Phosphatidylinositol 4,5-bisphosphate (PIP2)-enhanced G protein-coupled receptor kinase (GRK) activity. Location, structure, and regulation of the PIP2 binding site distinguishes the GRK subfamilies.

The G protein-coupled receptor kinases (GRKs) phosphorylate agonist occupied G protein-coupled receptors and play an important role in mediating receptor desensitization. The localization of these enzymes to their membrane incorporated substrates is required for their efficient function and appears to be a highly regulated process. In this study we demonstrate that phosphatidylinositol 4,5-bisphosphate (PIP2) enhances GRK5-mediated β-adrenergic receptor (βAR) phosphorylation by directly interacting with this enzyme and facilitating its membrane association. GRK5-mediated phosphorylation of a soluble peptide substrate is unaffected by PIP2, suggesting that the PIP2-enhanced receptor kinase activity arises as a consequence of this membrane localization. The lipid binding site of GRK5 exhibits a high degree of specificity and appears to reside in the amino terminus of this enzyme. Mutation of six basic residues at positions 22, 23, 24, 26, 28, and 29 of GRK5 ablates the ability of this kinase to bind PIP2. This region of the GRK5, which has a similar distribution of basic amino acids to the PIP2 binding site of gelsolin, is highly conserved between members of the GRK4 subfamily (GRK4, GRK5, and GRK6). Indeed, all the members of the GRK4 subfamily exhibit PIP2-dependent receptor kinase activity. We have shown previously that the membrane association of βARK (β-adrenergic receptor kinase) (GRK2) is mediated, in vitro, by the simultaneous binding of PIP2 and the βγ subunits of heterotrimeric G proteins to the carboxyl-terminal pleckstrin homology domain of this enzyme (Pitcher, J. A., Touhara, K., Payne, E. S., and Lefkowitz, R. J. (1995) J. Biol. Chem.270, 11707-11710). Thus, five members of the GRK family bind PIP2, βARK (GRK2), βARK2 (GRK3), GRK4, GRK5, and GRK6. However, the structure, location, and regulation of the PIP2 binding site distinguishes the βARK (GRK2 and GRK3) and GRK4 (GRK4, GRK5, and GRK6) subfamilies.

Repeated or continuous exposure of G protein-coupled receptors to agonists results in attenuated signaling, a process referred to as desensitization. The direct uncoupling of the receptor from its respective G protein underlies the rapid phase of this process and is mediated, at least in part, by receptor phosphorylation (reviewed in Ref. 1). Two classes of serine/ threonine kinases phosphorylate G protein-coupled receptors, the second messenger-dependent protein kinases (cAMP-dependent protein kinase and protein kinase C), and the G protein-coupled receptor kinases (GRKs). 1 Six members of the GRK family have been cloned to date (reviewed in Refs. 2 and 3) and, on the basis of structural similarities, have been divided into three subfamilies as follows: 1) rhodopsin kinase (RK or GRK1), 2) the ␤-adrenergic receptor kinase subfamily (␤ARK and ␤ARK2 also termed, respectively, GRK2 and -3), and 3) the GRK4 subfamily (GRK4, -5, and -6) (3). Characteristics common to all members of the GRK family include their substrate specificity and the membrane targeting of these kinases. All the GRKs exhibit a marked preference for agonist-occupied receptors as substrates and, since they phosphorylate membrane-incorporated substrates, require membrane localization for efficient function. The structural determinants responsible for mediating membrane association of the GRKs have been most clearly elucidated for RK and the ␤ARK subfamily. These enzymes exhibit stimulus-dependent membrane association, thus agonist occupancy of a receptor substrate results in the "translocation" of RK and ␤ARK from the cytosol to the membrane (4,5). The stimulus-dependent membrane targeting of RK and thus RKmediated rhodopsin phosphorylation has been shown to be dependent upon RK farnesylation (6,7). Interestingly, of the GRKs thus far identified, RK is unique in that it is the only member of this family to possess a CAAX box and thus to be modified by isoprenylation.
