Modulation of Pro-inflammatory Gene Expression by Nuclear Lysophosphatidic Acid Receptor Type-1*

Lysophosphatidic acid (LPA) is a bioactive molecule involved in inflammation, immunity, wound healing, and neoplasia. Its pleiotropic actions arise presumably by interaction with their cell surface G protein-coupled receptors. Herein, the presence of the specific nuclear lysophosphatidic acid receptor-1 (LPA1R) was revealed in unstimulated porcine cerebral microvascular endothelial cells (pCMVECs), LPA1R stably transfected HTC4 rat hepatoma cells, and rat liver tissue using complementary approaches, including radioligand binding experiments, electron- and cryomicroscopy, cell fractionation, and immunoblotting with three distinct antibodies. Coimmunoprecipitation studies in enriched plasmalemmal fractions of unstimulated pCMVEC showed that LPA1Rs are dually sequestrated in caveolin-1 and clathrin subcompartments, whereas in nuclear fractions LPA1R appeared primarily in caveolae. Immunofluorescent assays using a cell-free isolated nuclear system confirmed LPA1R and caveolin-1 co-localization. In pCMVEC, LPA-stimulated increases in cyclooxygenase-2 and inducible nitric-oxide synthase RNA and protein expression were insensitive to caveolea-disrupting agents but sensitive to LPA-generating phospholipase A2 enzyme and tyrosine kinase inhibitors. Moreover, LPA-induced increases in Ca2+ transients and/or iNOS expression in highly purified rat liver nuclei were prevented by pertussis toxin, phosphoinositide 3-kinase/Akt inhibitor wortmannin and Ca2+ chelator and channel blockers EGTA and SK&F96365, respectively. This study describes for the first time the nucleus as a potential organelle for LPA intracrine signaling in the regulation of pro-inflammatory gene expression.

In the mammalian system, LPA 1 signaling cascades regulate important cellular processes, including gene expression, cell proliferation and growth, cell survival and death, and cell motility and secretion (1)(2)(3). These plethora of activities are characteristic features of inflammation that occur in various physiological as well as pathological states (e.g. ontogenic change, wound healing, cancer, etc.) (1)(2)(3). In humans, physiological responses induced by LPA arise from specific interactions with at least three genetically identified receptors designated LPA 1 , LPA 2 , and LPA 3 (formerly referred to as EDG 2 , EDG 4 , and EDG 7 receptors, respectively), which belong to the heptahelical transmembrane-spanning G protein-coupled receptor (GPCR) superfamily (4). These receptors show a broad, virtually distinct distribution and may couple in a cell-dependent manner to numerous heterotrimeric G proteins. In this context, LPA 1 and LPA 2 receptors have been shown to interact with G i/o , G q/11/14 , and G 12/13 proteins, whereas the LPA 3 receptor combines with G i/o and G q/11/14 proteins (5). Although many responses induced by extracellular LPA can result from its interaction with plasma membrane GPCRs, there is circumstantial evidence for an intracrine mode of action of LPA. For instance, putative biogenesis (e.g. secretory and cytosolic calcium-dependent and -independent phospholipase A 2 , phospholipase D, and monoacylglycerol kinase) and degradation (e.g. phosphohydrolase and lysophospholipase) pathways for LPA have been detected at the nuclear membrane and/or within the nucleus of targeted cells (6 -11). Further support for the functionality of constitutive intracellular LPA receptors, specifically at the cell nucleus, is revealed by adjacent localization of required signaling effectors, which couple to the receptors. These accessory proteins include, among others, G proteins, ion channels, phospholipases A 2 , C, and D, adenylate cyclase, MAPKs, and NF-B (see Ref. 12 for review). Alternatively, LPA may exert intracellular actions by generating its own formation, inferring possible active intracellular binding sites. Along this line of thought, LPA signaling at discrete subcellular domains may be provided in part by intracellular conveyors such as gelsolin and the fatty acid-binding protein (8), by uptake of extracellular LPA bound to albumin (via albondin receptor; gp60) (13), and/or to lipocalins (14).
Lastly, several established pathways of GPCR regulation and desensitization driven by extracellular agonists have also been implicated in their intracellular relocalization, which results in delayed complementary signaling cascades (15,16). In this context, accumulating evidence suggests that nuclear translocation of peptide growth factors (e.g. angiogenin, PDGF, basic FGF, EGF, and PTH) and/or their integral membrane receptors, is mandatory for gene transcription associated with proliferating/growth events (17). Because most of these latter receptors (e.g. EGF, PDGF, and PTH receptors) co-localize within caveolae (18), we speculated that translocation to this specific organelle proceeds via receptor endocytosis through the caveolar compartment. Whether this phenomenon holds true for the transmission of LPA gene responses (1,2) is not yet known.
