Identification of Putative Sites of Interaction between the Human Formyl Peptide Receptor and G Protein*

Wild-type and 35 mutant formyl peptide receptors (FPRs) were stably expressed in Chinese hamster ovary cells. All cell surface-expressed mutant receptors bound N-formyl peptide with similar affinities as wild-type FPR, suggesting that the mutations did not affect the ligand-binding site. G protein coupling was examined by quantitative analysis ofN-formyl-methionyl-leucyl-phenylalanine-induced increase in binding of 35S-labeled guanosine 5′-3-O-(thio)triphosphate (GTPγS) to membranes. The most prominent uncoupled FPR mutants were located in the N-terminal part of the second transmembrane domain (S63W and D71A) and the C-terminal interface of the third transmembrane domain (R123A and C124S/C126S). In addition, less pronounced uncoupling was detected with deletion mutations in the third cytoplasmic loop and in the cytoplasmic tail. Further analysis of some of the mutants that were judged to be uncoupled based on the [35S]GTPγS membrane-binding assay were found to transduce a signal, as evidenced by intracellular calcium mobilization and activation of p42/44 MAPK. Thus, these single point mutations in FPR did not completely abolish the interaction with G protein, emphasizing that the coupling site is coordinated by several different regions of the receptor. Mutations located in the putative fifth and sixth transmembrane domains near the N- and C-terminal parts of the third cytoplasmic loop did not result in uncoupling. These regions have previously been shown to be critical for G protein coupling to many other G protein-coupled receptors. Thus, FPR appears to have a G protein-interacting site distinct from the adrenergic receptors, the muscarinic receptors, and the angiotensin receptors.

The N-formyl peptide receptor (FPR; also called FPR1) 1 is a seven-transmembrane domain G protein-coupled receptor (GPCR) expressed mainly in myeloid cells (for a review, see Ref. 1). FPR binds bacterial and mitochondrial N-formylated peptides with high affinity in a G protein-dependent fashion, resulting in an intracellular signal transduction cascade that induces chemotaxis, calcium mobilization, superoxide produc-tion, and degranulation. Based on a number of studies, the ligand is believed to occupy a ligand binding pocket within the transmembrane regions (for reviews, see Refs. [1][2][3]. The biological importance of FPR in host defense against bacterial infections was recently confirmed; Gao and co-workers (4) showed that mice with a targeted disruption of Fpr1, the mouse orthologue of human FPR1, had increased susceptibility to challenge with Listeria monocytogenes.
Previous studies analyzing G protein coupling to FPR have suggested that regions in the first and second cytoplasmic loop, fifth and sixth transmembrane domain, and the cytoplasmic tail of FPR are involved in the interaction with G protein; synthetic receptor-mimetic peptides corresponding to Gly 43 -Thr 61 , Ile 119 -Thr 133 , Asp 122 -Lys 144 , Gln 134 -Trp 150 , Phe 210 -Ile 224 , Lys 230 -Val 246 , and Arg 322 -Thr 336 disturbed the formation of a FPR-G i2 complex to various extents (IC 50 20 M to 1.4 mM), whereas Val 127 -Ser 140 , Lys 227 -Pro 239 , Met 304 -Ser 319 , Ala 315 -Thr 329 , Gln 330 -Glu 344 , and Asn 337 -Lys 350 had a minor effect or no effect (see Fig. 1A) (5,6). Another study using a similar approach provided evidence that peptides containing the amino acids Cys 126 -Arg 137 , Phe 308 -Arg 322 , and Ser 319 -Leu 340 interacted with G protein, whereas Lys 227 -Arg 241 and Thr 339 -Lys 350 did not (Fig. 1A) (7). Furthermore, mutagenesis studies suggested that a deletion of Lys 230 -Pro 239 and various substitutions between residues 226 and 234 in the putative midportion of the third cytoplasmic loop did not have a critical role in G protein coupling (8). A study using chimeras between FPR and C5a receptor showed that the first intracellular loop Arg 54 -Ile 62 is necessary for ligand-dependent activation of G ␣16 (9). The consensus from these studies is that the third cytoplasmic loop is not involved in G protein coupling, whereas all or parts of the other cytoplasmic domains do participate in the interaction.