For the ␤ARK subfamily, membrane association appears to be mediated, at least in part, by the carboxyl-terminal pleckstrin homology (PH) domain of these enzymes. The PH domain, an approximately 100-amino acid region of sequence homology, is found in numerous proteins involved in the processes of signal transduction and cell growth (8,9). The apparently conserved structure of this domain (10 -13) suggests a potentially conserved function, although this remains to be further investigated. In the case of the ␤ARK subfamily, the simultaneous binding of two PH domain ligands, the ␤␥ subunits of heterotrimeric G proteins (G␤␥), and phosphatidylinositol 4,5bisphosphate (PIP 2 ), has been shown to be required for membrane association of ␤ARK and ␤ARK-mediated ␤AR phospho-rylation in vitro (14). Thus, for the RK and ␤ARK subfamilies agonist-dependent membrane association is mediated via distinct structural determinants and mechanisms. The one commonality is the participation of a prenylated protein in this process, RK itself, or G␤␥ for the ␤ARK subfamily.
The mechanisms underlying the membrane targeting of the GRK4 subfamily remain less well defined. These enzymes are not isoprenylated, do not contain a PH domain, and fail to bind G␤␥. GRK4 and GRK6 are, however, modified by fatty acylation, a palmitoyl moiety being attached to carboxyl-terminal cysteine residues (15,16). Palmitoylation represents an attractive mechanism for affecting the membrane localization of these kinases since it provides the potential for regulation of their subcellular distribution by an acylation/deacylation cycle. GRK5 is not palmitoylated. This enzyme does, however, in contrast to GRK4 and GRK6, autophosphorylate (17). Comparison of the kinase activities of wild type GRK5 with an autophosphorylation-deficient mutant of this enzyme reveal equivalent abilities to phosphorylate soluble peptide substrates (17). In contrast, the autophosphorylation-deficient mutant exhibits a dramatically impaired ability to phosphorylate membraneincorporated receptor substrates (17). Thus, autophosphorylation, although apparently not directly increasing the catalytic activity of GRK5, facilitates GRK5-mediated receptor phosphorylation. GRK5 autophosphorylation has thus been proposed to participate in the membrane localization of this enzyme (17).
Members of the GRK4 subfamily have regions in both their carboxyl and amino termini rich in basic and polar amino acids. It has been speculated that these regions may facilitate membrane association of these GRKs via interaction with the negatively charged headgroups of phospholipids (2). Following the recent elucidation of the role of the negatively charged phospholipid PIP 2 in mediating membrane association of ␤ARK, we sought to examine a potential role for this and other lipids in mediating the membrane association of one member of the GRK4 subfamily, GRK5.

EXPERIMENTAL PROCEDURES
Materials-Bovine ␤ARK and GRK5 were overexpressed and purified from baculovirus-infected Sf9 cells (18,19), and G ␤␥ subunits were purified from bovine brain (20) according to previously published procedures. Cos7 cell extracts expressing GRK4, -5, and -6 were also utilized as a source of these kinases. GRK4, -5, and -6 were transfected into Cos7 cells using a standard DEAE-dextran procedure (21). Cells were subsequently harvested and lysed, and a soluble cell extract enriched in these kinases was obtained (7). Rod outer segment membranes, devoid of RK activity, were prepared as described previously (22). Purified lipids and soybean phosphatidylcholine (ϳ20% phosphatidylcholine (PC), termed "crude lipid") were obtained from Sigma.
Construction of Wild Type and Mutant GRK5 cDNAs-The cDNA encoding bovine GRK5 (19) was modified to remove all 3Ј-and 5Јuntranslated regions. Using standard polymerase chain reaction techniques as described previously for mutant ␤ARK constructs (7), a new 5Ј end containing EcoRI, BglII, and Kozak consensus sequences was added before the start codon, and a new 3Ј end containing an XbaI site was inserted immediately after the translational stop codon. The amplified fragment was inserted into the EcoRI/XbaI sites of the expression vectors pcDNAI (Invitrogen) and pRK5 (23). These constructs were used as the wild type GRK5 (GRK5 WT ) templates for all subsequent molecular manipulations. Mutated GRK5 cDNAs were constructed using standard polymerase chain reaction techniques. The carboxyl-terminal polybasic mutant was made using a sense primer starting from the GRK5 AccI site and mutating K547A, K548A, R553A, K556A, and R557A and an antisense primer ending at the 3Ј XbaI site. The amplified fragment was digested with AccI/XbaI and ligated with the EcoRI/ AccI fragment of GRK5 into the EcoRI/XbaI sites of pRK5 to form the expression plasmid pRK5-GRK5 CTPB . The amino-terminal polybasic mutant was made using an antisense primer starting at the GRK5 BalI site and mutating K22A, R23A, K24A, K26A, K28A, and K29A and a sense primer ending at the PstI site. The amplified fragment was digested with BalI/PstI and inserted into the BalI/PstI sites of pcDNAI-GRK5 WT to form pcDNA-GRK5 NTPB . The EcoRI/XbaI fragment of pcDNA-GRK5 NTPB was excised and inserted into the EcoRI/XbaI sites of pRK5 to form the expression plasmid pRK5-GRK5 NTPB . The autophosphorylation-deficient kinases, the S484A, T485A double mutant (GRK5 ST-AA ) and the S484D, T485D double mutant (GRK5 ST-DD ), were constructed as described previously for the double mutant GRK5 ST-AA (17). The amplified fragments were digested with BglII/ XbaI and inserted into the BglII/XbaI sites of the baculovirus transfer vector pVL1392 (PharMingen) to form pVL-GRK5 ST-AA and pVL-GRK5 ST-DD . Mutated constructs were verified by double-stranded dideoxynucleotide sequencing of cDNAs using chain termination with Sequenase 2 Polymerase (Amersham Corp./U. S. Biochemical Corp.).
