An Intact N Terminus of the Subunit Is Required for the G Stimulation of Rhodopsin Phosphorylation by Human -Adrenergic Receptor Kinase-1 but Not for Kinase Binding

Cleavage after lysine 32 in the G subtype and after lysine 36 in the G subtype of purified mixed brain Gβ by endoproteinase Lys-C blocks Gβ-mediated stimulation of phosphorylation of rhodopsin in urea-extracted rod outer segments by recombinant human β-adrenergic receptor kinase (hβARK1) holoenzyme while hβARK1 binding to rod outer segments is partially affected. This treatment does not attenuate the binding of the treated Gβ to C-terminal fragments of hβARK1 containing the pleckstrin homology domain. Lys-C proteolysis also does not alter the association of the Gβ with phospholipids, its ability to support pertussis toxin-catalyzed Gα/Gα ADP-ribosylation, or its ability to inhibit forskolin-stimulated platelet adenylate cyclase. The Gβ subunit remains noncovalently associated with the cleaved G fragments. Thus, in addition to recruiting hβARK1 to its receptor substrate, G contributes secondary and/or tertiary structural features to activate the kinase.

Cleavage after lysine 32 in the G␥ 2 subtype and after lysine 36 in the G␥ 3 subtype of purified mixed brain G␤␥ by endoproteinase Lys-C blocks G␤␥-mediated stimulation of phosphorylation of rhodopsin in urea-extracted rod outer segments by recombinant human ␤-adrenergic receptor kinase (h␤ARK1) holoenzyme while h␤ARK1 binding to rod outer segments is partially affected. This treatment does not attenuate the binding of the treated G␤␥ to C-terminal fragments of h␤ARK1 containing the pleckstrin homology domain. Lys-C proteolysis also does not alter the association of the G␤␥ with phospholipids, its ability to support pertussis toxin-catalyzed G␣ o /G␣ i ADP-ribosylation, or its ability to inhibit forskolin-stimulated platelet adenylate cyclase. The G␤ subunit remains noncovalently associated with the cleaved G␥ fragments. Thus, in addition to recruiting h␤ARK1 to its receptor substrate, G␥ contributes secondary and/or tertiary structural features to activate the kinase.
G-protein-coupled receptor responses to agonist ligands are modulated at multiple levels along the signal transduction pathway. Regulation includes short term effects on receptor coupling and on internalization of the receptor. Longer term effects include down-regulation of the cellular receptor content and mRNA levels (Tholanikunnel et al., 1995). Several different protein kinases have been shown to phosphorylate some of the seven transmembrane helix receptors on cytosolic portions of the molecule, reducing coupling of the receptors to their associated heterotrimeric G-proteins (reviewed by Kobilka (1992)). One family of protein kinases, the G-protein receptor kinases (GRKs), 1 has been defined, (Premont et al., 1995;Inglese et al., 1993). Three GRKs have been shown to target agonist-occupied receptors, phosphorylating multiple serine/ threonine residues adjacent to acidic amino acid residues in the primary amino acid sequence. There are presently six members of this protein kinase family that share a common catalytic domain, diverging in the N-and C-terminal extensions outside of this region.
The GRK2/GRK23 (␤ARK1 and ␤ARK2) subfamilies are en-coded by separate genes (Benovic et al., 1991). They were originally shown to phosphorylate the ␤ 2 -adrenergic receptor and thus were given the name ␤-adrenergic receptor kinases or ␤ARKs. The ␤ARKs are C-terminally extended relative to other members of the GRK family. The C-terminal 222 amino acids of ␤ARK1 and ␤ARK2 contain a domain responsible for the association of the enzyme with heterotrimeric G-protein G␤␥ subunits (Pitcher et al., 1992). Addition of G␤␥ subunits to an in vitro phosphorylation system stimulates ␤ARK phosphorylation of receptor substrates Haga, 1990, 1992), but not that of peptide substrates (Pitcher et al. 1992). This Cterminal domain of the kinase includes a region homologous to a domain of the platelet protein pleckstrin (PH domain) (Touhara et al., 1993), thought to mediate protein-protein interactions among signaling proteins (Musacchio et al., 1993;Gibson et al., 1994;Ingley and Hemmings, 1994). PH domains may be functionally analogous to the Src homology 2 and 3 domains of tyrosine kinase signaling systems (Mayer et al., 1993). While the portions of the ␤ARK C terminus involved in the association with G␤␥ subunits have been delineated , less is known about the determinants on the G␤␥ partner. The multiple subtypes of G␤ and G␥, the requirement for the G␤␥ heterodimer for cellular function, (Iniguez-Lluhi et al., 1992), and multiple post-translational modifications (Yamane and Fung, 1993) have impeded study. This paper describes the dissociation of G␤␥ binding to h␤ARK1 from G␤␥ stimulation of rhodopsin phosphorylation by this kinase after proteolysis with Lys-C. G␥ is cleaved by the protease at lysine 33 in G␥ 2 (lysine 36 in G␥ 3 ), and the G␥ fragments remain associated with the G␤ subunit.
