A potential target for liver cancer management, lysophosphatidic acid receptor 6 (LPAR6), is transcriptionally up-regulated by the NCOA3 coactivator

Lysophosphatidic acid receptor 6 (LPAR6) is a G protein–coupled receptor that plays critical roles in cellular morphology and hair growth. Although LPAR6 overexpression is also critical for cancer cell proliferation, its role in liver cancer tumorigenesis and the underlying mechanism are poorly understood. Here, using liver cancer and matched paracancerous tissues, as well as functional assays including cell proliferation, quantitative real-time PCR, RNA-Seq, and ChIP assays, we report that LPAR6 expression is controlled by a mechanism whereby hepatocyte growth factor (HGF) suppresses liver cancer growth. We show that high LPAR6 expression promotes cell proliferation in liver cancer. More importantly, we find that LPAR6 is transcriptionally down-regulated by HGF treatment and that its transcriptional suppression depends on nuclear receptor coactivator 3 (NCOA3). We note that enrichment of NCOA3, which has histone acetyltransferase activity, is associated with histone 3 Lys-27 acetylation (H3K27ac) at the LPAR6 locus in response to HGF treatment, indicating that NCOA3 transcriptionally regulates LPAR6 through the HGF signaling cascade. Moreover, depletion of either LPAR6 or NCOA3 significantly inhibited tumor cell growth in vitro and in vivo (in mouse tumor xenograft assays), similar to the effect of the HGF treatment. Collectively, our findings indicate an epigenetic link between LPAR6 and HGF signaling in liver cancer cells, and suggest that LPAR6 can serve as a biomarker and new strategy for therapeutic interventions for managing liver cancer.

Lysophosphatidic acid receptor 6 (LPAR6) is a G proteincoupled receptor that plays critical roles in cellular morphology and hair growth. Although LPAR6 overexpression is also critical for cancer cell proliferation, its role in liver cancer tumorigenesis and the underlying mechanism are poorly understood. Here, using liver cancer and matched paracancerous tissues, as well as functional assays including cell proliferation, quantitative realtime PCR, RNA-Seq, and ChIP assays, we report that LPAR6 expression is controlled by a mechanism whereby hepatocyte growth factor (HGF) suppresses liver cancer growth. We show that high LPAR6 expression promotes cell proliferation in liver cancer. More importantly, we find that LPAR6 is transcriptionally down-regulated by HGF treatment and that its transcriptional suppression depends on nuclear receptor coactivator 3 (NCOA3). We note that enrichment of NCOA3, which has histone acetyltransferase activity, is associated with histone 3 Lys-27 acetylation (H3K27ac) at the LPAR6 locus in response to HGF treatment, indicating that NCOA3 transcriptionally regulates LPAR6 through the HGF signaling cascade. Moreover, depletion of either LPAR6 or NCOA3 significantly inhibited tumor cell growth in vitro and in vivo (in mouse tumor xenograft assays), similar to the effect of the HGF treatment. Collectively, our findings indicate an epigenetic link between LPAR6 and HGF signaling in liver cancer cells, and suggest that LPAR6 can serve as a biomarker and new strategy for therapeutic interventions for managing liver cancer.
During the past years, multiple targets were characterized to develop new drugs for liver cancer; nevertheless, many of these drugs failed in Phase III trials (1,2). Currently, sorafenib is the only first-line drug in liver cancer treatment, yet the five-year survival rate remains low. Therefore, finding new targets and developing specific drugs is still an urgent issue (3)(4)(5). Hepatocyte growth factor (HGF) 3 is a multipotent cytokine secreted by mesenchymal cells, acting mainly on epithelial-derived cells (6). It binds to the c-Met receptor, activating tyrosine kinase cascade to regulate cellular physiological properties that endow an important role in angiogenesis, tissue regeneration, and tumorigenesis (7)(8)(9)(10). The abnormal HGF/MET signaling mediated by Met receptor mutation, cross-talk of Met with EGFR/ ERBB2/IGF1R, and the promoting effect to angiogenesis all make HGF an enhancer of tumorigenesis (11)(12)(13)(14). Surprisingly, HGF expression is lower in some liver tumor tissues compared with paracancerous tissues (15)(16)(17). We and others have previously reported that HGF can specifically inhibit the growth of HepG2 cells through ERK activation (18 -21). Although some potential targets and pathways were identified, a detailed regulatory mechanism and in vivo evidence of this interesting phenomenon remains poorly defined (22)(23)(24).
Lysophosphatidic acid receptor 6 (LPAR6), a G proteincoupled receptor that is highly expressed in epithelial cells and hair follicles, mediates cAMP accumulation and Rho-dependent cellular morphological changes (25,26). Some mutations in this gene have been found to cause hypotrichosis (27,28). Surprisingly, both in liver cancer cell lines and patient tumors, LPAR6 positively correlates with proliferative activity (29,30).
However, the underlying molecular mechanism is still largely unknown.
NCOA3 is a member of the steroid receptor coactivator family (31). NCOA3 has intrinsic histone acetyltransferase (HAT) activity and contains two transcriptional activation domains that recruit CBP/p300 and histone methyltransferases (32)(33)(34). Previous studies have revealed that NCOA3 expression is elevated in multiple tumor types (33). Also that NCOA3 overexpression contributes to cancer initiation, metastasis, and chemoresistance by mostly activating signaling cascades leading to uncontrolled proliferation (35). However, no correlation between NCOA3 and LPAR6 has been found so far.
In this study, we aimed to understand the role of LPAR6 in liver tumorigenesis and the underlying mechanism for LPAR6 regulation. We found that LPAR6 was overexpressed in liver tumor tissues and contributed to HepG2 cell proliferation. Moreover, HepG2 cells treated with HGF showed LPAR6 down-regulation in an NCOA3-dependent manner. More importantly, loss of either LPAR6 or NCOA3 significantly inhibited tumor cell growth in vitro and in vivo. Chromatin immunoprecipitation (ChIP) assay revealed that NCOA3 enrichment was closely correlated with the H3K27ac level at the LPAR6 locus in response to HGF treatment, indicating that NCOA3 transcriptionally regulates LPAR6 as part of the HGF signaling cascade. Moreover, HGF demonstrated strong inhibition toward HepG2-developed xenograft tumor growth, providing promising evidence for in vivo usage of HGF in treating liver cancer. Our study reveals a novel epigenetic regulatory mechanism for HGF inhibition on HepG2 cell growth and provides in vivo evidence for the therapeutic potential of HGF and its downstream targets.

