Identification of sites of phosphorylation by G-protein-coupled receptor kinase 2 in β-tubulin


  • Norihiro Yoshida,

    1. Department of Neurochemistry, Faculty of Medicine, University of Tokyo, Japan;
    Search for more papers by this author
    • Present address: Otsuka Pharmaceutical Co. Ltd, Research Institute for Pharmacological and Therapeutical Development, Tokushima, Japan.

  • Kazuko Haga,

    1. Department of Neurochemistry, Faculty of Medicine, University of Tokyo, Japan;
    2. Institute for Biomolecular Science, Faculty of Science, Gakushuin University, Tokyo, Japan
    Search for more papers by this author
  • Tatsuya Haga

    1. Department of Neurochemistry, Faculty of Medicine, University of Tokyo, Japan;
    2. Institute for Biomolecular Science, Faculty of Science, Gakushuin University, Tokyo, Japan
    Search for more papers by this author

T. Haga, Institute for Biomolecular Science, Faculty of Science, Gakushuin University, Mejiro 1-5-1, Toshima-ku, Tokyo 171-8588, Japan. Fax: + 81 35992 1034, Tel.: + 81 35992 1033, E-mail:


G-protein-coupled receptor kinase 2 (GRK2) is known to specifically phosphorylate the agonist-bound forms of G-protein-coupled receptors (GPCRs). This strict specificity is due at least partly to activation of GRK2 by agonist-bound GPCRs, in which basic residues in intracellular regions adjacent to transmembrane segments are thought to be involved. Tubulin was found to be phosphorylated by GRK2, but it remains unknown if tubulin can also serve as both a substrate and an activator for GRK2. Purified tubulin, phosphorylated by GRK2, was subjected to biochemical analysis, and the phosphorylation sites in β-tubulin were determined to be Thr409 and Ser420. In addition, the Ser444 in βIII-tubulin was also indicated to be phosphorylated by GRK2. The phosphorylation sites in tubulin for GRK2 reside in the C-terminal domain of β-tubulin, which is on the outer surface of microtubules. Pretreatment of tubulin with protein phosphatase type-2A (PP2A) resulted in a twofold increase in the phosphorylation of tubulin by GRK2. These results suggest that tubulin is phosphorylated in situ probably by GRK2 and that the phosphorylation may affect the interaction of microtubules with microtubule-associated proteins. A GST fusion protein of a C-terminal region of βI-tubulin (393–445 residues), containing 19 acidic residues but only one basic residue, was found to be a good substrate for GRK2, like full-length β-tubulin. These results, together with the finding that GRK2 may phosphorylate synuclein and phosducin in their acidic domains, indicate that some proteins with very acidic regions but without basic activation domains could serve as substrates for GRK2.


G-protein-coupled receptor


G-protein-coupled receptor kinase


phosphatase 2A


microtubule-associated protein


poly(vinylidene difluoride)


glutathione S-transferase

Many G-protein-coupled receptors (GPCRs) including rhodopsin, muscarinic acetylcholine receptors, and β-adrenergic receptors are known to be phosphorylated in a light-dependent or agonist-dependent manner by members of the protein kinase family called G-protein-coupled receptor kinases (GRKs) [1]. GRKs constitute a subgroup of the serine/threonine kinase superfamily and are characterized by their strict substrate specificity, i.e. they only recognize the stimulated forms of GPCRs. The phosphorylation sites in rhodopsin [2] and β2-adrenergic receptors [3] for GRK1 and GRK2, respectively, are located in their C-termini, and those in muscarinic acetylcholine receptor M2 subtypes (M2 receptors) [4], M3 receptors [5], and α2-adrenergic receptors [6] for GRK2 are in the central parts of their third intracellular loops. No strict consensus sequence for GRK-mediated phosphorylation has been found among these phosphorylation sites, except that acidic amino-acid residues near the phosphorylation sites may be required [7].

