ZRANB1 Drives Hepatocellular Carcinoma Progression through SP1-LOXL2 Axis

close relationship underlying and molecular mechanisms in Gene and protein expression of ZRANB1 in HCC tissues were determined by qRT-PCR, western blot and immunohistochemistry. A series of gain- and loss-of-function assays were used to investigated the role of ZRANB1 in HCC cells progression. Moreover, RNA-seq were used to identify the downstream targets of ZRANB1 in HCC cells. The interaction between ZRANB1 and SP1 was examined through co-IP experiment and in vitro ubiquitination assay.


Abstract
Background Deubiquitinase (DUB) zinc nger RANBP2-type containing 1 (ZRANB1/TRABID) has been reported to have a close relationship with cancers. However, its underlying role and molecular mechanisms in hepatocellular carcinoma (HCC) remain elusive.

Methods
Gene and protein expression of ZRANB1 in HCC tissues were determined by qRT-PCR, western blot and immunohistochemistry. A series of gain-and loss-of-function assays were used to investigated the role of ZRANB1 in HCC cells progression. Moreover, RNA-seq were used to identify the downstream targets of ZRANB1 in HCC cells. The interaction between ZRANB1 and SP1 was examined through co-IP experiment and in vitro ubiquitination assay.

Results
ZRANB1 was highly expressed in HCC tissues and ZRANB1 can regulate HCC cell growth and metastasis in vitro and in vivo. Through RNA-seq, we identi ed that Lysyl oxidase-like 2 (LOXL2) was the most signi cantly downregulated gene after ZRANB1 knockdown. Furthermore, the scatter plots indicated a signi cant positive correlation between ZRANB1 and LOXL2 expression in clinical HCC specimens.
Additionally, LOXL2 was essential for ZRANB1-mediated HCC growth and metastasis. More importantly, speci city protein 1 (SP1) was critical in ZRANB1-mediated regulation of LOXL2 expression.
Mechanistically, ZRANB1 bound with SP1 directly and stabilized the SP1 protein by deubiquitinating it. The expression patterns of ZRANB1, SP1 and LOXL2 were evaluated in HCC patients.

Conclusion
ZRANB1 overexpression facilitates the carcinogenesis of HCC through stabilizing and upregulating SP1 to promote LOXL2 expression, suggesting ZRANB1 can be novel prognostic biomarker for HCC treatment.
Background Hepatocellular carcinoma (HCC) is the second leading cause of cancer-related death worldwide and is currently one of the most heterogeneous and complex malignancies [1]. Although some clinicopathological factors are essential in in uencing the treatment outcome of HCC, there are no effective therapeutic strategies for patients with advanced HCC [2,3]. With a 5-year survival rate of 14-18%, the mortality of HCC is mostly due to its characteristics of proliferation, invasion and migration [4]. Therefore, further studies and identi cation of the molecular mechanism underlying HCC carcinogenesis may help us to identify potential targets for improving therapeutic effects on HCC.
Ubiquitination is one of the most versatile post-translational modi cations, and it plays an essential role in protein degradation through the proteasome pathway or by direct sorting to the lysosome [5]. The regulation and reversibility of ubiquitination is accomplished through deubiquitinases (DUBs) [6]. In recent years, accumulated evidence has shown that dysregulation of multiple DUBs is closely related to the development of numerous carcinomas [7,8]. The deubiquitinase (DUB) ZRANB1/TRABID (zinc nger RANBP2-type containing 1) is a member of the A20 OTU (ovarian tumor proteases) family [9]. It has been reported that ZRANB1 can cleave K29-, K33-, and K63-linked ubiquitin chains [10]. Another study also demonstrated that ZRANB1 can deubiquitinate and stabilize EZH2 in breast cancer and provided therapeutic strategies for targeting the EZH2 deubiquitinase ZRANB1 [11]. Recently, a study reported that high phosphorylation levels of ZRANB1 can activate its deubiquitinase activity, which is highly correlated with poor prognosis in HCC patients [12]. These ndings suggest that aberrant expression of ZRANB1 plays an oncogenic role in multiple carcinomas and may indicate novel therapeutic targets. However, the potential role of ZRANB1 in HCC and its underlying mechanisms in regulating HCC carcinogenesis remain elusive.
Here, we investigated the expression patterns of ZRANB1 in HCC samples and further explored the role of ZRANB1 in the proliferation, migration and invasion of HCC cells in vitro and in vivo. We found that ZRANB1 regulated HCC cell growth and metastasis through the SP1-LOXL2 axis. At the mechanistic level, we revealed that ZRANB1 can bind with SP1 directly and thus deubiquitinate and stabilize SP1, which contributes to LOXL2 overexpression and HCC carcinogenesis.

