Báo cáo khoa học: Molecular characterization of recombinant mouse adenosine kinase and evaluation as a target for protein phosphorylation

doi:10.1111/j.1432-1033.2004.04291.x

Eur. J. Biochem. 271, 3547–3555 (2004) (cid:1) FEBS 2004

Molecular characterization of recombinant mouse adenosine kinase and evaluation as a target for protein phosphorylation

Bogachan Sahin1, Janice W. Kansy1, Angus C. Nairn2,3, Jozef Spychala4, Steven E. Ealick5, Allen A. Fienberg3,6, Robert W. Greene1,7 and James A. Bibb1 1The University of Texas Southwestern Medical Center, Dallas, TX; 2Yale University School of Medicine, New Haven, CT; 3The Rockefeller University, New York, NY; 4University of North Carolina, Chapel Hill, NC; 5Cornell University, Ithaca, NY; 6Intra-Cellular Therapies Inc., New York, NY; 7Veterans Administration Medical Center, Dallas, TX, USA

kinases was screened for ability to phosphorylate recom- binant mouse AK. Data from these in vitro phosphorylation studies suggest that AK is most likely not an efficient sub- strate for PKA, PKG, CaMKII, CK1, CK2, MAPK, Cdk1, or Cdk5. PKC was found to phosphorylate recombinant AK efficiently in vitro. Further analysis revealed, however, that this PKC-dependent phosphorylation occurred at one or more serine residues associated with the N-terminal affinity tag used for protein purification.

Keywords: adenosine kinase; adenosine regulation; protein serine/threonine kinases; CNS.

The regulation of adenosine kinase (AK) activity has the potential to control intracellular and interstitial adenosine (Ado) concentrations. In an effort to study the role of AK in Ado homeostasis in the central nervous system, two iso- forms of the enzyme were cloned from a mouse brain cDNA library. Following overexpression in bacterial cells, the cor- responding proteins were purified to homogeneity. Both isoforms were enzymatically active and found to possess Km and Vmax values in agreement with kinetic parameters des- cribed for other forms of AK. The distribution of AK in discrete brain regions and various peripheral tissues was defined. To investigate the possibility that AK activity is regulated by protein phosphorylation, a panel of protein

compartments [5]. Adenosine kinase (AK), which catalyzes the transfer of the c-phosphate from ATP to the 5¢-hydroxyl of Ado, generating AMP and ADP, is one of several enzymes responsible for maintaining steady-state Ado levels [6]. The structure of AK has been determined at 1.5 A˚ resolution and consists of one large and one small a/b domain and two Ado binding sites [7]. AK has a low Km value [8] that falls within the range of extracellular Ado levels (25–250 nM) [9], suggesting that the reaction it catalyzes may be the primary route of Ado metabolism under physiological conditions. Moreover, AK inhibitors are effective pharmacological reagents for increasing inter- stitial Ado levels [10]. Thus, it is likely that mechanisms that might regulate AK activity would be important in the modulation of extracellular Ado concentrations. Adenosine (Ado) is a potent biological mediator and a key participant in cellular energy metabolism. In the central nervous system (CNS), extracellular Ado behaves primarily as a tonic inhibitory neuromodulator that controls neuronal excitability through its interaction with four distinct subtypes of G protein-coupled receptors, A1, A2A, A2B, and A3 [1]. A1 receptor signaling in the cholinergic arousal centers of the basal forebrain and brainstem reduces cholinergic CNS tone, facilitating the transition from waking to sleep [2]. A2A receptors in the striatum are involved in the modulation of locomotor activity, pain sensitivity, vigilance, and aggression [3]. Caffeine, the most widely used psychomotor stimulant substance in the world, is a well-known Ado antagonist of both A1 and A2A receptor subtypes [4].

