Structural and Functional Requirements for Agonist-induced Internalization of the Human Platelet-activating Factor Receptor*

The receptor for platelet-activating factor (PAF) is a member of the G-protein-coupled receptor family. To study the structural elements and mechanisms involved in the internalization of human PAF receptor (hPAFR), we used the following mutants: a truncated mutant in the C-terminal tail of the receptor (Cys317→ Stop) and mutations in the (D/N)P(X)2,3Y motif (Asp289 → Asn,Ala and Tyr293 → Phe,Ala). Chinese hamster ovary cells expressing the Cys317→ Stop mutant exhibited a marked reduction in their capacity to internalize PAF, suggesting the existence of determinants important for endocytosis in the last 26 amino acids of the cytoplasmic tail. Substitution of Asp289 to alanine abolished both internalization and G-protein coupling, whereas substitution of Tyr293 to alanine abolished coupling but not internalization. Inhibition or activation of protein kinase C did not significantly affect the internalization process. Receptor sequestration and ligand uptake was, at least in part, blocked by concanavalin A and blockers of endocytosis mediated by clathrin-coated pits. Our data suggest that the internalization of a G-protein-coupled receptor and coupling to a G-protein can be two independent events. Moreover, the C terminus tail of hPAFR, but not the putative internalization motifs, may be involved in the internalization of hPAFR.

Platelet-activating factor (PAF) 1 is a potent phospholipid mediator that produces a wide range of biological responses through activation of a specific receptor on target cell surface (1,2). The PAF receptor is part of the large family of seven transmembrane domain receptors (3)(4)(5)(6)(7). These receptors, when exposed to an agonist, activate a heterotrimeric G-protein that in turn activates and regulates events leading to cellular responses. Following repeated or prolonged agonist stimulation, their signal transduction potential becomes modulated and attenuated, a phenomenon termed desensitization (8). Three distinct events participate in the process of desensitization as follows: the functional uncoupling of the receptor from the G-protein, internalization or sequestration of the receptor from the cell surface toward the intracellular compartments, and finally, the down-regulation of receptor numbers, involving the degradation of the protein in lysosomes (9). The role of phosphorylation in receptor desensitization has been widely studied. It seems clear, at this time, that phosphorylation by second messenger kinases (PKC, PKA, PKG) and the GRKs (G-protein-coupled receptor kinases) helps to maintain the receptors uncoupled from the G-protein (10 -13). However, the role of phosphorylation in receptor sequestration and down-regulation is less defined and seems to vary from receptor to receptor (14,15).
Receptor internalization permits ligand entry and seems to participate in resensitization as well as desensitization (16 -20). The internalization process can be mediated by clathrincoated vesicles for some receptors (17,21), but not for others (22,23), and a phosphorylation step may or may not be essential (24). The structural elements involved in this phenomenon can be found in the intracellular loops (25)(26)(27) as well as the C-terminal tail (28 -32) of the individual receptors. No consensus motif has as yet been found, although both positive (28, 30 -32) and negative (29,33) regulatory elements have been identified for some receptors. The motif (D/N)P(X) 2,3 Y is found at the end of the 7th transmembrane domain of most receptors of this family and seems involved in the internalization process of some (17,34) but not all (35)(36)(37) receptors. In addition, the substitution of residues in this motif was found to affect other functions of the receptor, such as G-protein coupling, affinity for the ligand (37,38), and even the phosphorylation level of the receptor. These results suggest a global role for this region in the conservation of adequate receptor conformation. It has also been shown that the medium chains ( 1 and 2 ) of clathrinassociated protein complexes AP-1 and AP-2 can specifically interact with a tyrosine-based signal YXX⌽ (⌽ is an amino acid with a bulky hydrophobic side chain) motif (39). Both these motifs are found in the sequence DPVIYCFL present in the hPAFR.