LPAR6 is highly expressed in liver cancer and closely related to liver cancer patient survival
To determine the role of LPAR6 in hepatocellular carcinoma, we analyzed LPAR6 expression in liver cancer and matched paracancerous tissues. Immunostaining of liver specimens in IRS (immunoreactivity score) between tumors and paracancerous tissues is based on the intensity of LPAR6 staining. Image-Pro Plus 6.0 was used for further IRS analysis. Both histochemistry and integrated optical density (IOD)/area of LPAR6 positivity in images indicated significantly higher expression of LPAR6 in tumors (Fig. 1, A and B). Moreover, survival analysis of 63 patients revealed a negative correlation between LPAR6 expression and patient prognosis (Fig. 1C). Taken together, these results suggest that highly expressed LPAR6 in liver cancer patients may promote tumor growth and lead to poor survival.
To further confirm our findings, we verified LPAR6 expression by analyzing data from the TCGA database. We found that LPAR6 was highly expressed in liver cancer (p Ͻ 0.0001) compared with normal liver tissue (Fig. 1D). Ethnic analysis revealed a significant increase in LPAR6 expression in Caucasian and Asian liver cancer samples (Fig. 1E). Both tumor grade and stage showed that LPAR6 expression increased in tumor compared with paracancerous tissue (Fig. 1, F and G). In addi-tion, the ENCODE database showed that LPAR6 transcription was also significantly increased in HepG2 cells compared with normal liver cells (Fig. 1H). Studies show that proto-oncogene c-Myc or RhoA directly link to liver cancer (36,37). Thus, we analyzed their correlation with LPAR6 in liver cancer. Pearson correlation coefficient analysis showed a strong correlation between LPAR6 and c-Myc or RhoA expression levels (Fig. 1I), suggesting that LPAR6 may play a key role in carcinogenesis of liver cancer and may be a potential therapeutic target.

