The long noncoding RNA PCAT-1 links the microRNA miR-215 to oncogene CRKL-mediated signaling in hepatocellular carcinoma

The long non-coding RNA (lncRNA) PCAT-1 resides in the chromosome 8q24 cancer-risk locus and acts as a vital oncogene during tumorigenesis and progression. However, how PCAT-1 is post-transcriptionally regulated, for example, by small ncRNAs, such as microRNAs (miRNAs) is largely unknown. Here, we report how miRNAs regulate PCAT-1 expression and also investigate the biological significance of this regulation in hepatocellular carcinoma (HCC). We found that miR-215, a P53-inducible miRNA, is a key regulator of PCAT-1 expression in HCC and identified an interaction between miR-215 and PCAT-1 in dual luciferase reporter gene assays. We also found that post-transcriptional silencing of PCAT-1 by miR-215 or PCAT-1 siRNAs significantly inhibited proliferation of HCC cells and, conversely, that inhibition of endogenous miR-215 up-regulated PCAT-1 expression and promoted cell viability. The tumor-suppressing role of miR-215 was further confirmed in an in vivo mouse HCC xenograft model. Of note, gene profiling assays suggested that the kinase CRK-like proto-oncogene, adaptor protein (CRKL), is a potential downstream target of the miR–215–PCAT-1 axis in HCC, and we demonstrated that CRKL silencing significantly suppresses cell proliferation. Taken together and considering the essential role of CRKL in cancer cells, we propose that the TP53–miR-215–PCAT-1–CRKL axis might represent an important regulatory pathway in HCC. In summary, our results highlight the involvement of several ncRNAs in HCC and thus provide critical insights into the molecular pathways operating in this malignancy.

Hepatocellular carcinoma (HCC) 3 is the sixth most prevalent human malignancy globally (1)(2)(3). Its morbidity approximately matched its mortality, indicating the aggressive nature of this malignant disease (1)(2)(3). Chronic infection with the hepatitis B or C viruses (HBV or HCV), exposure to dietary aflatoxin B, as well as alcohol abuse are major risk factors of HCC (2,3). There are limited therapeutic options for this refractory disease. Most patients diagnosed at an advanced stage cannot receive surgical resection or liver transplantation (4,5). Prevention, intervention, diagnosis, and treatment for HCC are far from satisfactory (4,5). As a result, clarifying the underlying mechanism of HCC development and progression is urgently needed.
Long noncoding RNAs (lncRNAs) are functionally defined as transcripts Ͼ200 nucleotides in length with no protein-coding potential. Although more and more lncRNAs have been identified, the majority's biological functions are unclear (6 -10). Indeed, lncRNAs may contribute diverse functional roles considering their expression across a variety of cellular-and tissuespecific contexts (6 -10). Therefore, the exquisite regulation of lncRNA expression at either transcription or post-transcription levels can significantly impact disease development, i.e. malignant transformation. Our and others data declared the essential involvement of microRNAs (miRNAs) in post-transcription regulation of cancer-related lncRNAs, which may drive many important cancer phenotypes (11)(12)(13)(14)(15)(16).
High-throughput sequencing of the poly(A) ϩ RNA (RNA-Seq) of prostate tissues and cells lines led to the discovery of a novel lncRNA PCAT-1 (17). PCAT-1 is evidently overexpressed in a subset of prostate cancers and may contribute to prostate cancer cell proliferation. The interaction between PCAT-1 and polycomb repressive complex 2 (PRC2) could transcriptionally regulate target genes of PRC2 and further indicates its important role in prostate cancer progression (17). It is also notable that PCAT-1 resides in the 8q24 "gene desert" locus, which is also prostate cancer susceptibility locus identified by genome-wide association studies (GWAS). Through integrative analyses of the lncRNA transcriptome with genomic data and single nucleotide polymorphism data from prostate cancer genome-wide association studies, Guo et al. (18)  fied 45 candidate lncRNAs associated with risk to prostate cancer. Among these lncRNAs, PCAT-1 ranks the top hit. The risk-associated rs7463708 single nucleotide polymorphism increases binding of ONECUT2, a novel androgen receptorinteracting transcription factor, at a distal enhancer that loops to the PCAT-1 promoter, resulting in up-regulation of PCAT-1 upon prolonged androgen treatment (18). Besides the transcriptional regulation by the androgen receptor signaling, somatic copy-number alterations or structural variants may also contribute to PCAT-1 dysregulation in esophageal cancer (19). However, fine regulation of lncRNA PCAT-1 expression in HCC, especially at the post-transcriptional level, is still largely unknown. Two previous studies showed the role of lncRNA PCAT-1 in HCC (20,21). Wen et al. (20) found that the increase of lncRNA PCAT-1 worsened HCC. Also, the increased expression of PCAT-1 was associated with advanced clinical parameters and poor overall survival of HCC patients (21). In this study, we for the first time declared the fine regulation of lncRNA PCAT-1 by the TP53-response miR-215 and its potential downstream effector, oncogene CRKL. Ectopic expression of miR-215 or PCAT-1 siRNA significantly inhibits expression, thereby suppressing HCC proliferation in vitro and in vivo.

