The Amino-terminal Domain of G-protein-coupled Receptor Kinase 2 Is a Regulatory Gβγ Binding Site

G-protein-coupled receptor kinase 2 (GRK2) is activated by free Gβγ subunits. A Gβγ binding site of GRK2 is localized in the carboxyl-terminal pleckstrin homology domain. This Gβγ binding site of GRK2 also regulates Gβγ-stimulated signaling by sequestering free Gβγ subunits. We report here that truncation of the carboxyl-terminal Gβγ binding site of GRK2 did not abolish the Gβγ regulatory activity of GRK2 as determined by the inhibition of a Gβγ-stimulated increase in inositol phosphates in cells. This finding suggested the presence of a second Gβγ binding site in GRK2. And indeed, the amino terminus of GRK2 (GRK21–185) inhibited a Gβγ-stimulated inositol phosphate signal in cells, purified GRK21–185 suppressed the Gβγ-stimulated phosphorylation of rhodopsin, and GRK21–185 bound directly to purified Gβγ subunits. The amino-terminal Gβγ regulatory site does not overlap with the RGS domain of GRK-2 because GRK21–53 with truncated RGS domain inhibited Gβγ-mediated signaling with similar potency and efficacy as did GRK21–185. In addition to the Gβγ regulatory activity, the amino-terminal Gβγ binding site of GRK2 affects the kinase activity of GRK2 because antibodies specifically cross-reacting with the amino terminus of GRK2 suppressed the GRK2-dependent phosphorylation of rhodopsin. The antibody-mediated inhibition was released by purified Gβγ subunits, strongly suggesting that Gβγ binding to the amino terminus of GRK2 enhances the kinase activity toward rhodopsin. Thus, the amino-terminal domain of GRK2 is a previously unrecognized Gβγ binding site that regulates GRK2-mediated receptor phosphorylation and inhibits Gβγ-stimulated signaling.

Activated G-protein-coupled receptors are switched off by phosphorylation through G-protein-coupled receptor kinases (GRKs) 1 (1). GRKs are modular proteins consisting of at least three structural domains with different functions. The core kinase domain of GRK2 and GRK3, which represents the ␤-adrenergic receptor kinase isozymes, is flanked by an aminoterminal domain, which contains an RGS domain, and a carboxyl-terminal domain, which contains a pleckstrin homology domain (PH domain) (2)(3)(4). The activation of GRK2 and GRK3 requires the activation and dissociation of a heterotrimeric G-protein, i.e. the kinases are activated by free G␤␥ subunits (5,6). A G␤␥ binding site of GRK2 and GRK3 is localized in the carboxyl terminus of the kinase and overlaps the PH domain (7). Truncation of the PH domain of GRK2 generates a kinase with compromised regulation by G␤␥ subunits (7). The carboxyl-terminal G␤␥ binding site of GRK2 also regulates G␤␥stimulated signaling by sequestering free G␤␥ subunits (8). Analyzing the G␤␥ regulatory activity of proteins is a means of identifying G␤␥-binding proteins or localizing G␤␥ binding sites of proteins (9 -11). To find out whether the G␤␥ regulatory activity of GRK2 resides entirely in the carboxyl-terminal PH domain, we analyzed the G␤␥-sequestering activity of wildtype GRK2 and of carboxyl-terminal-truncated GRK2 mutants. The capacity of those proteins to inhibit a G␤␥-stimulated increase in inositol phosphates mediated by activation of phospholipase C␤ 2 was determined (12). We report here that truncation of the carboxyl-terminal G␤␥ binding site of GRK2 did not abolish the G␤␥ regulatory activity of GRK2. A previously unrecognized G␤␥ binding site of GRK2 was identified in the amino terminus of GRK2 that enhances the kinase activity of GRK2 toward the receptor substrate rhodopsin and which inhibits G␤␥-stimulated signaling.

EXPERIMENTAL PROCEDURES
Cell Culture and Cell Transfection-Human embryonic kidney cells (HEK-293) were cultured and transfected as described previously (13) with plasmids encoding human GRK2 or the indicated truncation mutants of GRK2 or GRK5. The mutants were generated by polymerase chain reaction and were sequenced entirely to confirm the identity of the mutants.
Determination of Cellular Inositol Phosphate Levels-Total inositol phosphate levels of HEK-293 cells were determined as described (13). For determination of the G␤␥ regulatory activity of wild-type GRK2 and of the different truncation mutants, cells were co-transfected with plasmids encoding the indicated GRK2 mutants and phospholipase C␤ 2 , G␤ 1 , and G␥ 2 (11).
