Identification of Heparin-Binding EGF-Like Growth Factor (HB-EGF) as a Biomarker for Lysophosphatidic Acid Receptor Type 1 (LPA1) Activation in Human Breast and Prostate Cancers

Lysophosphatidic acid (LPA) is a natural bioactive lipid with growth factor-like functions due to activation of a series of six G protein-coupled receptors (LPA1–6). LPA receptor type 1 (LPA1) signaling influences the pathophysiology of many diseases including cancer, obesity, rheumatoid arthritis, as well as lung, liver and kidney fibrosis. Therefore, LPA1 is an attractive therapeutic target. However, most mammalian cells co-express multiple LPA receptors whose co-activation impairs the validation of target inhibition in patients because of missing LPA receptor-specific biomarkers. LPA1 is known to induce IL-6 and IL-8 secretion, as also do LPA2 and LPA3. In this work, we first determined the LPA induced early-gene expression profile in three unrelated human cancer cell lines expressing different patterns of LPA receptors (PC3: LPA1,2,3,6; MDA-MB-231: LPA1,2; MCF-7: LPA2,6). Among the set of genes upregulated by LPA only in LPA1-expressing cells, we validated by QPCR and ELISA that upregulation of heparin-binding EGF-like growth factor (HB-EGF) was inhibited by LPA1–3 antagonists (Ki16425, Debio0719). Upregulation and downregulation of HB-EGF mRNA was confirmed in vitro in human MDA-B02 breast cancer cells stably overexpressing LPA1 (MDA-B02/LPA1) and downregulated for LPA1 (MDA-B02/shLPA1), respectively. At a clinical level, we quantified the expression of LPA1 and HB-EGF by QPCR in primary tumors of a cohort of 234 breast cancer patients and found a significantly higher expression of HB-EGF in breast tumors expressing high levels of LPA1. We also generated human xenograph prostate tumors in mice injected with PC3 cells and found that a five-day treatment with Ki16425 significantly decreased both HB-EGF mRNA expression at the primary tumor site and circulating human HB-EGF concentrations in serum. All together our results demonstrate that HB-EGF is a new and relevant biomarker with potentially high value in quantifying LPA1 activation state in patients receiving anti-LPA1 therapies.

LPA 1 is the most ubiquitous of all LPA receptors in organs and tissues both in human and mouse [13]. Lpar1 2/2 mice revealed that LPA 1 signaling influences the pathological process of many diseases. These animals are partially protected against bleomycininduced lung fibrosis [14] and Lpar1 2/2 mice immunized with type II collagen do not develop arthritis [15]. Additionally, Lpar1 2/2 mice are resistant to neuropathic pain induced by partial nerve ligation, and both allodynia and hyperalgesia induced in by intrathecal injection of LPA is totally abolished in these animals [16]. LPA 1 is highly involved in cancer development. LPA 1 has pro-oncogenic and pro-metastatic activities [17] and recent developments of anti-LPA 1 pharmacological drugs showed that targeting this receptor is a therapeutic option for metastasis suppression [18][19][20].
Therefore, LPA 1 is an attractive target in multiple clinical situations. However, current developments of LPA receptor inhibitors are based on the use of LPA receptor-null cells, such as the rat hepathoma RH7777 cells transfected to express recombinant LPA receptor specific subtypes, and measuring the capacity of the drugs inhibiting LPA-induced Ca 2+ release in the cytoplasm and GTPcS binding to the plasma membrane [21,22]. However, no specific LPA receptor biomarker has been defined so far, which impairs validation of specificity and efficacy of these drugs in more complex systems such as in cells expressing multiple subtypes of LPA receptors, as well as in vivo both in animals and humans. LPA 1 was shown to induce the secretion of IL-6 and IL-8 in ovarian and breast cancer cells [23,24]. However, LPA 2 and LPA 3 also induce the secretions of these cytokines [23,24]. Renal cells from Lpar1 2/2 mice showed impaired expression of CTGF indicating a close relationship between LPA 1 activity and CTGF expression [25]. However, TGF-b also controls CTGF expression [26]. Thus, the specificity of CTGF expression through LPA 1 activation requires further demonstration.
Here, we identified a set of genes specifically upregulated through the activation of LPA 1 based on transcriptomic analyses of nongenetically manipulated tumor cells selected on the bases of their different cancer of origin and on their distinct expression panel of LPA receptors. Among those genes, we demonstrated in vitro and in vivo that heparin-binding EGF-like growth factor (HB-EGF) is a new specific biomarker for LPA 1 activity in human breast and prostate cancers. Our findings revealed that HB-EGF is a potential new biomarker that will be useful to monitor the LPA 1 activation state in patients receiving anti-LPA 1 therapies.