miR-215 directly suppresses lncRNA PCAT-1 expression
To investigate the potential miRNA-lncRNA interactions, we first subcloned a 510-bp human PCAT-1 sequence after the firefly luciferase gene and named the construct as pGL3-PCAT-1 (Fig. 1E). Then SMMC7721 and Huh7 cells were co-transfected with this construct and miR-215 mimics or NC RNA. miR-215 suppressed a 70.0 or 67.7% luciferase activity compared with NC RNA in SMMC7721 or Huh7 cells (both p Ͻ 0.01) (Fig. 1F). We then introduced point substitutions disrupting the target sites of miR-215 in pGL3-PCAT-1 and named the construct as pGL3-Mut215 (Fig. 1E) After HCC cells were co-transfected with this construct and miR-215 mimics or NC RNA, we found that there was no significantly decreased luciferase activity caused by miR-215 (all p Ͼ 0.05) (Fig. 1F).

TP53 response-miR-215 inhibits HCC proliferation via suppressing lncRNA PCAT-1
Two previous studies demonstrated that miR-215 could be actively induced by TP53 in multiple cancers including lung cancer, colorectal cancer, and ostosarcoma (23,24). However, it is still unclear if this exists in HCC. To test this, we examined miR-215 expression after overexpression of TP53 in both SMMC7721 and Huh7 cells (Fig. 2, A and B). We found that tumor suppressor TP53 does up-regulate miR-215 expression in HCC and, thus, down-regulation of lncRNA PCAT-1 (Fig. 2, A and B). Additionally, miR-215 inhibitors could rescue TP53-induced lncRNA PCAT-1 down-regulation in HCC cells ( Fig. 2C and supplemental Fig. S1), which indicates that TP53 down-regulates PCAT-1 in a miR-215-dependent manner in HCC cells.

miR-215 inhibits migration and invasion of HCC cells
Because impacts of lncRNA PCAT-1 on HCC invasion and metastasis were still largely unclear, we examined how PCAT-1 siRNAs and miR-215 regulate migration and invasion of HCC cells. The wound-healing assays demonstrated that PCAT-1 siRNAs and miR-215 mimics impaired the motility of the HCC cells compared with control cells transfected with NC RNA (Fig. 4, A and B). On the contrary, miR-215 inhibitors (in-215) could significantly accelerate migration of HCC cells (Fig. 4, A  and B).

miR-215-PACT1-CRKL axis in HCC
Next, the impact of miR-215 on invasiveness of SMMC7721 and Huh7 cells was investigated using the Matrigel invasion assay system. Reduced invasion ability of HCC cells was observed after elevated expression of miR-215 (Fig. 4C). In line with this, PCAT-1 siRNAs can also inhibit the invasion of these HCC cells (Fig. 4C). We also confirmed this observation using inhibitors of miR-215 and found enhanced invasion ability of HCC cells transfected with its inhibitors (Fig. 4C).
Thirteen genes identified by microarray profiling were then selectively validated using qRT-PCR (Fig. 5B). These genes are ones that have been reported to either be involved in cancer development or relatively high expression in SMMC7721 cells. Among these successfully validated downstream genes, we chose the oncogenic kinase CRKL as the candidate gene to examine (25)(26)(27). As shown in Fig. 5, C and D, and supplemen-tal Fig. S2D, silencing CRKL expression could significantly suppress viability of HCC cells, in both dose-and time-dependent ways (all p Ͻ 0.05). Colony formation of HCC cells also support the oncogene nature of CRKL (Fig. 5E). In Fig. 5F, we found that overexpression of CRKL prevents the anti-tumorigenic properties of miR-215 in HCC.

miR-215 inhibits HCC growth in vivo
We found that the growth of tumors from SMMC7721 xenografts with miR-215 up-regulation or knockdowns of PCAT-1 (shPCAT-1) and CRKL (shCRKL) was significantly inhibited compared with that of tumors formed from control xenografts (Fig. 6A). Expression of miR-215, lncRNA PCAT-1, and CRKL in xenografts was examined (Fig. 6B). The pathological detection with H&E staining further confirmed the HCC nature of SMMC7721 xenografts (Fig. 6C). Immunohistochemistry assay showed that the CRKL protein expression was obviously inhibited in tumor tissues with high miR-215 expression or low PCAT-1 expression (Fig. 6C). A working model summarizing relationships between TP53, miR-215, lncRNA PCAT-1, and CRKL is shown in Fig. 6D.

