posttranscriptional levels has been extensively studied [17]. For example, Pan et al. revealed that miR-15b-5p might play an essential part in hepatocarcinogenesis by regulating RIPK4 [18]. Moreover, Cai et al. indicated that miR-330-3p suppressed the progression of ovarian cancer by targeting RIPK4 [19]. However, the posttranslational modulation of RIPK4 in ovarian cancer remains less explored.

In this study, we analyzed data from the PhosphoSitePlus® PTM database and identified ubiquitination and phosphorylation as the primary PTMs of RIPK4. We discovered that UCHL3 acts as a deubiquitinase for RIPK4, and that GSK3β phosphorylates RIPK4 at Ser420, significantly enhancing its stabilization by promoting the UCHL3/RIPK4 interaction. This interaction was found to drive cell proliferation and metastasis in ovarian cancer cell lines and xenografted models. Additionally, we observed a positive correlation between RIPK4 and UCHL3 protein expression, with high levels of either protein indicating poor prognosis in ovarian cancer patients. Our findings provide new insights into the crosstalk between phosphorylation and ubiquitination in RIPK4 stability and identify the UCHL3/RIPK4 axis as a potential therapeutic target for ovarian cancer.

Results

Identifying the deubiquitinase UCHL3 as a RIPK4-binding protein in ovarian cancer

Our single-cell sequencing data of ovarian cancer showed that RIPK4 was highly expressed in the metastatic group compared to the primary group (Fig. 1A, B). Additionally, western blotting results demonstrated a notable elevation of the RIPK4 expression in metastatic tissues compared to primary tissues in cases of ovarian cancer (Fig. 1C). Moreover, by analyzing data from The Cancer Genome Atlas (TCGA), we found that high RIPK4 levels were dramatically associated with shorter overall survival (OS) (Fig. 1D). These findings suggest that RIPK4 may play a key role in ovarian cancer progression. However, most studies have mainly focused on the transcriptional regulation mechanism of RIPK4 in ovarian cancer, few studies have explored its PTM mechanism. To address this question, we primarily performed Coimmunoprecipitation (Co-IP) and mass spectrometry (MS) analysis in Caov-3 cells to identify RIPK4-interacting proteins. Based on the number of unique binding peptides, the deubiquitinase UCHL3 might be a potential RIPK4-interacting protein (Fig. 1E and Supplementary Table 2). As expected, endogenous RIPK4 selectively interacted with UCHL3 in OVCAR-8 (Fig. 1F), SK-OV-3 and Caov-3 cells (Fig. S1A, B). Consistent with these results, exogenously expressed Myc-tagged RIPK4 could be detected in HA-tagged UCHL3 and vice versa (Fig. S1C). Moreover, Myc-RIPK4 was readily detected in HA-UCHL3 wild type (WT) or the enzymatically inactive HA-UCHL3 C95S from HEK-293T and Caov-3 cells (Fig. 1G and Fig. S1D), suggesting that the enzyme activity of UCHL3 was not needed for UCHL3 binding to RIPK4. We also performed in vitro Glutathione S-transferase (GST) pull-down assays by mixing purified Myc-RIPK4 with purified recombinant GST-UCHL3 or GST-UCHL3 C95S proteins. As shown in Fig. 1H, either GST-UCHL3 WT or GST-UCHL3 C95S could bind to Myc-RIPK4, but GST did not bind to Myc-RIPK4 alone, confirming that the interaction of RIPK4 with UCHL3 was straightforward (Fig. 1H). To explore the subcellular localization of RIPK4 and UCHL3, we performed immunofluorescence staining and proximity ligation assay (PLA) in ovarian cancer cells. Immunofluorescence data showed that endogenous UCHL3 and RIPK4 were mainly colocalized in the cytoplasm (Fig. 1I). Consistent with these results, the PLA data showed the direct interaction of endogenous UCHL3 with RIPK4 in the cytoplasm (Fig. 1J). To map the necessary region for their interaction, we made truncation constructs targeting the N- and C-domains of UCHL3 and the protein kinase, MD and ANK-domains of RIPK4 (Fig. 1K). We found that the C-domain of UCHL3 was necessary for binding to the MD-domain of RIPK4 (Fig. 1L-O). Collectively, our results provide evidence that UCHL3 is a bona fide RIPK4-interacting protein in ovarian cancer.

Fig. 1: RIPK4 interacts with UCHL3 in ovarian cancer.
figure 1

A, B RIPK4 expression in the primary and metastatic tissues of ovarian cancer analyzed from our single-cell sequencing data. C Western blotting analyses of RIPK4 in primary and metastatic tissues of ovarian cancer. PTL: primary tumor lysates, MTL: metastatic tumor lysates. D OS in patients with ovarian cancer in the RIPK4high and RIPK4low groups analyzed from the TCGA database. E Immunoprecipitation coupled with MS analysis to identify potential RIPK4-interacting proteins in Caov-3 cells. F OVCAR-8 cells were immunoprecipitated with anti-UCHL3 or anti-RIPK4 and analyzed. G HEK-293T cells were transfected with Myc-RIPK4, vector, HA-UCHL3 or HA-UCHL3 C95S plasmids. Lysates were immunoprecipitated with anti-Myc and analyzed. H Purified Myc-RIPK4 was incubated with GST, GST-UCHL3 or GST-UCHL3 C95S coupled to GSH-Sepharose, followed by Coomassie blue (CB) staining. I Cellular immunofluorescence assay. OVCAR-8, Caov-3, and SK-OV-3 cells were analyzed to determine the localization of UCHL3 (green) and RIPK4 (red). DAPI was used as a nuclear stain (blue). Scale bar, 10 μm. J In situ PLA. Endogenous UCHL3 and RIPK4 in OVCAR-8, Caov-3 and SK-OV-3 cells. Representative images are shown with merged PLA, and DAPI was used as a nuclear stain (blue). Each red dot indicates a UCHL3-RIPK4 complex. Scale bar, 10 μm. K Schematic diagram depicting the structure of the UCHL3 and RIPK4 proteins and their truncation constructs. The UCHL3 protein is composed of N- and C-domains, and the RIPK4 protein consists of the protein kinase, MD, and ANK-domains. L HA-tagged UCHL3 WT or short truncations and Myc-RIPK4 plasmids were coexpressed in Caov-3 cells. Then, the cell lysates were subjected to immunoprecipitation with an anti-HA antibody and immunoblotted with the indicated antibodies. M HA-tagged UCHL3 (WT or short truncations) and Myc-RIPK4 plasmids were coexpressed in HEK-293T cells. Then, the cell lysates were subjected to immunoprecipitation with an anti-Myc antibody and immunoblotted with the indicated antibodies. N Myc-tagged RIPK4 (WT or short truncations) and HA-UCHL3 plasmids were coexpressed in Caov-3 cells. Then, the cell lysates were subjected to immunoprecipitation with an anti-Myc antibody and immunoblotted with the indicated antibodies. O Myc-tagged RIPK4 (WT or short truncations) and HA-UCHL3 plasmids were coexpressed in HEK-293T cells. Then, the cell lysates were subjected to immunoprecipitation with an anti-HA antibody and immunoblotted with the indicated antibodies.

