Lysophosphatidic Acid Receptor LPAR6 Supports the Tumorigenicity of Hepatocellular Carcinoma

Hepatocellular carcinoma (HCC) is a very frequent malignancy, being the sixth most common cancer worldwide (1). In addition, the incidence of HCC has been increasing over the last years, especially in western countries (2). The etiology of HCC is multifactorial, and the disease is very often preceded by other conditions, including chronic inflammation, liver fibrosis, and cirrhosis, that are frequently associated with hepatitis B virus or hepatitis C virus infection, carcinogen/toxin exposure including alcohol and aflatoxin; gender and metabolic conditions are other risk factors (3, 4). Given the multifactorial etiology of HCC, a better understanding of the common molecular pathogenic mechanisms underlying the tumorigenic process of this disease is essential to help identify more effective therapeutic targets. In this context, different molecular mechanisms driving the process evolving from chronic inflammation and cirrhosis to the formation of dysplastic nodules have been identified (5). These mechanisms may not only involve hepatocytes but also liver microenvironment elements, which contribute to HCC growth and progression (6–8). The HCC microenvironment is considered as a metastasis-promoting element in itself, and although the presence of distant metastases in HCC is rarely observed, intrahepatic metastases are very frequent, being a clear sign of tumor progression and an adverse prognostic factor. Although different mechanisms and predisposing factors for the development of intrahepatic metastases have been investigated (9, 10), intrahepatic metastasis remains a serious clinical problem and the main cause of HCC recurrence.

Lysophosphatidic acid (LPA) receptors are G protein–coupled receptors (GPCR) that bind the bioactive phospholipid LPA and activate multiple cellular responses, including cell proliferation, cytoskeletal rearrangements, motility, and apoptosis (11–13). At least five LPAR (LPAR1-5) are relatively well characterized and currently under investigation (14). LPAR6 is a newly identified receptor, originally referred to as purinergic receptor P2Y5 that has been involved in inherited forms of hair loss, specifically hypotrichosis simplex (15) and woolly hair (16). However, other than these reports, the role of LPAR6 in human disease including cancer has not been investigated. Here, we show that LPAR6 is essential for maintaining the proliferation capacity and the tumorigenic phenotype of HCC, and that the tumorigenicity control exerted by LPAR6 occurs through the transcriptional activation of proto-oncogene Pim-3. This mechanism sustains HCC growth and progression.

Human HCC cell lines, Huh7, PLC/PRF/5, and HLE and were maintained in DMEM supplemented with 10% bovine calf serum (Gibco; Life Technology), as previously reported (17).

Cultured Huh7 were infected by the lentiviral particles, containing 3 target-specific constructs that encode 19–25 nt (plus hairpin) shRNA designed to knock down LPAR6 gene expression and isolated by puromycin selection to obtain clonal propagation. Clones LPAR6-shRNA#8 and LPAR6-shRNA#9 showing a significant silencing of LPAR6 gene were selected and used for further experiments. Control-shRNA lentiviral particles encoding a scrambled shRNA sequence that will not induce specific degradation of any known cellular mRNA were used as negative control. To generate human HCC cell lines stably expressing the LPAR6 gene, 1 × 10⁵ HLE cells per 24-mm dish were transfected with 2 μg of pcDNA 3.1/V5-His-LPAR6, mutants pcDNA 3.1/V5-His-Gln 155X, and pcDNA 3.1/V5-His-Lys125AsnfsX37 or pcDNA 3.1/V5-His-neo as negative control using TransIT-LT1 Transfection Reagent (Mirus) according to the manufacturer's instructions. Selection of transfectant clones was carried out by adding 300 μg/mL of neomycin (G418) to the culture medium.

Briefly, 1 × 10⁴ cells were plated onto 24-well plates and allowed to grow up to 5 days in the presence of 10% FBS. At each time points, cells were manually counted using Bürker chamber. All experiments were performed in triplicate, and the viable cells were determined by vital dye.

