Lysophosphatidic Acid (LPA) in Malignant Ascites Stimulates Motility of Human Pancreatic Cancer Cells through LPA 1 *

Cytokines and growth factors in malignant ascites are thought to modulate a variety of cellular activities of cancer cells and normal host cells. The motility of cancer cells is an especially important activity for invasion and metastasis. Here, we examined the components in ascites, which are responsible for cell motility, from patients and

Pancreatic cancer is a highly metastatic cancer characterized by widespread intraperitoneal dissemination and ascites for-mation, which frequently occur even after curative resection and constitute the major cause of death in pancreatic cancer patients (1,2). Therefore, the suppression of dissemination is an important issue in the treatment of pancreatic cancer. Peritoneal dissemination is thought to be composed of several processes, including cell adhesion, migration, invasion, and proliferation (3,4). In the ascites and pleural effusions of patients, a variety of cytokines and growth factors are present; these cytokines were produced and secreted from cancer cells, as well as from normal host cells, and have been suggested to affect these processes of peritoneal dissemination (5)(6)(7).
Lysophosphatidic acid (LPA) 1 has been shown to participate in diverse biological actions, including a change in cell shape, motility, and proliferation in a variety of cell types in association with the stimulation of early signaling events, such as Ca 2ϩ mobilization, change in cAMP accumulation, and activation of several protein kinases (8 -11). Extracellular LPA has also been shown to be involved in certain diseases, such as atherosclerosis (12,13) and cancer (11, 14 -18). In fact, LPA has been identified as a growth-promoting factor that promotes the proliferation of ovarian cancer cells in malignant ascites from ovarian cancer patients (14 -16). LPA has also been reported to be present in malignant effusions (19). Most of these LPA actions are mediated through the lipid-specific EDG family G protein-coupled receptors, i.e. LPA 1 /EDG-2, LPA 2 /EDG-4, and LPA 3 /EDG-7 (8 -10, 20), although recent studies suggested that LPA actions are potentially mediated through LPA 4 / GPR23, another type of G protein-coupled receptor for LPA (21) and peroxisome proliferator-activated receptor ␥, a transcriptional factor identified as an intracellular LPA receptor (22). Although there are many reports suggesting the potential role of LPA in cell growth, migration, and invasion of several cancer cells (14,18,(23)(24)(25)(26)(27)(28)(29)(30)(31), it was difficult to determine the extent to which LPA, among a variety of potent cytokines and growth factors present in ascites, serves as a mediator of these responses. LPA receptor antagonists, LPA-neutralizing antibod-ies, or the LPA-specific degrading enzyme would be very useful for this purpose. Especially, receptor-selective antagonists are strong tools for identifying the subtype of LPA receptors that are responsible for the lipid action and would be a platform to develop therapeutic agents.
We have recently developed a novel LPA receptor antagonist, Ki16425, which shows a preference for LPA 1 and LPA 3 over LPA 2 (32). Ki16425 showed an extremely high specificity to LPA and its receptors. Thus, Ki16425 may be a useful tool for evaluating the role of LPA in biological samples, such as ascites. In the present paper, we showed that a considerably high amount of LPA is present in ascites from pancreatic cancer patients and that the formation of LPA-rich ascites can be duplicated by an intraperitoneal injection of human pancreatic cancer cells in nude mice. Furthermore, our data indicated that LPA in malignant ascites is an important component for the motility of pancreatic cancer cells through Ki16425-sensitive LPA receptors, especially LPA 1 .

EXPERIMENTAL PROCEDURES
Materials-1-Oleoyl-sn-glycero-3-phosphate (LPA), L-␣-lysophosphatidylcholine palmitoyl (LPC, C16:0), and sphingosine 1-phosphate (S1P) were purchased from Cayman Chemical Co. (Ann Arbor, MI); fatty acid-free BSA was from Calbiochem-Novabiochem Co. ( Mice-Female BALB/c nude mice (5 weeks old) were obtained from Charles River Japan, Inc. (Tokyo, Japan). Sterile food and water were fed to the mice ad libitum. The mice were maintained in sterile cages on sterile bedding and housed in rooms at a constant temperature and humidity. All experiments using mice were performed according to procedures approved by the Gunma University Animal Care Committee.
Ascites from Cancer Patients-Eight Japanese patients with pancreatic cancer were available for this study at the Second Department of Surgery, Gunma University Faculty of Medicine from 2001 through 2003. Table I summarized the characteristics of the patients. A cytological examination demonstrated that the ascites contains cancer cells. Plasma or ascites were collected in the presence of EDTA (at a final concentration of 2-3 mM) and centrifuged at 1,000 ϫ g for 20 min to remove cells. The cell-free fluid was stored at Ϫ80°C until use. Informed consent was obtained from each patient for the use of samples.