UCHL3 stabilizes RIPK4

We have revealed the interaction of UCHL3 with RIPK4, and speculated that UCHL3, as a deubiquitinase, might target RIPK4 to modulate its expression. As expected, ectopic UCHL3 overexpression significantly increased the level of RIPK4 in a dose-dependent manner in HEK-293T and Caov-3 cells (Fig. 2A and Fig. S2A). Additionally, knockdown of UCHL3 using a specific shRNA (Fig. S2B, C) significantly decreased the level of endogenous RIPK4 (Fig. 2B, C), whereas combined UCHL3 overexpression markedly reversed this effect (Fig. 2B). However, no effects of UCHL3 on RIPK4 mRNA expression were observed (Fig. 2E, F and Fig. S2D), supporting the notion that UCHL3 regulates RIPK4 protein abundance at the posttranslational level in ovarian cancer. Notably, overexpression of UCHL3 WT, but not UCHL3 C95S, increased the level of RIPK4 (Fig. 2D), indicating that the enzyme activity of UCHL3 was responsible for mediating the RIPK4 protein level. To explore whether UCHL3 modulates the expression of RIPK4 protein by affecting its stability, we utilized cycloheximide (CHX) to inhibit protein translation. UCHL3 knockdown significantly shortened the half-life of the RIPK4 protein in OVCAR-8 and SK-OV-3 cells (Fig. 2G-2J). Conversely, ectopic expression of UCHL3 WT, but not UCHL3 C95S, significantly prolonged the half-life of the RIPK4 protein (Fig. 2K and Fig. S2E). Collectively, these results demonstrate that UCHL3 positively regulates RIPK4 protein stability at the posttranslational level in ovarian cancer.

Fig. 2: UCHL3 promotes RIPK4 stability in ovarian cancer.
figure 2

A Western blotting analyses of HA-UCHL3 and Myc-RIPK4 in HEK-293T cells transfected with an increasing amount of HA-UCHL3 (0 μg, 2.5 μg, 5.0 μg and 10.0 μg) and a constant amount of Myc-RIPK4. B Western blotting analyses of UCHL3 and RIPK4 in OVCAR-8 cells transfected with UCHL3 shRNAs or HA-UCHL3 plasmids. C Western blotting analyses of UCHL3 and RIPK4 in OVCAR-8 and SK-OV-3 cells transfected with UCHL3 shRNA plasmids. D HEK-293T and Caov-3 cells were transfected with Myc-RIPK4 and HA-tagged vector, UCHL3 or UCHL3 C95S plasmids. Lysates were immunoblotted with the indicated antibodies and analyzed. E The mRNA levels of UCHL3 and RIPK4 in OVCAR-8 cells transfected with UCHL3 shRNAs. ***P < 0.001. n.s: P > 0.05. F The mRNA levels of UCHL3 and RIPK4 in SK-OV-3 cells transfected with UCHL3 shRNAs. ***P < 0.001. n.s.: P > 0.05. G, H Western blot analysis of RIPK4 and UCHL3 expression in OVCAR-8 and SK-OV-3 cells transfected with UCHL3 shRNAs and treated with CHX (50 μg/ml) at the indicated time points (0, 3, 6 and 9 h). I, J Quantification of RIPK4 and UCHL3 expression in OVCAR-8 and SK-OV-3 cells transfected with UCHL3 shRNAs and treated with CHX at the indicated time points (0, 3, 6, and 9 h). ***P < 0.001. K Western blot analysis of RIPK4 and UCHL3 expression in Caov-3 cells transfected with UCHL3 and UCHL3 C95S and treated with CHX at the indicated time points (0, 3, 6, and 9 h). L Western blotting analyses of UCHL3 and RIPK4 in OVCAR-8 cells transfected with UCHL3 shRNAs and treated with dimethylsulfoxide (DMSO) or the proteasome inhibitor MG132. M Western blotting analyses of UCHL3 and RIPK4 in OVCAR-8 cells transfected with UCHL3 shRNAs and treated with DMSO or the autophagy-lysosomal inhibitor chloroquine (CQ).

Protein degradation is commonly regulated by the ubiquitin-proteasome system and the autophagy-lysosomal pathway [20]. As UCHL3 functions as a deubiquitinase, we hypothesized that UCHL3 might prevent RIPK4 degradation by restraining the proteasomal degradation pathway. Consistent with this hypothesis, the absence of UCHL3 significantly decreased the expression of RIPK4, while the addition of the proteasome inhibitor MG132 (Fig. 2L and Fig. S2F) but not the autophagy-lysosomal inhibitor chloroquine (CQ) (Fig. 2M) almost completely reversed the effect. Moreover, UCHL3, but not UCHL3 C95S, significantly enhanced RIPK4 levels, whereas MG132 treatment dramatically elevated RIPK4 expression in both UCHL3- and UCHL3 C95S-overexpressing Caov-3 cells (Fig. S2G). These data indicated that RIPK4 constitutively underwent degradation in a proteasome-dependent manner in ovarian cancer.

UCHL3 deubiquitinates RIPK4 at the K469 site by removing the K48-linked ubiquitin chains

As UCHL3 functions as a deubiquitinase, we hypothesized that UCHL3 might remove RIPK4 ubiquitination to prevent its degradation. Consistent with this hypothesis, we found that knockdown of UCHL3 led to the accumulation of ubiquitinated RIPK4 in OVCAR-8 and SK-OV-3 cells (Fig. 3A). In contrast, the overexpression of UCHL3 significantly reduced the ubiquitination level of RIPK4 in HEK-293T and Caov-3 cells in a dose-dependent manner (Fig. 3B). Moreover, we observed that HEK-293T and Caov-3 cells expressing UCHL3 C95S had no effect on the ubiquitination of RIPK4, suggesting that UCHL3 removes the ubiquitination of RIPK4 through its deubiquitinating enzymatic activity (Fig. 3C). We further confirmed these data by in vitro ubiquitination experiments (Fig. 3D).

Fig. 3: UCHL3 deubiquitinates RIPK4 in ovarian cancer.
figure 3

A OVCAR-8 and SK-OV-3 cells were transfected with His-Ub or UCHL3 shRNAs and treated with MG132. Cell lysates were subjected to immunoprecipitation with anti-RIPK4 antibody and immunoblotted with the indicated antibodies for the in vivo deubiquitination assay. B HEK-293T and Caov-3 cells were transfected with Myc-RIPK4, His-Ub, or HA-UCHL3 and treated with MG132. Cell lysates were subjected to immunoprecipitation with anti-Myc and immunoblotted with the indicated antibodies for in vivo deubiquitination assays. C HEK-293T and Caov-3 cells were transfected with Myc-RIPK4, His-Ub, HA-UCHL3, or HA-UCHL3 C95S and treated with MG132. Cell lysates were subjected to immunoprecipitation with anti-Myc and immunoblotting with the indicated antibodies for the in vivo deubiquitination assay. D Myc-RIPK4 and His-Ub proteins were incubated with purified HA-UCHL3 or HA-UCHL3 C95S in vitro and then immunoblotted with the indicated antibodies for the in vitro deubiquitination assay. E HEK-293T and Caov-3 cells were transfected with Myc-RIPK4, HA-UCHL3, and ubiquitin fragments His-Ub K48 or K63 and treated with MG132. Cell lysates were subjected to immunoprecipitation with anti-Myc and immunoblotted with the indicated antibodies for in vivo deubiquitination assays. F HEK-293T cells were transfected with His-Ub, HA-UCHL3, Myc-RIPK4 WT, Myc-RIPK4 ▲AD, Myc-RIPK4 AD or Myc-RIPK4 ▲PKD fragments and treated with MG132. Cell lysates were subjected to immunoprecipitation with anti-Myc and immunoblotted with the indicated antibodies for in vivo deubiquitination assays. G HEK-293T cells were transfected with His-Ub, HA-UCHL3, Myc-RIPK4 WT or Myc-RIPK4 fragments (K456, K457, K458, K469 and K472) and treated with MG132. Cell lysates were subjected to immunoprecipitation with anti-Myc and immunoblotted with the indicated antibodies for in vivo deubiquitination assays. H HEK-293T cells were transfected with His-Ub, HA-UCHL3, Myc-RIPK4 WT or Myc-RIPK4 K469 mutant (K469R) and treated with MG132. Cell lysates were subjected to immunoprecipitation with anti-Myc and immunoblotted with the indicated antibodies for in vivo deubiquitination assays.