Total RNA was isolated using the RNeasy Mini Kit (Qiagen) and underwent DNAse treatment (TURBO DNA-free Ambion). Reverse transcription (RT) was then carried out with random primers using 1 μg of total RNA from each sample. Quantitative RT-PCR was performed using predesigned primers and SYBR Green Master Mix (Bio-Rad). Housekeeping Glyceraldehyde 3-phosphate dehydrogenase gene (GAPDH) was used as endogenous control. The following primers were obtained from Integrated DNA Technologies: LPAR6: forward: 5′- TTT GCA CTG GCG TGT GGT T-3′, reverse: 5′-TCT GAG GCA TTG TTA CCC TGA-3′; Ki-67, forward: 5′-TTT CAT TTT ACA GGG CTG TTG ATG-3′, reverse: 5′-GGA GGG TTG TGT AGA AGT GGT GTT-3′; Pim-3, forward: 5′-TCA GTA CCT GTG TTT GTG TGA ATG C-3′, reverse: 5′-CCA GAC ATC TCA CTT TTG AAC TGA A-3′; GAPDH, forward: 5′-CCA CAT CGC TCA GAC ACC AT-3′, reverse: 5′-GCG CCC AAT ACG ACC AAA T-3′.

Cells (1 × 10⁵) in culture medium were mixed with a volume of 0.5% top agar twice higher and seeded in 60-mm plates onto a base layer of complete medium containing 0.5% agar in the presence of culture medium. The cells were incubated at 37°C in 5% CO2 in air. After 3 weeks, the number and the size of the colonies were counted and statistically analyzed.

The pGL4.10 (luc2) firefly luciferase reporter gene vectors (Promega) harboring the promoter of the human Pim-3 gene or the deleted mutants of the Pim-3 promoter were generated as previously described (18). For luciferase assays, HLE-LPAR6 or HLE-neo cells were transfected in 96-well plates with the same amount (0.1 μg) of pGL4 vectors containing deleted mutants (p809-luc, p264-luc, and p104-luc) or the entire Pim-3 promoter fragment (p2074-luc) using the TransIT-LT1 Transfection Reagent (Mirus), according to the manufacturer's instructions. Luciferase activities were measured 24 hours after transfection using a dual-luciferase reporter assay kit (Promega). Values are expressed as fold increase in luciferase activities compared with pGL4 empty vector control. All experiments were performed in triplicate.

All animal procedures were conducted in accordance with the national and international Guidelines for the Care and Use of Laboratory and were approved by the local Institutional Animal Care and Use Committee. For the evaluation of tumorigenicity after subcutaneous implantation, 4- to 5-week-old female CD-1 nude (nu/nu) athymic mice were subdivided into four groups. The mice were housed and received food and water ad libitum. Huh7 parental cells, control-shRNA, LPAR6-shRNA#8, and LPAR6-shRNA#9 from mid-log phase cultures were trypsinized, counted, and then resuspended in a 50% mixture of Matrigel (BD Biosciences) in PBS. A 0.2 mL volume of the cell suspension containing 5.0 × 10⁶ cells/mouse was injected s.c. in the right flank of each mouse. Tumor dimensions and body weights were recorded twice weekly. Tumor sizes (mm³) were calculated using the equation (w² × l)/2, where “w” and “l” refer to the width (mm) and length (mm) recorded at each measurement. For intrasplenic injection, 6-weeks-old male CD-1 nu/nu mice were subdivided into two groups, anesthetized by intraperitoneal injection of a mix of ketamine:xylasine (100:10 mg/kg) diluted in isotonic saline, and then splenically injected with 2 × 10⁶ HLE-LPA cells or HLE-neo in control mice using a 27-gauge needle. The spleen was removed 1 minute after injection and the wound closed using silk suture and surgical clips. Neoplastic progression was monitored based on the general health of the animals. Necropsies were carried out, and snap-frozen liver tumor samples were collected and sectioned at 5 μm for immunohistochemical and hematoxylin and eosin staining, or at 20 μm for RNA extraction.