Establishment of a Highly Peritoneal Metastatic Pancreatic Cancer Cell Line, YAPC-PD-A human pancreatic cancer cell line, YAPC-PD, was established from YAPC cells, which had been previously established in our laboratory (33). The YAPC is not a highly peritoneal metastatic cell line, but, in one nude mouse of 30, which were injected in the peritoneal cavity with YAPC cells (1 ϫ 10 7 per mouse), peritoneal dissemination with bloody ascites was detected 12 weeks after the injection. Cells in the bloody ascites were harvested and cultured in RPMI 1640 containing 10% fetal bovine serum. Intermingled mouse fibroblasts gradually decreased in number and finally disappeared within a 4-week culture. The rest of the cells attached on dishes were harvested, and 1 ϫ 10 7 cells were then injected in the peritoneal cavity of nude mice. The same procedure was repeated in a fifth cycle, and we obtained the YAPC-PD cell line, which induces peritoneal dissemination with high frequency. More than 90% of nude mice injected intraperitoneally with this cell line developed peritoneal dissemination with bloody ascites within 4 to 6 weeks.
Ascites from Mice Injected with the Human Pancreatic Cancer Cell Line, YAPC-PD-Ten million cells of YAPC-PD were injected into the peritoneal cavity of nude mice. Four to 6 weeks later, mice that developed abdominal distension were killed, and bloody ascites was collected in the presence of EDTA (at a final concentration of 2-3 mM). The bloody ascites was rendered cell-free by centrifugation as described above.
Cell Culture-Human pancreatic cancer cell lines, PK-1, PK-9, and Panc-1, were kindly provided by the Cancer Cell Repository, Tohoku University (Sendai, Japan), and MIA PaCa-2, BxPC-3, CFPAC-1, and HPAC were purchased from the American Type Culture Collection (Rockville, MD). The YAPC cells were transfected with a pEFneo empty vector alone or a pEFneo vector containing human LPA 1 receptor (34) by electroporation, and the neomycin (G418 sulfate at 1 mg/ml)-resistant cells were selected. All pancreatic cancer cell lines and the receptortransfected cells were cultured in RPMI 1640 containing 10% fetal bovine serum. Twenty-four hours before the experiments, the medium was changed to a fresh medium (without serum) containing 0.1% (w/v) BSA (fraction V) unless otherwise specified. Where indicated, PTX (100 ng/ml) was added to the culture medium 24 h before the experiments. Transfection of siRNA-Pancreatic cancer cell lines, YAPC-PD and Panc-1, were plated on 12 multiwell plates at ϳ2.0 ϫ 10 5 cells/well. Sixteen h later, siRNAs (total 30 nM) were introduced into cells using an RNAiFect reagent (Qiagen K.K., Tokyo, Japan) according to the manufacturer's instructions and the cells were further cultured for 24 h. The LPA receptor mRNA level was measured using real-time TaqMan technology, and a cell migration assay was performed 24 h after serum starvation as described later. The following 21-mer oligonucleotide pairs were used as siRNAs against human LPA 1  Evaluation of LPA-like Activity-LPA in malignant ascites and plasma (0.5 ml, unless otherwise stated) was selectively extracted as alkaline-soluble lipids as described previously (35). By this procedure, major lipid components, such as phosphatidylcholine, sphingomyelin, and other neutral lipids, can be removed. To evaluate the content of LPA in this extract, a sensitive and specific bioassay based on the ability of LPA to inhibit cAMP accumulation in LPA 1 -expressing RH7777 cells was used because vector-transfected RH7777 cells do not respond to LPA (20). The LPA 1 -expressing RH7777 cells (a generous gift from Prof. Kevin R. Lynch, University of Virginia School of Medicine) were cultured in minimal essential medium containing 10% fetal bovine serum. Three days before the experiments, cells were seeded on 12-well plates that were coated with rat-tail collagen (400 g/ml). Twenty-four hours before the experiments, the medium was changed to fresh minimal essential medium containing 0.1% BSA. The cells were washed twice with a HEPES-buffered medium (36) and incubated with the extract or LPA (C18:1) standard in a HEPES-buffered medium containing 10 M forskolin and 0.5 mM isobutylmethylxanthine at a final concentration of 500 l. After a 10-min incubation, the reaction was terminated by adding 100 l of 1 N HCl. Cyclic AMP in the acid extracts was measured (36). The cAMP-inhibiting activity of LPA or test samples was completely lost by MG lipase, an enzyme hydrolyzing monoglycerides, such as LPA, or Ki16425, an LPA antagonist. Furthermore, the activity was unchanged even when the LPA-like activity was measured after further purification with silica gel high performance thin-layer chromatography (HPTLC) (Merck) using a solvent system consisting of BuOH/acetic acid/water (3:1:1). These results support the specificity of this bioassay. By this method, we detected LPA C18:1 (1-oleoyl-sn-glycero-3-phosphate) equivalent activity that was as low as 1 pmol/assay well. The evaluated LPA-like activity in the test sample was presented as an LPA C18:1 equivalent level.
