Binding and Phosphorylation of Tubulin by G Protein-coupled Receptor Kinases*

Although the β-adrenergic receptor kinase (βARK) mediates agonist-dependent phosphorylation and desensitization of G protein-coupled receptors, recent studies suggest additional cellular functions. During our attempts to identify novel βARK interacting proteins, we found that the cytoskeletal protein tubulin could specifically bind to a βARK-coupled affinity column.In vitro analysis demonstrated that βARK and G protein-coupled receptor kinase-5 (GRK5) were able to stoichiometrically phosphorylate purified tubulin dimers with a preference for β-tubulin and, under certain conditions, the βIII-isotype. Examination of the GRK/tubulin binding characteristics revealed that tubulin dimers and assembled microtubules bind GRKs, whereas the catalytic domain of βARK contains the primary tubulin binding determinants. In vivo interaction of GRK and tubulin was suggested by the following: (i) co-purification of βARK with tubulin from brain tissue; (ii) co-immunoprecipitation of βARK and tubulin from COS-1 cells; and (iii) co-localization of βARK and GRK5 with microtubule structures in COS-1 cells. In addition, GRK-phosphorylated tubulin was found preferentially associated with the microtubule fraction during in vitro assembly assays suggesting potential functional significance. These results suggest a novel link between the cytoskeleton and GRKs that may be important for regulating GRK and/or tubulin function.

Tubulin, along with a variety of associated proteins, constitute microtubules, a major component of the cytoskeleton. Tubulin exists principally in two forms, as either cytosolic soluble tubulin heterodimers consisting of various ␣and ␤-tubulin isotypes or as insoluble assembled tubulin polymers (microtubules) (1). The ability of tubulin to cycle between these two states is one of its most fundamentally important features. In cells, microtubules are involved in such diverse and dynamic functions as maintaining cell shape, endocytosis, exocytosis, vesicle trafficking, cellular transport, and mitosis (1). These dynamic functions require the coordinated regulation of microtubule function and assembly, a process thought to reflect integration of various extracellular signals (2). Indeed, many studies have revealed functional relationships between tubulin and various cellular signaling molecules. For example, tyrosine kinases such as the insulin receptor and c-Src, as well as second messenger-responsive kinases such as the cAMP-dependent and Ca 2ϩ /calmodulin-dependent kinases, can regulate microtubule function and/or assembly through phosphorylation (Ref. 3 and references therein). Moreover, a series of recent studies have demonstrated direct binding interactions between G␣ and G␤␥ subunits and tubulin (4 -6) suggesting new modes of regulation for microtubule assembly. Other studies provide examples of tubulin or microtubules regulating either the activity or the localization of signaling molecules such as the A 1 adenosine (7) and ␥-aminobutyric acid (8) receptors, phospholipase C-␤1 (9), and Ki-Ras (10). Taken together, these studies provide support for a novel paradigm whereby information flow appears to be bi-directional between signaling and cytoskeletal molecules.
␤-Adrenergic receptor kinase (␤ARK) 1 is a member of a family of G protein-coupled receptor kinases (GRKs) that includes rhodopsin kinase (GRK1), ␤ARK (GRK2), ␤ARK2 (GRK3), GRK4, GRK5, and GRK6. GRKs phosphorylate the agonistactivated form of G protein-coupled receptors which in turn promotes the high affinity binding of a second family of proteins, called arrestins (11). This process functions to both uncouple the G protein from the receptor and to promote receptor endocytosis via clathrin-coated pits. ␤ARK is one of the best characterized GRKs and has been shown to be important for the phosphorylation and desensitization of a variety of receptors (11). Moreover, ␤ARK activity appears to be regulated through interactions with both free heterotrimeric G␤␥ subunits (12,13) and negatively charged membrane phospholipids (14 -16). These interactions are thought to be important largely for their ability to promote translocation of ␤ARK to the plasma membrane, facilitating interaction with receptor substrates. Recent studies have provided novel information regarding the function and cellular localization of ␤ARK. Specifically, it was shown that ␤ARK can traffic along with ␤ 2 -adrenergic receptors to the endosome following receptor activation (17). Mayor and co-workers (18) have also demonstrated an association of ␤ARK with microsomes that appears to be mediated via an unidentified ␤ARK-binding protein. Furthermore, ␤ARK has been implicated to function in developmental processes involved in cardiogenesis through currently undefined mechanisms (19). Given these and other areas of inquiry, it seems likely that additional cellular substrates and regulators of ␤ARK exist. Thus, in this study a biochemical approach was employed to explore this possibility. We describe here the identification of tubulin as a novel GRK substrate and present biochemical and cellular characterization of GRK/tubulin interaction.

EXPERIMENTAL PROCEDURES
Materials-Dodecylmaltoside, urea, and protein A-agarose were from Boehringer Mannheim, and [ 32 P]ATP was from NEN Life Science Products. Anti-␣-tubulin monoclonal antibody and enhanced chemiluminescence (ECL) reagents were from Amersham Pharmacia Biotech. Polyvinylidene difluoride membrane was purchased from Applied Biosystems. Nitrocellulose was from MSI, and LipofectAMINE TM and cell culture medium were from Life Technologies Inc. Horseradish peroxidase-conjugated goat anti-mouse and goat anti-rabbit antibodies and Affi-Prep 10 affinity chromatography matrix were from Bio-Rad. Horseradish peroxidase-conjugated horse anti-mouse antibody was from Vector Laboratories. Hemagglutinin (HA)-specific polyclonal antibody was from Babco. GRK5-specific polyclonal antibody was from Santa Cruz Biotechnology. Fluorescein isothiocyanate-conjugated anti-mouse antibody, tetramethylrhodamine-conjugated anti-rabbit antibody, and Slowfade TM mounting medium were from Molecular Probes Inc. pGEX-2T GST gene fusion vector was from Amersham Pharmacia Biotech. ATP, GTP, glutathione-agarose beads, paclitaxel (Taxol), crude soybean phosphatidylcholine, ␤-mercaptoethanol, iodoacetic acid, and all other chemicals were from Sigma.