RNA was extracted from cells and retro-transcribed into cDNA. The population of cDNAs was then bound to a streptavidin matrix via 3′-biotin shred to 50 to 500 bp fragments, and unbound fragments were discarded. The bound fragments were sequenced by next generation sequencing, starting at the fragmentation site, generating 50 to 500 bp “tags.” Tags were assembled into contigs and all tags annotated to database entries and counted and finally SNPs were identified. The ToppCluster multiple gene analyzer (Division of Biomedical Informatics, Cincinnati Children's Hospital Medical Center, Cincinnati, OH) was used to group the most significant genes according to their functions using Bonferroni correction and the Kamada–Kawai algorithm and the STRING (Search Tool for the Retrieval of Interacting Genes/Proteins) Database to predict protein interactions.

Samples of HCC tissues were obtained from 128 patients undergoing hepatic resection and classified according to the World Health Organization guidelines. Approval for the study was gained from the local ethics committee, and patients gave prior written informed consent to the use of their tissues. Immunohistochemistry and immunofluorescence were performed as previously described (2). For immunofluorescence analysis, cells or tissue sections were fixed with 4% paraformaldehyde and blocked with 1% BSA/PBS-T. Immunostaining was performed using the appropriate primary and secondary antibodies, and images were acquired using a Nikon eclipse E1000 microscope. For immunohistochemical analysis, frozen liver sections (8 μm) were fixed in acetone-chloroform for 10 minutes and incubated for 1 hour with a polyclonal antibody to LPAR6 (MBL International), and the binding was detected using the APAAP Kit (Dako). Slides were examined under a Nikon eclipse E1000 microscope. Tissues were quantitatively scored according to the percentage of positive cells and staining intensity.

The Kaplan–Meier method was used to estimate survival curves, and the log-rank test to test for differences between curves using Prism 5 (Graphpad Software, Inc.). For correlations, Pearson r coefficients were calculated. Data are represented as the mean ± SEM. Significant differences between groups were determined via unpaired two-way or one-way Student t test or ANOVA. P < 0.05 was considered statistically significant.

We examined mRNA expression levels of all known LPAR (LPAR1-6) in human HCC surgical specimens by real-time PCR. Among the six receptors investigated, LPAR6 was the isoform consistently most commonly expressed in analyzed tumor tissues (n = 35; Fig. 1A). We next confirmed the presence of elevated protein levels of LPAR6 in the same tissues by Western blotting. Interestingly, the LPAR6 protein was rarely expressed in paired peritumoral tissues and absent in normal control tissues (Fig. 1B). To evaluate the tissue distribution of LPAR6 protein, double immunofluorescence staining was done. This analysis revealed that LPAR6 is expressed by epithelial cancer cells in HCC, as shown by the colocalization with the hepatocyte marker Hepar-1, whereas no LPAR6 protein expression was detected in paired peritumoral tissues (Fig. 1C). mRNA expression levels of other LPAR (LPAR1-5) in peritumoral tissues are also shown (Supplementary Fig. S1A). Because LPAR6 is present in HCC tissues with different levels of expression, we stratified 128 patients according to their expression levels of LPAR6. Tissues were scored on the basis of the intensity of LPAR6 staining (low, medium, and high), and Kaplan–Meier survival analysis was performed. A significantly longer survival was shown in patients with lower LPAR6 expression levels (P < 0.001; Fig. 1D), indicating that expression levels of tumor-associated LPAR6 affect the clinical outcome in patients with HCC. Clinical characteristics of patients included in the survival analysis are shown in Fig. 1E.

