Regulation of cysteinyl leukotriene type 1 receptor internalization and signaling.

Cysteinyl leukotrienes activate the cysteinyl leukotriene type 1 receptor (CysLT1R) to regulate numerous cell functions important in inflammatory processes and diseases such as asthma. Despite its physiologic importance, no studies to date have examined the regulation of CysLT1R signaling or trafficking. We have established model systems for analyzing recombinant human CysLT1R and found regulation of internalization and signaling of the CysLT1R to be unique among G protein-coupled receptors. Rapid and profound LTD4-stimulated internalization was observed for the wild type (WT) CysLT1R, whereas a C-terminal truncation mutant exhibited impaired internalization yet signaled robustly, suggesting a region within amino acids 310-321 as critical to internalization. Although overexpression of WT arrestins significantly increased WT CysLT1R internalization, expression of dominant-negative arrestins had minimal effects, and WT CysLT1R internalized in murine embryonic fibroblasts lacking both arrestin-2 and arrestin-3, suggesting that arrestins are not the primary physiologic regulators of CysLT1Rs. Instead, pharmacologic inhibition of protein kinase C (PKC) was shown to profoundly inhibit CysLT1R internalization while greatly increasing both phosphoinositide (PI) production and calcium mobilization stimulated by LTD4 yet had almost no effect on H1 histamine receptor internalization or signaling. Moreover, mutation of putative PKC phosphorylation sites within the CysLT1R C-tail (CysLT1RS(313-316)A) reduced receptor internalization, increased PI production and calcium mobilization by LTD4, and significantly attenuated the effects of PKC inhibition. These findings characterized the CysLT1R as the first G protein-coupled receptor identified to date in which PKC is the principal regulator of both rapid agonist-dependent internalization and rapid agonist-dependent desensitization.

The cysteinyl (Cys) 1 leukotriene (LT) LTC4, and its conversion products LTD4 and LTE4, regulate numerous cell and organ system functions (1,2). Most notably, CysLTs have been identified as important mediators of asthmatic attacks and asthma pathogenesis; they are potent bronchoconstrictors (3), and also seem important in modulating airway inflammation and remodeling (4). Despite the detection of specific LTD4 binding to guinea pig lung membranes in 1993 by Metters et al. (5), the cloning of a high affinity CysLT receptor was frustrated for years, until the ultimate reporting of the human CysLT type 1 receptor (CysLT1R) by Lynch et al. (6) and Saura et al. (7) in 1999. The CysLT1R is expressed in spleen, peripheral blood leukocytes, and airway smooth muscle, has nanomolar affinity for LTD4, and couples to the heterotrimeric G protein G q to promote calcium flux. LTC4 is also a full agonist of the CysLT1R but is 10 times less potent. A second CysLT receptor subtype, CysLT type 2 receptor, has recently been cloned (8,9). CysLT type 2 receptor is expressed in leukocytes, heart, and brain, and binds LTD4 and LTC4 with equal affinity. Although no specific CysLT type 2 receptor antagonists currently exist, CysLT1R antagonists have been established as effective antiasthma drugs (10,11).
Since the initial characterization of the CysLT1R only a handful of studies, focused primarily on pharmacologic properties, have been published examining this receptor. This lack of studies is due primarily to difficulties in expressing recombinant CysLT1R in mammalian cells (6,12). Although expression of recombinant CysLT1R in HEK 293T cells appears useful for high-throughput screening of potential CysLT1R ligands using an automated assay of intracellular calcium mobilization (7,9), features of the CysLT1R beyond basic pharmacologic receptorligand interactions remain uncharacterized.
Understanding the mechanisms by which the responsiveness of a given G protein-coupled receptor (GPCR) is regulated not only can provide insight into the functional impact of the receptor in physiologic and disease states but also can identify regulatory molecules as potential therapeutic targets (13)(14)(15). For example, the responsiveness of the ␤2-adrenergic receptor (␤ 2 AR) is tightly regulated by phosphorylation of the receptor by GPCR kinases (GRKs) and the subsequent binding of arres-tin proteins. GRK-mediated phosphorylation partially uncouples the ␤ 2 AR from G s and promotes the binding of arrestins to the receptor, which in turn sterically inhibits ␤ 2 AR-G s interaction while promoting receptor internalization into endocytotic vesicles. Sensitivity of the ␤ 2 AR to this mode of regulation may explain a differential efficacy of ␤-agonists among airway cells that influences airway function and response to therapy (16) and may contribute to the pathology of chronic heart failure, in which ␤ 2 AR hyporesponsiveness is associated with elevated GRK levels in cardiac myocytes (17). Importantly, cardiac expression of a GRK2 "minigene" that effectively inhibits GRK2 activity can reverse ␤ 2 AR hyporesponsiveness and the pathologic phenotype in animal models of heart failure (18 -20), thereby establishing the utility of targeting GRK2, and possibly other GPCR regulatory molecules, in disease therapy. Moreover, differential sensitivity to GRK/arrestin-mediated regulation seems to explain, in part, differences in the signaling capacity and functional effects among GPCRs in a given cell type. The prostaglandin E2 EP2 receptor, which is resistant to GRK-mediated phosphorylation and arrestin binding, is much more efficacious than the ␤ 2 AR in stimulating cAMP production in analyses of both recombinant and endogenous receptors (21). The enhanced signaling capacity of EP2 receptors in human airway smooth muscle likely contributes to the significantly greater effect of prostaglandin E2 (relative to ␤-agonists) in modulating growth, migration, and contraction of human airway smooth muscle (22)(23)(24).
