Different Mechanisms Regulate Lysophosphatidic Acid (LPA)-dependent Versus Phorbol Ester-dependent Internalization of the LPA1 Receptor*

Lysophosphatidic acid (LPA) stimulates cells by activation of five G-protein-coupled receptors, termed LPA1–5. The LPA1 receptor is the most widely expressed and is a major regulator of cell migration. In this study, we show that phorbol ester (PMA)-induced internalization of the LPA1 receptor requires clathrin AP-2 complexes, protein kinase C, and a distal dileucine motif (amino acids 352 and 353) in the cytoplasmic tail but not β-arrestin. Agonist-dependent internalization of LPA1, however, requires a cluster of serine residues (amino acids 341–347) located proximal to the dileucine motif, β-arrestin, and to a lesser extent clathrin AP-2. The serine cluster of LPA1 is required for β-arrestin2-GFP translocation to the plasma membrane and signal desensitization. In contrast, the dileucine motif (IL) is required for both basal and PMA-induced internalization. Evidence for the β-arrestin independence of PMA-induced internalization of LPA1 comes from the observations that β-arrestin2-GFP is not recruited to the plasma membrane upon PMA treatment and that LPA1 is readily internalized in β-arrestin1/2 knock-out mouse embryonic fibroblasts. These results indicate that distinct molecular mechanisms regulate agonist-dependent and PMA-dependent internalization of the LPA1 receptor.

The plasma membrane expression of functional GPCRs 2 can be regulated by a variety of mechanisms such as biosynthetic transport, endocytosis, endosomal sorting, recycling, and degradation. Specific amino acids found in the cytoplasmically exposed domains of a variety of GPCRs influence their intracellular trafficking (1)(2)(3). Many GPCRs undergo basal and/or agonist-induced endocytosis through both clathrin-dependent and clathrin-independent pathways (e.g. caveolae, etc.) by forming associations with specific adaptor proteins (4 -7).
Depending on the characteristics of agonist-dependent arrestin association, two classes of GPCRs have been proposed, class A and class B, which differ in their affinities for ␤-arrestin1 and -2 and in their spatiotemporal association (13).
Other adaptor proteins like clathrin AP-2 can directly mediate the internalization of some GPCRs (PAR-1 and thromboxane A 2 ) (14 -16). Several studies have shown that discrete amino acid motifs such as dileucine-based (LL/IL) and tyrosine-based (YXX) motifs, which are present in the cytoplasmic tails of GPCRs, mediate the direct interaction with clathrin and AP-2 to promote endocytosis (15,17,18).
We have been studying the trafficking of the most widely expressed GPCR for the serum phospholipid lysophosphatidic acid (LPA), termed LPA 1 (7,19). LPA has growth factor-like properties and is involved in a variety of processes such as wound healing, cell proliferation and survival, neurite retraction, and cell migration (20). LPA stimulates these processes mainly through the activation of five distinct GPCRs, LPA [1][2][3][4][5] (21,22). We have previously shown that the LPA 1 receptor utilizes a clathrin-and ␤-arrestin-dependent pathway for agonist-induced internalization (7,19).
In this study, we investigated the amino acid sequences in the cytoplasmic tail of the LPA 1 receptor that are involved in receptor internalization. Two distinct motifs are required for agonistdependent versus agonist-independent internalization of the LPA 1 receptor. A serine cluster ( 341 SDRSASS 347 ) in the C-terminal tail is required for ␤-arrestin association, signal attenuation, and subsequent endocytosis after LPA stimulation. A more distal dileucine motif ( 352 IL 353 ) is required for agonistindependent and phorbol ester-induced internalization of the LPA 1 receptor. This type of internalization is independent of ␤-arrestin but requires the clathrin adaptor AP-2. This suggests that two distinct mechanisms regulate agonist-dependent and PMA-dependent internalization of the LPA 1 receptor.
Plasmids-HA-LPA 1 plasmid was generated by PCR, using a previously described FLAG-LPA 1 plasmid as template (19). HA-tagged truncation mutant receptors, as described in the figure legends, were generated using the Gene-Tailor site-directed mutagenesis kit (Invitrogen), according to manufacturer's instructions, with the HA-LPA 1 wild-type plasmid as template. All plasmids were subjected to DNA sequencing to confirm their sequences. ␤-Arrestin GFP plasmids were kindly provided by Dr. Stefano Marullo.