Materials And Methods
Human specimens and cell culture From June 2016 to April 2020, human HCC tissues and corresponding non-tumor tissues were obtained from 132 patients who underwent HCC resection at the Second A liated Hospital of Nanchang University. Fresh samples collected after resection were frozen immediately in liquid nitrogen and stored at -80 °C for further study. Informed consent of the patients was obtained and the investigation was permitted by research ethics committee of the Second A liated Hospital of Nanchang University. quantitative real-time PCR (qRT-PCR) and Western blot Total RNA was extracted by the standard Trizol-based protocol (Invitrogen, USA). Complementary DNA (cDNA) was synthesized using the PrimeScript RT Reagent Kit (Invitrogen, USA) and qRT-PCR was performed using SYBR Premix Ex Taq (TaKaRa Bio, Shiga, Japan), according to the manufacturer's instructions. Information about the gene-speci c primers were in Supplementary Table 1. Total protein extracts were prepared as described in [13]. Protein was extracted by RIPA buffer (Beyotime, Shanghai, China) containing protease and inhibitor mixes (Thermo Fisher Scienti c, New York, USA) on ice. BCA Protein Assay kit (Thermo Scienti c, Waltham, MA, USA) was used to analyze protein concentration. Before centrifugation for 10 min, all protein extracts were boiled in the lysis buffer for 30 min.
Cell proliferation assay 5-Ethynyl-20-deoxyuridine assay (EdU) and Real Time Cellular Analysis (RTCA) were described previously [15]. In EdU assay, cells were plated in 96-well plates (5x10 4 cells per well) and cultured for 24 hours. Then 5-ethynyl-20-deoxyuridine (EdU; Ribobio) was used to culture cells for 2 hours, followed by three washes with PBS. After cells being incubated with 300 µL of 1xApollo reaction cocktail for 30 minutes, Hoechst 33342 (5 mg/mL) was used to stain the DNA contents of the cells in each well for 25 minutes and was imaged through a uorescence microscope. In RTCA tests, the transfected cells (1x10 5 cells per well) were plated and incubated in E-plate 16 for 35 hours and the data during the period was recorded by A Real-Time Cell Kinetic Analyzer xCELLigence RTCA (ACEA Biosciences) to monitor cell proliferation dynamics. Data recorded by the machine was analyzed by the RTCA software.
Cell colony formation assay 0.6 × 10 3 cells were seeded in 6-well plates. Once the appropriate colony size had formed, cells were xed with 4% paraformaldehyde for 30 min, and then stained by 1.0% crystal violet for 30 min until formed visible clones. The number of colonies was counted in 10 different elds.

The scratching assays
When HCC cells were grown to 80-90% con uence in 6-well plates, scraping apipette tip was used to scratch across the cells surface. After washing the detached cells with PBS, the cells were incubated over time to allow the cells to migrate into the scratching area. The cells movements during wound closure were measured by phase-contrast photography at 37℃ for incubations of 0 and 36 hours, and three randomly selected wound areas were analyzed.

In vitro migration and invasion assays
Cell transwell migration and invasion assays were performed using a transwell system (Corning, NY, USA) with or without Matrigel matrix (BD bioscience) coated above the 8-µm pore polycarbonate membranes.
Fresh medium containing 10% FBS was added in the lower chamber as a chemoattractant. For the migration assay, a total of 1 x 10 5 cells were resuspended in serum-free medium and placed in the upper chambers; for invasion assays, a total of 1 x 10 5 cells were seeded in a Matrigel-coated chamber (BD Biosciences). After 24 hours (to examine migration) or 48 hours (to examine invasion) of incubation, the non-migrated cells on the upper surface of the membrane were removed, and the cells on the lower surface were xed with methanol and stained by 0.1% crystal violet. The cells in ve random microscopic elds were counted and imaged using a light microscope with a DP70 CCD system (Olympus Corp).