Peptides corresponding to these phosphorylation sites are generally poor substrates for GRK1 or GRK2, but their phosphorylation is greatly stimulated by rhodopsin [8], β2-adrenergic receptors [9] or M2 receptors [10] in a light-dependent or agonist-dependent manner. GRK2, but not GRK1, is also stimulated by G-protein βγ subunits, and this phosphorylation is synergistically stimulated by agonist-bound receptors and G-protein βγ subunits [11–14]. These results indicate that light-exposed rhodopsin, agonist-bound β-adrenergic receptors, or M2 receptors function both as substrates and activators, and explain, at least partly, why the substrates of GRK2 are restricted to agonist-bound receptors in spite of the absence of a strict consensus sequence among various phosphorylation sites. As phosphorylation site-deleted rhodopsin [15] and M2 receptors [10] also act as activators of GRK2, the activation sites are thought to be different from the phosphorylation sites. Possible activation sites in M2 receptors are suggested to be several portions of intracellular loops adjacent to transmembrane segments, because the peptides corresponding to these regions stimulated phosphorylation of synthetic peptides corresponding to the phosphorylation sites in M2 receptors [14]. These regions are assumed to undergo a conformational change on agonist binding and to be involved in the interaction with G-proteins [16,17]. Furthermore, mastoparan, which is known to mimic agonist-bound receptors and activates G-proteins [18], has been shown to stimulate GRK1 [15] and GRK2, particularly in the presence of G-protein βγ subunits [14]. All these peptides with GRK2-stimulating activity, including mastoparan, are basic peptides.

Recent studies have suggested that GRK may phosphorylate substrates other than the stimulated forms of GPCRs. Tubulin is the first nonreceptor protein found to be phosphorylated by GRK2 and GRK5, although its phosphorylation sites have not been identified yet [19–21]. Other nonreceptor substrates for GRK2 have been reported, including synucleins [22], phosducin, and phosducin-like protein [23]. The phosphorylation sites in synucleins and phosducin are located in their C-terminal domains, which include many acidic residues but few basic residues. It remains unknown, however, whether the C-terminal peptides serve as good substrates for GRK2 by themselves or do so only in the presence of activating domains in another part of these proteins. We have attempted to identify phosphorylation sites for GRK2 in tubulin as a first step to determining if tubulin serves as both a substrate and an activator for GRK2, as was shown in the case of stimulated forms of GPCRs.

Here, we show that tubulin is phosphorylated by GRK2 in a very acidic C-terminal domain and that the C-terminal peptide of tubulin is a good substrate for GRK2, suggesting that the presence of a basic activation domain is not necessary for the protein to be a substrate for GRK2. In addition, we present evidence that tubulin is phosphorylated in situ at the sites phosphorylated by GRK2.

Materials and methods


Phenyl-sepharose, heparin-sepharose, glutathione-sepharose 4B, sephadex G-50 fine, [γ-32P]ATP, the pGEX4T-3 vector, and an ECL chemiluminescence detection system were purchased from Amersham Pharmacia Biotech. Achromobacter protease I and endoproteinase Asp-N were purchased from Wako Pure Chemical Industries. KOD polymerase, Pfu turbo polymerase, the pBluescript vector, and restriction enzymes were from Toyobo. C18 RP-HPLC and DEAE-5PW columns were from Tosoh. A thermo sequence fluorescent-labeled primer cycle sequencing kit was purchased from Perkin–Elmer. TLC plates were purchased from Merck, and human erythrocyte phosphatase 2A (PP2A) was from Upstate Biotechnology Inc. Other reagents used were of the highest grade commercially available.