The autophosphorylation-deficient kinases, GRK5 ST-AA and GRK5 ST-DD , were expressed in and purified from baculovirus-infected Sf9 cells, using the purification procedure described in Ref. 19. In contrast, Cos7 cell extracts expressing the polybasic mutants GRK5 NTPB and GRK5 CTPB were used as the source of these kinases. Expression of the mutant GRK5 constructs was confirmed by Western blot analysis using GRK5 antibodies as described previously (19).
Purification and Reconstitution of the ␤ 2 -Adrenergic Receptor-The human ␤ 2 -adrenergic receptor (␤AR) was expressed and purified from baculovirus-infected Sf9 cells as described previously (14,24). Purified receptor was subsequently reconstituted into either soybean phosphatidylcholine vesicles (20% PC, termed "crude lipid environment") or into vesicles of defined lipid composition. To form vesicles, the required amounts of lipid in chloroform were dried under a stream of nitrogen, resuspended in 10 mM Tris-HCl, pH 7.2, 100 mM NaCl by vortexing, and sonicated with a microtip sonicator. Purified ␤AR was reconstituted into these vesicles as described previously (25). The ␤AR-containing vesicles were resuspended in 20 mM Tris-HCl, pH 7.5, 2 mM EDTA and the receptor concentration determined by radioligand binding using GRK-mediated ␤AR Phosphorylation-The ␤AR (40 nM) reconstituted in various lipid environments (described in the text and figure legends) was incubated with GRK in 20 mM Tris-HCl, pH 7.5, 2.0 mM EDTA, 10 mM MgCl 2 , 1 mM dithiothreitol containing 60 M ATP (ϳ6000 cpm/pmol) in a total volume of 25 l. Purified GRK5 or ␤ARK were used at a final concentration of 10 nM. The purified autophosphorylationdeficient mutants of GRK5 were utilized at equivalent peptide (RRREEEEESAAA) phosphorylation activities, approximately 15 and 7 nM, respectively, for the GRK5 ST-AA and the GRK5 ST-DD . When utilizing Cos7 cell extracts as the source of GRK, either equivalent amounts of protein or equivalent rhodopsin kinase activities (as indicated in the figure legend) were utilized in the ␤AR phosphorylation assay. All assays were performed in the presence of 50 M (Ϫ)-isoproterenol, and purified G ␤␥ subunits (100 nM) were also included where indicated. Reactions were incubated at 30°C, stopped by addition of an equal volume of SDS sample-loading buffer (8% SDS, 25 mM Tris-HCl, pH 6.5, 10% glycerol, 5% mercaptoethanol, 0.003% bromphenol blue), and electrophoresed on 10% SDS-polyacrylamide gels. The dried gels were subjected to autoradiography and phosphorimager (Molecular Dynamics) analysis to determine the pmol of phosphate transferred to the receptor substrate.