Protein Expression and Purification-␤ARK and GRK5 were overexpressed in and purified from Sf9 insect cells (20,21). Purified rhodopsin kinase was generously provided by Drs. J. Pitcher and R. Lefkowitz, and purified G␤ 1 ␥ 2 was generously provided by Dr. S. P. Kennedy. Partially purified tubulin (PPT) containing microtubule-associated proteinswaspreparedfromporcinebrainbyonetothreecyclesoftemperaturedependent microtubule assembly/disassembly as described previously (6,22). Purified tubulin, prepared from bovine brain by assembly/ disassembly followed by phosphocellulose chromatography and 99% free of microtubule associated proteins (23), was from Cytoskeleton and microSuppliers.
Gel Electrophoresis and Immunoblotting-SDS-polyacrylamide gel electrophoresis (SDS-PAGE) was performed using standard methods (24). Following electrophoresis, proteins were electroblotted onto nitrocellulose. The membrane was then incubated in blocking buffer (10 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.02% Tween 20, and 5% non-fat dried milk) at 22°C for 1 h followed by probing with either an ␣-tubulin-, ␤ARK-, GRK5-, or arrestin-specific primary antibody diluted in blocking buffer. Membranes were subsequently probed with a 1:2000 dilution of horseradish peroxidase-conjugated secondary antibody in blocking buffer for 20 -60 min at 22°C. All blots were developed by ECL following the manufacturer's guidelines.
Synthesis of ␤ARK-coupled Affinity Resin-50 ml of Affi-Prep 10 affinity chromatography matrix (20% v/v) was incubated with or without ␤ARK at a final concentration of 0.3 mg/ml in 20 mM HEPES, pH 7.2, 5 mM EDTA, 0.02% Triton X-100 for 5 h for 4°C. The coupling reaction was stopped by addition of 5 ml of 1 M Tris-HCl, pH 7.5, and incubation overnight at 4°C. A trace amount of [ 35 S]methionine-labeled ␤ARK was included in the initial coupling reaction to enable a determination of coupling efficiency (ϳ1 mg ␤ARK/ml resin). The resin was then washed extensively with 20 mM HEPES, pH 7.2, 200 mM NaCl, 5 mM EDTA, 0.02% Triton X-100 and stored at 4°C.
Identification of ␤ARK-binding Proteins-Fresh bovine calf brain was stripped of connective tissue and minced in 200 ml of homogenization buffer (20 mM Tris-HCl, pH 7.5, 5 mM EDTA, 100 mM NaCl, 5 mM benzamidine, 5 mM phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, 10 g/ml leupeptin, 0.02% Triton X-100) using a Brinkman Polytron (14,000 rpm, 30 s). The homogenate was centrifuged at 45,000 ϫ g for 20 min and the resulting supernatant at 300,000 ϫ g for 60 min. The final supernatant was aliquoted and stored at Ϫ70°C until use. One ml of ␤ARK-coupled affinity resin (20% v/v) or control resin was incubated with 10 ml of the soluble brain extract (or buffer) for 2 h at 4°C. The incubation mixture was centrifuged at 1000 ϫ g for 1 min, and the pellet was washed four times with buffer (20 mM Tris-HCl, pH 7.5, 5 mM EDTA, 100 mM NaCl, 0.02% Triton X-100). Bound proteins were released from the matrix by addition of 50 l of SDS sample buffer and 100 l of water to the final pellet followed by boiling for 10 min. The samples were then subjected to 7.5% SDS-PAGE, and the separated proteins were transferred to a polyvinylidene difluoride membrane. A specific 55-kDa protein band was identified by Ponceau S staining, excised, and then submitted to the Harvard Microchemistry Facility for sequence analysis.
Identification of Tubulin Isotypes Phosphorylated-Phosphorylation reactions were performed as above using either 0.025 or 2.5 M ␤ARK or GRK5 and 4.5 M tubulin dimers in a total volume of 20 l at 30°C for 15 min. Phosphorylated tubulin was then reduced and carboxymethylated essentially as described previously (25). Briefly, 0.5 M Tris-HCl, pH 8.6, 5 mM EDTA, 8 M urea, and 120 mM ␤-mercaptoethanol was added to each phosphorylation reaction in a final volume of 100 l and incubated under nitrogen at 22°C for 2 h. 250 mg/ml iodoacetic acid, dissolved in 1 N NaOH, was then diluted 1:10 into the reduced tubulin solutions and incubated at 22°C in the dark for 30 min. Reactions were stopped with SDS sample buffer, boiled, and subjected to 7.5% SDS-PAGE, Coomassie Blue staining, and autoradiography.
GST-␤ARK Fusion Binding Assay-Purified GST-␤ARK fusion proteins containing either the N-terminal domain (residues 1-184), the central catalytic domain (residues 185-467), or the C-terminal domain (residues 466 -689) of ␤ARK were generated essentially as described previously (26). Ten g of GST or GST-␤ARK fusion proteins immobilized on glutathione-agarose beads were incubated with 0.4 g of soluble tubulin dimers in 100 l of binding buffer (20 mM Tris-HCl, pH 7.5, 5 mM EDTA, 100 mM NaCl, and 0.02% Triton X-100) at 30°C for 60 min. Reactions were then chilled on ice for 5 min followed by precipitation in a microcentrifuge for 10 s. The pellet was washed three times with 400 l of binding buffer and then boiled with SDS sample buffer. Samples were subjected to 10% SDS-PAGE and Western blotting using a monoclonal ␣-tubulin-specific antibody.