Discussion
Cancer is a genetic disease that changes cellular information flow to modify cellular homeostasis and promote growth (6 -10). Accumulated evidences demonstrated that lncRNAs and miRNAs play an important role during carcinogenesis (6 -15). Although first identified in prostate cancer, lncRNA PCAT-1 has been shown to be involved in development of multiple cancers as an oncogene (19, 29 -31). However, its fineregulation at the post-transcriptional level and downstream signaling pathway in HCC were still unclear. To the best of our knowledge, we here for the first time revealed that miR-215 can suppress lncRNA PCAT-1 expression and HCC viability in vitro. Gene expression profiling data indicated that CRKL might be one of the key downstream effector genes in HCC. In line with this notion, in vivo xenografts results highlight that this miRNA-mediated epigenetic regulation might be therapeutically relevant for HCC.

miR-215-PACT1-CRKL axis in HCC
CRKL encodes a protein kinase containing SH2 and SH3 domains, which has been shown to activate the RAS and JUN kinase signaling pathways. The kinase is vital in malignant transformation of multiple cancers, including HCC (25,26,28,42,43). Liu et al. applied in situ proximity ligation assay and determined that the novel interaction, CRKL-FLT1, has a high centrality ranking, and the expression of this interaction is strongly correlated with the migratory ability of HCC. They also found that knockdown of CRKL in HCC cells leads to a decrease in cell migration (28). Therefore, we speculate that the miR-215-PCAT-1 axis may interrupt CRKL and its downstream signaling pathway and, thus, suppress malignant phenomena of HCC cells. Interestingly, Tamura et al. (43) found that the TP53 target miR-200b/200c/429 miRNAs are negative regulators of the CRKL oncogene in cancer cells. Endogenous CRKL expression was decreased in cancer cells through the introduction of the TP53 family and endogenous TP53 activation, which is consistent to our results that TP53-response
In all, we identified lncRNA PCAT-1 as a novel target gene of miR-215. This fine post-transcriptional regulation significantly impacts multiple malignant phenomena of HCC cells. The identification of CRKL as a potential downstream gene of the miR-215-PCAT-1 axis highlights the involvement of its downstream signaling pathways in HCC. Our findings highlight the interaction between miRNAs and lncRNA PCAT-1 during tumorigenesis and progression of HCC cells.

Quantitative reverse transcription-PCR (qRT-PCR)
Each RNA sample was treated with RNase-free DNase to remove genomic DNA (Invitrogen) after its isolation from culture cells with TRIzol reagent (Invitrogen). These RNA samples were reverse transcribed into cDNAs using Revert Ace kit (TOYOBO, Osaka, Japan). Human U6 and miRNAs were examined with their specific stem-loop RT-PCR primers (Ribobio, Guangzhou, China). SYBR Green qRT-PCR was used to detect expression of PCAT-1 and other potential PCAT-1 downstream genes. The expression of individual genes was calculated relative to the ␤-actin expression (11)(12)(13).

PCAT-1 reporter and mutant constructs
The pGL3-Control plasmid containing the firefly luciferase gene as a reporter (Promega) was digested with XbaI (Promega) and treated with mung bean nuclease (Promega) to form blunt ends. The sequence corresponding to the wild-type PCAT-1 (100 -610 nucleotides) was amplified with Huh7 cDNA using Pyrobest TM DNA Polymerase (TaKaRa). The PCR primer pair used was: 5Ј-CATCTGTACCCTTACAATTTG-3Ј/5Ј-GCGC-ACCCTTTGACCCTTGG-3Ј. The PCR products were ligated into the appropriately digested pGL3-Control. The resultant plasmid, designated pGL3-PCAT-1, was sequenced to confirm the orientation and integrity. The PCAT-1 reporter gene plasmid with the mutant miR-215-binding site was constructed with QuikChange Site-directed Mutagenesis kit (Stratagene) and named as pGL3-Mut215.
To address if miR-215 fails to inhibit proliferation in a model in which PCAT-1 is not down-regulated, we first cloned fulllength lncRNA PCAT1 in the pcDNA3.1 expression construct named as PCAT1 (wild-type). The PCAT-1 expression construct with the mutant miR-205-binding site was constructed with a QuikChange Site-directed Mutagenesis kit (Stratagene) and named as PCAT1 mutant.

miR-215-PACT1-CRKL axis in HCC
activities were examined at 48 h after transfection using a luciferase assay system (Promega). Three independent transfections were done for each luciferase construct (each in triplicate). Fold-changes were calculated by defining the activity of the pGL3-Control vector as 1.

