N-formyl peptide receptors cluster in an active raft-associated state prior to phosphorylation.

In response to ligand binding, G protein-coupled receptors undergo phosphorylation and activate cellular internalization machinery. An important component of this process is the concentration of receptors into clusters on the plasma membrane. Aside from organizing the receptor in anticipation of internalization, little is known of the function of ligand-mediated G protein-coupled receptor clustering, which has traditionally been thought of as being a phosphorylation-dependent event prior to receptor internalization. We now report that following receptor activation, the N-formyl peptide receptor (FPR) forms distinct membrane clusters prior to its association with arrestin. To determine whether this clustering is dependent upon receptor phosphorylation, we used a mutant form of the FPR, DeltaST-FPR, which lacks all phosphorylation sites in the carboxyl-terminal domain. We found that activation of the signaling-competent DeltaST-FPR resulted in rapid receptor clustering on the plasma membrane independent of Gi protein activation. This clustering required receptor activation since the D71A mutant receptor, which binds ligand but is incapable of transitioning to an active state, failed to induce receptor clustering. Furthermore we demonstrated that FPR-mediated clustering and signaling were cholesterol-dependent processes, suggesting that translocation of the active receptor to lipid rafts may be required for maximal signaling activity. Finally we showed that FPR stimulation in the absence of receptor phosphorylation resulted in translocation of FPR to GM1-rich clusters. Our results demonstrate for the first time that formation of a clustered activated receptor state precedes receptor phosphorylation, arrestin binding, and internalization.

In response to ligand binding, G protein-coupled receptors undergo phosphorylation and activate cellular internalization machinery. An important component of this process is the concentration of receptors into clusters on the plasma membrane. Aside from organizing the receptor in anticipation of internalization, little is known of the function of ligand-mediated G proteincoupled receptor clustering, which has traditionally been thought of as being a phosphorylation-dependent event prior to receptor internalization. We now report that following receptor activation, the N-formyl peptide receptor (FPR) forms distinct membrane clusters prior to its association with arrestin. To determine whether this clustering is dependent upon receptor phosphorylation, we used a mutant form of the FPR, ⌬ST-FPR, which lacks all phosphorylation sites in the carboxylterminal domain. We found that activation of the signaling-competent ⌬ST-FPR resulted in rapid receptor clustering on the plasma membrane independent of G i protein activation. This clustering required receptor activation since the D71A mutant receptor, which binds ligand but is incapable of transitioning to an active state, failed to induce receptor clustering. Furthermore we demonstrated that FPR-mediated clustering and signaling were cholesterol-dependent processes, suggesting that translocation of the active receptor to lipid rafts may be required for maximal signaling activity. Finally we showed that FPR stimulation in the absence of receptor phosphorylation resulted in translocation of FPR to GM1-rich clusters. Our results demonstrate for the first time that formation of a clustered activated receptor state precedes receptor phosphorylation, arrestin binding, and internalization.
A central question in understanding the process of cellular activation is defining the events that localize and restrict signaling activity at the membrane. This is of particular importance in chemotactic cells where cell polarization dictates the direction of movement. Activation of cell surface receptors initiates cellular signaling pathways that ultimately control cell polarity and migration. Polarization correlates with the redistribution of distinct lipid raft domains and their associated signaling molecules to the leading and trailing edges of the cell (1). However, at what point and even whether chemotactic receptors translocate into raft domains remain controversial (1,2).
Chemoattractant G protein-coupled receptors (GPCRs) 1 are seven transmembrane receptors that activate numerous cellular functions in part through the stimulation of heterotrimeric GTP-binding (G) proteins. Agonist binding to the GPCR triggers a conformational change within the receptor that facilitates G protein binding and activation. This binding leads to exchange of GDP for GTP on the G␣ subunit of the G protein followed by the dissociation of the ␣ subunit from the dimeric ␤/␥ subunit. These G protein subunits then regulate the activity of multiple cellular effectors such as adenylyl cyclase, phosphatidylinositol 3-kinase, ion channels, and phospholipase C (3). Within seconds to minutes of agonist activation, the receptor becomes phosphorylated by GPCR kinases; the phosphorylated receptor in turn promotes the binding of arrestins. While four isoforms of arrestin have been identified, only two isoforms, arrestin-2 (␤-arrestin) and arrestin-3 (␤-arrestin-2), are ubiquitously expressed in mammalian cells (3). Binding of arrestins to phosphorylated receptors can inhibit further association of G proteins and thereby inactivate classical receptormediated signaling, a process termed desensitization (4). In addition, arrestins can function as adapters to mediate receptor internalization via clathrin-coated pits (5) and to recruit effectors such as Src and mitogen-activated protein kinases, resulting in G protein-independent signaling (6,7).
