Quantitative Analysis of Formyl Peptide Receptor Coupling to Giα1, Giα2, and Giα3 *

The human formyl peptide receptor (FPR) is a prototypical Gi protein-coupled receptor, but little is known about quantitative aspects of FPR-Gi protein coupling. To address this issue, we fused the FPR to Giα1, Giα2, and Giα3 and expressed the fusion proteins in Sf9 insect cells. Fusion of a receptor to Gα ensures a defined 1:1 stoichiometry of the signaling partners. By analyzing high affinity agonist binding, the kinetics of agonist- and inverse agonist-regulated guanosine 5′-O-(3-thiotriphosphate) (GTPγS) binding and GTP hydrolysis and photolabeling of Gα, we demonstrate highly efficient coupling of the FPR to fused Giα1, Giα2, and Giα3without cross-talk of the receptor to insect cell G proteins. The FPR displayed high constitutive activity when coupled to all three Giα isoforms. The K d values of high affinity agonist binding were ∼100-fold lower than the EC50 (concentration that gives half-maximal stimulation) values of agonist for GTPase activation. Based on theB max values of agonist saturation binding and ligand-regulated GTPγS binding, it was previously proposed that the FPR activates G proteins catalytically, i.e. one FPR activates several Gi proteins. Analysis of agonist saturation binding, ligand-regulated GTPγS saturation binding and quantitative immunoblotting with membranes expressing FPR-Giα fusion proteins and nonfused FPR now reveals that FPR agonist binding greatly underestimates the actual FPR expression level. Our data show the following: (i) the FPR couples to Giα1, Giα2, and Giα3 with similar efficiency; (ii) the FPR can exist in a state of low agonist affinity that couples efficiently to G proteins; and (iii) in contrast to the previously held view, the FPR appears to activate Gi proteins linearly and not catalytically.

The extended ternary complex model of receptor activation assumes that GPCRs exist either in an inactive "R" state or an active "R*" state (4 -6). GPCRs can isomerize from R to R* spontaneously, and this process is referred to as constitutive activity. Receptor agonists stabilize the R* state and increase basal G protein activity, whereas inverse agonists stabilize the R state and reduce basal G protein activity (4,6,7).
Several questions regarding the quantitative aspects of FPR-G i ␣ coupling are still unresolved. Specifically, it is unclear why the K d value for high affinity [ 3 H]fMLF binding is ϳ3 nM, whereas the EC 50 value of fMLF at activating G i proteins in terms of GTP hydrolysis, AA-GTP labeling, and GTP␥S binding is much higher (ϳ0.1-1 M) (8,12,25,26). Similarly, it is unknown why fMLF is much more potent at activating chemotaxis than lysosomal enzyme release and superoxide radical formation (8, 19 -21). Furthermore, it is not clear whether the FPR couples differentially to G i ␣ isoforms. This latter question is intriguing in view of several findings. First, phagocytes express G i ␣ 2 at a much higher level than G i ␣ 3 , and G i ␣ 1 is not expressed at all (11,13,27,28). Second, certain GPCRs, including the receptors for the chemoattractant interleukin-8, differ substantially in their coupling efficiency to various G i ␣ isoforms (29 -34). Third, data obtained with the G i ␣ 2 knock-out mouse suggest that this G protein has unique functions in signal transduction (35,36).
Perhaps most importantly, G i ␣ 2 has a lower GDP affinity than G i ␣ 1 and G i ␣ 3 (37,38). For G s ␣ isoforms, it has been shown that the GDP affinity of G␣ has a substantial impact on the efficiency of receptor-G protein coupling. Particularly, G s ␣ L (the long splice variant of the ␣-subunit of the stimulatory G protein of adenylyl cyclase G s ) has a lower GDP affinity than G s ␣ S (the short splice variant) (7,39). Thus, the GPCR activation energy required for releasing GDP from G s ␣ L is lower than the corresponding activation energy needed for GDP release from G s ␣ S . Accordingly, the ␤ 2 AR catalyzes GDP release from G s ␣ L more readily than from G s ␣ S . Experimentally, this results in increased efficacy and potency of partial agonists and increased efficacy of inverse agonists when the ␤ 2 AR is coupled to G s ␣ L as compared with the corresponding ligand properties when coupling of the ␤ 2 AR to G s ␣ S is considered (7). In other words, G s ␣ L confers to the ␤ 2 AR the properties of a constitutively active GPCR. Intriguingly, when coupled to G i ␣ 2 , the FPR is constitutively active as well as assessed by strong inhibitory effects of the inverse agonist CsH on the high basal GTP␥S binding (25). Taken together, all these findings raise the question of the impact of the different G i ␣ isoforms on constitutive activity of the FPR.
The aim of our present study was to quantitatively analyze FPR coupling to the three G i ␣ isoforms. To achieve this aim, we fused the FPR to G i ␣ 1 , G i ␣ 2 , and G i ␣ 3 and expressed the fusion proteins in Sf9 cells. Fusion of GPCR to G␣ ensures a defined 1:1 stoichiometry and efficient coupling of the signaling partners and allows for the sensitive detection of differences in the coupling of a given GPCR to different G proteins (7, 40 -43).

