Expression and function of lysophosphatidic acid receptors (LPARs) 1 and 3 in human hepatic cancer progenitor cells

Hepatocellular carcinoma (HCC) is the most common primary cancer of the liver and is characterized by rapid tumor expansion and metastasis. Lysophosphatidic acid (LPA) signaling, via LPA receptors 1–6 (LPARs1–6), regulates diverse cell functions including motility, migration, and proliferation, yet the role of LPARs in hepatic tumor pathology is poorly understood. We sought to determine the expression and function of endothelial differentiation gene (EDG) LPARs (LPAR1–3) in human HCC and complimentary in vitro models. Human HCC were characterized by significantly elevated LPAR1/LPAR3 expression in the microenvironment between the tumor and non-tumor liver (NTL), a finding mirrored in human SKHep1 cells. Analysis of human tissue and human hepatic tumor cells in vitro revealed cells that express LPAR3 (HCC-NTL margin in vivo and SKHep1 in vitro) also express cancer stem cell markers in the absence of hepatocyte markers. Treatment of SKHep1 cells with exogenous LPA led to significantly increased cell motility but not proliferation. Using pharmacological agents and cells transfected to knock-down LPAR1 or LPAR3 demonstrated LPA-dependent cell migration occurs via an LPAR3-Gi-ERK-pathway independent of LPAR1. These data suggest cells that stain positive for both LPAR3 and cancer stem cell markers are distinct from the tumor mass per se, and may mediate tumor invasiveness/expansion via LPA-LPAR3 signaling.


INTRODUCTION
Hepatocellular carcinoma (HCC) is the most common primary tumor of the liver and accounts for more than 750,000 deaths/yr. In most instances HCC arises following exposure to known risk factors, the most common being viral hepatitis infection (HBV/HCV), prolonged, heavy alcohol use, aflatoxin ingestion and obesity [1]. Mainstay therapies for HCC rely on surgical intervention (resection, ablation, and/or transplant) with alternatives being limited in scope and efficacy [1,2]. Outcomes for HCC patients are often compromised by advanced tumor stage, impaired liver function as a result of underlying cirrhosis, and the presence of extra-and/ or intra-hepatic metastasis. As a result, the prognosis for HCC patients remains bleak, survival times of 7-12 months following diagnosis being typical [1,2].
Lysophosphatidic acid (LPA) is a small phospholipid molecule comprising of a phosphate group, glycerol backbone, and single fatty acid chain. Biologically, LPA exists in many structural variants based on degree of saturation, position of the C = C (when present), and number of carbons in the fatty acid chain [3,4]. Synthesis of LPA occurs from membrane phospholipids, primarily via lysophospholipase D (autotaxin) and lysophospholipase A1β [3,5,6]. Following synthesis LPA regulates diverse cell functions across a range of cell types including proliferation, survival, and migration [3]. To do so LPA acts as an extracellular agonist binding to G-protein-coupled LPA receptors (LPARs) of which 6 have been characterized to date (LPARs1-6) [3,7,8]. Each receptor differs in cell/tissue distribution, agonistbinding profile, and downstream intracellular signaling pathway(s) regulated following activation. Based on Oncotarget 2952 www.impactjournals.com/oncotarget structural and phylogenetic homology LPARs can be divided into two major sub-groups: the endothelial differentiation gene (EDG) sub-family (LPARs 1-3), and the non-EDG sub-family (LPARs 4-6) [7].
Given LPA's capability to regulate diverse basic cell functions, it is unsurprising that LPA signaling is also exploited by malignant cells and is altered in many cancers. This aberrant regulation is evident at various levels including escalation in LPA synthesis, changes in circulating LPA profile, and altered LPAR expression profiles [9][10][11], and occurs in various cancers including ovarian [12], breast [13], colon [14], and pancreatic tumors [15,16]. Unlike other organs the role of LPAR signaling in normal liver function has proven more ambiguous due to the [relative] lack of previously well-characterized LPARs (LPARs 1-5) in healthy liver/hepatocytes [4,[17][18][19]. Analysis of serum samples report elevated LPA levels in HCC patients [10,20] and animal models of liver disease [21]. Circulating LPA, and changes in LPA isoform composition, are also indicated as potential markers of HCV patient progression to HCC [21], and as early markers of HCC development [9,10]. Within cirrhotic patients, LPA signaling is linked with hepatic stellate cell activation [22,23] and tumor-derived LPA has been reported to be central to peritumoral fibroblast recruitment and transdifferentiation into myofibroblasts and accelerated tumor progression [20].
Studies by our group and others now report LPAR6, the most recently characterized LPAR subtype [24,25], is expressed in normal liver/hepatocytes, and is significantly elevated in human HCC [26,27] and regenerating rodent liver [28]. During the course of these studies we reported LPAR1 and LPAR3 expression was increased in a subset of human HCC and cirrhotic non-tumor liver (NTL) compared to liver from non-tumor burdened patients [27]. In the current study we further analyzed EDG-LPAR (LPARs1-3) expression and localization in human HCC specimens. These studies allowed us to determine that changes in LPAR1/LPAR3 expression in HCC tissue were confined to a subset of cells located at the HCC-NTL margin. Further analysis of these LPAR1/LPAR3 positive cells revealed they also express progenitor/ stem cell markers in the absence of hepatocyte markers. By screening established human hepatic tumor cells we determined the SKHep1 cell line exhibited a similar profile to the subset of cells that stain positive for both LPAR3 and cancer stem cell markers located at the HCC-NTL margin. Using SKHep1 cells in vitro we were able to conclude LPA stimulates cell migration in the SKHep1 cell line via an LPAR3-Gi-protein-MEK-ERK dependent mechanism, independent of Rho or PI3K-Akt signaling, both of which are present and activated following LPA stimulation of SKHep1 cells. Collectively these data provide detailed mechanistic evidence for a role for LPA-LPAR3 dependent signaling in a unique subset of cancer stem cells located at the tumor-NTL margin in HCC patients.

