Long noncoding RNA UPK1A-AS1 indicates poor prognosis of hepatocellular carcinoma and promotes cell proliferation through interacting with EZH2

Dysregulations of lncRNA are responsible for cancer initiation and development, positioning lncRNAs as not only biomarkers but also promising therapeutic targets for treatment. Growing number of lncRNAs reported in and mechanistic remain unclear. Gene Set Enrichment Analysis was used to investigate the molecular mechanism of lncRNA UPK1A antisense RNA 1 (UPK1A-AS1). CCK-8 assay, EdU assay, ow cytometry, western blot, and xenograft assay were used to conrm the role of UPK1A-AS1 in proliferation of HCC cells both in vitro and in vivo. Bioinformatics analysis and qRT-PCR were performed to explore the interplay between UPK1A-AS1 and Enhancer of Zeste Homologue 2 (EZH2). RNA immunoprecipitation, RNA-pull down assay, western blot, qRT-PCR, and were conducted to conrm the interaction between UPK1A-AS1 and EZH2. Finally, the expression level and prognosis value of UPK1A-AS1 in HCC were analyzed using RNA-seq data from TCGA datasets.


Western blot
Protein were separated on an SDS-polyacrylamide gel followed by transferring onto polyvinylidene uoride membranes (Bio-Rad, USA). The membranes were blocked with 5% BSA for 50 minutes at room temperature before incubated with primary antibody in 4 ℃ overnight. The membranes were incubated with secondary antibody conjugated to horseradish peroxidase, followed by signal detection by enhanced chemiluminescence Western blotting substrate (Bio-Rad, USA). The primary antibodies used here were listed in supplementary Table 2.
Xenograft assay Four-week-old male nude mice were subcutaneously injected with 1*10 7 UPK1A-AS1-overexpressing or negative control MHCC-97H cells. Tumor diameters was measured every other day and tumor volumes was calculated as (length *width 2 )/2. The mice were sacri ced before tumor removal at 4 weeks after injection. All procedures for animal study were approved by the Animal Used and Care Committee of Nanfang Hospital, Southern Medical University (Guangzhou, China).

Functional Enrichment Analysis
A total of 500 genes positively correlated with EZH2 in HCC samples from The Cancer Genome Atlas (TCGA) dataset were subjected to Functional Enrichment Analysis using online software Metascape (https://metascape.org/). Only terms with P < 0.01, minimum count of 3 and enrichment factor greater than 1.5 were identi ed as signi cant.

RNA immunoprecipitation (RIP)
RIP assay was performed using Magna RIP RNA-binding protein immunoprecipitation kit (Millipore, USA) following the manufacturer's recommendations. Cells were lysed with lysis buffer and cell lysates was immunoprecipitated with anti-EZH2 and IgG antibodies. Immunoprecipitated RNA were extracted, puri ed and reverse transcribed to cDNA. The transcribed cDNA was subjected to qRT-PCR using UPK1A-AS1 speci c primers. The primer sequences used for UPK1A-AS1 ampli cation were listed in supplementary Table 1.
RNA pull-down assay RNA pull-down assay was carried out by Magnetic RNA-Protein Pull-Down Kit (Pierce, USA) following the manual. Full length of UPK1A-AS1 and antisense were yield using RiboMAX Large Scale RNA Production Systems (Promega, USA). Biotin labeled UPK1A-AS1 and antisense were bound to the beads and incubated with the whole cell protein lysates for immunoprecipitation. The beads were washed before eluted with SDS-PAGE loading buffer. Sample eluted from beads was subjected to western blot analysis.

Statistical methods
All statistical analysis was carried out by the SPSS statistical software version 22 (Abbott Laboratories, USA). Student's t-test and one-way ANOVA were carried out for statistical analysis when appropriate.
Kaplan-Meier and log-rank test was used for survival analysis. A P-value < 0.05 (two-tailed) was statistically signi cant.