Internalization Assay-LPA 1 /HeLa cells or HeLa cells were plated on coverslips. For internalization assays involving different mutant LPA 1 receptors, HeLa cells were transfected with plasmid DNA as described in the figure legends. For experiments with siRNA-mediated reduction of AP-2, LPA 1 /HeLa cells were transfected with siAP-2 as described in the figure legends. The day before experimentation, the cells were serumstarved overnight. Cells were incubated with or without inhibitors as described in figure legends and were then transferred onto ice and incubated with mouse anti-HA primary antibodies (Covance) for 1 h to label cell surface HA-tagged LPA 1 (wildtype and mutant) receptors. Cells were then either left at 4°C (total surface) or transferred to 37°C and treated as described in the figure legends. After incubation at 37°C, antibodies bound to the cell surface were removed by rinsing the cells with 100 mM glycine, 20 mM magnesium acetate, 50 mM KCl, pH 2.2 (acid wash) (24), for 90 s. For experiments with co-transfected HA-LPA 1 receptors and ␤-arrestin2 GFP, the surface antibodies were not removed by acid wash. Cells were rinsed and processed for immunofluorescence as described below.
Immunofluorescence-HA-LPA 1 receptors were detected using mouse anti-HA antibody (Covance). Immunofluores-cence detection of AP-2 was done using a mouse anti-AP-2 antibody (AP-6) (Santa Cruz Biotechnology). For internalization assays, following anti-HA antibody incubations at 4°C and treatments at 37°C, the cells were then fixed in 2% formaldehyde in phosphate-buffered saline (PBS) for 10 min and rinsed with 10% fetal bovine serum containing 0.02% azide in PBS (PBS serum). For indirect immunofluorescence and AP-2 detection, following overnight serum starvation and subsequent treatments, cells were fixed and rinsed with PBS serum and were incubated with mouse anti-HA or anti-AP-2 antibodies diluted in PBS serum containing 0.2% saponin for 45 min. Following fixation and incubation with primary antibodies, the cells were washed three times with PBS serum and then incubated with fluorescently labeled donkey anti-mouse secondary antibodies (Jackson ImmunoResearch) diluted in PBS serum containing 0.2% saponin for 45 min, washed three times with PBS serum, washed once with PBS, and mounted on glass slides as described previously (19). Images were captured using a Hamamatsu digital camera mounted on a Leica Inverted microscope with a 100ϫ oil immersion objective. Images were processed with Adobe Photoshop 7.0 software.
Quantification of Internalization-For internalization assays, images were taken using a Hamamatsu digital camera mounted on a Leica inverted microscope with a 100ϫ oil immersion objective. The images were analyzed by Simple PCI software (Compix, Cranberry Township, PA), and total fluorescence for both internalized (vesicles/cell) and surface antibody levels were measured as described previously (25). Internalization (cell-associated fluorescence after acid wash) is expressed as a percentage of total fluorescence of initial surface-bound antibodies (4°C) for both wild-type and mutant LPA 1 receptors.
Confocal Microscopy for ␤-Arrestin2-GFP Translocation Assay-To observe ␤-arrestin2 GFP translocation to the plasma membrane in response to LPA or PMA stimulation, HeLa cells were plated onto 35-mm glass bottom microwell dishes (MatTek Corp., Ashland, MA). The cells were co-transfected with HA-tagged LPA 1 R and ␤-arrestin2-GFP and, after 24 h post-transfection, were treated with either 10 M LPA or 1 M PMA as described in the figure legends. The cells were then processed for immunofluorescence as described earlier. Images were taken using a Zeiss laser-scanning confocal microscope (LSM-510).
siRNA-mediated Reduction of AP-2-siRNA oligonucleotides to the -subunit of adaptin were purchased from Dharmacon (Lafayette, CO) and have been described previously (26). LPA 1 /HeLa cells were transiently transfected with 175 pmol (12-well plate) or 300 pmol (6-well plate) of siRNA to AP-2 (siAP-2) using Lipofectamine 2000 (Invitrogen). The transfection medium was replaced with complete medium (without penicillin/streptomycin) 5 h later, and the cells were trypsinized after an additional 2 h and plated in a 6-well plate with (immunofluorescence) or without (immunoblotting) coverslips and then incubated for 16 h. The cells were transfected a second time as above, and the medium was then replaced with serum-free medium and incubated for an additional 16 h before experimentation.
Phosphoinositide Hydrolysis-HeLa cells were plated at a density of 4 ϫ 10 4 cells/well into 24-well plates and transfected with plasmids encoding wild-type or mutant HA-LPA 1 using Exgen500 (Fermentas). At 24 h post-transfection, cells were labeled overnight with myo-[3H]inositol in inositol-and serumfree medium, treated as described in the figure legends, and then processed for analysis of phosphoinositide hydrolysis by anion exchange chromatography as described (7).
Statistical Analysis-The data are expressed as the mean Ϯ S.E. from the indicated number of independent experiments. Differences were analyzed by two-factor analysis of variance followed by a Tukey's statistical significance test.