MG Lipase Treatment-LPA (1 M), EGF (100 ng/ml), or ascites (10%) were treated with MG lipase at 10 units/ml for 30 min at 37°C in RPMI 1640 containing 0.1% BSA. In separate experiments, we added [ 3 H]LPA to the assay medium and confirmed that more than 98% of LPA in the test samples was degraded under these conditions. These MG lipase-treated samples were finally used by 10 times dilution with the assay medium.
Extraction of an Active Component from Ascites-The ascites were treated with 2 volumes of BuOH and separated into two phases. The BuOH-extracted components were further separated by HPTLC using a solvent system consisting of BuOH/acetic acid/water (3:1:1) (37). The silica gel with the resolved lipids (about 1-cm length each) was scraped off to obtain lipids covering the entire area of migration. The lipids were eluted and dried by evaporation. All fractions thus separated were dissolved in an assay medium containing 0.1% BSA and used at the final concentration corresponding to 1% ascites.
Cell Migration Assay-The migration experiment was performed using a Boyden chamber apparatus, as previously described (38). In brief, YAPC-PD cells and other cancer cells were harvested with 0.05% trypsin containing 0.02% EDTA, washed once, and resuspended with RPMI 1640 containing 0.1% BSA. The cells were loaded into the upper chamber, and test agents were placed in the lower chamber, unless otherwise specified. A membrane filter with 8-mm pores was precoated overnight at 4°C with 100 g/ml rat-tail collagen. When the effects of LPA antagonists were examined, the cells were preincubated for 10 min with antagonists before loading. The number of cells that had migrated for 4 h to the lower surface were determined by counting the cells in four places under microscopy at ϫ400 magnification. Unless otherwise stated, this Boyden chamber method was used for the evaluation of the migratory activity of the cells. In some experiments, Transwell chemotaxis chambers (6.5 mm diameter, 8 m pore size) (Costar, Inc.) were used. Chemotaxis filters were soaked in 100 g/ml rat-tail collagen overnight, similarly to the Boyden chamber method. The cells were first attached on the filters; unattached cells were then removed 1 h after seeding, and test agents were placed in the lower chamber. The number of the cells that migrated to the lower surface during a 4-h incubation was determined as described above.
Matrigel Invasion Assay-Cell invasion activity was assessed by using a Matrigel invasion chamber (BD Biosciences, San Jose, CA) according to the instructions provided by the manufacturer. The procedures are essentially the same as those for the migration assay using Transwell chemotaxis chambers, except that the incubation time was 24 h.
Cell Adhesion Assay-Cells were harvested by trypsin, seeded on 48-well plates at a density of 5 ϫ 10 4 cells per well in RPMI 1640 containing 0.1% BSA and test agents, and then incubated for 1 or 4 h. At each time, floating cells were aspirated, plates were rinsed with phosphate-buffered saline, and the attached cells were then evaluated by an 3-(4,5-dimethythiazol-2-yl)-diphenyltetrazolium bromide assay as described previously (33).
Cell Proliferation-Cells were seeded on 48-well plates at 2 ϫ 10 4 cells in 0.4 ml. Twenty-four hours before the experiments, the medium was changed to RPMI 1640 containing 0.1% BSA, and cells were exposed to the test agents for an additional 24 h. Cell proliferation was evaluated by an 3-(4,5-dimethythiazol-2-yl)-diphenyltetrazolium bromide assay (33).