Phosphorylation of Rhodopsin by GRK2-The kinase activity of GRK2 was assessed by phosphorylation of the receptor substrate rhodopsin in a total volume of 50 l of buffer (20 mM HEPES, pH 7.4) containing 20 nM GRK2, G␤␥ subunits as indicated, 400 nM rhodopsin, 10 mM MgCl 2 , 2 mM EDTA, and 50 M [␥-32 P]ATP. Phosphorylation was initiated by light and proceeded for 20 min at room temperature. After SDS-PAGE, receptor phosphorylation was assessed by autoradiography. Rhodopsin-enriched membranes were prepared from dark-adapted bovine retinae by sucrose gradient centrifugation (14).
To determine the activation of GRK2 by G␤␥ subunits, various concentrations of purified G␤␥ subunits from bovine brain were incubated * This work was supported by the Deutsche Forschungsgemeinschaft. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
in the phosphorylation mixture (16). To assess the effect of antibodies specifically cross-reacting with the amino terminus or with the carboxyl terminus of GRK2, polyclonal anti-GRK2 antibodies were immunoselected by affinity chromatography on GRK2 1-185 or on GRK2 561-689 , respectively, covalently coupled to Affi-Gel-10 (13). The purified antibodies were incubated in the phosphorylation mixture as indicated. Specificity and cross-reactivity of the antibodies with GRK2 1-185 or with GRK2 561-689 were analyzed in immunoblot.
Immunoblot Detection of Proteins-Proteins were separated on SDScontaining polyacrylamide gels, transferred to polyvinylidene difluoride membranes, and identified in immunoblot similarly as described (17). Antibodies specific for GRK2 or for G␤ have been characterized previously (14,18).
Binding of GRK2  to G␤␥-Purified GST-GRK2 1-185 (100 nM) coupled to glutathione-Sepharose was incubated with G␤␥ subunits (3 nM) in a total volume of 500 l of buffer (100 mM NaCl, 20 mM HEPES, pH 7.4). After extensive washing with the same buffer, bound G␤␥ subunits were eluted with SDS sample buffer, separated by SDS-PAGE, and identified in immunoblot with G␤-specific antibodies. As a control, GST-GRK5 1-200 was used instead of GST-GRK2  .

G␤␥ Regulatory Activity of a GRK2 Mutant with Truncated
Carboxyl-terminal PH Domain-The carboxyl-terminal PH domain of GRK2 is essential for the G␤␥-dependent phosphorylation of receptor substrates by GRK2 (7). Furthermore, PH domains can regulate G␤␥-stimulated signaling by sequestering free G␤␥ subunits (9). To analyze whether the PH domain of GRK2 is entirely responsible for the G␤␥ regulatory activity of GRK2, the PH domain of GRK2 was truncated, and the G␤␥ regulatory activity of the truncated GRK2 mutant (GRK2 1-558CVLL ) was analyzed in intact cells by determining the inhibition of a G␤␥-stimulated increase in inositol phosphates mediated by activation of phospholipase C␤ 2 (11,12). To exclude that a cytosolic localization of the truncated GRK2 mutant prevented the interaction with membrane-anchored G␤␥ subunits in cells, a membrane-anchoring CAAX motif was introduced in GRK2 1-558CVLL similarly as described (7). Wild-type GRK2, the carboxyl-terminal-truncated mutant GRK2 1-558CVLL , and the carboxyl terminus of GRK2 containing the PH domain, GRK2 561-689 , were expressed in HEK-293 cells (Fig. 1A, lanes 1-3) and analyzed for their G␤␥ regulatory activity (Fig. 1B). Wild-type GRK2 and GRK2 1-558CVLL inhibited the G␤␥-stimulated increase in inositol phosphates (Fig.  1B, columns 1 and 2 versus c). GRK2 561-689 comprising the PH domain of GRK2 also significantly decreased the G␤␥-stimulated signal (Fig. 1B, column 3). Expression levels of G␤␥ and of phospholipase C␤ 2 were similar in the different experiments (not shown). Together, these data demonstrate that the PH domain of GRK2 inhibits G␤␥-stimulated signaling in cells but that the G␤␥ regulatory activity of GRK2 is not entirely mediated by the carboxyl-terminal PH domain.