It has been shown that the PAFR undergoes a ligand-specific, temperature-dependent internalization in transfected cells (40). The aim of our study was to identify the structural elements and mechanisms involved in the internalization of the hPAFR. The role of the two motifs (D/N)P(X) 2,3 Y and YXX⌽ was evaluated with the mutant receptors Asp 289 3 Ala,Asn (41) and Tyr 293 3 Ala,Phe (this study). The presence of regulatory elements in the cytoplasmic tail was verified with the mutant Cys 317 3 Stop that does not contain the last 26 amino acids of the C-terminal tail. The involvement of clathrin-coated vesicles in hPAFR internalization was addressed with internalization blockers. Finally, to determine whether signalization was an important step in the initiation or regulation of the hPAFR internalization process, certain inhibitors and an activator of PKC, in addition to uncoupled mutants of the receptor, D63N (43), A230E (44), D289A (41), and Y293A (this study), were used.

EXPERIMENTAL PROCEDURES
Materials-Reagents were obtained from the following sources: oligonucleotides were synthesized at Life Technologies, Inc., Pwo polymerase was from Boehringer Mannheim, restriction endonucleases and T4 DNA ligase were from Promega and Pharmacia Biotech Inc., bovine serum albumin, 1-(5-isoquinolinylsulfonyl)-2-methylpiperazine, pertussis toxin, phenylarsine oxide, and concanavalin A were from Sigma, staurosporine and calphostin C were from BIOMOL Research Laboratories, Inc., NH 4 Cl and sucrose were from Fisher, lipid-free bovine serum albumin and hexadecyl-PAF were from Calbiochem, AG1-X8 resin (in formate form, 100 -200 mesh) was from Bio-Rad, cell culture media and LipofectAMINE were from Life Technologies, Inc., Construction of Epitope-tagged PAF Receptor cDNA and Mutants-A tagged hPAFR cDNA was generated by polymerase chain reaction (46) from Kp132 using the oligonucleotide 5Ј-CCACATGACTCCTCCCA-CATG-3Ј and the M13 sequencing primer (5Ј-GTAAAACGACGGC-CAGT-3Ј). The resulting fragment was then digested with Acc65I and subcloned into the Eco1CR1-Acc65I sites of the pJ3M vector. In this construction, the N-terminal initiator methionine was replaced by the peptide sequence MEQKLISEEDLSRGSPG, resulting in a c-myc epitope-tagged PAF receptor coding sequence. Mutated receptors were constructed by polymerase chain reaction using Kp132 as template. Tyr 293 3 Phe and Tyr 293 3 Ala substitutions were created using the oligonucleotides 5Ј-TGTTATCTTCTGTTTCC-3Ј and 5Ј-CCTGTTA-TCGCCTGTTTCC-3Ј with their reverse complements, respectively. The point mutation was then introduced in the epitope-tagged receptor coding sequence using BstEII and Acc65I restriction enzymes. Truncated forms, Cys 317 3 Stop and Lys 298 3 Stop, of the receptor were also constructed. To construct these mutants the polymerase chain reaction product generated with the oligonucleotides 5Ј-CCACATGACTCCTC-CCACATG-3Ј and 5Ј-GCCCGGGATCATTTCCGG-3Ј (Cys 317 3 Stop) or 5Ј-CCAATTCTAGGTGAGGAAAC-3Ј (Lys 298 3 Stop) were subcloned in the site Eco1CR1 of pJ3M vector. Mutations and the integrity of the coding sequence were confirmed by dideoxy sequencing (University of Alberta, Alberta, Canada).
Cell Culture and Transfections-COS-7 and CHO cells were grown in Dulbecco's modified Eagle's medium high glucose and Dulbecco's modified Eagle's medium F12, respectively, supplemented with 10% fetal bovine serum. Cells were plated in 30-mm dishes (2.0 ϫ 10 5 COS-7 cells/dish or 3.0 ϫ 10 5 CHO cells/dish), transiently transfected with the constructions encoding the WT and the mutant receptors using 4 l of LipofectAMINE and 1 g of DNA per dish and harvested 48 h after transfection. For studies of biphasic isotherms, 100-mm dishes with 3 ϫ 10 6 CHO cells/dish were transfected with constructions encoding the WT and Y293A mutant using 30 l of LipofectAMINE and 4 g of DNA.