RESULTS
Differential Sensitivity of LPA-Versus PMA-induced Internalization of the LPA 1 Receptor (LPA 1 R) to PKC Inhibitors-We were interested in defining the molecular determinants of the LPA 1 R that mediate its endocytic trafficking. In addition to LPA, a previous study indicated that the phorbol ester, PMA, induced PKC-dependent phosphorylation, signal desensitization, and internalization of the LPA 1 R in C9 rat hepatoma cells (27). To determine whether PMA also induced LPA 1 R endocytosis in HeLa cells, we compared the effects of LPA and PMA on the distribution of LPA 1 R by using indirect immunofluorescence microscopy after fixation (Fig. 1A). To label endosomes, these cells were incubated with Alexa-labeled transferrin (Tfn) prior to fixation. In untreated cells, LPA 1 Rs were localized predominantly to the plasma membrane and Alexa-Tfn-labeled punctate endosomal compartments. Following either LPA or PMA treatment, LPA 1 Rs co-localized  1 Rs to endosomes loaded with 10 mg/ml Alexa 594-Tfn as shown after fixation and indirect immunofluorescence microscopy. B, LPA and PMA stimulate the co-uptake of anti-HA antibodies bound to cell surface LPA 1 Rs into cells. HA-LPA 1 R/HeLa cells were incubated with mouse anti-HA antibodies at 4°C (30 min) and then warmed in the presence or absence (Untreated) of 10 M LPA or 1 M PMA for 30 min prior to rinsing with a mild acid wash, to remove remaining cell surface antibody. Cell-associated anti-HA antibody was visualized by staining with fluorescently labeled secondary antibodies as described under "Materials and Methods." C, internalized anti-HA antibodies localize to Alexa-Tfn-containing endosomes. HA-LPA 1 R/ HeLa cells were labeled with anti-HA antibodies at 4°C (1 h) and warmed to 37°C for 30 min in the presence or absence of 10 M LPA or 1 M PMA along with 10 mg/ml Alexa 594-Tfn, to label endosomes, prior to acid wash and fixation. Cell-associated anti-HA antibodies were visualized by staining with fluorescently labeled secondary antibodies as described under "Materials and Methods." D, PMA-induced internalization of LPA 1 Rbound anti-HA antibodies is inhibited by the PKC inhibitor Gö 6976. LPA 1 R/HeLa cells were pretreated with vehicle, 5 mM Gö 6976, or 5 mM Gö 6983 for 30 min at 37°C prior to cooling and labeling with anti-HA antibodies at 4°C for 1 h. The cells were warmed to 37°C in the presence or absence of LPA or PMA and the inhibitors as described above, acid-washed, fixed, and processed for immunofluorescence localization of internalized anti-HA antibodies. Internalized anti-HA antibodies were quantified using Simple PCI image analysis software as described under "Materials and Methods." Internalization was expressed as a percent of the total fluorescence of initial surface-bound antibody of the 4°C untreated control. **, p Ͻ 0.01 compared with control samples, not treated with PKC inhibitors. FEBRUARY 29, 2008 • VOLUME 283 • NUMBER 9

JOURNAL OF BIOLOGICAL CHEMISTRY 5251
with Alexa-Tfn in endosomal compartments, which indicated that both of these stimuli induced LPA 1 R endocytosis. Interestingly, these LPA 1 R ϩ and Alexa-Tfn ϩ endosomes had a more clustered appearance following PMA treatment.
To quantify LPA 1 R endocytosis, we took advantage of the N-terminal HA epitope tag on LPA 1 R, which is exposed to the medium at the cell surface, and used an HA antibody co-internalization assay (Fig. 1B). Mouse anti-HA antibodies were first bound to cells expressing HA-tagged LPA 1 Rs at 4°C, and the cells were either left at 4°C (total surface) or treated with vehicle, LPA, or PMA at 37°C. During incubation at 37°C, the HA antibody was co-internalized along with the LPA 1 R. The remaining surface-bound antibodies were removed with a mild acid wash, whereas the co-internalized antibody bound to the LPA 1 receptor was retained. At 4°C, the anti-HA antibodies bound to LPA 1 Rs, localized to the plasma membrane (Fig. 1B, left panel). When these cells were warmed to 37°C in the absence of any stimuli and then rinsed with a mild acid wash, very little anti-HA antibody was internalized (Fig. 1B, untreated, 2nd panel from the left). However, incubation with either LPA or PMA markedly increased internalization of the anti-HA antibody, which localized to endosomal-like structures.
To confirm the endosomal localization of the internalized anti-HA-bound LPA 1 Rs, we repeated these experiments and included Alexa-Tfn during the incubations at 37°C (Fig. 1C). In these experiments, the internalized anti-HA-bound LPA 1 Rs extensively co-localized with Alexa-Tfn-labeled endosomes. The results of these anti-HA antibody internalization experiments gave the same results as those obtained by indirect immunofluorescence after treatment and fixation. The bound anti-HA antibody did not alter the pattern of internalization or the endosomal localization of LPA 1 Rs, thus validating this antibody internalization assay to monitor LPA 1 R internalization.