The Amino-terminal Domain of GRK2 Regulates G␤␥-stimulated Signaling-In search for the second G␤␥ binding site of GRK2, the G␤␥ regulatory effect of the amino-terminal domain of GRK2 was analyzed. GRK2 1-185 was expressed in HEK-293 cells (Fig. 3A, upper panel, lane 1). GRK2 1-185 inhibited the G␤␥-stimulated increase in inositol phosphates by ϳ40% similarly to GRK2   (Fig. 3A, column 1 versus c and cf. Fig. 2B). Again the G␤␥ regulatory activity was independent of a CAAX membrane-anchoring motif (Fig. 3A, lane 2, column 2). As a control, the amino-terminal domain of GRK5 was expressed (Fig. 3B, upper panel, lanes 1 and 2) because GRK5 has been reported to phosphorylate receptor substrates independently of the addition of G␤␥ subunits (20). GRK5  or GRK5  did not significantly affect the G␤␥-stimulated increase in inositol phosphates under the applied experimental conditions (Fig. 3B, columns 1 and 2 versus c). In contrast, the amino terminus of GRK2 lacking the receptor interacting site (21), GRK2  , inhibited the G␤␥-stimulated signal similarly as did GRK2   (Fig. 3B, column 3, versus Fig. 3A). The expression levels of GRK5 1-200 , of GRK5 1-200CVLL , and of GRK2  were similar when determined in immunoblot with antibodies specific for GRK5 or for GRK2 (Fig. 3B, upper panel, lanes 1-3). Thus, the amino-terminal domain of GRK2 contains a G␤␥ regulatory site, which is apparently absent in GRK5.
GST as a control did not affect the phosphorylation of rhodopsin at concentrations Ͻ10 M (not shown). These findings demonstrate that the amino-terminal domain of GRK2 regulates a G␤␥-stimulated signal in cells and in vitro.
The Amino-terminal G␤␥ Binding Domain of GRK2 Regulates Kinase Activity-The carboxyl-terminal G␤␥ binding domain of GRK2 is essential for the G␤␥-dependent stimulation of the kinase activity toward receptor substrates (7). Does the amino-terminal G␤␥ binding domain also affect the kinase activity of GRK2? To determine the effect of the amino terminus on the kinase activity, the amino-terminal G␤␥ binding domain of GRK2 was targeted with immunoselected antibodies. The purified polyclonal antibodies used for this experiment specifically cross-reacted with GRK2 1-185 but did not interact with the carboxyl-terminal G␤␥ binding domain of GRK2 as determined in immunoblot (not shown). The presence of 100 nM antibodies to the amino terminus of GRK2 suppressed the G␤␥-stimulated (10 -40 nM) phosphorylation of rhodopsin by 20 nM GRK2 (Fig. 5, B, lanes 1-3, versus A, lanes 1-3), suggesting that the amino terminus of GRK2 is involved in phosphorylating rhodopsin.
To analyze whether the antibodies interfered with the binding of G␤␥ subunits to the amino terminus of GRK2, the concentration of the purified G␤␥ subunits was increased (Fig. 5, A  and B, lanes 4 -7). G␤␥ stimulated the phosphorylation of rhodopsin by GRK2 in the absence of antibodies (EC 50 ϭ 38 Ϯ 7 nM, Fig. 5A). The presence of 100 nM antibodies specifically cross-reacting with the amino terminus of GRK2 increased the EC 50 value of G␤␥ in stimulating GRK2-mediated rhodopsin phosphorylation more than 20-fold (Fig. 5, B versus A), but the G␤␥ subunits were capable of reversing the antibody-mediated inhibition of the GRK2-induced rhodopsin phosphorylation (Fig. 5B). A higher concentration of the antibodies (250 nM) further increased the EC 50 value of G␤␥ in stimulating GRK2 (not shown). As a control, unrelated antibodies not cross-reactive with GRK2 did not affect the phosphorylation of rhodopsin by GRK2 (not shown). Together these findings provide strong evidence that the antibodies compete with G␤␥ subunits for binding to the amino terminus of GRK2, thereby preventing the stimulatory interaction of G␤␥ with the amino terminus.  1-7). Data Ϯ S.E. are the means (n ϭ 3).

FIG. 6. Differentiation of the amino-and carboxyl-terminal G␤␥ binding sites of GRK2.