Understanding the functional consequences of ligand-induced GPCR activation, regulation, and trafficking is complicated by the fact that multiple processes occur simultaneously in the cell following agonist stimulation. These include G protein coupling, cellular activation, receptor phosphorylation, binding of arrestins and associated proteins, receptor cluster-ing, internalization, intracellular trafficking, and recycling of receptors to the cell surface. In particular, understanding the role of receptor clustering, phosphorylation, and arrestin binding in receptor dynamics is a distinct challenge. Activation of many receptors, including GPCRs, in the plasma membrane is thought to be regulated by the trafficking of the receptors into or out of discrete detergent-insoluble, glycolipid-rich microdomains termed rafts (8 -12). These rafts are enriched with cholesterol and sphingomyelin, often contain the protein caveolin, and contribute structural dimension to the membrane. Since the plasma membrane is a two-dimensional, fluid phase structure, the membrane domains where receptor clusters form are thought to control membrane organization during receptor internalization (11). Given the heterogeneity among rafts (11), it is unclear, however, whether receptor clustering may occur solely to initiate internalization of activated receptors or whether it serves a distinct function in signaling by conformationally activated states of the receptor.
Recently the roles of arrestins in GPCR function have been more definitively examined through the use of mouse embryonic fibroblasts (MEFs) that are deficient in either one or both arrestins. Signaling and trafficking of numerous GPCRs, including the ␤ 2 adrenergic receptor, angiotensin 1A receptor, protease-activated receptor, and N-formyl peptide receptor (FPR) (13)(14)(15), have been investigated in these arrestin-deficient cells. Whereas the ␤ 2 adrenergic receptor and angiotensin 1A receptor display a dependence on arrestin for receptor internalization, the protease-activated receptor and FPR do not. However, in wild type cells following ligand activation, the FPR colocalizes with arrestins predominantly on intracellular endosomes. Yet until recently a function for arrestin in the trafficking of the FPR has remained elusive (16). Recent studies by our laboratory have, however, revealed a role for arrestin in the intracellular trafficking and recycling of the FPR (14). In these studies, MEFs derived from knock-out mice that lacked both arrestin-2 and arrestin-3 were found to internalize the FPR but subsequently mistraffic the FPR, leading to a loss of receptor recycling (14). Furthermore a mutant preactivated arrestin, arrestin 3A, which binds to GPCRs in the absence of receptor phosphorylation, was found to inhibit FPR recycling, confirming a role for arrestins in the intracellular trafficking and recycling of GPCRs. 2 These results lead to the possibility that FPR clustering, a requisite event in receptor internalization, occurs in the absence of arrestin binding and therefore possibly prior to receptor phosphorylation.
In the present study, we investigated whether the stimulation of the FPR would promote redistribution of the receptor in the absence of receptor phosphorylation and arrestin binding. To determine whether receptor clustering represents a unique intermediate involved in cellular activation distinct from internalization, we used a non-phosphorylatable mutant FPR (17). We observed that the FPR was rapidly clustered on the plasma membrane following ligand binding independently of receptor phosphorylation. This clustering required receptor activation as an FPR mutant that binds ligand but is incapable of transitioning to an activated state failed to cluster. Conversely G protein activation was not required to induce clustering since pertussis toxin failed to prevent FPR clustering. Furthermore we demonstrated that receptor clustering and signaling were cholesterol-dependent processes independent of the phosphorylation state of the receptor. In addition, cholesterol depletion inhibited signaling, clustering, and internalization of the wild type FPR. Finally we showed that the FPR colocalized with GM1 glycolipid-rich raft domains following receptor activation. As many signal transduction components, including G proteins and downstream effectors, are known to reside in rafts (11), we conclude that the FPR in its activated state translocates to signaling domains to effect cell activation prior to receptor phosphorylation and internalization.