The biochemical mechanisms by which GPCRs, including LPA-Rs, modulate gene transcription are complex and not fully understood (19). Some mechanisms recently uncovered to explain GPCR-mediated gene induction implicate an endocytosisassociated ␤-arrestin-c-Src interaction leading to downstream activation of Ras and mitogen-activated protein kinases (MAPKs) through possibly metalloprotease-dependent transactivation of receptor tyrosine kinases (RTK) involving de novo release of their ligand (19). Herein, we postulated that endogenous LPA stimulates gene expression, specifically the proinflammatory genes cyclooxygenase-2 (COX-2) and inducible nitric-oxide synthase (iNOS), through a formerly undescribed mechanism, which involves the activation of nuclear LPA receptors that pre-exist at the nuclear envelope or originate from the internalization/endocytosis of plasmalemmal LPA receptors. In the present study, we focused on endothelial cells, which are known to express predominantly LPA 1 R (20) and consolidated our findings on LPA 1 R stably transfected HTC4 rat hepatoma cells (21) and rat liver tissue specimens. Our in vitro and in vivo findings support the existence of constitutive LPA 1 R at the cell nucleus, which upon stimulation mediates calcium transients and transcriptional signals of immediateearly response genes. Our results also suggest that (i) LPAinduced PLA 2 -dependent COX-2 expression is not reliant on prostaglandin, leukotriene, or epoxide production and (ii) contrary to common peptide growth hormone receptors, the sequestration and transcellular transport of LPA-R via caveolae to the nucleus is not a prerequisite for LPA-R activity on gene expression. This study unravels an as yet undescribed mechanism by which LPA modulates gene expression.
Animals-Experiments were performed on endothelial cells derived from Yorkshire piglet brain microvasculature (Fermes Ménard, Quebec, Canada) and hepatocytes from adult Sprague-Dawley male rats (Charles River, Quebec, Canada). Animal housing and experimental protocols were carried out in accordance with regulations set by the Canadian Council of Animal Care Committee and were approved by the Sainte-Justine Hospital Animal Care Committee.
Cell Culture and Fractionation-Primary endothelial cells obtained from porcine cerebral microvessels (26) and stably transfected, Geneticin-resistant LPA 1 -HTC4 rat hepatoma cells (21) were cultured and passaged, as previously reported. Cell lines were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 1% penicillin G, 1% streptomycin and used between passages 6 -13 at ϳ80% confluency unless otherwise stated. Prior to stimulation, cells were serum-starved overnight. Isolation of nuclei was achieved by cell fractionation using the hypotonic/Nonidet P-40 lysis method (27). Nuclear envelopes were prepared by incubating nuclear suspensions (8 ϫ 10 6 nuclei/ml) with DNase I (800 units, pancreas type II, Roche Applied Science) and RNase A (32 mg/ml, Promega) for 30 min at 37°C (28). Supernatants of homogenized cells (in lysis buffer) were sequentially centrifuged at 10,000 ϫ g for 15 min then 120,000 ϫ g for 60 min to obtain mitochondrial and plasma membrane fractions, respectively. Plasma and nuclear membranes and intact nuclei were stored at Ϫ80°C unless otherwise stated. Protein concentration was determined by Bradford protein assay using BSA as standard. The morphological integrity and purity (Ͼ98%) was routinely assessed by light microscopy after trypan blue staining and confirmed by electron microscopy (27) (Fig. 1D, inset). The plasma membrane marker 5Ј-nucleotidase activity (Sigma assay kit) in nuclear versus plasma membrane fraction was less than 7%.
Isolation of Subcellular Fractions of Rat Liver-Hepatocytes were harvested following digestion of liver with collagenase (type II, 0.05%, Sigma), as described (29). Isolation of liver nuclei was carried out by ultracentrifugation through a sucrose gradient according to Nicotera et al. (30). Plasma membrane microsomes and nuclear envelopes were prepared, and morphological integrity and purity were ascertained as indicated above; 5Ј-nucleotidase activity endowed in the isolated nuclear fraction was less than 2% of that of plasma membrane fraction. Purity of subcellular fractions was further substantiated by means of immunological methods using the specific organelle marker antigens alkaline phosphatase and CD51 (plasma membrane), lamin A/C (nuclei), and cytochrome c (mitochondria, cytosol), as depicted below in Fig.  1A. Fig. 1B exhibits electron micrographs of highly purified nuclei (Ͼ98%) and prepared nuclear envelopes (right panel), appearing as intact spheres, following this isolation procedure.