In the current study, we have provided mechanistic insight into the regulatory features of the signaling and trafficking of the human CysLT1R. Results demonstrated that the CysLT1R undergoes rapid agonist-dependent internalization, yet, unlike most GPCRs characterized to date, this effect appeared largely GRK-and arrestin-independent. Instead, internalization and desensitization were most dramatically affected by PKC activity, characterizing the CysLT1R as the only GPCR examined to date in which both agonist-dependent internalization and desensitization are primarily PKC-dependent phenomena.
Sources of other reagents either are identified below or are from previously identified sources (21).
Generation of Receptor Constructs-More detailed descriptions of construct generation are provided in supplemental material. A signal sequence (26) was inserted upstream of the 3-FLAG cassette in pcDNA3-3FLAG, and this plasmid was used to generate all WT and mutant CysLT1R clones. The open reading frame encoding the human CysLT1R in HG55 was amplified by PCR and inserted in-frame immediately downstream of the 3-FLAG cassette. C-terminal truncation mutants CysLT1R321stop, CysLT1R309stop, and CysLT1R300stop, as well as CysLT1RS(313-316)A, in which serines 313, 315, and 316 were mutated to alanines, were similarly generated by PCR cloning with specific antisense primers. Generation of pcDNA3 plasmid encoding FLAG-tagged human H1 histamine receptor (H1 HR) in pcDNA3 was described previously (27).
The sequence encoding the CysLT1R open reading frame, the signal sequence motif, and the 3-FLAG epitope were PCR-amplified and cloned into the shuttle vector pAdEGI to make pAdEGI-CysLT1R. Recombinant adenovirus was generated by co-transfection of pAdEGI-CysLT1R DNA with 5 viral DNA as described previously (25). The resulting virus, AdEGI-CysLT1R, was plaque-purified, expanded, and purified by two rounds of cesium chloride density centrifugation followed by exhaustive dialysis against 10 mM Tris, pH 8.0, 140 mM NaCl, 1 mM MgCl 2 , and 5% w/v sucrose. Viral particle numbers were determined by dilution in 0.1% SDS followed by measuring absorbance at 260 nm, and viral titers were determined by plaque analysis on 911 cells. The particle to infective/particle ratio was between 25 and 45 for all of the viruses used. Sequence fidelity of all of the clones was verified by dideoxynucleotide sequencing.
Cell Culture, Transfection, Infection-HEK 293 cells and COS-1 cells were maintained in Dulbecco's modified Eagles medium (DMEM) supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin. Cells were plated into 60-mm dishes 1 day prior to transfection so that they were 65-75% confluent immediately prior to transfection. All transfections were carried out using 10 l of FuGENE 6 transfection reagent and 3 g of total DNA. Eighteen h after transfection, cells were harvested and plated on 6-or 24-well plates for subsequent immunoblot analysis, receptor sequestration ELISA, analysis of phosphoinositide (PI) production, or onto poly-L-lysine-coated coverslips for immunocytochemical analysis of receptor subcellular distribution. For transfection of MEFs with pcDNA3-FLAG␤ 2 AR, 20 l of Lipofectamine 2000 (Invitrogen) in 500 l of OptiMEM (Invitrogen) were mixed with 8 g of pcDNA3-FLAG␤ 2 AR, and the mixture was added dropwise to MEFs seeded in 60-mm dishes; the cells were passaged 24 h later into 24-well plates. For infection of MEFs using AdEGI-CysLT1R, MEFs were harvested with trypsin/EDTA, washed in phosphate-buffered saline, and resuspended in DMEM/10% fetal bovine serum medium lacking antibiotics. 1 ϫ 10 10 particles of CysLT1R adenovirus plus 5 ϫ 10 9 particles of receptor plasmid VgRXR adenovirus, or 5 ϫ 10 9 particles of VgRXR alone, were combined with 500,000 cells and were plated into 60-mm dishes. Fifteen h later, cells were again harvested, washed, and plated into 24-well plates in the presence of 10 M (final concentration) ponasterone to induce expression. Twenty-four h later, receptor internalization was assessed by ELISA as described below.
