Direct or C5a-induced Activation of Heterotrimeric G i2 Proteins in Human Neutrophils Is Associated with Interaction between Formyl Peptide Receptors and the Cytoskeleton*

The binding of ligands to N -formyl peptide chemoattractant receptors in human neutrophils results in a rapid association of these receptors with a cytoskeletal fraction and a specific activation and release of G i2 (cid:97) -subunits from this fraction. In the present study we could show that pretreating neutrophils with GDP (cid:98) S prevented the fMet-Leu-Phe-induced association of its receptor with a cytoskeletal fraction and also blocked the release of G i2 (cid:97) -subunits from the same cytoskeletal fraction. In contrast, direct activation of G i2 proteins by addition of GTP (cid:103) S or AlF 4 (cid:50) not only caused a release of G i2 (cid:97) -subunits from the cytoskeleton but also an associ- ation of formyl peptide receptors with the cytoskeleton. The receptor for complement fragment 5a, which trans-duces its signaling through the same G i2 protein, trig- gers both a release of G i2 (cid:97) -subunits from the cytoskeleton fraction and, of even greater interest, an association between formyl peptide receptors and the cytoskeleton. The close relationship

Human neutrophils are phagocytic cells specialized in the destruction of invading microorganisms. To perform this important role in host defense, these cells emigrate from the blood vessels to the sites of infection. This directed locomotion, as well as other neutrophil responses, is triggered by chemotactic factors, e.g. complement fragment 5a (C5a), 1 leukotriene B 4 , and N-formylated peptides. Many of the receptors for these ligands have been cloned (Boulay et al., 1990;Gerard and Gerard, 1990) and shown to belong to a subclass of small seven-transmembrane-spanning G-protein-coupled receptors (Probst et al., 1992;Prossnitz et al., 1993). Related chemoattractant receptors show a greater degree of homology in their intracellular domains than in their extracellular domains, suggesting that they may have common mechanisms of signal transduction and receptor regulation (Ye et al., 1991). The neutrophil receptor for N-formylated peptides (FPR) is one of the most thoroughly studied chemoattractant receptors (for review, see Murphy, 1994). This receptor is a glycosylated protein with an apparent molecular mass of 50 -70 kDa, as estimated by photoaffinity labeling (Tardif et al., 1993;Sengeløv et al., 1994).
To be able to move in a chemotactic gradient, motile cells like the human neutrophil must respond to changes in the concentration of one or probably several stimuli. In part, this is achieved by an adaptive process that results in a blunted response despite the permanent presence of agonists. This process, which is also called desensitization, involves receptor down-regulation by internalization as well as more rapid mechanisms by which components of the signaling pathway are modified. In general, two types of desensitization have been characterized, and these are designated homologous and heterologous . Homologous desensitization affects only the receptor system that has been activated by an agonist, whereas heterologous desensitization also inactivates other receptors coupled to the same effector system. Both types of desensitization results in the uncoupling of receptors from their effector enzymes. Several potentially interrelated mechanisms for how the receptor-G-protein interaction are altered when cells get desensitized have been demonstrated and hypothesized McLeish et al., 1989).
Considering the neutrophil FPR, it has previously been suggested that binding to the cytoskeleton is an important step in the desensitization of this receptor to chemotactic peptides Jesaitis et al., 1986). In addition to the association of FPRs with the cytoskeleton, a lateral segregation of the FPR from its G-protein has also been observed (Jesaitis et al., 1989;for review, see Jesaitis, 1992). Such segregation could well be a consequence of the N-formyl-L-methionyl-L-1 The abbreviations used are: C5a, complement fragment 5a; FPR, formyl peptide receptor; fMet-Leu-Phe, N-formyl-L-methionyl-L-leucyl-L-phenylalanine; G-protein, GTP-binding protein; G i2 ␣-subunit, ␣-subunit of the G i2 protein; G n ␣, ␣-subunit of the G i2 protein; G s ␣-subunit, ␣-subunit of the G s protein; GDP␤S, guanosine-5Ј-O-(2-thiodiphosphate); GTP␥S, guanosine-5Ј-O-(3-thiotriphosphate); PVDF, polyvinylidene difluoride; TX-100, Triton X-100. leucyl-L-phenylalanine (fMet-Leu-Phe)-induced release of the ␣-subunits of the G i2 protein (G i2 ␣-subunits) from the cytoskeleton (Sä rndahl et al., 1993) and would, as part of the desensitization process, serve to keep the FPR from transducing its signal to its corresponding G-protein. Although both the activation and cellular handling of G i2 proteins and FPRs are most likely of importance in the desensitization process, it is still unclear whether they are in any way interrelated.