We next investigated whether LPAR6 overexpression was relevant to the pathogenesis of HCC, hypothesizing a role of this receptor in controlling tumor proliferation and the tumorigenic phenotype. To gain insight into the mechanism by which LPAR6 regulates growth and tumorigenicity, we used both in vitro and in vivo approaches. Firstly, we screened five different human HCC cells lines for the expression of both LPAR6 mRNA and protein, and found that HuH7 expressed the highest levels of LPAR6 (Fig. 2A and B). Expression levels of others LPAR (LPAR1-5) are also shown (Supplementary Fig. S1B). Next, we generated stable knocked down LPAR6 in HuH7 using lentiviral delivery of short hairpin RNAs and selected two different cell populations, namely LPAR6-shRNA#8 and LPAR6-shRNA#9, in which LPAR6 was significantly knocked down (Fig. 2C and D). We then performed five-day proliferation assays and found that LPAR6 knockdown significantly decreased cell growth (P < 0.05) in LPAR6-shRNA#8 and LPAR6-shRNA#9 cells as compared with control-shRNA and the parental cell line Huh7 (Fig. 2E). This result was corroborated by a reduced percentage of Ki-67–positive cells (Fig. 2F). Consistently, both LPAR6-shRNA cell populations showed a delayed cell-cycle progression, with a higher percentage of cells in phase G₀–G1 as compared with controls (Supplementary Fig. S2A–S2C), but with no effect on apoptosis (Supplementary Fig. S3A and S3B). To further characterize the LPAR6-knockdown phenotype, we performed Matrigel migration assays that demonstrated a significantly decreased migration of LPAR6-shRNA cells (P < 0.001) compared with controls (Supplementary Fig. S4A). We also noticed differences in cell morphology on Matrigel (i.e., a rounded morphology, loss of filopodia and lamellipodia) between LPAR6-shRNA cells and controls (Supplementary Fig. S4B). In addition, the distribution of cortical actin was dramatically decreased (P < 0.001) in both LPAR6-shRNA populations as shown in Supplementary Fig. S4C and quantified in Supplementary Fig. S4D. Finally, no difference in the number of cells adherent to Matrigel was observed (Supplementary Fig. S4E and S4F). Overall, these data indicate that LPAR6 gene knockdown impairs the proliferation ability and tumorigenic phenotype of HCC cells.

To assess the contribution of the LPAR6 to tumor growth and tumorigenicity in vivo, we developed a human HCC xenograft model. Four groups of 12 4-week-old female CD-1 nude (nu/nu) athymic mice were formed, and each group was injected with LPAR6-shRNA#8 Huh7 or LPAR6-shRNA#9 Huh7, control-shRNA Huh7, or the parental cell line Huh7. After 8 weeks, mice inoculated with LPAR6-shRNA#8 (n = 12) and LPAR6-shRNA#9 (n = 12) displayed no significant tumor formation, or very few, small tumors (2 of 24), whereas mice inoculated with control cells all developed (100%) sizeable tumors (mean volume ±SD 1,897.13 ± 10.31 mm³; Fig. 3A and B). To validate the persistence of gene silencing, after sacrifice we checked the expression of LPAR6 in the HCC xenografts and found that LPAR6 expression was still undetectable in LPAR6-shRNA tumors (Fig. 3C). In addition, the Ki-67 proliferative index was markedly reduced in these tumors (P < 0.05) compared with controls (Fig. 3D and E), indicating that LPAR6 knockdown reduces tumor growth in vivo.

