RNF31 promotes proliferation and invasion of hepatocellular carcinoma via nuclear factor kappaB activation

RNF31 is a multifunctional RING finger protein implicated in various inflammatory diseases and cancers. It functions as a core component of the linear ubiquitin chain assembly complex (LUBAC), which activates the nuclear factor kappaB (NF-κB) pathway via the generation of the Met1-linked linear ubiquitin chain. We aimed to clarify the role of RNF31 in the pathogenesis of hepatocellular carcinoma (HCC) and its relevance as a therapeutic target. High RNF31 expression in HCC, assessed by both immunohistochemistry and mRNA levels, was related to worse survival rates among patients with HCC. In vitro experiments showed that RNF31 knockdown in HCC cell lines led to decreased cell proliferation and invasion, as well as suppression of tumor necrosis factor (TNF)-α-induced NF-κB activation. Treatment with HOIPIN-8, a specific LUBAC inhibitor that suppresses RNF31 ubiquitin ligase (E3) activity, showed similar effects, resulting in decreased cell proliferation and invasion. Our clinical and in vitro data showed that RNF31 is a prognostic factor for HCC that promotes tumor aggressiveness via NF-κB activation.

immunohistochemical (IHC) staining.The number of cases classified in RNF31 expression grades 0/+1/+2 were 14/50/20, respectively (Fig. 1a).Sixty-four patients were classified as the low RNF31 expression group (grades 0 and +1) and 20 patients were classified in the high RNF31 expression group (grade +2).The 5-year overall survival rate was significantly better in the low RNF31 expression group than in the high RNF31 expression group (66.45% vs. 30.85%;P = 0.021) (Fig. 1b).The low RNF31 expression group had a better 5-year recurrence-free survival rate than the high RNF31 expression group (54.44% vs. 22.34%;P = 0.083).There was also a significant difference in overall survival when RNF31 expression was stratified into grades 0, +1, and +2 groups (P = 0.028) (Fig. S1).We confirmed these findings by analyzing the relationship between RNF31 mRNA expression and survival using data from The Cancer Genome Atlas database (n = 365) (Fig. 1c) 21 .Consistent with IHC results, high RNF31 mRNA expression was associated with significantly worse survival (P = 0.009).

Figure 1.
Correlation between RNF31 expression and survival of patients with hepatocellular carcinoma.(a) Representative immunohistochemical (IHC) staining of RNF31 in cancerous areas of an HCC specimen (original magnification, ×200).RNF31 expression was scored as grades 0, +1, and +2 according to staining intensity.(b) Overall survival and recurrence-free survival according to RNF31 IHC expression.(c) Survival according to RNF31 mRNA expression in The Cancer Genome Atlas database of patients with liver cancer (image credit: Human Protein Atlas, image available from v22.0 proteinatlas.org).OS, overall survival; RFS, recurrence-free survival.

RNF31 knockdown suppresses the proliferation and invasion of HCC cell lines.
The in vitro effects of RNF31 on HCC cells were analyzed next.First, we evaluated the mRNA and protein levels of RNF31 in six HCC cell lines, in addition to RBCK1 and SHARPIN, the two other components of LUBAC.RNF31 mRNA expression was highest in HepG2 cells, followed by Hep3B and PLC cells (Fig. 2a).Protein expression of RNF31 was also highest in HepG2 cells, followed by HLF and Hep3B cells (Fig. 2b, Fig. S3a).Both mRNA and protein expression of RBCK1 were highest in HepG2 cells (Fig. S4a, Fig. 2b, Fig. S3b).SHARPIN mRNA expression levels were particularly high in Hep3B and PLC, followed by HepG2 cells (Fig. S4b).SHARPIN protein expression was highest in HepG2 cells (Fig. 2b, Fig. S3c).For subsequent experiments, we selected HepG2, Hep3B, and PLC cells, in which the mRNA level of RNF31 was high and the other LUBAC components, RBCK1 and SHARPIN, were expressed in a balanced pattern.We performed RNF31 knockdown using three independent siRNAs, which induced a substantial reduction of RNF31 mRNA and protein expression in HepG2, Hep3B, PLC (Fig. 2c,d).RNF31 knockdown also reduced the protein level of RBCK1 in HepG2, Hep3B, and PLC (Fig. S5a).Protein level of SHARPIN was also decreased by RNF31 knockdown in HepG2 and PLC, but not in Hep3B cells (Fig. S5b).In all three cell lines, RNF31 knockdown significantly decreased the cell proliferation rate (P < 0.05) (Fig. 3a).Invasion was also significantly suppressed in RNF31-silenced cells (P < 0.05) (Fig. 3333b).