Ubiquitination contains seven types of linkages between ubiquitin molecules (lys-6, lys-11, lys-27, lys-29, lys-33, lys-48, and lys-63), and two of the most fully characterized forms of polyubiquitination occur by attachment to K48 or K63 [21, 22]. We sought to determine the type of ubiquitin chain on which UCHL3 removed RIPK4 and found that UCHL3 dramatically affected K48-linked ubiquitination in RIPK4, but it had no effect on K63- and other lys-linked ubiquitination (Fig. 3E and Fig. S3A). To determine which lysine sites on RIPK4 are involved in UCHL3-mediated deubiquitination, we generated several fragments of RIPK4 based on its protein domains. We observed that the MD domain (amino acids 286-484) was essential for UCHL3-mediated deubiquitination of RIPK4 (Fig. 3F). We then generated two mutants of the RIPK4 MD domain containing 6 K (mutated all other lysines in the MD domain except for K337, K343, K352, K367, K409 and K417) and 5 K (mutated all other lysines in the MD domain except for K457, K458, K459, K469 and K472). We found that the mutant containing 5 K was required for UCHL3 to deubiquitinate RIPK4 (Fig. S3B, C). To identify the specific target residue of RIPK4 modified by UCHL3, we constructed corresponding fragments at positions K456, K457, K458, K469 and K472 of RIPK4. The results showed that UCHL3 particularly deubiquitinated RIPK4 at K469 (Fig. 3G). Moreover, we observed almost no ubiquitination of RIPK4 when the K469 site of RIPK4 was mutated (Fig. 3H). Taken together, these data indicate that UCHL3 deubiquitinates RIPK4 by removing K48-linked ubiquitination at position K469 of RIPK4.

The enzymatic activity suppression of UCHL3 weakens the stability and deubiquitination of RIPK4

To improve the possibility of clinical transformation, we employed TCID, a small molecule inhibitor that mainly suppresses the enzymatic activity of UCHL3 in ovarian cancer cells [23, 24], to further confirm the UCHL3-mediated stability and deubiquitination of RIPK4. We primarily explored the effect of TCID on the expression and degradation of RIPK4. We observed that TCID suppressed the protein level of RIPK4 in a dose-dependent manner in OVCAR-8 and SK-OV-3 cells (Fig. 4A). In addition, pretreatment with MG132 restored the RIPK4 downregulation caused by pharmacological inhibition of TCID (Fig. 4B, C). We then examined whether UCHL3 inhibition modulated RIPK4 stability and found that cotreatment with TCID and CHX significantly shortened the half-life of the RIPK4 protein (Fig. 4D-G). Moreover, UCHL3, but not UCHL3 C95S, reversed this effect caused by TCID and dramatically prolonged the half-life of RIPK4 in Caov-3 cells (Fig. 4H, I). Importantly, TCID treatment failed to further significantly reduce RIPK4 levels in UCHL3-silenced cells, indicating that the enzymatic activity of UCHL3 plays a major role in RIPK4 stability (Fig. 4J, K). Furthermore, the deubiquitination activity of UCHL3 was almost completely abrogated by TCID treatment in HEK-293T and Caov-3 cells (Fig. 4L). Generally, these findings indicated that UCHL3, as a deubiquitinase for RIPK4, enhances RIPK4 protein stabilization and deubiquitination in an enzymatic activity-dependent manner in ovarian cancer.

Fig. 4: The enzymatic activity suppression of UCHL3 decreases the stability and deubiquitination of RIPK4 in ovarian cancer.
figure 4

A Western blotting analyses of RIPK4 in OVCAR-8 and SK-OV-3 cells treated with 0, 3, 6, and 9 μM TCID. B, C Western blotting analyses of RIPK4 in OVCAR-8 and SK-OV-3 cells treated with TCID or MG132. D Western blotting analyses of RIPK4 in OVCAR-8 cells treated with CHX alone or TCID + CHX for the indicated time periods (0, 4, 8 and 12 h). E Quantification of RIPK4 in OVCAR-8 cells treated with CHX alone or TCID + CHX for the indicated time periods (0, 4, 8 and 12 h). ***P < 0.001. F Western blotting analyses of RIPK4 in SK-OV-3 cells treated with CHX alone or TCID + CHX for the indicated time periods (0, 4, 8, and 12 h). G Quantification of RIPK4 in SK-OV-3 cells treated with CHX alone or TCID + CHX for the indicated time periods (0, 4, 8 and 12 h). ***P < 0.001. H Caov-3 cells were transfected with vector, UCHL3 or UCHL3 C95S followed by TCID treatment. RIPK4 protein expression was measured using western blotting. I Quantification of RIPK4 in Caov-3 cells transfected with vector, UCHL3 or UCHL3 C95S followed by TCID treatment. ***P < 0.001. J, K OVCAR-8 and SK-OV-3 cells were transfected with UCHL3 shRNAs or treated with TCID. The expression of RIPK4 and UCHL3 was measured using western blotting. L HEK-293T and Caov-3 cells were transfected with Myc-RIPK4, His-Ub, or HA-UCHL3 followed by TCID or MG132 treatment. Cell lysates were subjected to immunoprecipitation with anti-Myc and immunoblotted with the indicated antibodies for in vivo deubiquitination assays.

RIPK4 was found to drive the proliferation and metastasis of ovarian cancer in a UCHL3-dependent manner in vitro and in vivo

To explore the biological function of RIPK4 and UCHL3 in ovarian cancer, we primarily transfected a series of RIPK4 or UCHL3 plasmids in ovarian cancer cell lines, and then performed functional assays in vitro and in vivo. As expected, following UCHL3 knockdown, we observed a significant reduction in RIPK4 levels, whereas the reconstitution of RIPK4 observably enhanced its expression in OVCAR-8 and SK-OV-3 cells (Fig. S4A, B). Conversely, UCHL3 overexpression significantly increased the RIPK4 level, whereas combined knockdown of RIPK4 using a specific shRNA markedly reversed this effect (Fig. S4C). The colony formation and Transwell results indicated that UCHL3 knockdown significantly impaired cell proliferation and migration, whereas ectopic RIPK4 overexpression (Fig. S4A, B) largely abrogated these effects (Fig. 5A, B). To further verify our results, we established xenografts in nude mice by the subcutaneous injection of ovarian cancer cells. We observed that UCHL3 downregulation dramatically suppressed tumor growth, consequently leading to a decrease in tumor mass, weight, and volume compared to those in the control group (Fig. 5C, E, G). However, this inhibition was observably reversed by RIPK4 elevation (Fig. 5C, E, G). Moreover, UCHL3 depletion markedly reduced the number of metastatic nodules in the lung, but combined RIPK4 overexpression markedly reversed this effect (Fig. 5I, J). Conversely, ectopic UCHL3 overexpression (Fig. S4C) promoted tumor growth (Fig. 5D, F, H) and metastasis (Fig. 5K, L), but this enhanced protumor function was markedly suppressed by RIPK4 silencing. In aggregate, our results indicated that RIPK4 functioned as a key putative oncogenic role that facilitated ovarian cancer initiation and progression, probably mediated by UCHL3.