To mirror the tumorigenicity loss induced by LPAR6 knockdown, we next stably overexpressed LPAR6 in HLE cells (hereafter named HLE-LPAR6), an HCC cell line that does not express this receptor (Fig. 2A and B). As control, we generated HLE transfectants expressing the empty vector harboring the neomycin resistance gene (HLE-neo). Expression of LPAR6 in HLE-LPAR6 and HLE-neo was confirmed by real-time PCR and immunoblotting (Supplementary Fig. S5A and S5B). We also generated HLE expressing two different mutated forms of the LPAR6 gene, mutant c.463C>T, and mutant c.373_374delAA, encoding for the truncated nonfunctional protein p.X and p.Lys125AsnfsX37, respectively (Fig. 4A). Proliferation assays showed that LPAR6 overexpression efficiently increased proliferation in HLE-LPAR6 (P < 0.01) compared with HLE-neo. By contrast, increased proliferation was not observed in HLE harboring the truncated mutants Gln155X and HLE-Lys125AsnfsX37, indicating that LPAR6 is indeed required for proliferation (Fig. 4B). Consistently, the number of Ki-67–positive cells was significantly higher in HLE-LPAR6 (P = 0.0005) compared with control and LPAR6 truncated mutants (Fig. 4C). We also observed that HLE-LPAR6 cells migrated and invaded more efficiently compared with HLE-neo, whereas the mutants HLE-Gln155X and HLE-Lys125AsnfsX37 failed to increase migration and invasion (Fig. 4D). Finally, HLE-LPAR6 formed a significantly larger number and size of colonies in soft agar cells compared with HLE-neo and with HLE-Gln155X and HLE-Lys125AsnfsX37 truncated mutants (Fig. 4E). To further validate the effects of LPAR6 on tumorigenesis, we performed in vivo experiments. HLE cells are normally nontumorigenic in mice. To see whether the ectopic expression of LPAR6 confers the ability to form tumors in vivo, we carried out experiments in nude mice adopting two different injection methods of HLE-LPAR6 cells, namely subcutaneous and intrasplenic injection. After subcutaneous injection of HLE-LPAR6, few mice (4 of 20) showed a tumor burden and this was not statistically significant (data not shown). By contrast, after intrasplenic injection, although no macroscopic tumors were visible in the liver, we detected the presence of liver neoplastic microfoci in mice injected with HLE-LPAR6, whereas no neoplastic cells foci were detected in the HLE-neo control group (Fig. 4F, left). This prompted us to further investigate the capacity of HLE-LPAR6 to survive and generate hepatic tumor microfoci. By real-time PCR using human-specific GAPDH primers, we found that HLE-LPAR6 cells were present and surviving in the liver 68 days after injection, whereas no tumors were detected in control mice (Fig. 4F, right). Overall, these data indicate that nontumorigenic HCC cells expressing LPAR6 acquire tumorigenic properties and generate tumor microfoci or can survive in the liver in a dormant state for prolonged periods of time.

To identify candidate genes under the control of LPAR6, we analyzed transcriptome modifications in LPAR6-silenced HuH7 cells. Gene expression levels were measured by sequencing and counting a single tag originating from the 3′ end of mRNAs, a region with a high density of sequence polymorphisms. Using unique molecule identifiers (TrueQuant technique, GenXPro-GmbH), PCR copies were eliminated from the dataset to reduce false-positive SNPs and biased expression values. When compared with regular RNA-seq, the focus on this 3′-untranslated region increases the coverage by 10- to 20-fold and allows a reliable identification of SNPs at comparably lower sequencing costs (Supplementary Fig. S6A). The full list of up- and downregulated genes is reported in Supplementary Tables S1 and S2, respectively. The gene-related function and the network between the most significant up- and downregulated genes in LPAR6-silenced HuH7 cells, analyzed by the STRING database, are summarized in Supplementary Fig. S6B and S6C. Next, the 500 most significant up- and downregulated genes were grouped according to their functions (analysis for comparative cellular processes) using the publicly available software ToppCluster multiple gene analyzer. This analysis revealed that these genes fall into specific categories of function, including cytoskeleton, extracellular matrix (ECM) components, and tissue development (Supplementary Table S3). Because genes regulating tissue development are relevant to carcinogenesis, we focused on this category of genes and in particular on the proto-oncogene Pim-3 as a potential mediator of LPAR6-mediated tumorigenesis (Supplementary Fig. S6D and Supplementary Table S3).

To corroborate the molecular cross-talk between LPAR6 and Pim-3, we measured the mRNA expression levels of Pim-3 in LPAR6–knocked down Huh7 cells and found that Pim-3 was significantly downregulated (P < 0.001) in these cells compared with control-shRNA (Fig. 5A). Similarly, the expression levels of both Pim-3 mRNA and protein were significantly decreased (P < 0.01) in tumors originated from nude mice inoculated with LPAR6-shRNA HuH7 cells compared with controls (Fig. 5B). Consistently, Pim-3 mRNA levels were upregulated in HLE-LPAR6 cells compared with HLE-neo control cells (P < 0.001) and in mice xenotransplanted with HLE-LPAR6 compared with those (control group) transplanted with HLE-neo (Fig. 5C). These findings strongly suggest that Pim-3 is an important contributor to LPAR6-mediated tumor growth.