HOIPIN-8 suppresses the proliferation and invasive potential of HCC cell lines.
Next, we assessed the effects of HOIPIN-8, an α,β-unsaturated carbonyl-containing LUBAC-specific inhibitor 22,23 , on HCC cell lines.Similar to RNF31 knockdown, HOIPIN-8 significantly suppressed the proliferation rate of HCC cell lines (P < 0.05) (Fig. 5a).The IC50 of HOIPIN-8 on HepG2 and Hep3B cells was about 70-100 µM (Supplementary Table S2) and invasion was significantly suppressed at lower concentrations such as 10 or 30 µM. (P < 0.05) (Fig. 5b).HOIPIN-8 further suppressed the induction of NF-κB target genes (interleukin [IL]-6, IL-8, and BIRC3) (Fig. 5c) and the phosphorylation of NF-κB signaling factors upon TNF-α stimulation (Fig. 5d, Fig. S7).To understand if the effect of HOIPIN-8 on HCC cell lines was reversible, HOIPIN-8 was washed out after 24 h, and the subsequent proliferation rate was analyzed for 96 h.Cells that had wash-out of HOIPIN-8 showed a small but significantly higher proliferation rate when compared to cells which underwent continuous treatment with HOIPIN-8 (P < 0.05, Fig. S8).Lastly, we verified whether HOIPIN-8 could affect apoptosis induced by TNF-α and cycloheximide.The combined treatment of HOIPIN-8 with TNF-α and cycloheximide markedly enhanced the cleavage of PARP and caspase 3, suggesting that induction of the extrinsic apoptotic pathway could be enhanced by HOIPIN-8 (Fig. 5e).Similarly, the rate of apoptosis using the Annexin V/PI www.nature.com/scientificreports/apoptosis assay showed that late apoptosis was significantly increased in the combination treatment of HOIPIN-8 with TNF-α and cycloheximide (Fig. S9).

Discussion
The key findings of our study were as follows: RNF31 was associated with poor survival and was an independent prognostic factor for HCC.Knockdown of RNF31 decreased proliferation and invasion in HCC cell lines with decreased NF-κB activation upon TNF-α stimulation.Treatment with HOIPIN-8, a LUBAC-specific inhibitor, had similar inhibitory effects on cell proliferation, invasion, and NF-κB activation as those of RNF31 knockdown.Based on these results, we concluded that RNF31 is a prognostic biomarker and a potential therapeutic target for HCC.
Our analysis of patients with HCC showed a significant correlation between survival and RNF31 expression as assessed by immunohistochemistry.These findings were also confirmed by an analysis of The Cancer Genome Atlas database, which uses mRNA levels to evaluate RNF31 expression.Because RNF31 plays an important role in inflammatory signaling as a component of LUBAC, we initially expected that RNF31 could be related to an inflammatory background, such as liver cirrhosis or chronic viral infections.Shimizu et al. reported that depletion of RNF31 led to chronic inflammation and apoptosis in mouse liver parenchyma, resulting in increased hepatocarcinogenesis 24 .However, in our cohort, RNF31 expression was not correlated with viral hepatitis or liver cirrhosis.Chen et al. also reported that there was no correlation between RNF31 expression and HBV or cirrhosis 25 .
As a mechanism of RNF31 in HCC, Chen et al. suggested that RBCK1 could promote HCC metastasis and growth by stabilizing RNF31 25 .Because RNF31 and RBCK1 are both essential components of the LUBAC, we suspected that the stabilization of RNF31 by RBCK1 could increase LUBAC formation and subsequently trigger enhanced NF-κB activation via linear ubiquitination.Consistent with this, Chen et al. showed that gliotoxin, an inhibitor of LUBAC activity, has inhibitory effects on HCC cell lines 25 .However, they did not clarify whether the inhibitory effect of gliotoxin was due to LUBAC or NF-κB activity, as gliotoxin itself has cytotoxicity 23 .In this study, we showed that RNF31 knockdown could actually inhibit TNF-α-induced NF-κB activation.Furthermore, we used HOIPIN-8, a LUBAC-specific inhibitor, which suppresses the E3 activity of RNF31 by modifying the active Cys885 22,23 .This leads to robust inhibition of both LUBAC activation and the NF-κB pathway with low cytotoxicity 23 .Similar to the results of RNF31 knockdown, the treatment of HCC cell lines with HOIPIN-8 also resulted in decreased cell proliferation, invasion, NF-κB activation, and enhanced apoptosis.Therefore, reducing the expression and activity of LUBAC can be considered as effective treatment for HCC.
NF-κB plays a crucial role in cancer development and progression 10,13,26 .Therefore, various NF-κB inhibitors have been developed and investigated for their anti-tumor effects.However, the clinical application of these inhibitors is limited due to side effects and issues related to drug dosage.In addition to their effect on tumors, NF-κB inhibitors may reduce systemic immunity to infections and other diseases and, more importantly, could negatively impact anti-tumor immunity.In this study, we used HOIPIN-8, a refined derivative of HOIPIN-1, which was developed as a specific inhibitor of LUBAC.HOIPIN-1 and -8 effectively induced cell death in activated B cell-like diffuse large B cell lymphoma cells (ABC-DLBCL) 22,23 .In addition to its anti-tumor effect, HOIPIN-1 could enhance anti-tumor immunity in melanoma models, even when T cells were pretreated with HOIPIN-1 27 .Interestingly, Frey et al. showed that RNF31 depletion in pancreatic ductal carcinoma cell lines could increase both the infiltration and effector functions of CD8 + T cells in an orthotopic tumor model 28 .Unfortunately, we could not validate this effect of RNF31 on anti-tumor immunity in our cohort of patients with HCC, since we identified no correlation between tumor-infiltrating CD8 + cells and levels of RNF31 expression (Table 1).
In conclusion, our study demonstrated that RNF31 expression is an important prognostic factor for HCC, and RNF31 is involved in HCC tumor progression via the NF-κB pathway.These results suggest that RNF31 may be a potential therapeutic target for HCC as well as a novel biomarker.One limitation of this study was that it was a single-center, retrospective study with a limited sample size, which may have biased the clinical results.