Heterotrimeric G-proteins were isolated from frozen bovine brain, and the G␤␥ subunit complex (mixed subtypes of ␤ and ␥ subunits) purified by chromatography in sodium cholate on heptylamine-Sepharose (Sternweis and Pang, 1990) in the presence of GDP/AlMgF. Further resolution of the separated G␤␥ from G␣ subunits for ADP-ribosylation studies was achieved by ion exchange chromatography on a MonoQ (Pharmacia) column in 0.1% (w/v) Lubrol PX (Sternweis and Pang, 1990). The purified G␣ and G␤␥ subunits were stored in aliquots at Ϫ80°C. Two different batches of G␤␥ when assayed for effects of Lys-C proteolysis on rhodopsin phosphorylation and h␤ARK1 binding activity revealed no discernible differences. Carboxymethylation of the C terminus of the ␥ subunit of G␤␥ with S-adenosyl-L-[ 3 H-methyl]methionine was accomplished using a cholate-extracted brain membrane fraction (Fung et al., 1990). Bovine retinal rod outer segments (ROS) containing rhodopsin were purified under red light illumination and urea-extracted before use as a phosphorylation substrate (Phillips et al., 1989). SDS-PAGE was performed in 10% acrylamide, 0.267% bisacrylamide gels containing 0.1% SDS in the Laemmli buffer system (Laemmli, 1970). Rhodopsin phosphorylation was determined after separation of 32 P-labeled proteins by SDS-PAGE. Radioactive gels were fixed for 10 min in 25% methanol, 10% acetic acid, washed with distilled water for 10 min, and the gels dried at 80°C under vacuum before exposure to Kodak XAR-5 or X-Omat-LS film. The 32 P radioactivity of the rhodopsin band was quantitated from the film using a BioImage (Millipore) scanner. Tritiated proteins separated by SDS-PAGE were visualized after fixation in 25% methanol, 10% acetic acid, soaking in an Amplify enhancement solution for 45 min, and drying and exposing to film at Ϫ80°C for 3-7 days. ␤ARK1 and G␤ were determined after transfer of polypeptides separated by SDS-PAGE on a 10% acrylamide, 0.367% bisacrylamide in the Laemmli buffer system (Laemmli, 1970) to 0.2-m pore size nitrocellulose (LeVine and Sahyoun, 1988). An affinity-purified antibody directed against the C-terminal 222 amino acids of h␤ARK1 recognizing the PH domain of that kinase (Sallese et al., 1995) was used to detect the enzyme. G␤ subunits were detected with pan-␤ subunit-specific antibodies. ␤ and ␥ subunits of the heterotrimeric Gproteins were separated on 16.5% acrylamide, 0.5% bisacrylamide gels containing 6 M urea and 0.1% SDS in a Tris-Tricine buffer system (Schagger and von Jagow, 1987). The subunits were transferred as described for ␤ARK1 and were detected with G␤-or G␥-subtype subunit-specific antibodies. Quantitation of the ECL was standardized as described (Mahadevan et al., 1995).
Measurement of the Ability of G␤␥ to Stimulate Rhodopsin or Synthetic Peptide Phosphorylation by h␤ARK1-G␤␥ subunits (20 -1400 nM) were incubated in a 60-l reaction volume containing 20 mM Tris-HCl, pH 7.5, 2 mM EDTA, 6 mM MgCl 2 , 16 Ci/ml [␥-32 P]ATP, 50 M ATP, 10 M GTP␥S, and 160 pmol of rhodopsin as urea-extracted ROS. After 10 min on ice, 5 l of 50 g/ml recombinant h␤ARK1 purified from Sf9 cells were added, and the reaction was illuminated with a fluorescent table lamp at 30°C for 20 min. The rate of phosphorylation was linear for at least 30 min under these conditions. The reaction was terminated with SDS-sample buffer without boiling, and the proteins resolved by SDS-PAGE. Phosphorylation of the synthetic peptide RRREEEEESAAA (Immunodynamics, Inc., La Jolla, CA) was carried out in the same reaction buffer with 100 M peptide in place of ROS. The reaction was stopped by the addition of 0.5 volume of 30% trichloroacetic acid. Phosphorylated products were spotted onto Whatman P-81 phosphocellulose paper and washed with 75 mM phosphoric acid three times for 15 min each to remove the free nucleotide. Phosphorylated peptide was determined by liquid scintillation counting of the paper (Cook et al., 1982). Translocation of h␤ARK1 was performed according to Chuang et al. (1992) and involved a 3-min incubation of 300 pmol of rhodopsin (as urea-treated ROS), 5 l of 50 g/ml h␤ARK1, and absence or presence of 178 nM G␤␥ under a fluorescent table lamp at room temperature in 20 mM Tris-HCl, pH 7.8, 2 mM EDTA, 10 mM MgCl 2 , 2 mM dithiothreitol. The ROS were pelleted at 4°C at 109,000 ϫ g for 10 min in a TLA 100.3 rotor with microcentrifuge tube adapters. The pellets were resuspended in 50 l of 20 mM Tris-HCl, pH 8.0, and assayed as above for rhodopsin phosphorylation.