Receptor Internalization-Agonist-induced changes in cell surface receptor distribution were measured by ELISA. Forty-eight h after transfection, cells in 24-well plates were washed once and then incubated at 37°C for 30 min in plain DMEM with various inhibitors or vehicle (typically 0.03% Me 2 SO). After pretreatment, cells were stimulated with LTD4, histamine, isoproterenol, or vehicle (typically 0.05% EtOH) for 0 -30 min. The medium was aspirated, and cells were fixed with 3.7% formaldehyde for 10 min at room temperature, washed three times with Tris-buffered saline, and then blocked for 45 min with 1% bovine serum albumin in Tris-buffered saline. Cells were then incubated for 1 h with a 1:1200 dilution of an anti-Flag M2 antibody conjugated to alkaline phosphatase, washed three times with Trisbuffered saline, and then incubated with alkaline phosphatase substrate at 37°C until adequate color development was visible. Reactions were stopped by adding 0.1 ml from each sample well into 0.1 ml of 0.4 M NaOH, and the sample absorbances were read at 405 nm using a Bio-Rad microplate reader and Microplate Manager software.
Treatment of cells with 10 mM L-cysteine throughout the course of cell treatment with LTD4, in experiments assessing either receptor internalization or PI production, had no effect on results, suggesting that the metabolism of LTD4 in culture media was not of consequence. Experimental results were also qualitatively similar whether cells were maintained in serum-containing or serum-free medium for 18 h prior to acute pretreatment and treatment. Whenever possible, the effect of experimental variables (e.g. receptor mutations, pharmacologic inhibition) was tested in assays performed under optimally matched conditions (e.g. simultaneously performed using same passage or transfected cells, same prepared reagents) to minimize intra-and inter-experimental variability.

CysLT1R Regulation
Assay of Phosphoinositide Generation-Cells passaged into 24-well plates as described above were loaded with 2 Ci/ml myo-[ 3 H]inositol for 18 h. Cells were then washed briefly with phosphate-buffered saline, incubated in DMEM containing 5 mM LiCl plus various inhibitors or vehicle for 15 min, and then stimulated with LTD4, histamine, or vehicle for 30 min. Reactions were terminated by replacing medium with cold 20 mM formic acid. The inositol and PI fractions were separated by anion exchange chromatography, mixed with Ultima Gold and Ultima Flo AF scintillation fluid (Packard), respectively, and counted. To normalize for loading variability among wells, PI production was calculated as PI/(PI ϩ inositol), and values were reported as fold basal (vehicle-stimulated) production, as described previously (28).
Analysis of Agonist-induced Ca 2ϩ Flux-COS-1 cells grown on glass coverslips and transfected with WT CysLT1R, CysLT1RS(313-316)A, or H1 HR constructs were loaded with 10 M fura-2/AM, a ratiometric calcium indicator dye. All of the calcium measurements were carried out at 37°C in Hanks' balanced salt solution containing 10 mM HEPES, 11 mM glucose, 2.5 mM CaCl 2 , and 1.2 mM MgCl 2 (pH 7.4) using a dual excitation fluorescence photomultiplier system (IonOptix, Milton, MA). The ratio of intensities of fura-2 emissions at excitation wavelengths 340 and 380 nm was calculated every second, and the intracellular calcium concentration (nM) was calculated by extrapolation from a calibration curve, as described previously (29). Cells were pretreated for 15 min with either vehicle (0.03% Me 2 SO) or 10 M Bis I and then were stimulated with agonist (1 M LTD4 or 1 M histamine), and changes in the intracellular calcium concentrations were recorded for 3-4 min. Calcium flux upon agonist stimulation was calculated by subtracting the mean basal intracellular calcium concentration from that of the peak intracellular calcium concentration elicited by agonist. Comparable expression of WT CysLT1R and CysLT1RS(313-316)A was confirmed by the immunoblotting of lysates from cells plated in parallel in 6-well plates.