The present study was performed to reveal whether or not the activation and release of G i2 ␣-subunits from the cytoskeleton and the association of FPRs with the cytoskeleton are interdependent processes. This was done to gain further knowledge about the desensitization of FPRs that is necessary for neutrophil locomotion.

MATERIALS AND METHODS
Chemicals-All reagents used were of an analytical grade. Dextran and Ficoll-Paque were from Pharmacia Biotech (Sollentuna, Sweden), fMet-Leu-Phe was obtained from Sigma, and GDP␤S and GTP␥S were both from Boehringer Mannheim (Mannheim, Germany). The peroxidase-conjugated goat anti-rabbit antiserum was obtained from Dako A/S (Glostrup, Denmark), and the enhanced chemiluminescence detection system was from Amersham Int. (Buckinghamshire, U. K.). The affinity purified rabbit anti-FPR antibody was raised against amino acid 340 -350 of the carboxyl tail of the human FPR (i.e. the FPR C-terminal peptide; Tardif et al., 1993). The rabbit antiserum R16,17 was raised against a peptide corresponding to the 9-amino acid Cterminal sequence of G i and was shown to react specifically with Bokoch et al., 1988).
Isolation of Human Neutrophils-Peripheral human blood was obtained from healthy volunteers and collected in heparin-containing vacutainer tubes. After sedimentation on dextran, the neutrophils were isolated according to the method described by Böyum (1968). The obtained cell suspension, which consisted of approximately 98% neutrophils, was washed twice and then resuspended in a calcium-containing medium as described previously (Sä rndahl et al., 1993). The cells were routinely pretreated with 2.5 mM diisopropyl fluorophosphate for 10 min and then kept on ice pending further processing.
Neutrophil Permeabilization-Neutrophils were electrically permeabilized according to a previously described method (Sä rndahl et al., 1989;Fä llman et al., 1992). The cells were kept on ice and rendered permeable by repeated exposures (150 s each) to an electrical field of 1.7 kV/cm. After permeabilization, the cells were immediately exposed to stimuli, as indicated in the figure legends.
Preparation of the Cytoskeletal Fractions-Cytoskeletal fractions were prepared using Triton X-100 (TX-100) as described by Sä rndahl et al. (1989,1993). The obtained cytoskeletal preparations were washed once, pelleted, and prepared for electrophoresis by resuspending the pellets in a sample buffer previously described (Sä rndahl et al., 1993).