Fig. 5: UCHL3 participates in the development of ovarian cancer in a RIPK4-dependent manner.
figure 5

A Colony formation assay of OVCAR-8 and SK-OV-3 cells. OVCAR-8 and SK-OV-3 cells were transfected with negative control (Scr-shRNA), UCHL3 shRNAs, or RIPK4 plasmids followed by a colony formation assay. Representative images and quantification analysis are shown. **P < 0.01, ***P < 0.001. B Transwell assay of OVCAR-8 and SK-OV-3 cells. OVCAR-8 and SK-OV-3 cells were transfected with Scr-shRNA, UCHL3 shRNA or RIPK4 plasmids followed by transwell assays. Representative images and quantification analysis are shown. ***P < 0.001. C The image shows representative xenografts in the Scr-shRNA, UCHL3 shRNA, and UCHL3 shRNA+RIPK4 groups. n = 5. D The image shows representative xenografts in the Scr-shRNA, UCHL3, and UCHL3 + RIPK4 shRNA groups. n = 5. E Tumor weights were detected on day 28 after SK-OV-3 cell injection in each group. n = 5. ***P < 0.001. F Tumor weights were detected on day 28 after Caov-3 cell injection in each group. n = 5. ***P < 0.001. G Tumor volumes were detected on day 28 after SK-OV-3 cell injection in each group. n = 5. ***P < 0.001. H Tumor volumes were detected on day 28 after Caov-3 cell injection in each group. n = 5. ***P < 0.001. I The image shows the metastatic lung nodules for the Scr-shRNA, UCHL3 shRNA, and UCHL3 shRNA+RIPK4 groups. n = 5. J Quantification analysis of metastatic lung nodules in the Scr-shRNA, UCHL3 shRNA, and UCHL3 shRNA+RIPK4 groups. n = 5. ***P < 0.001. K The image shows the metastatic lung nodules for the Scr-shRNA, UCHL3, and UCHL3 + RIPK4 shRNA groups. n = 5. L Quantification analysis of metastatic lung nodules in the Scr-shRNA, UCHL3 and UCHL3 + RIPK4 shRNA groups. n = 5. ***P < 0.001.

GSK3β-mediated Ser420 phosphorylation of RIPK4 promotes RIPK4 interaction with UCHL3

Phosphorylation and ubiquitination are two important PTMs, and their interplay plays a critical role in regulating a variety of biological processes in various cancers [25, 26]. Based on the analysis from the PhosphoSitePlus® PTM database (https://www.phosphosite.org/), most phosphorylation sites were present on RIPK4. Importantly, we identified the phosphorylation site Ser420 of RIPK4 as the conserved GSK3β substrate motif (Fig. 6A), suggesting that RIPK4 might interact with GSK3β. Consistent with our hypothesis, endogenous RIPK4 directly interacted with GSK3β in OVCAR-8, SK-OV-3, and Caov-3 cells (Fig. 6B-D). To determine the regions mediating this interaction, we prepared several fragments of RIPK4 based on the protein domains. We observed that the MD regions (amino acids 286-484) of RIPK4 were needed for its interaction with GSK3β in Caov-3 and HEK-293T cells (Fig. 6E, F).

Fig. 6: GSK3β-mediated phosphorylation of RIPK4 at Ser420 promotes RIPK4 interaction with UCHL3.
figure 6

A The GSK3β consensus motif and phosphorylation of RIPK4 at the Ser420 site identified in the PhosphoSitePlus® PTM database (https://www.phosphosite.org/homeAction.action). B-D OVCAR-8, SK-OV-3 and Caov-3 cells were immunoprecipitated with anti-GSK3β or anti-RIPK4 antibodies and analyzed. E, F Myc-RIPK4 (WT or ▲AD, AD and ▲PKD short truncations) and Flag-GSK3β plasmids were coexpressed in Caov-3 and HEK-293T cells. Cell lysates were subjected to immunoprecipitation with anti-Myc antibody and immunoblotted with the indicated antibodies. G, H RIPK4 phosphorylation assay in HEK-293T and Caov-3 cells after control (NaCl) or GSK3β inhibitor (LiCl) treatment. Cell lysates were immunoprecipitated with anti-RIPK4 antibody and analyzed. I RIPK4 phosphorylation assay in SK-OV-3 and OVCAR-8 cells after control or LiCl treatment. Cell lysates were immunoprecipitated with anti-RIPK4 antibody and analyzed. J, K RIPK4 phosphorylation assays in HEK-293T and Caov-3 cells cotransfected with phosphorylation-activated Flag-GSK3β (Flag-GSK3β CA), Myc-RIPK4 WT or phosphorylation-deficient Myc-RIPK4 S420A. Cell lysates were immunoprecipitated with anti-Myc antibody and analyzed. L, M Caov-3 and HEK-293T cells were transfected with Scr-shRNA or GSK3β shRNAs. Cell lysates were immunoprecipitated with anti-RIPK4 antibody and analyzed. N, O HEK-293T and Caov-3 cells were transfected with His-Ub, HA-UCHL3 or GSK3β shRNAs and treated with MG132. Cell lysates were immunoprecipitated with anti-RIPK4 antibody and immunoblotted with the indicated antibodies for the in vivo deubiquitination assay. P, Q HEK-293T and Caov-3 cells were transfected with Flag-GSK3β, HA-UCHL3, Myc-RIPK4 WT or phosphorylation-deficient Myc-RIPK4 S420A. Cell lysates were immunoprecipitated with anti-HA antibody and analyzed. R, S HEK-293T and Caov-3 cells were transfected with phosphorylation-activated Flag-GSK3β CA, His-Ub, HA-UCHL3, Myc-RIPK4 WT or phosphorylation-deficient Myc-RIPK4 S420A and treated with MG132. Cell lysates were immunoprecipitated with anti-Myc antibody and immunoblotted with the indicated antibodies for the in vivo deubiquitination assay.

Since GSK3β is a highly conserved serine/threonine kinase and a potential GSK3β substrate motif Ser420 exists in the formation of RIPK4, we then explored whether RIPK4 could be phosphorylated at Ser420 by GSK3β. We initially validated that the GSK3β inhibitor LiCl [27] significantly restrained RIPK4 phosphorylation in HEK-293T, Caov-3, SK-OV-3, and OVCAR-8 cells (Fig. 6G-I). In addition, the replacement of Ser420 with Ala (A) (S420A), a phosphorylation-deficient mutant, abolished RIPK4 phosphorylation in HEK-293T and Caov-3 cells (Fig. 6J, K). Moreover, the phosphorylation-deficient S420A of RIPK4 dramatically shortened its half-life, while the phosphorylation mimetic of RIPK4 (S420D) observably extended the stability of RIPK4 in SK-OV-3 and Caov-3 cells (Fig. S5A-C). These findings suggested that GSK3β-induced RIPK4 phosphorylation at the S420 site served as another PTM to control RIPK4 protein expression and stability.