Radioligand Binding Assay-Competition binding curves were done on COS-7 cells expressing the WT and mutant receptor species. Cells were harvested and washed twice in Hepes-Tyrode's buffer (140 mM NaCl, 2.7 mM KCl, 1 mM CaCl 2 , 12 mM NaHCO 3 , 5.6 mM D-glucose, 0.49 mM MgCl 2 , 0.37 mM NaH 2 PO 4 , 25 mM Hepes, pH 7.4) containing 0.1% (w/v) bovine serum albumin (BSA) (47). Binding reactions were carried out on 5 ϫ 10 4 cells in a total volume of 0.25 ml in the same buffer with 10 nM 3 H-WEB2086 and increasing concentrations of nonradioactive WEB2086 or PAF for 90 min at 25°C. Reactions were stopped by centrifugation. The cell-associated radioactivity was measured by liquid scintillation.
Inositol Phosphate Determination-COS-7 cells were transfected as described above with the WT or mutant receptors and labeled the following day for 18 -24 h with myo-[ 3 H]inositol at 5 Ci/ml in Dulbecco's modified Eagle's medium (high glucose, without inositol). After labeling, cells were washed once in phosphate-buffered saline (PBS), pretreated or not with the indicated inhibitors (concanavalin A (0.25 mg/ml, 20 min), sucrose (0.45 M, 20 min), NH 4 Cl (10 mM, 10 min), phenylarsine oxide (80 M, 5 min with 30-min rest)) and preincubated 5 min in PBS at 37°C. At the end of this preincubation period, the PBS was removed, and cells were incubated in pre-warmed Dulbecco's modified Eagle's medium (high glucose, without inositol) containing 0.1% BSA and 20 mM LiCl for 5 min. Cells were then stimulated for 30 s with indicated concentrations of PAF. The reactions were terminated with the addition of perchloric acid followed by a 30-min incubation on ice. Inositol phosphates were extracted (48) and separated on Dowex AG1-X8 columns (49). Total labeled inositol phosphates were then counted by liquid scintillation.
Ligand Internalization-The evaluation of ligand internalization kinetics was done on CHO cells transiently transfected in 12-well dishes with constructions encoding mutant and WT receptors. After the incubation period, cells were washed two times with 1 ml of the same buffer but containing 2% BSA. Cells were then lysed in 0.1 N NaOH, and internalized radioactivity was evaluated by liquid scintillation.
Confocal Microscopy-Stably transfected CHO cells were grown on coverslips (25 mm). The cells were incubated in the presence or absence of sucrose and/or PAF (0.5 M) for 20 min and fixed with 4% paraformaldehyde (15 min at RT). The coverslips were then placed in 0.1% Triton (20 min RT) and then sequentially incubated with 5% milk (30 min RT) and 0.1% glycine (60 min RT). The cells were then incubated with anti-PAFR antibodies (polyclonal, directed at the C-terminal tail; Cayman Chemical, Ann Arbor, MI) followed with rhodamine-conjugated goat anti-rabbit antibodies (Bio/Can Scientific, Mississauga, Ontario). The cells were then analyzed on a Molecular Dynamics (Sunnyvale, CA) Multi-Probe 2001 confocal argon laser scanning system equipped with a Nikon Diaphot epifluorescence inverted microscope. Scanned images FIG. 1. Putative seven-transmembrane segment topography of the PAF receptor. Solid circles indicate the amino acids of the hPAFR that were mutated for this study (Tyr 293 ) and previous studies (Asp 63 , Ala 230 , and Asp 289 ). The substituting residues are illustrated. The mutants characterized in previous studies are identified by an asterisk (41,43,44). Deletion mutants (Lys 298 3 Stop and Cys 317 3 Stop) are indicated by an arrow.
were transferred onto a Silicon Graphics Indy 4000 workstation equipped with Molecular Dynamics' Imagespace analysis software.

Functional Characterization of the Mutant Receptors-
The mutants used in the present report are illustrated in Fig. 1. The deletion mutants, Cys 317 3 Stop and Lys 298 3 Stop, were made to analyze the involvement of the cytoplasmic tail in the internalization of the receptor. The tyrosine 293 putatively involved in the motifs (N/D)P(X) 2,3 Y and YXX⌽ was substituted with phenylalanine and alanine. The role of receptormediated signal transduction in internalization was studied with the help of G-protein-uncoupled mutants that were distributed in distinct areas of the receptor (D63N, A230E, D289A, and Y293A).