SDS-PAGE and Immunoblot Detection of FPRs and G i2 ␣-Subunits-
Cytoskeletal preparations were solubilized in sample buffer and boiled for 5 min. Thereafter, the proteins of each sample (1 ϫ 10 6 cell equivalents per lane) were separated on a 7.5 or 10% SDS-polyacrylamide gel to detect FPRs and G i2 ␣-subunits, respectively. The proteins were then blotted onto a PVDF membrane as described previously (Sä rndahl et al., 1993). The membranes were blocked with 5% (w/v) bovine serum albumin in phosphate-buffered saline, pH 7.3, overnight at 4°C, and then with 0.5% (w/v) dried milk for 30 min at 37°C. The immunoblotting was performed by exposing the PVDF membranes to the different primary antibodies, after which a peroxidase-conjugated secondary antibody was added and the immune reaction was detected as enhanced chemiluminescence. When using the anti-FPR antibody, control experiments were performed by adding the antibody (1:300 dilution) to the FPR C-terminal peptide against which the antibody had been raised (10 g/ml) for 24 h at 4°C. This mixture was then added to the PVDF membrane, and the immune reaction was detected as described above. Densitometric analysis was performed with an UltroScan XL enhanced laser densitometer (LKB, Bromma, Sweden). (Sä rndahl et al., 1989), we used radiolabeled ligand to detect FPR and found that pretreatment of permeabilized cells with GDP␤S appeared to prevent the binding of ligand-receptor complexes to the cytoskeleton. However, it is possible that these results actually reflected a decreased binding of the radiolabeled ligand to the receptor and not a true reduction in the binding of the ligandreceptor complex to the cytoskeleton. On the other hand, our original interpretation is supported by the present finding that the association between the FPR and the cytoskeleton was inhibited in GDP␤S-treated neutrophils, as shown by immunoblot analysis (Fig. 1A). The affinity purified anti-FPR antibody we used detected a rather thin and distinct protein band at approximately 55 kDa, which is somewhat different to the protein band(s) that Sengelv et al. (1994) observed as a smear around 50 -70 kDa when photolabeling were used. The specificity of the binding of the antibody in the present study was confirmed by the finding that if the anti-FPR antibody was preincubated with the peptide against which the antibody was raised, this resulted in a failure to detect the 55-kDa protein band (data not shown). Although not addressed in the present study, it is possible that the association of FPRs with the cytoskeleton is somehow associated with a structural modification of the receptor. Using the same batch of cells, GDP␤S was also noted to have an inhibitory effect on the fMet-Leu-Phe-FIG. 1. Immunoblot analysis of the association of FPRs and G i2 ␣-subunits with the cytoskeleton in GDP␤S-pretreated neutrophils. The GDP␤S pretreatment was carried out as described previously (Sä rndahl et al., 1989). Immediately after permeabilization, the neutrophils were exposed to 1 mM GDP␤S for 10 min at 4°C. The samples were then transferred to a 37°C water bath, incubated for an additional 10 min, and subsequently stimulated with 20 nM fMet-Leu-Phe for 30 s. The stimulation was stopped by putting the cells on ice and simultaneously adding ice-cold TX-100-containing medium. Neutrophils termed "unstimulated" were permeabilized but not treated with GDP␤S or fMet-Leu-Phe. The 1st and 2nd lanes, respectively, show the cytoskeletal fractions of unstimulated and fMet-Leu-Phe-stimulated GDP␤S-treated neutrophils; for comparison, the 3rd lane shows the cytoskeletal fraction of fMet-Leu-Phe-stimulated permeabilized cells not treated with GDP␤S. The proteins were detected with a 1:300 dilution of the anti-FPR antibody (A) and a 1:2500 dilution of the anti-G i2 ␣-antibody (R16, 17) (B).

Effects of GDP␤S on the Association of FPRs and G i2 ␣-Subunits with the Cytoskeleton-In a previous study
induced release of the 40-kDa G i2 ␣-subunit from the cytoskeleton (Fig. 1B). Densitometer analysis of these blots revealed that 94 Ϯ 1% (mean Ϯ S.E., n ϭ 4) of the 40-kDa protein band remained associated with the cytoskeletal fraction in GDP␤Spretreated neutrophils after fMet-Leu-Phe stimulation in contrast to 23 Ϯ 4% (mean Ϯ S.E., n ϭ 5) when not pretreated with GDP␤S. Despite the effects of GDP␤S noted in the presence of fMet-Leu-Phe, GDP␤S alone was not found to have any effect on the binding of either FPRs or G i2 ␣-subunits to the cytoskeleton (data not shown).