Materials and methods

Facilitated diffusion of Ado across the cell membrane via equilibrative nucleoside transporters closely couples baseline Ado concentrations in the intracellular and extracellular

Correspondence to J. A. Bibb, Department of Psychiatry, The University of Texas Southwestern Medical Center, 5323 Harry Hines Blvd., NC5.410, Dallas, TX 75390–9070, USA. Fax: + 1 214 6481293; Tel.: + 1 214 6484168; E-mail: james.bibb@utsouthwestern.edu Abbreviations: AK, adenosine kinase; Ado, adenosine; hAK, human adenosine kinase; mAK, mouse adenosine kinase. Note: Nucleotide sequence data for the long and short isoforms of mouse adenosine kinase are available in the DDBJ/EMBL/GenBank databases under the accession numbers, AY540996 and AY540997, respectively. (Received 24 March 2004, revised 29 June 2004, accepted 14 July 2004)

Chemicals and enzymes

All chemicals were from Sigma, except where indicated. Deoxyoligonucleotides were obtained from Integrated DNA Technologies, Inc. Restriction and DNA modifying enzymes were from New England Biolabs. Electrocompe- tent bacteria were from Life Technologies, Inc. Cloning and expression vectors were from Invitrogen and Novagen. Site-directed mutagenesis reagents were from Stratagene. [2,8-3H]Adenosine was from Amersham Biosciences. Protease inhibitors, dithiothreitol, isopropyl thio-b-D-gal- actoside, and ATP were from Roche. [32P]ATP[cP] was from PerkinElmer Life Sciences. The catalytic subunit of

3548 B. Sahin et al. (Eur. J. Biochem. 271)

(cid:1) FEBS 2004

dithiothreitol, with two changes of buffer. Eluted and dialyzed protein (10 lg) was analyzed for purity by SDS/ PAGE (15% acrylamide). In the final set of experiments (Fig. 5F), the N-terminal affinity tag was removed using biotinylated thrombin (Novagen) according to the manu- facturer’s recommendations.

AK activity assays

PKA was purified from bovine heart as previously described [11]. PKG and cGMP were purchased from Promega; MAPK, CaMKII, and calmodulin from Upstate; and CK1, CK2, and Cdk1 from New England Biolabs. Cdk5 and p25 were coexpressed in insect Sf9 cultures using baculovirus vectors. PKC (a mixture of Ca2+-dependent isoforms, a, b and c) was purified from rat brain [12]. Recombinant protein phosphatase inhibitor-1 and DARPP-32 were generated as previously described [13,14]. Recombinant tyrosine hydroxylase was kindly provided by P. Fitzpatrick and C. Daubner, Texas A&M University. Purified histone H1 and myelin basic protein were from Upstate Biotechnology. TLC plates were from Analtech (microcrystalline cellu- for phosphoamino acid analysis) and Bodman lose, for phospho- (polyethyleneimine-impregnated cellulose, peptide mapping). Biotinylated thrombin and streptavidin agarose were from Novagen.

Molecular cloning and site-directed mutagenesis

Kinetic analysis of AK activity was performed under empirically defined linear steady-state conditions. Reactions were carried out at 37 (cid:2)C in a final volume of 20 lL. Reaction mixtures contained 50 mM Tris/HCl, pH 7.5, 100 mM KCl, 5 mM MgCl2, 5 mM b-glycerol phosphate, 3 mM ATP, dilutions of [2,8-3H]adenosine with a specific activity of 20–50 CiÆmmol)1, and recombinant mAK-L or mAK-S. Reactions were stopped by incubation at 95 (cid:2)C and were spotted onto Grade DE81 DEAE cellulose discs. The discs were washed in 5 mM ammonium formate to remove unphosphorylated adenosine and subjected to liquid scintillation counting.

Immunoblot analysis

Mouse brain and peripheral tissues were rapidly dissected, homogenized by sonication, and boiled in 1% SDS. Appropriate measures were taken to minimize pain or discomfort in accordance with the Guidelines laid down by the NIH regarding the care and use of animals for experimental procedures. Protein concentrations were determined by BCA assay (Pierce). Twenty-five micro- grams of total protein from each sample was subjected to SDS/PAGE (15% acrylamide), followed by electrophoretic transfer to nitrocellulose membrane and detection by enhanced chemiluminescence. The blot was screened for the presence and abundance of AK using a mouse ascites fluid monoclonal antibody [15]. Known amounts of purified recombinant AK were included as standards for quantification. Results were quantitated using NIH IMAGE software.