Radioligand Binding Assays-Cells were seeded into 24-well plates (500 l of media/well) and allowed to reach 90% confluency (ϳ70,000 cells/well) before beginning experiments. Quiescent cells were washed thrice with PBS containing BSA faf (0.1%), soybean trypsin inhibitor (100 g/ml), and phenylmethylsulfonyl fluoride (1 mM) then incubated in the same buffer at 4°C for 90 min with 10 nM radioactive tracer [ 3 H]oLPA (50 Ci/mmol). Thereafter, cells were rinsed twice with cold incubation buffer (1 ml), lysed with sodium hydroxide (0.1 N), and transferred into scintillation vials. Nonspecific binding was determined in the presence of unlabeled LPA (10 M). Radioactivity of samples was measured with a ␤-counter. In parallel, cells from untreated wells within the same plate were harvested with trypsin and counted thrice with a hemacytometer for cell count normalization.
Electron and Cryomicroscopic Immunohistochemistry of LPA 1 R-Male Sprague-Dawley rats (250 -300 g) were used. Rat liver tissue sectioning and pre-embedding immunogold staining was done as described in detail previously (33). Vibratome sections (50 m) of liver were incubated with the primary antibody (a rabbit anti-LPA 1 R-C antibody) (1:50) overnight at 4°C followed by another overnight incubation with goat anti-rabbit gold (10 nm)-conjugated IgG (1:50) (British Biocell International). In control sections, the primary antibody was either removed or pre-absorbed with its cognate peptide (10-fold excess by weight, Upstate Biotechnology). Thereafter, specimens were postfixed in 1% osmium tetroxide, subsequently dehydrated in graded ethanol, and embedded in Epon according to standard technique. Ultrathin sections were cut using a Reichert Ultracut ultramicrotome, mounted on Formvar-coated copper grids, stained with uranyl acetate and lead nitrate, and examined with a transmission electron microscope (Philips 410LS, Netherlands).
Electron Cryomicroscopy-All procedures were based on previously reported methods (34). Rat liver sections were cut at Ϫ100°C in an FC4D cryochamber using a Reichert-Jung ultramicrotome and transferred to Formvar-coated copper grids. For immunolabeling assays, sections were incubated for 30 min at RT with rabbit polyclonal antihuman LPA 1 R-N or -C antibody (1:10) followed by incubation with a goat anti-rabbit gold (10 nm)-conjugated IgG (1:20) (Sigma Chemical Co.). Negative controls were analyzed by omitting primary antibodies.
Frozen sections were contrasted and embedded as described (35) and viewed through of a Philips 400T electron microscope.
Indirect Immunofluorescence Staining-Isolated nuclei from pCM-VECs were resuspended in blocking buffer constituted of 10 mM Tris-HCl, 20 mM NaCl, 3 mM MgCl 2 , 300 mM sucrose, complete protease mixture inhibitor, 5% horse serum, and 5% fetal calf serum. Double staining was achieved by incubating the nuclear suspension (1 h at RT) with a polyclonal rabbit anti-LPA 1 R-N (1:50) and a monoclonal mouse anti-caveolin-1 antibody (1:25). The nuclear suspension was diluted in a 5-fold excess of blocking buffer then centrifuged (800 ϫ g, 10 min) and resuspended in similar buffer. The nuclei were subsequently incubated 60 min with a goat anti-rabbit AlexaFluor 488-conjugated IgG (2 g/ml) and a chicken anti-mouse AlexaFluor 647-conjugated IgG (10 g/ml) (Molecular Probes) and washed as indicated above. The nuclear suspension was then placed on poly-L-ornithine-treated glass coverslips (25 mm) and examined with a laser-scanning confocal microscope. Samples were imaged using a 60ϫ differential interference contrast oil immersion objective lens on a Nikon TE300 microscope with a Bio-Rad Radiance 2000 confocal accessory. The two images were collected by using the same Z values and were merged using a Silicone graphic (SGI) software. Nuclear staining was realized at the end of the experiment using the DNA dye Syto-11 (100 nM) (Molecular Probes). For negative controls the primary antibodies were omitted.