LPAR1 and LPAR3 expression is significantly increased in human HCC samples and localizes to the tumor margin
Immunohistochemical (IHC) analysis was performed on archived human HCC samples from patients with varying underlying etiologies (Supplementary Table 1) using antibodies specific against LPAR1 or LPAR3 ( Figure 1A and 1B, Supplementary Figure S1). Using this approach we demonstrated significantly increased LPAR1 expression in HCC vs. NTL ( Figure 1C, IHC score 0.58 ± 0.08 vs. 0.21 ± 0.04; HCC vs. NTL; *p < 0.05). Overall, LPAR1 expression was increased in 71% of patients (15/21) and was most apparent at the NTL-HCC margin ( Figure 1A). Analysis of LPAR3 also demonstrated significantly increased expression in HCC vs. NTL ( Figure 1C, IHC score 1.13 ± 0.12 vs. 0.28 ± 0.05, HCC vs. NTL, *p < 0.001). Of note, increased LPAR3 in HCC was more pronounced than that observed for LPAR1 and occurred in 89% of patients (17/19), the most significant expression again being localized to the HCC-NTL margin ( Figure 1B).

SKHep1 cells express LPAR1 and LPAR3
LPAR1-6 mRNA expression was examined in 3 human tumor cell lines of hepatic origin (SKHep1, HepG2, and HuH7). These data demonstrate the LPAR mRNA profile of SKHep1 cells resembled the in vivo LPAR profile detected in cells localized to the HCC-NTL margin; SKHep1 cells expressing LPAR1 and LPAR3 mRNA ( Figure 2A). Conversely, LPAR6 mRNA was readily detectable in HepG2 and HuH7 cells but not SKHep1 cells, and mRNAs for LPAR S2, 4 and 5 were barely detectable, if at all, in any of the cell lines ( Figure 2A). LPAR protein was measured by Western blot and LPAR1 and LPAR3 were detected in SKHep1 cells but not in normal liver from non-HCC patients (NL) ( Figure 2B).