Quantitative Reverse Transcriptase-PCR Using Real-time TaqMan Technology-Total RNA was isolated using Tri-Reagent (Sigma) according to the instructions from the manufacturer. After DNase I (Promega, Madison, WI) treatment to remove possible traces of genomic DNA contaminating the RNA preparations, 5 g of the total RNA was reverse transcribed using random priming and Multiscribe reverse transcriptase according to the instructions from the manufacturer (Applied Biosystems, Foster City, CA). To evaluate the expression levels of the LPA 1 , LPA 2 , LPA 3, and LPA 4 /GPR23 mRNAs, quantitative reverse transcriptase-PCR was performed using real-time TaqMan technology with a sequence detection system model 7700 (Applied Biosystems, Foster City, CA). The human LPA 1 -, LPA 2 -, LPA 3 -, LPA 4 /GPR23-, and glyceraldehyde-3-phosphate dehydrogenase-specific probes were obtained from Assay-on-Demand products (Applied Biosystems, Foster City, CA). The ID numbers of the products are Hs00173500 for LPA 1 , Hs00173704 for LPA 2 , Hs00173857 for LPA 3 , Hs00271072 for LPA 4 / GPR23, and Hs99999905 for glyceraldehyde-3-phosphate dehydrogenase. Amplification reaction was performed using the TaqMan universal PCR master mixture following the instructions from the manufacturer (Applied Biosystems). The thermal cycling conditions were as follows: 2 min at 50°C, 10 min at 95°C, 40 cycles of 15 s at 95°C, and 1 min at 60°C. The procedure for the calculation of their expression was essentially the same as that described previously (39). The expression level of the target mRNA was normalized to the relative ratio of the expression of glyceraldehyde-3-phosphate dehydrogenase mRNA. Each reverse transcriptase-PCR assay was performed at least three times, and the results are expressed as mean Ϯ S.E.
Northern Blot Analysis-Total RNA was prepared as described above. Twenty g of total RNA was electrophoresed in a 1% agarose gel containing 3.7% formaldehyde in a 20 mM MOPS buffer and blotted onto a nylon membrane (Hybond-N) with 20ϫ standard saline citrate (SSC). The cDNA probe of LPA 1 (20 ng) was labeled with [␣-32 P]dATP by random oligonucleotide priming and added to the blots at a concentration of about 5 ϫ 10 6 disintegrations/min in 5 ml of a hybridization buffer as described previously (34). The hybridization was carried out at 60°C. Following hybridization, the blots were washed at 60°C with 0.2ϫ SSC and 0.1% SDS as described previously (34).
Data Presentation-All experiments were performed in duplicate or triplicate. The results of multiple observations are presented as the mean Ϯ S.E. or as representative results from more than three different  (14) Ascites from mouse 624 Ϯ 213 (6) FIG. 1. LPA and ascites stimulate the migration of YAPC-PD cells but not their adhesion. A, the migration activity of YAPC-PD cells was examined using a Boyden chamber in the presence or absence of LPA (100 nM) or Case 2 ascites (1%) in the lower and upper chambers, as indicated. B, the migration activity of the cells was examined using a Transwell chamber in the presence or absence of EGF (10 ng/ml), LPA (100 nM), or ascites (1%) from two pancreatic cancer patients (Case 1 and Case 2) in the lower chamber. C, the adhesion of YAPC-PD cells to collagen-coated dishes was examined in the presence or absence of the indicated agents. Concentrations were the same as those for B.

FIG. 2. Inhibition by Ki16425 and PTX of the migration response to malignant ascites and LPA but not to EGF in YAPC-PD cells.
A, dose-dependent increase in the migration response to ascites and its inhibition by Ki16425. B, dose-dependent inhibition by Ki16425 of the migration response to ascites (1%). C, migration response to the increasing dose of ascites from a mouse (PD-1) that was injected with YAPC-PD cells and its inhibition by Ki16425. D, inhibition by Ki16425 of the migration response to ascites (1%) obtained from YAPC-PD cell-injected mice (PD-2-PD-6). E, dose-dependent inhibition by Ki16425 of the migration response to LPA but not to EGF (10 ng/ml). F, inhibition by PTX of the migration response to LPA (100 nM) and ascites (1%) but not to EGF (10 ng/ml). batches of cells unless otherwise stated. Statistical significance was assessed by the Student's t test; values were considered significant at p Ͻ 0.05.

Presence of a High Level of LPA in Malignant Ascites from Pancreatic Cancer Patients and Cancer Cell-injected Mice-
Because there was no previous information concerning the presence of LPA in ascites of patients with pancreatic cancer, we first measured the LPA content in ascites of pancreatic cancer patients. Many forms of LPA have been identified on the basis of differences in the species of fatty acid, the position of the fatty acid substituent (1-or 2-position of the glycerol backbone), and the attachment to the backbone (acy, alkyl, or alkenyl) (11). In the present study, we evaluated the amount of LPA on the basis of its ability to stimulate the LPA 1 receptor after a partial purification as alkaline-soluble lipids from other lipid and protein components. The LPA levels in ascites and plasma are summarized in Table I. There was no significant difference in the plasma LPA levels between cancer patients and normal healthy volunteers (Table I). In all the cases of ascites (8 patients), however, we detected a significantly high amount of LPA, and its level was always higher than that in the plasma of individuals.