Invitrophosphorylation reactions

Long and short forms of mouse AK (mAK-L and mAK-S) were amplified by PCR from a mouse brain cDNA library (courtesy of L. Monteggia, UT Southwestern, Dallas, TX). the Primers: 5¢-GGTGCATATGGCAGCTGCGG for 5¢ end; 5¢-TCCACTCCACAGCCTGAGTT for the 3¢ end. PCR products were TA-cloned into the bacterial vector pCR II-TOPO (Invitrogen) and subjected to automated fluores- cent DNA sequencing using primers specific for the T7 and Sp6 promoters. For protein expression, a 5¢-primer including an NdeI restriction site and a 3¢-primer containing a BamHI restriction site were used to subclone mAK-L and mAK-S cDNA sequences into a hybrid bacterial expression vector based on pET-16b and incorporating the multiple cloning region of pET-28a (Novagen). Primers: 5¢-CGTGGGGT GCATATGGCAGCTGCG for the 5¢ end of mAK-L; 5¢-GTAGGTGCACATATGACGTCCACC for the 5¢ end of mAK-S; 5¢-ATATAGGATCCTCAGTGGAAGTC TGG for the 3¢ end of both clones. Consensus PKC phosphorylation sites were selected for site-directed muta- genesis using SCANSITE software, a web-based program for motif prediction (http://scansite.mit.edu). Site-directed mutants were generated at these and other sites using a standard kit (Stratagene) and following the manufacturer’s recommendations for mutagenic primer design. Mutations were confirmed by DNA sequencing along both strands, using primers specific for the T7 promoter and T7 terminator.

Purification of mAK-L and mAK-S protein

All reactions were carried out at 30 (cid:2)C in a final volume of at least 30 lL containing 10 lM substrate, 100 lM ATP, and 0.2 mCiÆmL)1 [32P]ATP[cP]. The PKC reaction solu- tion included 20 mM MOPS, pH 7.2, 25 mM b-glycerol phosphate, 1 mM sodium orthovanadate, 1 mM dithiothre- itol, 1 mM CaCl2, 10 mM MgCl2, 0.1 mgÆmL)1 phospho- tidylserine, 0.01 mgÆmL)1 diacylglycerol. PKA reactions were conducted in 50 mM HEPES, pH 7.4, 1 mM EGTA, 10 mM magnesium acetate, and 0.2 mgÆmL)1 bovine serum albumin; PKG reactions in 40 mM Tris/HCl, pH 7.4, 20 mM magnesium acetate, and 3 lM cGMP; MAPK reactions in 50 mM Tris/HCl, pH 7.4, 10 mM MgCl2, and 20 mM EGTA; and Cdk5 reactions in 30 mM MOPS, pH 7.2, and 5 mM MgCl2. For CaMKII, CK1, CK2, and Cdk1, reaction buffers provided by the suppliers were used. As positive controls, reactions were conducted using proteins previously defined as physiological substrates for each protein kinase. Specifically, protein phosphatase inhibitor-1 was used in the PKA, MAPK, Cdk1 and Electrocompetent BL21 (DE3) cells were transformed with hybrid pET-28a/16b expression vectors incorporating the cDNA of mAK-L or mAK-S downstream from a vector- encoded polyhistidine tag and thrombin cleavage site. Cultures were grown to log phase and induced with isopropyl thio-b-D-galactoside at room temperature for 20 h. Following lysis by French press and centrifugation at 10 000 g, cleared lysates were incubated with Ni-NTA agarose beads (Qiagen). The beads were washed and applied to an elution column. Bound protein was eluted using a linear gradient of 0–500 mM imidazole. Both AK isoforms eluted at approximately 150 mM imidazole. Samples were dialyzed overnight in 10 mM Tris/HCl, pH 7.5, and 1 mM

Recombinant mouse AK as a protein phosphorylation target (Eur. J. Biochem. 271) 3549

(cid:1) FEBS 2004

visualize the phosphoamino acid standards. Samples were visualized by autoradiography.