Ethic statement
The mice used in our study were handled according to the rules of Décret Nu 87-848 du 19/10/1987, Paris. The experimental protocol was reviewed and approved by the Institutional Animal Care and Use Committee of the Université Claude Bernard Lyon-1 (Lyon, France). Studies were routinely inspected by the attending veterinarian to ensure continued compliance with the proposed protocols. Male BALB/C nude mice, 4 weeks of age, were housed under barrier conditions in laminar flow isolated hoods. Autoclaved water and mouse chow were provided ad libitum. Animals bearing tumor xenografts were carefully monitored for established signs of distress and discomfort and were humanely euthanized when these were confirmed. Studies involving human primary breast tumors were performed according to the principles embodied in the Declaration of Helsinki. Tissue biopsies were obtained as part of surgical treatments for the hormone receptor content determination. Remaining samples were included anonymously in this study. All human experiments were approved by the Experimental Review Board from the Laennec School of Medicine that waived the need for consent.

Drugs and reagents
Lysophosphatidic acid (LPA, Oleoyl C18:1) was obtained from Avanti Polar Lipids. The competitive inhibitors of LPA signaling pathways dependent on LPA 1 and LPA 3 receptors, Ki16425 was obtained from Cayman and Debio0719 was obtained from Debiopharm SA.

Reverse transcription and polymerase chain reaction (RT-PCR)
Total RNA from cells and mouse xenograph tumors were extracted using Nucleospin RNAII kit (Macherey-Nagel) and cDNA were synthesized using iScript cDNA Synthesis kit (Biorad).

Affymetrix analysis
Large-scale mRNA expression profiling was done on MDA-MB-231, MCF-7 and PC3 cells untreated and treated with LPA (1 mM) for 45 min. Total RNAs were purified as described above. Labeled cRNA probes from two independent replicates were hybridized to GeneChip Human Genome U1033 plus 2.0 (Affymetrix). Data were analyzed using Affymetyrix expression console (v.1.1) software by ProfilExpert (Lyon, France). Mean fluorescent signals of duplicated probe sets were calculated. All probe sets with more than 1.3-fold increase in presence of LPA defined LPA-dependent upregulated genes. The data discussed in this publication have been deposited in NCBI's Gene Expression Omnibus [29] and are accessible through GEO Series accession number GSE56265 (http://www.ncbi.nlm.nih.gov/geo/query/ acc.cgi?acc = GSE56265).

Microarray correlation analysis
Publicly available gene expression data for Prostate Tumor (GSE2109); Lung Tumor (GSE43580) and Colon Tumor (GSE21510) were obtained. Log2 tranformed values for HB-EGF and LPA1 expression were extracted from these database using R2 genomics analysis and visualization platform. Scatter plots were constructed. Analysis of correlation and computation of linear regression of the data were performed using Prism v5.0b (GraphPad Software, Inc.). P values ,0.05 were considered significant.

Patients and tumor characteristics
Patients were selected according to the following criteria: primary breast tumor without inflammatory feature, no previous treatment. Patients tumors were provide by three medical centers (Centre Hospitalier Régional Annecy, Chirurgie Oncologique Centre Hospitalier Universitaire Lyon-Sud, and Clinique Mutualiste Saint Etienne, France) in which patients were included between October 1994 and October 2001. Breast cancer tissue biopsies were obtained by surgery, selected by the pathologist and immediately stored in liquid nitrogen until processing. The biopsies were pulverized using a Mikro-Dismembrator (B. Braun Biotech International, Melsungen, Germany) and total RNAs were extracted using TRI Reagent (Sigma). To remove any genomic DNA contamination, total RNAs were treated with RNAse-free DNAse I and purified using RNeasy microcolumns (Qiagen). Quality of RNAs was verified using an Agilent Bioanalyser 2100 (Agilent).

Animal studies
Tumor xenograph experiments were performed using PC3 cells. Cells were suspended at a density of 10 6 cells in 100 ml of PBS and inoculated subcutaneously into the flank of male BALB/C nude mice at 4 weeks of age (Charles River). Tumor size was assessed by external measurement of the length (L) and width (W) of the tumor using a Vernier caliper. Tumor volume (TV; expressed in mm 3 ) was calculated using the following equation: TV = (LxW 2 )/2. Thirty-five days post-tumor cell injection animals were treated daily for 5 days with subcutaneous injection of Ki16425 (25 mg/ kg). Then animals were sacrificed. At that time the serum of each animal was collected for human HB-EGF quantification by ELISA (RayBiotech Inc) and PC3 xenograph tumors were submitted to total RNA extraction as described above.