Luciferase reporter assay
The luciferase experiment was carried out using LOXL2 re y luciferase and transfection control renilla luciferase plasmids. Plasmids were forward-transfected into cells using lipofectamine 3000 and PLUS reagent (Thermo Fisher Scienti c). Cells were stimulated 24 h post transfection and luciferase activities were evaluated using the Dual-Luciferase Reporter Assay System (Promega).

Co-immunoprecipitation experiment
Cell lysates were incubated with 50 µl of protein G beads (Thermo Scienti c), and 1 µg of the indicated antibody overnight at 4 °C. The protein G beads were collected by centrifugation. Loading buffer was added to the tube and heated for 15 min at 100 °C. Then the immunoprecipitated proteins were examined by SDS-PAGE and immunoblotting analysis. The intensity of protein bands was analyzed by Quantity One software (Bio-Rad, Hercules, CA, USA).

Animal experiments
For tumorigenicity assays, control and ZRANB1 stably knockdown or ZRANB1 stably overexpression luciferase-labelled HCC cells (2 x10 6 cells per mouse) were subcutaneously injected into the anks of 8week-old female BALB/c-nude mice (Hunan SJA Laboratory Animal Co., Ltd.

Statistical analysis
All results are shown as mean ± SD from at least three independent experiments. One-way ANOVA for multiple comparisons was used to compare EdU, colony-forming assays, the wound-healing assay, transwell migration and invasion assay, and normalized luciferase values. Student's t-test was used to compare the mice tumor volume and other related comparisons. All the data were analyzed using GraphPad Prism 5 software and P < 0.05 was considered signi cant.

ZRANB1 expression is increased in human HCC tissues and HCC cells
To determine the expression levels of ZRANB1 in HCC tissues, we performed quantitative real-time PCR on 72 HCC specimens and their corresponding non-tumor tissues. Upregulation of ZRANB1 mRNA level was markedly observed in 55 of the 72 pairs (Fig. 1a). Consistently, we detected the protein expression of ZRANB1 in these 72 pairs of HCC tissues and corresponding normal tissues. Quantitative analyses showed that ZRANB1 was signi cantly overexpressed in HCC tissues compared with non-tumor tissues (Fig. 1b). Consistently, we analyzed the relationship between ZRANB1 expression and clinicopathological features in 72 HCC patients. The results showed that ZRANB1 overexpression was signi cantly correlated with advanced tumor size, TNM stage, vascular invasion and intrahepatic metastasis (Table 1).
Furthermore, we performed immunohistochemical staining to examine ZRANB1 expression in 132 pairs of HCC tissues (including the 72 fresh HCC specimens) and found that ZRANB1 expression was markedly upregulated in HCC tissues compared with paired adjacent normal tissues (Fig. 1c, d). Kaplan-Meier analyses of the overall survival (OS) rate revealed that the high ZRANB1 expression group (i.e., IHC staining with strong and moderate expression) had a worse prognosis than the low ZRANB1 group (i.e., IHC staining with weak and negative expression) (Fig. 1e). Then, we explored ZRANB1 expression in HCC cell lines and two normal human hepatocyte lines (HL7702 and THLE-3). As expected, the expression level of ZRANB1 was signi cantly higher in HCC cell lines (Huh7, MHCC97H, HepG2, and HCCLM3) than that in HL7702 and THLE-3 cells at both the protein and mRNA levels (Fig. 1f, g). Furthermore, univariate and multivariate logistic regression analyses indicated that ZRANB1 was an independent predictor of poor prognosis for patients with HCC (Table 2). Our results revealed that ZRANB1, which is overexpressed in HCC, might play an essential role in HCC carcinogenesis and prompted us to explore the precise function of ZRANB1 in HCC.