Cell Lines and cDNA-The cDNA encoding the FPR was obtained from a human HL-60 granulocyte library (20). The ⌬ST-FPR mutant in which all 11 of the serine and threonine residues have been replaced (S319A, T325G, S328A, T329A, T331A, S332G, T334G, T336G, S338G, T339A, and S342G) (21) and the D71A mutant (22) have been described previously. The D71A/⌬ST-FPR chimeric mutant was generated by digesting both the ⌬ST and D71A FPR plasmids with BclI, ligating the appropriate fragments, and sequencing the resulting construct to ensure the presence of both mutations. Arresin-2 and -3-green fluorescent protein (GFP) were kindly provided by Dr. Jeffrey Benovic (Thomas Jefferson University). Arrestin-3-mRFP1 was generated by substitution of the GFP cDNA in arrestin-3-GFP with a PCR-amplified cDNA of mRFP1 generously provided by Dr. Roger Tsien (23). Plasmid DNA was stably transfected into U937 cells using Effectene (Qiagen). Briefly cells were plated overnight, transfected with 1 g of plasmid DNA, and then selected for 14 -21 days in G418. Surviving cells were pooled, and the expression levels were analyzed by flow cytometric analysis of 10 nM 6-pep-FITC binding. U937 cell lines were maintained at 37°C in 5% CO 2 in RPMI supplemented with 2 mM glutamine, 100 units/ml penicillin, 100 g/ml streptomycin, 10 mM HEPES (pH 7.4), 10% heatinactivated fetal bovine serum.
Receptor Internalization-To examine the internalization of receptor in response to ligand stimulation, U937 cells stably expressing a wild type or mutant FPR were resuspended in serum-free RPMI and allowed to incubate at 37°C for 10 min prior to stimulation with ligand. Cells were then stimulated with 1 M fMLF for the indicated time periods. Plunging the cells in ice cold RPMI and incubation on ice for at least 15 min arrested the internalization. Cells were washed three times in cold Hanks' buffered saline solution to remove uninternalized ligand. Receptors remaining on the cell surface were labeled in the presence of 10 nM 6-pep-FITC and analyzed for fluorescence intensity on a FACScan flow cytometer (BD Biosciences). Only viable cells were included in the assay as determined by a gate on forward and side scatter. Nonspecific binding was determined in the presence of 1 M fMLF. Receptor internalization was expressed relative to the total number of receptors on untreated cells.
Colocalization Studies-U937 cells were transiently transfected by electroporation as follows. U937 cells stably expressing the wild type or a mutant of the FPR were pelleted and resuspended in serum-free RPMI at a concentration of 2 ϫ 10 7 cells/ml. Cells (8 ϫ 10 6 ) were transferred to an electroporation cuvette (0.4-mm gap) and pulsed (200 V/2000 microfarads for an exponential decay (t1 ⁄2 ) of ϳ50 ms) with 25 g of DNA. Cells were allowed to recover for 10 min before being transferred to 10 ml of complete RPMI. Following electroporation cells were incubated for 12-16 h at 37°C in 5% CO 2 .
To examine the response of FPR-expressing U937 cells to stimulation with formyl peptides, cells were stimulated with ALEXA 546-6-pep for the indicated time points and plunged immediately into 1 volume of ice-cold 2% paraformaldehyde for 15 min. Cells were subsequently pelleted and resuspended in 2% paraformaldehyde for an additional 15 2 T. A. Key and E. Prossnitz, submitted for publication. min on ice. For immunofluorescence staining following fixation, the cells were stained with anti-flotillin antibodies and rinsed three times in 1ϫ phosphate-buffered saline. The cells were then labeled with a secondary FITC-conjugated anti-rabbit antibody. Cells were pelleted and resuspended in Vectashield, transferred to a glass slide, and examined by confocal microscopy.
For MEF cell lines, cells were plated on glass coverslips overnight in complete medium at 37°C in 5% CO 2 and then transiently transfected (LipofectAMINE 2000) with pEGFP (Clontech) fused to the carboxyl terminus of wild type FPR, the ⌬ST mutant of FPR, the D71A mutant of FPR, or the ␤ 2 adrenergic receptor. Plated cells were stimulated with ALEXA 546-6-pep for the indicated time periods, and coverslips were flooded with 1 volume of ice-cold 2% paraformaldehyde. Following incubation of at least 15 min on ice, fresh 2% paraformaldehyde was added for an additional 15 min on ice. Cells were mounted in Vectashield, transferred to a glass slide, and examined by confocal microscopy.
Confocal images were acquired at room temperature using a Zeiss LSM510 system equipped with argon and HeNe lasers for excitation at 488 and 543 nm. Samples were viewed with the 40 ϫ 1.3 oil immersion objective lens.
Calcium Mobilization-To assess functional signaling, calcium mobilization in response to ligand stimulation was measured as described previously (21). Briefly cells were harvested by centrifugation, washed once in phosphate-buffered saline, and resuspended at 5 ϫ 10 6 cells/ml in Hanks' buffered saline solution. The cells were incubated with 5 M Indo1-AM for 30 min at 37°C, washed once with Hanks' buffered saline solution, resuspended at 10 6 cells/ml in Hanks' buffered saline solution, and stored on ice. Cells were allowed to equilibrate at 37°C for 2 min, stimulated with fMLF, and monitored by continuous fluorescence measurement using a Quantimaster QM 2000-6 spectrofluorometer (Photon Technologies International) detecting at 400 and 490 nm.