PIP 2 -dependent Association of GRK5 with Lipid Vesicles-Vesicles composed of either 100% PC or 95% PC, 5% PIP 2 were incubated in 7 ϫ 20-mm polycarbonate tubes (Beckman) with purified GRK5 (0.5 g). Incubations were performed for 10 min at 4°C in phosphate-buffered saline (PBS); the final lipid concentration was 1.7 mg/ml and reaction volume 30 l. Tubes were subsequently centrifuged at 100,000 rpm (TL-100 rotor) for 15 min at 4°C. The supernatant was removed and the pellet rinsed once with PBS. The pellet was subsequently resuspended in 15 l of PBS and transferred to a clean tube. SDS sample-loading buffer was added to the supernatant and pellet fractions, and the samples were electrophoresed on 4 -20% gradient polyacrylamide gels (Novex) and subjected to Western blot analysis (ECL, Amersham Corp.) using anti-GRK5 antibodies (19). The distribution of the GRK5 between the pellet (P) and the supernatant (S) was determined by densitometric analysis of the Western blot.
GRK-mediated Phosphorylation of a Soluble Synthetic Peptide Substrate-A stock solution of the purified peptide (RRREEEEESAAA) was prepared and the pH adjusted to 7.2 by the addition of Tris base. GRK-mediated peptide phosphorylation was determined by incubating peptide (1 mM) and the purified GRK, either GRK5 or ␤ARK (10 nM), in 20 mM Tris-HCl, pH 7.2, 2 mM EDTA, 7.5 mM MgCl 2 , and [␥-32 P]ATP (ϳ2000 cpm/pmol) at the concentrations indicated. The final reaction volume was 25 l, and incubations were performed at 30°C for 15 min. Phosphorylation reactions were linear over this time. Reactions were stopped by spotting onto P-81 phosphocellulose paper (2 ϫ 2-cm squares). Free [␥-32 P]ATP was subsequently removed by washing in 75 mM phosphoric acid as described previously (26). For each reaction condition (i.e. lipid concentration) GRK-mediated peptide phosphorylation was determined by subtracting the counts incorporated in the absence of peptide from the counts incorporated in the presence of this substrate.

RESULTS
A role for PIP 2 in facilitating ␤ARK-mediated phosphorylation of the ␤AR has recently been elucidated. Using purified proteins in a reconstituted system, the binding of G␤␥ and PIP 2 to the ␤ARK PH domain has been shown to be required for membrane association of this enzyme and for ␤ARK-mediated ␤AR phosphorylation (14). To determine if PIP 2 plays a role in facilitating ␤AR phosphorylation mediated by other members of the GRK family, the ␤AR reconstituted into vesicles of defined lipid composition was utilized as a substrate for GRK5. GRK5, a member of the GRK4 subfamily, does not possess a PH domain and fails to bind G␤␥ (19). This enzyme does contain regions rich in basic amino acids within both its carboxyl and amino termini that potentially represent lipid binding domains.
GRK5-mediated receptor phosphorylation has previously been assessed utilizing as substrates either rhodopsin in rod outer segment membranes or alternatively purified G proteincoupled receptors reconstituted in vesicles composed of heterogeneous and ill-defined lipid mixtures (termed crude lipid environment) (19,27). As demonstrated in Fig. 1, under these conditions GRK5 mediates agonist-dependent receptor phosphorylation. In marked contrast, the ␤AR reconstituted in lipid vesicles composed of purified phosphatidylcholine (100% PC) fails to serve as a substrate for this kinase even in the presence of agonist (Fig. 1). These results suggest the presence of essential lipid cofactors for GRK5 in the crude lipid mixture. To examine a potential role for PIP 2 in GRK5-mediated ␤AR phosphorylation, this lipid was incorporated into the receptor containing vesicles. The addition of 5% PIP 2 to PC vesicles (95% PC, 5% PIP 2 ) results in a dramatic enhancement of GRK5mediated agonist-dependent ␤AR phosphorylation (Fig. 1). PIP 2dependent GRK5 phosphorylation occurs in the absence of G␤␥ subunits (Fig. 1) consistent with previous studies demonstrating the G␤␥ independence of this enzyme (19).