A, effect of 100 nM immunoselected antibodies specifically cross-reacting with the carboxyl terminus of GRK2 (ϩanti-C) on the GRK2-mediated phosphorylation of rhodopsin (Rh) in the presence of increasing concentrations of G␤␥ subunits as indicated. As a control, the rhodopsin phosphorylation was determined in the absence of antibodies (Ϫanti-C). B, increasing concentrations of G␤-specific antibodies (anti-G␤) used as G␤␥ scavenger inhibited the rhodopsin phosphorylation by GRK2 (Ϫanti-C), and 100 nM antibodies specifically cross-reacting with the carboxyl terminus of GRK2 did not alter this inhibition (ϩanti-C). C, shielding of the amino terminus of GRK2 with increasing concentrations of site-directed antibodies (anti-N) inhibited the rhodopsin phosphorylation by GRK2 in the presence of 10 nM G␤␥ (Ϫanti-C). Again, 100 nM antibodies specifically cross-reacting with the carboxyl terminus of GRK2 did not alter this inhibition (ϩanti-C). The rhodopsin phosphorylation is expressed as % of control (i.e. the phosphorylation mediated by GRK2 in the presence of 10 nM G␤␥ but in the absence of antibodies). Data Ϯ S.E. are the means of three independent experiments (upper panels), and the bottom panels show autoradiograms of representative experiments.
The interaction of G␤␥ with the amino-terminal G␤␥ binding domain of GRK2 may thus contribute to the G␤␥-dependence of GRK2.
Differentiation of the Amino-and Carboxyl-terminal G␤␥ Binding Sites of GRK2-To differentiate between the aminoand carboxyl-terminal G␤␥ binding sites of GRK2, the effect of domain-specific antibodies to the carboxyl terminus was assessed in the rhodopsin phosphorylation assay. Antibodies specifically cross-reacting with the carboxyl-terminal domain, GRK2 561-689 (anti-C), inhibited the stimulatory effect of G␤␥ at concentrations ranging from 20 nM to 1 M (Fig. 6A). Interestingly, these antibodies did not alter the GRK2-mediated rhodopsin phosphorylation in the presence of less than 20 nM G␤␥ as did the antibodies cross-reacting with the amino terminus ( Fig. 6A versus Fig. 5B). Considering that these low G␤␥ concentrations in the rhodopsin phosphorylation assay are achieved without the addition of purified G␤␥ because the G␤␥ subunits come from the rhodopsin-enriched membranes as determined in immunoblot (not shown), this finding is in good agreement with previous observations; a GRK2 mutant lacking the carboxyl-terminal G␤␥ binding site is capable of phosphorylating rhodopsin but lacks the G␤␥-enhancing effect exerted by the addition of purified G␤␥ subunits (7). Together these data indicate that the amino-and carboxyl-terminal G␤␥ binding sites of GRK2 are functionally different.
The GRK2 activity that was not blocked by the carboxylterminal-specific antibodies was still G␤␥-dependent because increasing concentrations of G␤-specific antibodies used as G␤␥ scavenger inhibited the residual rhodopsin phosphorylation entirely (Fig. 6B). Similar results were obtained with several other G␤␥-binding proteins such as G␣ o or the Raf kinase (not shown). Because the carboxyl-terminal-specific antibodies did not interfere with the G␤␥ scavenger (Fig. 6B), these findings reveal again the second G␤␥ binding site in GRK2, which is distinct from the carboxyl-terminal site (Fig. 6B). For comparison, antibodies to the amino terminus of GRK2 inhibited the GRK2-mediated rhodopsin phosphorylation under similar conditions in a concentration-dependent manner (Fig. 6C). As controls, the antibodies to the amino terminus of GRK2 did not bind G␤␥ (not shown), and carboxyl-terminal antibodies did not interfere with the inhibition exerted by the amino-terminalspecific antibodies (Fig. 6C). Thus, the second G␤␥ binding site in the amino terminus of GRK2 is functionally different from the carboxyl-terminal site and is involved in rhodopsin phosphorylation at low concentrations of G␤␥ (Ͻ20 nM).