Results
Upregulation of UPK1A-AS1 promotes proliferation of HCC cells To investigate the molecular mechanism of UPK1A-AS1, we conducted GSEA of TCGA cohort and found that high UPK1A-AS1-expressing groups were enriched for cell cycle-related gene sets ( Figure 1A), suggesting UPK1A-AS1 may hold a function on cell proliferation. To con rm its function on cell proliferation, lentivirus with full length of UPK1A-AS1 or negative control were introduced into HCC cells and the proliferation rate of HCC cells was examined. UPK1A-AS1 was successfully overexpressed in HCC cells, and upregulation of UPK1A-AS1 signi cantly promoted HCC cell proliferation, as detected by CCK-8 assay ( Figure 1B-C). Since upregulation of UPK1A-AS1 correlated with cell cycle-related gene sets, we further determine whether UPK1A-AS1 could affect HCC cell cycle progression. We then performed EdU dye assay to examine the ratio change of cells entering the S phase. The results showed that more UPK1A-AS1-overexpressing cells entered S phase than its control cells ( Figure 1D-G). Taken together, overexpression of UPK1A-AS1 could promote HCC proliferation.

UPK1A-AS1 downregulation inhibits HCC cell proliferation
To further con rm the regulatory function of UPK1A-AS1 on cell proliferation, we knocked down UPK1A-AS1 expression in HCC cells using siRNAs and shRNAs. The knocked down e ciency was veri ed using qRT-PCR. CCK-8 assays showed that downregulation of UPK1A-AS1 visibly inhibited HCC cell proliferation ( Figure 2A-D). Locked nucleic acids (LNAs) speci c targeting UPK1A-AS1 were introduced into HCC cells to further veri ed the effect of UPK1A-AS1 downregulation on HCC proliferation. Consistently, downregulation of UPK1A-AS1 by LNAs also impaired HCC proliferation ( Figure 2E). Moreover, cells in LNAs treatment groups that entered S phase were signi cantly less than its control groups ( Figure 2 F-G).
In summary, knocked down of UPK1A-AS1 inhibits HCC cell proliferation.
UPK1A-AS1 accelerates G1/S transition of HCC cells It is well accepted that rapid cell cycle progression accounts for cancer proliferation. Above results showed that upregulation of UPK1A-AS1 correlated with cell cycle-related gene sets and UPK1A-AS1 promotes cell proliferation. This led us to hypothesize that UPK1A-AS1 might regulate cell cycle progression. To this end, we carried out ow cytometry analysis to detect cell cycle distributions in HCC cells after UPK1A-AS1 overexpression or downregulation. The results showed that HCC cells with UPK1A-AS1 overexpression had a decreased rate of G1 phase cells and increased rate of S phase cells (Figure3 A-D). Consistently, CyclinD1, one of the most important modulators in G1/S transition, was highly expressed in cells with UPK1A-AS1 overexpression ( Figure 3E). On the contrary, si-UPK1A-AS1 resulted in an evident cell cycle arrest at the G1/G0 phase (Supplementary Figure 1A-B), and CyclinD1 was visibly decreased in cells with UPK1A-AS1 downregulation (Supplementary Figure 1D).
We also explored the effect of UPK1A-AS1 on apoptosis and drug resistance. Lower levels of apoptosis were founded in UPK1A-AS1-overexpressing cells, indicating that overexpression of UPK1A-AS1 could protect HCC cells from cis-platinum toxicity ( Figure 3F-G). Consistently, the expression level of wellde ned apoptosis markers, including cleaved caspase3 and cleaved PARP, were obviously decreased in UPK1A-AS1-overexpressing cells after cis-platinum exposure. This suggests that UPK1A-AS1 may boost the resistance to chemotherapy with cis-platinum in HCC cells. In conclusion, upregulation of UPK1A-AS1 accelerates G1/S transition of HCC cells.