EXPERIMENTAL PROCEDURES
Materials-The cDNAs of G i ␣ 1 , G i ␣ 2 and G i ␣ 3 in pGEM-2 were kindly provided by Dr. R. Reed (Howard Hughes Medical Institute, Johns Hopkins University, Baltimore, MD) (44 fMLF, fMW, and MLF were from Sigma. Unlabeled nucleotides were obtained from Roche Molecular Biochemicals. All restriction enzymes and T4 DNA ligase were from New England Biolabs. Cloned Pfu DNA polymerase (for generation of fusion protein DNAs) was from Stratagene. Taq polymerase (for analysis of mRNA expression in Sf9 cells) was from Sigma. All other reagents were of the highest purity available and were from standard suppliers.
Construction of the cDNAs Encoding Fusion Proteins-For generation of FPR-G i ␣ fusion proteins, in PCR 1A, the DNA sequence of the C terminus of the FPR was amplified with pGEM3Z-SF-FPR26-His 6 as template by using a sense primer 5Ј of the newly created EcoRI site of the FPR (25) (sense EcoRI primer) and an antisense primer encoding the hexahistidine tag. The cDNAs of G i ␣ 1 , G i ␣ 2 , and G i ␣ 3 were amplified in PCR reactions 1B 1 , 1B 2 , and 1B 3 , respectively, using pGEM-2-G i ␣ 1,2,3 plasmids as templates. The sense primers annealed with the first 18 bp of the 5Ј-end of the respective G i ␣ and included the 18 bp of the hexahistidine tag in their 5Ј-extensions. The antisense primers encoded the 5 C-terminal amino acids of the respective G i ␣ followed by the stop codon and an extra XbaI site for cloning purposes in the 3Ј-end extension. In PCRs 2A-2C, the cDNA fragments from PCR 1A and PCRs 1B 1 , 1B 2 , and 1B 3 , respectively, were annealed and amplified using the sense EcoRI primer and the antisense primers of PCR 1B 1 , 1B 2 , and 1B 3 , respectively. In this way, a fragment encoding the C terminus of FPR, a hexahistidine tag, and the respective G i ␣ followed by an XbaI site was obtained. The fragments were digested with EcoRI and XbaI and cloned into pGEM3Z-SF-FPR26-His 6 digested with EcoRI and XbaI. PCRgenerated DNA sequences were confirmed by restriction enzyme analysis and enzymatic sequencing. For cloning into the baculovirus expression vector pVL 1392, the cDNAs encoding fusion proteins in pGEM-3Z were digested with HindIII at the 5Ј-end of the SF region, blunted with DNA Polymerase I (Klenow fragment), and then digested with XbaI at the 3Ј-end of G i ␣. Digested fusion protein DNAs were then transferred into pVL 1392 that had been digested with BglII, blunted with Klenow fragment, and subsequently digested with XbaI. The DNA for the ␤ 2 AR-G i ␣ 2 fusion protein (used as immunoblotting standard for the determination of G i ␣ 2 expression) was prepared by overlap extension PCR analogous to the FPR-G i ␣ fusion protein DNAs. PCR-generated sequences were confirmed by enzymatic sequencing.
Generation of Recombinant Baculoviruses, Cell Culture, and Membrane Preparation-Generation of baculoviruses, cell culture, and membrane preparation were performed exactly as described (25,41). Sf9 cells were co-infected with recombinant baculoviruses encoding nonfused FPR, G i ␣ 2 , and G protein ␤ 1 ␥ 2 complex or fusion proteins plus G protein ␤ 1 ␥ 2 complex.
Analysis of FPR-G i ␣ Fusion Proteins by Agonist Saturation Binding, GTP␥S Binding, and GTPase Activity-[ 3 H]fMLF saturation binding, GTP␥S saturation binding, and time course of GTP␥S binding in Sf9 membranes expressing FPR-G i ␣ fusion proteins plus ␤ 1 ␥ 2 complex were performed exactly as described for Sf9 membranes expressing nonfused FPR, G i ␣ 2 , and ␤ 1 ␥ 2 complex (25). Steady-state GTPase activity with different substrate concentrations was determined as described for Sf9 membranes expressing ␤ 2 AR-G s ␣ L fusion protein except that the MgCl 2 concentration was 5 mM instead of 1 mM (41).
Photoaffinity Labeling of FPR-G i ␣ Fusion Proteins-AA-GTP was prepared from [␣-32 P]GTP as described (47). Labeling of Sf9 membranes was performed essentially as described (47). Briefly, Sf9 membranes expressing FPR-G i ␣ fusion proteins (200 g/tube) were incubated for 20 min at 25°C in a buffer containing 100 mM NaCl, 5 mM MgCl 2 , 1 M GDP, 5 g/ml soybean trypsin inhibitor, and 30 mM HEPES/NaOH, pH 7.5. Reaction mixtures additionally contained 1 Ci of AA-GTP and solvent (control) or 10 M fMLF. After centrifugation for 10 min at 15,000 ϫ g and 4°C, membranes were suspended in a buffer containing 100 mM NaCl, 5 mM MgCl 2 , 5 g/ml soybean trypsin inhibitor, 2 mM dithiothreitol, and 30 mM HEPES/NaOH, pH 7.5. Membranes were UV-irradiated for 3 min at 302 nm and separated by SDS-PAGE, and AA-GTP-labeled proteins were identified by autoradiography.