We quantified the amount of anti-HA antibody that was internalized relative to what was initially bound at 4°C by using Simple PCI image analysis software as published previously (7) (Fig. 1D). Cells incubated in the absence of LPA and PMA only internalized about 9% of the initial bound anti-HA antibody (Fig. 1D, untreated, white bar). This represents the basal internalization of LPA 1 Rs. Incubation with LPA for 30 min induced ϳ38% internalization of anti-HA antibody-bound LPA 1 Rs (Fig.  1D, LPA, white bars) and PMA induced ϳ28% internalization of anti-HA-bound LPA 1 Rs (PMA, white bars).
To test the role of PKC in LPA 1 R internalization, we examined the effects of the PKC inhibitors Gö 6976, which inhibits both conventional PKCs and PKD/PKC (28), and Gö 6983, which inhibits conventional and novel PKC isoforms but not PKD/PKC (28,29). Preincubation with 5 mM Gö 6976 (Fig.  1D, PMA, black bar), but not Gö 6983 (Fig. 1D, PMA, gray bar), inhibited PMA-induced internalization of LPA 1 Rs by ϳ70%. Neither inhibitor inhibited LPA-induced internalization (Fig.  1D, LPA). These results indicated that PKC, and perhaps PKD/ PKC, was required for PMA-induced but not LPA-induced internalization. This also raised the possibility that different mechanisms regulated LPA 1 R internalization in response to these two stimuli.
The Cytoplasmic Tail of LPA 1 Rs Is a Critical Determinant of LPA Signal Strength-The cytoplasmic tails of many GPCRs are critical for the regulation of receptor signaling and/or trafficking (30,31). Inspection of the amino acid sequence of the LPA 1 R cytoplasmic tail reveals several discrete amino acid motifs, including a di-cysteine palmitoylation motif (aa 328 and 329), a serine-rich domain (aa 341-347), a di-leucine motif (aa 352 and 353), and a PDZ binding domain (aa 362-364) (Fig. 2A).
We investigated the importance of these motifs to LPA 1 R signaling and intracellular trafficking by creating several truncated HA-tagged receptor mutants ( Fig. 2A). Truncation of the cytoplasmic tail to eliminate the putative palmitoylation site, as well as the other motifs, resulted in a misfolded LPA 1 R, which was retained in the endoplasmic reticulum (data not shown). Therefore, we focused on four mutants indicated in Fig. 2A to investigate the role of the serine-rich domain, di-leucine motif, and the PDZ binding domain. The effects of these truncated mutants on LPA signaling was investigated in transiently transfected HeLa cells by using a phosphoinositide (PI) hydrolysis assay, which measures G␣ q activation in response to LPA stimulation (7,14). Cells expressing wild-type (WT) LPA 1 displayed a 5-fold increase in PI hydrolysis upon LPA stimulation (Fig.  2B). A mutant LPA 1 R that lacked the PDZ binding domain (⌬361) displayed a slight decrease in LPA-stimulated PI hydrol-FIGURE 2. Truncation of the LPA 1 R cytoplasmic tail increases LPA-induced phosphoinositide hydrolysis. A, amino acid sequence of the cytoplasmic tail of LPA 1 R is listed along with annotation of amino acid motifs of interest and sites of truncation mutants. B, serine-rich domain is required for signal attenuation. HeLa cells were transiently transfected with plasmids encoding either wild-type LPA 1 R (WT) or the indicated truncation mutants of LPA 1 R. The cells were labeled with radioactive inositol and incubated in the absence (untreated) or presence (10 M LPA) of agonist for 1 h, and radioactively labeled inositol phosphates were isolated as described. The radioactivity recovered in the different samples was normalized to both total cellular protein and total receptor surface expression as determined by Western blotting (data not shown). The data are presented as the mean Ϯ S.E. of triplicate measurements from a representative experiment that was repeated three times. *, p Ͻ 0.05; **, p Ͻ 0.01 comparing the responses of the mutant receptors to WT LPA 1 Rs. ysis but was indistinguishable from WT LPA 1 Rs in basal activation. However, successive truncation of the cytoplasmic tail led to a progressive increase in LPA-induced PI hydrolysis (Fig.  2B, see ⌬353, ⌬347, and ⌬340). These results indicated that the region between residues 340 and 361 dampen LPA signaling and suggest that they are important for signal attenuation.
The Serine-rich Region Is Critical for ␤-Arrestin Association-Because the region between residues 340 and 361 dampened LPA 1 R signaling, we hypothesized that this region was critical for signal desensitization and therefore ␤-arrestin interaction. We next sought to determine whether the serine-rich region is required for ␤-arrestin interaction. We have previously showed that LPA 1 Rs transiently (i.e. following 2 min of LPA stimulation) associate with ␤-arrestins at the plasma membrane and that this association is required for signal attenuation (7). In cells expressing LPA 1 Rs and ␤-arrestin2-GFP, the latter protein translocates to the plasma membrane and co-localizes with LPA 1 Rs and clathrin AP-2 complexes following a 2-min LPA stimulation (7).