In addition to the cytoplasmic amino acids, certain highly conserved amino acids in the predicted transmembrane domains have been found to affect G protein coupling to various GPCRs. These residues include an asparagine in the first transmembrane domain (TMI), aspartic acid in the second transmembrane domain (TMII), aspartic acid and arginine in the third transmembrane domain (TMIII: (D/E)RY motif; DRC in FPR), and asparagine and tyrosine in the seventh transmembrane domain (TMVII: (N/D)PXXY motif). The most comprehensive analysis of G protein coupling was carried out through site-directed and computer-simulated mutagenesis of the ␣ 1B -adrenergic receptor. These studies showed that substitution of asparagine in TMI with an alanine (N63A) resulted in a constitutively active receptor (10). In addition, a D142A mutation in the DRY motif of TMIII caused high constitutive activity, whereas a R143A mutation, also in the DRY motif, abolished inositol phosphate accumulation in response to ligand binding (10). Similarly, Asp 71 in TMII and Arg 123 in the DRC motif of FPR appeared to be required for G protein coupling (11). Based on molecular dynamics simulations of the ␣ 1B -adrenergic receptor, Arg 143 of the DRY motif is involved in hydrogen bonding interactions with Asp 91 , the conserved aspartic acid in TMII. This amino acid in turn forms a highly conserved polar pocket including the conserved asparagine in TMI and asparagine and tyrosine in TMVII (10). Similar hydrogen bonding is also thought to take place in other GPCRs. In the wild-type gonadotropin-releasing hormone receptor, the conserved aspartic acid in TMII has been replaced with an asparagine, and the conserved asparagine in TMVII has been replaced with an aspartic acid. A N87D mutation resulted in loss of ligand binding and a D318N mutation resulted in abolished inositol phosphate production (12,13). However, a double mutant N87D/D318N re-established inositol phosphate production (12,13). Similarly, a D120N substitution in TMII of serotonin 5-HT 2A receptor eliminated coupling, but a D120N/ N376D double mutant restored receptor function (14). The purpose of the hydrogen bonding network is to stabilize the inactive form of the receptor (14). Upon ligand binding, this hydrogen bonding network may be altered, allowing certain amino acid residues to interact with G protein (10). To test the importance of equivalent amino acids in FPR, we made the following mutations of corresponding putative hydrogen bonding residues in FPR; N44A (TMI), D71A (TMII), D122A, R123A, R123G (TMIII), N297A, Y301A (TMVII), and D71N/ N297D. In addition, we examined the effect of several other transmembrane domain point mutations and cytoplasmic deletion mutations. Although our results supported the hydrogen bonding network model, they also suggested that many of the G protein-interactive sites of FPR are distinct from other extensively studied GPCRs.

EXPERIMENTAL PROCEDURES
Oligonucleotide-directed Mutagenesis and Cell Transfections-cDNAs encoding point-mutated human FPRs were generated by oligonucleotide-directed mutagenesis using standard techniques (15,16). The mutations were confirmed by dideoxy sequencing (17). cDNA inserts coding for wild-type or mutant FPR were subcloned into the EcoRI site of pBGSA (18). This vector contains an SR␣ promoter (19), transcription termination and polyadenylation signals from the human growth hormone gene, and a neomycin resistance cassette. The constructs were transfected into Chinese hamster ovary (CHO) cells using lipofectACE (Life Technologies, Inc.; Ref. 20). Transfectants were selected with 0.5 mg/ml G-418 sulfate (Calbiochem-Novabiochem). 16 or 32 single colonies were isolated and tested for expression and cellular localization by immunofluorescence microscopy.
Cell Culture and Immunofluorescence Microscopy-Transfected CHO cells were maintained in selection medium containing 0.5 mg/ml G418 in ␣-modified Eagle's medium containing 5% fetal bovine serum, 50 units/ml penicillin and 50 g/ml streptomycin. 14 -16 h before each experiment, increased expression was induced by adding 6 mM sodium butyrate to the medium (21). To examine receptor expression levels and localization in the cell, cells on coverslips were fixed and stained as described previously (22).