Colony formation assays
A total of 1,500 SMMC7721 or Huh7 cells were seeded into a 6-well cell culture plate and transfected with 20 nmol/ liter of miR-215 mimics, siPCAT-1-1, siPCAT-1-2, siCRKL-1, siCRKL-2, or NC RNA, respectively. Cells were washed twice with cold PBS and fixed with 3.7% formaldehyde 10 days later. Colony number in each well was counted after cells were dyed with crystal violet.

Wound healing and transwell assays
The wound healing and transwell assays were performed as reported previously (46). In details, the SMMC7721 or Huh7 cell layer was scratched when reaching about 90% confluence. HCC cells were then continued to be cultured in a 37°C CO 2 incubator. The average extent of wound closure was quantified. During transwell assays, the transwell chambers were coated with 100 ml of BD Matrigel overnight in a cell incubator. SMMC7721 or Huh7 cells transfected with miR-215 mimics, PCAT-1 siRNAs (siPCAT-1-1 and siPCAT-1-2), miR-215 inhibitors (in-215), or NC RNA were added to upper transwell chambers (pore 8 mm, Corning). A medium containing 10% FBS (650 ml) was added to the lower wells. After 48 h incubation, cells were fixed and stained, and the nonmigratory cells were scraped from the upper part of the filter. Cells migrated to the lower wells through pores were stained with 0.2% crystal violet solution and counted.

Gene expression profiling
Huh7 cells were transfected with 20 nmol/liter of NC RNA, miR-215 mimics, or siPCAT-1-1 and total RNA was extracted. Gene profiling of these samples was measured using OneArray Plus chips (Phalanx Biotech Group). Differentially expressed genes between the NC RNA group and the miR-215 group, or the NC RNA group and the siPCAT-1-1 group were identified separately. The differentially expressed genes were identified following the criteria log2 (fold-change) Ն0.585 and p Ͻ 0.05. The gene profiling data have been deposited at the National Center for Biotechnology Institute Gene Expression Omnibus (GEO) repository under accession number GSE92648.

HCC xenograft
We next evaluated the tumor suppressor role of miR-215 in vivo via HCC xenografts. The miR-215 mature sequence was first cloned after the CMV promoter into the pcDNA3.1 vector. The plasmid was named as miR-215 and transfected into SMMC7721 cells for G418 (Geneticin) selection. We isolated a stable cell clone with relative high expression of miR-215. We purchased 5-week-old female nude BALB/c mice from Vital River Laboratory (Beijing, China). A total of 1 ϫ 10 8 SMMC7721 cells with stable transfection of miR-215 or the pcDNA3.1 vector (NC) were inoculated subcutaneously into the fossa axillaris of 8 nude mice (n ϭ 4 per group). We measured tumor volumes every day after tumor volumes equaled to or were greater than 50 mm 3 . All procedures involving mice were approved by the institutional Review Board of Cancer Hospital affiliated to Shandong University.

Histologic and immunohistochemical analyses
The HCC xenograft tissue sections were stained with H&E for histologic analyses. The slides were viewed and photographed. Formalin-fixed, paraffin-embedded HCC xenograft tissue samples were used for immunohistochemical analysis. Antibody against CRKL (Abcam) was used to determine proliferation abilities of tumor cells. Stained slides were read independently by two pathologists.

Statistics
Student's t test was used to calculate the difference between two groups. Differences between three or more groups were analyzed via one-way analysis of variance. A p value of less than 0.05 was used as the criterion of statistical significance. All analyses were performed with the SPSS software package (Version 16.0, SPSS Inc.).
Author contributions-M. Y. and Y. R. contributed to conception and design. Y. R., J. S., and J. L. contributed to acquisition of data, or analysis and interpretation of data. M. Y., Y. R., W. L., Z. Z., and J. Y. drafted the article.
Note added in proof-There were several errors in the version of this article that was published as a Paper in Press on September 8, 2017. In Fig. 6C, the miR-215 and shPCAT1 panels for the CRKL immunohistochemical analysis was inadvertently duplicated. The actin immunoblot in Fig. 5C was inadvertently used in supplemental Fig.  S2F. There was also some textual overlap with Ge et al. (11). These errors have now been corrected and do not affect the results or conclusions of this work.