Immunoprecipitation of LPA 1 R and Western Blotting-Cellular and nuclear extracts (500 g) of pCMVEC were lysed, immunoprecipitated by an anti-LPA 1 R-N polyclonal antibody (4 g/ml), separated by SDS-PAGE (12%) and transferred onto nitrocellulose membrane. Immunoblotting was performed with either an anti-LPA 1 R-N polyclonal antibody (serving as positive control) (1:2000), anti-clathrin (1:1000), or caveolin-1 (1:1000) monoclonal antibody. Cellular and nuclear extracts of EC were subjected to extraction in Triton X-100 followed by sucrose density gradient centrifugation to isolate caveolae (35). Caveolae-enriched fractions (low density and Triton X-100-insoluble materials) were verified by SDS-PAGE and probed for caveolin-1. Equal amounts of proteins from caveolar and non-caveolar membranes were then immunoblotted for LPA 1 R as indicated above.
Western Blot of COX-2 and iNOS Proteins-pCMVEC were seeded into 10-cm dishes, rendered quiescent with starving medium, and treated for 6 h with either the vehicle (bidistilled and deionized water), sLPA (1-10 M) or human IL-1␤ (10 ng/ml). Cells were then washed twice with cold PBS (5 ml) and harvested in the same buffer by gentle scraping. Sedimented cells were then lysed by brief sonification in buffer consisting of Tris-HCl (10 mM), pH 7.4, EDTA (5 mM), and complete protease mixture inhibitor and then centrifuged at 200 ϫ g for 10 min at 4°C. The supernatant was collected and submitted to an ultracentrifugation (100,000 ϫ g for 60 min at 4°C). Resulting crude membrane and cytosolic extracts were concentrated (Centricon Plus 20, PL-10, Millipore) and analyzed for protein content. Denatured membrane proteins (30 g) were resolved by SDS-PAGE on 7.5% gel and transferred onto PVDF membranes, which were subsequently cut and probed with either an anti-human Von Willebrand factor polyclonal antibody (1:200), an anti-COX-2 polyclonal antibody (1:1000), or an anti-␤-actin polyclonal antibody (1:500). Denatured cytosolic proteins (30 g) were processed in a similar way and immunoprobed using a mouse anti-iNOS monoclonal antibody (1:200). Relative quantities of proteins were expressed in terms of a densitometric ratio to ␤-actin. Densitometric analysis was carried out by integrating the intensity of all pixels in a representative area, excluding background, using Image-Proϩ software (Version 4.1; Media Cybernetic, Silver Spring, MD).
Measurement of Ca 2ϩ Signals by LPA in Isolated Nuclei-Nuclear Ca 2ϩ signals were measured by fura-2-AM technique as essentially described (32) with minor modifications. Briefly, isolated nuclei (1-5 ϫ 10 6 nuclei/ml) placed in incubation buffer (Hepes (25 mM), pH 7.2, KCl (125 mM), K 2 HPO 4 (2 mM), MgCl 2 (4 mM), CaCl 2 (400 nM)), with an osmolarity adjusted with glucose to 300 mosM, were loaded in the dark with fura-2-AM (7 M) (45 min at 4°C). Fura-2-loaded nuclei were then diluted (1:10), centrifuged at 600 ϫ g for 10 min, and resuspended with the incubation buffer. Leakage of fura-2 from nuclei was minimal (Ͻ2%) during the time frame of experiments as reported (31). The [Ca 2ϩ ] nuclear was determined in 1 ml of fura-2-loaded nuclei (ϳ5 ϫ 10 6 nuclei/ml) at 37°C under constant stirring using the ratio of excitation wavelengths 350/380 nm with emission set at 500 nm. Nuclear calcium signals were measured by means of a spectrofluorometer (model LS50, PerkinElmer Life Sciences, UK). Calibration of maximal (R max ) fluorescence signals was determined by sequential addition of ionomycin (5 mM) and calcium (1 mM), whereas that of minimal (R min ) fluorescence signals was obtained with Triton X-100 (1%) followed by EGTA (25 mM). The [Ca 2ϩ ] n was calculated according to Grynkiewicz et al. (43) by the equation: [Ca 2ϩ ] n ϭ K d (224 nM) [(R Ϫ R min )/(R max Ϫ R)] (S f2 /S b2 ), where K d is the dissociation constant (224 nM for fura-2) and S f2 /S b2 is the ratio of fluorescent intensity at 380-nm wavelength. In another set of experiments, confocal microscopy was used to monitor movements of nuclear Ca 2ϩ induced by LPA (1 pM to 1 M) in a single nucleus. For this purpose, nuclear suspension was loaded with fluo-4-AM (30 M) at 25°C for 45 min then diluted in 20ϫ excess volume, spun down, and resuspended in the above-mentioned buffer. Loaded nuclei (in 500 l) were settled on glass coverslips and examined with a Molecular Dynamics (Sunnyvale, CA) Multi Probe 2001 confocal argon laser-scanning system as described previously (44). Rapid linescan imaging was used to monitor temporal oscillations of intranuclear calcium. Stimulated nuclei were recorded at a rate of 320 ms per scan for a total of 250 frames. Each frame consisted of 32 lines/scan (512 pixels) and a pixel size of 0.17 m.