Analysis of PKC Translocation-COS-1 cells were transiently transfected with pcDNA3wtCysLT1R or vector control or were co-transfected with a construct encoding PKC␤-GFP plus either pcDNA3wtCysLT1R or pcDNA3 vector and subsequently passaged into 60-mm dishes. Cells were serum-starved and then stimulated for 20 min with vehicle, 100 nM LTD4, or 100 nM PMA. Cells were harvested by scraping them into 500 l of homogenization buffer (20 mM Tris, pH 7.5, 0.25 M sucrose, 10 mM EDTA, 2 mM EGTA, 10 M phenylmethylsulfonyl fluoride, 10 g/ml benzamidine, 10 g/ml aprotinin, and 10 g/ml leupeptin). Lysates were passed through a 26-gauge needle three times and were centrifuged at 100,000 ϫ g for 60 min. The resultant supernatant was saved as the cytosolic fraction, and the pellet (crude membrane fraction) was resuspended in 125 l of homogenization buffer containing 0.1% Triton. 25 l of each fraction were resolved by SDS-PAGE and immunoblots were probed with anti-PKC␣ or anti-GFP antibody.
Immunofluorescence Microscopy and Immunoblotting-Visualization of the agonist-induced changes in CysLT1R distribution was performed on cells plated on poly-L-lysine-coated coverslips using a Nikon Eclipse E800 fluorescence microscope, as described previously for EP receptors (21). Immunoblot analysis of WT and mutant CysLT1R expression was performed by harvesting cells and scraping them into Laemmli buffer (0.75 M Tris, pH 6.5, 5% ␤-mercaptoethanol, 10% glycerol, and 4% SDS). For samples generated for radioligand binding studies and analysis of membrane/cytosolic distribution of CysLT1R, cells were scraped into homogenization buffer (as described above) and homogenized with a Dounce homogenizer. To generate membrane and cytosolic fractions, whole cell lysates were centrifuged at 30,000 ϫ g for 30 min, and the supernatant was saved and the pellet resuspended in homogenization buffer. For radioligand binding studies, resuspended pellets of the 30,000 ϫ g fraction were centrifuged a second time and resuspended in binding buffer. For immunoblot analysis, protein samples were resolved by SDS-PAGE, transferred to nitrocellulose membrane, and then probed with the appropriate antibodies. Bands were visualized by enhanced chemiluminescence (Pierce).

RESULTS AND DISCUSSION
Initial attempts to generate a useful recombinant CysLT1R produced constructs that either expressed poorly or, when expressed, resulted in proteins that migrated at too small a size Samples were filtered onto Whatman GF/C filters and were washed, filter-bound radioactivity was assessed, and specific binding (spec) was calculated as described under "Experimental Procedures." C, PI production by recombinant CysLT1R. COS-1 cells expressing CysLT1R or H1 HR were plated into 24-well plates and loaded with [ 3 H]myoinositol for 18 h. Cells were washed and pretreated with 5 mM LiCl in DMEM for 15 min and then were stimulated with either 100 nM LTD4 or 10 M histamine for 30 min. Inositol fractions were subsequently resolved by ion exchange chromatography, and PI production was calculated as described under "Experimental Procedures." In untransfected cells, neither LTD4 nor histamine stimulated PI production beyond basal levels (data not shown). conc., concentration.

CysLT1R Regulation
(ϳ35 kDa), suggesting either folding problems or difficulties in processing through the endoplasmic reticulum and the Golgi (data not shown). Ultimately a construct encoding an N-terminal 3-FLAG epitope and an upstream signal sequence to facilitate endoplasmic reticulum processing (26) was generated that expressed well in both COS-1 and HEK 293 cells and migrated at the predicted size (Fig. 1A). Multiple bands ranging from ϳ45 to 52 kDa in size were detected (as well as a possible dimer at ϳ90 kDa), suggesting the existence of multiple glycosylation sites. Importantly, antibodies against the N-terminal FLAG epitope and the C terminus of the human CysLT1R identified the same size bands (see below). Radioligand binding analysis of this construct expressed in COS-1 revealed saturable binding of [ 3 H]LTD4 (B max ϭ 450 fmol/mg protein) at levels ϳ10fold higher than that reported for the original HG55 clone (6), yet with a similar K d value (0.4 nM) (Fig. 1B). Comparable levels of CysLT1R were expressed among groups in which additional constructs were co-expressed (see supplemental materials). In assays of agonist-stimulated PI production (Fig.  1C), signaling capacity of wt CysLT1R was robust (ϳ5-fold of basal (vehicle-stimulated)), albeit less than that mediated by the H1 histamine receptor (H1 HR).