To determine if the enhanced ␤AR phosphorylation observed in the presence of PIP 2 arises as a consequence of the direct activation of GRK5, GRK5-mediated phosphorylation of a soluble peptide substrate was examined. As shown in Fig. 2, addition of lipid vesicles composed of 100% PC; 95% PC, 5% PIP 2 (0.5 g of PIP 2 ); or 80% PC, 20% PIP 2 (2.0 g of PIP 2 ) had no significant effect on the initial rate of GRK5-mediated pep-tide phosphorylation. Under these conditions, both the maximal rate of peptide phosphorylation and the concentration of ATP required for half-maximal activation of GRK5 (the K m for ATP) were unaffected by PIP 2 addition. These results suggest that the PIP 2 -dependent enhancement of GRK5-mediated ␤AR phosphorylation, observed in Fig. 1, arises as a consequence of increased membrane association of GRK5 rather than direct activation of this kinase. That PIP 2 indeed promotes vesicle association of GRK5 is shown in Fig. 3. Addition of PIP 2 to PC vesicles significantly increases membrane association of GRK5 with 5 Ϯ 5.0% and 70.2 Ϯ 13.2% of the enzyme, respectively, being pelleted in the absence and presence of PIP 2 (Fig. 3). PIP 2 -dependent membrane association of GRK5 was observed in both the presence (Fig. 3) and absence (data not shown) of the ␤AR substrate. Thus, as with ␤ARK (14), PIP 2 appears to enhance GRK5-mediated ␤AR phosphorylation by promoting the membrane localization of this kinase.
As shown in Fig. 2, low concentrations of PIP 2 (0.5-2.0 g) have no effect on GRK5-mediated peptide phosphorylation and thus no direct effect on GRK5 activity. In contrast, high concentrations of this lipid inhibit GRK5 (Fig. 4A). Addition of 25 g of 100% PIP 2 dramatically and specifically impairs the ability of this kinase to phosphorylate a soluble peptide substrate (Fig. 4A). Similar effects of PIP 2 are observed for ␤ARKmediated peptide phosphorylation (Fig. 4B). Thus, the addition of vesicles containing 2.0 g of PIP 2 has no effect on ␤ARK-mediated peptide phosphorylation (Fig. 4B). High concentrations of vesicles of 100% PIP 2 (25 g), however, dramatically inhibit the maximal rate of ␤ARK-mediated peptide phosphorylation.
These results serve to clarify the somewhat confusing literature concerning the role PIP 2 plays in regulating ␤ARK activity. A recent report from our laboratory demonstrates that in vitro the coordinated binding of this lipid and G␤␥ to the PH domain of the ␤ARK is required for ␤ARK-mediated ␤AR phosphorylation (14). In contrast others (28,29) have reported that PIP 2 inhibits ␤ARK activity. The data shown in Fig. 4B provide an explanation for these apparently conflicting results. The differential effects of PIP 2 reported by us (14)  ␤ARK-mediated ␤AR phosphorylation is observed at concentrations of PIP 2 that have no effect on ␤ARK-mediated peptide phosphorylation (Fig. 4B) (14). In contrast the addition of high concentrations of this lipid directly inhibit ␤ARK, resulting in inhibition of ␤ARK-mediated receptor and peptide phosphorylation. The PIP 2 concentrations required to observe inhibition of ␤ARK activity are approximately 12-100-fold higher than those required to observe PIP 2 -dependent ␤ARK activity and are equivalent to the inhibitory concentrations of PIP 2 shown in Fig. 4, A and B.
GRK5 undergoes rapid intramolecular phosphorylation to stoichiometries approaching 2.0 mol of P i /mol of kinase (17,19). Autophosphorylation has been shown to be stimulated nonspecifically by phospholipids and is proposed to play a role in the membrane localization of GRK5 (17). A role for autophosphorylation in membrane localization is suggested by the observation that, as compared with the wild type enzyme, an autophosphorylation-deficient mutant of GRK5 is specifically impaired in its ability to phosphorylate membrane incorporated receptor substrates (17). PIP 2 may thus potentially facilitate membrane association of GRK5 and thus enhance ␤AR phosphorylation by stimulating the autophosphorylation of this kinase. To test this hypothesis, the two principal autophosphorylation sites of GRK5 (Ser-484 and Thr-485) were mutated to either alanine (GRK5 ST-AA ) or aspartic acid residues (GRK5 ST-DD ) to create, respectively, an autophosphorylation-deficient mutant mimicking the unphosphorylated wild type GRK5 and a mutant in which the autophosphorylation sites are replaced with negatively charged amino acids. The substitution of negatively charged amino acids for phosphorylatable residues has been proposed to mimic the functional effects of phosphorylation (30). The GRK5 ST-DD mutant would thus be proposed to be functionally similar to a fully autophosphorylated form of GRK5. These two mutant GRK5s were subsequently assessed for their ability to phosphorylate the ␤AR in either the presence or absence of PIP 2 (Fig. 5). PIP 2 enhances ␤AR phosphorylation mediated by either of the GRK5 mutant kinases. Thus, although at equivalent peptide kinase activities the GRK5 ST-AA mutant is less active than the GRK5 ST-DD mutant at phosphorylating the ␤AR, consistent with previously published observations (17), the activity of both kinases is enhanced by PIP 2 (Fig. 5). The PIP 2 -dependent membrane association of GRK5 and enhanced GRK5-mediated ␤AR phosphorylation thus occurs via an autophosphorylation-independent mechanism.