LPAR6 promotes HepG2 cell proliferation
To test whether LPAR6 does play a role in liver carcinogenesis, we knocked down LPAR6 with shRNA ( Fig. 2A) and performed a proliferation activity assay using EdU staining. We found that EdU-positive cells were remarkably reduced with LPAR6 knockdown (Fig. 2B). Clonogenicity was also diminished in LPAR6 knockdown cells (Fig. 2, C and D). In addition, LPAR6 knockdown significantly attenuated HepG2 and Huh7 cell proliferation (Fig. 2, E and F), which may be due to cell cycle arrest because LPAR6 depletion caused G1 phase arrest (Fig.  2G). More importantly, LPAR6 overexpression significantly promoted HepG2 and Huh7 cell proliferation (Fig. 2, H and I).
Together, these data indicate that LPAR6 is a key factor that promotes HepG2 cell growth.

HGF inhibits HepG2 proliferation by down-regulating LPAR6
Because HGF can specifically inhibit HepG2 cell proliferation (19,20), we sought to investigate the underlying mechanisms. We first tested the proliferation of HepG2 cells cultured with 50 ng/ml of HGF for 5 days, and found that HGF indeed inhibited HepG2 proliferation and clonogenicity (Fig. 3, A-C). To further test the anti-tumor effect of HGF, we cultured HepG2 cells with HGF at a final concentration of 50 ng/ml for 4 days and then implanted cells into the dorsal ventral skin of nude mice. The tumor volume was monitored for 30 days, and the tumor size in the HGF-treated group was much smaller than that in the control group (Fig. 3D). IHC results revealed that LPAR6 expression was reduced in HGFtreated tumor tissues (Fig. 3E). These results suggest that LPAR6 may be involved in the anti-proliferative effect of HGF in cancer.
To explore genes potentially involved in HGF-induced HepG2 proliferation inhibition, we performed whole-genome transcriptome analysis by RNA-seq. For the convenience of sequencing data analysis, we set T to represent the HGF-treated group and C as untreated cells. To study the differentially expressed genes shared between the replicates, we divided the experimental data into three groups (T1 versus C1, T2 versus C2, T1 ϩ T2 versus C1 ϩ C2) and performed differential gene expression analyses between the groups. We found 723 differentially expressed genes between HGF-treated and untreated cells (T1 ϩ T2 versus C1 ϩ C2) and 653 differentially expressed genes shared between the three groups (Fig. 3F). Among the significantly changed genes, 587 genes were up-regulated and 136 genes were down-regulated by HGF in T1 ϩ T2 versus C1 ϩ C2 (Fig. 3G). Interestingly, LPAR6 was transcriptionally downregulated by HGF (Fig. 3H), and the reduction was also confirmed by immunoblotting and quantitative real-time PCR NCOA3 up-regulates LPAR6 transcription in liver cancer (qPCR) (Fig. 3, I and J). Immunofluorescence staining with LPAR6 antibody showed weaker fluorescence signals in HGFtreated HepG2 cells compared with untreated cells (Fig. 3K). These results indicate that LPAR6 is a key downstream gene regulated by HGF in liver cancer. Studies implied that HGF inhibited HepG2 cell growth through the MEK/ERK signaling pathway, thus we tested whether MEK inhibitor U0126 blocked HGF-induced LPAR6 decrease. Indeed, U0126 rescued LPAR6 expression in HepG2 cells treated with HGF ( Fig. 3L), indicating the regulation of LPAR6 expression by HGF was also mediated by the MEK/ERK signaling pathway. In addition, LPAR6 overexpression stimulated cell proliferation, suggesting that LPAR6 could partially rescue HGFinduced HepG2 cell proliferation arrest (Fig. 3M). These results further support that LPAR6 is important for liver cancer growth.