Protein expression and purification

GRK2 was overexpressed in and purified from Sf9 insect cells as described previously with some modifications [10]. The infected Sf9 cells were homogenized in 20 mm Hepes/KOH (pH 7.0), containing 2 mm MgCl2, 1 mm dithiothreitol, and 0.5 mm phenylmethanesulfonyl fluoride (solution A; 20 mL per cell pellet from 1 L culture). The homogenate was centrifuged, and then the pellet was homogenized in solution A supplemented with 0.5 m KCl. Most of the GRK2 activity was recovered in the supernatant obtained by centrifugation at 42 000 g for 20 min. Ammonium sulfate was added to the extract to a saturation level of 20%. After centrifugation, a saturated ammonium sulfate solution was added to the supernatant to give a final concentration of 30%, and then the suspension was centrifuged and the resulting pellet dissolved in solution A (15 mL). This solution was applied to a phenyl-Sepharose column (5 mL) equilibrated with 20 mm Hepes/KOH (pH 7.0), containing 1 mm dithiothreitol, 0.5 mm phenylmethanesulfonyl fluoride, and 1 m ammonium sulfate at a flow rate of 1 mL·min−1. After the column had been washed thoroughly, proteins were eluted with a linear gradient of 1.0–0 m ammonium sulfate (1.0 and 0 m, 20 mL each) and collected in fractions of 2 mL each. The fractions containing tubulin-phosphorylating activity were combined, dialyzed against 20 mm Hepes/KOH (pH 7.0)/50 mm NaCl, and then applied to a heparin column (1 mL) equilibrated with the dialysis buffer. Proteins were eluted with a linear gradient of 50–500 mm NaCl in 20 mm Hepes/KOH (pH 7.0) (30 mL in total) and collected in fractions of 1.5 mL each. Each fraction was assayed for GRK2, and subjected to SDS/PAGE on a 12% acrylamide gel. The purified GRK2 was mixed with an equal volume of glycerol as a stabilizer and stored at −80 °C until use. Crude tubulin, which contained microtubule-associated proteins (MAPs), was prepared from porcine brains by the polymerization–depolymerization procedure, which was performed three times, as described previously [24]. The tubulin was further purified by phosphocellulose chromatography [25].

Phosphorylation of tubulin

Tubulin was phosphorylated with GRK2 as described previously with some modifications [19]. Briefly, various concentrations of tubulin were incubated with 40 nm GRK2 in a buffer comprising 50 µm[γ-32P]ATP (100 c.p.m.·pmol−1), 20 mm Tris/HCl (pH 7.4), 50 mm KCl, 2 mm EDTA, 0.5 mm EGTA, and 5 mm MgCl2 at 30 °C, followed by SDS/PAGE. Incorporation of radioactivity into the tubulin was visualized by autoradiography and quantified with a Fuji BioImage BAS2000 analyzer.

Overlaying and detection of GRK2

Purified tubulin was reduced and carboxymethylated essentially as described previously [26]. Purified tubulin was lyophilized and then dissolved in 140 µL 6 m guanidinium hydrochloride in 0.1 m Tris/HCl (pH 8.5), 40 µL propan-2-ol, and 2 µL 2-mercaptoethanol by incubation at room temperature for 2 h. The tubulin solution was then carboxymethylated by mixing it with 1 µL 1 m iodoacetic acid in 1 m NaOH, followed by incubation of the mixture at room temperature in the dark for 40 min. The reaction was terminated by the addition of excess 2-mercaptoethanol, and then the mixture was passed through a column of Sephadex G-50 fine (2 mL) previously equilibrated with 50 mm ammonium carbonate (pH 9.0). The carboxymethylated tubulin was subjected to SDS/PAGE and then transferred to a poly(vinylidene difluoride) (PVDF) membrane [27,28]. The PVDF membrane was incubated in blocking buffer [0.1% (v/v) Tween 20 and 5% (w/v) nonfat dry milk in NaCl/Pi] for 1 h at 4 °C, and subsequently washed three times with binding buffer [0.1% (v/v) Tween 20 and 0.5% (w/v) nonfat dry milk in NaCl/Pi]. The PVDF membrane was then incubated with GRK2 in binding buffer overnight at 4 °C. After the PVDF membrane had been washed three times with binding buffer, GRK2 was detected by incubating the PVDF membrane with anti-GRK2 IgG. For immunological detection, horseradish peroxidase-conjugated anti-IgG antibodies and an ECL chemiluminescence system were used according to the manufacturer's instructions.

Digestion of tubulin

Phosphorylated and then carboxymethylated tubulin (100 µg) was treated with 1 µg Achromobacter protease I (EC in 200 µL 100 mm ammonium carbonate buffer (pH 9.0) at 37 °C for 60 min. The digested peptides were applied to a DEAE-5PW column equilibrated with 50 mm ammonium carbonate (pH 9.0)/50 mm NaCl at a flow rate of 0.5 mL·min−1. After the column had been washed, the peptides were eluted with a linear gradient of 50–500 mm NaCl in 50 mm ammonium carbonate, pH 9.0 (30 mL in total) and collected in fractions of 1 mL each. Radioactivity was detected by Cerenkov counting. The fractions containing the phosphopeptides were combined and digested overnight with 10 µg endoproteinase Asp-N at 30 °C. The reaction product was applied directly to a C18 RP-HPLC column, which was eluted with a linear gradient of 0–50% acetonitrile containing 0.1% trifluoroacetic acid in 50 min at a flow rate of 0.3 mL·min−1. The amino-acid sequences of the radioactive peptides were determined with a Hewlett–Packard G1000A Protein Sequencer.