UPK1A-AS1 promotes tumor growth in vivo
Based on the results of UPK1A-AS1 in vitro assay, we speculated that UPK1A-AS1 may take an important part in tumor growth in vivo. HCC cells with stable UPK1A-AS1-overexpressing or negative control were then subcutaneously injected into nude mice. The tumors formed in UPK1A-AS1-overexpressing group grew faster than those in negative control group. The tumor weight and volume were signi cantly higher in UPK1A-AS1-overexpressing group than in negative control group ( Figure 4A-C, Supplementary Figure  2). UPK1A-AS1 was remarkedly overexpressed in UPK1A-AS1-overexpressing group, as detected by qRT-PCR ( Figure 4D). In addition, the positive rate of proliferation marker Ki-67, was obviously increased in tumor with UPK1A-AS1-overexpressing ( Figure 4E). Collectively, UPK1A-AS1 boosts tumor growth in vivo.

UPK1A-AS1 correlates with EZH2-mediated cell cycle progression
To dissect the molecular mechanism involved in HCC progression through UPK1A-AS1, GSEA was carried out with the HCC tumor samples in TCGA dataset. GSEA results suggested that high expression of UPK1A-AS1 was correlated with EZH2 targets ( Figure 5A). Interestingly, GSEA putputs showed that high expression of EZH2 was positively correlated with cell cycle gene sets ( Figure 5B). Additionally, the functions of EZH2 and its correlated genes in HCC were predicted by analyzing GO and KEGG in Metascape. The top 20 Go enrichment items also suggested that EZH2 was associated with cell cycle. It has been long recognized that EZH2 has a crucial role in regulating cancer cell proliferation [17]. Consistently with previous studies, downregulation of EZH2 with siRNA signi cantly inhibited HCC cell proliferation. CyclinD1, CDK2, CDK4, which accelerates cell cycle progression, were reported to be downstream targets of EZH2 [18]. Not surprisingly, these genes were signi cantly downregulation after EZH2 silencing in HCC cells ( Figure 5F). We also founded that CCNB1 and CCNB2 were signi cantly decreased after EZH2 silencing in HCC cells ( Figure 5F). These results suggested that CyclinD1, CDK2, CDK4, CCNB1, and CCNB2 were direct targets of EZH2 in HCC. We further investigated the correlation between EZH2 and its targets from TCGA dataset. Strong positive correlations between EZH2 and CyclinD1, CDK2, CDK4, CCNB1, and CCNB2 were found in HCC samples (Supplementary Figure 3A). These results suggested that EZH2 promoted HCC proliferation by regulating cell cycle-related genes.

UPK1A-AS1 interacts with EZH2
To further investigate the molecular mechanisms by which UPK1A-AS1 contributes to the progression of HCC, we examined the subcellular distribution of UPK1A-AS1 in HCC cells by fractionation assays.
UPK1A-AS1 located in both nucleus and cytosol in HCC cells, indicating that UPK1A-AS1 could function as a modulator of gene transcription ( Figure 6A). It is reported that one-fth of human lncRNAs identi ed physically interacted with polycomb repressive complex 2 (PRC2), consisting of EZH2, SUZ12, and EED, with EZH2 as a crucial component of PRC2 [19]. Our results that UPK1A-AS1 regulates EZH2-mediated cell cycle progression triggered us come up with an assumption that UPK1A-AS1 may interact with and bind to EZH2. To test our hypothesis, RNA immunoprecipitation assay (RIP) against EZH2 was performed. RIP assay showed that UPK1A-AS1 was signi cantly enriched with the EZH2 antibody compared with negative control (IgG) in HCC cells ( Figure 6B-C). To further con rm our assumption, the interaction of UPK1A-AS1 with EZH2 was determined using RNA pull-down assay. The results showed that biotinlabeled UPK1A-AS1, but not antisense, exhibited the ability to harbor EZH2 protein ( Figure 6D). These results demonstrated that UPK1A-AS1 could physically interact with EZH2.
We next wondered if UPK1A-AS1 had an impact on the expression level of EZH2. Western blot showed that neither overexpression nor downregulation of UPK1A-AS1 altered the expression of EZH2 ( Figure 6E).
Moreover, no signi cant correlation was found between UPK1A-AS1 and EZH2 expression level ( Figure  6F). These results demonstrated that UPK1A-AS1 interacted with EZH2 without changing the expression of EZH2. Surprisingly, overexpression of UPK1A-AS1 increased the trimethylation of H27K3 which was caused by PRC2 activation. On the contrary, silencing UPK1A-AS1 led to obvious reduction of trimethylation on H27K3 ( Figure 6E), suggesting that interaction between UPK1A-AS1and EZH2 led to PRC2 activation.
It has been reported that lncRNA physically interacts with and binds to proteins to alter their subcellular distribution [20]. Fractionation assays showed that overexpression of UPK1A-AS1 decreased the cytoplasmic expression of EZH2, but increased the expression level of EZH2 in the nucleus ( Figure 6G). Immuno uorescence experiment also con rmed that overexpression of UPK1A-AS1 induced translocation of EZH2 from the cytoplasm to the nucleus ( Figure 6H). EZH2, SUZ12 and EED form complex in the nucleus for PRC2 activation. An increased interaction between EZH2 and SUZ12 was found after UPK1A-AS1 overexpression ( Figure 6I). In brief, UPK1A-AS1 interacted with EZH2 and mediated its nucleus translocation, reinforced its binding to SUZ12, leading to the increased trimethylation of H27K3.