The formation of peritoneal dissemination and ascites with a significant level of LPA was reproduced in mice by an intraperitoneal injection of YAPC-PD, a highly invasive pancreatic cancer cell line (Table I). The mean LPA equivalent level was 2728 nM in ascites from cancer patients and 624 nM in the malignant ascites from mice; these values are high enough to bind to and activate the known LPA receptor subtypes; the apparent dissociation constant of LPA receptors is around 100 nM (8 -10, 20, 21). These results raised the possibility that LPA in malignant ascites may participate in the regulation of the cellular activities of cancer cells as well as host normal cells.
LPA and Malignant Ascites Stimulate the Motility of YAPC-PD Cells-In Fig. 1, we examined whether LPA and ascites can affect the cellular activities of the pancreatic cancer cell line. When LPA or ascites were loaded into the lower chamber of the Boyden apparatus, YAPC-PD cells migrated toward the lower chamber. The migration activity was still effective even when the same concentration of LPA or ascites was present in both chambers, suggesting that LPA and ascites simulate random migration (chemokinetics) as well as directional migration (chemotaxis) of pancreatic cancer cells (Fig.  1A). We also used Transwell chambers, in which the cells were first attached on the filter and the migration activity in response to test agents was then determined. LPA, ascites from two patients, and EGF significantly stimulated the migration (Fig. 1B). In separate experiments, we observed no significant effect by these test agents on the adhesion of cells on the dishes (Fig. 1C). These results support the idea that LPA and ascites enhance the migratory activity of the cells but do not stimulate their adhesion activity.
LPA Is a Component Responsible for the Stimulation of Motility in Malignant Ascites-To assess the role of LPA in malignant ascites in cell migration activity, we examined the effects of Ki16425, a novel LPA receptor antagonist with high specificity to LPA (32). As shown in Fig. 2, Ki16425 at 1 to 10 M markedly inhibited the migration of YAPC-PD cells by the ascites from a pancreatic cancer patient (Fig. 2A). The inhibition by Ki16425 was also observed in a dose-dependent manner for malignant ascites from other patients (Fig. 2B). The migration of YAPC-PD cells was also induced by ascites obtained from mice injected with the same cells and the response to the ascites was again markedly inhibited by Ki16425 (Fig. 2, C and  D). Ki16425 was also an effective inhibitor of the migration response to LPA but not to EGF, supporting the specificity of Ki16425 (Fig. 2E). Because Ki16425-sensitive LPA receptors seem to be coupled to PTX-sensitive G-proteins (32), the effect of PTX on the migration of YAPC-PD cells was examined in Fig.  2F. As expected, PTX treatment almost completely inhibited the migration response to LPA and ascites but not to EGF. These results suggest that the ascites-induced migration response is mediated by Ki16425-sensitive LPA receptors that are coupled to PTX-sensitive G-proteins.
To further confirm the involvement of LPA in the induction of migration responses to malignant fluids, we performed several experiments (Fig. 3). The ascites-induced action was inhibited by a prior treatment of ascites with MG lipase, an enzyme hydrolyzing monoglycerides (Fig. 3A). Although it is uncertain whether LPA is a substrate for MG lipase under physiological conditions, we confirmed that this enzyme degraded more than 98% of LPA in the test samples. In fact, the LPA-induced migration was markedly inhibited by the enzyme treatment. These results suggest that LPA or related monoglycerides in ascites may participate in the induction of migration. In Fig. 3B, we extracted the active components from ascites that induce migration. About 70% of the active component(s) sensitive to MG lipase was extracted in the BuOH fraction. The recovery rate was almost identical to the recovery rate of [ 3 H]LPA in the same lipid extraction (data not shown). The BuOH fraction was further processed by HPTLC separation of the active components. The highest cell migration ac- tivity was detected in fraction 3, which is the LPA-richest fraction (Fig. 3B). Because S1P has a similar property to LPA with respect to its solubility to BuOH and migration on HPTLC, it is not easy to clearly separate LPA and S1P in biological samples. However, compared with LPA, S1P exerted only a small effect on the migration of YAPC-PD cells (Fig. 3C). In this figure, we also examined the effect of LPC, which may be present in ascites at a high level, but its effect on migration was marginal (Fig. 3C). These results support the idea that LPA is an important component for the migration response to ascites. Fig. 4 we examined the migration activity in pancreatic cancer cell lines other than YAPC-PD. Among the seven cell lines used, a significant migration response to EGF, LPA, and ascites was observed in PK-1, PK-9, Panc-1, BxPC-3, CFPAC-1, and HPAC. In MiaPaCa-2 cells, however, no significant effect was observed by these test agents (Fig. 4). In all pancreatic cell lines responsive to ascites, Ki16425 (Fig. 4) and MG lipase (data not shown) specifically inhibited the migration response to LPA and ascites, as was the case for YAPC-PD cells. Table II shows the expression of the mRNA of LPA receptor subtypes by real-time PCR. In all cell lines responsive to LPA and ascites, a significant amount of LPA 1 mRNA expression was detected, although there are variations in their expression among cell types. In MiaPaCa-2, an unresponsive cell line to LPA and ascites, however, LPA 1 mRNA expression was not detected by real-time PCR (Table II)  rable with the level of LPA 2 mRNA. Thus, in PK-1 and HPAC, cell lines that are highly responsive to LPA and ascites, only a marginal or no significant expression of LPA 3 mRNA was detected. As for LPA 4 /GPR23, its expression was marginal or not significant. Although the expression of receptor proteins was not successfully detected in the present study, LPA 1 expression seems to correlate with the migration activity by LPA and ascites in pancreatic cancer cell lines. Thus, there was a significant correlation between LPA-induced migration response in eight cell lines and their mRNA expression of LPA 1 but not LPA 2 or LPA 3 mRNA (Table II).