Given that the MD regions of RIPK4 were identified as a coincident domain for RIPK4 interactions with GSK3β and UCHL3, we speculated that GSK3β-mediated phosphorylation of RIPK4 might affect the interaction of UCHL3 with RIPK4. In accord with our hypothesis, GSK3β knockdown dramatically restrained the interaction of exogenous and endogenous RIPK4 with UCHL3 (Fig. 6L, M). In keeping with this result, the phosphorylation-inactivated GSK3β KD led to a reduced interaction between RIPK4 and UCHL3, while the phosphorylation-activated GSK3β CA observably enhanced their interaction in HEK-293T and Caov-3 cells Fig. S5D, E). Moreover, GSK3β knockdown dramatically shortened the half-life of RIPK4 in UCHL3-overexpressing Caov-3 cells (Fig. S5F, G). We further investigated whether GSK3β-mediated phosphorylation affected RIPK4 ubiquitination. We found that individual knockdown of GSK3β dramatically inhibited UCHL3-induced deubiquitination of RIPK4 in HEK-293T and Caov-3 cells (Fig. 6N, O). We finally explored whether the phosphorylation site Ser420 of RIPK4 is linked to the GSK3β-mediated interaction of RIPK4 with UCHL3. We found that the phosphorylation-deficient S420A of RIPK4 dramatically reduced the GSK3β-enhanced interaction between RIPK4 and UCHL3 (Fig. 6P, Q). Moreover, we observed that GSK3β could not enhance the UCHL3-induced deubiquitination of RIPK4 when HEK-293T and Caov-3 cells were transfected with the phosphorylation-deficient S420A of RIPK4 (Fig. 6R, S). Together, these results suggest that RIPK4 phosphorylation at the Ser420 site mediated by GSK3β promotes RIPK4 interaction with UCHL3, leading to enhanced stabilization of RIPK4 in ovarian cancer.

RIPK4 protein expression positively correlates with UCHL3 and predicts a poor prognosis in ovarian cancer patients

To further clarify the protein expression and correlation of UCHL3 and RIPK4 in ovarian cancer, we first performed western blotting analysis in the ovarian normal and cancer cell lines. We found elevated expression of RIPK4 and UCHL3 in ovarian cancer cell lines compared with normal cells (IOSE80 and HOSE) (Fig. 7A). In addition, UCHL3 showed a positive correlation with RIPK4 expression in ovarian cell lines (Fig. 7B, P = 0.0199, Pearson r = 0.75). In keeping with these data, we found significantly higher UCHL3 and RIPK4 expression in ovarian cancer tissues than in normal tissues (Fig. 7C-E), and a moderately positive correlation was observed between UCHL3 and RIPK4 protein levels (Fig. 7F, P < 0.0001, Pearson r = 0.7276). We subsequently performed immunohistochemical (IHC) staining for UCHL3 and RIPK4 in tissue microarrays containing samples from human ovarian cancer patients with available survival data. We observed that the expression of UCHL3 was positively correlated with the RIPK4 protein in ovarian cancer (Fig. 7G, H, P < 0.0001, Pearson r = 0.3415). Furthermore, Kaplan-Meier plotter data demonstrated that high protein levels of UCHL3 or RIPK4 were significantly associated with shorter OS in ovarian cancer patients (Fig. 7I, J, P < 0.0001). Taken together, these data indicate that RIPK4 protein levels are positively associated with UCHL3 levels, and elevated RIPK4 or UCHL3 expression is strongly associated with poor prognosis in ovarian cancer patients.

Fig. 7: RIPK4 protein expression positively correlates with UCHL3 and predicts a poor prognosis in ovarian cancer patients.
figure 7

A Western blotting analysis of RIPK4 and UCHL3 expression in the ovarian normal (IOSE80 and HOSE) and cancer cell lines (A2780, SK-OV-3, Caov-3, HEY, ES-2, OVCAR-8 and SW626). B Pearson correlation analysis of UCHL3 and RIPK4 protein levels in the ovarian normal and cancer cell lines. C Western blotting analysis of RIPK4 and UCHL3 expression in ovarian normal (N) and tumor (T) tissues. D, E Statistical analysis of UCHL3 and RIPK4 expression in ovarian N and T tissues. ***P < 0.001. F Pearson correlation analysis of UCHL3 and RIPK4 protein levels in ovarian tissues. G Representative images of UCHL3 and RIPK4 staining in tissue microarrays of ovarian cancer analyzed by IHC staining. Scale bar, 400 µm (left), 50 µm (right). H Pearson correlation analysis of UCHL3 and RIPK4 protein levels in the ovarian cancer tissue microarray. I, J Kaplan-Meier plots of OS in ovarian cancer patients dichotomized into ‘high’ and ‘low’ based on median UCHL3 and RIPK4 expression scores. Survival differences were calculated with the log-rank test. K Schematic of the PTM mechanism of RIPK4 in ovarian cancer. UCHL3 deubiquitinates and stabilizes RIPK4. Importantly, GSK3β-induced phosphorylation of RIPK4 enhances the interaction of RIPK4 with UCHL3, resulting in enhanced deubiquitination and stabilization of RIPK4 to accelerate ovarian cancer progression.

Discussion

Our study clarifies the mechanism by which phosphorylation and deubiquitination govern RIPK4 stabilization to promote tumor metastasis in ovarian cancer. We primarily identified UCHL3 as a binding protein for stabilizing RIPK4 through deubiquitination. More importantly, GSK3β-induced phosphorylation of RIPK4 at the Ser420 site enhanced the interaction of UCHL3 with RIPK4, resulting in increased deubiquitination and stabilization of RIPK4. Functional analyses revealed that RIPK4 functioned as a tumor promoter to accelerate tumor progression in a UCHL3-dependent manner in ovarian cancer cell lines and xenografted nude mice. Finally, we observed that RIPK4 protein expression was positively associated with UCHL3, and both were identified as predictors of poor prognosis in patients with ovarian cancer. Our findings revealed that UCHL3/RIPK4 may be a promising therapeutic target for ovarian cancer.

Ample evidence has revealed that RIPK4 is frequently overexpressed and promotes tumor progression in various cancers [28, 29]. However, no study has explored the PTM mechanisms that affect the expression and stability of RIPK4 under certain circumstances in ovarian cancer. In the current study, we identified the deubiquitinase UCHL3 as a RIPK4-interacting partner. Moreover, UCHL3 directly deubiquitinated RIPK4 at the K469 site by removing K48-linked ubiquitin chains to protect RIPK4 against proteasome-mediated degradation in ovarian cancer. In keeping with our results, Ren et al. provided compelling evidence confirming that UCHL3 could negatively regulate proteasome activity by trimming K48 ubiquitin chains on proteasome-bound substrates [30].