Binding and Phosphorylation of Taxol-stable Microtubules by GRKs-Tubulin assembly reactions contained, in a total volume of 50 l, 60 g of tubulin (or control buffer), 1 mM GTP, 30% glycerol, and 10 M Taxol in 100 mM PIPES, pH 6.9, 1 mM EGTA, 1 mM MgCl 2 . Reactions were incubated at 37°C for 30 min followed by pelleting at 150,000 ϫ g for 10 min at 37°C. The supernatant was removed and the Taxol-stable microtubule pellets were gently resuspended in 50 l of reaction buffer (100 mM PIPES, pH 6.9, 1 mM EGTA, 7.5 mM MgCl 2 , 100 M [␥-32 P]ATP (5 cpm/fmol), 1 M Taxol) in the absence or presence of 5 g of ␤ARK or GRK5 and incubated for 30 min at 30°C. Tubulin was then repelleted, and both the supernatant and pellet fractions were boiled with SDS sample buffer and subjected to 10% SDS-PAGE, Coomassie Blue staining, and densitometry. After autoradiography, the 32 P-labeled tubulin bands were excised and counted to determine the picomoles of phosphate transferred.
Cell Culture and Transfection-COS-1 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 100 units/ml penicillin G, and 100 g/ml streptomycin sulfate at 37°C in a humidified atmosphere containing 5% CO 2 . Cells grown to 75-95% confluence were transfected with 10 g of either pcDNA3, pcDNA3-␤ARK, pBC12BI, or pBC-GRK5 DNA using 65 l of Lipo-fectAMINE TM per T75 flask according to the manufacturer's instructions. For immunoprecipitation experiments, cells were harvested 60 h after transfection. For immunofluorescence experiments, cells were trypsinized and plated onto 12-mm glass coverslips in a 24-well dish 2 h after transfection and analyzed at 24 -36 h after transfection.
Immunoprecipitation-COS-1 cells were transfected either with pcDNA3 or pcDNA3-␤ARK as described above. At 60 h after transfection, T75 flasks were rinsed with PBS, and the cells were scraped into 1 ml of extraction buffer (10 mM Tris-HCl, pH 7.4, 1 mM EDTA, 5 mM dithiothreitol, 100 mM NaCl, 0.02% Triton X-100, 1 mM phenylmethylsulfonyl fluoride, 10 g/ml leupeptin, 10 g/ml pepstatin A, and 10 g/ml aprotinin) and lysed by freeze-thaw at Ϫ70°C. Lysates were then centrifuged at 55,000 ϫ g for 1 h at 4°C, and the supernatant was stored at Ϫ70°C. For immunoprecipitation, 100 l of lysate was incubated with either a ␤ARKor HA-specific polyclonal antibody or in the absence of antibody for 30 min at 4°C followed by addition of 50 l of 50% protein A-agarose pre-equilibrated in extraction buffer and an additional 60 min incubation at 4°C. Samples were then centrifuged at 2000 rpm for 30 s in a microcentrifuge. The pellet was washed three times with extraction buffer for 30 min at 4°C and then eluted in 50 l of SDS sample buffer. Samples were subjected to 10% SDS-PAGE and Western blot analysis using either ␤ARK-specific or ␣-tubulin-specific monoclonal antibodies.
Immunofluoresence Microscopy-Transfected COS-1 cells grown on glass coverslips were rinsed twice with PBS at 37°C and fixed with methanol at Ϫ20°C for 10 min. Alternatively, some cells were preincubated with Dulbecco's modified Eagle's medium containing 10 M (Ϫ)-isoproterenol for either 20 or 60 min at 37°C prior to fixing. Fixed cells were subsequently rinsed three times with phosphate-buffered saline (PBS) and permeabilized with PBS containing 0.2% Triton X-100 for 10 min at 22°C. These cells were blocked for 30 min at 37°C with PBS containing 0.05% Triton X-100 (PBS/Triton) and 5% non-fat dried milk and subsequently incubated with an ␣-tubulin-specific mouse monoclonal antibody and either a ␤ARK-, GRK5-, or arrestin-3-specific rabbit polyclonal antibody diluted in the same buffer for 1 h at 37°C. Cells were rinsed three times with PBS/Triton and then incubated at 37°C for 30 min. The cells were then incubated with a tetramethylrhodamine-conjugated goat anti-mouse antibody and a fluorescein isothiocyanate-conjugated goat anti-rabbit antibody both diluted 1:100 in PBS/ Triton and 5% non-fat dried milk for 1 h at 37°C. Cells were rinsed five times with PBS/Triton and then incubated at 37°C for 30 min followed by fixing with methanol at Ϫ20°C for 5 min. Cells were rinsed with PBS and mounted on a slide with Slowfade TM mounting medium. Fluorescence microscopy was performed on a Bio-Rad MRC600 laser scanning confocal microscope (Hemmelholsteadt, UK) attached to a Zeiss Axiovert 100 microscope, using a Zeiss Plan-Apo 63 ϫ 1.40 NA oil immersion objective at a zoom factor of 1.4.
Assembly of GRK-phosphorylated Tubulin Dimers-Purified tubulin dimers were subjected to a 10 min pre-spin at 150,000 ϫ g at 4°C to precipitate any existing tubulin aggregates. The supernatant was removed and utilized in the subsequent phosphorylation assay. Phosphorylation reactions (40 l total) were incubated at 30°C for 45 min and contained 60 g of tubulin dimers, 200 M GTP, 100 mM PIPES, pH 6.9, 1 mM EGTA, and 3 mM MgCl 2 , in the absence or presence of 100 M [␥-32 P]ATP (5 cpm/fmol) and either 4 g of ␤ARK or GRK5. Assembly of phosphorylated or control tubulin was initiated by the addition of glycerol (5-30% final concentration) and raising the temperature to 37°C for 5-60 min. Assembly was quantitated by pelleting the microtubules at 150,000 ϫ g for 10 min at 37°C. The supernatant and pellet fractions were then subjected to 10% SDS-PAGE, Coomassie Blue staining, and densitometry. Distribution of the 32 P-labeled tubulin was assessed by autoradiography, followed by excision and counting of the radiolabeled bands.