The binding characteristics of the WT receptor and the mutants, Cys 317 3 Stop, Y293A, and Y293F, were examined in transiently transfected COS-7 cells using the agonist (PAF) and an antagonist (WEB2086) (Fig. 2). The Lys 298 3 Stop mutant was not illustrated as it did not bind either of the ligands, despite comparable cell-surface expression to the WT construct, as examined by cytofluorimetry with an anti-c-myc antibody (results not shown). No significant difference was found in the affinities for WEB2086 ( Fig. 2A) or PAF (Fig. 2B) between the different mutants and the WT receptor. Similar levels of expression were also obtained in COS-7 as well as in CHO cells ( To further characterize these receptors, the signaling response of the mutant (Cys 317 3 Stop, Y293A, and Y293F) and WT receptors was measured by IP accumulation (Fig. 3A). In COS-7 cells, hPAFR-induced IP production would be mediated by PLC␤1 activation via G q/11 as the IP production is not blocked by pertussis toxin or tyrosine kinase inhibitors (44). Transiently transfected COS-7 cells were exposed for 30 s to graded concentrations of PAF, and total IP accumulation was measured. As shown, the Cys 317 3 Stop deletion of the cytoplasmic tail did not impair the response of the receptor. Since nearly half the potential phosphorylation sites of the cytoplasmic tail were eliminated by this deletion, the slightly augmented response of this mutant could reflect the role played by these residues in controlling the response to PAF. On the other hand, the Y293A mutant did not generate an IP response, except at very high concentrations of PAF (1 M). In contrast, the substitution of Y293F had no significant effect on the response of the receptor.
When conditions of transfection are such that low levels of hPAFR receptors are expressed (44), receptors can be observed in both a G-protein-coupled state (high affinity) and uncoupled state (low affinity). As shown in Fig. 3B, under these conditions of receptor expression, the mutant Y293A appears to exist only in its uncoupled, low affinity state, when compared with the WT, which might explain its inability to activate PLC after PAF stimulation.
Characterization of the Internalization Pathways-The effects of different inhibitors of internalization on receptor sequestration and endocytosis of 3 H-PAF are illustrated in Fig. 4. After a 20-min stimulation of stably transfected hPAFR-CHO cells with 0.5 M PAF, sequestration was studied by confocal microscopy with an antibody raised against a peptide sequence of the C-terminal portion of the hPAFR (Fig. 4A). Cells were pretreated with medium (a and c) or with hyperosmolar concentration of sucrose (b and d) to block clathrin-mediated internalization before stimulation with PAF (c and d). PAF induced the translocation of receptors toward the intracellular compartment (c), whereas sucrose blocked the process of hPAFR sequestration (d).
In addition, sucrose as well as other inhibitors of clathrinmediated endocytosis significantly blocked the internalization of 3 H-PAF (Fig. 4B), indicating that receptor-specific endocytosis of PAF is mediated via clathrin-coated vesicles. Fig. 4C shows that inhibition of PAFR endocytosis was not a nonspecific effect of the inhibitors on ligand binding or receptor signal transduction. The level of IP accumulation was measured after PAF stimulation in the presence of the inhibitors, which, except for phenylarsine oxide, did not significantly inhibit PAFR response. In fact, the apparent decrease in PAF-induced response in cells pretreated with phenylarsine oxide was due to enhanced basal levels of IP production.
The Role of Signal Transduction in hPAFR Internalization-The different mutants were compared with the WT receptor for their capacity to internalize 3 H-PAF (Fig. 5). The Lys 298 3 Stop mutant was used as a control for non-receptor-mediated 3 H-PAF internalization since it did not detectably bind PAF. The results indicate that the majority of mutants that are uncoupled from G-proteins (Fig. 5A) were impaired in their capacity to internalize PAF, except for the Y293A mutant. This mutant indicated that receptor signalization does not participate in ligand-mediated internalization of hPAFR. Although PAF-induced mitogen-activated protein kinase activation in guinea pig PAFR-transfected CHO cells was reported to be pertussis toxin-sensitive (47), internalization of human PAFR in stably transfected CHO cells was not affected by pertussis FIG. 4

. Effect of inhibitors of internalization on hPAFR sequestration (A), 3 H-PAF endocytosis (B), and response (C).