H3K27ac enrichment is elevated at LPAR6 promoter in HepG2 cells
To address the underlying mechanism regarding how HGF regulate LPAR6 expression, we mainly focused on the epige-netic regulation on LPAR6 transcription. First, we analyzed the profile of histone modifications at LPAR6 locus. Compared with normal donor hepatocytes, H3K27ac (active transcription marker) was significantly enriched in the promoter region of LPAR6 in HepG2 cells, whereas H3K9ac (active transcription marker), H3K9me3 and H3K27me3 (both are suppressive transcription markers) had slight differences (Fig. 4A, Fig. S1). Therefore, we hypothesized that certain histone acetylationmodifying enzymes or co-activators might be involved in LPAR6 regulation. To identify these regulators, we performed qPCR targeting all histone acetylation modifiers and found that NCOA3 transcription was significantly up-regulated in HepG2 compared with liver cell line LO2 (Fig. 4B). Thus, we speculated that NCOA3 might serve as an epigenetic regulator of LPAR6 transcription. Western blotting results indicated that LPAR6 expression was indeed down-regulated with NCOA3 knockdown (Fig. 4C).
To verify whether NCOA3 also plays a role in promoting cancer growth, we knocked down NCOA3 in HepG2 cells and performed functional analyses. Cell proliferation measured by Log-rank test shows statistically significant differences between high and low groups (p ϭ 0.0034). According to the LPAR6 optical density of IHC specimens and survival status events (0 for survival, 1 for death), the cutoff value was obtained by ROC curve analysis. IHC specimens were divided into high and low expression groups by cutoff value. D, analysis of the LPAR6 expression level between normal liver tissues versus liver cancer tissues. Data represent mean Ϯ S.D. Significance of expression level differences was determined using Student's t test (p Ͻ 0.0001). E-G, expression level of LPAR6 in different race, grade, and tumor stage on the basis of liver cancer compared with normal liver tissues. Data represent mean Ϯ S.D. Significance of expression level differences was determined using Student's t test (p Ͻ 0.05, p Ͻ 0.001, and p Ͻ 0.0001). H, comparison of LPAR6 transcription levels in HepG2, hepatocytes, and normal liver cells of donor. TPM, transcripts per million reads. I, linear regression analyses of LPAR6, C-Myc, and RhoA, respectively. Pearson correlation coeffients show strong correlation of expression level between LPAR6 and c-Myc or RhoA. Raw data were obtained from OncoLnc database.

NCOA3 up-regulates LPAR6 transcription in liver cancer
various methods were significantly reduced in HepG2 and Huh7 cells lacking NCOA3, whereas overexpression of NCOA3 accelerated Huh7 proliferation and colony formation (Fig. 4, D-I). As LPAR6 was down-regulated after NCOA3 knockdown, we further asked whether compensation of LPAR6 could rescue cell proliferation deficiency caused by NCOA3 depletion. The results showed that LPAR6 overexpression was able to rescue cell proliferation from NCOA3-knockdowninduced inhibition in HepG2, Huh7, and SK-Hep1 (Fig. 4J). Earlier we showed that U0126 also rescued NCOA3 expression (Fig. 3G). These combined results suggest that LPAR6 is transcriptionally regulated by NCOA3, and HGF may inhibit liver cancer proliferation by blocking NCOA3 and LPAR6 expression.

NCOA3 regulates H3K27ac enrichment at LPAR6 locus in response to HGF stimulation
To gain further insights into the role of NCOA3 regulation on LPAR6 expression, we investigated the distribution of histone modifications at the LPAR6 gene locus. H3K27ac enrichment was remarkably reduced in the LPAR6 promoter (site 2 and site 3) and coding sequence region (site 1) in NCOA3knockdown cells (Fig. 5, A and B), whereas H3K9ac, H3K4me3, H3K9me3, and H3K27me3 enrichment was only slightly changed (Fig. S2). The reduced H3K27ac enrichment at the LPAR6 promoter was not due to diminished H3 expression, because knocking down NCOA3 did not affect H3 expression (Fig. 4C). These results suggest that NCOA3 regulates LPAR6

NCOA3 up-regulates LPAR6 transcription in liver cancer
transcription by manipulating H3K27 acetylation in HepG2 cells. The mechanism of NCOA3 regulating H3K27ac enrichment at the LPAR6 locus was similarly replicated in Huh7 cells (Fig. 5C).
Because HGF inhibited NCOA3 and LPAR6 expression (Fig.  5D), we speculate that HGF may also affect H3K27ac deposition at LPAR6 locus. As we expected, H3K27ac enrichment at the LPAR6 promoter (site2 and site3) and coding sequence NCOA3 up-regulates LPAR6 transcription in liver cancer region (site 1) was also significantly reduced in response to HGF treatment in HepG2 and Huh7 cells (Fig. 5, E and F). H3K36me3 enrichment also showed remarkable reduction at the LPAR6 coding sequence region (site1), but not at the promoter region (sites 2 and 3, Fig. S3).
Meanwhile, HGF treatment also significantly reduced NCOA3 enrichment at the LPAR6 loci (sites 1-3, Fig. 5G), and the NCOA3 distribution pattern was consistent with H3K27ac. Taken together, these data indicate that HGF-induced downregulation of LPAR6 transcription attributes mainly to H3K27ac enrichment reduction at LPAR6 loci, and partially to the H3K36me3 reduction in the LPAR6 coding-sequence region (Fig. S3). Generally, acetylation removes positive charges on the histones. As a consequence, the condensed chromatin is transformed into a more relaxed structure that is associated with active transcription. In addition, the gene body marked with H3K36me3 is also associated with active transcription. Therefore, the above results indicate that HGF may regulate transcription by remodeling the chromatin structure to fulfill its anti-tumor function in liver cancer.