Flow Cytometry-Transfected CHO cells were detached from culture plates with a cell scraper after incubation with 1 mM Na-EDTA (pH 8.0) in Ca 2ϩ /Mg 2ϩ -free phosphate-buffered saline. Cells were harvested by centrifugation and resuspended in phosphate-buffered saline containing 5% fetal bovine serum. Cells were filtered through a 44-m Spectra/ Mesh macroporous filter (Spectrum, Houston, TX), and N-formyl-Nle-Leu-Phe-Nle-Tyr-Lys-fluorescein derivative (Molecular Probes, Inc., Eugene, OR) was added to 1-ml cell volumes to give final concentrations ranging from 50 pM to 40 nM (or in some cases to 160 nM). Nonspecific binding was determined in the presence of a 1000-fold excess of fMLF. Following a 30-min incubation at 0°C, 10,000 cells from each peptide incubation mixture were analyzed using a FACScan flow cytometer (Becton Dickinson, San Jose, CA), and the K d values for the various receptors were determined by least squares analysis of the mean fluorescence intensity using the Prism program (GraphPad Prism, San Diego, CA). Quantum 25 fluorescein microbeads (Flow Cytometry Standards Corp., San Juan, PR) containing molecules of equivalent soluble fluorochromes ranging from 5 ϫ 10 4 to 2 ϫ 10 6 were used as standards to determine the average number of FPRs on the cell surface. The B max was calculated based on the sigmoidal binding curve.
Preparation of Membranes and [ 35 S]GTP␥S Assay-Cells were harvested as above and resuspended in 10 mM HEPES, 100 mM KCl, 10 mM NaCl, 3.5 mM MgCl 2 , 1 mM ATP, and 1 mM phenylmethylsulfonyl fluoride, pH 7.4. Membranes were prepared by nitrogen cavitation at 4°C for 15 min at 400 p.s.i. The cavitates were cleared by a low speed centrifugation after which the membranes were pelleted for 30 min at 45,000 rpm in a Ti75 rotor. The membranes were resuspended in membrane binding buffer containing 10 mM HEPES, 100 mM NaCl, 10 mM MgCl 2 , and 1 mM phenylmethylsulfonyl fluoride, pH 7.4. The amount of membrane protein was determined using the BCA protein assay kit (Pierce). The membranes were either used immediately or frozen at Ϫ70°C.
[ 35 S]GTP␥S assays were performed using 30 g of membrane protein in 1 ml of membrane binding buffer. Samples were incubated with or without 1 M fMLF in the presence of 0.5 M GDP for 30 min at 30°C. [ 35 S]GTP␥S was added to 0.04 Ci/tube with or without 10 M GTP␥S to analyze for nonspecific [ 35 S]GTP␥S binding. The reaction was allowed to proceed for 30 min at 30°C and terminated by filtration through Whatman GF/C filters (Whatman International, Maidstone, United Kingdom), and the filters were washed three times with 2 ml of buffer containing 50 mM Tris-HCl, pH 7.5, and 5 mM MgCl 2 . Filters were placed in 5 ml of scintillation fluid and counted in liquid scintillation counter. The counts generally ranged from 100 to 200 cpm for the control vials with excess unlabeled GTP␥S to 1000 -2000 cpm in the presence of fMLF and absence of unlabeled GTP␥S.
Calcium Release Assay-Cells were detached from culture plates as above and incubated with 2.5 M Fura-2/AM (Molecular Probes) in phosphate-buffered saline plus Ca 2ϩ /Mg 2ϩ and 5% fetal bovine serum for 45 min at 37°C. Cells were centrifuged and resuspended in the above buffer at room temperature, where they were left until the assay. Each measurement contained about 5 ϫ 10 6 CHO cells in a volume of 1 ml. Continuous fluorescent measurements of calcium-bound and free Fura-2 were made at 37°C with a double excitation monochromator fluorescence spectrofluorometer (Photon Technology International, Monmouth Junction, NJ) with excitation at 340 and 380 nm and emission at 510 nm. 10 M ATP was added as a heterologous ligand to provide a standard stimulus for calcium mobilization. The amount of fMLF-induced calcium release was expressed as the percentage of ATPinduced calcium release (equal to 100%) using the value immediately before the fMLF addition (t ϭ 48 s) as base line (equal to 0%).
MAPK Assay-CHO transfectants were incubated with 100 nM fMLF at 37°C for 0, 1, 5, or 10 min and lysed with boiling Laemmli sample buffer. In some experiments, cells were preincubated 20 min with 500 ng/ml pertussis toxin or 100 nM wortmannin prior to a 5-min incubation with 100 nM fMLF. An equivalent of about 1 ϫ 10 5 cells was run on two 10% SDS-polyacrylamide gels. The proteins were transferred from the gels onto nitrocellulose membranes, and the MAPKs were detected with two different antibodies. The phosphorylated MAPKs, p44 and p42, were detected using a mouse monoclonal antibody against human p42/44 MAPK phosphorylated on Thr 202 and Tyr 204 (New England Biolabs, Beverly, MA). The total p42/44 MAPKs were detected using a rabbit polyclonal antibody that recognizes both phosphorylated and nonphosphorylated protein (New England Biolabs). The primary antibody incubations were followed by horseradish peroxidase-conjugated secondary antibody incubations and detection was carried out by enhanced chemiluminescence (Amersham Pharmacia Biotech).