RESULTS AND DISCUSSION
Identification of Tubulin as a ␤ARK-interacting Protein-In an effort to identify novel GRK-interacting proteins, a crude bovine brain cytosolic extract was chromatographed over a ␤ARK affinity column. The column was washed extensively, and bound proteins were eluted with boiling SDS sample buffer and subsequently subjected to SDS-PAGE and Coomassie Blue staining. This strategy identified a total of six proteins that specifically bound to the ␤ARK column (Fig. 1). The predomi-nant interacting protein migrated at ϳ55 kDa, and additional proteins of ϳ50, ϳ46, ϳ44, ϳ38, and ϳ35 kDa were also observed. Since ␤ARK is known to bind G␤␥ subunits, we initially probed this blot using a ␤ common antibody. This revealed that the ϳ35-kDa protein was a G␤ subunit (data not shown).
Importantly, tubulin is known for its ability to cycle between soluble ␣␤ dimers and insoluble assembled microtubules (1), and it is of interest to consider which form is responsible for the observed ␤ARK binding. Although this was not assessed directly, the temperature and buffering conditions used to generate the brain extract, as well as those used in the binding experiment, inhibit tubulin assembly. Thus, it is likely that the tubulin was principally in the soluble form implying that ␤ARK can bind soluble tubulin dimers.
Phosphorylation of Tubulin by GRKs-The ␤ARK/tubulin interaction was confirmed by the ability of purified ␤ARK to rapidly phosphorylate highly purified bovine brain soluble tubulin dimers to a stoichiometry of ϳ1.0 mol of P i /mol of tubulin dimer (Fig. 2). Addition of ␤ARK activators such as G␤ 1 ␥ 2 or phosphatidylcholine-dodecylmaltoside mixed micelles increased the stoichiometry to ϳ1.5 and ϳ1.6 mol/mol, respectively, or to ϳ2.0 mol/mol when added together. Addition of 10 mM dodecylmaltoside alone, however, was without effect (data not shown).
We also investigated the ability of tubulin to serve as a substrate for other members of the GRK family (Fig. 2). Rhodopsin kinase weakly phosphorylated tubulin (stoichiometry of ϳ0.45 mol of P i /mol of tubulin dimer), whereas GRK5 rapidly phosphorylated tubulin to a stoichiometry of ϳ1.6 mol/mol. Interestingly, whereas both GRK5 and ␤ARK appear capable of phosphorylating at least two sites on the tubulin dimer, there are apparent differences in the nature of this phosphorylation as demonstrated by a lower mobility phosphorylated species generated in the presence of GRK5 that is not observed in the presence of ␤ARK ( Fig. 2A). This difference suggests specificity in the molecular nature of the GRK/tubulin interaction and allows the potential for functional differences to be attributed to each of these phosphorylation reactions.
In order to identify which tubulin isotypes (␣ or ␤) are substrates for GRKs, tubulin was initially phosphorylated by either ␤ARK or GRK5 and then carboxymethylated. Tubulin ␣and ␤-isotypes are differentially susceptible to carboxymethylation allowing for their electrophoretic separation (27). Moreover, among the various ␤-isotypes, ␤III has a distinct susceptibility to carboxymethylation that resolves it from both the ␣and other ␤-isotypes (␤*) (27), as seen in the Coomassie Bluestained lanes in Fig. 3. Phosphorylation of tubulin by 0.025 M kinase for 15 min resulted in a preferential phosphorylation of One ml of ␤ARK-coupled affinity resin (20% v/v) (␤ARK) or control resin (control) was incubated with 10 ml of a soluble bovine brain extract (extract) or buffer for 2 h at 4°C. The incubation mixture was centrifuged at 1000 ϫ g for 1 min, and the pellet was washed extensively as described under "Experimental Procedures." Final pellets were resuspended in SDS sample buffer, electrophoresed on a SDS-polyacrylamide gel, and stained with Coomassie Blue. Known proteins are indicated on the right and molecular mass markers (Std) on the left. ␤-tubulin isotypes by both ␤ARK and GRK5 (Fig. 3). Interestingly, at higher GRK levels (2.5 M kinase), a qualitative difference in the phosphorylation selectivity was observed with ␤ARK and particularly GRK5, both exhibiting a preference for the ␤III-isotype (even though ␤III represents only ϳ25% of the total ␤-tubulin (28)). Although we have not more thoroughly investigated the molecular basis of this apparent switch in isotype specificity, it is possible that the formation of GRKtubulin hetero-oligomers at higher GRK concentrations contributes to this process. Total cellular concentrations of ␤ARK and GRK5 were estimated to be in the range of 0.01-0.2 M based on Western analysis of several cell lines (data not shown). Thus, while 2.5 M kinase may not seem physiological it is likely that GRKs can exist in much higher local concentrations in various cellular micro-domains (e.g. at the plasma membrane/cytoskeleton interface), suggesting that both concentrations of GRK used in this experiment may be physiologically relevant.
The above data permit a comparison of our findings with previously characterized tubulin phosphorylation studies. Most striking is the fact that the brain-specific ␤III-tubulin isotype appears to be the principal tubulin isotype phosphorylated in brain tissue in vivo (27,29). Additional studies have revealed that Ser 444 of ␤III-tubulin is a critical phosphorylation site (29) and that casein kinase II may be partly responsible for this in vivo phosphorylation (29 -31). Interestingly, casein kinase II and ␤ARK have previously been shown to preferentially phosphorylate substrates containing acidic residues N-terminal to the phosphorylation site (32). Moreover, these kinases are also both effectively inhibited by micromolar concentrations of heparin, which was previously used to demonstrate that a microtubule-associated kinase activity was likely casein kinase II (20,30,31). Thus, GRK-mediated tubulin phosphorylation appears to be consistent with several aspects of the physiological phosphorylation of brain tubulin previously characterized (27, 29 -32).
It is important to note that our studies have been limited to analysis of brain tubulin. Although it is clear that the neuronal-specific ␤III-isotype is the preferred physiological phosphorylation substrate in brain, relatively few studies have characterized the tubulin isotypes phosphorylated in non-neuronal tissues. Since GRKs are expressed ubiquitously, it is of interest to consider which tubulin isotypes may be phosphorylated in other tissues. Of specific interest is the cardiac myocyte, where the levels of some ␤-tubulin isoforms have been shown to increase significantly in congestive heart failure (33)(34)(35). The increased levels of microtubules are thought to be, at least in part, a consequence of increased microtubule stability (34,35) and might be related to the extent of post-translational modification such as phosphorylation. Interestingly, both ␤ARK and GRK5 levels have also been shown to increase (ϳ2-3-fold) in the failing heart and are thought to contribute to the pathology of this condition (36,37). Thus, future study of the specific ␤-tubulin isotypes phosphorylated by GRKs, as well as the possible functional consequences, in non-neuronal cells such as cardiac myocytes is warranted.