Phosphoamino-acid analysis by TLC

A portion of the radioactive peptides eluted from the C18 column was lyophilized, resuspended in 6 m HCl, and then hydrolyzed by incubation at 110 °C for 60 min. The hydrolysate was lyophilized and then subjected to TLC with pyridine/acetic acid/water (1 : 10 : 189, v/v). The radioactive phosphoamino acids were visualized by autoradiography. The TLC plate was sprayed with 0.7% ninhydrin in acetone and heated in an oven at 65 °C to visualize the standard phosphoamino acids.

Cloning of β-tubulin and mutagenesis of its phosphorylation sites

Poly(A)-rich RNA was prepared from rat and mouse brains with Moloney murine leukemia virus reverse transcriptase (Toyobo) and then used to construct a cDNA library. The DNA fragment encoding the full-length rat βI-tubulin (accession No. AB011679) or mouse βIII-tubulin (accession no. NM_023279) was amplified by PCR using the rat or mouse brain cDNA library as a template. The PCR products were digested with EcoRI–NotI and then cloned into plasmid vector pBluescript II KS(–). For construction of mutant βI-tubulin and βIII-tubulin, Thr409 and Ser420 of βI-tubulin and Thr409, Ser420 and Ser444 of βIII-tubulin were replaced with Ala using the inverted amplification method [29]. Oligonucleotide primers were designed in inverted tail-to-tail directions to amplify the cloning vector together with the inserts. PCR was performed with Pfu turbo polymerase. cDNAs encoding the wild-type and mutant β-tubulins were excised as EcoRI–NotI fragments and then subcloned into EcoRI–NotI-digested expression vector pGEX4T-3, followed by transformation into Escherichia coli and expression as fusion proteins with glutathione S-transferase (GST; GST-β-tubulins). A fusion protein with GST of a peptide corresponding to positions 393–445 of rat βI-tubulin was also expressed in E. coli using an expression vector, pGEX4T-3 (GST-β-tubulinC). These GST fusion proteins were purified using glutathione-Sepharose by the procedure recommended by the manufacturer, as described previously [10].

Dephosphorylation of tubulin

Tubulin purified from porcine brains was subjected to dephosphorylation with PP2A (0.2 U) at 30 °C for 60 min. The dephosphorylation buffer contained 50 mm Mes (pH 6.8), 0.1 mm EDTA, 1 mm EGTA, 5 mm MgCl2, 0.2 mg·mL−1 BSA, and 1 mm 2-mercaptoethanol. The dephosphorylation reaction was terminated by adding 10 nm okadaic acid, followed by phosphorylation of the dephosphorylated tubulin by GRK2 at 30 °C as above. In the control sample, PP2A was incubated with 10 nm okadaic acid before the addition of tubulin. To follow the time course of dephosphorylation, tubulin was first phosphorylated in the presence of [γ-32P]ATP by GRK2 and then subjected to dephosphorylation by PP2A, followed by SDS/PAGE and quantification of the radioactivity remaining in the tubulin.


GRK2 binds specifically to β-tubulin

The α and β isotypes of tubulin were separated from each other by carboxymethylation and subsequent SDS/PAGE. After electrophoresis and Western blotting, the PVDF membrane was incubated with a purified preparation of GRK2 and then GRK2 antibodies. As shown in Fig. 1, GRK2 was found to interact only with β-tubulin. This is consistent with the study by Carman et al. [21], in which GRK2 phosphorylated β-tubulin but not α-tubulin. Therefore, these results indicate that GRK2 binds to and phosphorylates β-tubulin specifically.

Figure 1.