Analysis of FPR-G i ␣ Fusion Protein Expression on the Protein Level-Sf9 membranes were analyzed by immunoblotting using the monoclonal M1 antibody, which recognizes the N-terminal FLAG epitope of the FPR and ␤ 2 AR (25, 41), anti-His 6 Ig, which recognizes the Cterminal His 6 epitope of the FPR and ␤ 2 AR (25, 41), anti-G i ␣ 1/2 Ig, and anti-G i ␣ 3 Ig. The M1 antibody was used at a dilution of 1:1000. The other antibodies were used at a dilution of 1:500. SDS-PAGE and immunoblotting were performed exactly as described (25).
Analysis of FPR-G i ␣ Fusion Protein Expression on the mRNA Level-mRNA from Sf9 cells infected with recombinant baculoviruses was isolated with the RNeasy kit from Qiagen and treated with RNasefree DNase. mRNA was reverse-transcribed using the First Strand cDNA synthesis kit from Amersham Pharmacia Biotech. The cDNAs of G i ␣ 1 and G i ␣ 2 in FPR-G i ␣ fusion proteins were amplified using appropriate primer pairs. PCR products were digested with various restriction enzymes, separated by electrophoresis on gels containing 2% (w/v) agarose, and visualized by ethidium bromide staining.
Miscellaneous-Protein was determined using the Bio-Rad DC protein assay kit. Data were analyzed by nonlinear regression, using the Prism program.

Analysis of FPR-G i ␣ Fusion Protein Expression on the Protein and mRNA Level--
The human FPR expressed in Sf9 cells has a molecular mass of 40 kDa (25), and the molecular mass of G i ␣ proteins is 40 -41 kDa (11,13). Thus, the expected molecular mass of FPR-G i ␣ fusion proteins is 80 -81 kDa. Indeed, the M1 antibody, which recognizes the FLAG epitope attached to the N terminus of the FPR (25) detected antigens of the appropriate mass in membranes prepared from baculovirus-infected Sf9 cells (Fig. 1A). The anti-G i ␣ 1/2 Ig detected the FPR-G i ␣ 1 and FPR-G i ␣ 2 fusion proteins but not FPR-G i ␣ 3 (Fig. 1B). In contrast, the anti-G i ␣ 3 Ig reacted with the FPR-G i ␣ 3 fusion protein but not with FPR-G i ␣ 1 or FPR-G i ␣ 2 (Fig. 1C). There was no indication for proteolytic degradation of fusion proteins. Because the anti-G i ␣ 1/2 Ig cannot discriminate between G i ␣ 1 and G i ␣ 2 (Fig. 1B), we performed reverse transcriptase-PCR analysis on mRNA from Sf9 cells infected with the FPR-G i ␣ 1 and FPR-G i ␣ 2 baculoviruses to differentiate between the two G i ␣ isoforms. G i ␣ 1 has a unique SacI site at position 382, and G i ␣ 2 has a unique BamHI site at position 653 (Fig. 1D). The G i ␣ portions of fusion proteins were amplified by PCR and digested. SacI generated the expected 382-and 711-bp fragments with G i ␣ 1 , whereas G i ␣ 2 cDNA was not cut. In contrast, digestion with BamHI yielded the expected 443-and 653-bp fragments with G i ␣ 2 , whereas G i ␣ 1 cDNA was not cut. Taken together, the immunoblotting and reverse transcriptase-PCR data document the specific expression of FPR-G i ␣ 1 , FPR-G i ␣ 2 , and FPR-G i ␣ 3 fusion proteins in Sf9 cell membranes.  Table I summarizes the data for all three fusion proteins. For comparison, the [ 3 H]fMLF binding data for nonfused FPR co-expressed with G i ␣ 2 are included as well in Table I. FPR-G i ␣ fusion proteins bound the agonist [ 3 H]fMLF (0.2-30 nM) according to single-site saturation curves. The K d and B max values of [ 3 H]fMLF binding to FPR-G i ␣ 1 , FPR-G i ␣ 2 , and FPR-G i ␣ 3 were similar to each other and comparable with the values for the co-expression system. An increase of the [ 3 H]fMLF concentration up to 300 nM in the fusion protein and co-expression systems did not further increase B max of agonist binding as compared with a [ 3 H]fMLF concentration of 30 nM (data not shown), indicating that a possible low affinity agonist-binding site of the FPR cannot be unmasked by means of agonist saturation binding. Binding of GTP␥S to G␣ uncouples GPCRs from G proteins and reduces high affinity agonist binding (7,25,48). By analogy to nonfused FPR co-expressed with G i ␣ 2 , GTP␥S substantially reduced the B max of [ 3 H]fMLF binding to FPR-G i ␣ fusion proteins and, to a variable extent, increased the K d for [ 3

H]fMLF.