ZRANB1 facilitates HCC cell proliferation and metastasis in vitro
To investigate how aberrant ZRANB1 levels affect HCC carcinogenesis, we inhibited ZRANB1 expression with shRNA (small hairpin RNA). Then, the ZRANB1 knockdown e ciency was con rmed by western blot and qRT-PCR analysis ( Supplementary Fig. 1a, b). To examine the proliferation ability of HCC cells transfected with control shRNA or shRNA targeting ZRANB1, we performed EdU and colony formation assays. The EdU assay results suggested that ZRANB1 knockdown markedly inhibited HCCLM3 cell and Huh7 cell growth (Fig. 2a, b). Similarly, the colony formation assay results also showed that decreased ZRANB1 expression reduced the colony-forming capacity of HCC cells (Fig. 2c, d). Furthermore, we investigated the proliferation ability of ZRANB1-overexpressing HCC cells. The results of both the EdU and colony formation assays suggested that ZRANB1 overexpression enhanced HCC cell growth ( Supplementary Fig. 2a, b).
To further verify the impact of dysregulated ZRANB1 expression on HCC cell metastasis, we performed transwell migration and invasion assays and the scratching assay. The transwell migration and invasion assay results showed that ZRANB1 knockdown signi cantly suppressed HCCLM3 and Huh7 cell migration and invasion (Fig. 2e, f). In addition, the scratching assay results suggested that ZRANB1 suppression can inhibit HCC cell migration (Fig. 2g, h). Consistent with these results, ZRANB1 upregulation in HCCLM3 and Huh7 cells contributed to the opposite effects, as evaluated by transwell migration, invasion and the scratching assays ( Supplementary Fig. 2c, d). These results indicated that ZRANB1 can facilitate HCC cell proliferation, migration and invasion in vitro.

ZRANB1 drives HCC carcinogenesis in vivo
To further understand the biological effects of ZRANB1 on HCC carcinogenesis, we performed tumorigenicity assays and lung metastasis experiments in animal subjects. In the tumorigenicity assay, nude mice were injected subcutaneously with shNC or shZRANB1 HCCLM3 cells expressing a luciferase reporter. Using IVIS to monitor tumor growth in mice, imaging analysis showed that ZRANB1 stable knockdown decreased tumor growth signi cantly (Fig. 3a). Tumor weights and volumes of the nude mice were recorded throughout the experimental period. Decreased ZRANB1 expression signi cantly inhibited HCC growth compared with that in the control group (Fig. 3b, c). Additionally, the Ki-67 level was markedly decreased in tumors with ZRANB1 suppression, as assessed through immuno uorescence experiments (Fig. 3d). In contrast, ZRANB1 upregulation promoted HCC progression in nude mice compared with that in the control group ( Supplementary Fig. 2e, f).
To further evaluate the contribution of ZRANB1 to HCC metastasis in vivo, we established a lung metastasis model via tail vein injection. Luciferase-labelled HCCLM3 cells transfected with shNC or shZRANB1 were injected into the tail vein of nude mice. Luciferase uorescence intensity of the progressive HCC in nude mice was detected by IVIS. The results showed a striking reduction of the metastatic capacity of ZRANB1 knockdown HCCLM3 cells to the lung (Fig. 3e). Additionally, H&E staining and the number of lymph node metastases results showed that the incidence of lung metastasis was signi cantly reduced in the ZRANB1 inhibition group compared with the control group (Fig. 3f, g).
Furthermore, we demonstrated that ZRANB1 upregulation promoted HCC lung metastasis ( Supplementary Fig. 2g). Collectively, these results showed that ZRANB1 can enhance HCC growth and metastasis in vivo.

LOXL2 expression is highly correlated with ZRANB1 in HCC
To obtain a more robust understanding of the mechanism by which ZRANB1 regulates HCC carcinogenesis, we performed RNA-seq on shNC and shZRANB1 HCCLM3 cells to identify the potential downstream targets of ZRANB1. Our data demonstrated that LOXL2 was the gene with the most marked decrease in expression after ZRANB1 suppression (Fig. 4a). Consistent with this nding, we detected the expression level of LOXL2 in shNC and shZRANB1 HCCLM3 cells and found that LOXL2 expression was reduced at both the protein and mRNA levels after ZRANB1 inhibition (Fig. 4b, c). In contrast, LOXL2 expression showed a signi cant increase in HepG2 cells with ZRANB1 upregulation (Fig. 4d, e). Importantly, we detected the mRNA and protein levels of ZRANB1 and LOXL2 in 72 HCC samples, representative images of ZRANB1 and LOXL2 protein expression in HCC tissues and adjacent non-tumor tissues are shown in Figure 4f. Additionally, the scatter plots suggested that LOXL2 expression was signi cantly correlated with ZRANB1 expression at both the mRNA and protein levels (Fig. 4g, h). These results revealed a signi cant relationship between ZRANB1 and LOXL2 in HCC.