Statistical analysis
Data were analyzed with the GraphPad Prism v5.0c software. Differences between groups were determined by one-way ANOVA followed by Bonferroni post-test and single comparisons were carried out using two-sided unpaired t-Test. Clinical multiple group comparisons were performed using Kruskall-Wallis followed Dunn's post-test. P,0.05 was considered significant.

Results
Determination of LPA 1 -specific early genes upregulated by LPA LPA 1 is the most ubiquitous LPA receptor in mammalian tissues. As a consequence, its expression in cells and tissues is frequently associated with other LPA receptors. Therefore, to identify LPA 1 activated genes independently of cell backgrounds we defined a strategy based on the use of nongenetically modified cell lines expressing different patterns of LPA receptors. In addition, to avoid activation of cell type-specific genes we selected the cell lines from two different human cancers of origin, breast (MDA-MB-231 and MCF-7) and prostate (PC3). For breast cancer cells we used two different types of cells, an estrogen positive cell line (MCF-7) and a triple negative cell line (MDA-MB-231). Hama and colleagues in 2004 showed that PC3 cells express LPA 1 , LPA 2 and LPA 3 , MDA-MB-231 cells express at least LPA 1 and LPA 2 , and MCF-7 cells do not express LPA 1 nor LPA 3 but LPA 2 [30]. These cells were subjected to total RNA extractions that were labeled and cRNA were probed on Affimetrix GeneChip Human Genome U1033 plus 2.0. In order to define the LPA receptor expression profiles in each cell line, we first extracted the fluorescent values obtained for each Affimetrix probe set corresponding to LPA receptor genes ( Figure 1A). Our results confirmed that the cell lines used in our study expressed differently LPA 1 , LPA 2 and LPA 3 receptors ( Figure 1B). Moreover, we could determine the complete LPA receptor patterns in PC3 cells (LPA 1 , LPA 2 , LPA 3 and LPA 6  . Those genes were good candidates for additional analyses because of the potential use of their products as biomarkers detectable in biological fluids. We have described previously in the context of bone metastasis induced by breast cancer cells that LPA through LPA 1 controls the expression of IL6, IL-8, CSF2(GM-CSF), and CXCL1(Groa) [18]. Here, we selected DTR coding for HB-EGF for further investigation because links between LPA 1 and HB-EGF expression have not been described so far. We found that purified LPA (1 mM) was as potent as FBS (10%) in upregulating HB-EGF mRNA expression in PC3 cells after 45 min of stimulation ( Figure 2A). The effect of FBS was totally blocked when cells were incubated in the presence of the LPA 1-3 antagonist, Debio0719 [19]. This result was also observed in PC3 cells treated with another LPA 1-3 antagonist, Ki16425 [21] ( Figure 2B). Moreover, Ki16425 completely abolished FBS-induced expression of HB-EGF in MDA-MB-231 cells ( Figure 2B). These cells do not express LPA 3 (Figure 1A), supporting a prevalent role of LPA 1 in HB-EGF expression. To confirm the functional implication of LPA 1 in HB-EGF expression, we used three subclones of human MDA-B02 cells (MDA-B02/GFP-bGal, MDA-B02/LPA1, MDA-B02/ shLPA1) that we generated in previous studies [18,27,28,31]. We first validated by RT-QPCR the expression levels of LPA 1 in these cells. As expected, MDA-B02/GFP-b Gal cells express LPA 1 that is stably overexpressed in MDA-B02/LPA1 cells and stably downregulated in MDA-B02/shLPA1 cells ( Figure 2C). RT-QCR analyses on these cells cultured in the presence of FBS (10%) showed that expression of HB-EGF was significantly higher in MDA-B02/LPA1 cells and lower in MDA-B02/shLPA1 cells than in MDA-B02/GFP-bGal cells, respectively, confirming the direct control of HB-EGF expression through LPA 1 activation ( Figure 2D). We then tested the capacity of LPA to induce the expression of HB-EGF after a more extended time of stimulation. PC3 cells were treated for 24 h with LPA. Cells were harvested for RNA extraction and mRNA quantifications, and culture media were collected for secreted protein analysis. We observed that LPA treatment upregulated the expression of HB-EGF mRNA ( Figure 2E) and the secretion of the protein ( Figure 2F) by PC3 cells. Both LPA-induced upregulation of HB-EGF mRNA and protein secretion were totally inhibited in the presence of Ki16425 (Figures 2E, 2F).