In order to characterize the GRK-mediated phosphorylation better, kinetic analysis was performed by varying the tubulin concentration in the presence of the various GRKs (Table I).  Phosphorylation reactions were performed at 30°C for 5 min, a time within the linear region of phosphorylation time courses for all kinases examined (Fig. 2). These studies revealed a K m of 0.42 M and a V max of 17.4 nmol of P i /min/mg for the ␤ARKmediated tubulin phosphorylation reaction. Although addition of G␤ 1 ␥ 2 had little effect on the maximum velocity of this reaction (V max ϭ 15.0 nmol of P i /min/mg), the apparent affinity of ␤ARK for tubulin was increased nearly 2-fold (K m ϭ 0.23 M). Conversely, the primary effect of the phospholipid addition was on the catalytic activity of ␤ARK yielding a V max of 25.8 nmol P i /min/mg. Concomitant addition of both G␤ 1 ␥ 2 and phospholipid was additive yielding a combined enhancement of both K m and V max . These two regulators of ␤ARK activity have previously been suggested to elicit their effects largely by facilitating membrane translocation of ␤ARK, thereby achieving closer proximity to membrane-bound receptor substrates (12)(13)(14)(15). Importantly, the effects of G␤ 1 ␥ 2 and phospholipid on the kinetic parameters of tubulin phosphorylation are apparent in this soluble system underscoring their potential role in the direct activation of ␤ARK. Interestingly, however, membranebound forms of tubulin have been observed in a variety of systems (10, 38 -40), and tubulin palmitoylation has recently been characterized in detail (41). Thus, it is likely that the properties of membrane translocation also are important aspects of the ␤ARK/tubulin interaction in cells. For example G␤␥ and phospholipid may direct the specific phosphorylation of the palmitoylated membrane-bound forms of tubulin, in a fashion analogous to that of G protein-coupled receptors. Alternatively, palmitoylated tubulin itself may be capable of supporting ␤ARK membrane translocation thus regulating the localization and/or function of ␤ARK.
The kinetic analysis of GRK5 revealed a relatively modest catalytic activity toward tubulin with a V max of 8.7 nmol of P i /min/mg, approximately one-third that of activated ␤ARK. However, the affinity of GRK5 for tubulin was observed to be ϳ4.5-fold greater (K m ϭ 0.064 M) than that of activated ␤ARK. These distinctions in kinetic properties support the observed differences in ␤ARKand GRK5-mediated tubulin phosphorylation (Fig. 2), although the significance of these observations remains unclear. Rhodopsin kinase phosphorylated tubulin with a K m of 4 M and a V max of 5.3 nmol of P i /min/mg making it a relatively inefficient kinase for tubulin. Importantly, however, the affinities exhibited by ␤ARK, GRK5, and rhodopsin kinase are all within physiological ranges. For example, the affinities of ␤ARK and GRK5 for an in vivo substrate, the ␤ 2 -adrenergic receptor, have previously been assessed to be 0.25 (42) and 0.54 M (21), respectively. Moreover, the total cellular concentration of tubulin in fibroblasts has been estimated to be ϳ20 M (1). Finally, measured affinities for several well characterized tubulin kinases have revealed K m values for tubulin of 20 M for casein kinase II (30) and 11.3 M for Ca 2ϩ /calmodulin-dependent kinase (43). Taken together, these data suggest that tubulin may serve as a physiological substrate for GRK-mediated phosphorylation.
In Vitro Interaction of GRKs with Soluble and Assembled Tubulin-Since the GRK/tubulin association was discovered via their direct binding (Fig. 1), we were interested in further characterizing this aspect of the interaction. Thus, we assessed the binding of soluble tubulin dimers to GST-␤ARK fusion proteins containing either the N-terminal, central catalytic, or C-terminal domains of ␤ARK. When these GST-␤ARK fusions were incubated with purified soluble tubulin dimers and subsequently precipitated with glutathione-agarose beads, it was found that the central catalytic domain of ␤ARK (residues 185-467) contained the major binding determinant for tubulin (Fig. 4). These data are consistent with the ability of multiple GRKs to phosphorylate tubulin as the catalytic domain is the most highly conserved region among the GRKs (11). Whereas the N-terminal domain possessed a small amount of specific binding, the C-terminal GST-␤ARK fusion which includes the PH domain and G␤␥ binding determinants demonstrated no tubulin binding capacity. This raises the interesting possibility that ␤ARK, which appears to have distinct regions for interaction with G␤␥ (the C-terminal domain) and tubulin (the catalytic domain), may be able to concomitantly associate with both tubulin and G␤␥ to form a ␤ARK⅐G␤␥⅐tubulin ternary complex. This is consistent with our observation that the apparent affinity of ␤ARK for tubulin is increased ϳ2-fold by the presence of G␤␥ (Table I) and is further supported by the recent demonstration that G␤␥ can directly associate with microtubules (6). These observations suggest potentially interesting implications for ␤ARK regulation.