LOXL2 is critical for ZRANB1-mediated HCC progression
Considering the correlation between ZRANB1 and LOXL2, we further investigated whether LOXL2 plays an essential role in ZRANB1-mediated HCC progression. After upregulating LOXL2 expression in ZRANB1 knockdown HCCLM3 cells, we assessed ZRANB1 and LOXL2 protein expression by western blot. Our results showed that although LOXL2 overexpression increased the LOXL2 protein level in HCCLM3 cells, ZRANB1 suppression markedly abated the increased LOXL2 (Fig. 5a). We then performed EdU assay and transwell migration and invasion assays to explore the impact on HCC cell growth and metastasis. Similarly, LOXL2 upregulation signi cantly enhanced HCCLM3 cell proliferation, migration and invasion; however, ZRANB1 inhibition effectively decreased the enhanced proliferation and metastasis capacities of HCCLM3 cells caused by LOXL2 overexpression (Fig. 5b, c). We further performed rescue experiments in vivo and assessed tumor growth, tumor weight and lung metastasis in nude mice. Consistent with the observations in vitro, our results showed that ZRANB1 knockdown also signi cantly suppressed the enhancement of tumor growth and HCC cells lung metastatic ability induced by LOXL2 upregulation (Fig.  5d, e, f).
In comparison, we also transfected LOXL2 shRNA into ZRANB1-overexpressing HepG2 cells and performed western blot and a series of functional assays. Consistent with the results of the previous experiments, ZRANB1 upregulation markedly reversed the decreased LOXL2 expression in LOXL2 knockdown HepG2 cells (Supplementary Fig. 3a). Furthermore, the EdU and transwell assays results showed that LOXL2 suppression signi cantly inhibited the increased proliferation, migration and invasion capacities of HepG2 cell induced by ZRANB1 overexpression (Supplementary Fig. 3b, c).

ZRANB1 promotes LOXL2-mediated HCC carcinogenesis through SP1
We next investigated how ZRANB1 regulates LOXL2 in HCC cells. Our co-IP results showed that ZRANB1 cannot interact with LOXL2 directly (Supplementary Fig. 4a, b). A previous study showed that SP1 can bind to the LOXL2 promoter and regulate its expression directly in breast cancer [17]. In addition, it has reported that SP1 can be regulated by the deubiquitinase USP33 [18]. We thus hypothesized that ZRANB1 might regulate LOXL2 through SP1. We evaluated the protein expression of SP1 and LOXL2 in HepG2 cells transfected with shZRANB1 RNA or His-ZRANB1 overexpression plasmids and, as expected, found that the protein expression levels of SP1 and LOXL2 were signi cantly correlated with the ZRANB1 expression level (Fig. 6a). Importantly, ZRANB1 knockdown decreased the transcriptional activity of the LOXL2 luciferase reporter, whereas upregulation of ZRANB1 showed the opposite effects (Fig. 6b). However, neither up-nor downregulation of ZRANB1 had a signi cant effect on SP1 mRNA expression ( Fig. 6c). Furthermore, we performed rescue experiments to explore the relationships among ZRANB1, SP1 and LOXL2 in HCC cells. The western blot results showed that ZRANB1 downregulation decreased the upregulation of SP1 and LOXL2 in SP1-overexpressing HCCLM3 cells (Fig. 6d). Moreover, the SP1 upregulation-induced enhancement of LOXL2 luciferase reporter transcriptional activity was alleviated by ZRANB1 knockdown (Fig. 6e). Through RTCA and transwell migration and invasion assays, ZRANB1 knockdown was found to reverse the increases in the proliferation and metastasis capacities of HCCLM3 cells induced by SP1 overexpression (Fig. 6f, g). Taken together, these results provide further support for functional associations among ZRANB1, SP1 and LOXL2 in HCC carcinogenesis.