Expression of HB-EGF correlates with LPA 1 expression in human primary tumors of breast, prostate, lung and colon cancers
We then asked if the activity of LPA through LPA 1 inducing HB-EGF expression observed in our breast cancer cell lines could be also observed in clinical human samples. We analyzed the expression LPA 1 and HB-EGF by real-time quantitative PCR in a series of 234 primary tumor biopsies from patients with breast cancer. We then distributed tumors form patients into four quartiles according to the expression values of LPA 1 (Q1, Q2, Q3, Q4) and reported the mean relative expression value of HB-EGF in each group. We observed that the LPA 1 -Q4 group, which has the highest level of LPA 1 expression, has a significantly higher expression of HB-EGF than groups with lower expression of LPA 1 (Q1 and Q2) ( Figure 3A). We also observed that there was a statistically significant positive correlation between LPA 1 and HB-EGF in these samples ( Figure 3B). To further confirm our findings, we then looked into publically available databases for Prostate Tumor, Lung Tumor, and Colon Tumor. We observed that in the prostate tumor database there was a moderate but highly statistically significant positive correlation between LPA 1 and HB-EGF ( Figure 3C). In addition, in lung and colon tumor databases we observed a low but still highly statistically significant   Figure 3D and 3E). Therefore, this result supports the hypothesis for a positive correlation between LPA 1 expression and HB-EGF expression in vivo.

Pharmacological blockade of LPA 1 in vivo inhibits HB-EGF secretion by human PC3 xenographs
To confirm if LPA 1 activity may control HB-EGF expression in vivo, we generated xenograph tumors by subcutaneous injection of with LPA (1 mM) in absence or presence of Ki16425 (10 mM). (C) Expression of LPA 1 mRNA was measured by real-time quantitative PCR and normalized to housekeeping L32 gene in MDA-B02/GFP-bGal, MDA-B02/shLPA1 and MDA-B02/LPA1 cells cultured in the presence of 10% FBS. (F) Quantification of HB-EGF concentration in the conditioned culture media of PC3 cells treated for 24 h with LPA (1 mM) in absence or presence of Ki16425 (10 mM). All values were the mean6SD of at least three experiments. *p,0.05; **p,0.01; ***p,0.001 using one-way ANOVA with a Bonferroni post-test. doi:10.1371/journal.pone.0097771.g002 PC3 cells in the right flank of male BALB/C nude mice. After 35 days, post-tumor cell injection animals were treated subcutaneously with Ki16425 (25 mg/kg) for 5 d. Animals were then sacrificed. At that time, we collected the tumors for RNA preparation and the serum of animals for protein quantification. First, we found that a systemic treatment of animals for five days with the LPA 1 blocker did not alter tumor growth as judged by the absence of difference in the volume of PC3 tumors in animals treated with Ki16425 compared to vehicle-treated mice ( Figure 4A). This result was in agreement with previous reports of our laboratory and others using breast cancer mouse models [19,20]. We also found that Ki16425 treatment did not affect the expression levels of LPA receptors in these tumors ( Figure 4B). However, we found that the level of HB-EGF mRNA was significantly decreased by 28% in tumors of animals treated with Ki16425 compared to tumors of animals treated with the vehicle (p = 0.0231; 0.016960.0011 vs. 0.0232760.0028) ( Figure 4C). Moreover, we observed that the concentration of human HB-EGF was significantly decreased by 44% in the serum of animals treated with Ki16425 compared to that of animals treated with the vehicle (p = 0.0321; 25.3563.14 vs. 44.9366.87) ( Figure 4D). These results indicate that blocking LPA 1 activity in vivo inhibits the expression and secretion of HB-EGF in animals bearing PC3 xenograph tumors.