UCHL3 serves as a putative oncogene, as evidenced by its overexpression and positive association with unfavorable prognosis in multiple human cancers, such as cervical, anaplastic thyroid, and ovarian cancers [23, 31, 32]. Our findings agree with these data and reiterate the upregulation, protumorigenic functions and prognostic significance of UCHL3 in ovarian cancer. Our functional experiments revealed that RIPK4 functioned as a novel downstream target of UCHL3 to promote the proliferation and metastasis of ovarian cancer, as evidenced by the fact that reintroduction of RIPK4 largely attenuated the biological effects of UCHL3 knockdown. Moreover, we observed a strong positive correlation between UCHL3 and RIPK4 protein levels in ovarian cancer, and UCHL3high or RIPK4high expression predicted a poor prognosis. These findings support the notion that UCHL3 likely serves as a protumorigenic mediator of RIPK4 at the protein level in ovarian cancer. Notably, some studies have indicated that UCHL3 overexpression renders anaplastic thyroid [23] and breast cancer [33] cells resistant to chemotherapy. Although we have demonstrated the role of the UCHL3-RIPK4 axis in promoting cell proliferation and metastasis, the correlation and specific posttranslational regulatory mechanism of UCHL3 associated with chemoresistance in ovarian cancer remain largely unclear and warrant further exploration. In addition, most studies have shown that ovarian cancer progression is related not only to genetic information but also to their own energy metabolism, such as glycometabolism [34, 35]. Fan et al. reported that UCHL3 appeared to have a carcinogenic function by promoting aerobic glycolysis in pancreatic cancer [36]. However, whether UCHL3 could promote tumor metastasis by regulating glycometabolism in ovarian cancer remains unknown and requires further exploration. Moreover, the data analyzed from the PhosphoSitePlus® PTM database showed that phosphorylation was the common PTM for UCHL3. Ample evidence has revealed that UCHL3 phosphorylation promotes the cleavage activity of polyubiquitin chains and enhances the stability and interaction of proteins [30, 33]. However, the mechanism of phosphorylation activation of UCHL3 and its effect on the stability of interacting proteins remain unknown in ovarian cancer, which warrants further experimental exploration.

Another important finding of the current study was the functional link between phosphorylation and ubiquitination in governing the stability of RIPK4 in ovarian cancer. On the one hand, phosphorylation is known to prevent or promote ubiquitination to mediate substrate degradation. For instance, Xiao et al. reported that ERK-mediated phosphorylation of PD-1 at Thr234 was essential for inhibiting PD-1 ubiquitination and subsequently enhancing its stabilization [10]. Conversely, Wang et al. suggested that the Ser79 phosphorylation of PROX1 induced by AMPK promoted PROX1 ubiquitination and degradation [37]. On the other hand, phosphorylation can regulate substrate ubiquitination, independent of proteolysis. For example, Lu et al. reported that RECQL4 phosphorylation caused by CDK1/2 enhanced the MRE11/RECQL4 interaction and RECQL4 recruitment to DNA double-strand breaks to prevent cellular senescence [38]. We found that phosphorylation of RIPK4 at the Ser420 site induced by GSK3β promoted RIPK4 interaction with UCHL3, leading to enhanced deubiquitination and stabilization of RIPK4 in ovarian cancer. Conversely, GSK3β inhibition can reverse this process to destabilize RIPK4. GSK-3β is a highly conserved serine/threonine kinase that phosphorylates glycogen synthase and participates in almost every cell physiological process [39]. Most studies have indicated that GSK3β-mediated phosphorylation interacts with ubiquitination to modulate the stabilization of individual target proteins. For example, Li et al. reported that GSK3β increased myeloid cell leukemia-1 (MCL-1) Ser159 phosphorylation and ultimately induced MCL-1 ubiquitination and degradation to reverse radioresistance in oral squamous cell carcinoma [40]. Zhang et al. revealed that GSK3β was able to phosphorylate DHX33 at T482 to mediate protein ubiquitination and degradation in lung cancer [41].

Inevitably, we note several limitations to our study. First, RIPK4 serves as a serine/threonine kinase and could function as a regulator involved in mediating the stabilization of other proteins to modulate tumor development. For example, phosphorylation of plakophilin-1 by RIPK4 regulates epidermal differentiation and skin tumorigenesis [42]. Huang et al. reported that phosphorylation of DVL2 by RIPK4 favored canonical Wnt signaling, which might contribute to the growth of certain tumor types [43]. Inadequately, although we characterized the molecular events leading to RIPK4 stability, the targets or pathways that could be activated following RIPK4 stabilization were not investigated, which warrants further exploration. Second, ovarian cancer is typically diagnosed at a late stage with peritoneal metastases in approximately 75% of cases [44]. However, the mouse model used in our study only exhibited significant metastatic changes in the lungs. Therefore, to further replicate the clinical features, a mouse model of ovarian cancer peritoneal metastases should be established in future research. Third, emerging evidence has demonstrated that UCHL3 maintains cancer stem cell properties in various cancers, such as prostate and gastric cancers [45, 46]. However, whether a similar function of UCHL3 exists in stem cells of ovarian cancer remains largely unknown and requires further exploration. Fourth, to systematically evaluate the role of RIPK4 in ovarian cancer, a more efficient inhibitor should be explored. In line with our ideas, Liu et al. applied RIPK4 small interfering RNA for bladder cancer therapy [47]. Finally, we cannot rule out that other uncharacterized mediators beyond UCHL3 for RIPK4 exist in ovarian cancer, which warrants further exploration.

In summary, we demonstrated that UCHL3 interacted with, deubiquitinated and stabilized RIPK4 to promote tumor metastasis of ovarian cancer. More importantly, we verified that GSK3β-induced phosphorylation of RIPK4 at the Ser420 site enhanced the interaction between UCHL3 and RIPK4, resulting in increased deubiquitination and stabilization of RIPK4 (Fig. 7K). Therefore, our results suggest that suppressing UCHL3 might represent a promising alternative approach to effectively inhibit RIPK4-mediated tumor progression. Generally, RIPK4 is a promising drug target for cancer therapy, especially in patients with high UCHL3 expression. Thus, targeting RIPK4 stabilization through UCHL3 inhibition represents a potential therapeutic strategy in ovarian cancer.

Materials and methods

Cells, antibodies, and reagents

Human HEK-293T, SK-OV-3, Caov-3, A2780, HEY, SW626, ES-2 and OVCAR-8 cells were obtained from Procell Life Science & Technology Co. LTD (Wuhan, China). The HOSE cell line was supplied by Yaji Biotechnology Co. LTD (Shanghai, China). The IOSE80 cell line was obtained from Fu Heng Biotechnology Co. LTD (Shanghai, China). All cell lines were tested for mycoplasma contamination and were confirmed by short tandem repeat (STR) DNA fingerprinting.

The following is the information on the antibodies used in the study. Protein A/G magnetic beads (Cat# HY-K0202), anti-HA magnetic beads (Cat# HY-K0201), anti-His magnetic beads (Cat# HY-K0209), anti-Flag magnetic beads (Cat# HY-K0207), anti-GST magnetic beads (Cat# HY-K0222) and anti-Myc magnetic beads (Cat# HY-K0206) were purchased from MedChemExpress (MCE, Shanghai, China). GST-Tag mouse mAb (Cat# 2624), GST-Tag rabbit mAb (Cat# 2625), His-Tag antibody (Cat# 2365), His-Tag rabbit mAb (Cat# 12698), HA-Tag rabbit mAb (Cat# 3724), Myc-Tag mouse mAb (Cat# 2276), Myc-Tag rabbit mAb (Cat# 2278), His-Tag mouse mAb (Cat# 2366), ubiquitin mouse mAb (Cat# 3936), ubiquitin rabbit mAb (Cat# 20326), p-Ser (Cat# 9615), ubiquitin mouse mAb (Cat# 14049) and ubiquitin rabbit mAb (Biotinylated, Cat# 91112) were purchased from Cell Signaling Technology (CST, Boston, USA). UCHL3 rabbit pAb (Cat# 12384-1-AP), GSK3β rabbit mAb (Cat# 22104-1-AP), β-actin rabbit mAb (Cat# 81115-1-RR), β-actin mouse mAb (Cat# 66009-1-Ig), HRP-conjugated AffiniPure goat anti-rabbit IgG (H + L) (Cat# SA00001-2) and HRP-conjugated AffiniPure goat anti-mouse IgG (H + L) (Cat# SA00001-1) were obtained from Proteintech (Wuhan, China). RIPK4 mouse pAb (Cat# PA5-97085) was purchased from Invitrogen (California, USA).