Involvement of the LPA 1 Receptor in the Migration of Pancreatic Cancer Cell Lines-In
The ability of LPA 1 to stimulate the migration activity was confirmed by overexpressing LPA 1 in YAPC, a parent cell line of YAPC-PD cells (Fig. 5). YAPC cells showed less expression of LPA 1 mRNA and less responsiveness to LPA than YAPC-PD cells. Transfection of LPA 1 in YAPC cells, however, resulted in an enhancement of the migration response to LPA to the level of YAPC-PD cells (Fig. 5).
To further examine the LPA receptor subtypes responsible for the migration response, we performed two lines of experiments. In the first line of experiments, siRNAs against LPA 1 receptors were transfected into YAPC-PD cells (Fig. 6A) and Panc-1 cells (Fig. 6C) to decrease the expression of the receptor mRNA. The siRNA transfection resulted in a marked reduction of the expression of LPA 1 mRNA without a significant change in the expression of LPA 2 and LPA 3 mRNAs, which was accompanied by a remarkable inhibition of migration response to LPA and ascites but not to EGF (Fig. 6, B and D). Thus, LPA 1 -specific siRNA inhibited LPA-and ascites-induced migration responses. These results are consistent with the idea that LPA 1 is a critical receptor for the migration response to LPA and ascites in pancreatic cancer cell lines, at least in YAPC-PD and Panc-1.
In the second line of experiments, we compared the effects of several LPA receptor antagonists. Ki16425 has preference for LPA 1 and LPA 3 over LPA 2 (32), whereas VPC12249 showed preference for LPA 1 and LPA 3 but not for LPA 2 (40), and DGPP 8:0, only for LPA 3 (41) (Table III). In the case of YAPC-PD cells, the migration response to LPA and ascites was significantly inhibited by VPC12249 as well as Ki16425, antagonists for LPA 1 and LPA 3 , but not by DGPP 8:0, an LPA 3 -specific antagonist. None of these LPA receptor antagonists affected the EGF-induced migration, suggesting the specificity of the antagonists. In other cell lines as well, the migration response to ascites was sensitive to both Ki16425 and VPC12249 but not to DGPP 8:0 (Table III). These pharmacological studies also support a critical role of LPA 1 in the migration response to ascites in pancreatic cancer cell lines.
Ascites and LPA Induce Matrigel Invasion in a Manner Sensitive to Ki16425-Proteolysis of extracellular matrix proteins is an important step for the invasion and metastasis of cancer cells. The possibility that LPA and ascites stimulate the invasion of pancreatic cancer cells was studied using a Matrigel invasion assay. YAPC-PD cells were loaded on Matrigel-coated pore filters for 24 h in the absence or presence of EGF, LPA, or ascites in the lower chamber, and the cells that had migrated to  The mRNA expression of LPA receptor subtypes was assessed by real-time PCR in the indicated pancreatic cancer cell lines. Results are expressed as the relative ratios to glyceraldehyde-3-phosphate dehydrogenase mRNA expression (ϫ1000). A ratio of less than 0.00001 (described as Ͻ0.01 in the table) is not significant by this method. A correlation coefficient between the expression of each LPA receptor subtype mRNA and the LPA-induced migration activity in eight cell lines was calculated as 0.761 (p Ͻ 0.05) for LPA 1 , 0.045 (not significant) for LPA 2 , and 0.647 (not significant) for LPA 3 , where the migration activities shown in Fig. 3C were used for YAPC-PD and those in Fig. 4 were for other cell lines. the lower surface of the filters were counted. As shown in Fig.  7, ascites and other agents stimulated the invasion, and the response to ascites and LPA, but not to EGF, was markedly inhibited by Ki16425, suggesting that LPA is also an important component for the invasive activity of ascites.