Although the data presented in Figs. 1-4 demonstrate that GRKs can bind and phosphorylate soluble tubulin, these data do not test whether assembled microtubules may also serve as substrates for GRK-mediated phosphorylation. In order to directly address this question, stable microtubules were formed by assembly in the presence of Taxol and then tested for their ability to specifically bind and be phosphorylated by ␤ARK or GRK5. Taxol-stable microtubules were sedimented, and the pellet was resuspended in buffer containing Taxol, [␥-32 P]ATP, and either ␤ARK or GRK5. After incubation at 30°C for 30 min, the microtubules were repelleted and examined by SDS-PAGE and Coomassie Blue staining and densitometry. The stable microtubules were found to specifically sediment a significant portion of ␤ARK (54 Ϯ 3%; n ϭ 8) and GRK5 (69 Ϯ 5%; n ϭ 4). When subjected to autoradiography it was found that the stable microtubules also served as a substrate for ␤ARKmediated (0.13 Ϯ 0.01 mol P i /mol tubulin dimer; n ϭ 3) and GRK5-mediated (0.19 Ϯ 0.01 mol P i /tubulin dimer; n ϭ 3) phosphorylation. Thus, similar to previous findings with casein kinase II (30), GRKs can directly bind and phosphorylate preformed Taxol-stabilized microtubules. Whereas the form of tubulin (soluble dimers or assembled microtubules) that is the preferred substrate for GRKs remains unclear, it is noteworthy that under similar conditions, the phosphorylation stoichiometries of assembled (above) and unassembled (␤ARK ϭ 0.14 Ϯ 0.02 mol P i /mol tubulin dimer and GRK5 ϭ 0.10 Ϯ 0.01 mol P i /mol tubulin dimer) tubulin are comparable.
In Vivo Interaction of GRKs with Tubulin-To address the potential physiological relevance of the GRK/tubulin interaction, we assessed the in vivo association of GRKs and tubulin using three distinct approaches. First, a partially purified tubulin preparation (PPT) was generated from porcine brain tissue by two successive cycles of tubulin assembly (at 37°C) and FIG. 4. Binding of tubulin dimers to GST-␤ARK fusion proteins. 10 g of GST-␤ARK fusion proteins representing either the N-terminal (residues 1-184), the central catalytic (residues 185-467), or the C-terminal (residues 466 -689) domain of ␤ARK were incubated with 0.4 g of soluble tubulin dimers, followed by pelleting and washing as described under "Experimental Procedures." Samples were subjected to SDS-PAGE and Western analysis using a monoclonal tubulin antibody. The gel shown is representative of four separate experiments. Bands, in duplicate, represent tubulin precipitated by each of the GST-␤ARK fusion proteins. Forty and 160 ng of purified tubulin, representing 10 and 40% of the total amount loaded in each experiment, are shown on the right. disassembly (at 4°C) each followed by centrifugation. In this way tubulin, as well as microtubule-associated proteins, are significantly enriched within one or two cycles (22). This enrichment was quantitated by subjecting varied amounts of PPT to SDS-PAGE, Coomassie Blue staining, and densitometry revealing that tubulin represents ϳ93% of the protein in this preparation, and the remaining ϳ7% is composed of microtubule-associated proteins and other associated proteins (data not shown). We then tested various amounts of PPT for the presence of ␤ARK by Western blotting using a ␤ARK-specific antibody. This analysis revealed a significant amount of ␤ARK present in the PPT preparation (0.05-0.1% of the total protein, Fig. 5A). A similar analysis of the initial soluble porcine brain extract revealed that ␤ARK is present at ϳ0.02% of the total protein (data not shown) suggesting a 2.5-5-fold enrichment of ␤ARK in PPT. Analysis of equal amounts of the warm (37°C) and cold (4°C) supernatant and pellet fractions revealed that ␤ARK is associated to a similar extent with the "microtubule fractions" (i.e. the warm pellet and cold supernatant) and the "non-microtubule fractions" (i.e. warm supernatant and cold pellet). Given the above data that suggest ␤ARK may bind to both soluble and assembled forms of tubulin, it is not surprising that ␤ARK does not strictly associate with the microtubule fraction. Even so, a significant amount of ␤ARK is present in the 2-cycle PPT (Fig. 5A) as well as in a 3-cycle PPT (data not shown). Thus, the presence of ␤ARK in these tubulin preparations is suggestive of an in vivo association between ␤ARK and tubulin that by necessity includes association with microtubules, as ␤ARK does not precipitate on its own under these conditions (data not shown). In order to demonstrate the specific nature of the ␤ARK/tubulin association, the arrestins present in the brain extract and PPT were also quantitated. While arrestins were found to be expressed at ϳ10-fold higher levels than ␤ARK (ϳ0.2% of the protein in the pig brain extract), no arrestin was detected in 40 g of PPT (data not shown).
A second approach to examine the GRK/tubulin interaction in vivo involved the immunoprecipitation of either endogenous or overexpressed ␤ARK from COS-1 cells using a ␤ARK-specific polyclonal antibody. COS-1 cells were transfected either with pcDNA3 or pcDNA3-␤ARK DNA using LipofectAMINE TM . Western analysis revealed detectable levels of endogenous tubulin and ␤ARK, as well as a robust overexpression of ␤ARK in cells transfected with pcDNA3-␤ARK (Fig. 5C). Immunoprecipitation of ␤ARK from cell lysates using the ␤ARK-specific polyclonal antibody was detectable in both cell lysates and was proportional to the respective ␤ARK expression levels (data not shown). When the ␤ARK immunoprecipitates were blotted and probed with an ␣-tubulin-specific antibody, a significant coprecipitation of tubulin was detected in lysates from both the pcDNA3-and pcDNA3-␤ARK-transfected cells (Fig. 5B), clearly demonstrating the in vivo association of ␤ARK and tubulin. In contrast, no tubulin was found in control immunoprecipitations performed either in the absence of the ␤ARK antibody or with an HA-specific polyclonal antibody.