G␤␥ Subunit-mediated ADP-ribosylation of G␣ o and G␣ i by Pertussis Toxin-The ability of G␤␥ subunits to support pertussis toxin-catalyzed [ 32 P]ADP-ribose transfer from [␣-32 P]NAD ϩ to G␣ o /G␣ i was determined essentially as described by (Kwon et al., 1993), with 5 M NAD ϩ (2 Ci/assay), 0.5 mM dimyristoyl phosphatidylcholine, and 7.5 g/ml activated pertussis toxin in a final volume of 50 l of 75 mM Hepes buffer, pH 8.0, 2 mM dithiothreitol, 1 mM MgCl 2 , 1 mM EDTA, and 10 M GDP. The dose response for G␤␥ subunits was determined at a constant G␣ o /G␣ i concentration. After SDS-PAGE, the incorporation of 32 P into polypeptides migrating in the G␣ region was determined by densitometry from autoradiograms.
Adenylate Cyclase Inhibition by G␤␥ Subunits-Dose-dependent inhibition of platelet membrane type I adenylate cyclase by G␤␥ subunits was determined as described by Kwon et al. (1993), except that cAMP was determined by Scintillation Proximity immunoassay (Amersham) according to the manufacturer. Dilutions of G␤␥ subunits were made in 20 mM Tris-HCl, pH 7.6, 0.1% (v/v) Lubrol PX. Reactions were run in a 100-l final volume containing 25 mM Tris-HCl, pH 7.6, 2.5 mM EGTA, 2 mM MgCl 2 , 0.1 mM isobutylmethylxanthine, 10 M propanolol, 0.2 mM ATP, 0.01 mM GTP, 0.8 mM phosphoenolpyruvate, 10 M forskolin, and 10 g/ml pyruvate kinase. Ten microliters of diluted G␤␥ subunits were preincubated on ice in the reaction mixture with 5 l (15 g) of platelet membrane protein for 5 min. The tubes were transferred to a 30°C water bath and the incubation continued for 15 min. One hundred microliters of ice-cold 5% (w/v) trichloroacetic acid were added to stop the reaction on ice. Cyclic AMP content was determined on aliquots of the supernatant after centrifugation at 16,000 ϫ g for 10 min at room temperature to remove precipitated material. Adenylate cyclase activity was linear in both membrane protein and time over the ranges used.
Proteolysis of G␤␥ by Endoproteinase Lys-C-Cleavage of the G␤␥ complex by Lys-C protease was performed in 50 mM Tris-HCl, pH 8.6, 1 mM dithiothreitol, 0.1 mM EDTA, 0.8% cholate, 0.05% Lubrol PX for 30 min at 30°C using a protease:protein ratio of 1:10 (w/w). The digestion was terminated by the addition of 20 g/ml leupeptin. To facilitate comparison of protease treatment, the different functional assays were carried out with the same batches of G␤␥ stored at Ϫ70°C that had been kept on ice, incubated without protease at 30°C, incubated with leupeptin-inactivated protease at 30°C, or incubated with active Lys-C at 30°C for 30 min. Thirty minutes of incubation of G␤␥ at 30°C without protease had no discernible effect on its measured biochemical properties.
Binding of G␤␥ Subunits to PH Domain-containing C-terminal ␤ARK1 Fragments-G␤␥ subunit binding to immobilized PH domains was determined by immunoblot analysis with pan-anti-␤ subunit antibodies as described by Mahadevan et al. (1995).
Amino Acid Sequencing of Lys-C-cleaved G␤␥-Separation of the G␤ and G␥ subunits was achieved by C4 reverse phase chromatography under the conditions used to separate the farnesylated ␥ subunit of transducin on a C18 column (Parish and Rando, 1994). Western blot analysis showed that the G␤ and G␥ subunits were separated under these conditions (data not shown). Fifty g of Lys-C-treated G␤␥ (ϳ1.2 nmol) was diluted with an equal volume of 6 M guanidine HCl, 50 mM dithiothreitol and injected onto a Brownlee 4.6 ϫ 30-mm 300-Å pore size C4 reverse phase column equilibrated with 5% (v/v) acetonitrile containing 0.1% (v/v) trifluoroacetic acid at 0.75 ml/min. After a 10-min wash period, the proportion of acetonitrile was linearly increased to 65% over 40 min and held at 65% for 5 min. The effluent was monitored during the separation at 214 nm, and 0.75-ml fractions were collected. Fractions containing the ␥ subunits as judged by comigration of digested [ 3 H]carboxymethylated G␤␥ (on a separate HPLC run) and the appearance of new 214 nm absorbing peaks were reduced to near dryness under vacuum on a Savant SpeedVac. The fractions were subjected to 15 cycles of automated Edman degradation on an Applied Biosystems 477A protein sequencer and the phenylthiohydantoinamino acids identified with the integral HPLC unit of the sequencer.