UPK1A-AS1 functions through EZH2
To explore whether EZH2 mediated the regulative effect of UPK1A-AS1 on HCC cell proliferation, we cotransfected EZH2 siRNA and UPK1A-AS1 vectors into HCC cells and analyzed the expression of EZH2 targets related to cell cycle. Overexpression of UPK1A-AS1 increased the expression of CCND1, CDK2, CDK4, CCNB1 and CCNB2. Downregulation of EZH2 eliminated upregulation of these genes caused by UPK1A-AS1 overexpression ( Figure 7A-B). Consistent with results of qRT-PCR, EdU assay showed that more UPK1A-AS1-overexpressing cells entered S phase than its control cells. The increase of S phase ratio by UPK1A-AS1 overexpression was reversed in part by silencing EZH2 (Figure 7C-D). Taken together, targeting EZH2 with speci c siRNA impaired UPK1A-AS1-mediated upregulation of proliferation and cell cycle progression related genes.
High expression of UPK1A-AS1 predicts poor prognosis for patients with HCC UPK1A-AS1 is a newly identi ed lncRNA and little is known about its clinical implication in cancers.
Analysis from genotype-Tissue Expression (GTEx) benign tissue RNA-seq revealed that UPK1A-AS1 was highly expressed in the bladder, but scarcely in other tissues (Supplementary Figure 4A). However, data from TCGA datasets showed UPK1A-AS1 was relatively induced in some kinds of cancers, including HCC (Supplementary Figure 4B), indicating its important role in development and progression of malignancies.
To explicit the clinical implication of UPK1A-AS1 in HCC, UPK1A-AS1 expression level in HCC was analyzed with RNA-seq data from TCGA dataset. UPK1A-AS1 was highly expressed in HCC ( Figure 8A). To eliminate the possibility that the signi cant difference between HCC tissues and non-tumor tissues was caused by imbalanced of sample size, paired HCC and corresponding non-tumor samples was reanalyzed. The results convinced that UPK1A-AS1 was signi cantly overexpressed in HCC ( Figure 8B). Moreover, high expression of UPK1A-AS1 positively correlated with tumor stage of HCC ( Figure 8C). Survival analysis showed that patients with high expression of UPK1A-AS1 exhibited worse overall survival (OS) as compared with those with low UPK1A-AS1 expression group ( Figure 8D). Because UPK1A-AS1 expression correlated with HCC stage, we reanalyzed the data from subgroups. Patients with high UPK1A-AS1 level of UPK1A-AS1 presented shorter OS than those with low UPK1A-AS1 expression, though the difference did not reach statistical signi cance ( Figure 8E-F). Vascular invasion is a sign of poor prognosis for patients with HCC. Survival analysis showed that in vascular invasion group, patients with high UPK1A-AS1 level of UPK1A-AS1 suffered poorer OS. Due to limitation in sample size, the difference did not reach statistical signi cance ( Figure 8F). Since infection of hepatitis virus and alcohol abuse were risk factors for HCC, we also clari ed correlation between UPK1A-AS1 expression level and prognosis in patients with HCC risk factor exposure. It is shown that patients with high UPK1A-AS1 expression suffered shortened OS in patients with HCC risk factors ( Figure 8G). Furthermore, univariate Cox regression analysis demonstrated that the OS risk of patients with HCC was signi cantly associated with upregulation of UPK1A-AS1 (Table 1).
We also explored the clinical signi cance of EZH2 in cancer. Data from TCGA datasets showed that EZH2 was highly expressed in various cancers, including HCC (Supplementary Figure 5A). Overexpression of EZH2 predicted poor prognosis in various cancer, suggesting its oncogenic role in tumorigenesis (Supplementary Figure 5B). A series of HCC datasets from Gene Expression Omnibus (GEO) con rmed that EZH2 was highly expressed in HCC (Supplementary Figure 5C) Our results showed that UPK1A-AS1 functioned through EZH2, at least by part. Consistently, patients with simultaneous high UPK1A-AS1 and EZH2 expression also exhibited shorter OS. Collectively, UPK1A-AS1 was signi cantly upregulated in HCC, and upregulation of UPK1A-AS1 predicted poor prognosis for patients with HCC.