Kinetics of GTP␥S Binding-Ligand-regulated GTP␥S binding to G proteins is a sensitive method for studying GPCR-G protein coupling, specifically for analyzing the time course of G protein activation and the stoichiometry of coupling (22,25,49). Fig. 2C shows a representative time course experiment for the GTP␥S binding to FPR-G i ␣ 1 , and Fig. 2D shows a representative GTP␥S saturation binding experiment for FPR-G i ␣ 1 . Table II summarizes the GTP␥S binding data for all three fusion proteins. For comparison, the GTP␥S binding data for the FPR co-expressed with G i ␣ 2 are included in Table II as well. As reported previously for the FPR co-expressed with G i ␣ 2 in Sf9 cells (25), the FPR fused to G i ␣ 2 displayed high constitutive activity as assessed by the strong inhibitory effect of the inverse agonist CsH on GTP␥S binding. CsH also displayed strong inhibitory effects at the FPR fused to G i ␣ 1 and G i ␣ 3 . The K d values for fMLF-stimulated and CsH-inhibited GTP␥S binding were similar for all FPR-G i ␣ fusion proteins studied and compare favorably with the K d values for GTP␥S binding with nonfused FPR co-expressed with G i ␣ 2 . With respect to the time course of GTP␥S binding, fMLF reduced whereas CsH increased t1 ⁄2 , but there was some variability between the different systems studied.
Because there is a defined 1:1 stoichiometry of GPCR to G␣ in fusion proteins (7,(41)(42)(43), 1 mol of activated GPCR can maximally stimulate the binding of 1 mol of GTP␥S to its fused G␣ partner. Thus, one would expect that the B max of receptor ligand binding is very similar to the B max of ligand-regulated GTP␥S binding. The B max of ligand-regulated GTP␥S binding Summaries of the data for FPR-G i ␣ 1 and the other fusion proteins as well as the co-expression system consisting of nonfused FPR plus G i ␣ 2 are given in Tables I-IV. Tables I-IV also provide details on the specific type of nonlinear regression analysis performed with each data set.
(B max total ) is the difference between maximum GTP␥S binding stimulated by fMLF and minimum GTP␥S binding inhibited by CsH. As is evident from the comparison of B max values for [ 3 H]fMLF binding and B max total values of GTP␥S binding (compare Tables I and II and Fig. 2, A, B, and D), B max total of GTP␥S binding surpasses B max of [ 3 H]fMLF binding by 4.7-fold for FPR-G i ␣ 1 , by 5.9-fold for FPR-G i ␣ 2 , and by 2.5-fold for FPR-G i ␣ 3 .

Resolution of the Discrepancies between the B max of [ 3 H]fMLF
Binding and Ligand-regulated GTP␥S Binding-Two explanations for the discrepancies between the B max values of [ 3 H]fMLF binding and ligand-regulated GTP␥S binding have to be considered. First, the fused FPR could interact not only with its fused G i ␣ partner but also with G proteins of the host cell. Indeed, cross-talk of a fused G i protein-coupled GPCR to endogenous G proteins occurs in some mammalian expression systems (50,51). Second, the FPR could still couple only to its fused G␣ partner but may do so in a state of low agonist affinity. This low agonist affinity state may not be detected in the [ 3 H]fMLF binding assay but in the GTP␥S binding assay because the receptor promotes binding of GTP␥S to the G protein. The efficient coupling of GPCRs to G proteins in a state of low agonist affinity had already been postulated in previous studies (48,52).
To address the first explanation, we visualized FPR-activated G proteins by photoaffinity labeling with AA-GTP (12,13,47). These studies were performed in the presence of NaCl (see "Experimental Procedures") to facilitate detection of agonist responses and to suppress agonist-independent labeling of G proteins as the result of high constitutive receptor activity (25). fMLF efficiently stimulated the incorporation of AA-GTP into all three FPR-G i ␣ fusion proteins (Fig. 3). However, fMLF did not stimulate the incorporation of AA-GTP into ϳ40 -45-kDa proteins, i.e. endogenous insect cell G proteins (46). These data rule out the possibility of cross-talk of the fused FPR to endogenous G proteins of Sf9 cells. In agreement with the data regarding fused FPR, nonfused FPR also does not couple to insect cell G proteins (25,53).
To address the second explanation, we determined the expression of FPR-G i ␣ fusion proteins by a method that is independent of radioligand binding. Specifically, we took advantage of the FLAG and His 6 epitopes that are located at the N and C termini, respectively, of the GPCR portions of fusion proteins (see "Experimental Procedures") (41,43). The previously characterized ␤ 2 AR-G s ␣ L fusion protein can be used as standard to assess the expression level of other proteins bearing the same epitopes because the expression level of ␤ 2 AR-G s ␣ L can be unequivocally determined by receptor antagonist saturation binding (41,43). Fig. 4A shows that the immunoreactivity of FPR-G i ␣ 1 expressed at 1.4 pmol/mg ([ 3 H]fMLF saturation binding) was even higher than the immunoreactivity of ␤ 2 AR-G s ␣ L expressed at 8.6 pmol/mg ([ 3 H]DHA saturation binding) using the anti-His 6 Ig. The signals with FPR-G i ␣ 2 (Fig. 4A) and FPR-G i ␣ 3 (data not shown) expressed at 1.0 pmol/mg each were slightly smaller than the signals obtained with ␤ 2 AR-G s ␣ L expressed at 8.6 pmol/mg. By using the anti-FLAG Ig, the immunoreactivity with ␤ 2 AR-G s ␣ L expressed at 8.6 pmol/mg was moderately higher than with FPR-G i ␣ 1 expressed at 1.4 pmol/mg. The difference in sensitivity between the anti-FLAG Ig and anti-His 6 Ig could be due to the fact that the FPR is heavily glycosylated at the extreme N terminus (25,53,54) and that this glycosylation could interfere with the recognition of the N-terminal FLAG epitope by the M1 antibody.