SKHep1 cells and LPAR3 positive cells in the HCC-NTL margin express stem cell but not hepatocyte markers
The SKHep1 cell line is a human tumor cell line that lacks hepatic parenchymal cell (hepatocyte) markers that was originally derived from the ascites fluid of a patient presenting with an adenocarcinoma [29]. Recently, Eun et al. reported extensive expression of mesenchymal and cancer stem cell markers in SKHep1 cells [30]. These findings led us to examine the expression of Hep par 1 (an hepatocyte marker [31]), CD44, Cytokeratin-19 (CK-19), epithelial cell adhesion molecule (EpCAM), and vimentin (markers of mesenchymal stem cells, hepatic tumor/ www.impactjournals.com/oncotarget progenitor cells, and cells of mesenchymal origin [32][33][34][35]). By Western blot analyses we demonstrate NL readily expresses Hep par 1 in the absence of CD44, CK-19, EpCAM or vimentin ( Figure 2C). Analysis of human cell lines revealed only SKHep1 expresses CD44 and vimentin ( Figure 2C), while EpCAM was most readily detectable in HepG2 and HuH7 cells and CK19 was expressed in SKHep1 and HepG2 cells ( Figure 2C). In light of these Oncotarget 2954 www.impactjournals.com/oncotarget in vitro data sections from human HCC-NTL margin were then analyzed by immunofluorescent histochemistry (IFHC) using antibodies against LPAR3 and EpCAM, CD44, or Hep par1. These data demonstrated extensive co-localization of LPAR3 with CD44 and EpCAM in the HCC-NTL margin ( Figure 3) in the absence of LPAR3-Hep par1 co-localization ( Figure 3).

Exogenous LPA stimulates SKHep1 cell motility
We next examined the effect of exogenous LPA on SKHep1 cell function. Analysis of cell proliferation demonstrated LPA did not significantly affect SKHep1 growth in serum-depleted medium (SDM; 0.1% FBS) at LPA levels ≤ 50 µM. At LPA levels ≥ 100 µM cell (B) LPAR1 and LPAR3 protein expression in lysates prepared from SKHep1 cells (SK), normal liver from non-HCC burdened patients (NL). Positive control (+ve) for LPAR1 was rat brain lysate and PC3 human prostate cancer cell lysate for LPAR3. Membranes were stained with amido black as an indicator of total protein loading. (C) Representative Western blots of lysates prepared from normal liver (NL), SKHep1 (SK), HepG2 (H2) and HuH7 (H7) cells probed with antibodies against Hep par1, CD44, CK19, EpCAM or vimentin (Vim). Membranes were stripped and probed with GAPDH as a house-keeping protein/loading control.
We next assessed the effect of exogenous LPA on SKHep1 motility in vitro. Using a 2D migration assay we demonstrated LPA significantly increased SKHep1 migration toward the unseeded area of the plate ( Figure 4B, *p < 0.05 LPA vs. Control). Concurrently SKHep1 cells were pre-treated with the LPAR1-3 inhibitor Ki16425 (Ki; 10 µM) [36] followed by LPA. Using this approach Ki abolished the effect of LPA on cell migration ( Figure 4B, *p < 0.05 LPA vs. Ki + LPA). Parallel experiments were performed to determine 3D cell migration in responses to LPA (10 µM) using a Transwell insert assay. Using this approach LPA significantly increased SKHep1 cells actively migrating through the membrane ( Figure

LPAR3 mediates Gi-ERK-dependent migration in response to LPA in SKHep1 cells
SKHep1 cell lines with down regulated LPAR1 or LPAR3 expression were created by stable transfection with shRNA constructs. From these (a minimum of) two clones were selected in which target LPAR mRNA was reduced by ≥ 70% (Figure 7A and 7B, *p < 0.05 vs. untransfected (C) or cells transfected with scrambled sequence (Scr)). Clones were expanded and decreased target protein expression confirmed by Western blot (Figure 7A and 7B). Inhibition of target LPARs did not significantly alter expression of other LPAR isoforms (data not shown).
Inhibition of LPAR1 failed to significantly alter the effect of LPA on either AKT or ERK 1/2 -MAPK activity ( Figure 8A