Western Blot of MAP Kinases and Akt Activation-Hepatocyte-derived nuclei (100 g of protein) were treated or not with oLPA (1-10 M) for 0 -30 min in the above-mentioned buffer for calcium signal assay. In concomitant experiments, nuclear suspensions were pretreated with PI 3-kinase/Akt inhibitor wortmannin (100 nM) for 15 min at 37°C and processed as above. Freezing samples in liquid nitrogen terminated the reaction. Proteins were resolved by SDS-PAGE on 9% gel, transferred onto PVDF membranes, and then probed with Erk1/2, phospho-Erk (1/2), Akt, and phospho-Akt antibodies (diluted 1:1000) according to manufacturer's instructions. Autoradiograms were scanned and analyzed by densitometry (ImagePro 4ϩ software).
Statistical Analysis-Data were analyzed by one-way analysis of variance factoring for treatments. Comparison among means was performed by Tukey-Kramer method. Statistical significance was set at p Ͻ 0.05. Data are presented as mean Ϯ S.E. (Fig. 1B) from rat liver (Fig. 1C) and pCMVEC (Fig. 1D) displayed binding sites for [ 3 H]oLPA (10 nM) specific for LPA surrogates (oLPA, sLPA, and pLPA), because lipids structurally distinct from LPA were significantly less efficient or not able to displace the radioactive tracer. pCMVEC and derived nuclei contain 10-to 20-fold more binding sites for LPA than rat liver nuclei.

[ 3 H]oLPA Binding to Subcellular Fractions-Both intact nuclei and nuclear envelopes
Western Blot Analysis of LPA 1 R-Immunoprobing of subcellular fractions from native and recombinant cell systems with an anti-LPA 1 R-N polyclonal antibody revealed a prominent immunoreactive band at ϳ49 kDa ( Fig. 2A), in agreement with Zheng et al. (25). This is suggestive of a receptor with putative carbohydrate adducts, at the nucleus and plasma membrane of the cell types investigated. Similar results were obtained with the anti-LPA 1 R-C polyclonal antibody. The presence of glycan moieties comprised in LPA 1 R was established by assessing the electromobility shift of LPA 1 R protein to the predicted theoretical molecular mass (ϳ38 kDa) (31) following the treatment of rat liver cell membrane with PNGase F (Fig. 2B), or pCMVEC with tunicamycin (Fig. 2C); these oligosaccharide moieties account for ϳ37% of the overall LPA 1 R molecular weight. In pCMVEC, two glycoforms (47 and 49 kDa) of LPA 1 R were detected by performing a low concentration SDS-PAGE (9%) (Fig. 2C). These results conformed to the quasi-universal posttranslational modifications occurring in the GPCR family (45), including that of lysophospholipid receptor system (46).
In Situ Detection of Perinuclear and Intranuclear LPA 1 R by Electron Microscopy-The intracellular localization of LPA 1 R was ascertained in vivo by high resolution immunogold electron microscopy and cryomicroscopy on rat liver tissue sections. LPA 1 Rs were identified at the plasma membrane of hepatocytes (Fig. 3A) and endothelial cells (Fig. 3B), as expected, and notably intracellularly, including the perinuclear and nuclear regions (Fig. 3B) mostly confined to euchromatin structures; these nuclear associated receptors account for 7-15% of overall cellular immunoreactivity. Negative controls revealed no specific labeling when the primary antibody was pre-absorbed with its cognate peptide (Fig. 3C) or was omitted (Fig. 3D). Ultrastructural localization of immunoreactive LPA 1 R in rat liver sections was further substantiated by electron cryomicroscopy using polyclonal rabbit anti-LPA 1 R-C (Fig. 3E) and anti-LPA 1 R-N antibodies (Fig. 3F), which target two distinct epitopes. In both cases, a broad and similar distribution pattern for LPA 1 R was revealed namely at the hepatic plasmalemma (not shown), the endoplasmic reticulum, the perinuclear envelope (arrowheads), as well as inside the nucleus (Fig. 3, E  and F). Sporadic cytoplasmic detection of LPA 1 R was also observed at higher magnification, presumptively in vesicular compartments (not shown).