The RGS Domain of GRK2 Does Not Interfere with G␤␥ Binding-The amino terminus of GRK2 contains a previously identified RGS domain (Ref. 2 and 3 and Fig. 7D). Does the RGS domain overlap with the amino-terminal G␤␥ binding domain? Two different GST fusion proteins were prepared, GST-GRK2 1-53 and GST-GRK2 54 -185 , encompassing the RGS domain (Fig. 7D). Although GRK2 1-53 inhibited the G␤␥-stimulated phosphorylation of rhodopsin by GRK2 similarly to GRK2   (Fig. 7A, upper panel versus Fig. 4A), the RGS domain, GRK2 54 -185 , had no significant effect when applied at similar concentrations (Fig. 7A, lower panel). This finding strongly suggests that the RGS domain and the amino-terminal G␤␥ regulatory site of GRK2 do not overlap. In addition to the G␤␥ binding site, GRK2 1-53 contains other regulatory sites such as a receptor interacting site (21) or a calmodulin binding site (22). To exclude the possibility that GRK2 1-53 interfered with the kinase activity of GRK2 in a G␤␥-independent manner, we analyzed the G␤␥ regulatory effects of this protein in another G␤␥-dependent assay, the G␤␥-mediated enhancement of the pertussis toxin-catalyzed ADP-ribosylation of G␣ o (19). GRK2 1-53 inhibited the enhancing effect of G␤␥ on the ADP-ribosylation of G␣ o (Fig. 7B) with similar potency and efficacy as did GRK2 1-185 (IC 50 , 290 Ϯ 20 and 360 Ϯ 30 nM of GRK2 1-53 and GRK2 1-185 , respectively). In contrast, the RGS domain, GRK2 54 -185 , had no significant effect at concentrations Ͻ1 M (Fig. 7B). Thus, GRK2 1-53 encompasses the functionally important portion of the amino-terminal G␤␥ regulatory site of GRK2, whereas the RGS domain of GRK2, GRK2 54 -185 , did not interfere significantly with G␤␥ binding. The in vitro findings were confirmed in cells. Although GRK2 1-53 inhibited G␤␥-stimulated signaling similarly to GRK2 1-185 , the RGS domain GRK2 54 -185 did not affect the G␤␥-stimulated increase in inositol phosphates mediated by PLC-␤ 2 (Fig. 7C, first through fourth columns). By contrast, the RGS domain-containing GRK2 1-185 and GRK2 54 -185 efficiently inhibited a G␣ q -stimulated inositol phosphate signal mediated by PLC-␤ 1 that was not affected by GRK2   (Fig. 7C, fifth through eighth columns). DISCUSSION The kinase activity of GRK2 and GRK3 toward receptor substrates is strongly enhanced by G␤␥ subunits. The G␤␥ dependence links the kinase activity of these GRKs to the activation of a heterotrimeric G-protein. A carboxyl-terminal PH domain in GRK2 and GRK3 is essentially involved in the G␤␥ dependence of GRKs (7). Here we present strong evidence that the amino-terminal domain of GRK2 contains a second G␤␥ binding site that contributes to the regulation of GRK2 by low concentrations of G␤␥ subunits; (i) the amino terminus of GRK2 inhibited G␤␥-stimulated signaling in cells and in vitro, (ii) GRK2 1-185 interacted directly with purified G␤␥ subunits, (iii) targeting of the amino-terminal G␤␥ binding domain of GRK2 by site-directed antibodies suppressed the GRK2-mediated phosphorylation of rhodopsin, and (iv) this inhibition was released by an excess of free G␤␥ subunits.
The amino terminus of GRK2 contains several important structural elements, an RGS-domain affecting G␣ q -stimulated signaling (2, 3), a calmodulin binding site (22), which is regulated by protein kinase C phosphorylation (23), and a receptor interacting site (21). A receptor interacting site and a calmodulin binding site were also localized in the carboxyl-terminal domain of GRK2 (22,24). These functional similarities of the amino-and the carboxyl-terminal domains of GRK2 are complemented by the localization of a previously unrecognized G␤␥ binding site in the amino terminus of GRK2. The novel aminoterminal G␤␥ binding site is involved in the G␤␥ dependence of GRK2 in addition to the carboxyl terminus. A topological model of GRK2 appears to consist of a core kinase domain flanked by two structurally and functionally different G␤␥ binding domains. With two G␤␥ binding sites, GRK2 activity is tightly controlled by G␤␥ subunits over a wide concentration range. Thereby G␤␥ subunits translate the intensity of a G-proteinstimulated signal into GRK2 activity to switch off the signalgenerating receptor.
Apart from the functional importance of the newly identified G␤␥ binding site in the amino terminus of GRK2, this domain may constitute a novel target allowing the selective inhibition of GRK2-mediated receptor phosphorylation by pharmacological tools. Site-directed antibodies to the kinase amino terminus suppressed the phosphorylation of rhodopsin by GRK2. Because such an inhibition was released by the addition of an excess of G␤␥ subunits, blockade of GRK2 activity by pharmacological compounds binding to the amino terminus of GRK2 would be reversed upon excessive G-protein activation, i.e. G␤␥ release. The proposed mechanism could allow the design of fine-tuning GRK inhibitors, which would amplify low threshold signals and maintain desensitization of excessive stimuli. Additional experiments will have to identify such compounds to validate the proposed principle under physiological conditions.