Patients and samples
We analyzed tumor tissues of 84 patients with HCC who underwent surgical treatment at the Department of General Surgical Science Gunma University (Maebashi, Japan) between 1996 and 2014.The tumor stage was classified according to the 6th Japanese Tumour-Node-Metastasis (TNM) Classification of the Liver Cancer Study Group of Japan 29 .All clinical samples and patient data were analyzed following Gunma University Hospital's institutional guidelines (approval number: HS2020-124) and the Declaration of Helsinki, with a waivor of informed consent, using the opt-out method.All experimental protocols were approved by Gunma University Ethical Review Board for Medical Research Involving Human Subjects.Patient consent was obtained by using the opt-out method.

Tissue microarrays
Formalin-fixed paraffin-embedded samples obtained from HCC patients were stored in the Clinical Department of Pathology at Gunma University Hospital.The formalin-fixed paraffin-embedded tissue blocks were marked with two representative tumor areas after examining the slides stained with hematoxylin and eosin.Tumor cores with a diameter of 2.0 mm were removed using a cylinder.A manual arraying instrument (Beecher Instruments, Sun Prairie, WI, USA) was used to assemble paraffin blocks into tissue microarrays as previously described 30 .

Immunohistochemistry analysis
The tissue microarray blocks were cut into 4-µm-thick slices, and immunostaining was performed using a primary antibody against RNF31 (1:200; anti-HOIP antibody ab133818; Abcam, Cambridge, UK).Each mounted section was de-paraffinized, rehydrated, and incubated with 0.3% hydrogen peroxide in methanol for 30 min at room temperature to block the endogenous peroxidase activity.Antigen retrieval for RNF31 was performed by boiling slides in 10 mmol/L citrate buffer (pH 6.0) at 98 ℃ for 45 min.Non-specific binding sites were blocked for 30 min at room temperature using Protein Block Serum-Free (Dako, Osaka, Japan).The sections were then incubated with a primary antibody for 24 h at 4 ℃.Following washing with phosphate-buffered saline (PBS), they were coated with Histofine Simple Stain MAX-PO (Multi) Kit (Nichirei Biosciences, Tokyo, Japan) for 1 h at room temperature.Chromogen 3,3′-diaminobenzidine tetrahydrochloride (Dojindo Laboratories, Kumamoto, Japan) was applied as a 0.02% solution containing 0.005% H 2 O 2 in 50 mM ammonium acetate-citrate acid buffer (pH 6.0).Mayer's hematoxylin was lightly counterstained and mounted on each section.Negative controls consisted of PBS containing 0.1% bovine serum albumin instead of the primary antibody.The degree of cytoplasmic staining was evaluated in three levels.The cells with no staining at all were defined as grade 0, cells with clear staining as grade +2, and cells with weak staining (between grades 0 and +2) as grade +1.Grades 0 and +1 were defined as low expression and grade +2 was defined as high expression.Intratumoral CD8-positive cytotoxic lymphocytes (CD8 + ) were counted using light microscopy at 200× magnification in three selected hotspots.We defined patients with a cytotoxic lymphocyte count greater than 10 as belonging to the high cytotoxic lymphocyte infiltration group in our previous study 31 .The Ki-67 labelling index was calculated as the percentage of positive tumor cell nuclei, using more than 500 tumor cells, regardless of staining intensity 31 .