CHO cells stably transfected with the WT receptor were used for the following studies. A, cells were incubated with medium (a and c) or with sucrose (b and d) and then exposed (c and d) to 0.5 M PAF, 20 min at 37°C before fixation and permeabilization; sequestration of hPAFR was then revealed in confocal microscopy with an anti-PAFR antibody and a rhodamine-conjugated goat anti-rabbit antibody, as described under "Experimental Procedures." B, cells were incubated with 2 nM 3 H-PAF for 30 min at 37°C and washed in the presence of 2% BSA to remove the extracellular PAF. 3 H-PAF internalization was determined after sub-traction of nonspecific values that were obtained in the presence of 5 M PAF. The effect of inhibitors on IP production was examined in COS-7 cell transiently transfected with the WT receptor (C). Cells were stimulated with PAF 100 nM for 30 s after a pretreatment with medium (control) or the inhibitors. IP quantification was as described under "Experimental Procedures." The results are representative of three independent experiments, each done in duplicate. toxin treatment (data not illustrated). In contrast, the Cys 317 3 Stop mutant was impaired in its capacity to mediate PAF internalization (Fig. 5B), despite its efficient signal transduction, indicating that the last 26 amino acids of the C-terminal tail could be important for internalization. As Cys 317 3 Stop is expressed less than the WT, we expressed the WT at different levels and compared the internalization. When the WT is expressed at 30% maximal level (comparable to Cys 317 3 Stop expression), its internalization is only 25% less than maximal WT internalization, whereas Cys 317 3 Stop internalizes 90% less than the WT. In addition, our results indicate (Fig. 5B) that the substitution of the residues Asp 289 or Tyr 293 , within the putative internalization motifs (N/D)P(X) 2,3 Y and YXX⌽, by conserved residues (Asn and Phe, respectively) partially decreased but did not abolish the potential for internalization.
Phosphorylation by PKC does not appear to be involved in the initiation or control of PAFR internalization. Modulation of PKC activity by inhibitors (calphostin C, staurosporine, and H7) or an activator, in its active 4␤-PMA or inactive 4␣-PMA form, did not affect PAF-induced internalization (Fig. 6). DISCUSSION In agreement with data obtained by Takano et al. (50), with a truncated guinea pig PAF receptor, our data indicate that the last 26 amino acids of the C-terminal end of the hPAFR are not required for functional coupling to a G-protein, as removal of this region yielded a receptor that was as efficient as the WT in mediating phospholipid hydrolysis. In contrast, as observed with our Lys 298 3 Stop mutant, a larger deletion in the cytoplasmic tail can confer a loss of binding capacity for both PAF and WEB2086. This loss of effector function could be the consequence of a structural change in the 7th transmembrane domain, as substitution of Asp 285 3 Ile, in this domain, leads to a similar phenotype (41). Our data indicate, however, that the intracellular C-terminal tail of the hPAFR may be involved in endocytosis of the receptor-ligand complex, as internalization of 3 H-PAF was dramatically decreased, although not completely suppressed, in the Cys 317 3 Stop mutant. These results indicate either that other parts of the receptor might also be involved in controlling the internalization process or that 3 H-PAF could be partially internalized by constitutive endocytosis; a similar deletion in the thyrotropin-releasing hormone receptor led to a constitutively internalized and recycled receptor (51). The role of different intracellular domains of G-proteincoupled receptors in triggering internalization has been demonstrated in several studies using mutagenesis. Some segments in the intracellular loops (25)(26)(27) as well as in the C-terminal domain of the yeast pheromone ␣-factor, thyrotropin, gastrin, parathyroid hormone, and angiotensin receptors were found to be specifically involved in the sequestration process (5,6,8,51), whereas deletion mutants of the ␤-adrenergic and muscarinic M1 receptors were internalized to a similar extent compared with the WT receptors (4,30).