Co-localization of LPA 1 R and Caveolin-1 in pCMVEC-isolated Nuclei-Many endocytic and transcytotic transport processes involve subcellular redistribution of transmembrane proteins, including possibly GPCRs, which are mainly mediated via either the clathrin or caveolae membrane system (16,18). Therefore, we sought to investigate which of these endocytic systems may be monopolized for the nuclear translocation of cell surface LPA 1 R, if indeed it is occurring. Immunoprecipitation experiments demonstrated that LPA 1 R is dually re-trieved in cell membrane clathrin-coated pits and caveolae while being only present in the caveolar compartment in the nucleus (Fig. 4A). Further support for the nuclear co-localization of LPA 1 R with caveolae in pCMVEC was obtained from Western blot experiments depicting the presence of LPA 1 R primarily in cholesterol-rich detergent-resistant caveolar domains of purified nuclei (Fig. 4B) and from immunofluorescence studies describing the double labeling of LPA 1 R and caveolin-1 in more than 90% of nuclear organelles examined. LPA 1 R co-localized with caveolin-1, as anticipated not exclusively.

Role of Caveolae and Phospholipase A 2 -derived Products in LPA-induced COX-2 and iNOS Gene and Protein Expression-
The induction of immediate-early response genes COX-2 and iNOS by LPA has recently been reported for various cell types. For instance, LPA-induced transcriptional up-regulation of COX-2 enzyme has been shown in renal mesangial cells (47), fibroblasts (48), and astrocytes (49), whereas iNOS has been induced in cardiomyocytes (50). We extended these observations by showing that sLPA (as well as oLPA) increases, in a concentration-dependent manner, COX-2 and iNOS protein expression in brain vascular endothelia (Fig. 5A), whereas the classic endothelial cell antigen von Willebrand factor was unaffected. The concentration-dependent effects of LPA on COX-2 FIG. 2. A, Western blot of LPA 1 R in cell and nuclear fractions of different cell types. Proteins were extracted from: whole nuclei (WN), nuclear envelopes (NE), and plasma membranes (PM) from rat livers (RL), porcine cerebral endothelial cells (EC), and HTC4 rat hepatoma cells. Proteins (25 g) were separated on SDS-PAGE (12%) and probed for LPA 1 R using a rabbit polyclonal anti-LPA 1 R-N as described under "Experimental Procedures." B, immunoblots of rat liver plasma membrane (RLPM) treated or not with PNGase F. Proteins (50 g) were treated as in A. C, immunoblots of total cellular extract of EC pretreated or not with tunicamycin. Proteins were separated on SDS-PAGE (9%) and probed for LPA 1 R as described above. Note the increase of immunoreactivity of antibody toward its N-terminal epitope in the absence of N-linked glycans at LPA 1 R N-terminal segment. Results are representative of three to four experiments. F). LPA 1 R immunoreactivity on plasma membrane of hepatocyte (A) and in perinuclear/nuclear regions of vascular endothelial cell (B) (arrowheads). The absence of immunoreactivity when the anti-LPA 1 R-C antibody is pre-absorbed with its cognate peptide (C) or when the primary antibody is omitted (D). Intracellular distribution of LPA 1 R in Ultrathin sections of rat liver (E and F). Specimens were incubated with either a rabbit polyclonal anti-human LPA 1 R-C (E) or LPA 1 R-N antibody (F) (1:10) followed by incubation with a goat anti-rabbit gold (10 nm)-conjugated IgG (1:20). There was no labeling in the absence of primary antibody (not shown). gene expression were further substantiated by semi-quantitative RT-PCR (Fig. 5B). Based on the abovementioned results demonstrating co-localization of LPA 1 R with caveolin-1 more so in nuclear than plasma microsomal fractions (Fig. 4, A-C), we sought to explore whether caveolae was instrumental in LPA 1 R-mediated COX-2 expression by facilitating its transport to the cell nucleus. The caveolae-disrupting agents, filipin and methyl-␤-cyclodextrin, both used at effective inhibitory concentrations (37,38), failed to modify LPA activity. Similarly, inhibition of clathrin-dependent endocytosis by means of sucrose hypertonic solution (15) did not affect COX-2 expression (not shown). Conversely, LPA-induced COX-2 gene transcription was entirely blocked by the PLA 2 inhibitors mepacrine and cytidine-5Ј-diphosphocholine suggesting a role for endogenous arachidonic acid and/or LPA in this phenomenon; partial impairment (ϳ50%) was also observed with RTK inhibitor tyrphostin AG 1478 and MEK inhibitor PD 98059 (Fig. 5B). The participation of cyclooxygenase-, lipoxygenase-, and epoxygenase-derived products of arachidonic acid and PAF was ruled out based on the inefficiency of corresponding specific enzyme inhibitors to block LPA responses (Fig. 5B). Thus, it appears that LPA signaling in inflammatory gene expression proceeds, independent of caveolae, via PLA 2 -triggered intracellular messengers, possibly LPA per se, acting conceivably on perinuclear/ nuclear receptor sites (see Figs. 1-3). To corroborate this premise, we examined whether a direct stimulation of isolated nuclei with LPA would reproduce or mimic gene responses evoked by extracellular LPA. Results from these series of experiments are presented in the following section.