Exposure of COS-1 cells expressing wt CysLT1R to LTD4 caused a rapid internalization of receptor as demonstrated by the loss of cell surface binding of anti-FLAG antibody in ELISA experiments ( Fig. 2A) and as suggested by punctate aggregation of receptor, visualized by immunocytochemistry, following the addition of LTD4 (Fig. 2B). In unstimulated cells, CysLT1R was visualized primarily at the plasma membrane, although some puncta were frequently observed, suggesting that the tendency of receptors to internalize in the absence of agonist. Co-expression of CysLT1R with either arrestin-2 or arrestin-3 increased LTD4-stimulated internalization (Fig. 2C), suggesting a role for arrestins in mediating agonist-dependent internalization similar to that demonstrated for the ␤ 2 AR and numerous other GPCRs (30). Co-expression of arrestin mutants (ARR2(R169E) or ARR2(R170E)), of which the binding to GPCRs is largely independent of receptor phosphorylation (21,31), was no more effective in increasing CysLT1R internalization than was the corresponding wild type arrestin construct.  (56)). Cells were passaged into 24-well plates, and CysLT1R internalization in response to a 30-min treatment with 100 nM LTD4 was assessed by ELISA. Data represent mean Ϯ S.E. values from six experiments. D, effect of arrestin expression on CysLT1R-mediated PI production. COS-1 cells were transfected and plated as described in C, and LTD4-stimulated PI production was assessed as described under "Experimental Procedures." Data represent mean Ϯ S.E. values from four experiments.
Co-expression of arrestin-2 or arrestin-3 also reduced LTD4stimulated PI production, suggesting the capacity of arrestins to promote uncoupling/desensitization as well as internalization of the CysLT1R (Fig. 2D).
In an attempt to establish structural regions of the CysLT1R important in LTD4-stimulated internalization and desensitization, we generated CysLT1R mutants in which the receptor C terminus was progressively truncated (Fig. 3). Although lacking canonical GRK phosphorylation sites (32), the CysLT1R C-tail contains multiple serine/threonine residues that might serve as phosphorylation sites. Constructs encoding CysLT1R, truncated after amino acid 321 or 309, were expressed in COS-1 cells (Fig. 3B), and agonist-stimulated internalization was assessed. Truncation at amino acid 321 (CysLT1R321stop) had no effect on LTD4-stimulated internalization (Fig. 3C). Truncation at amino acid 309 (CysLT1R309stop) resulted in a receptor that internalized ϳ20% compared with 44% for wt CysLT1R. However, LTD4-stimulated PI production mediated by CysLT1R309stop was slightly greater than that observed for wt CysLT1R; PI production by CysLT1R321stop was slightly lower (data not shown). These data implicate the region spanning amino acids 310 to 321 as important in mediating internalization of the CysLT1R.
A mutant lacking the entire cytosolic C-tail (CysLT1R300stop) was also generated and was shown to express at levels approximately one-half that of wt CysLT1R (estimated by immunoblot analysis and cell-surface fluorescence detected by ELISA). However, CysLT1R300stop did not internalize or stimulate PI production in response to LTD4 (data not shown), thus precluding any clear interpretation of data derived from this mutant.
Although the observed effects of co-expression of arrestin-2/3 suggest a role for GRKs and arrestins in CysLT1R regulation, the relevance of these findings is unclear given the supraphysiologic levels of arrestins expressed in this model. More definitive evidence of the role of arrestins in CysLT1R internalization and desensitization was therefore sought by examining the competitive effect of dominant-negative arrestin mutants on wt CysLT1R internalization in HEK 293 cells. The magnitude of FIG. 3. Role of C terminus in CysLT1R internalization. A, topography of human CysLT1R C terminus. FLAG-tagged mutants were generated by PCR cloning (as described under "Experimental Procedures") in which the C-tail of FLAG-CysLT1R was progressively truncated at amino acids 321 (t321), 309 (t309), and 300 (t300). B, immunoblot analysis of wt CysLT1R, CysLT1R321stop, and CysLT1R309stop expression. The CysLT1R300stop construct expressed but did not signal and was therefore not pursued. C, LTD4-stimulated internalization of wt CysLT1R (wt) and truncation mutants CysLT1R321stop (321stop) and Cys-LT1R309stop agonist-induced internalization of wt CysLT1R in HEK 293 cells was greater than that observed in COS-1 cells (ϳ60% loss of cell surface receptors compared with ϳ35-40% loss in COS-1 cells) and was also much higher than that observed for isoproterenol-stimulated ␤ 2 ARs (Fig. 4A). Surprisingly, expression of LIELD/F391A arrestin-2 (previously shown to bind GPCRs but not clathrin/AP-2, thus effectively inhibiting GPCR internalization mediated by endogenous arrestins) or R169E/LIELD/ F391A arrestin-2 (the same construct with high affinity for GPCRs independent of receptor phosphorylation state) (33) had only a small effect on wt CysLT1R internalization (ϳ15% inhibition