Cholesterol Depletion/Reconstitution-Cells were incubated in the presence of 10 mM methyl-␤-cyclodextrin (M␤CD) at 37°C for 30 min to deplete the cells of cholesterol. To reconstitute cellular cholesterol following cholesterol depletion, cholesterol was restored by incubation with 10 mM water-soluble cholesterol for 30 min at 37°C. For calcium measurements, cells were loaded with Indo1-AM as described above following treatments with M␤CD. Cholesterol-depleted or cholesterol-reconstituted cells were assayed for calcium mobilization as described above.

RESULTS
Activation of seven transmembrane receptors is initiated by ligand binding followed by receptor clustering and internalization. As a result of receptor phosphorylation, arrestins bind to receptors and contribute to the termination of the signaling event. It is unclear whether the clustering of GPCRs at sites of internalization is dependent upon arrestin binding or whether receptor clustering takes place prior to the recruitment of arrestins. In the case of the CCR5 receptor (24) and the well characterized ␤ 2 adrenergic receptor arrestin binding to the receptor mediates interactions with internalization machinery (clathrin and the adapter AP-2), suggesting that arrestin is directly involved in the recruitment and clustering of receptors to clathrin-coated pits (25). Whether clustering of the ␤ 2 adrenergic receptor can take place in the absence of arrestins is unknown. However, as the FPR can internalize in the absence of arrestins (14), it should therefore be capable of clustering in the absence of arrestin binding.
To determine whether FPR clustering is a distinct event from receptor internalization and whether the binding of arrestins precedes or follows FPR clustering in cells that express endogenous arrestins, we examined the kinetics of FPR clustering simultaneously with membrane localization of arrestin-2 or arrestin-3 in the U937 promonocytic leukemia cell line. Following incubation with fluorescently labeled ligand at 37°C, the activated receptors began to form distinct clusters within 20 -40 s that persisted as the receptor internalized (Fig.  1, A and B). In contrast, significant colocalization of arrestin did not take place until 3-4 min after receptor stimulation well after the t1 ⁄2 for internalization (ϳ2 min, Fig. 2). These results suggest that substantial arrestin binding may not occur until more than 50% of the receptor is internalized and that arrestin binding may not be required to promote clustering of the receptor or internalization.
To determine whether receptor clustering took place in response to receptor phosphorylation, we used a mutant form of the FPR, ⌬ST-FPR, which lacks all of the serine and threonine residues in the carboxyl-terminal domain that are normally phosphorylated in response to receptor activation. This mutant, when activated, fails to bind arrestins or become desensitized as assessed by calcium mobilization (17). Characterization of this receptor in U937 cells revealed that while the FPR has a t1 ⁄2 for internalization of ϳ2 min, the ⌬ST-FPR failed to internalize (Fig. 2) as we have demonstrated previously (17). Similar to the wild type receptor, the ⌬ST-FPR mutant receptor formed distinct clusters on the membrane following exposure to ligand for ϳ2 min (Fig. 3, A and B). In contrast to the wild type FPR, the ⌬ST-FPR exhibited no colocalization with arrestin even at extended times. The somewhat slower kinetics of ⌬ST-FPR clustering as compared with wild type FPR clustering could be the result of wild type receptor phosphorylation, which may serve to accelerate its clustering and internalization. These results suggested that even in the absence of a phosphorylated tail and receptor internalization the FPR is capable of forming clusters within the plasma membrane.
We next questioned whether G protein activation was required for clustering of the FPR or whether clustering occurred as a function of receptor activation prior to the activation of G proteins. To this end, we pretreated ⌬ST-FPR cells with pertussis toxin, which leads to ADP-ribosylation of the G␣ i subunit of the heterotrimeric G protein, rendering the G protein unable to interact with and be stimulated by ligand-activated GPCRs. Under these conditions, stimulation of the wild type and ⌬ST-FPR resulted in receptor clustering (Fig. 4, A and B). The effectiveness of the pertussis toxin in completely blocking fMLF-induced calcium mobilization was determined in parallel (data not shown). Therefore, we concluded that G protein activation and therefore cellular signaling were not required for clustering of the FPR.