ZRANB1 can deubiquitinate and stabilize SP1
Having con rmed the correlation between ZRANB1 and SP1, we aimed to explore the interaction of them.
It has been reported that ZRANB1 can deubiquitinate EZH2 and signi cantly increase its stability in breast cancer [11]. Additionally, our results in Fig 6 a and c show that ZRANB1 expression had a notable impact on SP1 expression only at the protein level. Moreover, another study reported that SP1 can be degraded through the ubiquitin-proteasome pathway and can be deubiquitinated by USP33 [18]. We thus hypothesized that ZRANB1 might deubiquitinate SP1 and stabilize it. As expected, the results of co-IP experiments showed that ZRANB1 can bind SP1 directly in HCCLM3 cells (Fig. 7a). We also found that ZRANB1 can bind with SP1 directly in HCCLM3 cells cotransfected with His-ZRANB1 and Flag-SP1 (Fig.  7b). To evaluate whether ZRANB1 can regulate SP1 degradation through the ubiquitin-proteasome pathway, we treated HepG2 cells transfected with ZRANB1 shRNA or His-ZRANB1 plasmids with or without 15 μM MG132. Our results showed that neither knockdown nor upregulation of ZRANB1 had a notable effect on the SP1 protein level in HepG2 cells treated with MG132 compared with those not treated with MG132 (Fig. 7c). Furthermore, we used a 20 μM dose of the translation inhibitor cycloheximide (CHX), and at the indicated times, we detected the SP1 protein level in HepG2 cells transfected with ZRANB1 shRNA or His-ZRANB1 plasmids. Our data demonstrated that ZRANB1 overexpression signi cantly decreased the degradation rate of SP1, while decreased ZRANB1 levels had the opposite effect (Fig. 7d, e).
Additionally, we further investigated the dose-dependent impact of ZRANB1 on SP1. Our data showed that a dose increase in the ectopic ZRANB1 level contributed to SP1 protein accumulation in HCC cells (Fig. 7f). Next, an equal dose of Flag-SP1 and an enhancing dose of His-ZRANB1 were cotransfected into HCCLM3 and Huh7 cells, and a dose-dependent effect of ZRANB1 overexpression on SP1 protein levels was observed (Fig. 7g). Finally, our results revealed that ZRANB1 suppression markedly increased the ubiquitination level of SP1, while ZRANB1 upregulation showed the opposite impact on SP1 (Fig. 7h, i). Thus, our data indicated that ZRANB1 can act as a deubiquitinase of SP1 and stabilize it.
Clinical signi cance of ZRANB1, SP1 and LOXL2 in HCC tissue To extend our ndings to the clinical setting, we analyzed the expression patterns of ZRANB1, SP1 and LOXL2 in 72 paired human HCC specimens. Our results showed that ZRANB1, SP1 and LOXL2 were upregulated in HCC tissues, whereas in adjacent normal tissues, the opposite results were observed (Fig.  8a). Notably, there was a marked correlation between ZRANB1 and SP1 protein expression but no signi cant correlation at the mRNA level (Fig. 8b, c). For SP1 and LOXL2, signi cant correlations were observed at both the mRNA and protein levels (Fig. 8d, e).
In summary, our study highlights the novel role and molecular mechanism of ZRANB1 in HCC carcinogenesis. ZRANB1 promotes HCC carcinogenesis by directly deubiquitinating SP1 and stabilizing it via the SP1-LOXL2 axis (Figure 9). Our ndings may provide a promising alternative therapeutic strategy for HCC.