To validate the direct contribution of Pim-3 to LPAR6-mediated cell growth, we applied our gain- and loss-of-function model. Firstly, we transfected LPAR6-shRNA Huh7 cells with a construct expressing PIM3-GFP and found that ectopic expression of PIM3 completely restored growth in these cells. Then, we treated the same cells with 5 μmol/L of SGI-1776, a pharmacologic PIM3 inhibitor, and observed that PIM3 failed to restore cell growth (Fig. 5D). The effect of SGI-1776 alone on tumor growth in vivo (subcutaneous and intrasplenic HCC xenografts) and in vitro (LPAR6-expressing HLE) is shown in Supplementary Fig. S7A–S7C. Lastly, we found that inhibition of Pim-3 by both siRNAs and SGI-1776 decreased cell proliferation in HLE-LPAR6 and PLC/PRF/5-LPAR6, two different cell lines overexpressing LPAR6 (Fig. 5E and F). These data confirm the contribution of PIM3 to LPAR6-induced cell proliferation. Finally, we determined whether a correlation between expression of LPAR6 and Pim-3 was detectable in our cohort of patients with HCC, whose clinical and pathologic characteristics are reported in Fig. 1E. We stratified and scored the intensity of LPAR6 staining as high- (n = 37) or low-LPAR6 (n = 33) expression. Representative immunofluorescence images are shown in Fig. 6A. We found elevated PIM3 tissue expression levels in patients with high levels of LPAR6 (P < 0.01; Fig. 6B) and a significant correlation between LPAR6 and PIM3 in these patients (P < 0.0001; R = 0.55; Fig. 6C). We also sought to determine the correlation between LPAR6 expression and the proliferative marker Ki-67 by measuring stained areas of tissues. We detected elevated Ki-67 levels in high-LPAR6 tissues and low levels of Ki-67 in low-LPAR6 tissues (P < 0.001; Fig. 6D and E). A Pearson scatterplot showing the significant correlation (P = 0.0002; R = 0.54) between Ki-67 and LPAR6 is shown in Fig. 6F. This positive correlation between LPAR6 and Pim-3 expression and with the proliferative marker Ki-67 in patients with HCC is consistent with our findings showing a contribution of PIM3 to LPAR6-induced cell proliferation.

To test the hypothesis that LPAR6 might be required for the transcriptional activation of proto-oncogene Pim-3, we investigated the effects of LPAR6, ectopically expressed in HLE cells, on the 5′-flanking region containing the promoter of the human Pim-3 gene (Fig. 7A). To do so, we transiently transfected HLE-LPAR6 cells with a construct (pGL4.10) containing the luciferase reporter gene (luc) upstream of the entire promoter sequence of Pim-3 (p2074-luc). This promoter sequence contains the binding sites of several transcription factors, including STAT3, SP1, ETS1, and nuclear factor NF-κB. The results of the assay revealed that luciferase activity was present in HLE-LPAR6, whereas no activity was detected in HLE-neo control cells (Fig. 7B). Luciferase activity was also detectable in HLE-LPAR6 cells transfected with the deleted mutant p-809-luc (deletion between the position −2074 bp and −809 bp), which conserves the same binding sites as the p2074-luc vector (Fig. 7C). Conversely, decreased or no luciferase activity was observed in HLE-LPAR6 cells transfected with p264-luc (lacking the binding sites for STAT3) or p104-luc (no binding sites) vectors, respectively (Fig. 7B). Taken together, these data indicate that LPAR6 acts on the Pim-3 gene promoter. Moreover, the use of deleted promoter sequences revealed that STAT3 is an important binding site for the activation of the proto-oncogene Pim-3 promoter. To corroborate this observation, we evaluated the luciferase activity of HLE-LPAR6 expressing the Pim-3 promoter vector or the deleted mutants following STAT3 gene silencing by STAT3 siRNAs or treatment with STA-21 (30 μmol/L), a STAT3 selective inhibitor that inhibits the STAT3 binding to DNA (19). We found that both STAT3 siRNAs and STA-21 failed to increase luciferase activity in HLE-LPAR6 expressing the STAT3-binding sites (p2074-luc and p809-luc) compared with controls (HLE-LPAR6 control-siRNA and HLE-LPAR6 + vehicle; Fig. 7D). Figure 7E shows Western blot analysis demonstrating effective knockdown of STAT3 in HLE-LPAR6 expressing the pGL4 vectors harboring the entire Pim-3 promoter fragment (p2074-luc) or the deleted mutants (p809-luc, p264-luc, and p104-luc). Together, these data demonstrate that the STAT3-DNA binding site on the Pim-3 promoter is indeed relevant for the activation of Pim-3 by LPAR6.