The deficiency of LPAR6 and/or NCOA3 limits cell proliferation in vivo
Both NCOA3 and LPAR6 knockdown significantly suppresses the growth of HepG2 cells in vitro. Therefore, we compared HepG2-derived xenograft tumor growth among the control, LPAR6-knockdown, and NCOA3-knockdown cells. Either LPAR6-or NCOA3-knockdown HepG2 cells and control cells (n ϭ 6/group) were, respectively, inoculated into the dorsal ventral side of nude mice. After 30 days of growth, the average volume and weight of LPAR6-knockdown (shLPAR6) tumors were obviously smaller than those in the control group (NT), suggesting LPAR6 deficiency affected cell proliferation in vivo (Fig. 6A). Similarly, the mean volume and weight of tumors with NCOA3-knockdown were also much smaller than the control tumors when monitored at Day 20. Given H3K27ac enrichment at the LPAR6 promoter, these data imply that LPAR6 requires H3K27ac and NCOA3 to drive proliferation (Fig. 6B).

HGF significantly limits liver cancer cell proliferation in vivo
Given that HGF inhibits HepG2 cell proliferation, we speculated that HGF could limit liver cancer cell proliferation in vivo.
To test this possibility, we generated xenografts with HepG2 cells and cultured them with HGF in vitro for 4 days. Twentyeight days after inoculation, we found that the tumor size and weight were significantly smaller in the HGF-treated group (Fig. 6C). EdU staining and apoptosis testing of HepG2 cells suggested that the cells were viable before being implanted into mice (Fig. S4).
To further examine the effect of HGF treatment, we directly injected HGF into the tumors in mice. After treating with HGF for seven times, the mean volume of HGF-treated tumors was smaller (Fig. 6D). However, during the treatment period, the mean tumor dimension for the HGF-treated group was slightly larger than the control group (data not shown), which may be due to other HGF-induced physiological functions such as angiogenesis promotion. Therefore, to limit this side effect of HGF on tumor, we carried out a combinatorial treatment with sorafenib, an anti-cancer drug inhibiting vascular endothelial growth factor receptor, platelet-derived growth factor receptor, and Raf family kinases. As we expected, the size and weight of tumors treated with the sorafenib-HGF combined therapy were much smaller than those treated with sorafenib only (Fig. 6E).

HGF treatment inhibited LPAR6 expression in liver cancer xenograft
Thus far, our data demonstrated that NCOA3 transcriptionally up-regulated LPAR6 in liver cancer cells, and HGF inhibited LPAR6 expression. However, how HGF signals through NCOA3 to regulate LPAR6 in liver cancer patients, and whether NCOA3 and LPAR6 function in concert have not yet been fully determined. To this end, we first examined the expression pattern of HGF and LPAR6 in the same patient tumor tissue. Immunohistochemistry (IHC) results suggest that HGF and LPAR6 expression levels were negatively correlated (Fig. 7A). Interestingly, our results showed that tumors expressed low levels of HGF, much lower than their matched paracancerous tissues, but high levels of LPAR6 (Fig. 7, B and C).
Moreover, liver cancer xenograft tumor tissues showed significantly reduced LPAR6 expression in the combined treatment group compared with sorafenib-only group (Fig. 7D). The statistical analysis of histochemical indicators showed remarkable LPAR6 reduction in the combined treatment group (Fig.  7E). Through the analysis, we also found an HGF expression pattern that is favorable in normal liver tissue instead of tumor tissues (Fig. 7, F-H). Considering the remarkable anti-tumor effect of the sorafenib-HGF combination protocol in vivo, these results further support that LPAR6 is a potential target of liver cancer.