FIG. 3. In situ ultrastructural localization of nuclear LPA 1 R on rat liver tissue sections by electron microscopy (A-D) and cryomicroscopy (E and
Signaling Effectors of LPA-stimulated Calcium Transients and iNOS Expression in Highly Purified Rat Liver Nuclei-It is now becoming clear that the nucleus possesses autonomous calcium signaling machinery (e.g. ion channels, protein kinases, and phospholipases) where the nuclear envelope serves as the nuclear calcium pool (12). Furthermore, nuclear calcium is instrumental in regulating many processes, including gene transcription (12). We therefore examined whether changes in nuclear Ca 2ϩ can be elicited by LPA 1 R stimulation by means of spectrofluorometry and live imaging confocal microscopy using different fluorescent Ca 2ϩ -sensitive probes (fura-2-AM and fluo-4-AM). Incubation of isolated hepatic nuclei with oLPA induced a concentration-dependent increase in nuclear Ca 2ϩ levels (Fig. 6A); indistinguishable responses were seen with LPA analogues (1 M) comprising distinct fatty acid species (sLPA and pLPA). Similar responses were observed in LPA 1 R-HTC4 transformed hepatic cell-derived nuclei. Moreover, the Ca 2ϩ chelator EGTA and putative receptor-operated Ca 2ϩ channel blocker SK&F96365 prevented LPA-induced Ca 2ϩ transients, as seen with PTX suggestive of G i/o coupling, whereas the Ca 2ϩ ATPase pump inhibitor thapsigargin was inoperative (Fig. 6, A and B). Rapid nuclear calcium mobilization in response to LPA was visualized in a single nucleus by confocal microscopy (Fig. 6B). Live scan imaging clearly illustrated a progressional transient Ca 2ϩ movement across the nucleus (Fig. 6B, third panel). Notably, these discrete nucleoplasmic Ca 2ϩ responses evoked by LPA were concentration-dependent (with an apparent affinity in the nanomolar range) (Fig. 6C), reversible and seemingly desensitizable (Fig. 6D) suggestive of a receptor-operated mechanism. Alternatively, this latter fading of organelle responsiveness upon repeated challenges of high concentration of LPA (1 M) might reflect a re-localization and sequestration of calcium in unvisualized nuclear foci.
Direct stimulation of nuclei with oLPA (as well as sLPA) also provoked dose-dependent, calcium-and PTX-sensitive iNOS gene induction (Fig. 7), consistent with results from calcium mobilization assays (see Fig. 6A). Along with the presence of nuclear Akt (Fig. 7A, inset), LPA effects were also associated with phosphorylation (and resultant activation) of this kinase (Fig. 7B), which in turn participated in iNOS gene induction. Accordingly, this latter effect was abolished by PI 3-kinaseactivated protein kinase Akt inhibitor wortmannin (Fig. 7, A  and B). In similar experimental conditions, LPA failed to induce phosphorylation of nuclear Erk1/2 kinases (not shown). DISCUSSION The generally accepted view of signaling to immediate-early growth-related genes by cell surface GPCRs for lysophospholipid growth factors centers on ras-and rho-dependent activation of transcription through the multicomponent serum-response element (1). The results of our present study support an alternative signaling pathway by which LPA-LPA 1 R complexes in the nucleus more directly evoke transcriptional events. In this study, we unveil for the first time the presence of functional LPA 1 Rs at the cell nucleus. The novel distribution of LPA 1 Rs was discovered in cultured microvascular endothelial cells from brain, where native LPA-Rs, specifically the LPA 1 Rs, are particularly abundant (32, 51, 53), in HTC4 rat hepatoma cells stably transfected with recombinant LPA 1 Rs (21) and more importantly, in vivo, in rat liver tissues, which express high levels of native LPA-Rs (51), presumably of the LPA 1 R type (54), which are highly responsive to LPA, as assessed by Ca 2ϩ mobilization assays (55,56). Our findings also reveal that these nuclear LPA receptors are functional and are coupled to G i/o proteins, calcium signaling, PI 3-kinase/Akt activation, and subsequent inflammatory gene transcription.