Discussion
In our research, we provided a comprehensive investigation of the role and mechanism of ZRANB1 in regulating HCC carcinogenesis. Compared with other DUBs, such as USP7, BAP1 and Otubain1 [13,19,20], ZRANB1 has attracted little attention. Previous studies have shown that ZRANB1 positively regulates the WNT/CTNNB1 signaling pathway [10]. Recently, some studies have focused on the oncogenic role of ZRANB1 in promoting cancer progression [11,12,21]. However, the impact and mechanisms of ZRANB1 in HCC remain unclear. Our study is the rst to report the mechanism by which ZRANB1 modulates HCC carcinogenesis. We revealed signi cant overexpression of ZRANB1 in HCC specimens and found that its expression was closely correlated with tumor size, TNM stage, vascular invasion and intrahepatic metastasis. Furthermore, we demonstrated that ZRANB1 facilitated HCC cell growth, migration and invasion in vitro and in vivo, identifying the oncogenic function of ZRANB1 in HCC carcinogenesis.
In our study, we rst discovered that LOXL2 is a critical mediator of ZRANB1-induced HCC carcinogenesis.
LOXL2 belongs to the lysyl oxidase gene family and has been shown to promote the development of multiple carcinomas [21][22][23]. As an essential oncogenic protein, LOXL2 can also modify histone tails and reduce cell polarity, which can enhance the capability for carcinoma metastasis [23]. Recent studies have shown that upregulation of LOXL2 promotes HCC cell progression and metastasis and that LOXL2speci c inhibitors can inhibit tumor growth [24]. In the present study, through RNA-seq, we identi ed LOXL2 as an essential downstream regulated target of ZRANB1 in HCC cell. To obtain a comprehensive understanding of the relationship between ZRANB1 and LOXL2, we analyzed clinical HCC samples and found that LOXL2 expression was signi cantly correlated with ZRANB1 expression in HCC. More importantly, through a series of gain-and loss-of-function experiments, we revealed that LOXL2 was critical for ZRANB1-mediated HCC proliferation and metastasis.
It has been shown that ZRANB1 is required for the deubiquitylation of EZH2 in breast cancer [11].
Considering the function of ZRANB1 in deubiquitylation, we hypothesized that ZRANB1 could bind with LOXL2 and deubiquitinate it directly. However, our results showed that there was no direct interaction between ZRANB1 and LOXL2. A previous study reported that SP1 can promote LOXL2 expression by binding to the LOXL2 promoter and thus enhance the metastasis of pancreatic ductal adenocarcinoma [17]. Additionally, our laboratory previously demonstrated that the deubiquitinase ubiquitin-speci c protease 33 (USP33) can bind with SP1 directly and deubiquitinate SP1 to stabilize it [18]. Furthermore, the transcription factor speci city protein 1 (SP1) belongs to the SP transcription factor family and is highly regulated by post-translational modi cations [25]. Contributing to the "hallmarks of cancers", SP1 is overexpressed in a variety of cancers and is involved in numerous cellular functions, including proliferation, apoptosis and angiogenesis [26,27]. We thus assume that there exists an association between ZRANB1 and SP1. First, we suggested that SP1 was essential for ZRANB1-mediated LOXL2 expression and for HCC cell proliferation and metastasis. Next, given our data showing that ZRANB1 had a signi cant effect on SP1 protein expression only at the mRNA level, we assumed that ZRANB1 may interact with SP1 in protein level. We therefore performed co-IP and found direct binding between ZRANB1 and SP1. Then, we demonstrated that ZRANB1 can deubiquitinate SP1 and signi cantly increase its stability. Together, these results suggest a role for the ZRANB1-LOXL2 axis in regulating HCC carcinogenesis.
Furthermore, this is the rst time that we identi ed that ZRANB1, SP1 and LOXL2 are simultaneously overexpressed in HCC samples. Additionally, we analyzed the relationships among ZRANB1, SP1 and LOXL2 and discovered that at the protein level, ZRANB1 expression is highly correlated with SP1 expression, whereas at the mRNA level, there is no signi cant correlation. SP1 and LOXL2 are markedly correlated with each other at both the mRNA and protein levels.

Conclusion
In summary, our study revealed the novel role and regulatory mechanism of ZRANB1 in HCC carcinogenesis. ZRANB1 targets the SP1-LOXL2 axis to enhance the growth, invasion and migration of HCC cells. Our ndings provide strong evidence to highlight the potential role of ZRANB1 as a prognostic indicator and a therapeutic target in HCC.