We investigated the lipid specificity of GRK5. To this end the ␤AR was reconstituted in purified PC vesicles containing either 3 or 20% of the lipids to be tested (Fig. 6, A and B). As shown in Fig. 6, lipid binding to GRK5 appears highly specific. In vesicles composed of 97% PC and 3% of various lipids, only PIP 2 was capable of promoting GRK5-mediated ␤AR phosphorylation. Increasing the concentration of lipids in the PC background to 20% increased the extent of GRK5-mediated PIP 2dependent ␤AR phosphorylation (Fig. 6B). At a concentration of 20%, phosphatidylinositol 4-phosphate (PIP) also supported GRK5-mediated ␤AR phosphorylation, although to a lesser extent than the equivalent concentration of PIP 2 (Fig. 6B). Under these conditions, none of the other lipids tested enhanced GRK5-mediated receptor phosphorylation (Fig. 6). These results support the hypothesis that GRK5 possesses a highly specific binding domain for PIP 2 .
To identify which region of GRK5 participates in the binding of PIP 2 , mutant GRK5 enzymes were constructed. Since regions rich in basic amino acids have been implicated as potential sites of interaction with negatively charged phospholipids, two such regions, one in the carboxyl and one in the amino terminus of GRK5, were mutated. The amino-terminal basic amino acid residues Lys-22, Arg-23, Lys-24, Lys-26, Lys-28, and Lys-29 or the carboxyl-terminal basic amino acids Lys-547, Lys-548, Arg-553, Lys-556, and Arg-557 were mutated to alanines. These kinases are termed, respectively, GRK5 NTPB and GRK5 CTPB . The ability of these kinases to phosphorylate the ␤AR in either the presence or absence of PIP 2 is shown in Fig.  7. Extracts derived from Cos7 cells overexpressing these enzymes exhibited very different PIP 2 -dependent ␤AR kinase activities (Fig. 7). Thus, although both the wild type enzyme (GRK5) and the carboxyl-terminal GRK5 mutant (GRK5 CTPB ) exhibited PIP 2 -dependent ␤AR phosphorylation, the aminoterminal GRK5 mutant (GRK5 NTPB ) failed to phosphorylate this receptor substrate. These results suggest that the site of PIP 2 interaction may reside within the amino terminus of GRK5 and that the basic amino acids, Lys-22, Arg-23, Lys-24, Lys-26, Lys-28, and Lys-29, play an important role in mediating this interaction. Interestingly, this basic amino-terminal region is highly conserved between members of the GRK4 subfamily (Fig. 8A) suggesting that GRK4 and GRK6 may also be regulated by this lipid. Indeed, cell extracts overexpressing either of these kinases exhibited enhanced ␤AR phosphorylation in the presence of PIP 2 (Fig. 8B). That the PIP 2 dependence of the ␤AR phosphorylation is a regulatory mechanism operating at the level of the GRK rather than at the level of the ␤AR is supported by the observation that cAMP-dependent protein kinase-mediated ␤AR phosphorylation is unaffected by the lipid composition of the receptor containing vesicles (Fig.  8B). Thus, all the currently identified members of the GRK4 subfamily appear to bind PIP 2 at their amino termini. The binding of this lipid facilitates membrane association of these kinases and promotes phosphorylation of vesicle-incorporated receptor substrates. DISCUSSION GRK5, a member of the GRK4 subfamily of G protein-coupled receptor kinases, phosphorylates in an agonist-dependent FIG. 7. The PIP 2 binding site of GRK5 is located within the amino terminus of the enzyme. The PIP 2 -dependent ␤AR kinase activity of wild type (GRK5) and two mutant GRK5s (GRK5 NTPB and GRK5 CTPB ) was assessed. GRK5 NTPB and GRK5 CTPB are mutants in which, respectively, the amino-terminal basic amino acids (Lys-22, Arg-23, Lys-24, Lys-26, Lys-28, and Lys-29) or the carboxyl-terminal basic amino acids (Lys-547, Lys-548, Arg-553, Lys-556, and Arg-557), are mutated to alanine residues. The ability of these enzymes to phosphorylate ␤AR (1 pmol) reconstituted in vesicles composed of 100% PC or 95% PC, 5% PIP 2 (indicated on the figure) was subsequently assessed. Cos7 cell extracts derived from cells expressing either of these three GRKs were used as the source of