The following is the information of reagents used in the study. MG-132 (Cat# HY-13259), CHX (Cat# HY-12320), RIPA lysis buffer (Cat# HY-K1001), penicillin-streptomycin (Cat# HY-K1006), serum/protein-free cell freezing medium (Cat# HY-K1012), Renilla-firefly luciferase dual assay kit (Cat# HY-K1013), SYBR Green qPCR master mix (Cat# HY-K0501) and RT master mix for qPCR II (Cat# HY-K0510A) were purchased from MCE. Fetal bovine serum (FBS, Cat# 16140071) was obtained from ThermoFisher Scientific (Waltham, USA). RPMI 1640 (Cat# PM150110), DMEM (Cat# PM150210) and McCoy’s 5A (Cat# PM150710) media were obtained from Procell Life Science & Technology Co. LTD. SDS loading buffer (Cat# P0015A), puromycin (Cat# ST551-10 mg), BeyoRT™ II cDNA (Cat# D7168M) and polybrene (Cat# C0351-1 ml) were purchased from Beyotime Biotechnology (Guangdong, China). A proximity ligation assay reagent (PLA, Cat# DUO92020) and DAPI (Cat# 28718-90-3) were obtained from Sigma (Missouri, USA). TCID (Cat# S7140) and LiCl (Cat# E0153) were purchased from Selleck (Texas, USA).

Single-cell sample preparation and sequencing

The matched primary and metastatic tumor tissues were collected from three ovarian cancer patients without treatment. Single-cell suspension were prepared using TrypLE for 10 min at 37 °C and quantified for viability using Vi-Cell XR (Beckman Coulter). Before proceeding, all samples were ensured to have >90% viability to minimize dead cell carryover in sequencing, and then were processed into nuclei according to the 10× Multiome ATAC + Gene Expression (GEX) protocol (CGOOO338). The samples were washed with PBS (with 0.04% BSA), lysed with chilled lysis buffer for 4 min, washed three times with wash buffer and resuspended in 10× nuclei buffer. Nuclear samples were processed using the Chromium 10× genomics instrument with a target cell number of 7000-10,000. Library preparations were established by the 10× Single Cell Multiome ATAC + Gene Expression v1 kit according to the manufacturer’s instructions. Library sequencing was performed using the NovaSeq 6000 System (Illumina).

Cell culture and virus infection

ES-2 and SK-OV-3 cells were cultured in McCoy’s 5A supplemented with 10% FBS and 1% penicillin-streptomycin. The other cell lines were cultured in RPMI 1640 or DMEM supplemented with 10% FBS and 1% penicillin-streptomycin. All cells were cultured at 37 °C in an atmosphere of 5% CO2. All cell lines were confirmed by STR DNA fingerprinting and tested for mycoplasma contamination.

HEK-293T cells were used to package lentiviruses to generate UCHL3, RIPK4 and other shRNA lentiviral vectors. shRNA oligonucleotides against human UCHL3, RIPK4 and GSK3β are listed in Supplementary Table 1. For each 10 cm dish, cells were cotransfected with 4 μg psPAX2, 3 μg pMD2.G and 4 μg pLKO.1-shGFP/target gene plasmids. Seventy-two hours posttransfection, medium with secreted virus particles was harvested and filtered using a 0.45 μm syringe filter. To generate cell lines with stable UCHL3 or RIPK4 overexpression, human UCHL3 or RIPK4 was inserted into the lentiviral expression vector pCDH-CMV-GFP puro. For infection of HEK-293T and ovarian cancer cells, cells were first coincubated with lentiviruses in the presence of polybrene (6 μg/mL). Forty-eight hours after infection, cells were selected using puromycin (2 μg/mL for ovarian cancer cells and HEK-293T cells) for 3 days. Lentivirally transduced cells were validated by western blotting.

Immunofluorescence assay

SK-OV-3, Caov-3, and OVCAR-8 cells were fixed in 4% paraformaldehyde and permeabilized with 0.1% Triton X-100 for 20 min. All cells were incubated with anti-UCHL3 and anti-RIPK4 antibodies overnight at 4 °C. Then, the cells were washed three times and incubated with the appropriate Alexa Fluor 488 or Alexa Fluor 594-conjugated secondary antibodies for 2 h at room temperature. Nuclei were stained with DAPI. Respective images were taken under a fluorescence microscope (Zeiss LSM900, Oberkochen, Germany).

In situ PLA

PLA was performed using a DUOLinkTM kit (OLINK; Uppsala, Sweden) according to the manufacturer’s instructions. Briefly, SK-OV-3, Caov-3 and OVCAR-8 cells were fixed with cold methanol for 1 h at 4 °C. After permeabilization with 0.1% Triton X-100 for 1 h and blocking with 5% BSA for 1 h, the cells were incubated with antibodies against UCHL3 and RIPK4 at 4 °C overnight. Cells were treated with PLA plus and minus probes and incubated for 1 h at 37 °C. The probes were hybridized using a ligase to form a closed circle. The cells were then stained with DAPI to visualize nuclear DNA. Respective images were detected by a fluorescence microscope (Zeiss LSM900, Oberkochen, Germany).

Immunoprecipitation coupled with MS analysis for identifying RIPK4-interacting proteins

Caov-3 cells transfected with RIPK4-overexpressing plasmids were lysed. The supernatants of cell lysates were incubated with anti-RIPK4 in a rotating incubator for 4 h at 4 °C. The resin was washed five times with NETN buffer (20 mM Tris, pH 8.0, 100 mM NaCl, 0.5% NP-40, and 1 mM EDTA). The samples prepared from immunoprecipitation were sent to the MS facility. MS analysis and data processing were performed by an MS specialist from the MS facility at Jingjie Biotechnology Co. Ltd. (Hangzhou, China). Three biological replicates were performed for the experiment.

Protein half-life analysis

OVCAR-8, SK-OV-3, and Caov-3 cells were treated with 50 µM CHX for the indicated time points and harvested for immunoblot analysis using the indicated antibodies. The protein level was quantified using ImageJ. The quantitative value of the objective protein band was normalized to RIPK4/β-actin and then compared to the 0 h point.

Quantitative reverse transcription PCR (qRT-PCR) analysis

Total RNA from cells was extracted using TRIzol reagent (Invitrogen), and reverse transcription reactions were performed using BeyoRT™ II cDNA. After mixing the cDNA templates with primers and SYBR qPCR Master Mix, reactions were measured by a 7500 Real-Time PCR System (Applied Biosystems). The following primers were used for qRT-PCR. For human UCHL3: 5-CAAACAATCAGCAATGCCTGTGG-3 (forward), 5-GGCTCATTGACACAGATTCCTCC-3 (reverse); for human RIPK4: 5-CTGAAACCGAGGACCTGTGTGA-3 (forward), 5-GCTGTAGTCGTTATCGAAGGTGG-3 (reverse); for human GAPDH: 5-GGAGCGAGATCCCTCCAAAAT-3 (forward), 5-GGCTGTTGTCATACTTCTCATGG-3 (reverse).