We also assessed GRK/tubulin interaction by immunofluorescence microscopy of COS-1 cells overexpressing either ␤ARK or GRK5. Transfected cells were plated on glass coverslips, fixed, permeabilized, and then incubated with tubulinand ␤ARKor GRK5-specific primary antibodies followed by fluorescently labeled secondary antibodies. This analysis revealed that GRKs are found associated with the microtubule cytoskeleton (Fig. 6). Specifically, ␤ARK is found enriched in a perinuclear region of the cells coincident with the microtubule organizing center (top panels) as well as in other regions containing dense microtubule structures (middle and bottom panels), a pattern comparable with that of casein kinase II microtubule co-localization in NRK cells (31). A generally similar co-localization of GRK5 with microtubules was also observed, with the exception that GRK5 was not significantly localized to the perinuclear region (data not shown). Although co-localization is apparent at areas of high microtubule content, GRKs are not strictly associated with microtubules. This observation may, in part, be due to the relatively high level of GRK expression in these cells or the apparent ability of GRKs to interact with both soluble and assembled forms of tubulin. In addition, the effect of endogenous microtubule-associated proteins on GRK/microtubule interactions is presently unknown. In any case, it appears that at least a portion of the GRKs are specifically associated with distinct microtubule structures in cells. Taken together, the co-purification (Fig. 5A) and co-immunoprecipitation (Fig. 5B) of ␤ARK and tubulin, as well as the co-localization of ␤ARK and GRK5 with microtubules ( Fig. 6 and data not shown), demonstrates that GRK/tubulin interactions occur in vivo.
Analysis of the Functional Role of GRK/Tubulin Interaction-While providing important in vitro and in vivo evidence FIG. 5. In vivo interaction between ␤ARK and tubulin. A, a crude porcine brain extract was enriched for tubulin by two rounds of assembly and disassembly as detailed under "Experimental Procedures." Aliquots of 10, 20, and 30 g of PPT were subjected to SDS-PAGE and Western analysis using a ␤ARK-specific monoclonal antibody. Standards of 2, 5, and 10 ng of purified ␤ARK are shown on the left. B, COS-1 cells, transfected with pcDNA3 or pcDNA3-␤ARK, were harvested after 60 h and lysed. Lysates were immunoprecipitated without antibody (none), with a ␤ARK-specific polyclonal antibody (␤ARK), or an HA-specific polyclonal antibody (HA) as described under "Experimental Procedures." Samples were subjected to SDS-PAGE and Western analysis using an ␣-tubulin-specific mouse monoclonal antibody. Ten ng of purified tubulin dimer are shown on the right (Std). C, the ␤ARK and tubulin content in the cell lysates were assessed by Western analysis using ␤ARKand tubulin-specific monoclonal antibodies. Twenty five ng of purified ␤ARK (Std) is shown on the left, and 10 ng of purified tubulin (Std) is shown on the right.
for GRK/tubulin association, the above experiments fail to define any functional role for this interaction. Formally, three principal possibilities exist with respect to the functional significance of this interaction: (i) GRKs might regulate the assembly of tubulin; (ii) GRKs might regulate one or more of the many diverse functional properties of microtubules (i.e. independently of effects on tubulin assembly); or (iii) tubulin and/or microtubules may effect GRK function by regulating its cellular localization. Importantly, none of these possibilities are mutually exclusive, allowing the potential for complex functional relationships in cells.
Whereas the cellular functions of tubulin are diverse and complex, the ability of tubulin to dynamically cycle between soluble and assembled states is a fundamental property readily amenable to in vitro analysis (1,2,22,23). Indeed, tubulin assembly is known to be subject to a wide variety of regulatory post-translational modifications including phosphorylation (3). Therefore, we initially focused on tubulin assembly as the most experimentally tractable and likely property of tubulin to be influenced by GRK-mediated phosphorylation. In this regard, we attempted to examine whether ␤ARKor GRK5-mediated tubulin phosphorylation could affect the properties of tubulin assembly in vitro. Unfortunately, since the assembly assay is highly sensitive to buffering conditions and requires high tubulin concentrations, suboptimal phosphorylation conditions were required resulting in phosphorylation stoichiometries of only 0.1-0.25 mol of P i /mol tubulin dimer with either ␤ARK or GRK5. Thus, as might be expected, we did not observe any effect of this phosphorylation on the rate or extent of tubulin assembly. However, although we observed no obvious effects of GRK-mediated tubulin phosphorylation on assembly, the above experiments did reveal one striking and extremely reproducible observation. When we examined the relative distribution of the GRK-phosphorylated tubulin species, we found a disproportionate amount of the radiolabeled tubulin associated with the pellet fraction consistently in over 70 separate reactions under a variety of conditions. This observation is illustrated in Fig. 7 using tubulin phosphorylated by either ␤ARK FIG. 6. Co-localization of ␤ARK with microtubules. COS-1 cells were transfected with pcDNA3-␤ARK and fixed and permeabilized at 24 -36 h after transfection. Cells were probed with both ␤ARKand tubulin-specific primary antibodies followed by fluorescently labeled secondary antibodies as described under "Experimental Procedures." ␤ARK was labeled with fluorescein isothiocyanateconjugated anti-rabbit antibody and is shown in the left panels (␤ARK), and tubulin was labeled with tetramethylrhodamine-conjugated anti-mouse antibody and is shown in the center panels (MT). Tubulin and ␤ARK immunofluorescence are overlaid in the panels on the right (␤ARK/MT). Yellow represents co-localization of tubulin and ␤ARK. or GRK5 to average stoichiometries of 0.14 Ϯ 0.02 and 0.10 Ϯ 0.01 mol of P i /mol of tubulin dimer, respectively. When assembly was promoted by 30% glycerol for 30 min, ϳ90% of the total tubulin and ϳ95% of the phosphorylated tubulin were found associated with the pellet fraction. However, when assembly was promoted by 5% glycerol, only ϳ20% of the total tubulin was pelleted, whereas ϳ90% of the ␤ARKor GRK5-phosphorylated tubulin was pelleted.