Since G protein activation and the resulting downstream signaling events are not required to induce receptor clustering, we postulated that the FPR in its ligand-bound, conformationally active state was capable of clustering in the membrane. To determine whether conversion of the FPR to an activated state was required to induce clustering, we examined the clustering of the D71A FPR mutant stably expressed in U937 cells. Asp-71 of the FPR is located in the second transmembrane domain and represents a conserved site within many GPCRs believed to be intimately involved in receptor activation (26,27). While the D71A FPR mutant is capable of binding fMLF, it does not bind to or activate G proteins and does not undergo ligand-mediated phosphorylation, arrestin binding, or internalization (22,26,27). It therefore bears all the hallmarks of a receptor trapped in the inactive conformation. Examination of the membrane distribution of the ligand-bound D71A mutant revealed that the receptor failed to form clusters on the membrane, remaining diffuse over the entire cell membrane (Fig. 4C). This suggested that receptor activation was indeed essential for FPR clustering. It should be noted that there existed some minor heterogeneity in the distribution of the ligand-bound D71A mutant, but this was similar to the extent observed in unstimulated wild type and mutant forms of the FPR.
To confirm that the specific loss of serine and threonine residues or alternatively the introduction of alanine and glycine residues in their place in the carboxyl-terminal domain of the ⌬ST-FPR did not in and of itself introduce receptor clustering, we examined the membrane localization of a chimeric receptor containing both the D71A and ⌬ST-FPR mutations. We hypothe-sized that if the ⌬ST mutation did not create a dominant effect, the addition of the D71A mutation should prevent the conformational activation of the ⌬ST-FPR and thus prevent receptor clustering. As predicted, the D71A/⌬ST-FPR did not cluster following stimulation with fluorescently labeled ligand (Fig. 4D), demonstrating a localization pattern indistinguishable from the D71A mutant (cf. Fig. 4C). This further indicated that clustering of the receptor results specifically from a transition of the ligand-bound receptor to an activated state.
Since transient interactions with arrestins may still contribute to the clustering of the ⌬ST-FPR in U937 cells, we investigated the ability of the ⌬ST-FPR to cluster in arrestin-deficient cells. In addition to leukocytes, other cell types such as human fibroblasts have been shown to express an endogenous FPR (28). Therefore we used MEFs, which lack expression of both arrestin-2 and arrestin-3 (MEF arr2 Ϫ/Ϫ arr3 Ϫ/Ϫ ) to assess the clustering of the FPR in the absence of arrestins. Use of these cells would also confirm that FPR clustering was not unique to myeloid cells. Arrestin-deficient MEFs internalize the wild type FPR with kinetics similar to those of arrestinexpressing cells but fail to internalize the ⌬ST receptor, demonstrating that in MEFs, receptor phosphorylation is also required for FPR internalization (14). To examine the ability of the MEF arr2 Ϫ/Ϫ arr3 Ϫ/Ϫ cells to induce clustering of activated receptors, we transiently transfected these cells with wild type and ⌬ST and D71A mutant forms of the FPR fused to the enhanced GFP. Whereas both the stimulated wild type FPR-GFP and the ⌬ST-GFP mutant were capable of forming clusters in the absence of arrestin expression, the D71A-GFP mutant exhibited no detectable clustering upon activation (Fig.  5A) similar to what we observed in U937 cells.
Arrestins are nonetheless believed to be required for clustering of many GPCRs, such as the ␤ 2 adrenergic receptor, during internalization through clathrin-coated pits. To assess whether clustering of the ␤ 2 adrenergic receptor could also occur in the absence of arrestins, arrestin-deficient MEFs expressing the ␤ 2 adrenergic receptor as a GFP fusion (␤ 2 AR-GFP) were stimulated with isoproterenol. We found that the ␤ 2 adrenergic receptor failed to form membrane clusters comparable to those formed by the ⌬ST or wild type FPR, remaining uniformly distributed on the plasma membrane (Fig. 5B). Reconstitution of arrestin expression with arrestin-3-mRFP1 restored internalization and therefore clustering of the ␤ 2 adrenergic receptor in arrestin-deficient MEFs. Therefore, we concluded that unlike the arrestin-mediated clustering of the ␤ 2 adrenergic receptor, clustering of the FPR in fibroblasts as well as myeloid cells is independent of arrestin association.