Discussion
The development of liver cancer drugs is almost stagnant. Even more frustrating is that the only first-line drug for liver cancer, sorafenib, provides poor benefits. Finding new targets and developing more drugs is necessary but challenging in the area of liver cancer. Here we identified a novel mechanism where HGF exerted the anti-proliferation effect in liver cancer by blocking NCOA3 activity in H3K27 acetylation at the LPAR6 promoter region, leading to down-regulated LPAR6 expression. In addition, the combination of HGF and sorafenib sufficiently inhibited the growth of HepG2-derived xenografts.
Overexpression of HGF in liver cancer models has revealed both tumor-promoting and tumor-inhibiting effects of HGF (16). HGF/Met signaling contributes to tumor angiogenesis, invasiveness, and oncogenesis. Tumor metastasis in many cancer types involves the HGF/Met pathway, leading to the rapid growth of HGF/Met pathway-targeted anticancer drug developmental programs (16,39). Nonetheless, several studies have also reported that HGF expression is down-regulated in some liver cancer tissues (15,16). In vitro experiments show that HGF can specifically inhibit the growth of HepG2 cells, but no repression effect was observed in Huh7 and SK-Hep1 cells (data not shown) (21,25). The low repression effect of HGF on SK-Hep1 cells may be due to loss of p53, whereas HepG2 has WT p53. Therefore, these results support the hypothesis that HGF expression is important for liver cancer development and may serve as a potential drug candidate for certain types of liver cancer. Further understanding of the molecular details downstream of the HGF-Met pathway may present additional therapeutic strategies.
LPAR6 is involved in cell growth, motility, and morphological changes (25,27). By analyzing the data from the TCGA database, we found high expression of LPAR6 in liver cancer (p Ͻ 0.0001), which was closely related to the poor prognosis of patients (p Ͻ 0.05). Through IHC staining, we also found that 88% of cancer samples showed high LPAR6 expression compared with paracancerous tissues. This observation is supported by an earlier study reporting that overexpression of LPAR6 in liver cancer specimens was associated with poor survival. Knockdown of LPAR6 inhibit liver cancer cell proliferation by reducing proto-oncogene PIM3 (29).
The HGF-Met signaling pathway plays an essential role in diverse developmental processes, and its dysregulation contributes to metastatic phenotypes of human cancers (40). Our study using liver cancer cell lines demonstrated that LPAR6 and NCOA3 functioned as novel key downstream effectors of the HGF-Met signaling pathway. HGF down-regulates LPAR6, suppressing HepG2 proliferation by arresting the cell cycle at the G1 phase. Our observation supports previous findings that knockdown of LPAR6 leads to proliferation arrest of liver cancer cell lines, and liver cancer patient tumor tissues with LPAR6 overexpression showed higher proliferative activity (29,30). These results suggest that LPAR6 is a downstream target of HGF and could serve as a potential therapeutic drug target.
Our study showed that H3K27ac was significantly enriched in the LPAR6 promoter region in HepG2 cells compared with normal liver cells. Given that the effects of histone acetylation on gene transcription activation are primarily mediated through relaxing the chromatin high order structure to allow transcription factors and RNA polymerases to access gene promoter, we postulated that H3K27ac was a major epigenetic modification involved in the up-regulation of LPAR6 transcription in liver cancer cells. By screening, we found that NCOA3 transcription was significantly higher in HepG2 than normal liver cells. As a transcriptional coactivator, NCOA3 has intrinsic HAT activity as well as the capacity of recruiting the CBP/ p300-associated factor and CREB-binding protein to form a transcriptional activation complex (34,41,42). Our further investigation revealed NCOA3 knockdown in HepG2 downregulated LPAR6. Moreover, HepG2 cells treated with HGF showed decreased expressions of NCOA3 and LPAR6 with reduced H3K27ac and H3K36me3 enrichment at LPAR6 loci. The decreased transcription of LPAR6 is further confirmed in NCOA3-depleted cells. Together, we show for the first time that HGF down-regulates NCOA3, leading to reduced transcription of LPAR6 with diminished enrichment of H3K27ac on its promoter locus, thus demonstrating anti-tumorigenesis property.
Currently, sorafenib serves as a first-line anti-liver cancer drug. Therefore, we evaluated the effect of the sorafenib-HGF combination therapy on liver cancer and found that the combinatorial therapy remarkably suppressed the growth of HepG2-derived xenografts. Thus, the HGF-sorafenib combination protocol might be a promising therapeutic strategy for liver cancer patients that have high LPAR6 but low HGF expression. Given the critical role of HGF signaling and the sharing of some downstream signaling components among different growth factor signaling pathways, LPAR6 and NCOA3 may participate in other growth factor signaling pathways beyond the present study. Future investigation of the potential involvement of LPAR6 and NCOA3 in individual growth factor

NCOA3 up-regulates LPAR6 transcription in liver cancer
signaling pathways may shed additional light on the anti-cancer effect beyond HGF.