The Effect of Transmembrane and Cytoplasmic Mutations on FPR Cell Surface Expression and Ligand
Binding Affinity-To examine the role of specific FPR structural features in receptor-G protein interaction, a number of amino acid residues, predicted to be either transmembrane or cytoplasmic, were mutated or deleted, and the mutant receptors were expressed in CHO cells. The amino acids that were mutated are shown in Fig. 1B and Table I. The following mutations led to intracellular retention of the receptor in transfected CHO cells and were not examined further: single-amino acid substitutions in five positions (N44A, R54A, D122A, P213A, and P298A), two alanine insertion mutations (D122-AAA-R123 and L243-AAA-S244), and three deletion mutations (⌬128 -135, ⌬316 -322, and ⌬323-329) (see Fig. 1B, amino acids within thick circles). All mutant receptors that were found to be expressed on the cell surface showed a similar ligand binding affinity as wildtype FPR ( Table I), suggesting that these mutations may not cause major structural changes that might account for uncoupling from G protein.
Amino Acid Substitutions at Seven Different Sites Significantly Affected G Protein Coupling to FPR-To examine whether the mutations affect ligand-induced coupling to G protein, we analyzed the binding of [ 35 S]GTP␥S to membranes from CHO transfectants in the absence and presence of fMLF. The results are summarized in Table I (23) and are believed to participate in G protein coupling in the ␣ 1B -adrenergic receptor through a hydrogen bonding network (10). The D71A substitution caused the most dramatic uncoupling and a substitution with asparagine instead of alanine (D71N) did not restore the coupling activity, suggesting that the charged residue is required in this position. A fourth residue also considered important in the hydrogen bonding is asparagine in the (N/D)PXXY motif in the seventh transmembrane domain of GPCRs (10). Contrary to our expectations, based on results from homologous substitutions on other GPCRs (12,14,24), the N297A mutation did not affect G protein coupling. This result was confirmed using another N297A clone from a separate transfection. However, a double mutant, D71N/N297D, restored the D71N defect and was able to couple to G protein, as has been previously shown for the gonadotropin-releasing hormone receptor (13).
In various GPCRs, amino acids in the amino-and carboxylterminal interfaces between the third cytoplasmic loop and the fifth and sixth transmembrane domains have been shown to be important for G protein coupling (see "Discussion"). We mutated multiple sites in these regions (see Fig. 1B); however, none of these mutations decreased the G protein coupling in response to fMLF. We also mutated amino acids in the first cytoplasmic loop (His 57 ) and the carboxyl-terminal interface between the first cytoplasmic loop and the second transmembrane domain (Ser 63 ). This is a region that is highly conserved among members of the FPR family including human FPR1, FPRL1 (also called the lipoxin A 4 receptor; Ref. 25) and FPRL2, rabbit FPR1, and mouse Fpr1, Fpr-rs1, Fpr-rs2, Fpr-rs3, Fpr-rs4, and Fpr-rs5 (26,27). His 57 in the middle of the putative first cytoplasmic loop is conserved between FPR1 and the  (23). A, FPR-mimetic peptides shown to be involved in G protein coupling. Regions of importance, as reported by Bommakanti et al. (6), are shown as black circles with white letters, whereas black letters on gray circles indicate inactive peptides (peptides that did not inhibit G protein-FPR interaction) (5,6). When the inactive and active peptides overlap, the active peptides are shown with thick circles. The active peptides described by Schreiber et al. (7) are indicated with thick lines, and the inactive peptides are indicated with thin lines. The extracellular domains of FPR are not shown. B, site-directed and deletion mutants analyzed for G protein coupling in this study. The shaded circles indicate amino acids that do not appear to be required for G protein coupling. The black circles with white letters indicate those amino acids that upon amino acid substitution or deletion (indicated with lines) resulted in uncoupling from G protein. The letters within thick circles indicate residues that upon deletion or substitution resulted in intracellular retention of the receptor. The boldface numbers show the fMLF-induced increase in GTP␥S binding for the deletion mutants. The quantitative results are shown in Table I. ***, p Ͻ 0.0001; *, p Ͻ 0.01; n.s., nonsignificant.