Once the receptor is organized into clusters, we hypothesized that ligand binding and thus receptor activation might be required to maintain localization of the receptor in these domains. To address this question we used t-Boc, a competitive antagonist of the FPR, to determine whether dissociation of agonist from the receptor-binding pocket disrupts receptor clustering in U937 cells. U937 cells expressing the ⌬ST-FPR fused to enhanced GFP were incubated in the presence of fluorescent ligand for 8 min, which provided ample time for clusters to form, at which time the ligand was prevented from rebinding upon dissociation by the addition of 2 M t-Boc. The cells were incubated at 37°C for an additional 6 min to allow time for inactive receptor to disperse from the clusters (Fig.  6A). Displacement of the ligand led to diminished clustering of the receptor, suggesting that once the receptors were in the clustered state, loss of ligand from the receptor led to a loss of the active receptor conformation and the concomitant loss of the receptor from the cluster. While weak clusters persisted in some cells, we observed that these clusters appeared to be due to incomplete displacement of the labeled fMLF from the receptors by the t-Boc. These results indicate that the translocation of the FPR into and out of clusters is a dynamic process potentially related to the signaling capacity of the receptor.
Removal of agonist from the receptor should result in a decrease in signaling potential and thus may suggest that receptor-mediated signaling is related to the ability of the receptor to cluster. We have demonstrated previously that both the FPR and ⌬ST-FPR are capable of mobilizing calcium in response to ligand binding (17). Therefore, to determine whether loss of ligand binding was sufficient to disrupt signaling, we examined the kinetics of calcium mobilization in cells following displacement of ligand with t-Boc (Fig. 6B). While intracellular calcium fluxes decayed slowly in stimulated ⌬ST-FPR cells over time, addition of 2 M t-Boc initiated a rapid decline in intracellular calcium concentrations consistent with the known dissociation rate of fMLF (t1 ⁄2 , ϳ5 s (29)). As would be expected, this result confirms that cellular signaling via calcium mobilization is proportional to receptor occupancy and that prolonged calcium signaling requires continued receptor occupancy. The slower decline in the calcium response from fMLF-stimulated cells not treated with t-Boc is likely due to cellular homeostasis mechanisms downstream from the receptor that serve to moderate intracellular calcium responses over time. From these observations we conclude that, in the absence of agonist binding, inactive receptors repartition out of membrane clusters/rafts. It should be noted that both the initial rise in intracellular calcium concentrations following fMLF stimulation and the subsequent decline upon t-Boc addition occur more quickly than the microscopically observed clustering and declustering of the FPR. This suggests that the size of the FPR clusters at early times following activation may be below the resolution of confocal microscopy, representing nanodomains (30), only yielding larger observable clusters at later time points.
Membrane lipid composition and structure are known to play an important role in the signal transduction capacity of many receptors, including some GPCRs (11,31). To examine whether the classical raft component cholesterol is involved in the clustering and signaling of the FPR, we used M␤CD to deplete cells of membrane cholesterol. Depletion of cholesterol from the membrane abolished clustering of the ⌬ST-FPR (Fig. 7A). Since M␤CD treatment can lead to additional perturbations of cell membranes in addition to the removal of cholesterol (32), we replaced membrane cholesterol in M␤CD-treated cells by incubation with cholesterol-loaded M␤CD to test for nonspecific effects of M␤CD treatment. Restoration of cholesterol resulted in the re-establishment of FPR clustering comparable to completely untreated cells (Fig. 7B). To determine whether cholesterol depletion and the associated loss of FPR clustering affected the ability of the receptor to mediate calcium mobilization, cells were stimulated, and the elevation of intracellular calcium was compared with untreated U937 cells expressing the ⌬ST-FPR. Depletion of membrane cholesterol significantly abrogated calcium mobilization in response to ligand stimulation (Fig. 7C). Restoration of membrane cholesterol resulted in a significant but incomplete restoration of calcium signaling. These results suggested that cholesterol, an essential component of rafts, is required to initiate calcium mobilization by the ⌬ST-FPR as well as mediate the clustering of the receptor.
The effect of cholesterol depletion on ⌬ST-FPR clustering and function raised the question as to the effects of cholesterol depletion on the wild type FPR, which normally undergoes internalization following activation. In particular, if receptor clustering is inhibited, receptor internalization is likely to be much less efficient or absent. We tested this by treating U937 cells expressing the wild type FPR with M␤CD as we did for the ⌬ST-FPR. Cholesterol depletion reduced FPR internalization by ϳ50% compared with untreated cells (Fig. 8A). Restoration of membrane cholesterol completely restored the rate and magnitude of receptor internalization. Surprisingly, while fMLF could bind and induce limited internalization of the FPR, the receptor failed to cluster to an observable degree (Fig. 8B), although receptor clustering could again be restored by reconstituting the cells with cholesterol. These results suggest that the FPR internalization that occurs under cholesterol-limiting conditions may result in receptor-containing endosomes whose size is beyond the resolution of light microscopy. In addition, although arrestin recruitment to the internalized FPR occurs in cholesterol-replete cells, arrestin translocation to the membrane of cholesterol-depleted cells was not observed.