Our current study reveals novel findings concerning the pathogenic role of LPAR6 in liver cancer and shows that LPAR6 functions as a promoter of tumorigenicity of HCC via the transcriptional activation of several genes, including Pim-3. This highlights the role of the LPA axis in HCC. We have previously demonstrated that LPA promotes HCC progression by stimulating the recruitment and transdifferentiation of peritumoral fibroblasts into carcinoma-associated fibroblasts (17). Here, we used a gain- and loss-of-function approach to elucidate how LPAR6 sustains the HCC tumorigenic process. In addition, we employed deletion mutants to corroborate the involvement of LPAR6 in this process. We also performed an extensive gene profiling analysis by Massive Analysis of cDNA Ends (MACE) to gain insight into genes under the control of LPAR6. Lastly, we show that data obtained in patients on the involvement of LPAR6 in HCC progression, as well as on the existing cross-talk between LPAR6 and Pim-3, are consistent with the experimental data. A descriptive study by Sokolov and colleagues (20) recently reported increased expression levels of LPAR6 in HCC. Our findings have extended these observations, disclosing a disease mechanism and attributing to LPAR6 a potential role as a biomarker predicting clinical outcome in patients with HCC. This is corroborated by the findings that LPAR6 levels are correlated with overall survival in patients with HCC, and that patients with high levels of LPAR6 have a poor prognosis. In our patient analysis, LPAR6 score seems to correlate with tumor size, so poor outcome could be expected. However, because LPAR6 sustains tumor growth and tumorigenicity giving the cancer cell a proliferative advantage over its neighbors, it is likely that larger tumors contain more LPAR6-expressing cells. In addition, we show the importance of LPAR6 expression in HCC regardless of circulating LPA levels. Blood LPA levels may vary in patients with HCC, although they are generally elevated compared with cirrhotic and healthy donors (17). This supports the idea that both elevated circulating LPA concentrations and LPAR6 tumor levels are important contributors to human disease.

Of similar importance is our finding demonstrating the role of LPAR6 in maintaining the tumorigenic potential of cancer cells. A possible explanation for this striking effect could be that LPAR6 orchestrates the function of multiple genes, as shown by our MACE experiments. Aberrant LPAR6 expression dysregulates genes involved in fundamental biologic processes, including tissue development, lipid metabolism and cytoskeleton, and ECM organization, and this may explain the tumorigenic activity in cancer cells. When LPAR6 is ectopically expressed, the growth potential is significantly increased, in an anchorage-independent manner, this being a typical hallmark of tumorigenicity. Our results suggest that the C-terminal domains of LPAR6 protein are involved in exerting this effect, because truncated LPAR6 mutants failed to promote tumor growth in our model. Interestingly, these mutants are responsible for lack of hair follicle growth and the generation of a genetic form of alopecia called hypotrichosis simplex (15). It is likely that a molecular mechanism involved in hair growth could be dysregulated during cancer growth and that the C-terminal portion of LPAR6 could be functionally implicated.