mouse Fpr clones but is an arginine in FPRL1 and FPRL2 (Fig.  1B). Substitution of this highly polar residue with serine did not affect G protein coupling (Table I). Ser 63 in the putative second transmembrane domain is found in human and rabbit FPR1 and in mouse Fpr-rs2 and Fpr-rs4, whereas the residue is a cysteine in human FPRL1 and FPRL2, and mouse Fpr-rs3 and Fpr-rs5. A more dramatic substitution is found in mouse Fpr-rs1 and Fpr-rs2, where the residue is a tryptophan (26). The C5a receptor also has a tryptophan in this position (28). A S63W substitution in FPR resulted in uncoupling of G protein (Table I). This result was somewhat unexpected, since the receptors in the FPR family and the C5a receptor are thought to be coupled to the same classes of heterotrimeric GTP-binding proteins. We also made a S63C substitution and, due to a planning error, an additional Y64F substitution, generating a S63C/Y64F double mutant (the C5a receptor has a phenylalanine in position 64). Like S63W, this S63C/Y64F double mutant was uncoupled from G protein. The serine residue in position 63 may thus be unique for G protein coupling to certain members of the FPR family of receptors.
As shown in Table I, we mutated a number of transmembrane cysteine residues to serines. The purpose for these mutations was 2-fold: The first was to confirm the importance of Cys 124 -Cys 126 in G protein coupling and explain the differen-  Fig. 1A). The second was to prepare mutant FPR to carry out site-directed spin labeling studies. C73S substitution in the second transmembrane domain did not affect G protein coupling. In addition to human FPR, only Fpr1 has a cysteine in this position; all other mouse Fprs and rabbit FPR have a serine in this position. In contrast, Cys 124 and Cys 126 are conserved throughout the FPR-family. Cys 124 is found in the DRC sequence in the putative third transmembrane domain (Fig. 1B). Most G proteincoupled receptors have a tyrosine in this position, forming the conserved DRY motif. Other amino acids in this position are tryptophan and histidine (29). The amino acid residue equivalent to Cys 124 is not conserved among GPCRs. Substitution of these cysteines with serines in FPR (C124S/C126S) resulted in an uncoupled receptor, suggesting that this conservation may be important for the function of the FPR family receptors. Together with the peptide competition studies mentioned above, the results support the conclusion that the midportion of the second cytoplasmic loop is inactive and the active G protein coupling portion is in the N-terminal part of the loop (6). Finally, Cys 295 in the seventh transmembrane domain, two residues from the NPXXY sequence (Fig. 1B), is found in all  FPR family receptors and several other chemoattractant receptors (29). A serine substitution resulted in uncoupling of FPR from G protein, although the effect was not as significant as seen with the C124S/C126S double mutant. Next we analyzed the effect of various 6 -8-amino acid deletions on G protein coupling. A deletion of 6 amino acids from the putative midsection of the third cytoplasmic loop of FPR resulted in a reduction in G protein coupling, but this reduction was not highly significant, suggesting that this region may play a structural rather than a direct role in G protein binding (Fig.  1B, Table I). Since parts of the cytoplasmic tail of FPR has previously been shown to participate in G protein coupling based on mutagenesis and peptide competition assays (5, 7, 11), we generated mutants with 7-amino acid deletions. Two out of four mutants were not expressed on the cell surface, presumably due to misfolding. A deletion including three threonine and one serine residue, ⌬330 -336, caused a significant decrease in G protein coupling, supporting previously obtained results showing that a synthetic peptide corresponding to amino acids 322-336 inhibits binding of G protein to FPR (Fig.  1A) (5,6). We also carried out substitutions of Asp 327 and Asp 333 to proline to examine whether a possible conformational change caused by the proline residue would affect G protein coupling. These mutations resulted in a small but insignificant decrease in GTP␥S binding, suggesting that the substitutions do not perturb the interaction with G protein.