Results

Two isoforms of AK are expressed in mouse brain

(Fig. 1)

AK was cloned from a mouse brain cDNA library using primers specific for the 5¢- and 3¢-UTRs of human AK (hAK) [8]. Ten randomly selected clones were subse- quently sequenced. Nine of these sequences were identical and showed extensive homology with the long isoform of hAK (hAK-L), while one was homologous to hAK-S. The deduced amino acid sequences further like their human homologues, mAK-L illustrated that, and mAK-S are identical except at their respective N-termini, where the first 20 amino acids of mAK-L (MAAADEPKPKKLKVEAPQA) are replaced by four residues (MTST) in mAK-S. This results in a length of 361 and 345 amino acids for mAK-L and mAK-S, respectively. Cdk5 reactions [13,16]; myelin basic protein in the PKC reaction [17]; histone H1 in the PKG reaction [18]; tyrosine hydroxylase in the CaMKII reaction [19]; and DARPP-32 in the CK1 and CK2 reactions [20,21]. Time-course reactions were performed by removing 10 lL aliquots from the reaction solution at various time points and adding an equal volume of 5· SDS protein sample buffer to stop the reaction. Kinetic parameters were determined using the results of four experiments performed under empirically defined linear steady-state conditions. In all cases, [32P]phosphate incorporation was assessed by SDS/ PAGE (15% acrylamide) and PhosphorImager analysis. To calculate reaction stoichiometries, radiolabeled reaction products and radioactive standards were quantitated using IMAGEQUANT software (Amersham Biosciences). Standards consisted of 5 lL aliquots of serial dilutions of the reaction mixtures, with the moles of phosphate defined using the ATP concentration. Division of the signal per mole of substrate by the signal per mole of phosphate yielded the reaction stoichiometry (moles phosphate per moles substrate).

Mouse and human AK were found to be 89% homologous. Non-identical residues between the two species were dispersed throughout the sequence, although residues known to be involved in catalytic activity, such as those responsible for substrate and cation binding, were 100% conserved. At the time of this analysis, it was also noted that only one mouse AK sequence had been reported to date and that this existing sequence corres- ponded to an N-terminal truncated form [22]. That sequence has since been replaced in the database with what is reported here as mAK-L. To the best of our knowledge, this is the first report of the deduced amino acid sequence of mAK-S.

Two-dimensional phosphopeptide map and phosphoamino acid analysis Dry gel fragments containing 32P-labeled phospho-mAK were excised, rehydrated, washed, and incubated at 37 (cid:2)C for 20 h in 50 mM ammonium bicarbonate, pH 8.0, containing 75 ngÆmL)1 trypsin. The supernatant containing the tryptic digestion products was lyophilized and the lyophilate washed up to four times with water and once with electrophoresis buffer, pH 3.5 (10% acetic acid, 1% pyridine; v/v/v). The final lyophilate was resuspended in electrophoresis buffer, pH 3.5, and 10% of the total volume was set aside for amino acid analysis. The remainder of the sample was spotted on a TLC plate for one-dimensional electrophoresis. Separation in the second dimension was achieved by ascending chromatography. Resulting phosphopeptide maps were visualized by autoradiography. Smearing was consistently observed in the first dimension when microcrystalline cellulose TLC plates (Analtech) were used. After testing a number of different TLC plates, buffer compositions, and electrophoresis conditions, this issue was resolved by the use of polyethyleneimine-impregnated cellulose TLC plates (Bodman). To our knowledge, this electrophoretic separ- ation problem may be unique to AK, as a number of other phosphoproteins similarly analyzed by phosphopeptide mapping have shown little or no smearing on microcrystal- line cellulose TLC plates.

In order to study the function and regulation of mouse AK in vitro, both isoforms were subcloned into a pET expression vector encoding an N-terminal polyhistidine tag for affinity purification. Recombinant protein was purified to homogeneity by affinity-column chromatography. SDS/ PAGE analysis of the pure fractions indicated an apparent molecular weight of 45 and 43.5 kDa for polyhistidine- tagged recombinant mAK-L and mAK-S, respectively (Fig. 2A). Moreover, in vitro AK activity assays demon- strated that the two recombinant proteins were enzymati- cally active, efficiently catalyzing the phosphorylation of Ado to AMP (for mAK-L, Km ¼ 20 ± 4 nM; Vmax ¼ 16 ± 1.6 nmolÆmin)1Ælg)1, n ¼ 8) (Fig. 2B). No significant difference was noted between mAK-L and mAK-S with respect to Km and Vmax (data not shown). These kinetic parameters were also in agreement with previously reported values for other forms of AK [8].