RESULTS
The G␤␥ subunits of the heterotrimeric G-proteins form a heterodimer (␤ ϭ 35 kDa or 36 kDa, ␥ϭ 6748 -8321 kDa), depending on the subunit subtypes and post-translational modifications. This non-covalent complex is not dissociable under non-denaturing conditions and associates reversibly with an undetermined variety of G␣ subtypes and with a protein kinase A substrate, phosducin (Bauer et al., 1992), as well as a host of effector molecules such as adenylate cyclase, ion channels, and phospholipases (Clapham and Neer, 1993). The G␥s are multiply post-translationally modified by N termini N-acetylation (Wilcox et al., 1994) and C-terminal cysteine geranylgeranylation and carboxymethylation (reviewed by Yamane and Fung (1993)). Although these modifications are not required for assembly of the G␤␥ dimer, prenylation apparently modulates interactions with other proteins (Kisselev et al., 1994) and their interaction with the lipid bilayer (Muntz et al., 1992;Iniguez-Lluhi et al., 1992). G␤␥ requires small amounts of detergent to remain in solution.
Lys-C Proteolysis of G␤␥-The structure of the G␤␥ dimer is compact, judging from its hydrodynamic properties (Huff and Neer, 1986) and by the resistance of the native protein to proteases such as trypsin (Fung and Nash, 1983;Tamir et al., 1991;Winslow et al., 1986). Proteolysis by the endoproteinase Lys-C from L. enzymogenes was carried out in sodium cholate, a detergent with a low aggregation number (small number of detergent molecules per micelle) (Calbiochem, 1993) to minimize interference with the hydrophobic C terminus of G␤␥.
Reducing SDS-PAGE analysis of the proteolytic product as a function of protease:G␤␥ ratio revealed a size range for the G␥ fragment of M r 3400 -5200 by silver staining, a reduction from M r 6900 -8600, corresponding to an M r of 3500. There was no discernible effect on the M r of the G␤ subunit (Fig. 1). The mixed G␤␥ preparation isolated from frozen bovine brain contains multiple subtypes of G␤ and G␥ subunits, the latter accounting for the smear around M r 6500. Antibodies raised to synthetic peptides corresponding to the N-terminal 17 residues of the G␥ subunits or 21 amino acids of the G␤ subunits (Santa Cruz Biotechnology) are subtype-selective and can be used to determine the distribution of the subtypes in the preparation. Western immunoblotting of two separate purifications revealed that the purified bovine brain G␤␥ used in these studies has the following distribution of G␤ and G␥ subunits: ␤ 4 Ͼ Ͼ ␤ 2 Ͼ ␤ 1 Ͼ ␤ 3 ; ␥ 2 Ͼ Ͼ ␥ 3 Ͼ Ͼ ␥ 5 Ͼ 7 (data not shown). This quantitation assumes that the subtype-selective antibodies are equally capable of detecting antigen at 1.6 g/ml antibody. In addition, the purification of the G␤␥ subunits could influence the recovery and thus the apparent distribution of the different subtypes.
Methylation of the C-terminal cysteine (Cys 68 ) carboxyl moiety of the purified brain G␤␥ with S-adenosyl-L-[ 3 H-methyl]methionine (Fung et al., 1990) provided a marker for the C termi-nus of the G␥ subunit. Lys-C treatment of the tritiated G␤␥ released a labeled fragment of M r 3500 (Fig. 2), mirroring the shift on SDS-PAGE to lower M r seen with silver staining (Fig.  1). This implies that the extreme C terminus of the G␥ subunit remains intact. Immunoblotting after SDS-PAGE showed that N-terminal G␥ subunit immunoreactivity (residues 2-17) was eliminated by Lys-C treatment (Fig. 3). By contrast, for the G␤ subunit, neither the N-terminal immunoreactivity (data not shown) nor the size on SDS-PAGE ( Fig. 1) was affected by Lys-C treatment. Gel permeation chromatography of the [ 3 H]carboxymethylated G␤␥ on a Superose 12 column in 0.8% cholate before or after Lys-C treatment demonstrated that the fragments of the cleaved G␥ subunit remained complexed with the G␤ subunit (Table I). When the Superose 12 column fractions were assayed by enzyme-linked immunosorbent assay and slot blotting onto nitrocellulose, virtually no loss of the immunoreactive N-terminal fragment(s) of G␤␥ was observed. By contrast, denaturing SDS-PAGE and subsequent Western blot analysis of the same fractions revealed a significant loss of N-terminal G␥ subunit immunoreactivity (Table I).
Amino Acid Sequence of the Lys-C-treated G␥ Subunit-G␤␥ subunits were digested with Lys-C, dissociated with 3 M guanidine hydrochloride ϩ 50 mM dithiothreitol, and separated by C4 reverse phase chromatography as described under "Experimental Procedures." Fig. 4 shows the migration of the [ 3 H]carboxymethyl marker of the C terminus of the G␥ subunit on a separate HPLC run. The arrow marks the position of the truncated C-terminal G␥ sequences (Ala 33 -Cys 67 for G␥ 2 ; Ala 37 -Cys 71 for G␥ 3 ). The two G␥ subtypes nearly comigrated on reverse phase chromatography so their sequences were determined simultaneously. This was possible because both sequences are known, they were present in different amounts, and they differ at several positions. The asterisk shows the position of the undigested G␥ 2 and G␥ 3 polypeptides, while the cluster of radioactive peaks around fraction 40 represent undissociated G␤␥ and the unidentified carboxymethylated M r 24,000 protein in brain membranes (see Fig. 1 and Fung et al.