To answer the question of whether [ 3 H]fMLF saturation binding also underestimates the expression level of nonfused FPR, we compared the immunoreactivity of membranes expressing the ␤ 2 AR or the FPR plus G i ␣ 2 plus ␤ 1 ␥ 2 complex using the anti-FLAG Ig. Fig. 4C shows that the FPR runs as a much more diffuse band in SDS-PAGE than the ␤ 2 AR, indicating that the FPR is more heavily glycosylated than the ␤ 2 AR. Despite this difference in glycosylation pattern of the two GPCRs, it is evident that the immunoreactivity of the ␤ 2 AR expressed at 3.9 pmol/mg ([ 3 H]DHA saturation binding) is comparable with the immunoreactivity of the FPR expressed at 1.1 pmol/mg ([ 3 H]fMLF saturation binding). Unfortunately, we H]fMLF binding not competed for by 10 M unlabeled fMLF, was subtracted from both sets of binding data. Data were analyzed for best fit to single-site and two-site saturation curves. All curves were best fit to single-site saturation curves. Data shown are the means Ϯ S.D. of three experiments with different membrane preparations performed in triplicates. The data for the nonfused FPR co-expressed with G i ␣ 2 were taken from Ref. 25.  . Time course data were analyzed for best fit to monophasic and biphasic saturation curves. All data best fit to a monophasic saturation curve. Data shown are the means Ϯ S.D. of three experiments with different membrane preparations performed in triplicates. The data for the nonfused FPR co-expressed with G i ␣ 2 were taken from Ref. 25. could not visualize nonfused GPCRs with the His 6 Ig. Apparently, the His 6 Ig cannot efficiently recognize the His 6 tag when directly located at the C terminus of the antigens studied.
Taken together, the quantitative immunoblotting studies demonstrate that [ 3 H]fMLF saturation binding underestimates the actual expression level of the fused and nonfused FPR by a factor of ϳ4 -8, depending on which antibody is used for the detection of the GPCRs. This factor agrees well with the factors obtained in the [ 3 H]fMLF and GTP␥S saturation binding studies (Tables I and II). The similarity of the fusion protein expression levels as determined by immunoblotting and GTP␥S saturation binding also implies that the majority, if not all, of the expressed fusion protein molecules are functionally active. Of interest, by comparing the B max values of agonist and antagonist binding at the adenosine A 1 receptor co-expressed with, or fused to, G i ␣ proteins, it was also shown that agonist binding underestimates GPCR expression level by ϳ3-6-fold (55). Unfortunately, an antagonist radioligand for the FPR is not available, but the combination of immunoblotting and ligand-regulated GTP␥S binding can compensate for this deficiency.
We also assessed the potency of fMLF at activating the fused G i ␣ partner. If the FPR in a state of low agonist affinity does indeed couple to G i proteins, then the EC 50 values for fMLF in functional assays should be considerably higher than the K d values in the agonist binding studies. In fact, at all three fusion proteins, fMLP activated GTP hydrolysis with EC 50 values that are ϳ80 -125-fold higher than the K d values for high affinity [ 3 H]fMLF binding (compare Fig. 2 (A, B, and E) and Tables I  and III). A similar discrepancy between agonist affinity and agonist potency was reported for nonfused FPR expressed in HL-60 leukemia cells and Sf9 cells (25,26).
GTP Turnover Measurements-The defined 1:1 stoichiometry of GPCR and G␣ in fusion proteins allows determination of ligand-regulated GTP turnover in a membrane system (41,42). Specifically, GTPase activities are divided by B max values of receptor antagonist saturation binding to calculate ligand-regulated GTP turnover in fusion proteins (41,42). However, because agonist saturation binding largely underestimates the actual expression level of FPR-G i ␣ fusion proteins (Tables I and  II and Figs. 2, A, B, and D, and 4), we divided GTPase activities by the B max values of ligand-regulated GTP␥S binding. Fig. 2F shows a representative experiment for the kinetics of ligandregulated GTP turnover at FPR-G i ␣ 1 , and Table IV summarizes the data for all three fusion proteins. The K m values of fMLF-stimulated GTP hydrolysis for FPR-G i ␣ fusion proteins were similar to each other as were the K m values for CsHinhibited GTP hydrolysis. The V max values of ligand-regulated GTP turnover were similar for the three FPR-G i ␣ fusion proteins. The kinetic parameters of the GTPase of FPR-G i ␣ fusion proteins are also similar to the corresponding parameters of an ␣ 2 -adrenoreceptor-G i ␣ 1 fusion protein (42).