Cells and cell culture
HepG2, Huh7, and SK-Hep1 cells were purchased from the Cell Bank of the Chinese Academy of Sciences in 2016. Cells were generally passaged less than 5 times, and freshly thawed cells were maintained in culture for no more than 2 weeks before conducting experiments. Cells were routinely tested for mycoplasma contamination using PCR. All cells were cultured in Dulbecco's modified Eagle's medium (HyClone) containing 10% fetal bovine serum (PAN) and 100 units/ml of penicillin/ streptomycin (HyClone) at 37°C in a humidified incubator with 5% CO 2 .

Cell-proliferation assay and EdU-staining assay
For the cell-proliferation assay, cells were seeded at a density of 6 ϫ 10 3 /well in 96-well-plates and maintained with HGF for 5 to7 days. Cell proliferation was measured using a CCK8 kit at serial time points as indicated in the figures (Dojindo Laboratories). For the EdU assay, cells were preincubated with 50 M EdU (RiboBio) for 2 h, and fixed with 4% paraformaldehyde for 30 min and neutralized with 2 mg/ml of glycine. After washing with PBS for three times, cells were stained with Apollo for 30 min. Last, cells were treated with 0.5% Triton X-100 for 10 min and cellular DNA was stained with Hoechst for 10 min. After washing with PBS for three times, images were acquired using Zeiss LSM-710 microscope.

Colony assay and cell cycle analysis
Cells were plated at a density of 1 ϫ 10 3 /well into 6-wellplates, and cultured with Dulbecco's modified Eagle's medium containing 10% fetal bovine serum for 10 days. Cell colonies were detected with crystal violet staining. The photos of colonies were obtained using Molecular Imager Gel Do XRϩ system (Bio-Rad). For cell cycle analysis, cells were fixed with 70% ethanol overnight, and digested with trypsin for 30 min at 37°C before propidium iodide staining (KeyGEN BioTECH). At least 10,000 live cells were subject to FACS analysis on a FACS Calibur flow cytometer (BD Biosciences). Experimental data were analyzed using FlowJo software.

Quantitative real-time PCR
Total RNA was extracted using TRIzol reagent (Thermo-Fisher). Briefly, homogenized samples were incubated for 5 min and 0.2 ml of chloroform was added to every milliliter of sample homogenized with TRIzol reagent. After centrifuging at 12,000 ϫ g for 15 min at 4°C, the aqueous phase was transferred into a new tube with 500 l of isopropyl alcohol and incubated for 10 min. The pellets were collected by centrifuging at 12,000 ϫ g for 10 min at 4°C and washed twice with 75% ethanol. The resulted pellets were dissolved in diethyl pyrocarbonate-treated dH 2 O for further experiments. Reverse transcription was carried out using the PrimeScript TM RT reagent kit with gDNA Eraser (Takara) according to the manufacturer's protocol. Quantitative PCR using SYBR Green Supermix (Bio-Rad) was performed using CFX96 Real-time PCR System (Bio-Rad).

Virus infection
Lentivirus was packaged with short hairpin RNA targeting LPAR6 and NCOA3. The constructed shRNA plasmids of pLKO.1-LPAR6-puro or pLKO.1-NCOA3-puro were co-transfected with packing plasmids psPAX2 and PMD2.G into 293T cells. The viral supernatants were harvested, pooled, and filtered with 0.45-m PES membrane filter, and used to infect target cells. The cells were selected with puromycin.