TABLE III
Effects of LPA antagonists on the migration response to EGF, LPA, and ascites The indicated cell lines were used for the evaluation of the ability of LPA antagonists to inhibit the migration response to EGF (10 ng/ml), LPA (100 nM), and malignant ascites (Case 2 or Case 3 at 1%) from patients with pancreatic cancer. The results are expressed as percentages of the control migration activity (increment from basal value) induced by the respective agent without any antagonist. LPA antagonists hardly affected the basal activity without test agents. The control activity (cell numbers per 4 macroscopic fields) was 57 Ϯ 2 for EGF, 87 Ϯ 12 for LPA, 85 Ϯ 13 for ascites from Case 2, and 94 Ϯ 14 for ascites from Case 3 in YAPC-PD cells; and these values for ascites in other cell lines were 100 -200, as described in Fig. 4 cells at 1-10 M and in Panc-1 cells at 0.1-1 M, but we observed only inhibitory effects by ascites at up to 10% (data not shown). In PK-9 cells, these malignant fluids significantly promoted their proliferation (Fig. 8). The cell proliferation, however, was insensitive to Ki16425 (Fig. 8A) and MG lipase (Fig. 8B); both compounds are potent inhibitors of the migration of pancreatic cancer cell lines, including PK-9. Furthermore, LPA had no stimulatory effect at 100 nM or less (Fig. 8), and the effect was inhibitory at more than 1 M (data not shown). Thus, the LPA component in these fluids does not seem to be an important regulator for the proliferation in this cell line. Thus, it remains unclear whether LPA participates in the regulation of the proliferation of pancreatic cancer cells. In any event, components in ascites other than LPA may be involved in the regulation of the proliferation of the cells.

DISCUSSION
In the present study, we showed that ascites from patients with pancreatic cancer contain a considerably high LPA-like activity, which was assessed by the ability of LPA to stimulate LPA 1 receptors. This bioassay excludes LPA species that cannot stimulate LPA 1 , although all the LPA species usually present in biological samples seem to stimulate LPA 1 (20). Moreover, LPA C18:1 (1-oleoyl-sn-glycero-3-phosphate), used as a standard, is usually the most potent species for LPA receptors. Therefore, the true LPA content may be higher than that estimated in the present study. Thus, the bioassay appears to be unsuitable for a quantitative measurement of LPA but may be superior to other methods, such as the combination of chromatography and mass spectrometry (42,43), for the evaluation of the active LPA that stimulates LPA receptors.
Pancreatic cancer is a highly metastatic disease characterized by widespread intraperitoneal dissemination and ascites formation. Indeed, we showed that intraperitoneal injection of a highly invasive pancreatic cancer cell line, YAPC-PD, into mice induced the formation of LPA-rich ascites. However, the molecular mechanism by which LPA accumulates in malignant ascites remains to be elucidated. One possible mechanism is that LPA might be produced from extracellular LPC by lysophospholipase D or autotaxin (44,45). In our preliminary experiments, we detected significant lysophospholipase D activity, which was assessed by choline formation from exogenous LPC, in malignant ascites from pancreatic cancer patients and YAPC-PD cell-injected mice. In any event, the pancreatic cancer cell-injected mice may provide a useful in vivo model system for characterization of the mechanism of the formation of LPA in malignant ascites.
Intraperitoneal dissemination is composed of several processes, including adhesion, migration, invasion, and proliferation (3,4). We demonstrated in the present study that LPA in malignant ascites is an important factor for the stimulation of motility, e.g. migration and invasion in vitro. This is supported by the following observations. First, LPA receptor-selective antagonists, including Ki16425 and VPC12249, inhibited the migration response to ascites as well as LPA. Second, treatment of ascites with MG lipase, an enzyme that hydrolyzes monoglyceride, including LPA, markedly inhibited the migration response. Third, a fractionation study of ascites by lipid extraction and subsequent HPTLC indicated that the active component was recovered in the LPA fraction. Finally, the LPA equivalent level in ascites was high enough to stimulate the migration and invasion of pancreatic cancer cells.