Determination of Agonist Efficacies and Potencies-The data obtained with the full agonist fMLF and the full inverse agonist CsH in GTP␥S binding and GTPase studies (Tables II-IV) point to high constitutive activity of the FPR coupled to all G i ␣ isoforms. We wished to explore the hypothesis that there are, nonetheless, subtle differences in the constitutive activity of the FPR coupled to G i ␣ isoforms that can only be detected with partial agonists. According to the extended ternary complex model, an increase in constitutive activity is accompanied by an increase in partial agonist efficacy and potency (7,56,57). The GTPase assay is particularly suitable for determination of ligand efficacies and potencies because it monitors, unlike the GTP␥S binding assay, receptor-G protein coupling at steady state (7,41). Moreover, the GTPase assay has already successfully been used to dissect differences in the constitutive activity of a given receptor coupled to different G␣ isoforms (7).
We analyzed a panel of peptides and identified fMW as moderately strong partial agonist and the nonformylated peptide MLF as weak partial agonist at the FPR. Fig. 2E shows representative concentration-response curves for various FPR ligands at FPR-G i ␣ 1 , and Table III provides a summary for all three FPR-G i ␣ fusion proteins studied. In all systems studied, FPR ligands activated GTP hydrolysis in the following order of efficacy: fMLF Ͼ fMW Ͼ MLF Ͼ BocFLFLF (ineffective). There were no systematic differences in agonist efficacy and potency for a given fusion protein compared with another fusion protein. Specifically, fMW and MLF were not more efficacious and potent at FPR-G i ␣ 2 than at FPR-G i ␣ 1 and FPR-G i ␣ 3 .
Expression Level of Nonfused G i ␣ 2 in the Co-expression System-When the FPR-G i ␣ 2 fusion protein was compared with the corresponding co-expression system, we were puzzled that not only the B max values of [ 3 H]fMLF binding were similar to each other but also the B max values of ligand-regulated GTP␥S binding (compare Tables I and II). However, if the FPR had activated G i ␣ catalytically in the co-expression system, we would have expected a much higher B max of ligand-regulated GTP␥S binding in this system than in the fusion protein system. Thus, we had to address the hypothesis that the low B max of ligand-regulated GTP␥S binding in the co-expression system was due to a low expression level of G i proteins, resulting in poor availability of G proteins for activation by GPCR. We used a ␤ 2 AR-G i ␣ 2 fusion protein expressed at 15 pmol/mg ([ 3 H]DHA saturation binding) as standard for analyzing the expression level of nonfused G i ␣ 2 . The immunoreactivity with the anti-G i ␣ 1/2 Ig in membranes expressing FPR plus G i ␣ 2 was ϳ20 -30 times higher than the immunoreactivity with similar amounts of protein from membranes expressing ␤ 2 AR-G i ␣ 2 . Thus, in the co-expression system, G i ␣ 2 is expressed at a level of ϳ300 -450 pmol/mg. Similar expression levels in Sf9 cells were also reported for other G proteins (41,58). Given an estimated FPR expression level of ϳ4 pmol/mg in the co-expression system (Fig. 4C), these data clearly show that limited expression of G i ␣ 2 cannot account for the surprisingly low B max of ligandregulated GTP␥S binding in this system.

Highly Efficient Receptor-G Protein Coupling in a FPR-G i ␣ 2
Fusion Protein-The aim of our study was to quantitatively analyze the coupling of the FPR to the three G i ␣ isoforms. The fusion protein technique provides a defined 1:1 stoichiometry of the signaling partners and is a sensitive system for dissecting differences in the coupling of a given GPCR to different G␣ isoforms (7,43). However, because the fusion of a receptor C terminus to the G␣ N terminus is artificial, one may be concerned that the fusion substantially alters the properties of the receptor and the G protein as well as their coupling to each other. Moreover, in certain expression systems, G i proteincoupled receptors cannot only couple to their fused G i ␣ partner but also to the endogenous G proteins of the host cells (50,51). To address these concerns, we compared the coupling of the FPR to G i ␣ 2 in the fused and nonfused state and visualized activated G proteins by photoaffinity labeling. There were only minor differences in FPR-G i ␣ 2 coupling between the fused and nonfused state (Tables I and II), and there is no cross-talk between the fused FPR and G proteins of the host cell (Fig. 4). Thus, the fusion protein technique is a valid approach to analyze potential differences in the coupling of the FPR to G i ␣ isoforms.