FIG. 2. Reduction in M r of C-terminal [ 3 H]carboxymethylated bovine brain G␤␥ by Lys-C proteolysis visualized by 3 H autora-
diography. 100 g of bovine brain G␤␥ were carboxymethylated with 0.5 mg of cholate-extracted bovine brain membrane protein and 50 Ci of S-adenosyl-L-[ 3 H-methyl]methionine (77 Ci/mmol) in 0.5 ml of carboxymethylation buffer and the labeled proteins re-extracted with cholate as described under "Experimental Procedures." Lane 1, [ 3 H]carboxymethylated proteins incubated for 30 min at 30°C in the absence of protease; lane 2, ϩ 2 g of endoprotease Lys-C. The M r 24,000 labeled protein is an endogenous substrate for the carboxymethyltransferase derived from the brain membrane enzyme source. This contaminant represents a minor fraction of the 3 H label incorporated. Lys-C treatment leads to a M r ϳ3500 decrease in size of the labeled G␥ subunit. The leading ion front ran to the very bottom of the gel.
(1990)). Table II shows the sequences obtained from the major peaks of 214 nm absorbance produced by Lys-C treatment of unlabeled isolated G␤␥ subunits. G␥ 2 , the major immunoreactive G␥ subtype in the purified brain G␤␥ preparation, was likewise the prominent proteolytic fragment represented. Both G␥ 2 and G␥ 3 yielded peptides beginning with Ala 33 (numbering system of G␥ 2 , A36 of G␥ 3 ) immediately following Lys 32 , consistent with the cleavage specificity of Lys-C after lysine. The [ 3 H]carboxymethylation experiments indicate that this C-terminal peptide extends to the modified penultimate cysteine. The removal of 32(35) amino acid residues is consistent with the M r change in G␥ seen by SDS-PAGE (ϳ3500, Figs. 1 and 3). G␥ sequences corresponding to other potential Lys-C proteolytic fragments of G␥ 2 , G␥ 3 , or other known G␥ subtypes were not detected. Minor 214 nm absorbance peaks contained short C-terminal G␥ fragments that were unrelated to Lys-C cleav-age specificity and probably represent co-purified proteolytic fragments in the initial brain preparation. Small amounts of sequences corresponding to N-terminal methionine-containing and non-acetylated G␥ 2 and G␥ 3 were also detected. The same sequencing results were obtained for two independent preparations of G␤␥ subunits.
Functional Effects of Cleavage of the G␥ Subunit: Biochemical Activities of G␤␥ Subunits-Since G␤␥ subunits do not possess an intrinsic enzymatic activity, the effect of Lys-C truncation on several extrinsic measures of functionality of the treated and control G␤␥ subunits was evaluated. Their ability to bind to ROS and to modulate rhodopsin phosphorylation by full-length recombinant h␤ARK1 was measured. Their ability to bind to the GST-␤ARK C-terminal domain of 222 amino acids (Pro 466 -Leu 689 ) and to bind to a more restricted region of the kinase, GST-or His 6 -tagged ␤ARK PHϩC domain (␤ARK G556-S670), was determined. Finally, the ability of the treated and control G␤␥ subunits to associate with and to modulate G␣ subunit activities and to regulate the activity of an effector, Type I adenylate cyclase, were also assessed.
Modulation of ␤ARK1-mediated Receptor Phosphorylation-The Lys-C-treated G␤␥ subunit preparation supported substantial translocation of human ␤ARK1 to isolated retinal rod outer segments (Fig. 5A), but failed to stimulate phosphorylation of rhodopsin (Fig. 5B). G␤␥ subunits treated with leupeptin-inhibited Lys-C fully supported both translocation and phosphorylation. Fig. 5C shows that Lys-C proteolysis of G␥ 2 and G␥ 3 abrogates stimulation of ␤ARK1 phosphorylation of receptor substrates without abolishing binding of the kinase. The saturation of the dose response of phosphorylation observed at low concentrations of Lys-C-digested G␤␥ contrasts with the roughly linear increase with control incubated G␤␥. The low stimulation seen in the Lys-C-treated G␤␥ may be residual undigested G␤␥, or it may represent an intrinsically lower stimulatory activity. Phosphorylation of rhodopsin was linearly related to the amount of ␤ARK1 and G␤␥ present in FIG. 3. Removal of N-terminal G␥ subunit immunoreactivity by Lys-C proteolysis. A total of 10 g of bovine brain G␤␥ was incubated with the indicated ratio of endoprotease Lys-C for 30 min at 30°C. The control lane was material incubated without Lys-C. The digest was then separated by Tris-Tricine-urea SDS-PAGE and transferred to nitrocellulose as described under "Experimental Procedures." The blot was reacted with a 1:1 mixture of G␥-antibodies specific for the N-terminal residues (2-17) of the G␥ 2 and G␥ 3 subtypes and the immunoreactivity quantitated as referenced under "Experimental Procedures." The experiment was repeated twice as single determinations with similar results. A, raw film data.  H]carboxymethylated G␤␥ were treated with either buffer or 1 g of Lys-C in a final volume of 100 l for 1.5 h at 30°C. The samples were dissociated with guanidine hydrochloride and subjected to reverse phase chromatography as described under "Experimental Procedures." The 3 H content of the 0.75-ml fractions collected was determined by liquid scintillation counting. Circles, incubated without protease; squares, ϩ 1 g of Lys-C. Arrow, C-terminal G␥ peptides (see text); asterisk, undigested G␥ 2 , G ␥ 3 polypeptides. the assay. The lack of effect of G␤␥ subunits on synthetic peptide substrate RRREEEEESAAA was unaltered by proteolysis (data not shown).