by each construct), whereas significantly inhibiting ␤ 2 AR internalization. Co-expression of CysLT1R with the dynamin mutant DynK44A (34) only partially inhibited (ϳ50%) LTD4-stimulated internalization, whereas ␤ 2 AR internalization was completely inhibited, suggesting that CysLT1R inter- nalization may occur by both dynamin-dependent and dynamin-independent pathways. To further establish a lack of requirement for arrestins in LTD4-induced internalization, we examined the capacity of wt CysLT1R to internalize in MEF cultures derived from transgenic mice in which arrestin-2 and arrestin-3 expression was ablated (arr2/3Ϫ/Ϫ) (35). Although expression of wt CysLT1R via FuGENE-mediated transfection of pcDNA3-3FLAG CysLT1R proved unsuccessful in MEFs, conditions were established for expression of a human 3-FLAG CysLT1R using a recombinant adenovirus (see "Experimental Procedures") that enabled analysis of receptor internalization by ELISA. When expressed in arr2/3Ϫ/Ϫ MEFs, the CysLT1R internalized in response to LTD4 at a level (18% loss of cell surface receptors) comparable with that observed when expressed in MEFs derived from matched nontransgenic controls (ϳ20%, Fig. 4B). Conversely, agonist-stimulated ␤ 2 AR internalization was minimal in arr2/3Ϫ/Ϫ MEFs. Collectively, results depicted in Fig. 4 suggest that the actions of arrestins are not required to effect CysLT1R internalization and led us to consider a role for other regulatory molecules.
We therefore examined the effect of numerous protein kinase inhibitors on LTD4-stimulated internalization of wt CysLT1R expressed in both COS-1 and HEK 293 cells. Pretreatment with H89, wortmannin, U0126, and genistein all failed to affect the magnitude of CysLT1R internalization (Fig. 5A (COS-1) and B (HEK)). Collectively these data suggest no role of PKA, phosphoinositide 3-kinase, p42/p44 mitogen-activated protein kinase, or tyrosine kinases, respectively, in regulating CysLT1R internalization. However, the specific PKC inhibitor Bis I caused a profound inhibition of internalization in both cell types, whereas structurally similar Bis V, which lacks the ability to inhibit PKC (36) had no effect. Bis IX, as well as Ro-32-0433, the latter more selective for conventional PKC isoforms than is Bis I, were both similar in effect to Bis I (data not shown). By comparison, PKC inhibition had no effect on agonist-promoted internalization of the H1 HR (data not shown), similar to our previous findings (27). Translocation of the PKC␣ isoform from the cytosol to membrane fractions was evident in cells expressing CysLT1R stimulated with LTD4, although the effect was modest compared with that induced by stimulation with PMA (Fig. 5C). Because low transfection efficiency could limit the appreciation of CysLT1R-induced endogenous PKC isoform translocation, we co-expressed a commercially available PKC␤-GFP chimera with CysLT1R and assessed LTD4-stimulated translocation of this chimera. Under these conditions, CysLT1R activation can be shown to more strongly promote the translocation of this PKC isoform (Fig. 5D).
Favorable substrates for PKC include serine/threonine residues, preferably flanked by a lysine or arginine at positions Ϫ3, Ϫ2, or ϩ2 with a hydrophobic residue at ϩ1 (37), although this consensus is not an absolute requirement because protein secondary structure is also an important determinant. On the basis of our data demonstrating the resistance of the truncation mutant CysLT1R309stop to internalization, we generated a CysLT1R mutant (CysLT1RS(313-316)A) in which serines 313, 315, and 316 (representing putative PKC sites) were mutated to alanines. Comparison of this construct with wt CysLT1R revealed that the selected mutations caused an ϳ50% loss of maximal internalization in both COS-1 (Fig. 6, A  and B) and HEK 293 (Fig. 6, C and D) cells. Interestingly, the EC 50 was also slightly decreased for the CysLT1RS(313-316)A mutant in both cell types (see Fig. 6 legend). In addition, the effect of Bis I pretreatment on CysLT1RS(313-316)A was considerably less. PKC inhibition also dramatically influenced LTD4-stimulated PI production. Pretreatment with Bis I (Fig. 7, A and D) or Bis IX, but not Bis V or other AGC kinase or tyrosine kinase inhibitors (data not shown), significantly increased maximal LTD4-stimulated PI generation by wt CysLT1R while decreasing the EC 50 for LTD4 (see Fig. 7 legend). LTD4-stimulated PI production was 2-3-fold higher in CysLT1RS(313-316)A (Fig.  7, B and E) compared with that mediated by wt CysLT1R. Similar to our previous findings (27), only a very small effect of PKC inhibition was observed for histamine-stimulated PI production in cells expressing the H1 HR (Fig. 7, C and F), suggesting that PKC inhibition had minimal effect on either G q or phospholipase C and that the receptor was the locus of regulation by PKC in LTD4-stimulated PI production. This interpretation was further supported by the observation that PI production in cells expressing CysLT1RS(313-316)A was minimally affected by PKC inhibition (Fig. 7, B and E).