DISCUSSION
In this study we report the microenvironment in human hepatocellular carcinoma (HCC) and non-tumor liver (NTL) is characterized by a subset of cells expressing LPAR1 and LPAR3, a profile shared by the human hepatic tumor-derived SKHep1 cell line. Stimulating SKHep1 cells with LPA in vitro led to G-protein-dependent activation of Rho (via G 12/13 -protein signaling) and Akt/ ERK-MAPK (via Gi-protein signaling) resulting in cell migration in the absence of changes in proliferation.
Using an shRNA approach we demonstrate LPA-mediated migration in SKHep1 cells is dependent on LPA-LPAR3 signaling via a Gi-ERK-MAPK signaling cascade. Conversely, while LPA-LPAR1 signaling led to activation of a Rho-dependent signaling cascade, inhibiting this pathway failed to alter LPA-dependent SKHep1 migration. Finally, re-examining human HCC revealed cells staining positive for LPAR3 in the HCC-NTL margin also stained positive for cancer stem cell (but not hepatocyte) markers, features shared by SKHep1 cells.
In the healthy liver LPAR1/LPAR3 are expressed at relatively low levels compared with other tissue types [4,[17][18][19]. Indeed, the relative absence of wellcharacterized LPAR subtypes (LPARs1-5) within the liver presents somewhat of a paradox in understanding how LPA affects liver function and hepatic pathogenesis. For example, in animal models, correlations between circulating LPA and degree of underlying liver injury are reported [41] yet LPARs 1-5 are barely detectable (if at all) in rodent hepatocytes/liver tissue [28]. Similarly, cirrhotic and HCC patients exhibit elevated serum LPA levels, while the relative expression of LPA species in patients at risk for developing HCC suggest serum LPA profiles may be beneficial in identifying those patients with HCC [10]. The characterization of the purinergic p2ry5 receptor as a novel non-EDG LPAR, LPAR6, and reports of LPAR6 expression in human and rodent liver, and human HCC tissue may help explain these findings. However, unlike NTL, HCC tissue, and other human hepatic tumor cell lines, we failed to detect significant LPAR6 expression in SKHep1 cells (relative to LPAR 1 and 3 mRNA). Additionally, we failed to detect the hepatocyte marker, Hep par1, in cells that stained for LPAR1/LPAR3 in human HCC tissue. These data suggest SKHep1 cells may not have originated as a result of hepatocyte transformation, and that LPAR6-dependent signaling is not a significant factor in mediating LPAdependent activity in these cells.
Studies in other organ systems report LPA-LPAR1-3 signaling is involved in tissue fibrosis [42,43]. Hepatic cirrhosis is the most common precursor for progression to HCC, and changes in LPA-LPAR1-3 signaling during hepatic stellate cell (HSC) activation is reported [44]. In addition to a role for LPA-LPAR1-3 signaling during hepatic fibroge nesis, studies by Mazzocca et al., report tumor-derived LPA increases peritumoral fibroblast recruitment, and subsequent transdifferentiation to myofibroblasts [20]. Myofibroblasts, in turn, play a crucial role in mediating the crosstalk between transformed hepatocytes and non-parenchymal, stromal liver cells [20,45]. These findings suggest tumor-derived LPA can act as a paracrine signaling factor within the tumor microenvironment to promote tumor expansion by stimulating the accumulation and/or activity of myofibroblasts within the tumor [20]. We also demonstrated (the majority of) LPAR1/LPAR3 detected by IHC occurred at the NTL-HCC margin. Because we failed to detect significant LPAR1/LPAR3 expression within the larger tumor mass, or more distant NTL, and cells that stained for LPAR1/LPAR3 did not stain for the hepatocyte marker Hep par1, we hypothesized LPAR1/ LPAR3 expression was confined to a non-HCC, nonhepatocyte cell population. While it remains possible this cell population is derived from HSCs (HCCs often being www.impactjournals.com/oncotarget characterized by fibrous encapsulation [1]), the relative absence of staining for LPAR1/LPAR3 in NTL, even in cirrhotic patients, and the histologic appearance of LPAR1/ LPAR3 positive cells led us to examine alternatives.