kinase in these experiments. The amount of Cos7 cell protein added to each ␤AR phosphorylation reaction was adjusted such that all enzymes were utilized at equivalent rhodopsin kinase activities. A representative autoradiogram is shown. The experiment was repeated at least three times with similar results. fashion G protein-coupled receptors when these substrates are presented in crude lipid environments (Fig. 1A) (19,27). Interestingly, however, this enzyme has negligible activity against the ␤AR reconstituted in vesicles composed of 100% PC (Fig. 1). Thus, GRK5-mediated ␤AR phosphorylation exhibits a specific lipid dependence. When a variety of lipids are incorporated into PC and ␤AR containing vesicles and tested for their ability to restore GRK5 function, only PIP 2 and to a lesser extent PIP are effective at promoting GRK5-mediated receptor phosphorylation (Fig. 6). Under these conditions, PIP 2 directly interacts with GRK5 ( Fig. 4) but has no effect on GRK5-mediated peptide phosphorylation (Fig. 2). These results suggest that the enhanced ␤AR phosphorylation observed in the presence of this lipid is due to PIP 2 -dependent membrane association of GRK5 rather than to direct activation of this enzyme. PIP 2 -mediated membrane localization presumably enhances GRK5-mediated ␤AR phosphorylation by placing the kinase in close proximity to its receptor substrate. PIP 2 is a negatively charged phospholipid. Regions of the GRK5 rich in basic amino acids are thus likely to represent potential binding sites for this lipid. Mutation of the charged amino acids within two such regions of GRK5, one in the amino and one in the carboxyl terminus of the enzyme, implicates amino-terminal basic amino acids as a site of PIP 2 interaction. Notably, the distribution of basic amino acids within this region of GRK5 bears a marked similarity to that of the PIP 2 binding site of gelsolin (31) (shown in Fig. 8A). Furthermore, this region is highly conserved among members of the GRK4 subfamily (Fig. 8A). Indeed all the members of the GRK4 subfamily exhibit PIP 2 -enhanced ␤AR phosphorylation (Fig. 8B). GRK4 has recently been shown to exist as four distinct splice variant proteins (16). The longest of these, GRK4␣, was utilized in this study and displays PIP 2 -dependent receptor kinase activity. Interestingly, two of the shorter GRK4 splice variants (GRK4␤ and -␦) lack 32 amino acids at their amino termini and thus lack the postulated PIP 2 binding site (16). The potentially differing lipid requirements of the GRK4 splice variants remain to be investigated.
As described previously, PIP 2 plays a role in facilitating ␤ARK-mediated ␤AR phosphorylation (14). Using purified components in a reconstituted system, the simultaneous bind-ing of G␤␥ and PIP 2 to the PH domain of ␤ARK has been shown to be required for the membrane localization of this kinase and ␤ARK-mediated ␤AR phosphorylation. Thus, PIP 2 plays a role in enhancing the receptor kinase activity of both the ␤ARK and the GRK4 subfamilies of the GRKs. However, the location and structure of the PIP 2 binding site as well as the mechanisms regulating the binding of this lipid distinguish the GRK subfamilies. Thus, members of the GRK4 subfamily bind PIP 2 via an amino-terminal peptide sequence rich in basic amino acids, a region with similarity to the PIP 2 binding site of gelsolin (31). In contrast, the ␤ARK subfamily binds PIP 2 via a carboxylterminal PH domain (14). Furthermore, PIP 2 binds to members of the GRK4 subfamily and promotes receptor kinase activity in the absence of additional ligands. PIP 2 binding to members of the ␤ARK subfamily, however, depends on the binding of a second ligand (G␤␥), which increases the apparent affinity of the ␤ARK for PIP 2 (14). In contrast to the GRK4 subfamily the presence of PIP 2 alone is insufficient to promote ␤ARK-mediated ␤AR phosphorylation (14). The binding of PIP 2 to either GRK subfamily has, however, the similar functional consequence of promoting the membrane association (and thus indirectly the receptor kinase activity) of these enzymes.