Uncoupling of Signal Transduction in FPR Mutants May Not
Be Complete, as Evidenced by the Calcium Mobilization Assay and the MAPK Assay-To further examine the effects of the mutations on signal transduction in transfected CHO cells, we determined whether ligand binding to mutant receptors fails to induce intracellular calcium release. Unexpectedly, we found that the two mutant receptors that were tested, D71A and R123A, mobilized calcium from intracellular stores in response to fMLF (Fig. 2). However, the amount of fMLF required for mobilization varied compared with wild-type FPR. D71A and R123A gave essentially no calcium response with 10 nM fMLF (Fig. 2, left panel). At 1 M fMLF, the calcium release was clearly noticeable, although the response was smaller compared with wild-type FPR (Fig. 2, right panel). To confirm that the cells were in good condition during the assays and to provide a standard stimulus, calcium mobilization was also triggered with ATP, a ligand that binds the purinergic receptor, resulting in release of intracellular calcium (Fig. 2). The numbers below the fMLF-induced calcium peaks indicate the percentage of calcium mobilization relative to ATP-induced calcium mobilization, as defined under "Experimental Procedures." As a control, we also examined the calcium release mediated by the D71N/N297D double mutant. As seen in Fig. 2, the extent of calcium mobilization in response to fMLF was similar to wild-type FPR. Nontransfected CHO cells did not release intracellular calcium in response to 1 M fMLF, indicating that the effect of fMLF was specific (data not shown).
To further analyze the signal transduction, we determined whether fMLF binding to the mutant receptors would activate a downstream protein kinase cascade. Activation of human neutrophils by fMLF is known to induce phosphorylation of a MAPK, p42/44 MAPK, also known as extracellular signal-regulated kinase (30). We incubated the cells in the presence of fMLF for 0, 1, 5, or 10 min and analyzed the cell lysate for the presence of total and phosphorylated p42/44 MAPK. As seen in the Western blot shown in Fig. 3A, increased phosphorylation of p42 was detected after a 5-10-min fMLF incubation of CHO cells expressing either wild-type FPR or the D71A or R123A mutants. To assure that the amount of total p42/44 MAPK was equal in each lane, a nitrocellulose membrane prepared in parallel was blotted with an antibody recognizing both phosphorylated and nonphosphorylated p42/44. As seen in Fig. 3B, the amount of total p42/44 MAPK did not significantly vary between the different incubation time points. Hence, we conclude that no single amino acid change that affects the interaction between FPR and G protein appears capable of completely blocking the function of the receptor, as shown through ligand-induced calcium release and p42/44 MAPK activation.
To verify that the p42/44 MAPK phosphorylation occurs through a G i -mediated pathway, cells were incubated with pertussis toxin, which ADP-ribosylates G i . As shown in Fig. 4, A and C, this treatment reduced the amount of phosphorylated p42 MAPK. In addition, activation of p42 MAPK required phosphoinositide 3-kinase, as indicated by the large reduction in phosphorylation in the presence of wortmannin (Fig. 4, A and  C), thus confirming previous findings (31,32). Therefore, the activation of p42/44 MAPK occurs through a pertussis toxinsensitive pathway, presumably through G i , although the possibility of alternative pathways cannot be completely excluded. DISCUSSION The sites for G protein coupling have been mapped for several GPCRs using receptor chimeras, site-directed mutants, deletion mutants, and inhibitory peptides. These studies have implicated the second cytoplasmic loop in receptor-G protein coupling for the adrenergic receptors (33), muscarinic receptors (34,35), interleukin-8 receptor (36,37), and vasopressin 1a receptor (38) as well as FPR (5)(6)(7). In our studies, we identified three amino acid residues (Arg 123 , Cys 124 , and Cys 126 ) in the N-terminal second cytoplasmic loop and the interface of the third transmembrane domain as putative sites of interaction with G protein (Fig. 1B). Our finding confirmed the results from two earlier studies. One showed that a R123G substitu-tion resulted in a mutant FPR that was unable to mediate calcium mobilization (11). The other showed that the C-terminal portion of the second cytoplasmic loop N-terminal to Val 125 contains residues critical for coupling (6). It has been postulated that Asp 122 and Arg 123 , which form the conserved (D/ E)RY motif (DRC in FPR), participate in an extensive hydrogen bonding network that stabilizes the inactive form of the receptor (10,14,39). Based on mutagenesis and computational analysis, ligand binding alters the hydrogen bonding network, and certain amino acid residues, such as arginine in the DRY motif, become exposed to interaction with G protein (14,39). To test this hypothesis, we constructed a mutant FPR with 3 alanine residues inserted between the aspartic acid and the arginine (D122-AAA-R123) to artificially alter the position of the arginine residue and possibly prevent the hydrogen bond formation. However, this mutation resulted in intracellular retention of the receptor and was not studied further. Likewise, our attempts to examine the role of Asp 122 in G protein coupling failed due to intracellular retention. Finally, we constructed a double mutant with Cys 124 (from the DRC motif) and Cys 126 , each substituted with a serine. This mutant was uncoupled from G protein, suggesting that amino acids in close proximity to DR may also participate either structurally or directly in G protein coupling.