To confirm that signaling initiated by the wild type FPR required cholesterol and therefore receptor clustering, as observed for the ⌬ST-FPR, we assessed M␤CD-treated U937 FPR cells for their ability to mobilize calcium. While untreated and cholesterol-replenished cells were capable of mobilizing calcium, loss of membrane cholesterol abrogated the calcium mo- bilization in response to ligand stimulation (Fig. 8C). These results suggested that cholesterol and therefore raft association are required to initiate calcium mobilization by the FPR as well as to mediate the clustering of the receptor and as a result receptor internalization.
Cholesterol-and sphingolipid-containing rafts are known to concentrate the ganglioside GM1, which as a result is a well characterized marker for rafts. To determine whether the clustering of the FPR occurred in such rafts, we examined the colocalization of the ⌬ST-FPR with a fluorescently labeled GM1. ⌬ST-FPR cells were incubated with fluorescently labeled ALEXA 488-GM1 and then stimulated with a fluorescent formyl peptide (ALEXA 568-6-pep) in suspension. Following an 8-min incubation at 37°C we observed that a subset of the ligand-bound ⌬ST-FPR clusters colocalized with the GM1 clusters (Fig. 9A). Interestingly there were individual clusters of either the FPR or GM1 that were distinct from the clusters containing both active receptor and GM1. We also examined colocalization between the FPR and flotillin-1, a marker for caveolae-like rafts in cells that lack caveolin, such as hematopoietic cells (33)(34)(35). While flotillin-1 has been observed to colocalize with GM1 following cell activation (35), we were unable to observe colocalization of the FPR with flotillin-1 following ligand binding by the receptor, suggesting that these signaling domains may be distinct in non-polarized cells (Fig.  9B). As a result, we concluded that the activated form of the FPR is concentrated in a subset of GM1-rich rafts prior to receptor phosphorylation and internalization. DISCUSSION We used promonocytic U937 cells stably transfected with the wild type FPR or a phosphorylation-deficient FPR mutant to examine whether receptor clustering designates an as of yet unrecognized state of GPCR activation. We observed that the wild type FPR clustered prior to its association with arrestins, suggesting that FPR internalization and even phosphorylation may not be required for receptor clustering. Such clustering might be the result of conformational changes within the receptor or signal transduction events initiated by receptor activation, leading to its reorganization within the plasma membrane. Due to the inherent difficulty in trapping wild type receptor intermediates, we used a mutant form of the FPR, ⌬ST-FPR, which signals to G proteins upon ligand binding but does not undergo phosphorylation, associate with arrestins, or become internalized.
In this study, we observed that the ⌬ST-FPR clustered on the plasma membrane following ligand binding. While G protein activation was not required to induce this clustering, transition of the receptor to an active conformation was found to be an essential factor since the D71A mutant, which binds ligand but is incapable of assuming an active conformation, failed to cluster. Furthermore we demonstrated that the clustering process was reversible and thus dynamic as loss of ligand from the clustered receptor, in response to treatment of the cells with antagonist, resulted in a loss of clusters commensurate with a cessation of signaling activity. This suggested that FPR clusters may be sites of active signal transduction. We subse-quently demonstrated that clustering and signaling of the phosphorylation-deficient and wild type forms of the FPR were cholesterol-dependent processes since treatment with M␤CD inhibited both responses, indicating that the very rapid response of calcium mobilization as well as the subsequent reorganization of receptors into clusters/rafts involves cholesterolrich domains. In addition, we confirmed the existence of these rafts by visualization of the clustered ⌬ST-FPRs within a subset of GM1-containing raft domains.
Previous studies examining the clustering of GPCRs suggested that cells organize GPCRs into clusters solely to mediate internalization (10,12,36). However, a number of GPCRs (such as the endothelin, M2 muscarinic acetylcholine, EDG-1, bradykinin B2, and kinin B1 receptors) have been shown to translocate into caveolae as a result of receptor activation (8,9,11,(37)(38)(39). Whether this receptor reorganization/clustering represents a mechanism critical for signaling or internalization is largely unclear. Interestingly the bradykinin B1 receptor is believed not to undergo phosphorylation or significant internalization following agonist stimulation, suggesting that its cholesterol-dependent clustering into caveolae may mediate signaling (39). These results support our conclusions that activated GPCRs are capable of translocating to various raft-like domains independently of receptor phosphorylation and arres- tin binding. Furthermore our results suggest that FPR-containing domains in unstimulated cells, if they exist, may initially be extremely small in size (below the resolution of light microscopy) but coalesce or translocate over time into observable clusters. A related model has recently been proposed for the epidermal growth factor receptor, a tyrosine kinase receptor, in which unstimulated receptors are localized to caveolae and upon stimulation move out of this domain, eventually translocating to clathrin-coated pits where the receptor is internalized (40). In total, these results indicate that the modulation of receptor localization in the plasma membrane is a complex process and likely highly regulated.