As for proliferation, however, we failed to demonstrate a positive role of LPA in the proliferative activity of malignant ascites. In PK-9 cells, ascites significantly stimulated proliferation, but their effect was insensitive to Ki16425, an LPA receptor antagonist, and MG lipase, an LPA-hydrolyzing enzyme. Furthermore, LPA was ineffective for the induction of the stimulation of proliferation in PK-9 cells (Fig. 8). These results suggest that cytokines and growth factors other than LPA may be responsible for the proliferative activity of ascites in pancreatic cancer cells. It should be noted, however, that LPA stimulated the proliferation of YAPC-PD cells, even though a stimulatory effect was not detected by ascites and effusions (data not shown). Thus, the regulatory mechanism of the proliferation appears to differ in different pancreatic cancer cell lines.
On the other hand, the motility of pancreatic cancer cells seems to be regulated by similar mechanisms involving a G protein-coupled receptor LPA 1 . First, the migration response to ascites and LPA was almost completely suppressed by PTX  (Fig. 2F), a toxin that ADP-ribosylates and inactivates G i/o proteins, suggesting a mediation by G i/o protein-coupled receptors. Among four LPA receptor subtypes, EDG family receptors (LPA 1 , LPA 2 , and LPA 3 ), but not LPA 4 , have been shown to be coupled to the PTX-sensitive G proteins (9,10,20,21). Second, the migration response appears to be correlated with the expression of LPA 1 mRNA ( Fig. 4 and Table II). A significant amount of LPA 1 mRNA is expressed in pancreatic cancer cell lines with high migration activity for ascites and LPA, but not in MiaPaCa-2 cells with low migration activity. In all cells, however, LPA 2 mRNA was similarly expressed, and the expression of LPA 3 and LPA 4 /GPR23 mRNAs was low or marginal except for LPA 3 mRNA in BxPC-3. Third, transfection of LPA 1 in YAPC cells, a parent cell line for YAPC-PD cells, resulted in an enhancement of the migration response to LPA to the level of that in YAPC-PD cells (Fig. 5). Fourth, Ki16425, an antagonist for all EDG family LPA receptors, albeit with less preference for LPA 2 , and VPC12249, an LPA 1 -and LPA 3 -selective antagonist, but not DGPP 8:0, an LPA 3 -selective antagonist, inhibited the migration response (Table III). Ki16425 was also effective for inhibiting the invasion response to LPA and ascites (Fig. 7). Finally, siRNA against LPA 1 markedly inhibited the migration response to LPA and ascites in association with a specific inhibition of the receptor mRNA expression (Fig. 6). Taken together, these results suggest that LPA 1 plays a critical role in the migration response to LPA and malignant ascites in pancreatic cancer cells. Although no intracellular signaling mechanism other than the involvement of G i/o proteins has been analyzed in the present study, recent studies have shown that two small molecule G proteins, Rho and Rac, which regulate actin cytoskeleton rearrangement, are involved in LPAinduced random and directional migration (chemokinetics and chemotaxis, respectively) and tumor cell invasion (24, 25, 29 -31).
Thus, the present study suggests the importance of LPA 1 in the migratory response in pancreatic cancer cells, but this finding can never rule out the possible involvement of LPA receptor subtypes other than LPA 1 receptor, especially LPA 2 , in LPA-induced actions, including the migration response. In the case of LPA 3 , the LPA 3 receptor antagonist DGPP 8:0 was ineffective for the migration response, and LPA 3 receptor mRNA expression was marginal in the highly responsive pancreatic cell lines including HPAC and PK-1, suggesting the minor role of the LPA 3 receptor. However, LPA 2 mRNA is expressed at a high level in all pancreatic cancer cells and might be involved in the signaling pathways leading to migration in collaboration with LPA 1 . In such a situation, inhibition of either receptor would lead to the loss of the response. The stimulatory role of LPA 2 in migration has been shown in Tlymphoma, whereas LPA 1 has a rather inhibitory effect on the migration response in the same cells (27). With respect to proliferation, LPA 2 has been suggested to be stimulatory for ovarian cancer cells (26) and colon cancer cells (31), although the role of LPA receptors in cell proliferation remains to be elucidated in pancreatic cancer cells. Thus, LPA and its receptors are involved in the migration, invasion, and proliferation of various types of cancer cells and may be potential targets for therapy. Indeed, the acceleration of LPA degradation by the introduction of lipid phosphate phosphohydrolase-3 appears to be effective for the control of tumor growth in ovarian cancer (46).
In conclusion, LPA in malignant ascites is an important component for the migration of pancreatic cancer cells through a G protein-coupled LPA receptor LPA 1 . Malignant ascites and LPA also induced the in vitro invasion of pancreatic cancer cells, which was inhibited by Ki16425, an LPA receptor antagonist. Thus, LPA receptors, especially LPA 1 , may be therapeutic targets for the control of the migration and invasion of pancreatic cancer cells. Ki16425 is a potential drug for this purpose.