Several authors postulate the significance of hepatic stem cells during hepatocyte repopulation, hepatocyte transformation, and tumor progression [46][47][48]. Using antibodies against CD44 and EpCAM1 (markers of mesenchymal stem cells and liver cancer progenitor cells) we identified cells expressing LPAR3 in the HCC-NTL margin also expressed these stem cell markers. A review of the literature reveals SKHep1 cells, while widely reported and employed as human HCC cells of mesenchymal origin, do not express hepatic genes or exhibit endogenous  Western blots demonstrating total Akt/pAkt in cell lysates prepared from SKHep1 cells stably transfected with an off-target scrambled sequence (Scr) or an shRNA against LPAR1 (shLPAR1 clones 11 and 8 (Cl-11/Cl-8)) or LPAR3 (shLPAR3, clones 8 and 9 (Cl-8/Cl-9) in the absence (-) or presence (+) of LPA (10 µM). (B) Representative Western blots demonstrating total ERK 1/2 /pERK 1/2 in cell lysates prepared from SKHep1 cells stably transfected with an off-target scrambled sequence (Scr) or an shRNA against LPAR1 (shLPAR1 clones 11 and 8 (Cl-11/Cl-8)) or LPAR3 (shLPAR3, clones 8 and 9 (Cl-8/Cl-9) in the absence (-) or presence (+) of LPA (10 µM). (C) Rho activity in SKHep1 clones stably transfected with an off-target scrambled sequence (Scr) or an shRNA against LPAR1 (shLPAR1) or LPAR3 (shLPAR3) in the absence (-) or presence (+) of LPA (10 µM). n = 3 independent experiments performed in duplicate, *p < 0.05 LPA vs. pair-matched non-LPA-treated, *p < 0.05 shLPAR1 + LPA vs. Scr + LPA and shLPAR3 + LPA. www.impactjournals.com/oncotarget liver cell (hepatocyte) functions [29], factors further supported by our failure to detect LPAR6 [27,28]. Unlike many immortalized HCC cell lines, the SKHep1 line was not derived from a surgically resected liver tumor. Rather, the SKHep1 line was derived from ascites fluid of a patient with a hepatic adenocarcinoma, indicating a cell line with existing metastatic properties [29,30,49]. This evidence is further supported in a report by Eun et al., in which a detailed analysis of SKHep1 cells (using 22 different markers/antibodies) reports SKHep1 are characterized by classical mesenchymal stem cell markers in the absence of endothelial and hematopoietic markers. Furthermore, SKHep1 cells are able to differentiate to adipocytes and osteoblasts and form widespread, multi-organ site metastatic masses using an in vivo/SCID mouse model. Collectively Eun et al., conclude SKHep1 cells represent a transformation of mesenchymal stem cells into cancer stem cells [30].
Our data demonstrate that, in addition to a similar LPAR1/LPAR3 profile, the stem cell markers used on human tissue were also present in SKHep1 cells in vitro. These findings led us to address the effect and mechanisms of action by which LPA affects SKHep1 cells as a means to investigate the potential function of cells staining positive for LPAR3 in the HCC-NTL margin. All members of the LPAR family are G-protein coupled receptors and others report LPAR1 and 3 transduce signals via Gαi, Gαq and Gα 12/13 (LPAR1) or Gαi and Gαq (LPAR3) subunits, the LPAR-Gα-protein interaction being dependent on factors that include cell type, agonist/ LPA isoform, and the relative expression/interaction of the different LPAR subtypes within the cell [3,4,50]. Our studies demonstrate that intracellular signaling in SKHep1 cells in response to LPA is dependent on the LPAR subtype activated. Activation of LPAR1 stimulates Rho activity but LPA-LPAR1-Rho signaling does not of 2D cell migration for SKHep1 clones stably transfected with an off-target scrambled sequence (Scr) or an shRNA against LPAR1 (shLPAR1; data shown are for clone 8) or LPAR3 (shLPAR3; data shown are for clone 8) in the absence (Control (C)) or presence (+) of LPA (10 µM). Data from ≥ 2 independent clones repeated in 8 replicates were collected and expressed as fold change vs. pair-matched non-LPA-treated control (C) *p < 0.05 C vs. LPA, *p < 0.05 shLPAR3 + LPA vs. Scr + LPA and shLPAR1 + LPA. (B) Representative images of 3D cell migration for SKHep1 clones stably transfected with an off-target scrambled sequence (Scr) or an shRNA against LPAR1 (shLPAR1; data shown are for clone 11) or LPAR3 (shLPAR3; data shown are for clone 9) in the absence (Control (C)) or presence (+) of LPA (10 µM). Data from ≥ 2 independent clones repeated in triplicate were collected and expressed as change vs. pair-matched non-LPAtreated control (C) or Scrambled sequence non-LPA-treated control. *p < 0.05 C vs. LPA, *p < 0.05 shLPAR3 + LPA vs. Scr + LPA and shLPAR1 + LPA.