β-Tubulin binds to GRK2. Purified tubulin and carboxymethylated tubulin (CM-tubulin) were subjected to SDS/PAGE and then stained with Coomassie Brilliant Blue or transferred to PVDF membranes. The PVDF membranes were incubated with purified GRK2 (0.6 µg·mL−1) and then with GRK2 antibodies, as described in Materials and methods.

Partial digestion of phosphorylated tubulin

Tubulin phosphorylated by GRK2 was cleaved on the C-terminal side of Lys with Achromobacter protease I and on the C-terminal sides of Lys and Arg with trypsin. The cleaved peptides were subjected to analysis by SDS/PAGE with 18% acrylamide in Tricine. The smallest fragment obtained on treatment with Achromobacter protease I or trypsin had an apparent molecular mass of 6 kDa (data not shown). We examined the amino-acid sequence of β-tubulin, looking for the region with the expected length after the Achromobacter protease I or trypsin treatment, and found that the C-terminus of the β-tubulin had the most likely sequence to be phosphorylated by GRK2. The sequence between 392 and 430 is the same in the βII and βIII tubulin isotypes of pig, and βI, βII, βIII, and βIV isotypes of mouse (Table 1). The sequence between residue 431 and the C-terminus differs from one isotype to another, but there are no Ser or Thr residues in the region except for Ser444 in βIII-tubulin [30]. Neither Lys nor Arg is present between Lys392 and the C-terminus except for Lys450 in βIII-tubulin. Thus, the fragment obtained by treatment with Achromobacter protease I or trypsin is expected to have 53–58 residues, which corresponds to the phosphorylated 6-kDa band shown by SDS/PAGE. This C-terminal region of β-tubulin from Ala393 to Lys450 is extremely acidic with 20 acidic residues and only two basic His396 and Lys450 residues, and it has three Ser and three Thr residues.

Table 1. Sites and potential sites for GRK2-mediated phosphorylation. Phosphorylation sites for GRK2 were identified for rhodopsin [2], β2-adrenergic receptors [3], and α-synuclein and β-synuclein [22]. Potential phosphorylation sites for GRK2 are indicated for α2A-receptors [39], M2 receptors [36], M3 receptors [5], phosducin and phosducin-like protein [23]. The phosphorylation sites and potential sites are indicated in bold type as S or T. Acidic and basic amino acids are denoted by italics and underlining, respectively.
α2A-Adrenoceptor (human, 279–323,  third intracellular loop)EPAPAGPRDTDALDLEESSSSDHAERPPGPRRPERGPRGKGKARA
Phosducin-like protein  (rat, 279–301, C-terminal)VLVLTSVRNSATCHSEDSDLEID-OH

Separation of radiolabeled peptides and analysis of phosphoamino acids

Phosphorylated tubulin was carboxymethylated and then digested with Achromobacter protease I. The digested sample was loaded on to a DEAE column. All the radioactivity bound to the column, none being detected in the flow-through fraction. As shown in Fig. 2A, most of the radioactivity was eluted as a single peak. The N-terminal sequence determined for the 6-kDa fragment was AFLHWYTGEG-, which was identical with the sequence of residues 393–402 in the C-terminus of porcine β-tubulins [30].

Figure 2.

Elution profile of 32P-labeled peptides on DEAE and C18 RP-HPLC columns. (A) Phosphorylated tubulin was carboxymethylated, digested with Achromobacter protease I at 37 °C for 30 min, and then applied to a DEAE column (0.75 × 7.5 cm). The elution of peptides and radioactivity was monitored by measuring A280 (upper line) and Cerenkov radiation (lower line), respectively. Edman degradation of the fraction, which included the major phosphopeptide, revealed the N-terminal sequence to be AFLHWYTGEG(393–402). (B) The pooled fractions from the DEAE column were digested overnight with endoproteinase Asp-N at 30 °C and then applied to a C18 RP-HPLC column (0.46 × 25 cm). Elution was monitored by UV absorption at 214 nm (upper line), and radioactivity was measured by Cerenkov counting (lower line). Two phosphopeptides were eluted from the C18 RP-HPLC column, which were subjected to Edman degradation and sequence determination. The sequence of peptide 1 was determined to be DEMEFTEAESNMN(404–416) and that of peptide 2 to be DLVSEYQQYQ(417–426).