Discussion
Despite the profound advances made in HCC therapeutic strategies, the long-term prognosis of HCC patients remains poor as a result of limited understanding of the underlying mechanisms of tumor initiation and development [21]. Dysregulation of lncRNAs has been recognized to be involved in the onset and progression of malignancies, suggesting their clinical potential as biomarkers for diagnosis and prognosis, as well as therapeutic targets. Here, we demonstrated that UPK1A-AS1 was highly expressed in HCC, and high expression of UPK1A-AS1 predicted poor prognosis in patients with HCC. Biological experiments showed that UPK1A-AS1 promoted proliferation and tumor growth by accelerating G1/S transition of HCC cells. Furthermore, we also found that overexpression of UPK1A-AS1 could protect HCC cells against cis-platinum toxicity, suggesting that UPK1A-AS1 may promote the resistance to chemotherapy in HCC cells. Our ndings suggested that UPK1A-AS1 may serve as a novel prognostic biomarker and potential therapeutic targets for HCC.
UPK1A-AS1 is a newly identi ed lncRNA with little information about its functional role and clinical signi cance in cancers. UPK1A-AS1, downregulated in ESCC, inhibited the proliferation, migration, and invasion of ESCC cells by serving as a miRNA decoy [16]. On the contrary, our nding showed that UPK1A-AS1 was upregulated in HCC, and overexpression of UPK1A-AS1 promoted proliferation by regulation of cell cycle progression. It is well accepted that lncRNA constantly acts in a more tissue-speci c or diseasespeci c manner [22]. RNA-seq data from GTEx revealed that UPK1A-AS1 was highly expressed in bladder, and modestly expressed in esophagus, cervix, and vagina, but hardly expressed in other tissues ( Supplementary Fig. 4A), indicating the expression of UPK1A-AS1 was tissue-speci c, and the biological role of UPK1A-AS1 may vary depending on organic context. Tissue-speci c or disease-speci c context of UPK1A-AS1 may account for the distinct roles of UPK1A-AS1 in ESCC and HCC.
EZH2 serves as the enzymatic core subunit of PRC2, a complex that has the ability to methylate lysine 27 of histone H3 and facilitates chromatin remodeling and transcriptional silencing [23]. Growing number of evidences have implicated EZH2 in the progression of a variety of human malignancies [24]. Consistently, our ndings also con rmed that EZH2 was highly expressed in various cancer type, including HCC [25].
Here, we found that high level of EZH2 correlated with the development and progression of HCC.
Upregulation of EZH2 predicted poor prognosis in patients with HCC. Moreover, in patients undergoing sorafenib treatment, EZH2 was a factor obviously effecting their survival, indicating that the expression level of EZH2 may distinguish patients who would bene t from sorafenib treatment. It has been reported that the subcellular localization of EZH2 correlates with mechanism of EZH2 oncogenic activity. EZH2 present in the cytoplasm may participate in actin polymerization to in uence tumor dissemination [26].
However, EZH2, SUZ12, and EED form a complex in nucleus and transcriptionally regulate gene expression [27]. The association of EZH2, SUZ12, and EED is responsible for PRC2 activation. Here, we found that UPK1A-AS1 induced translocation of EZH2 from the cytoplasm to the nucleus. Furthermore, UPK1A-AS1 increased the interaction between EZH2 and SUZ12, promoted the methylation of lysine 27 in histone H3, indicating that UPK1A-AS1 contributes to the formation and activation of PRC complex. EZH2 mediated PRC2 activation contributes to the transcriptionally silencing of tumor suppress genes, leading to the activation of NOTCH [28], JAK-STAT [29], or β-catenin signaling pathways [30], and upregulation of cell cycle genes, like CDK2, CDK4, CyclinD1, et. al. However, UPK1A-AS1 overexpression did not change the expression of p-STAT3 (data not shown) and β-catenin (Fig. 6G), while si-EZH2 abolished the upregulation of CDK2, CDK4, and CyclinD1 caused by UPK1A-AS1, suggesting that UPK1A-AS1 upregulated the aforementioned genes via EZH2, but not EZH2-mediated JAK-STAT, and β-catenin signaling activation. Growing evidence have manifested that EZH2 can function through a PRC2independent to facilitate transcriptional activation rather than repression [31][32][33]. EZH2 could directly bind to the promoter regions of CyclinD1 and promote its transcriptional activation [34]. Whether UPK1A-AS1-mediated upregulation of CDK2, CDK4, and CyclinD1 was EZH2-dependent transcriptional activation or not still requires further investigation.