Since concentration of the FPR in the plane of the plasma membrane must by definition occur prior to its internalization, it is likely that the clusters we observed represent sites at which the receptor will become phosphorylated and subsequently internalized. This is in contrast to the ␤ 2 adrenergic receptor, which requires the binding of arrestin to cluster in clathrin-coated pits. Unlike the ␤ 2 adrenergic receptor, however, the FPR does not require arrestin for internalization and internalizes in a dynamin-and clathrin-independent manner (16). Whether this occurs from the clusters that colocalize with GM1 or the non-coincident clusters or both is currently under investigation. Caveolae-mediated internalization is also not a viable mechanism for FPR internalization for a number of reasons. 1) Hematopoietic cells, such as U937, do not express caveolin (41). 2) Flotillin, which is expressed in hematopoietic cells and may functionally substitute for caveolin, was not found to colocalize with the activated FPR clusters (Fig. 9). 3) Caveolae-mediated internalization requires dynamin, the inhibition of which does not impair FPR internalization (16). Thus FPR internalization appears to be mediated by an as of yet undescribed mechanism.
The FPR clusters/rafts we observed were shown potentially to consist of two distinct types. A significant fraction of the FPR clusters colocalized with a fluorescently labeled GM1. As opposed to commercially available fluorescent GM1, which is labeled on the lipid tail and thus may distort its partitioning and biological properties (42), our GM1 is labeled on the carbohydrate moiety and thus behaves more like, if not identically to, endogenous GM1. 3 Interestingly numerous FPR and GM1 clusters also showed no clear coincidence with each other, suggesting that the FPR-containing GM1 domains are distinct from the FPR-free GM1 domains, consistent with recent observations of raft heterogeneity (43). We are currently examining the dynamics of these processes during cell polarization and chemotaxis to determine whether these various raft populations are stable over time and whether they segregate and redistribute in response to directional stimulation. While we have performed sucrose flotation experiments with the phosphorylation-deficient form of the FPR, we have been unable to extract the receptor in a low density membrane fraction. This, however, is not surprising as an ever increasing number of known raft-residing proteins are being shown to be extracted from membranes with detergents such as Triton X-100 (11).
In conclusion, we have characterized a previously unidentified state of GPCR activation wherein the activated receptor transiently clusters in signaling domains prior to its phosphorylation, desensitization, and internalization. These results describe a novel paradigm in which arrestin binding, resulting in receptor concentration in clathrin-coated pits, is not required for receptor clustering but rather conformational activation of the receptor by agonist appears to be sufficient to mediate clustering in rafts. As GM1-rich signaling domains are thought to persist throughout the maturation of endosomes (34), internalized receptors might be capable of continued G proteinmediated signaling until either the ligand is displaced from the activated GPCR, the receptor is segregated from G proteins, or proteins such as arrestin bind, preventing further association with G proteins. Our recent demonstration of GPCR-mediated apoptosis in arrestin-deficient cells, a process that requires internalization of signaling-competent receptors, further supports this concept (44). Further studies will be required to reveal the role(s) of GPCR-raft associations in receptor signaling and trafficking.
FIG. 9. GM1-rich domains but not flotillin-rich domains colocalize with the activated FPR. A, ⌬ST-FPR cells were preincubated with 10 nM GM1-AL-EXA 488 for 30 min at 37°C. The cells were washed and then stimulated with 10 nM ALEXA 546-6-pep for 8 min. The cells were fixed in 2% paraformaldehyde, phosphate-buffered saline and imaged by confocal immunofluorescence microscopy. B, ⌬ST-FPR cells were stimulated with 10 nM ALEXA 546-6-pep for 8 min and fixed in 2% paraformaldehyde, phosphate-buffered saline. Flotillin was subsequently labeled with anti-flotillin antibodies followed by anti-rabbit secondary antibodies conjugated to FITC. Cells were imaged by confocal immunofluorescence microscopy. Data are representative of three independent experiments.