Western blotting

The lysates from cell lines or tissues were mixed with SDS loading buffer for boiling for 10 min. Protein samples were loaded onto 12-well SDS-PAGE, electrophoresed, and then transferred to 0.45 μm PVDF membranes (Millipore, USA). The membranes were blocked with TBST containing 5% nonfat milk at room temperature for 1 h, and blotting was performed with the primary antibodies anti-β-actin, anti-UCHL3, or anti-RPIK4 or other indicated antibodies at 4 °C overnight, followed by the secondary antibodies anti-rabbit or anti-mouse IgG. The membrane was incubated with enhanced chemiluminescence (ECL) to acquire images (Bio-Rad, USA).

GST pull-down assay

Myc-RIPK4 protein was purified from HEK-293T cells. Recombinant GST-UCHL3 and GST-UCHL3 C95S were purified from Escherichia coli. GST pull-down assays were performed by incubating GST, GST-UCHL3 or GST-UCHL3 C95S immobilized on glutathioneSepharose resin with Myc-RIPK4 at 4 °C. GST pull-down products were washed five times, and SDS loading buffer was added for boiling 10 min before running SDS-PAGE and analyzed by immunoblotting with the indicated antibodies.

Co-IP assay

Cells were lysed with 100 μl of lysis buffer and immunoprecipitated with HA, Myc, RIPK4, UCHL3 or GSK3β antibodies at 4 °C overnight. Protein A/G magnetic beads were then added to the lysates and incubated at 4 °C for 3 h. The samples were added to SDS loading buffer for boiling for 10 min. Then, these samples were resolved by SDS-PAGE and analyzed by immunoblotting with the indicated antibodies.

In vivo deubiquitination assay

Cells were lysed with 100 μl of lysis buffer and boiled for 20 min. Next, 900 μl of dilution buffer, containing 150 mM NaCl, 1% Triton X-100, 2 mM EDTA, and 10 mM Tris-HCl (pH 8.0), was added. Samples were incubated with anti-RIPK4 or anti-Myc beads at 4 °C for 90-120 min with rotation. The beads were boiled after extensive washing with buffer containing 1 M NaCl, 1% NP-40, 1 mM EDTA, and 10 mM Tris-HCl (pH 8.0). The samples were added to SDS loading buffer for boiling for 10 min. Then, these samples were resolved by SDS-PAGE and analyzed by immunoblotting with the indicated antibodies.

In vitro deubiquitination assay

HA-UCHL3 and HA-UCHL3 C95S were purified from HEK-293T cells. Ubiquitinated Myc-RIPK4 was purified from HEK-293T cells previously cotransfected with UCHL3 shRNA.1, Myc-RIPK4 and His-Ub. HA-UCHL3 or HA-UCHL3 C95S was then incubated with Myc-RIPK4 in deubiquitylation buffer (50 mM Tris-HCl, pH 7.4, 1 mM MgCl2 and 1 mM DTT) at 37 °C for 2 hours. Western blot analysis was then performed. The samples were added to SDS loading buffer for boiling for 10 min. Then, these samples were resolved by SDS-PAGE and analyzed by immunoblotting with the indicated antibodies.

Cell colony formation assay

SK-OV-3 and OVCAR-8 cells were seeded in 6-well plates at an initial density of 1200 cells/well and grown for 14 days. The colonies were fixed with methanol and stained with 0.1% crystal violet for 15 min at room temperature. The plates were photographed after washing.

Transwell assays

SK-OV-3 and OVCAR-8 cells were seeded in 24-well plates at an initial density of 8×104 cells/well. Cells were resuspended in serum-free medium, and then 200 μL of the cell suspension and 800 μL of medium with 20% FBS were added to the upper and lower chambers with 60 μl of 1 mg/ml Matrigel (Millipore, USA), respectively. The migrated cells that reached the lower surface of the membrane were fixed with methanol after 48 h and stained with 0.1% crystal violet. The plates were photographed after extensive washing, and the number of invaded cells was quantified.

Human tissues and tumor tissue microarray

Ovarian cancer and normal tissues from patients were obtained from the First Affiliated Hospital of USTC with patient consent. The use of pathological specimens and the review of all pertinent patient records were approved by the Institutional Ethics Committee of the First Affiliated Hospital of USTC. Additionally, the tissue microarray of ovarian cancer was purchased from Shanghai Outdo Biotech (Shanghai, China). The microarray with available survival data contained 2 benign tumors, 2 adjacent nontumors, 135 primary tumors, and 12 metastatic tumor samples.

Hematoxylin-eosin (HE) staining analysis

For HE staining, lung tissues from each mouse were fixed in 4% paraformaldehyde and embedded in paraffin blocks. The paraffin blocks were sectioned (5 μm) for HE staining (Beyotime Biotech) followed by coverslipping. We applied an Aperio VERSA 8 (Leica) multifunctional scanner to acquire images.

IHC staining analysis

IHC staining for RIPK4 and UCHL3 was performed on the tissue microarray. The protein expression was estimated based on the intensity and extent of stained tumor cells. Briefly, the staining intensity was assessed by two experienced pathologists. The staining intensity was scored as follows: 0, negative; 1, weak; 2, moderate and 3, strong. The staining extent was graded from 0 to 4 based on the percentage of immunoreactive tumor cells: 0 (0-4% positive cells), 1 (5-24% positive cells), 2 (25-49% positive cells), 3 (50-74% positive cells) or 4 (75-100% positive cells). The integrated score was calculated by multiplying the extent score and the intensity score, resulting in a low level or a high level for each sample.

Tumor xenograft mouse models

Female BALB/c nude mice (4-5 weeks old, USTC) were raised in specific pathogen-free conditions and randomized into groups. The animal care and experimental protocols were in accordance with guidelines established by the Animal Care and Use Committee of the First Affiliated Hospital of USTC. SK-OV-3 cells (5 × 106 cells/mouse) transfected with Scr-shRNA, UCHL3 shRNA.1 or UCHL3 shRNA.1 + RIPK4 plasmids were subcutaneously injected into nude mice (ten female mice/group). Similarly, Caov-3 cells (5 × 106 cells/mouse) transfected with Scr-shRNA, UCHL3, or UCHL3 + RIPK4 shRNA.1 plasmids were subcutaneously injected into nude mice (ten female mice/group). Tumor volume was measured every three days and was calculated according to the following formula: (length × width2) × 0.5. After 28 days, five mice in each group were sacrificed, and tumor samples were collected for further analysis. After 42 days, the other five mice in each group were sacrificed. Lung tissues were embedded in paraffin for HE staining, and the number of metastatic nodules was calculated. Investigators who conducted the animal experiments were blinded to allocation during experiments and outcome assessments.

Statistical analysis

Statistical analyses were conducted using SPSS 23.0. Quantitative data are presented as the mean ± SD, as indicated by at least three independent experiments. Statistical significance was determined using Student’s t-test for comparisons between two groups and one-way analysis of variance (ANOVA) for comparisons among more than two groups. Pearson’s correlation analysis was used for statistical analysis of the correlation between UCHL3 and RIPK4 protein levels. The Kaplan-Meier method was adopted to generate graphs, and the survival curves were analyzed with the log-rank test. P values < 0.05 were considered statistically significant.