Conclusions
Taken together, our ndings uncovered the biological function and underlying mechanism of a newly identi ed lncRNA, UPK1A-AS1, which promotes HCC progression by accelerating cell cycle G1/S transition in an EZH2-dependent manner. Moreover, UPK1A-AS1 was highly expressed in HCC, and high expression of UPK1A-AS1 predicts poor prognosis in patients with HCC, suggesting that UPK1A-AS1 may serve as a potential biomarker for HCC prognosis and therapy.   Figure 1 Upregulation of UPK1A-AS1 promotes proliferation in HCC cells. A. Results of Gene Set Enrichment Analysis (GSEA)were plotted to visualize the correlation between UPK1A-AS1 and cell cycle gene signatures in HCC sample from TCGA dataset (P < 0.05). B-C. CCK-8 assay was performed to determine the effect of UPK1A-AS1 overexpression on proliferation in SK-Hep-1 (B) and MHCC-97H (C) cells (**P < 0.01, ***P < 0.001). D-G. Overexpression of UPK1A-AS1 in SK-Hep-1 (D-E) and MHCC-97H (F-G) cells promoted more cells into S phase than negative control as detected by EdU assay (*P < 0.05, **P < 0.01).    and MHCC-97H cells. β-Actin was used as a cytosol marker, U6 was used as a nuclear marker, both NEAT1 and MALAT1 were nucleus located lncRNAs. B-C. RIP assay was performed in MHCC-97H cells and the co-precipitated RNA was subjected to qRT-PCR for UPK1A-AS1(***P < 0.001). D. RNA pull-down assay was carried out to con rm the association between UPK1A-AS1 and EZH2. E. Effect of UPK1A-AS1 overexpression on expression level of EZH2 and H3K27M3 was measured by western blot. F. Correlation between UPK1A-AS1 and EZH2 was analyzed in HCC samples from TCGA dataset. G. EZH2 expression level in the cytoplasm or nucleus of MHCC-97H cells. β-Actin was used as a cytosol marker, LaminB1 served as a nuclear marker. H. Translocation of EZH2 from the cytoplasm to the nucleus was detected by immuno uorescence assay. I. Immunoprecipitation assay identi ed the increased interaction between EZH2 and SUZ12 in UPK1A-AS1-overexpressing MHCC-97H cells. β-Actin was used as negative control.