In the co-expression system, there is an ϳ100-fold molar excess of G i ␣ 2 over FPR as assessed by quantitative immunoblotting with defined standards (Figs. 4C and 5). Despite the large excess of G protein relative to receptor in the co-expression system and the high absolute expression levels of FPR and G i ␣ 2 , the signaling efficiency of the co-expression system as assessed by the B max of ligand-regulated GTP␥S binding was not greater than that of the fusion protein system, which has only a 1:1 stoichiometry of receptor to G protein and a much lower G protein expression level (Table II). These data show  50 values for CsH, corresponding to ligand potencies, were obtained from fitted sigmoidal saturation curves. The efficacies of ligands were calculated by dividing the maximal ligand-stimulated (or CsH-inhibited) GTPase activity by the maximal fMLF-stimulated GTPase activity. The maximum effect of fMLF was set 1.00. The maximum ligand effects were derived from sigmoidal saturation curves, too. Data shown are the means Ϯ S.D. of three experiments with different membrane preparations performed in triplicates. NA, not applicable (because of the minimal effects of BocFLFLF, precise potencies could not be calculated).   For each GTP concentration, the basal GTPase activity was subtracted from the GTPase activity observed in the presence of fMLF to calculate the increase in GTPase activity caused by fMLF. From each basal GTPase activity value, the GTPase activity observed in the presence of CsH was subtracted to obtain the decrease in GTPase activity caused by CsH. Kinetic data were analyzed for best fit to single-site and two-site saturation curves. All data fit best to single-site saturation curves. To obtain GTP turnover numbers, GTPase activities (expressed as pmol/mg/min) were divided by the B max total (pmol/mg) of ligand-regulated GTP␥S binding of the respective membrane (see Table II) because this is a more precise measure of fusion protein expression level than receptor agonist binding (see Table I and Fig. 4). V max total is the sum of fMLF-and CsHregulated GTP turnover (ligand-regulated GTP turnover). Data shown are the means Ϯ S.D. of three experiments with different membrane preparations performed in triplicates. that in the co-expression system, the vast majority of G proteins is not engaged in coupling, whereas in the fusion protein system, most if not all G proteins participate in coupling. Thus, the fusion induces an optimal positioning of G i ␣ relative to the FPR. Similar conclusions were obtained for ␤ 2 AR-G s ␣ and adenosine A 1 receptor-G i ␣ fusion proteins (41,55,59).
Similarly Efficient Coupling of the FPR to G i ␣ 1 , G i ␣ 2 , and G i ␣ 3 -A stimulus for our present study was observations regarding the importance of the GDP affinity of G proteins for receptor-G protein coupling. Specifically, the GDP affinity of G i ␣ 2 is lower than the GDP affinity of G i ␣ 1 and G i ␣ 3 (37,38). In addition, the FPR co-expressed with G i ␣ 2 has the properties of a constitutively active GPCR (25). Moreover, G s ␣ L has a lower GDP affinity than G s ␣ S , and the ␤ 2 AR coupled to G s ␣ L is constitutively active, too (7,39). An explanation for these findings could be that the lower the GDP affinity of a given G␣ is, the lower is the GPCR activation energy required to catalyze GDP release from a G protein. Because even an agonist-free GPCR can efficiently promote GDP release from a G protein with low GDP affinity, this G protein accordingly confers to the coupled GPCR the properties of constitutive activity.
Based on the above findings and considerations, we predicted that the FPR coupled to G i ␣ 2 would have a higher degree of constitutive activity than the FPR coupled to G i ␣ 1 and G i ␣ 3 . Surprisingly, however, the relative inhibitory effects of CsH at the total ligand-regulated GTP␥S binding and GTP hydrolysis were very similar among the three G i ␣ isoforms (Tables II-IV). Another indicator for increased constitutive activity of a GPCR is elevated efficacy and potency of partial agonists (7,56,57). However, we could not detect consistent increases in these parameters for FPR-G i ␣ 2 as compared with FPR-G i ␣ 1 and FPR-G i ␣ 3 (Table IV). Taken together, our data show that the different G i ␣ isoforms, which differ from each other in GDP affinity, do not have a specific impact on the constitutive activity of the FPR. These data clearly demonstrate that the observations made for the coupling of G s ␣ splice variants to the ␤ 2 AR cannot be readily extrapolated to other GPCR-G protein pairs. Possibly, the intrinsic constitutive activity of the FPR is so high that the modulation of constitutive activity by GDP affinity of G i ␣ isoforms is too subtle to become effective.
The lack of substantial differences in the coupling of the FPR to G i ␣ isoforms is somewhat unexpected in view of the fact that many GPCRs show differences in coupling efficiency to G i ␣ isoforms, that G i ␣ 2 is the major G protein in phagocytes, and that G i ␣ 1 is not expressed in these cells (27,29,(31)(32)(33)(34). Of interest, the first, second, and third intracellular loops as well as the C terminus of the FPR are involved in G protein coupling (9,60), whereas for most other receptors, the areas involved in G protein coupling appear to be more confined (61)(62)(63). Thus, the extensive contact area of the FPR with G proteins may provide the structural basis for the lack of differential coupling to G i ␣ isoforms. Based on our findings, one can assume that in vivo, G i ␣ 2 and G i ␣ 3 are used interchangeably by the FPR.