Consistent with these data characterizing PKC-dependent regulation of PI production, pretreatment of COS-1 cells expressing wt CysLT1R with Bis I caused a large increase in LTD4-stimulated peak Ca 2ϩ flux (Fig. 8). Conversely, Bis I pretreatment had only a small effect on peak Ca 2ϩ flux in cells expressing CysLT1RS(313-316)A and no effect on cells expressing H1 HR. Moreover, peak Ca 2ϩ flux mediated by CysLT1RS(313-316)A was significantly higher than that mediated by wt CysLT1R.
To explore the potential role of PKC in mediating heterologous (nonagonist-specific) CysLT1R regulation, CysLT1R internalization and signaling were examined in both COS-1 and HEK 293 cells. In COS-1 cells, treatment with 100 nM PMA for 30 min induced a small degree (ϳ10%) of internalization of both wt CysLT1R and CysLT1RS(313-316)A, similar to that observed for the H1 HR (Fig. 9). PMA pretreatment also caused a small (ϳ20%) inhibition of LTD4-stimulated PI production (suggesting heterologous desensitization) by the wt CysLT1R that was not observed for either CysLT1RS(313-316)A or H1 HR (Fig. 9C). The observed effects of PMA treatment on internalization and signaling were all partially reversed by Bis I pretreatment. Qualitatively similar results were obtained in HEK 293 cells (Fig. 9, B and D).
The role of second messenger-dependent kinases in GPCR regulation is frequently limited to the context of heterologous desensitization. Activation of second messenger kinases by other receptor-dependent or -independent pathways can promote the phosphorylation of a GPCR not occupied by its cognate ligand, and this phosphorylation can diminish subsequent agonist-stimulated receptor-G protein coupling. Many GPCRs exhibit this form of heterologous desensitization (38). For a handful of receptors including the CXCR4 (39), somatostatin 2A (40), and endothelial differention gene 1 (41) receptors, receptor phosphorylation by PKC seems sufficient to induce internalization of the receptor in the absence of agonist. For agonist-mediated internalization or desensitization, second messenger-dependent kinases typically play little if any role. For the overwhelming majority of GPCRs examined to date, GRK-mediated phosphorylation and arrestin binding are required for and are the principal determinants of agonist-promoted internalization and desensitization, as suggested by the effective inhibitory actions of dominant-negative GRKs and arrestins, and the lack of effect of various inhibitors of other kinases (42). For some GPCRs including the type 1 Angiotensin II (43), endothelial differentiation gene 1 (41), CXCR4 (44), chemokine receptor 5 (45), and somatostatin 2A (40,46) receptors, PKC appears to play a role in agonist-stimulated receptor phosphorylation. However, the effect of pharmacologic inhibition of PKC is typically only a small to moderate reduction in receptor phosphorylation with little if any effect on agonistinduced receptor desensitization and internalization. These results suggest that PKC-mediated phosphorylation is an innocuous or redundant action that follows receptor stimulation and that GRKs and arrestins are the principal regulators of agonist-occupied GPCRs.
Receptors in which second messenger kinases may play a more significant role in agonist-induced receptor internalization or desensitization are the secretin receptor (a class II GPCR coupled to G s ), the metabotropic glutamate receptor 1a (mGlu1a; a seven-transmembrane domain receptor belonging to a unique class of receptors that bear little sequence or structural homology to class 1 or class 2 GPCRs) and mouse thromboxane A2 receptor (TP). In HEK 293 cells, PKA inhibitors do not inhibit secretin receptor desensitization but reduce secretin-stimulated receptor phosphorylation by 50% (47). Interestingly, dominant-negative arrestins do not inhibit secretin receptor internalization, and GRK overexpression increases receptor internalization only in the presence of PKA inhibitors (48). Thus PKA seems to have some role in secretin receptor internalization without altering signaling.