To investigate the role of the cytoplasmic tail in LPA 1 R/␤arrestin interactions, HeLa cells were co-transfected with different HA-tagged LPA 1 Rs and ␤-arrestin2-GFP, stimulated with LPA for 2 min, and then analyzed by confocal fluorescence microscopy (Fig. 3A). In cells expressing WT, ⌬347, ⌬353, or ⌬361 LPA 1 Rs, ␤-arrestin2-GFP translocated to the plasma membrane and co-localized with LPA 1 Rs (Fig. 3A, see arrows), after brief LPA stimulation (Fig. 3A, WT). In contrast, ␤-arres-tin2-GFP failed to translocate to the plasma membrane in cells expressing ⌬340 LPA 1 Rs (Fig. 3A, ⌬340). This suggested that the serine-rich region between aa 340 and 347 was critical for interaction of LPA 1 Rs and ␤-arrestin2, but that neither the dileucine motif nor the PDZ binding domain was required for ␤-arrestin recruitment.
We next quantified the percentage of cells where ␤-arres-tin2-GFP co-localized with LPA 1 Rs following brief LPA stimulation (Fig. 3B). In the cells expressing WT, ⌬353, and ⌬361 LPA 1 Rs, ␤-arrestin2-GFP co-localized with LPA 1 Rs in ϳ90% of the cells. In contrast, ␤-arrestin2 GFP marginally co-localized with LPA 1 Rs in only ϳ10% of the cells expressing the ⌬340 LPA 1 R. ␤-Arrestin2-GFP co-localized with the ⌬347 LPA 1 R in 70% of the cells. These results suggest that the serine-rich region between aa 340 and 347 is required for ␤-arrestin interaction with the LPA 1 Rs.
Different Motifs in the Cytoplasmic Tail of the LPA 1 Receptor Are Required for LPA-Versus PMA-induced Endocytosis-We next sought to examine the role of the different amino acid motifs in the cytoplasmic tail of LPA 1 R in either LPA-induced or PMA-induced endocytosis. HeLa cells were transiently transfected with different HA-tagged LPA 1 receptors and tested for their ability to co-internalize anti-HA antibody following stimulation with either 10 M LPA or 1 M PMA for 30 min (Fig. 4).
Following LPA treatment, WT, ⌬347, ⌬353, and ⌬361 LPA 1 Rs internalized anti-HA antibody into endosomal compartments, which suggested that these mutant receptors retained the ability to undergo LPA-induced internalization (Fig. 4A, upper panels). In contrast, cells expressing ⌬340 LPA 1 Rs did not internalize anti-HA antibodies, which sug-gested that this receptor was not internalized in response to LPA. Because ⌬340 LPA 1 Rs lack the serine-rich domain, which is retained in all of the other receptor mutants, these results indicate that the serine-rich domain (aa, 340 -347) is essential for LPA-induced endocytosis.
In contrast to LPA-induced internalization, neither the ⌬340 nor ⌬347 LPA 1 R mutants were internalized in response to PMA stimulation; only WT, ⌬353, and ⌬361 LPA 1 Rs were internalized upon PMA treatment (Fig. 4A, bottom panels). This indicated that the di-leucine motif (aa 352 and 353) was critical for PMA-induced internalization but not for LPA-induced internalization. MetaMorph quantification of anti-HA internalization indicated that both ⌬340 and ⌬347 LPA 1 Rs displayed reduced basal endocytosis in addition to the inhibition of LPA-induced or PMA-induced endocytosis, respectively (Fig. 4B). Taken together, these results indicated that the serine-rich domain in the cytoplasmic tail, which is important for A, HeLa cells were co-transfected with WT or mutant HA-LPA 1 R and ␤-arres-tin2-GFP plasmids as described under "Materials and Methods." Surface receptors were labeled with mouse anti-HA antibodies at 4°C for 1 h. The cells were washed and then incubated in the absence (data not shown) or presence of 10 M LPA for 2 min at 37°C. The cells were fixed and processed for confocal fluorescence microscopy as described under "Materials and Methods." Arrows point to regions of co-localization. A schematic diagram of the cytoplasmic tail is shown above the micrographs with the deletion mutants indicated. B, ␤-Arrestin2-GFP translocation to the plasma membrane after 2 min of LPA or PMA stimulation was quantified by scoring 50 cells for each receptor construct for co-localization with LPA 1 Rs. The data from four separate experiments were averaged and reported as the mean Ϯ S.E. **, p Ͻ 0.01 in comparison to WT LPA 1 R-expressing cells. FEBRUARY 29, 2008 • VOLUME 283 • NUMBER 9 ␤-arrestin interaction, mediates LPA-induced endocytosis of LPA 1 Rs but that a novel di-leucine motif, and perhaps also the serine-rich domain, is responsible for basal and PMA-induced internalization.