The 6-kDa fragment was further digested with endoproteinase Asp-N and then subjected to RP-HPLC on a C18 column. Phosphopeptides were eluted at ≈ 20% acetonitrile as two peaks with a linear gradient of 0–50% acetonitrile (Fig. 2B). The phosphopeptides obtained were analyzed by two-dimensional mapping on TLC plates. Each of the two peak fractions from the C18 column gave a single spot on TLC mapping (data not shown). Edman sequence analysis of each peptide (peptide 1 and peptide 2) revealed that peptide 1 had the sequence DEMEFTEAESNMN(404–416), and peptide 2 had the sequence DLVSEYQQYQ(417–426). Acid hydrolysis followed by TLC analysis revealed only labeled phosphothreonine on peptide 1 and labeled phosphoserine on peptide 2 (Fig. 3). These results indicate that the phosphorylation sites for GRK2 are Thr409 and Ser420, but not Ser413.

Figure 3.

[32P]Phosphoamino-acid analysis on TLC plates.[32P]Phosphopeptides eluted from a DEAE or C18 RP-HPLC column (peptide 1 and peptide 2) were partially hydrolyzed with HCl and then analyzed by TLC. Autoradiograms of the TLC plates are shown, together with standard phosphoamino acids.

Phosphorylation by GRK2 of GST fusion proteins of full-length βI-tubulin, βIII-tubulin, and C-terminal peptides of βI-tubulin expressed in E. coli

We cloned the βI-tubulin and βIII-tubulin genes. GST fusion proteins of βI-tubulin (GST-βI-tubulin), its C-terminal peptide (393–445) (GST-βI-tubulinC), and βIII-tubulin (GST-βIII-tubulin) were expressed in E. coli and then subjected to phosphorylation by GRK2 with different substrate concentrations (Fig. 4). Each substrate was found to be phosphorylated by GRK2 at similar rates. The Km values for GST-βI-tubulin, GST-βIII-tubulin and GST-βI-tubulinC were estimated to be 2.6, 6 and 12 µm, respectively. The Km values for GST-βI-tubulin and GST-βIII-tubulin are comparable to those reported for the phosphorylation of tubulin purified from porcine brains (0.4–3 µm) [19–21].

Figure 4.

Phosphorylation by GRK2 of GST fusion proteins of βI-tubulin and βIII-tubulin(GST-β-tubulin) and the C-terminal peptide of βI-tubulin(GST-βI-tubulinC). The indicated concentrations of GST fusion proteins were subjected to phosphorylation with GRK2 in the presence of 50 µm[γ-32P]ATP and 40 nm GRK2 for 10 min, followed by SDS/PAGE, and radioactivity counting of the tubulin band. Molar concentrations of fusion proteins were calculated from the molecular mass: GST, 27.5 kDa; GST-βI-tubulin and GST-βIII-tubulin, 82.5 kDa; GST-βI-tubulinC, 33.5 kDa. Curves were fitted to the Michaelis–Menten equation, and Km values were estimated to be 2.5 µm (GST-βI-tubulin), 6 µm (GST-βIII-tubulin) and 12 µm (GST-βI-tubulinC). These experiments were repeated three times with essentially the same results.

Phosphorylation by GRK2 of GST fusion proteins of β-tubulin mutants

To confirm that Thr409 and Ser420 are the only phosphorylation sites for β-tubulin, we constructed mutants of βI-tubulin with Ala409 and/or Ala420 in place of Thr409 and Ser420. In addition, we constructed mutants of βIII-tubulin with Ala409, Ala420, and Ala444 or Ser444 to examine whether Ser444 in βIII-tubulin is phosphorylated by GRK2. These mutant forms were expressed in E. coli as GST fusion proteins and then analyzed with respect to their phosphorylation by GRK2. As demonstrated in Fig. 5A, compared with the wild-type βI-tubulin (GST-βI-tubulin), the mutant βI-tubulin (T409A and S420A) was less than 50% phosphorylated by GRK2 and the double mutant βI-tubulin (T409A/S420A) was hardly phosphorylated at all. These results confirm that Thr409 and Ser420 are the only residues in βI-tubulin phosphorylated by GRK2. On the other hand, compared with wild-type βIII-tubulin (GST-βIII-tubulin), the double mutant βIII-tubulin (T409A/S420A) was ≈ 30% phosphorylated, and the triple mutant βIII-tubulin (T409A/S420A/S444A) was hardly phosphorylated at all (Fig. 5B). This result indicates that Ser444 of βIII-tubulin is also a site of phosphorylation.