The hPAFR contains two putative internalization motifs in its sequence 289 DPVIYCFL 296 at the terminus of the 7th transmembrane domain. The sequence (N/D)P(X) 2,3 Y was proposed as an internalization motif by analogy with the endocytic motif NPXY (52) found in low density lipoprotein, transferrin, epidermal growth factor, and insulin receptors. However, the role played by this motif is not as well defined in G-protein-coupled receptors since substitution of some of these residues may affect other functions such as G-protein coupling (37,38). We have shown that this is the case for the hPAFR with the substitutions of Asp 289 (41) and Tyr 293 to alanine. On the other hand, a substitution by their isosters Asn and Phe, respectively, barely affected the coupling and only slightly influenced the kinetics of internalization. Therefore it is the structure of the amino acid that is important, not its polarity or side chain charge, indicating that these residues may form hydrogen bonds with others in the vicinity. As already suggested (38), the residues of this motif seem to participate in the preservation of an adequate conformation of the receptor and do not directly participate in internalization. This conclusion is supported by the fact that the mutant Y293A allows for the endocytosis of 3 H-PAF just as efficiently as the WT receptor. The lack of effect of the Y293A substitution on internalization also indicates that the other putative internalization motif YXX⌽ does not have such a function in the hPAFR.
The molecular mechanisms responsible for receptor-mediated endocytosis are still poorly understood. The potential involvement of PLC activation in the internalization process was suggested following the discovery that the complex AP-2 could bind to phosphatidylinositol (4,5)bisphosphate (53) and inositol (1,4,5)trisphosphate (54). It has also been shown by pertussis toxin treatment that coupling and activation of the G-protein was important for the internalization of somatostatin receptors (55). Despite the fact that loss of functional coupling seems to correlate with a decreased internalization for other receptors (3,31) and for most of our uncoupled mutant receptors (D63N, A230E, and D289A), results obtained with pertussis toxin treatment and the Y293A mutant indicate that G-protein coupling and signalization do not participate in hPAFR internalization. Similar discrepancies between internalization and signalization were also observed for muscarinic receptors (56) and for the type 1 angiotensin II receptor (57).
Moreover, results with an activator and inhibitors of PKC show that phosphorylation by this kinase does not trigger or modulate the internalization process. However, for some receptors, phosphorylation by a second messenger-dependent kinase might play a role in this process. It has been reported for the neurotensin receptor that the inhibition of PLC by U-73122 impaired the agonist-induced internalization of this receptor in transfected CHO cells and in N1E-115 neuroblastoma cells (58). Phosphorylation could also increase the probability of a protein to become a substrate for GRK2 (59), and in certain circumstances, the action of these kinases could modulate the internalization process (41,60,61). In addition, some GRKs can be influenced by phosphorylation cascades, as GRK2 activity and translocation can be increased through direct activation of PKC (62). GRK1, -2, and -3 are cytoplasmic kinases that translocate to the plasma membrane after a stimulus-mediated receptor activation. The translocation and activation of GRK2 and -3 is dependent on their association with specific ␤␥ Gprotein subunits (63,64) via their pleckstrin homology domains (65), in contrast to GRK5 and -6 that are associated with the membrane. The actions of these kinases are therefore dependent on G-protein coupling, and this could explain why the internalization of some receptors (55), but not of others, like hPAFR and type 1 angiotensin II receptor (57), is enhanced by G-protein activation as this might depend on which GRK will phosphorylate the activated receptor in a given cell type.
Our study indicates that ligand-induced signalization and the motifs (N/D)P(X) 2,3 Y and YXX⌽ are not involved in the process of internalization, whereas the use of internalization inhibitors suggests the participation of clathrin-coated vesicles in hPAFR internalization. Recently, it has been shown that arrestins could play a role in internalization by acting as adapter proteins between the receptors and clathrin and that GRK-mediated phosphorylation could augment this interaction (59 -61). The possibility that this mechanism is involved in the internalization of hPAFR is presently under investigation in our laboratory.