Most tissues express more of one AK isoform than the other

For phosphoamino acid analysis, the aliquot set aside in the previous step was hydrolyzed at 100 (cid:2)C for 1 h in 6 M HCl under an N2 atmosphere. The reaction was stopped by a sixfold dilution in water and the mixture was lyophilized. The lyophilate was resuspended in electrophoresis buffer, pH 1.9 (8% acetic acid, 2% formic acid; v/v/v) and spotted on a microcrystalline cellulose TLC plate along with phosphoserine, -threonine, and -tyrosine standards. Elec- trophoresis was performed over half the length of the TLC plate using electrophoresis buffer, pH 1.9, at which point the plate was transferred into the pH 3.5 buffer and electrophoresis was carried out to completion. A 1% (v/v) ninhydrin solution in acetone was sprayed onto the plates to Quantitative immunoblot analysis of AK expression in mouse brain and peripheral tissues using a monoclonal antibody anti-hAK [15] showed highest levels of AK expression in the liver, testis, kidney, and spleen (Fig. 3). AK protein was present at intermediate levels in the brain, with most forebrain structures and the cerebellum showing somewhat higher levels of expression than the midbrain and

3550 B. Sahin et al. (Eur. J. Biochem. 271)

(cid:1) FEBS 2004

Fig. 1. Deduced amino acid sequence alignment of the long and short isoforms of human and mouse AK. Sequences are divided into two domains (yellow and green blocks) based on crystal structure for the shorter splice variant of human AK [7]. Yellow blocks constitute the catalytic domain. The regulatory domain (green blocks) folds over the catalytic domain and forms a hydrophobic pocket for Ado phosphorylation. Residues that make close contacts with Ado are indicated by red letters. Green letters denote residues that form the ATP/secondary Ado-binding site. One Mg2+ ion is coordinated between the active site and this ATP-binding site by hydrogen-bonding interactions mediated by water and the residues designated by blue letters. Stars indicate nonidentical residues.

Fig. 3. Quantitative immunoblot analysis of AK expression in mouse brain and peripheral tissues. The three lanes on the far right were used to blot 10, 50 and 100 ng of pure recombinant mAK-S for quantifi- cation purposes. Recombinant mAK-S standards have a higher apparent molecular weight than mAK-S in the sample lanes due to N-terminal polyhistidine tags. Quantification of relative AK abun- dance in each tissue examined is also shown.

Fig. 2. Preparation of active recombinant AK. (A) Purification of recombinant mAK-L and mAK-S by affinity-column chromatogra- phy. SDS/PAGE of UIT, uninduced total cellular protein; S10, lysates at 10 000 g; P10, supernatant after centrifugation of cell insoluble pellet after centrifugation of cell lysates at 10 000 g; FT, flow- through, or unbound protein, after incubation of S10 with Ni-NTA agarose beads; F1, 2 and 3, eluted peak fractions. (B) Lineweaver– Burke analysis of mAK-L activity. Values represent the average of four experiments using duplicate samples.

brainstem. Moreover, in most tissue homogenates, two protein species of different molecular mass were detectable with this antibody. These two closely migrating bands are

Recombinant mouse AK as a protein phosphorylation target (Eur. J. Biochem. 271) 3551

(cid:1) FEBS 2004

Fig. 4. Phosphorylation of recombinant mAK-L by a panel of protein kinases. PKC, PKA, PKG, MAPK, CaMKII, CK1, CK2, Cdk1 and Cdk5 were used to phosphorylate mAK-L as well as control substrates in vitro. I1, protein phosphatase inhibitor-1; MBP, myelin basic protein; H1, histone H1; TH, tyrosine hydroxylase; D32, DARPP-32. The multiple H1 bands visible by Coomassie stain and PhosphorImager analysis of the PKG reaction correspond to degradation products of the protein. The two higher molecular weight species appearing as radiolabeled bands above the AK signal in the CaMKII reaction represent autophosphorylation of the different CaMKII isoforms present in this commercial enzyme preparation. At least one of these CaMKII bands is also present in the TH lanes. The other is likely too close to the more prominent TH band to be visible.