PKC inhibition has been shown to inhibit both agonist-induced internalization and agonist-induced desensitization of the metabotropic glutamate receptor 1a, although both of these effects have been shown to be GRK-and arrestin-dependent (49 -51). Both PKC and GRKs/arrestins seem to play significant roles in agonist-induced desensitization of the G q -coupled mouse TP receptor and the human homologue thromboxane A2␣ receptor. The PKC inhibitor staurosporine partially reversed (ϳ40%) homologous desensitization of the mouse TP receptor and also markedly attenuated agonist-induced receptor phosphorylation (52). Dominant-negative GRKs also inhibited agonist-induced phosphorylation and enhanced TP signaling (53), demonstrating a significant role for GRKs as well in homologous TP receptor desensitization. Agonist-induced phosphorylation of the human TP receptor is also significantly reversed by PKC inhibition (54). Interestingly, the TP-␣ receptor does not exhibit agonist-induced internalization, except when co-expressed with arrestin-2 or arrestin-3 (55).
Unlike other class 1 GPCRs examined to date, the CysLT1R exhibits rapid agonist-induced internalization that is primarily dependent on PKC. Because the effects of dominant-negative arrestin overexpression and arrestin2/3 gene ablation are minimal on agonist-induced CysLT1R internalization, the mechanism by which PKC acts is dissociated from that involving GRKs/arrestins and therefore is novel among all GPCRs.
Moreover, PKC also appears to profoundly influence homologous desensitization, identifying CysLT1R as the only GPCR examined to date in which both agonist-induced internalization and agonist-induced desensitization are primarily PKCdependent and arrestin-independent. LTD4-stimulated PI production and calcium mobilization are greatly increased in the presence of Bis I in cells expressing wt CysLT1R but not in cells expressing either the CysLT1RS(313-316)A, which lacks putative PKC phosphorylation sites, or the H1 HR, which similarly couples to (apparently desensitization-resistant in COS-1 and HEK 293) G q and phospholipase C. Although the disparate effects of PKC inhibition on the regulation of wt CysLT1R versus CysLT1RS(313-316)A suggest that PKC-mediated phosphorylation of CysLT1R on residues within 313-316 mediates the effects of PKC on CysLT1R desensitization and internalization, it is important to note that this remains to be demonstrated directly. Our attempts to date to establish PKCdependent phosphorylation of the CysLT1R have been frustrated by an inability to express the receptor at levels sufficient to permit detection of 32 P incorporation in receptors immunoprecipitated from orthophosphate-loaded cells. Our experience is that to detect agonist-dependent phosphorylation of GPCRs, expression levels of at least 1 pmol/mg protein must be achieved. To date, we know of no group that has been able to achieve expression of the CysLT1R, in either cell lines or transgenic mice designed for overexpression (56), that approaches this level. Possible explanations for the failure to achieve such expression is either that the cellular synthetic machinery is constrained by inherent structural features of the receptor or that a high level of expression of CysLT1R may be toxic to the cell.
Of interest is the slight difference in magnitude of effect of PKC inhibition on CysLT1R internalization in COS-1 versus HEK 293 cells. In COS-1 cells, the effect of Bis I pretreatment on wt CysLT1R internalization is profound (Ͼ80% inhibition), whereas in HEK 293 cells the effect is less dramatic (ϳ50%) but still prominent. In both cell types, maximal internalization of the CysLT1RS(313-316)A mutant is low (ϳ50% of that of the wild type) and is only minimally affected by Bis I. Why Bis I has a greater effect in COS-1 cells is unclear. One possible explanation is the greater levels of arrestins (57) and GRK2 (58) in HEK 293 cells may favor a slightly greater role for GRK/arrestin-mediated regulation, consistent with the results of Fig. 2 demonstrating that increased arrestin-2 or arrestin-3 expression can augment CysLT1R internalization. Moreover, in both COS-1 and HEK 293 cells, CysLT1RS(313-316)A retains some ability to internalize that is not inhibited by Bis I (Fig. 6) yet is slightly inhibited by dominant-negative arrestin 2 (Fig. 9), suggesting that a non-PKC (arrestin?)-dependent mechanism may acquire significance under certain conditions. Collectively, these findings suggest the intriguing possibility that PKC and GRKs/arrestins may be competitive with respect to CysLT1R regulation, and their relative effects may depend on relative expression levels and subcellular distribution, the latter influenced by competition among intracellular substrates for kinase activity. Future studies examining CysLT1R regulation employing strategies similar to those utilized by Kohout et al. (35) to assess arrestin selectivity among GPCRs, as well as analyses of primary cell types relating endogenous kinase:arrestin:receptor stoichiometry and CysLT1R signaling, will help clarify the physiologic roles of PKC, GRKs, and arrestins in CysLT1R regulation.