Cell cultures
In this study, we used the human hepatoblastoma cell line HepG2 and the human HCC cell lines Hep3B, PLC/ PRF/5 (PLC), Huh7, HLE, and HLF.Cell lines were purchased from the JCRB Cell Bank and American Type Culture Collection (Manassas, VA, USA).All cells used in the experiments were free of mycoplasma contamination.The cells were cultured in Dulbecco's modified Eagle's medium (Wako, Richmond, VA, USA) supplemented

Immunoblotting analysis
Protein extraction was performed using a lysis buffer containing 50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1% Triton X-100, and a complete protease and phosphatase inhibitor cocktail for 20 min on ice.SDS-PAGE with 10% Bis-Tris gels was used to separate the proteins, which were then transferred to nitrocellulose sandwiches (#12369; Cell Signaling Technology, Danvers, MA, USA).After blocking with 5% bovine serum albumin or 5% skim milk for 1 h at room temperature, the membranes were incubated with primary antibodies overnight at 4 °C.A membrane used for detection of multiple proteins were cut prior to incubation with primary antibodies (i.e. a membrane would be cut at the height of 75 kilodalton, then the upper membrane used for immunoblotting of RNF31 and the lower membrane for RBCK1

Reverse-transcription quantitative polymerase chain reaction
RNA was extracted using an RNeasy Mini Kit (#217004; Qiagen, Venlo, the Netherlands) and quantified using an ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA).Reverse-transcription (RT) quantitative polymerase chain reactions (qPCRs) were performed using an RT kit (Toyobo, Osaka, Japan) and Power SYBR Green PCR master mix (Life Technologies, Rockville, MD, USA) according to the manufacturer's instructions.The qPCR was performed with a Step-One-Plus PCR system (Applied Biosystems, Wilmington, DE, USA) using the 2 −ΔΔCt method.The primers used are listed in Supplementary Table S4.GAPDH was used to normalize RNA input for all RT-qPCR analyses.

Statistical analysis
Statistical significance was analyzed using the Mann-Whitney U test, analysis of variance, or Welch's t-test for continuous variables; the χ 2 test or Fisher's exact test was performed for categorical variables.When the analyses of variance results were significant, Tukey's multiple comparison tests were performed to assess the differences between groups.Survival curves were calculated using the Kaplan-Meier method.Differences between the survival curves were analyzed using log-rank tests.Prognostic factors were examined using univariate and multivariate analyses and the Cox proportional hazards model.Results were considered statistically significant

Figure 2 .
Figure 2. Expression of RNF31 in HCC cell lines and evaluation of RNF31 knockdown.(a) The mRNA level of RNF31 was evaluated in the HepG2, Hep3B, PLC, Huh7, HLE, and HLF cell lines by qPCR.mRNA level of RNF31 in Huh7 cells was used as reference.(b) The protein expression levels of RNF31, RBCK1, and SHARPIN were evaluated by Western blotting.(c)The effect of RNF31 knockdown was evaluated using Western blotting for HepG2, Hep3B, and PLC cells transfected with control (siCont) and RNF31-specific siRNAs (siRNA1-3).For quantification, normalized RNF31/b-actin level in parental cells was used as reference.d.The effect of RNF31 knockdown was evaluated using qPCRs.mRNA level of RNF31/GAPDH in control cells was used as reference.*P < 0.05; ns, non-specific; n.s., not significant; siCont, control siRNA.

Figure 3 .
Figure 3. Knockdown of RNF31 leads to decreased cell proliferation and invasion in HCC cell lines.(a) The effect of RNF31 knockdown on the proliferation rate of HepG2, Hep3B, and PLC cells was evaluated at 24, 48, 72, and 96 h after siRNA transfection.(b) The effect of RNF31 knockdown on invasion of HepG2, Hep3B, and PLC cells was evaluated at 48 h after seeding cells in the invasion chambers.Representative images of invading cells are shown in the lower panel.Cell nuclei are stained in purple.*P < 0.05; siCont, control siRNA.

Table 2 .
Univariate and multivariate analyses of RNF31 expression and clinicopathological features of patients with hepatocellular carcinoma.HR hazard ratio; CI confidence interval; AFP α-fetoprotein; PIVKA-II protein induced by vitamin K absence II.*P < 0.05.