Effects of GTP␥S and AlF 4 Ϫ on the Association of FPRs and G i2 ␣-Subunits with the Cytoskeleton-The association of FPRs with the cytoskeleton was further examined by incubating the cells with GTP␥S or AlF 4 Ϫ to activate the G-protein in a ligandindependent manner. As shown in Fig. 2A (2nd and 5th lanes), GTP␥S and AlF 4 Ϫ , respectively, induced an association between the receptor and the cytoskeleton, even in the absence of fMet-Leu-Phe. These effects were quite comparable with the association obtained when using fMet-Leu-Phe alone ( Fig. 2A, 3rd  and 6th lanes). Parallel immunoblot analysis also confirmed that both GTP␥S and AlF 4 Ϫ induced a release of G i2 ␣-subunits from the cytoskeletal fraction (Fig. 2B, 2nd and 5th lanes) that was quite similar to that observed upon stimulation with fMet-Leu-Phe alone (Fig. 2B, 3rd and 6th lanes, and Fig. 1B). Densitometer analysis revealed that only 22 Ϯ 7 and 26 Ϯ 6% (mean Ϯ S.E., n ϭ 6 and 5) of the 40-kDa protein band remained associated with the cytoskeletal fraction after GTP␥Sor AlF 4 Ϫ activation, respectively.

Effects of C5a on the Association of FPRs and G i2 ␣-Subunits
with the Cytoskeleton-The receptor for C5a, which is known to couple to G i2 proteins, caused a release of G i2 ␣-subunits from the cytoskeleton (Fig. 3B, 2nd lane) similar to that caused by fMet-Leu-Phe (Fig. 3B, 3rd lane, and Fig. 1B). Densitometer analysis revealed that only 33 Ϯ 5% (mean Ϯ S.E., n ϭ 4) of the 40-kDa protein band remained associated with the cytoskeletal fraction after C5a stimulation. Of even greater interest, C5a stimulation led to an association between FPRs and the cytoskeleton (Fig. 3A, 2nd lane). Since these findings were obtained with a natural ligand that is known to activate G i2 proteins, they confirm the results gained by direct manipulation of the G i2 proteins with GTP␥S and AlF 4 Ϫ (Fig. 2). Temporal Changes in the Association of FPRs and G i2 ␣-Subunits with the Cytoskeleton-To examine the kinetics of the release of G i2 ␣-subunits and the interaction of FPRs with the cytoskeleton, neutrophils were stimulated with fMet-Leu-Phe at 15°C. At that temperature, the receptor is converted into a high affinity form and becomes associated with the cytoskeletal fraction but is not internalized Sklar et al., 1984); in contrast, internalization occurs rapidly at 37°C (t 1 ⁄2 15-20 s; Janeczek et al., 1989). Only very few FPRs could be detected in the cytoskeletal fraction of unstimulated neutrophils (Fig. 4A, inset, 1st lane). The fMet-Leu-Phe-induced association of FPR with the cytoskeletal fraction occurred rapidly and then leveled off (Fig. 4A, and inset, 2nd through 4th lanes). Concurrently, the number of 40-kDa G i2 ␣-subunits associated with the cytoskeletal fraction of unstimulated neutrophils (Fig.  4B, inset, 1st lane) decreased upon stimulation with fMet-Leu-Phe. Comparison of the cellular handling of FPRs and G i2 ␣-subunits reveals that the fMet-Leu-Phe-induced association of FPRs with the cytoskeleton precedes the release of G i2 ␣-subunits from the same cellular fraction (Fig. 4). DISCUSSION Using the immunoblot technique and antibodies directed against the FPR or the G i2 ␣-subunit, we studied the interdependence of the binding of chemotactic peptide receptors to the cytoskeleton and the activation and release of G i2 ␣-subunits from the cytoskeleton. Exposing neutrophils to GDP␤S, a GDP analogue that keeps the G-protein in an inactivated state, prevented the chemotactic peptide-induced release of G i2 ␣-subunits from the cytoskeletal fraction and also inhibited the association of FPRs with the cytoskeleton. These findings suggest that binding of FPRs to the cytoskeleton is regulated by a G-protein, as previously proposed (Sä rndahl et al., 1989). In the cited investigation, we used a radiolabeled ligand to detect FPRs when measuring the effects of GDP␤S; hence, it is possible that the results actually reflect a reduced binding affinity between the ligand and its receptor. This alternative interpretation is based on the fact that both GDP␤S and GTP␥S cause a concentration-dependent reduction in the affinity of the FPR for its ligand (Koo et al., 1983;Posner et al., 1994). However, in the present study we employed an immunoblot technique to directly detect the receptor protein itself and also found that GDP␤S and GTP␥S affected the FPR cytoskeletal association in completely different ways. We are therefore convinced that