Substantial evidence supports the presence of cell surface GPCRs, mostly for peptide ligands (e.g. angiotensin II, somatostatin, substance P, and neurotensin) in perinuclear/nuclear domains of various cells following their stimulation with extracellular agonists (12). In most cases, however, neither the mechanism of GPCR intracellular distribution nor the functional consequence of receptor internalization and nuclear translocation have been explicitly established, with the exception perhaps of angiotensin II (12). The presence of putative nuclear localization signal sequences comprised within C-terminal segments of GPCRs is believed, although equivocally, to pertain to their nuclear sequestration via a Ran-GTP/importin mechanism as suggested for angiotensin II AT 1 receptors (12). However, human and rodent LPA 1 Rs do not bear such amino acid sequences implying that other nuclear import mechanisms may prevail, possibly brought into play following receptor biogenesis and/or endocytosis. In this context, we showed that in pCMVEC membrane, LPA 1 Rs are partitioned in both clathrin and caveolar endocytic microdomains (Fig. 4); the former domain usually being ascribed for receptor recycling or degradation (57), whereas the latter is thought to serve as a protein carrier responsible for intracellular redistribution of various signaling components (18). Notably, other transmembrane receptors such as ␤ 2 adrenergic (58), endothelin ET A (59), and interferon ␥ receptors (60), have been reported to be dually compartmentalized into caveolae and clathrin-coated pits in cell membranes. On the other hand, our results do not support a nuclear translocation of cell surface LPA 1 R-evoking inflammatory (COX-2) gene expression mediated through endocytic caveolar pathway (Fig. 5B), despite LPA 1 -R and caveolin-1 co-localization at the nucleus (Fig. 4, A-C). Rather, our data suggest a major role for released PLA 2 products in LPA induced-COX-2 expression (Fig. 5B); these PLA 2 products are unlikely to derive from cyclooxygenase, lipoxygenase, or cytochrome P450 enzymatic systems, but possibly LPA. Interestingly, a similar PLA 2 -dependent mechanism for ␣ 2B -adrenergic receptor-mediated mitogen-activated protein kinases (MAPKs) activation has been recently documented (36).
A novel pathway has recently been reported to explain LPAevoked signaling of cellular proliferation, which involves MAPK activation through RTK autophosphorylation (e.g. EGF and PDGF receptors) (19,61). The presumption of auto-induced generation of LPA in extracellular LPA-evoked COX-2 expression is strengthened from experiments describing the limited effectiveness of the RTK-specific inhibitor tyrphostin and MEK kinase inhibitor PD 98059 (Fig. 5B) compared with PLA 2 inhibitors. This may suggest that LPA-induced COX-2 up-regu-lation could proceed via a dual signaling system involving both PLA 2 and RTK activities, functioning at least in part independently. Alternatively, affecting the intracellular redox state could stimulate LPA induced-COX-2 gene transactivation. Of relevance, hydrogen peroxide is known to stimulate PLA 2 (62) and phospholipase D (63) enzymes, and peroxide concentration is increased in response to exogenous LPA (64). These phospholipases, which also catalyze LPA formation, can be detected in their active state at the nuclear membrane (8 -12, 65). Taken together, these observations infer the nucleus as a putative organelle in LPA formation at the vicinity of its cognate perinuclear receptors involved in gene transcription.
The nuclear network ensures the commutation and coordination of numerous biochemical signals that are critical in gene expression. In this study, we have begun to address the molecular mechanisms underlying LPA 1 R-mediated inflammatory gene expression. Our results suggest that perinuclear LPA 1 Rs regulate iNOS gene expression through G i/o proteins, calcium channels, and PI 3-kinase/Akt activation in accordance with the increasing ordnance of intrinsic downstream effectors found in the nucleus (10,12,66). The lack of potent and selective LPA receptor antagonists currently precludes unequivocal establishment of a sole dependence of LPA nuclear signaling on LPA 1 Rs and leaves open the possibility that other promiscuous LPA receptors participate meaningfully. Nonetheless, the inference of a predominant role for nuclear LPA 1 R-transducing transcriptional events stems from our observations. The presence of high levels of immunoreactive LPA 1 R on plasma and nuclear membranes of hepatic cells (Figs. 2 and 3) and the PTX sensitivity of LPA-evoked calcium and gene responses (Figs. 6A and 7A), independently of the degree of fatty acid saturation of LPA, all reflect common characteristic features of this LPA-R type (1,4). Overall, the demonstration of intracellular receptor sites for LPA provides an alternative concept to explain the duality in responses to GPCR ligands (e.g. PTH and sphingosine 1-phosphate) in activation of their intracellular versus extracellular targets (12,67,52).