Binding of G␤␥ Subunits to the ␤ARK1 PH Domain-There was no notable difference between Lys-C-treated and untreated G␤␥ subunit association with the PH domain contained within the C-terminal 222-amino acid fragment of h␤ARK1 (GST-(P466-L689)) ( Fig. 6), or the shorter (PH ϩ C-terminal helix) (G556-S670) GST-fusion or His 6 -tagged (PHϩC) domains (data not shown) over a range of G␤␥ concentrations. Thus, Lys-C proteolysis of G␥ does not detectably alter association of G␤␥ subunits with the kinase in the absence of the catalytic and N-terminal sequences of h␤ARK1.

Effects of N-terminal G␤ and G␥ Synthetic Peptides on G␤␥ Subunit Interactions with h␤ARK1-Antibodies (Santa Cruz
Biotechnology) to N-terminal residues (2-17) of G␥ (␥ 2 or ␥ 3 ) or the N-terminal region of G␤ (residues 25-40 of ␤ 2 or ␤ 4 ; 20 g/ml antibody) or the synthetic peptides in 1 mg/ml gelatin supplied by Santa Cruz Biotechnology) corresponding to these regions (40 g/ml, 25 M) were used to try to block or enhance ␤ARK1 binding to G␤␥ to immobilized PH domains or to block or stimulate ␤ARK1 phosphorylation of rhodopsin in ROS without significant effect (data not shown). Similar results were seen for peptides corresponding to residues 8 -34 of G␥ 2 or residues 3-29 of G␤ 2 (200 M each). This lack of effect suggests either that the most N-terminal segments of G␥ and G␤ are not involved in the assayed interactions with ␤ARK1 or that the interactions with the free peptides are of low affinity.
Catalysis of Pertussis Toxin-mediated ADP-ribosylation of G␣ Subunits-Pertussis toxin, an ADP-ribosylating enzyme, modifies a cysteine in the G␣ o and G␣ i family of G-proteins only when the G␣␤␥ heterotrimer is formed (Neer et al., 1984). Isolated mixed G␣ i /G␣ o from brain was titrated with proteolyzed and control G␤␥ subunits. The reconstituted G-proteins were treated with pertussis toxin as described under "Experimental Procedures." A dose-dependent increase to saturation was observed with increasing amounts of G␤␥ subunits added to G␣ subunits. For the particular amount of G␣ i /G␣ o used, saturation was achieved around 600 nM G␤␥ subunits. No significant difference was noted for any of the treatments of the G␤␥ subunits (Fig. 7A). Heating of the G␤␥ subunits at 65°C for 10 min abolished their ability to support ADP-ribosylation of G␣ subunits.
Inhibition of Platelet Adenylate Cyclase by G␤␥ Subunits-The type I adenylate cyclase is inhibited by G␤␥ subunits at a site distinct from that modulated by G␣ (Taussig et al., 1994). Titration of forskolin-stimulated platelet membrane Type I adenylate cyclase activity with G␤␥ stored on ice, G␤␥ subunits treated with Lys-C inhibited with leupeptin, and Lys-C-treated G␤␥ subunits were indistinguishable (Fig. 7B). Proteolytic nicking of the G␥ subunit, therefore, did not detectably alter the adenylate cyclase modulatory function of G␤␥ subunits. DISCUSSION G␤␥ subunits participate in a diverse range of biological interactions, from anchoring and modulating the GTPase ac-tivity of G␣ subunits, to activation or inhibition of transmembrane signaling effectors such as phospholipase C ␤ and ␥ isozymes, phospholipase A 2 , adenylate cyclase isozymes, and ion channels (reviewed by Clapham and Neer (1993)). They are also critically involved in regulating the activity of several of the G-protein coupled receptor kinases (GRK2 and GRK3) toward receptor substrates . The multiple isotypes (5-␤ and 7-␥ isotypes) of G␤␥ subunits provide a daunting array of possible combinations, which may be significant (Kleuss et al., 1993;Kleuss et al., 1992). The assembly of the dimeric structure is further complicated by their subsequent extensive post-translational modification and proteolytic processing. Chimeric studies in COS7 cells taking advantage of the differential interaction of G␥ 1 with G␤ 1 but not G␤ 2 , while G␥ 2 will interact with both G␤s (Simonds et al., 1991), defined multiple sites of interaction of the ␤ subunit with the ␥ subunit, and assigned subtype selectivity to G␤ 1 residues 215-340 (Garritsen and Simonds, 1994) or residues 210 -293 (Katz and Simon, 1995). A series of G␥ 2 truncation experiments suggest that a region between residues 45-59 in the G␥ subunit are involved with dimerization with G␤ 1 (Mende et al., 1995).