The N-and/or C-terminal portions of the third cytoplasmic loop proximal to or within the predicted transmembrane domains have been found to be involved in G protein interaction with the adrenergic receptors (33, 40 -45), the muscarinic receptors (35, 46 -48), the angiotensin receptor (49,50), interleukin-8 receptor (36), and platelet-activating factor receptor (51) as well as FPR (6). Based on application of the Baldwin model of GPCRs to FPR, the third cytoplasmic loop is predicted to be short, only 13 amino acids, compared with the adrenergic receptor family, with up to 78 amino acids (23). Two independent studies examining the role of this domain of FPR using inhibitory peptides indicated that its most hydrophilic part is not required for G protein coupling (6,7). In addition, point mutations and a deletion mutant (⌬230 -239) in this region of the receptor did not affect G protein coupling to the receptor (8). We therefore decided to first examine whether specific amino acid residues in the fifth or sixth transmembrane domains near the third cytoplasmic loop indeed are required for G protein coupling. Despite extensive mutagenesis in these regions (see Fig.  1B), we were unable to identify amino acids that are involved in G protein-coupling (Table I). The synthetic peptides Bommakanti et al. (6) used to perturb the FPR-G protein physical association extended deeper into the transmembrane bilayer by the more recent model. The N-terminal peptide comprised amino acids Phe 210 -Ile 224 , and the C-terminal peptide comprised amino acids Lys 230 -Val 246 (Fig. 1A). These peptides could have functioned by their mimicry of FPR in binding to G protein binding site or by perturbing FPR structure by binding to a complementary site in FPR. This latter interpretation is supported by the improper folding of the Pro 213 mutant, although it requires a direct test of the Bommakanti et al. (6) peptides in the function of GTP␥S binding. It is also supported by additional deletion/insertion experiments. An insertion mutant with three alanines between Leu 243 and Ser 244 (L243-AAA-S244) was used to examine whether perturbation of the placement of these putative transmembrane residues facilitates an interaction with G protein, as was previously demonstrated for the m2 muscarinic receptor (47,48). This mutant was also misfolded. These results support the importance of transmembrane/interfacial amino acid residues in maintaining proper structure of FPR. Finally, we generated a mutant with 6 amino acids deleted from the midsection of the third cyto- plasmic loop (⌬229 -234; Fig. 1B). This deletion mutant was expressed on the cell surface and showed a reduced coupling to G protein; however, the decrease was less pronounced than for the point mutants discussed above. In previous studies, a peptide comprising the amino acids Lys 227 -Pro 239 and Lys 227 -Arg 241 , and a deletion mutation ⌬K230-P239, had no effect on FPR-G protein interaction (Fig. 1A) (6 -8). In addition, H229A and K230Q point mutations did not increase the EC 50 of fMLFinduced Ca 2ϩ mobilization (8). Based on the above studies, it is likely that the 6-amino acid deletion in our ⌬229 -234 mutant is not directly involved in G protein coupling. Instead, the deletion may affect the receptor conformation and thus indirectly reduce the efficiency for G protein coupling. This interpretation is also supported by the finding that three out of six deletion mutants were retained in the endoplasmic reticulum, presumably due to misfolding. Thus, any conclusions based on the deletion mutants must be made carefully.
The asparagine and tyrosine residues in the conserved (N/ D)PXXY motif in the seventh transmembrane domain of GPCRs are thought to participate in the hydrogen-bonding network together with aspartic acid and arginine in the (D/ E)RY motif, aspartic acid in the second transmembrane domain (Asp 71 in FPR), and an asparagine residue in the first transmembrane domain (Asn 44 in FPR) (10). However, the N297A mutant in the (N/D)PXXY motif did not alter FPR-G protein coupling, as expected. To confirm the result, the mutation was resequenced, and the construct was retransfected into CHO cells. Identical results were obtained with the new transfec-tants. Although Asn 297 was not required for G protein coupling of FPR, we found that an amino acid residue in that position can affect coupling, since a D71N/N297D double mutant was found to rescue the D71N uncoupled phenotype (Table I). Alanine substitution of Tyr 301 in the (N/D)PXXY motif resulted in a 3-4-fold reduction in G protein coupling, suggesting that the tyrosine residue may be involved in the interaction, as previously shown for other GPCRs (52,53).