Stoichiometry of FPR-G i ␣ Coupling: Implications for Signaling in Vivo-Because FPR-G i ␣ fusion proteins have a defined 1:1 stoichiometry of GPCR and G␣ and because there is no cross-talk of the fused receptor with endogenous G proteins, we expected very similar B max values for [ 3 H]fMLF binding and ligand-regulated GTP␥S binding, reflecting the fact that one FPR molecule can activate one G i ␣ molecule. However, the B max total values of GTP␥S binding were up to 6-fold higher than the B max values of [ 3 H]fMLF binding (Tables I and II). Thus, a substantial fraction of the FPRs appears to exist in a state of low agonist affinity that cannot be detected in the agonist binding assay but which, nonetheless, efficiently couples to G proteins to stimulate guanine nucleotide exchange. This model is supported by the fact that the EC 50 values of fMLF at activating GTP hydrolysis and guanine nucleotide binding to G i ␣ in fused and nonfused systems are ϳ100-fold higher than the K d values for [ 3 H]fMLF binding (12,13,25,26). Additionally, quantitative immunoblotting with anti-His 6 Ig and anti-FLAG Ig using ␤ 2 AR-G s ␣ L as standard demonstrated that agonist saturation binding greatly underestimates the actual expression level of FPR fusion proteins (Fig. 4, A and B). The underestimation of FPR expression level by [ 3 H]fMLF binding is also true for nonfused receptors (Fig. 4C).
The fact that the agonist binding assay underestimates the actual expression level of FPRs by a factors of ϳ2.5-6 renders it necessary to re-evaluate previously obtained conclusions regarding the stoichiometries of FPR-G i protein coupling in nonfused systems. We and other groups had proposed that the FPR activates G i proteins catalytically, i.e. a single FPR molecule activates several G i ␣ molecules (22)(23)(24)(25). These numbers were obtained by dividing B max of ligand-regulated GTP␥S binding by B max of [ 3 H]fMLF binding. If one uses the actual B max of FPR expression for these calculations, it emerges that a single FPR in average does not activate much more than a single G i protein, i.e. there is linear rather than catalytical G protein activation. This model of linear signal transfer is further supported by the fact that in the co-expression system, despite very high abundance of G i proteins, the total number of activated G i proteins was not higher than in the fusion protein system (Tables I and II and Fig. 5). In a co-expression system of the ␤ 2 AR and G s ␣ L we also could not obtain evidence for catalytical G protein activation by receptors (41). Collectively, our data suggest that the 1: 1 stoichiometry of GPCR and G␣ obtained in fusion proteins may reflect the in vivo stoichiometry of receptor-G protein coupling more closely than was previously appreciated (22,43,64).
Linear G protein activation does have to result in a loss of signal amplification and stimulus sensitivity. Given the fact that the FPR can couple to G i proteins even in a state of low agonist affinity (Tables I-III and Fig. 2E) (8,12,48), our model of linear signal transfer could explain the different phagocyte responses over a large range of fMLF concentrations. Specifically, at low fMLF concentrations, phagocytes migrate along a fMLF gradient toward their target, i.e. the formyl peptidereleasing bacteria (20). Once the phagocyte is approaching the FIG. 5. Analysis of the expression level of G i ␣ 2 in the co-expression system consisting of FPR and G i ␣ 2 by quantitative immunoblotting. Sf9 membranes expressing ␤ 2 AR-G i ␣ 2 and nonfused FPR plus G i ␣ 2 were prepared. The expression level of ␤ 2 AR-G i ␣ 2 was determined by antagonist saturation binding ([ 3 H]DHA being the radioligand) and is indicated in the figure. Defined amounts of protein (expressed as g/lane) from membranes expressing ␤ 2 AR-G i ␣ 2 or nonfused FPR plus G i ␣ 2 were separated by SDS-PAGE and probed with anti-G i ␣ 1/2 Ig as described under "Experimental Procedures." Numbers on the left indicate molecular masses of marker proteins. Shown are the horseradish peroxidase-reacted nitrocellulose membranes of gels containing 10% (w/v) acrylamide. Similar results were obtained in three independent experiments. bacteria, the concentration of formyl peptide increases, and more G proteins are activated, resulting in stimulation of cytotoxic cell functions such as lysosomal enzyme release and superoxide radical production and destruction of bacteria (8,20,21). Thus, linear signal transfer in phagocytes may help to prevent premature activation of cytotoxic cell functions and harmful destruction of host tissue.
Finally, the results of our present study indirectly provide information about the FPR expression level in vivo. Most GPCRs are expressed at levels in the fmol/mg range (see, for example, Refs. 40, 65, and 66), whereas the FPR is expressed at levels of ϳ1 pmol/mg even when assessed by [ 3 H]fMLF saturation binding (22,24,26). Based on our present data, it is likely that the actual FPR expression level in phagocytes is considerably higher than 1 pmol/mg. Thus, the FPR is presumably a GPCR with one of the highest physiological expression levels.
In conclusion we have shown that the human FPR couples to the G i proteins G i ␣ 1 , G i ␣ 2 , and G i ␣ 3 with similar efficiency. By taking advantage of the defined 1:1 stoichiometry of GPCR and G␣ in fusion proteins, we obtained insights into quantitative aspects of FPR-G i protein coupling that could not have been achieved by another approach. Our data suggest that the FPR activates G i proteins in a linear fashion and not catalytically.