Figure 5.

Phosphorylation of GST-βI-tubulin and βIII-tubulin mutants. (A) Residues Thr409 and Ser420, and both Thr409 and Ser420 residues in βI-tubulin were replaced with alanine residues, yielding mutants T409A, S420A, and T409A/S420A, respectively. (B) Residues Thr409 and Ser420 and/or Ser444 residues were replaced with alanine residues, yielding mutants T409A/S420A and T409A/S420A/S444A. GST fusion proteins of these mutants were expressed in E. coli and then purified as described in Materials and methods. These β-tubulin mutants were subjected to phosphorylation with GRK2. The values are the means of three independent experiments for each βI-tubulin and βIII-tubulin mutants with similar results and are expressed as percentages of the control value for wild-type GST-βI or βIII-tubulin. Error bars represent means ± SD.

Phosphorylation of phosphatase-treated tubulin

Tubulin purified from porcine brains was phosphorylated with GRK2 and then dephosphorylated with PP2A. About 80% of the phosphate was removed from the tubulin on treatment with 0.2 U PP2A for 40 min at 30 °C, as was the case for βIII-tubulin [31,32](Fig. 6). We treated the purified tubulin with PP2A and then phosphorylated it with GRK2. As shown in Fig. 7, the amount of phosphorylation doubled on pretreatment with PP2A. This result indicates that tubulin had been phosphorylated when purified and that the endogenous phosphorylation is susceptible to PP2A and that the site can be phosphorylated by GRK2.

Figure 6.

Dephosphorylation by PP2A of phosphorylated tubulin. Tubulin purified from porcine brains was phosphorylated with [γ-32P]ATP by GRK2 at 30 °C for 30 min as described above and then incubated with or without PP2A (0.2 U) for the indicated time at 30 °C, followed by SDS/PAGE and counting of radioactivity on tubulin band. The indicated values are means of triplicate determinations in one of three independent experiments with similar results and are expressed as percentages of the amount of tubulin phosphorylated before addition of PP2A. Error bars represent means ± SD.

Figure 7.

Effects of PP2A treatment on phosphorylation of tubulin by GRK2. Purified tubulin was incubated with or without PP2A for 60 min at 30 °C followed by phosphorylation by GRK2 at 30 °C for 30 min. In samples designated PP2A + okadaic acid, PP2A was preincubated with okadaic acid before the phosphorylation reaction. The indicated values are means of triplicate determinations in one of three independent experiments with similar results. Error bars represent means ± SD.


In this report, we have shown that the phosphorylation sites in tubulin for GRK2 reside in the C-terminal domain of β-tubulin, and that two (Thr409 and Ser420) of five Ser or Thr residues in this domain are phosphorylated. As the four isotypes of β-tubulin, βI, βII, βIII and βIV, have the same sequence around the phosphorylation sites, it is most likely that all these isotypes serve as substrates for GRK2.

The extent of phosphorylation by GRK2 was found to be increased when tubulin was pretreated with PP2A after its purification from porcine brain. This result indicates that tubulin is phosphorylated in situ at sites from which phosphate may be removed by PP2A and to which phosphate may be added by GRK2. One of the most likely candidate sites is Ser444 in βIII-tubulin, although it is also possible that Thr409, Ser420, and other residues are the relevant sites. Evidence for this is that Ser444 has been identified as the phosphorylation site in brain-specific βIII-tubulin phosphorylated in cultured cells [33] and in the brain [34]. Furthermore, Khan and coworkers have reported that phosphate on the Ser444 residue of βIII-tubulin is resistant to a wide variety of phosphatases, except human erythrocyte PP2A [31], which is known to bind to polymerized tubulin [32]. Moreover, we demonstrated phosphorylation of Ser444 by GRK2 using recombinant tubulin mutants, although the phosphorylation was not detected for tubulin purified from porcine brain. These results suggest that GRK2 is the kinase that phosphorylates Ser444, although we cannot exclude the involvement of other kinases such as casein kinase II [35]. This assumption is supported by the observation that GRK2 is localized with microtubules in intact cells and the localization is facilitated by agonist-bound GPCRs [20].