The role of the PIP 2 /␤ARK interaction has been the subject of some controversy in the literature. Thus although three different laboratories have demonstrated an interaction between the intact ␤ARK and this lipid, different functional consequences arising as a result of this interaction have been reported (14,28,29). Research from this laboratory suggests a role for PIP 2 in promoting ␤ARK activity (as described above and in Ref. 14) while others have reported PIP 2 -mediated inhibition of ␤ARK activity (28,29). These contradictory results can be explained when the different experimental conditions and amounts of PIP 2 utilized are considered. As demonstrated in Fig. 4B, the amounts of PIP 2 (0.5-2.0 g) utilized in this and in a previous study from our laboratory (14) have no effect on ␤ARK-mediated phosphorylation of a soluble peptide substrate. PIP 2 binding to the ␤ARK thus promotes receptor phosphorylation by facilitating the membrane localization of this enzyme. In contrast, the amounts of PIP 2 utilized in the studies of DebBurman et al. (28) and Onorato et al. (29) range between 10 and 50 g per assay. Under these conditions, PIP 2 mediates direct inhibition of the catalytic activity of ␤ARK (Fig. 4B) (29). Thus, high PIP 2 concentrations inhibit ␤ARK-mediated ␤AR phosphorylation. In addition to the amount of PIP 2 utilized, one other significant difference distinguishes our studies from those of others, namely the manner in which PIP 2 is presented to the GRKs. In the studies of DebBurman et al. (28) and Onorato et al. (29), variable amounts of lipid vesicles composed of 100% PIP 2 are directly added to the phosphorylation reactions. In studies from this laboratory, however, the effect of PIP 2 is assessed using receptor containing lipid vesicles composed of a mixture of PC and PIP 2 . Under these conditions the total amount of lipid added per assay remains constant, and the only difference between assay conditions is the variable PIP 2 content of the vesicles. PIP 2 is thought to represent approximately 0.25-1.0% of total membrane lipid (32,33). Thus, in our studies the composition of lipid vesicles was varied between 100% PC, 0% PIP 2 , and 80% PC, 20% PIP 2 . The lower concentrations of PIP 2 utilized by us reflect the use of these, presumably, more physiological membrane environments. PIP 2 -mediated membrane association represents only one of a number of potential mechanisms for targeting the GRKs to their appropriate substrates. GRK5 autophosphorylation may also play a role in mediating membrane association of this kinase, since an autophosphorylation mutant of GRK5 has a specifically reduced ability to phosphorylate membrane-incor- FIG. 8. PIP 2 enhancement of ␤AR phosphorylation is common to all the members of the GRK4 subfamily. A, alignment of the amino-terminal basic domain of GRK4, -5, and -6. The residues postulated to represent the PIP 2 binding site of GRK5 are boxed. The sequence of the PIP 2 binding site of gelsolin is also shown. B, PIP 2 -dependent phosphorylation of the ␤AR by members of the GRK4 subfamily. Purified ␤AR reconstituted in PC in either the presence or absence of 5% PIP 2 was utilized as a substrate for the enzymes indicated. Cos7 cell extracts expressing the GRKs and purified cAMP-dependent protein kinase (PKA) were used as the source of kinase. A representative autoradiogram is shown. The experiment was repeated three times with similar results. porated receptor substrates (17). That PIP 2 and autophosphorylation represent two independent mechanisms for effecting GRK5 membrane localization is indicated by the observation that both wild type GRK5 and an autophosphorylation-deficient mutant, GRK5 ST-AA , exhibit PIP 2 -dependent ␤AR phosphorylation (Fig. 5). In the case of GRK4 and GRK6 the binding of PIP 2 coupled with the palmitoylation of these enzymes presumably facilitates their membrane association and thus interaction with receptor substrates (15,16). For the ␤ARK subfamily PIP 2 -and G␤␥-independent mechanisms of membrane localization may include the interaction with phospholipids other than PIP 2 . High concentrations of phosphatidylserine have been shown to bind to and directly activate ␤ARK (28,29). Additionally, the association of ␤ARK with specific membranelocalized "anchoring" protein(s) may represent an important mechanism for effecting the membrane localization of this enzyme subfamily (34,35). The GRKs phosphorylate membraneincorporated receptor substrates. The membrane localization of these enzymes is thus a prerequisite for efficient function and as such appears to be subject to complex regulatory processes.