These differences in the extent of total versus phosphorylated tubulin pelleted in the presence of 5% glycerol might be explained by one of two possibilities. First, the observed differences could be generated by a preferential phosphorylation of pre-formed microtubules. Importantly, the tubulin preparation is subjected to centrifugation prior to the phosphorylation reaction to remove any insoluble tubulin aggregates, ensuring that the tubulin added to the phosphorylation assay is at least initially in a soluble form. However, the phosphorylation assay is by design compatible with, although not optimal for, tubulin assembly. Thus, it is likely that during the phosphorylation reaction a small fraction of the tubulin assembles into microtubules. Indeed, centrifugation immediately after the phosphorylation step generally pellets 5-10% of the tubulin. In addition, the glycerol and GTP added to initiate assembly do not inhibit GRK activity. Thus, phosphorylation can continue as microtubules form, and it is conceivable that microtubules become preferentially phosphorylated by GRKs leading to the observed increase in radiolabeled tubulin in the pellet. A second possibility is that the phosphorylated species could prefer-entially be driven into the assembled form thus causing a disproportionate increase in phosphorylated tubulin in the pellet. This would suggest a direct functional role of GRK-mediated phosphorylation in promoting tubulin assembly. This possibility is consistent with the finding that under similar phophorylation conditions ␤III-tubulin is preferentially phosphorylated by both ␤ARK and GRK5 (Fig. 3). This is of interest as phosphorylation of ␤III has previously been shown to promote the assembly of tubulin and the stability of microtubules (27,31). Although we have currently been unsuccessful in further discriminating between these two possibilities, this effect is highly reproducible and likely represents an important aspect of the GRK/tubulin interaction.
An additional consideration in our in vitro functional analysis was that additional components not present in the assembly assay may be critical. Therefore, we also examined the effects of overexpression of GRKs on microtubule stability in COS-1 cells by immunofluorescence confocal microscopy. These experiments did not reveal any major GRK-dependent alterations in microtubule stability. Similarly, co-expression of the ␤ 2 -adrenergic receptor and ␤ARK or GRK5 in COS-1 cells also had no effect on the microtubule cytoskeleton even after a 20-or 60min activation with ␤-agonist (data not shown). The possibility remains that potential changes are confined to distinct microdomains in the cell, perhaps involving regions of the plasma membrane/cytoskeleton interface, and therefore go undetected by this relatively insensitive method of analysis. Additionally, we currently have no appreciation of how microtubule-associated proteins, important physiological regulators of tubulin assembly, may effect the GRK/tubulin interaction. Whereas the data presented in Fig. 7 suggests a potential role for GRKs in regulating tubulin assembly, more sophisticated analysis will be required to establish the precise role of GRKs in this process.
As stated above, a second possible result of the GRK/tubulin interaction might be the disruption or enhancement of some aspect of microtubule function. For example, phosphorylation has been demonstrated to regulate the ability of tubulin to interact with membranes (40), the ability of microtubules to cross-link with actin microfilaments (44), and the ability of microtubule motor proteins to function in transport (45). Thus, ␤ARK-mediated phosphorylation of microtubules could conceivably regulate any one of the many diverse microtubule functions in cells. However, identification and characterization of such a regulatory function would be limited by the diversity and complexity of the various possibilities.
Finally, a third possible result of the GRK/tubulin interaction is the tubulin-or microtubule-mediated regulation of the cellular localization and/or function of GRKs. It has been demonstrated that tubulin and microtubules are capable of influencing cellular signaling through direct interaction with a variety of signaling molecules (4 -9). The data presented in this article demonstrate the binding of GRKs to both soluble tubulin dimers and to assembled microtubules in vitro. Importantly, our immunofluorescence data reveal the association of ␤ARK and GRK5 with microtubules in cells suggesting that GRKs may be sequestered by the microtubule cytoskeleton, perhaps directing currently unappreciated GRK interactions. Such regulation has recently been demonstrated for the A 1 adenosine receptor (7) and Ki-Ras (10). Additionally, tubulin is known to become palmitoylated and to associate with a variety of cellular membranes including the plasma membrane, endosomes, microsomes, and various organelles (10, 38 -40). Thus, the potential exists that tubulin, perhaps in coordination with membrane phospholipids and/or G␤␥ subunits, could be responsible for directing GRKs to defined membrane microdomains in the cell. Finally, a particularly intriguing possibility is that such FIG. 7. Association of GRK-phosphorylated tubulin with microtubules. Forty l reactions containing 60 g of tubulin dimers, 4 g of ␤ARK or GRK5, and 100 M [␥-32 P]ATP were incubated at 30°C for 30 min followed by addition of GTP and glycerol (5 or 30%), incubation at 37°C for 30 min, and centrifugation as described under "Experimental Procedures." Samples were subjected to SDS-PAGE, Coomassie Blue staining, densitometry, and autoradiography. A, Coomassie Blue staining (Coomassie) and autoradiography ( 32 P) of a representative experiment is shown with the supernatant (S) and pellet (P) fractions for each experimental condition (5% glycerol or 30% glycerol) indicated. B, quantitation was achieved through densitometry of Coomassie Blue staining and cutting and counting of the radiolabeled tubulin bands. The relative amount of protein staining or radioactivity associated with the pellet fraction was expressed as a fraction of the total (supernatant ϩ pellet). Values are mean Ϯ S.E. from three separate experiments.
interactions may be dynamically regulated in coordination with dynamic functions of the microtubule cytoskeleton during processes such as mitosis and cell differentiation.
In summary, these data identify a novel interaction between GRKs and tubulin in which both ␤ARK and GRK5 are capable of phosphorylating tubulin in a stoichiometric fashion with a preference for ␤-tubulin and specifically the ␤III-isotype under certain conditions. Importantly, these observations are suggested to be physiologically relevant due to the ability of ␤ARK to co-purify with tubulin from brain tissue, the ability of tubulin to co-immunoprecipitate with ␤ARK from cell lysates, and the observed co-localization of ␤ARK and GRK5 with microtubules in cells. Whereas the potential functional consequences of GRK/tubulin interaction are largely unknown, GRKs appear to either selectively interact with microtubules (compared with soluble tubulin) and/or play a role in regulating tubulin assembly. Thus, GRK/tubulin interaction may represent a novel link between cell signaling and cytoskeletal molecules. As further investigations proceed into both the molecular nature and functional consequences of these interactions, it may be anticipated that GRKs will be capable of coordinating specific signaling events with changes in the stability and/or function of the cytoskeleton. Additionally, it is likely that just as tubulin has been shown to regulate a variety of other signaling molecules, tubulin may similarly affect the localization and/or function of GRKs, possibly directing distinct signaling events that remain to be identified.