The phosphorylation sites for GRK2 have been determined for rhodopsin [2], β2-adrenergic receptors [3], and synucleins [22]. Serine and threonine clusters have also been shown to be phosphorylation sites for GRK2 in M2 receptors [36], M3 receptors [5], α2A-adrenergic receptors [37], and phosducin [23], although the phosphorylated amino-acid residues have not been determined definitely. These phosphorylation sites and phosphorylation site candidates are shown in Table 1. Each of these phosphorylation sites resides in an acidic domain with a fairly long span. It may be a prerequisite for phosphorylation by GRK2 that the phosphorylation sites are in an acidic domain. However, it is not the only condition for phosphorylation that the Ser and Thr residues are in the acidic domain, as GRK2 phosphorylated Thr409, Ser420 and Ser444 but not Thr399, Ser413, and Thr429 in the C-terminal domain of βIII-tubulin. Further research is necessary into what discriminates phosphorylated from nonphosphorylated residues.

Initially, we hypothesized that tubulin may serve as both a substrate and an activator for GRK2 and that it contains a basic GRK2-activating domain besides a substrate domain. This working hypothesis is not supported by the present findings that the C-terminal peptide of βI-tubulin (βI-tubulinC), which is very acidic and does not contain a basic domain, is as good a substrate as full-length tubulin. Even if tubulin contains a basic GRK2-activating domain, the effect of the putative domain should not be important because the Km values for βI-tubulin and βI-tubulinC only differ by a factor of 5. Therefore, it is likely that tubulin is a substrate for GRK2 for a different reason from that in the case of agonist-bound GPCRs.

Synucleins and phosducin have been reported to be substrates for GRK2, but it is not known if they have basic domains which serve as activators for GRK2. However, we have noticed a common characteristic of tubulin, synucleins, and phosducin, i.e. all three proteins have very acidic C-terminal domains that include phosphorylation sites. The C-terminal domain of βIII-tubulin contains 20 acidic residues in a span of 58 residues (35%) with only two basic residue (His and Lys). The C-terminal domains of synucleins (α and β) also contain phosphorylation sites in very acidic domains, with 37–40% of acidic residues and no basic residues (Table 1). The phosphorylation sites in GPCRs are also in an acidic domain, but the acidic nature is much less evident. The presence of very acidic domains, particularly in the C-termini of nonreceptor substrates, may constitute a criterion for phosphorylation by GRK2.

The C-terminal domain containing Thr409 and Ser420 has been shown to form an α-helix (H12, 408–423 residues) and to be located on the outermost surface of microtubules [38]. The C-terminal residues including Ser444 of β-tubulin, which are lacking in the structure model, are also thought to be located on the outermost surface of microtubules. This is consistent with the findings that microtubules as well as tubulin dimer can be phosphorylated by GRK2 and that phosphorylated tubulin can polymerize into microtubules [19]. The C-termini of α and β tubulin are thought to be involved in the binding of MAPs and motor proteins [39,40]. MAPs and motor proteins are known to have major roles in microtubule assembly, organelle transport, and mitosis. It is possible that phosphorylation of the C-terminus of β-tubulin by GRK2 affects the microtubule dynamics or cellular mechanisms by affecting the binding of MAPs or motor proteins. In addition, a series of recent studies have demonstrated that Gα or Gβγ subunits interact directly with tubulin [41–43] and that muscarinic receptor activation induces transient translocation of tubulin to the plasma membrane [44]. Furthermore, microtubules have been suggested to mediate the internalization of β-adrenergic receptors [45]. The GRK-mediated phosphorylation of tubulin may affect physiological processes including GPCRs, and the interaction of GRK2 with tubulin may have an effect on the function of GRK2.


We thank Professor K. Matsushima and Mr Y. Terashima for their help in determining the peptide sequences. This work was supported in part by grants from the Japan Society for the Promotion of Science (Research for Future Program), and from the Japan Science and Technology Corporation (CREST).