Our results and recent results by others confirm the importance of the cytoplasmic tail of GPCRs in G protein interaction. A 3-amino acid consecutive sequence in the cytosolic tail of angiotensin receptor was found to be essential in the activation of G protein (54), and a deletion of the cytoplasmic tail of FPR led to uncoupling from G protein (55). In addition, a FPR peptide and a fusion peptide comprising amino acids 322-336 and 319 -340, respectively, inhibited G protein coupling to FPR (5, 7), whereas other peptides had no effect (Fig. 1A) (5,6). We therefore examined the effect of various cytoplasmic tail deletion mutants on G protein coupling to FPR. Deletion of amino acids 309 -315 showed a small (ϳ2-fold) decrease in G protein coupling. This result is similar to a previous observation showing that a R309G/E310A/R311G FPR mutant was partly uncoupled from G protein (11). However, our result was not statistically significant due to a large S.E. Deletions 316 -322 and 323-329 resulted in intracellular protein retention and were not studied further. Deletion of amino acids 330 -336 caused a significant decrease in G protein coupling, in agreement with the results from the peptide inhibition assays (5, 7). An aspar- tic acid in position 333 did not appear to be important, since a substitution with proline did not significantly reduce G protein coupling. We conclude that one or more of the amino acids in the cytoplasmic tail sequence 330 -336 may participate in the interaction with G protein, but the aspartic acid in position 333 is not required.
Mutant "Uncoupled" Receptors Can Activate Intracellular Calcium Mobilization and p42/44 MAPK Phosphorylation-Coupling of G protein to a receptor is generally analyzed by measuring binding of [ 35 S]GTP␥S to membranes or production of either cAMP or phosphatidylinositol. We decided to examine the effect of the FPR mutations on mobilization of intracellular calcium. We found that the mutations did not completely inhibit this event, but a higher concentration of ligand was required to obtain a response (Fig. 2). It thus appears that although the results from the [ 35 S]GTP␥S assay suggested that some of the mutants were uncoupled from G protein, the uncoupling may be partial. Similarly, ligand binding to the D71A and R123A mutants led to activation of the p42/44 MAPK pathway (Fig. 3). To confirm that the p42/44 MAPK activation was dependent on a G protein-mediated signal, we examined the effect of pertussis toxin on the phosphorylation. We observed a ϳ45% decrease in the amount of phosphorylated p42 MAPK after a 20-min preincubation with 500 ng/ml pertussis toxin, suggesting that G i plays a major role in the activation. Similar results with ϳ49 and ϳ70% inhibition, respectively, were reported for fMLF-stimulated neutrophils that were preincubated with pertussis toxin for 2.5 h at 500 ng/ml or 24 h at 100 ng/ml (56,57). We cannot completely rule out the possibility that alternative pathways for p42/44 MAPK activation may exist; cells expressing the M 3 muscarinic receptor showed that polyclonal antibodies against the small G proteins ARF1/3 and RhoA co-immunoprecipitated the M 3 receptor upon exposure to agonist (58). In the FPR system, small G protein interactions have been demonstrated by Polakis et al. (59). These investigators showed that small G proteins copurify with FPR in a GTP␥S-dependent manner, thus suggesting a functional link between the receptor and these proteins. Alternatively, the recently discovered constitutive activity of FPR is suggestive of several levels of activation that could stem from conformationally different forms of FPR (60). Such differences could be functionally represented by a spectrum of G protein-coupled activities that may be limited in mutated forms of FPR. It is therefore possible that the small G proteins also mediate FPRstimulated intracellular signal transduction, although our results suggest that the pertussis toxin-sensitive pathway is of major importance in the signal transduction pathway activating p42/44 MAPK.
In summary, we have confirmed the requirement of certain highly conserved amino acids in G protein coupling, such as Asp 71 (TMII: 10), Arg 123 (TMIII: 25), and Tyr 301 (TMVII: 21). We have also shown that the N-and C-terminal regions in the putative interface between the third cytoplasmic loop and the fifth and sixth transmembrane domains are probably not involved in G protein coupling to FPR. Further mutagenesis results also suggest that the following residues may be involved in the interaction with G protein: Ser 63 in the putative interface of the first cytoplasmic loop and second transmembrane domain; Cys 124 (DRC motif) and/or Cys 126 in the interface between the third transmembrane domain and second cytoplasmic loop; and Cys 295 in the seventh transmembrane domain. The results thus confirm that an extensive contact is required to form the G protein coupling site.