These data are both supportive and contraindicative to previous studies using tumor cell lines in culture, including the SKHep1 cell line. In a study by Park et al., LPA augmented SKHep1 invasion via LPAR1 signaling and matrix metalloproteinase 9 (MMP9). This conclusion is derived by the analysis of LPARs 1-3 mRNA expression and in vitro studies that stepwise analyze intracellular signaling mechanisms. Using this approach the authors determine that SKHep1 invasion following exogenous LPA requires PKCδ/p38 MAPK and PI3K/Akt activation, data that are initially mirrored by our findings. However, in their studies the authors report only LPAR1 mRNA, and not LPAR3 mRNA, expression in SKHep1 cells [51]. In contrast our data report readily detectable LPAR3 mRNA and protein in SKHep1. Furthermore, inhibition of LPAR1 or PI3K did not significantly alter LPAmediated cell migration in our studies, whereas inhibition of LPAR3-Gi-protein-ERK-MAPK signaling abrogated LPA-dependent migration. Several reasons may explain such differences. At a fundamental level our group used different primers to measure LPAR3 mRNA expression. However, protein measurement by Western blot supported our LPAR3 mRNA expression data. Alternatively, an increasing body of literature addresses consistency within 'banked' cell lines, data that highlights the potential that many commonly employed cell lines exhibit differences due to prolonged propagation in culture [52]. Finally, there are inherent differences between the assays performed by Park et al., in which they measure invasion whereas as we measured migration [51].
While our data provides evidence that LPA stimulates cell movement in the SKHep1 cell line via an LPAR3-Gi-ERK MAPK pathway it also raises other intriguing questions. For example, we clearly demonstrate LPA stimulates Rho activity in SKHep1 cells and inhibition of LPAR1 expression indicates an LPA-LPAR1-Rho activation pathway. Studies using other cell lines, including cancer cells, report Rho is a significant factor in mediating a range of cell responses, including cell movement, and Rho is an important factor in determining the metastatic potential of cancer cells [53,54]. Furthermore, several studies report LPAR1 is an abundantly expressed LPAR in a variety of cancers, including pancreas [15], colon, lung and breast [13], and Rho activation is mediated via Giα 12/13 [55,56], a Gαprotein associated with LPAR1 [3,4,57]. Conversely, ovarian cancer appears to be characterized by decreased LPAR1 expression [12] and tumor aggressiveness is associated with altered LPAR 2 and LPAR3, LPAR3 being capable of signaling via Gi-proteins [7,58]. This appears to be mirrored in SKHep1 cells as inhibition of Gi-signaling (PTx) abolished the effects of LPA on migration. Interestingly, activation of Gi-proteins has the potential to exert dual effects via α-subunit (inhibition of adenylyl cyclase-cAMP-protein kinase A) and βγ-dimer (stimulation of MEK-ERK MAPK) activity [59]. Our data suggest LPA-LPAR3 dependent cell migration in SKHep1 cells is most likely to occur via a G-βγ-dimer dependent mechanism.
In conclusion, our data provide further evidence to support the notion that SKHep1 cells are not human HCC cells but rather, are a human cancer stem cell line. Furthermore, LPA stimulates migration (but not proliferation) in the SKHep1 cell line via LPAR3-Giprotein-ERK-MAPK signaling. These data are particularly intriguing as a similar LPAR profile to that of SKHep1 cells exists in a sub-population of cells at the HCC-NTL margin, a cell population that also stains positive for stem cell, but not hepatocyte, markers. These data, along with previously published studies, suggest an important role for LPA-LPAR signaling within the tumor microenvironment whereby LPA is likely to act via either LPA-LPAR6 signaling in HCC cells, or via LPA-LPAR1/LPAR3 signaling in hepatic myofibroblasts and hepatic cancer progenitor cells. Given that healthy, non-parenchymal cells in the liver express LPAR1 and/or LPAR3 at low levels, this raises the possibility of developing selective therapies against LPAR1 and/or LPAR3 to slow or regress HCC progression.