likely to represent the long and short isoforms of the enzyme. Many of the tissues included in this analysis also showed a prevalence of one isoform over the other. For instance in the spleen, the short isoform is the predominant AK species, whereas in the testis and kidney, the long isoform is more abundant. Most brain regions, with the exception of the cerebellum, express detectable levels of only the short isoform. In the cerebellum, both isoforms are present at nearly equal levels. metry of 0.99, 0.31, 0.61 and 0.97 molÆmol)1 by PKA, MAPK, Cdk1 and Cdk5, respectively. Consistent with the existence of multiple PKC sites in myelin basic protein [23], the PKC-dependent phosphorylation of this control sub- strate reached a maximal stoichiometry of 2.35 molÆmol)1. Histone H1 was phosphorylated to a stoichiometry of 0.32 molÆmol)1 by PKG, tyrosine hydroxylase to a stoi- chiometry of 0.94 molÆmol)1 by CaMKII, and DARPP-32 to a stoichiometry of 0.49 and 0.92 molÆmol)1 by CK1 and CK2, respectively.

Phosphorylation of recombinant mouse AK by a protein kinase panel Phosphorylation of recombinant mouse AK by PKC

than 0.30 molÆmol)1

this phosphorylation occurs at

A time-course phosphorylation reaction conducted using an excess of PKC and 10 lM AK displayed linear conversion of substrate to phosphoprotein over the first 5 min and near saturation by 20 min, with a maximal stoichiometry greater (Fig. 5A). mAK-L and mAK-S served as equally efficient substrates for PKC in vitro (Fig. 5B). Kinetic analysis of the PKC- dependent phosphorylation of mAK-L revealed a Km of 6.9 ± 1.1 lM and Vmax of 68 ± 3 lmolÆmin)1Ælg)1 for this reaction (Fig. 5C, n ¼ 8). Similar values were obtained using the short isoform as a substrate (data not shown). A phosphopeptide map of mAK-L preparatively phos- phorylated by PKC showed two major spots (Fig. 5D, first panel). Phosphoamino acid analysis of the same material indicated that serine (Fig. 5D, second panel). Similar results were obtained with mAK-S (data not shown). Mutation of four PKC consensus sites to alanine (Ser48Ala, Ser85Ala, Ser272Ala, and Ser328Ala) had no Motif prediction analysis of the mouse AK sequence indicated the presence of putative phosphorylation sites for several protein kinases, including PKA, PKC, CaMKII, CK1 and CK2 (http://scansite.mit.edu). To investigate the possibility that AK activity may be regulated by protein phosphorylation, a panel of these protein kinases and others was tested for ability to phosphorylate recombinant mouse AK in vitro (Fig. 4). PKC was able to phosphorylate mAK- L efficiently. PKA, PKG, MAPK, CK2, and Cdk1 did not detectably phosphorylate mAK-L. Faint radiolabeling of mAK-L could be detected in reaction mixtures for CaM- KII, CK1, and Cdk5. However, maximal reaction stoi- chiometries were 0.007, 0.008 and 0.003 molÆmol)1, respectively, precluding subsequent biochemical analysis. Similar results were obtained when mAK-S was used as the putative protein kinase substrate (data not shown). In contrast, all control substrates were efficiently phosphoryl- ated by their respective protein kinases. At 60 min, protein phosphatase inhibitor-1 was phosphorylated to a stoichio-

3552 B. Sahin et al. (Eur. J. Biochem. 271)

(cid:1) FEBS 2004

of the first N-terminal 17 amino acids by thrombin cleavage (MGSSHHHHHHSSGLVPR/GSH, thrombin site indica- ted by forward slash) substantially diminished the PKC- dependent phosphorylation of mAK-L (Fig. 5F). Similarly, mutation of the five N-terminal serine residues in the affinity tag sequence of mAK-L resulted in a fusion protein that was no longer phosphorylated by PKC (Fig. 5G). effect on the phosphorylation of mAK-L by PKC (Fig. 5E). Mutants generated at the remaining nine conserved serine residues were also efficient PKC substrates (data not shown). In considering these observations, it was realized that in addition to six histidines and a thrombin cleavage site, the N-terminal affinity tag encoded by the expression vector incorporates five serine residues. Indeed, enzymatic removal