Digital expression profiling shows that genes with an altered expression as a consequence of LPAR6 knockdown are genes known to participate in the tumorigenic process. Among these, connective tissue growth factor (CTGF), FGFR1, ITGA2, BRCA2, and PIM3 were significantly downregulated. The involvement of CTGF in cancer and in particular in HCC is documented (21–23). Our gene profiling data also reveal alterations in genes controlling lipid metabolism and homeostasis, suggesting the existence of a molecular cross-talk between LPA and other lipid pathways. Moreover, genes involved in ECM organization such as MMP19, ITGA6, and several collagen genes were significantly downregulated. It is noteworthy that the ITGA6 gene encoding integrin α6 plays a role in the tumorigenic process of many cancers including HCC (24, 25). Among genes that control tissue development, we focused on Pim-3 because, being an oncogene, it could largely explain the LPAR6-mediated protumorigenic phenotype observed in HCC. Indeed, we have demonstrated that LPAR6 acts on the promoter and finely regulates the expression of Pim-3. This is consistent with the fact that Pim-3 is downregulated in LPAR6-shRNA cells and upregulated in LPAR6-overexpressing cells, and with the observation that Pim-3 expression is increased in patients with HCC overexpressing LPAR6. Whether Pim-3 is a predictor of clinical outcome and survival of patient with HCC was not an issue addressed in our study. Nevertheless, Zheng and colleagues (26) have suggested that Pim-3 may be useful to predict the prognosis of patients with gastric cancer. In terms of gene regulation, our findings clearly demonstrate that the LPAR6 signaling axis activates the transcription of Pim-3 by acting on the gene promoter that contains the binding sites for several transcription factors, including STAT3, SP1, ETS1, and nuclear factor NF-κB. Although the role of LPA in inducing NF-κB activation has been described (27), no report has yet shown the activation of STAT3 by LPA. Our results suggest that the STAT3-binding site is important for the activation of Pim-3 by LPAR6 because experiments using the deleted mutant p264, lacking the binding sites for STAT3, show a significantly decreased luciferase activity.

In conclusion, we provide evidence that LPAR6 is essential to maintain the tumorigenic properties of cancer cells, and that this occurs through the regulation of multiple genes including proto-oncogene Pim-3. Both experimental evidence and patients data reported in this study support the role of LPAR6 as a driver of tumorigenicity and growth in HCC. Future studies will evaluate the genetic and epigenetic factors that act on the LPAR6 signaling axis during the tumorigenic process, and whether these factors affect the LPAR6 gene in HCC precancerous conditions such as liver cirrhosis and dysplastic nodules. Moreover, it will be important to determine whether changes in LPAR6 expression or downstream signaling effectors are critical in different types of cancers other than HCC, and if they are predictive of patient survival. Lastly, this study provides a preclinical rationale for therapeutic targeting of LPAR6-Pim-3 in HCC.

No potential conflicts of interest were disclosed.

Conception and design: A. Mazzocca

Development of methodology: A. Mazzocca, F. De Santis, N. Mukaida

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Mazzocca, R.C. Betz

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Mazzocca, A. Filannino, N. Mukaida

Writing, review, and/or revision of the manuscript: A. Mazzocca

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A. Mazzocca, Y.-Y. Li

Study supervision: A. Mazzocca

Other (performed experiments): F. Dituri, A. Filannino, C. Lopane

Other (helped with discussion and comments on the project): C. Tortorella, C. Sabbà

The authors thank Dr. Letizia Porcelli for her technical contribution with flow cytometry and Mary V. Pragnell, B.A., for language revision. They also thank Sandra M. Pasternack for establishing LPAR6 constructs.

R.C. Betz is a recipient of a Heisenberg Professorship of the German Research Foundation (DFG; BE 2346/4-1). This work was supported by grants from University of Bari, Italian Ministry of Health “5 per mille” - 2011 grant from IRCCS “S. de Bellis,” National Institute for Digestive Diseases (A. Mazzocca), and Italian Association of Cancer Research grant N.11389 (G. Giannelli).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.