FIG. 2. Effects of GTP␥S and AlF 4
؊ on the association of FPRs and G i2 ␣-subunits with the cytoskeleton. Activation with GTP␥S and AlF 4 Ϫ was carried out as described previously (Sä rndahl et al., 1993). Neutrophils were exposed to AlF 4 Ϫ (10 M AlCl 3 ϩ 20 mM NaF) for 20 min at 37°C or, immediately after permeabilization, to 100 M GTP␥S for 10 min at 4°C and then transferred to a water bath and incubated at 37°C for an additional 10 min. The stimulation was stopped by putting the cells on ice and simultaneously adding ice-cold TX-100-containing medium. Unstimulated indicates cytoskeletal fractions of permeabilized or unpermeabilized neutrophils not treated with GTP␥S or AlF 4 Ϫ (1st and 4th lanes). 2nd and 5th lanes show the cytoskeletal fractions of GTP␥Sand AlF 4 Ϫ -treated neutrophils, respectively. For comparison, 3rd and 6th lanes show the cytoskeletal fraction of unpermeabilized fMet-Leu-Phe-stimulated cells. The proteins were detected with a 1:300 dilution of the anti-FPR antibody (A) and a 1:2500 dilution of the anti-G i2 ␣-antibody (R16,17) (B).

FIG. 3. Immunoblot analysis of the association of FPRs and G i2
␣-subunits with the cytoskeleton in C5a-stimulated neutrophils. The cells were preincubated for 5 min at 37°C and then stimulated with 20 nM C5a for 30 s. The stimulation was stopped by putting the cells on ice and simultaneously adding ice-cold TX-100-containing medium. Neutrophils termed unstimulated were subjected to the 5-min equilibration period at 37°C but were not stimulated (1st lane). 2nd lane shows the cytoskeletal fractions of C5a-treated neutrophils. For comparison, the 3rd lane shows the cytoskeletal fraction of fMet-Leu-Phestimulated cells. The proteins were detected with a 1:300 dilution of the anti-FPR antibody (A) and a 1:2500 dilution of the anti-G i2 ␣-antibody (R16,17) (B). the major effect of GDP␤S, in this context, is to inhibit the association of FPR with the cytoskeleton.
The impaired association of FPR with the cytoskeleton obtained by incubating the cells with GDP␤S was obtained in the presence of the commonly used ligand fMet-Leu-Phe. On the other hand, and perhaps of greater interest in the present study, both GTP␥S and AlF 4 Ϫ were found to induce association of FPRs with the cytoskeleton, even in the absence of any natural FPR ligand ( Fig. 2A). This means that association of the FPR with the cytoskeleton requires neither the binding of a natural ligand nor the immediate conformational change that occurs in the receptor as a result of that protein-protein interaction. Furthermore, these findings are in agreement with a regulatory role of G-proteins in the process of FPR cytoskeleton association. The fact that stimulation with either GTP␥S or AlF 4 Ϫ triggers an association between FPRs and the cytoskeleton and a parallel release of G i2 ␣-subunits from the cytoskeleton (Fig. 2B) suggests that the FPR cytoskeletal association might be related to an activation-dependent release of G i2 ␣-subunits from the cytoskeleton. Analysis of the time dependence of both the binding of FPRs to and the release of G i2 ␣-subunits from the cytoskeleton (at 15°C) revealed that the former association of FPRs occurred faster than the latter release of G i2 ␣-subunits. This suggests that a mechanism more complex than a simple protein-for-protein exchange is involved in the FPR cytoskeletal association; however, such a conclusion might be uncertain due to the difference in number of FPRs and G i2 ␣-subunits.