Differential LPA 1 R Endocytosis
PMA-induced Internalization of the LPA 1 Receptor Is ␤-Arrestin-independent-The results of the previous experiment suggested that PMA-induced internalization might be independent of ␤-arrestins because ⌬347 LPA 1 Rs, which can recruit ␤-arrestin to the plasma membrane and are able to undergo LPA-induced internalization, were not internalized in response to PMA.
To test this hypothesis, we asked whether PMA treatment induced ␤-arrestin2-GFP translocation to the plasma membrane using confocal fluorescence microscopy (Fig. 5). In untreated cells, as shown previously, LPA 1 Rs localized predominantly to the plasma membrane distribution, whereas ␤-arres-tin2-GFP had a diffuse cytoplasmic distribution. After LPA treatment for 2 min, both the LPA 1 Rs and ␤-arrestin2-GFP co-localized to punctate structures at the plasma membrane (Fig. 5A, 2 min/LPA). Prolonged exposure to LPA for 30 min led to the endosomal localization of LPA 1 Rs and ␤-arrestin2-GFP localized to the cytoplasm (Fig. 5A, 30 min/LPA).
In contrast to LPA stimulation, treatment with PMA for 2 min did not lead to the recruitment of ␤-arrestin2-GFP to the plasma membrane (Fig. 5A, 2 min/PMA). Prolonged PMA treatment also led to the endosomal localization of LPA 1 Rs, whereas ␤-arrestin2-GFP remained in the cytoplasm (Fig. 5B, 30 min/PMA). We quantified the percentage of cells in which ␤-arrestin2-GFP was translocated from the cytosol to the plasma membrane after brief LPA or PMA stimulation (Fig.  5B). These results confirmed that ␤-arrestin2-GFP is recruited to the plasma membrane by LPA 1 Rs upon LPA treatment but not after PMA treatment. To further investigate whether PMAinduced internalization of LPA 1 Rs was ␤-arrestin-independent, we compared the internalization of WT LPA 1 Rs, after stimulation by LPA or PMA, in MEFs derived from ␤-arres-tin1/2 double knock-out mice (Fig. 6).
We transiently transfected MEFs derived from wild-type mice (MEF WT) and those derived from ␤-arrestin1/2 KO mice (MEF KO) with WT and HA-tagged LPA 1 Rs and tested for their ability to internalize anti-HA antibody as before (Fig. 6). Cell surface LPA 1 Rs were readily labeled at 4°C with anti-HA antibodies suggesting that LPA 1 Rs are transported to the plasma membrane in both MEF WT and MEF KO cells (Fig. 6,  untreated, left panels). In the absence of stimuli, little to no anti-HA antibody was internalized in either MEF WT or MEF KO cells. Both LPA and PMA stimulation for 30 min induced LPA 1 R internalization in MEF WT cells. LPA-induced internal- Surface receptors were labeled with mouse anti-HA antibody (Covance) at 4°C for 1 h, and the cells were either left at 4°C or transferred to 37°C in the presence or absence of either 10 M LPA or 1 M PMA for 30 min at 37°C. The remaining surface-bound antibodies were removed by mild acid wash, and the cells were fixed and processed for immunofluorescence staining. A schematic diagram of the cytoplasmic tail is shown above the micrographs with the deletion mutants indicated. B, internalized anti-HA antibodies were quantified using Simple PCI software as described under "Materials and Methods." Internalization was expressed as a percent of the total surface-bound antibody of the 4°C untreated control. *, p Ͻ 0.05; **, p Ͻ 0.01 relative to WT LPA 1 R-expressing cells. ization of LPA 1 Rs was blocked in MEF KO cells (Fig. 6), as we have previously shown (7). Remarkably, PMA stimulation was able to induce robust LPA 1 R internalization in MEF KO cells. Taken together, these results indicate that PMA-induced internalization of LPA 1 Rs is independent of ␤-arrestins.
Basal and PMA-induced Endocytosis of LPA 1 Rs Is Sensitive to siRNA Depletion of Clathrin AP-2-Several studies have shown that tyrosine-and dileucine-based motifs can interact with the clathrin adaptor protein AP-2 and mediate internalization of receptors (32). We have shown that PMA-induced internalization of the LPA 1 Rs is ␤-arrestin-independent but requires a dileucine-based motif in the cytoplasmic tail (Figs. 4 -6). Therefore, we asked whether clathrin AP-2 was an alternate adaptor protein for PMA-induced internalization of LPA 1 Rs. We used siRNA to reduce AP-2 protein levels in HeLa cells and to determine its effects on LPA 1 R internalization. Treatment of cells with siAP-2 but not siControl RNA reduced the abundance of AP-2 by more than 70% (Fig. 7B). Immunofluorescence labeling confirmed the reduction of AP-2 in cells (Fig.  7A). We tested the effects of siAP-2 or siControl RNA on anti-HA antibody internalization (Fig. 7C). Unstimulated cells exhibited a 16% basal internalization of LPA 1 R, which was inhibited by ϳ85% after siAP-2 treatment (Fig. 7C, untreated). Likewise, cells treated with PMA showed an ϳ28% internalization, which was also inhibited by ϳ86% following siAP-2 treatment. LPA stimulation led to ϳ33% internalization, and siAP-2 treatment led to a partial inhibition of LPA 1 R internalization (ϳ50% inhibition). These results indicated that both basal and PMA-induced internalization of LPA 1 Rs was strongly inhibited by depletion of AP-2, whereas LPA-induced internalization was only partially inhibited.