Institutional assurances and patient demographics
The Institutional Review Board (IRB) at Carolinas Medical Center granted approval to analyze surgical specimens of patients undergoing HCC resection. We obtained informed, written consent from patients for participation in the study and for collecting and storing their information for research purposes. A board certified pathologist confirmed HCC diagnosis and underlying pathology and patient demographics were recorded from 23 patients undergoing hepatic resection or transplant for HCC (n = 18 male: age 61.0 ± 11.4y, range 43

mRNA and protein expression
Relative mRNA expression was determined by quantitative real-time PCR (qRT-PCR) as previously reported [27] (primer sequences and reaction conditions are described in Supplementary Table 2B). Relative protein expression was determined by Western blot using whole cell lysates [27].

Cell proliferation
Following cell plating (12-well plates; ≈ 90,000 cells/well) and attachment (18 Hrs), culture medium was replaced with SDM (0.1% (v/v) charcoal-stripped FBS (Gemini Bio-Products, Sacramento, CA) for 24 h followed by LPA treatment (0-100 μM, 24 h). At the end of this period cells were detached and cell proliferation measured by cell count-viability (trypan blue exclusion) analysis (Countess Automated Cell Counter, Life Technologies, Grand Island, NY) [60]. Data are means from a minimum of 4-wells per treatment group from ≥ 3 independent experiments.

Cell migration
Two dimensional cell migration experiments were conducted using an Oris Cell Migration Assay (Platypus Technologies, Madison, WI) . Briefly, SKHep1 cells were fluorescently stained (CellTracker Green, Life Technologies) and plated at 30,000 cells/well into 96-well plates equipped with rubber plugs in the absence or presence of inhibitors. After 24 hrs the plugs were removed, LPA added (10 µM final concentration) and cells incubated overnight. Migration was measured as relative fluorescence within the original plug "footprint". Alternatively, cells were rinsed, fixed, stained with crystal violet, and images captured and analyzed for cell occupancy within the original unseeded plug "footprint" using the NIH ImageJ software. Data are means from a minimum of 4 independent experiments performed in 8-well replicates per treatment group.
Three-dimensional migration was measured using 8 µm pore Transwells (BD Biosciences, San Jose, CA) as previously reported. Briefly, cells were collected after 24 h in SDM (0.1% (v/v) charcoal-stripped FBS) and added to the plastic inserts of the Transwell chamber in the absence or presence of inhibitors and 0.7 ml SDM, with or without LPA (10 µM), was added to the lower chamber. Following overnight incubation the inserts were removed, rinsed in phosphate buffered saline (PBS), and cells wiped from the internal surface. Cells on the outer surface were fixed in methanol, stained with crystal violet, rinsed in distilled water, and 5 random fields counted/experimental condition. Data are from a minimum of three independent experiments/group.

Intracellular signaling activity
Cells were placed in SDM for 24 h then treated with LPA (10 µM) or vehicle. Cells were harvested, lysed and resultant total/active protein expression detected by Western blot [27].

Rho activity
Rho activation was measured by Western blot analysis following pull-down of active Rho with an Active-Rho pull-down/detection kit as per the manufacturer's instructions (Thermo-Fisher Scientific, Waltham, MA) and compared to total Rho. Rho activity was also measured using a G-LISA RhoA Activation Kit and a Total RhoA ELISA as per the manufacturer's instructions (Cytoskeleton Inc.).