Discussion
Studies using genetically modified cells and mice have revealed the functional involvements of LPA 1 in multiple pathological processes. Overexpression of Lpar1 driven by the MMTV promoter in the mammary gland of transgenic mice induces the formation of spontaneous breast tumors within the first year [17]. Using immune compromised mice we showed that expression of LPA 1 confers a high propensity of inducing bone metastasis to human breast cancer cells [18]. In that context, treatments with the LPA 1-3 antagonist Ki16425 inhibited the progression of osteolytic bone metastases [18]. In addition, we recently described that the LPA 1-3 antagonist Debio0719 prevented spontaneous metastasis dissemination of mouse 4T1 mammary cancer cells to bone and lungs [19] with additional induction of tumor cell dormancy at secondary sites [20]. However, validating that LPA 1-3 antagonists were inhibiting LPA 1 -specific pro-oncogenic and pro-metastatic activities could not be achieved in these studies because of missing LPA 1 -specific biomarkers.
Our transcriptomic analysis confirmed previous findings showing that LPA induces the expression of the pro-osteoclastic cytokines IL6 and IL8 through LPA 1 in breast cancer cells [18,28]. The capacity of LPA 1 in inducing the secretion of these cytokines was further extended to ovarian cancer cells [23]. LPA 1 shares this activity with other LPA receptors as LPA 3 also mediates both IL6 and IL8 expressions and LPA 2 was identified to be the most efficient receptor in linking LPA to IL-6 and IL-8 production in ovarian cancer cells [24]. Therefore, IL6 and IL8 are unlikely relevant LPA 1 -specific biomarkers. Among other cytokines, LPA induces the expression of CYR61 and CTGF in epithelial cells [25,[32][33][34]. Analyses of Lpar1 2/2 animals or pharmacological blockade of LPA receptors with Ki16425 revealed a major role of LPA 1 in LPA-induced CYR61 and CTG expressions in bacterial infection and induction of tissue fibrosis in kidney and lungs [25,[32][33][34]. We observed that LPA induced the expression of both CYR61 and CTGF in prostate and breast cancer cells (data not shown). Expression of CYR61 and CTGF in MDA-MB-231 and MCF-7 cells were previously reported, corroborating that expression of these two cytokines are not restrained to LPA 1 -expressing cells [35]. CYR61 and CTGF genes are also upregulated by TGFb and commonly found in TGF-b gene signatures in different cellular models [26,36,37]. Recent reports showed that LPAdependent expression of CTGF requires transactivation of TGF-b receptor in myoblasts [38]. Thus, the specificity of CTGF expression through LPA 1 activation requires further demonstration. Based on cross-comparison analyses of whole genome transcriptomic profiles of PC3, MDA-MB-231 and MCF-7 cells in response to LPA, we determined a list of early genes upregulated through LPA 1 activation. In the perspective of establishing a new biomarker easily accessible for sample examination in clinic, we focused our study on the secreted growth factor HB-EGF. We found that HB-EGF expression was detected at an early time point following LPA stimulation (45 min). Its expression was sustained at least up to 24 h of stimulation in vitro and 5 d in vivo, suggesting the potential use of HB-EGF as a relevant marker in the perspective of mid-to long-term follow-up. Based on the quantification of HB-EGF in vitro, our results demonstrated that the treatment of MDA-MB-231 breast and PC3 prostate cancer cells with LPA 1-3 antagonists (Ki16425, De-bio0719) mimicked Lpar1 gene silencing in MDA-B02 breast cancer cells. Also, we found a significantly higher expression of HB-EGF mRNA in human primary breast tumors that highly expressed LPA 1 . Unfortunately, we were not able to confirm this correlation at the protein level because biological samples from patients of this cohort were not available. However, we were able to quantify by ELISA the concentration of human HB-EGF secreted by PC3 cells in the blood circulation of PC3 tumorbearing animals. Moreover, we validated that the pharmacological blockade of LPA 1 can be monitored in vivo through the inhibition of HB-EGF expression both at mRNA levels in the primary tumors and at protein levels in the blood circulation. LPA was previously shown inducing the shedding of proHB-EGF in VeroH cells through a Ras-Raf-MEK-dependent pathway [39]. LPA also induces the shedding of proHB-EGF from the membrane of in human bronchial epithelial cells involving activation of matrix metalloproteinases leading to down-stream stimulation of EGFR and the secretion of IL-8 [40]. However, these works did not characterize which LPA receptor transmitted LPA signals. Our results demonstrated at mRNA and protein levels that LPA induces HB-EGF expression through an LPA 1 -dependent mechanism. Our study was based on manipulating breast and prostate cancer cells and on analyzing publically available tumor databases of breast, prostate, lung and colon cancers. Therefore, the significance of HB-EGF as a biomarker of LPA 1 activity in other pathologies, such as tissue fibrosis [14,25], obesity [41], rheumatoid arthritis [15], osteoporosis [5], and neuropathic pain [42], remains to be determined.
Disease-free and overall survivals are frequent primary endpoints validating drug efficacies in clinical trials. However, using intermediate endpoints and biomarkers are required during the course of therapies to validate drugs having reached their dedicated targets. Our findings revealed that HB-EGF is a new biomarker downstream stimulation of LPA 1 that would be extremely valuable to quantify the LPA 1 activation state in patients receiving anti-LPA 1 therapies.