DISCUSSION
The endocytosis and intracellular trafficking of GPCRs is an important mechanism by which cells modulate their responsiveness toward a variety of extracellular stimuli. Studies of a variety of GPCRs have identified specific amino acid sequences in the cytoplasmically exposed domains of GPCRs that are critical for receptor endocytosis (33,34).
In this study, we investigated role of sequences within the cytoplasmic tail of the lysophosphatidic acid receptor, LPA 1 R, on its basal, LPA-induced, and PMA-induced endocytosis. LPA-induced internalization and signal attenuation was dependent upon ␤-arrestins and upon a serine-rich domain in the cytoplasmic tail. This domain was also essential for ␤-arrestin interaction with LPA 1 Rs. In contrast, both basal and PMAinduced endocytosis were dependent upon clathrin AP-2 adaptors and a di-leucine sequence, distal to the serine-rich domain. Unlike LPA stimulation, PMA stimulation did not induce the recruitment of ␤-arrestins from the cytoplasm to the plasma membrane. Furthermore, PMA-induced internalization was not dependent upon ␤-arrestins. These findings indicate that distinct mechanisms regulate basal and PMA-induced versus LPA-induced endocytosis of LPA 1 Rs.
Molecular Determinants of Agonist-induced LPA 1 R Endocytosis-Inspection of the cytoplasmic tail of LPA 1 Rs indicated the presence of several distinct amino acid motifs as follows: a di-cysteine palmitoylation motif, a serine-rich domain, a di-leucine motif, and a terminal PDZ-binding domain ( Fig. 2A). Analysis of a panel of LPA 1 R truncation mutants indicated that LPA-induced internalization of LPA 1 Rs required the serinerich domain but not the di-leucine motif or the PDZ-binding domain (Fig. 4). The serine-rich domain was also required for LPA-induced recruitment of ␤-arrestin to the plasma membrane (Fig. 3). Interestingly, progressive truncation of the LPA 1 R from the C-terminal PDZ binding domain toward the serine-rich domain led to a corresponding increase in the magnitude of LPA-induced PI hydrolysis (Fig. 2B). We hypothesize that efficient ␤-arrestin interaction with LPA 1 Rs requires both the serine-rich domain and additional amino acids distal to this region. This serine-rich domain ( 341 SDRSASS 347 ) bears a resemblance to the serine clusters found in GPCRs that form stable interactions with ␤-arrestins, which endure even upon delivery to endosomes (30). However, the interaction between LPA 1 Rs and ␤-arrestins appears to be transient as ␤-arrestins are recruited to the plasma membrane by LPA 1 Rs following a 2-min stimulation with LPA but return to the cytoplasm by 30 min, a time when LPA 1 Rs localize to endosomes. These results are consistent with the well established model for ligand-induced GPCR internalization where G protein receptor kinases phosphorylate specific serine or threonine residues within the third intracellular loop or cytoplasmic tail of GPCRs, which in turn promotes the binding of ␤-arrestins (35). ␤-Arrestin binding attenuates G protein signaling and targets the GPCR to clathrin-coated pits for endocytosis.
We have previously shown that both clathrin and ␤-arrestins are critical for agonist-induced LPA 1 R endocytosis (7). ␤-Arrestin binds directly to clathrin heavy chain and to clathrin AP-2 (8). siRNA knockdown of clathrin heavy chain inhibited LPA 1 R internalization by more than 90% (7). However, we found that siRNA knockdown of clathrin AP-2 only inhibited agonist-induced LPA 1 R internalization by ϳ50% (Fig. 7). Previous studies have shown that only a subset of all clathrincoated pits contain AP-2 (26). Alternative adaptor proteins such as AP-180 and CALM also mediate clathrin-coated pit assembly (36). Thus, our results indicating only a partial inhibition of LPA 1 R internalization upon knockdown of AP-2 is consistent with a partial depletion of clathrin-coated pits. We hypothesize that the remaining AP-2-independent clathrincoated pits mediate the agonist-induced LPA 1 R internalization by virtue of the ␤-arrestin-clathrin interaction.