At high concentrations trypsin will cleave the G␤ subunit at Arg 129 , generating two proteolytic fragments of M r 26,000 and 15,000 without noticeable effect on the G␥ subunit in native G␤␥ subunits (Tamir et al., 1991). The accessibility of the cleavage site in the ␥ subunit to Lys-C proteolysis demonstrated in the present work illuminates the quaternary structure of the G␤␥ dimer. To minimize possible interactions of the prenyl moiety with lipid bilayers and potential steric hindrance, the proteolysis was performed in cholate, an anionic detergent with the small aggregation number of 2 cholate molecules/micelle (Calbiochem, 1993). There are 7 (G␥ 2 ) or 6 (G␥ 3 ) lysines in the G␥ subunit, which could potentially be available for Lys-C proteolysis. The major identified G␥ products of this protease begin with Ala 33 (G␥ 2 ) and Ala 37 (G␥ 3 ). Monitoring of the migration of the C-terminal [ 3 H]carboxymethylated G␥ on HPLC gave no evidence for specific smaller or larger cleavage products. Thus, the majority of the treated G␥ retains the prenylated C-terminal cysteine residue. This part of the molecule is thought to interact with the seven-transmembrane helix receptor in the fashion of transducin of the visual system. The farnesylated C terminus of G t ␤ 1 ␥ 1 , stabilizes the active MII form of rhodopsin, which activates G t (transducin) in the presence of GTP (Kisselev et al., 1994). Hints as to the relative positioning of the ␤ and ␥ subunits come from several sources. Copper o-phenanthroline-mediated cross-linking of G␤ 1 and G t ␥ 1 through proximal intersubunit cysteine residues in transducin indicates that ␤ Cys 25 and ␥ Cys 36 or Cys 37 are nearby in the three-dimensional protein structure (Bubus and Khorana, 1990). This was interpreted as a point of close contact between the subunits. Alignment of transducin G␥ 1 with G␥ 2 and G␥ 3 places the reactive Cys 36 /Cys 37 at the site of Lys-C digestion determined here. Perhaps lysines C-terminal to this position are not accessible to the protease because of ␤ subunit-␥ subunit interactions or interactions involving the prenyl group. A  series of sequential amino acid replacements in G␥ 2 (35-37) and G␥ 1 (38 -40) are sufficient to specify the appropriate ␤-␥ selectivity (Lee et al., 1995). Interestingly, this motif is immediately C-terminal to the Lys-C cleavage site. G␤␥ subunits have been shown to mediate GRK2 and GRK3 (␤ARK1 and ␤ARK2) translocation to retinal ROS containing the light receptor substrate rhodopsin, cell membranes (Pitcher et al. 1992), or to phospholipid vesicles (Kim et al., 1993). In addition, they robustly stimulate the activity of the kinase 10 -20-fold toward receptor substrates but much less so toward synthetic peptide substrates (Pitcher et al. 1992). Cleavage of the G␤-␥ complex with Lys-C retained the two fragments 1-32(36) and 33(37)-68(71) non-covalently associated with the G␤ subunit. Tryptic fragments of the G␤␥ complex cleaved within G␤ similarly remain attached (Fung and Nash, 1983;Thomas et al., 1993). G␥ is not cleaved by trypsin in the G␤␥ complex (Tamir et al., 1991). The Lys-C G␤␥ fragment complex retained the ability to function as a binding site for the PH graphic data are shown at the top. The first five lanes are G␤␥ subunits treated with Lys-C in the presence of 10 g/ml leupeptin, while the final four lanes are Lys-C-digested G␤␥ subunits. Lane 1, no G␤␥ subunits; lanes 2 and 6, 27.9 nM G␤␥ subunits; lanes 3 and 7, 55.8 nM; lanes 4 and 8, 112 nM; lanes 5 and 9, 223 nM.