Immunohistochemistry and immunofluorescence
Specimens of liver cancer tissues were collected for IHC analysis of LPAR6 and HGF. Antibodies to LPAR6

NCOA3 up-regulates LPAR6 transcription in liver cancer
(ab135447) and HGF (ab83760) were purchased from Abcam. For immunofluorescence analysis, the prepared cells were fixed with 4% paraformaldehyde and blocked with PBS containing 1% BSA and 0.1% Triton X-100. Slides were incubated with primary antibodies overnight at 4°C. Following four washes with 0.1% BSA/PBS/Triton X-100, cells were incubated with fluorescein-conjugated secondary antibody for 1 h at room temperature. After four washes, cells were stained with 0.1 g/ml of 4Ј,6-diamidino-2-phenylindole in PBS for 15 min at room temperature followed by two washes with PBS. Images were acquired using Zeiss microscope. HepG2 cells were infected with lentivirus containing shRNA against LPAR6 or nontargeting scrambled shRNA. After knockdown validation with immunoblotting, infected cells were implanted into the dorsal flanks of 5-week-old nude mice (NT, n ϭ 6; shLPAR6, n ϭ 6). After 30 days, we analyzed xenograft tumor volume and tumor weight. p value was analyzed with unpaired t test (***, p Ͻ 0.001). B, NCOA3 knockdown represses xenograft growth. Cell infection and implantation procedures were conducted as described in A (NT, n ϭ 3; shNCOA3, n ϭ 3). After 20 days, xenograft tumor volume and weight were analyzed. Data represent mean Ϯ S.D. *, p Ͻ 0.05 and **, p Ͻ 0.01, analyzed with unpaired t test. C, pre-treatment with HGF delayed xenograft tumor formation. HepG2 cells cultured with or without HGF treatment for 4 days, and then the same number of cells (HGF-versus HGFϩ) were implanted into the dorsal flanks of 5-week-old nude mice for 28 days and monitored (HGFϪ, n ϭ 6; HGFϩ, n ϭ 6). *, p Ͻ 0.05, analyzed with unpaired t test. D, HGF treatment inhibited xenograft tumor growth. HepG2-derived xenograft mice were divided into two groups based on similar tumor size (HGFϪ, n ϭ 6; HGFϩ, n ϭ 6), and then subjected to intratumoral injection of HGF continuously for 7 days. Control was intratumorally injected with 0.1% BSA as a same volume with HGF. Tumor growth was monitored for 30 days, and xenograft tumor volume and weight were analyzed at the end point. **, p Ͻ 0.01; ***, p Ͻ 0.001. E, xenograft mice were generated as described in D and then treated with either sorafenib-only or HGF-sorafenib combined therapy (HGFϪ, n ϭ 6; HGFϩ, n ϭ 6) for 10 days. Tumor growth was monitored for 54 days. *, p Ͻ 0.05 and **, p Ͻ 0.01, analyzed with unpaired t test.

RNA-seq library preparation and data analysis
Briefly, mRNA purified from total RNA using poly(A) selection was chemically fragmented and converted into singlestranded cDNA using random hexamer priming. Doublestranded (ds) cDNA was generated for TruSeq library construction. Short ds-cDNA fragments were linked with sequencing adapters, and suitable fragments were separated by agarose gel electrophoresis. Constructed TruSeq RNA libraries were quantified using quantitative PCR, and the quality was assessed by electrophoresis (Bioanalyzer 2100, Agilent Technologies). To analyze sequencing data, the transcript counts for gene expression levels were calculated and the relative transcript abundance was determined as fragments per kilobase of exon per million fragments mapped (FPKM) using Cufflinks software. Raw data were extracted as FPKM values across all samples, and samples with zero values across more than 50% of the genes were excluded.

Animal study
All animal procedures were performed with respect to the national and international Guidelines for the Care and Use of Laboratory, and also were approved by the Institutional Animal Care and Use Committee from Sichuan University. Four-weekold male Balb/c nude mice and NOD/SCID mice were purchased from Vital River Laboratory Animal Technology Co., Ltd (Beijing). The mice were housed under standard conditions. Mice were administered under pathogen-free conditions.

Tumor-xenograft assay
All animal procedures were performed with respect to the national and international Guidelines for the Care and Use of Laboratory Animals, and also were approved by the local Institutional Animal Care and Use Committee. Cells containing shLPAR6, shNCOA3, or control shRNA were implanted at a density of 9 ϫ 10 6 cells per mouse into the dorsal ventral side of 5-week-old nude mice, each with the same amount of Matrigel (Corning). Neoplasms were monitored for nearly 2 months. HepG2-derived xenograft nude mice were divided into two groups based on the tumor size 2 weeks post-implantation. HGF with or without sorafenib was injected into tumors once per day for a total of 10 days, compared with 0.1% BSA (with or without sorafenib) injection, respectively. All the mice were housed and supplied with water and food ad libitum. Tumor volumes (mm 3 ) were calculated with the following equation: volume ϭ (length ϫ width 2 )/2.