In this context, it should be mentioned that receptor phosphorylation is another step that is most likely involved in desensitization of the FPR, as has previously been demonstrated in the visual system (Wilden et al., 1986) and in the ␤-adrenergic receptor system (Lefkowitz et al., 1990). This supposition is based on the finding that the FPR in differentiated HL60 cells undergoes phosphorylation soon after stimulation (t 1 ⁄2 around 1 min, Tardif et al., 1993) and that there is a correlation between the phosphorylation and the desensitization of this receptor (Ali et al., 1993). Recently, it has also been suggested based on in vitro experiments that the FPR might be phosphorylated by a cytosolic kinase that has features similar to those of GRK2, a G-protein-coupled receptor kinase (Prossnitz et al., 1995). The present finding that G-protein activation could be an initial step in the association of FPRs with the cytoskeleton is compatible with the idea that phosphorylation of the FPR is a subsequent step in the desensitization process; this is true because an initial activation of G-proteins is known to activate several downstream protein kinases. A G-protein-induced phosphorylation of the FPR may be required to allow this receptor to become associated with the cytoskeleton and segregated from its G i2 protein (which at the same time is released from the cytoskeleton). If this is the case, then it is not surprising that desensitization had no effect on the affinity in FPR-G-protein interactions as seen in reconstitution experiments (Klotz and Jesaitis, 1994). Additional experiments are needed, however, to determine whether phosphorylation of FPR precedes or follows the association of FPR with the cytoskeleton.
Desensitization of the N-formyl peptide receptor was demonstrated several years ago in a number of laboratories (Showell et al., 1979;Jesaitis et al., 1986). Notwithstanding, the molecular basis of this process is still unknown, although both homologous and heterologous desensitization appear to be involved (Didsbury et al., 1991;McLeish et al., 1989). Interestingly, Didsbury and co-workers (1991) proposed that a third type of desensitization, referred to as "class desensitization," also exists. These researchers observed that two chemotactic receptors in human neutrophils, namely the receptors for fMet-Leu-Phe and C5a, which are both coupled to the G i2 protein, are able to desensitize each other but that the ␣ 1adrenergic receptor, which is coupled to a different G-protein (i.e. G q ), is not affected by pre-exposure to either fMet-Leu-Phe or C5a. Our experiments show that the FPR associates with the cytoskeleton not only when it is activated by its ligand but, more importantly, also when it is activated by another chemotactic stimulus, namely C5a. These findings are supported by the fact that the FPR also becomes associated with the cytoskeleton when the G i2 protein is directly activated by GTP␥S or AlF 4 Ϫ in the absence of a ligand. Taken together, these FIG. 4. Time correlation between the interactions of G i2 ␣-subunits and FPRs with the cytoskeleton. Neutrophils were preincubated for 5 min at 15°C and then stimulated with 20 nM fMet-Leu-Phe. At different time points (abscissa), cells were withdrawn and added to ice-cold TX-100-containing medium, as described previously (Sä rndahl et al., 1989). The graphs show the relative amount of FPR (A) and G i2 ␣-subunits (B) in the cytoskeletal fractions of fMet-Leu-Phe-stimulated neutrophils, as revealed by densitometric analysis of immunoblots obtained by using either the anti-FPR antibody (a 1:300 dilution) or the anti-G i2 ␣-antibody R16, 17 (a 1:2500 dilution). Data are given as mean Ϯ S.E. for four separate experiments and expressed as percent of the maximum value for each individual experiment. The insets show representative immunoblots of the cytoskeletal association of both FPRs (A) and G i2 ␣-subunits (B). results agree with the idea of class desensitization (Didsbury et al., 1991) and suggest that the FPR can be desensitized by other receptors at the G-protein level. Moreover, in the present study stimulation with C5a released the ␣-subunits of its transducing G-protein (i.e. G i2 protein) from the cytoskeleton. This finding provides additional support to the hypothesis that cellular segregation of FPRs and G i2 proteins is an essential part of the mechanism underlying the termination and/or desensitization of FPR signaling properties (Jesaitis et al., 1989; for review see Jesaitis 1992). Furthermore, the previous observation that fMet-Leu-Phe-stimulation causes a selective release of G i2 ␣-subunits without affecting the association of G s ␣-subunits to the cytoskeleton (Sä rndahl et al., 1993) provides at least a partial explanation for the selectivity of class desensitization that is indicated in the model proposed by Didsbury and co-workers (1991).