FIG. 5. Modulation of ␤ARK1-mediated receptor phosphorylation. Incubations of G␤␥ subunits with ROS and ␤ARK1 and phosphorylation were performed as described under "Experimental Procedures." A, translocation of ␤ARK1 to rod outer segment membranes containing rhodopsin. The amount of ␤ARK1 binding to light-activated urea-extracted ROS in the absence, presence of G␤␥ subunits, and presence of Lys-C digested G␤␥ subunits was determined after SDS-PAGE by immunoreactivity with 0.1 g/ml anti-␤ARK1 antibody. Quantitation of the ECL reaction was by densitometry. Raw film data are shown at the top. Lane 1, translocation of ␤ARK1 to ROS in the absence of G␤␥ subunits; lanes 2 and 3, Lys-C digested G␤␥ subunits; lanes 4 and 5, G␤␥ subunits treated with Lys-C in the presence of 10 g/ml protease inhibitor leupeptin. The experiment was repeated three times with similar results. B, lack of stimulation of rhodopsin phosphorylation by proteolyzed G␤␥ subunits. ␤ARK1 phosphorylation of rhodopsin in ROS was determined after translocation of the kinase in the absence, presence, and presence of Lys-C-digested G␤␥ subunits. Following SDS-PAGE, the 32 P incorporated into rhodopsin was quantitated by densitometry of the autoradiogram. C, dose response of G␤␥ subunits on rhodopsin phosphorylation. ROS were phosphorylated with ␤ARK1 and the indicated concentrations of control incubated or Lys-C digested G␤␥ subunits as indicated under "Experimental Procedures" without removing unbound proteins. Following SDS-PAGE the 32 P incorporated into rhodopsin was quantitated by densitometry of the autoradiogram. Circle, control incubated G␤␥ subunits; square, Lys-C digested G␤␥ subunits. Data are single determinations, and the experiment was repeated twice with similar results. The raw autoradio-FIG. 6. Effect of Lys-C proteolysis of G␤␥ subunits on binding to the isolated PH domain of ␤ARK1. The indicated concentrations of treated G␤␥ subunits were incubated in a 50-l reaction volume with 0.5 M GST-␤ARK1 C-terminal (Pro 466 -Leu 689 ) fusion protein immobilized on glutathione-Sepharose as indicated under "Experimental Procedures." The beads were washed and subjected to SDS-PAGE, and the proteins transferred to nitrocellulose for immunodetection of the G␤ with a pan-G␤-specific antibody. A, raw film data. Lanes 1, 5, and 9, 106 nM G␤␥ subunits; lanes 2, 6, and 10, 319 nM; lanes 3, 7, and 11, 957 nM; lanes 4, 8, and 12, 2874 nM. Lanes 1-4 represent G␤␥ subunits incubated on ice, lanes 5-8 are G␤␥ subunits treated with Lys-C at 30°C for 30 min in the presence of 10 g/ml leupeptin and lanes 9 -12 are G␤␥ subunits treated at 30°C with Lys-C for 30 min. B, quantitation by densitometry. Circle, G␤␥ subunits incubated on ice; square, G␤␥ subunits treated with Lys-C at 30°C for 30 min in the presence of 10 g/ml leupeptin to inhibit the protease; triangle, G␤␥ subunits treated at 30°C with Lys-C for 30 min. Data are representative of four experiments with single determinations. domain of ␤ARK1, to productively associate with G␣ o /G␣ i to allow those proteins to serve as pertussis toxin substrates, and to inhibit Type I adenylate cyclase activity. On the other hand, the cleaved complex only weakly stimulated ␤ARK1 phosphorylation of rhodopsin. Lys-C cleavage may not be unique in diminishing G␤␥ stimulation of rhodopsin phosphorylation by ␤ARK1. The effect of trypsin cleavage of G␤ has not been assayed.
The inability of N-terminal G␤ or G␥ peptides or immunoprecipitating antibodies to these peptides to affect G␤␥-dependent rhodopsin phosphorylation or binding to the PH domain suggests that a region of the G␤␥ other than the N terminus is involved in these interactions. This suggestion is consistent with the findings of Wang et al. (1994) that ␤ARK PH domain binding may be mediated through the C-terminal five WD40 motifs of G␤ rather than through G␥.
There may be multiple points of interaction between G␤␥, ␤ARK, and other molecules. Some regions may specify binding to target molecules while others may modulate the catalytic activity of ␤ARK in conjunction with the ligand-activated receptor substrate. The modulation of ␤ARK activity could also be occurring at the level of G␤␥ interaction with the receptor substrate. While the fragments of Lys-C-digested G␥ appear to remain non-covalently associated in solution as the G␤␥ dimer, either they are displaced in the receptor complex with G␤␥, or they fail to assume the appropriate relationship to the rest of the members of the complex to activate ␤ARK phosphorylation of the receptor. Nicking of the G␥ subunit by proteolysis with Lys-C indicates that the G␤␥ dimer provides active modulation rather than a passive kinase binding scaffold for the phosphorylation of G-protein-coupled receptor substrates by ␤ARK1. FIG. 7. Functional assay of truncated G␤␥ subunits. A, G␤␥ subunit-mediated ADP-ribosylation of G␣ o /G␣ i by pertussis toxin. The reaction was carried out as described under "Experimental Procedures." Results are expressed as average of duplicates Ϯ S.D. The experiment was repeated twice with similar results. Circle, G␤␥ subunits incubated on ice; square, G␤␥ subunits treated with Lys-C at 30°C for 30 min in the presence of 10 g/ml leupeptin to inhibit the protease; triangle, G␤␥ subunits treated at 30°C with Lys-C for 30 min. B, inhibition of forskolin-stimulated platelet type I adenylate cyclase by G␤␥ subunits. The effect of treated G␤␥ subunits on platelet membrane adenylate cyclase activity was determined as described under "Experimental Procedures." Circle, G␤␥ subunits incubated on ice; square, G␤␥ subunits treated with Lys-C at 30°C for 30 min in the presence of 10 g/ml leupeptin to inhibit the protease; triangle, G␤␥ subunits treated at 65°C for 10 min to inactivate the proteins. Data shown are the average of triplicates Ϯ S.D. Basal adenylate cyclase activity ϭ 200 pmol/ min/mg protein, forskolin-stimulated adenylate cyclase activity (100%) ϭ 2200 pmol/min/mg protein. The experiment was repeated twice.