PMA Stimulation of LPA 1 R Endocytosis-Avendano-Vazquez et al., (2005) previously showed that, in addition to LPA, PMA stimulated the phosphorylation and endocytosis of LPA 1 Rs in rat hepatocytes (27). We confirmed these observations in HeLa cells, where PMA also stimulated the endocytosis of LPA 1 Rs. Similar to Avendano-Vazquez et al. ((27), we found that bisindolylmaleimide I also inhibited PMA-induced internalization of LPA 1 Rs. 3 We further investigated the role of PKC by using the inhibitors Gö 6976, which inhibits conventional PKC isoforms and inhibits PKD/PKC, but not by Gö 6983, which inhibits conventional and novel PKC members but not PKD/PKC (29). Surprisingly, we found that PMA-induced internalization was inhibited by Gö 6976 but not by Gö 6983 (Fig. 1D). This suggests that PKD/PKC may be important for PMA-induced LPA 1 R internalization; however, its precise role in LPA 1 R trafficking will require additional studies. Interestingly, PKD/ PKC is activated by LPA signaling in intestinal epithelial cells via a pertussis toxin-sensitive pathway (37). PKD/PKC is known to regulate trans-Golgi network-to-plasma membrane transport, by promoting the fission of transport vesicles, and is implicated in the membrane recycling of integrins (38,39).
Phorbol esters stimulate the production of LPA in cervical and ovarian cancer cells (40). Thus, it is a formal possibility that PMA-induced internalization of LPA 1 R is simply because of the stimulation of LPA secretion by cells, which in turn is expected to bind to LPA 1 Rs and stimulate internalization. However, if this scenario were true, then one would expect that PMA-induced internalization exhibits the same requirements as LPAinduced internalization (e.g. ␤-arrestin dependence, amino acid motifs, and sensitivity to AP-2 depletion). Because our data indicate that there are distinct differences in the requirements for LPAversus PMA-stimulated internalization of LPA 1 Rs (Figs. 4 -7), we believe that LPA and PMA promote LPA 1 R internalization through distinct mechanisms.
Several studies have shown a role for PKC in heterologous desensitization and internalization of GPCRs such as purinergic P2Y1 receptors, sphingosine 1-phosphate binding S1P1 receptors, and gastrin-releasing peptide receptors (10,(41)(42)(43). In the case of S1P1 receptors, which belong to the lysophospholipid-binding family of GPCRs, both G-protein receptor kinasedependent and PKC-dependent mechanisms regulate the agonist-dependent and agonist-independent endocytosis of these receptors, respectively (10). Our studies show a second member of the lysophospholipid-binding family of GPCRs, the LPA 1 receptor, also utilizes distinct mechanisms for agonistdependent and agonist-independent internalization.
Molecular Determinants of Basal and PMA-induced LPA 1 R Endocytosis-Surprisingly, we found that both basal and PMAinduced endocytosis of LPA 1 Rs required the distal di-leucine motif (Fig. 4). We also found that PMA did not promote ␤-arrestin recruitment to the plasma membrane and that PMA stimulated LPA 1 R endocytosis in ␤-arrestin1/2 double knockout MEFs (Figs. 5 and 6). Thus, we conclude that PMA-induced endocytosis of LPA 1 Rs is independent of ␤-arrestins.
Many receptors that utilize di-leucine-based internalization motifs directly bind to clathrin AP-2 adaptor complexes (15,44). GPCRs such as the human cytomegalovirus-US28 receptor, which uses a di-leucine internalization motif, and ␣ 1B -adrenergic receptors, which does not use a di-leucine motif, directly bind to AP-2 adaptors during their endocytosis (23,45). In addition, the thrombin receptor PAR1 has been shown to bind to AP-2 during thrombin-activated endocytosis; however, the interaction with AP-2 is mediated by a tyrosine-based endocytic motif (15).
We hypothesized that clathrin AP-2 may serve as an alternate adaptor to facilitate PMA-induced endocytosis of LPA 1 Rs. Indeed, depletion of AP-2 through siRNA knockdown strongly inhibited PMA-induced endocytosis of LPA 1 Rs (Fig. 7). Surprisingly, AP-2 depletion strongly inhibited basal endocytosis of LPA 1 Rs as well, suggesting that basal and PMA-induced endocytosis may utilize similar endocytic mechanisms.
Why do LPA 1 Rs utilize two different adaptor complexes for endocytosis? During LPA-stimulated internalization, ␤-arrestin binding serves to attenuate receptor signaling and secondarily promotes clathrin-dependent endocytosis. However, during basal or PMA-induced endocytosis, ␤-arrestin does not engage LPA 1 Rs. Because ␤-arrestin binding is preceded by G-protein receptor kinase phosphorylation of agonist-bound receptors, it is likely that basal and PMA-stimulated endocytosis is more efficiently facilitated through the direct interaction between the di-leucine motif of LPA 1 Rs and clathrin AP-2.
We have shown here that distinct amino acid motifs within the cytoplasmic tail of LPA 1 Rs as well as distinct clathrin adaptor proteins are responsible for agonist-dependent and -independent endocytosis. These observations reveal an additional layer of regulation for LPA 1 Rs. Future studies should provide details about the functional significance that these two endocytic mechanisms may provide to LPA receptor signaling.