Recombinant mouse AK as a protein phosphorylation target (Eur. J. Biochem. 271) 3553

(cid:1) FEBS 2004

Fig. 5. Phosphorylation of recombinant mouse AK by PKC in vitro. (A) Time-course analysis of the phosphorylation of mAK-L by PKC. The radiographic image shown in the middle panel was used to derive the plotted values for phosphate incorporation. (B) Phosphorylation of mAK-L and mAK-S by PKC in vitro. The two panels represent SDS/ PAGE analysis of Coomassie-stained (top) and 32P-labeled (bottom) mAK-L and mAK-S. Reaction times are indicated at the top. (C) Lineweaver–Burke analysis of PKC phosphorylation of mAK-L. The plot represents the results of four reactions conducted under identical linear conditions using duplicate samples. (D) Phosphopep- tide mapping (PPM) and phosphoamino acid analysis (PAAA) of mAK-L preparatively phosphorylated by PKC. (E) Site-directed mutagenesis analysis of PKC phosphorylation of mAK-L. The Coo- massie stain and autoradiogram depict various forms of mAK-L phosphorylated by PKC and subjected to SDS/PAGE. The results of four in vitro phosphorylation reactions are shown in which PKC was used to phosphorylate Ser fi Ala mutants at four PKC consensus sites for 60 min. The stoichiometry of each reaction is quantified in the histogram as a percentage of the stoichiometry of PKC-dependent phosphorylation of wild-type mAK-L. (F) The effect of thrombin cleavage on the phosphorylation of mAK-L by PKC. SDS/PAGE analysis of Coomassie-stained (top) and 32P-labeled (bottom) mAK-L is shown. Reaction times are indicated at the top. (G) The effect of five Ser fi Ala mutations in the N-terminal affinity tag on the phos- phorylation of mAK-L by PKC. The two panels represent SDS/PAGE analysis of Coomassie-stained (top) and 32P-labeled (bottom) mAK-L and a quintuple mutant of mAK-L (5XS>A) incorporating serine-to- alanine mutations at the five serine residues of the N-terminal affinity tag. Reaction times are indicated at the top.

Ado and intracellular adenylate levels in the CNS and periphery. AK-knockout mice undergo normal embryo- genesis, but develop microvesicular hepatic steatosis within 4 days of birth, dying by the end of two weeks with fatty liver [25]. Conditional gene knockout may therefore provide a useful tool for studying the role of AK in other tissues at later developmental time points. Notably, inhibitors of AK have already been used effectively to elevate extracellular Ado levels [28] and shown some promise in animal models of stroke [29], seizure [30], and pain and inflammation [31]. Therefore, AK continues to be the subject of intensive study for the development of neuroprotective, cardioprotective, and analgesic agents, as well as drugs to treat sleep disorders and enhance vigilance.

Discussion

In this study, we report the cDNA and deduced amino acid sequences for two isoforms of AK expressed in the mouse brain. To date, the existence of AK splice variants has been described in several mammalian species, namely mouse [24,25], rat [26] and human [8,27]. A search for multiple forms of AK in other species is likely to generate similar results.

Although pharmacological and biochemical studies point irrefutably to the importance of AK in Ado homeostasis, the question of whether AK activity is regulated remains largely unanswered. Insulin has been shown to induce AK expres- sion in rat lymphocytes [32]. Studies in the brain have suggested that AK activity exhibits diurnal variations [33,34]. Most recently, a kainic acid-induced mouse model of epilepsy was used to demonstrate that AK expression is up-regulated in the epileptic hippocampus, coincident with pronounced astrogliosis, which may partly explain the postlesion increase in AK immunoreactivity in this region [24]. Thus, several lines of evidence indicate that AK levels and enzyme activity are modulated in a number of systems, most likely through the transcriptional and/or translational control of AK it remains unclear whether post- expression. However, translational mechanisms also exist for the direct regulation of AK activity. A better understanding of AK regulation, with regard to gene expression as well as protein structure and function, may reveal specific signaling pathways that control this enzyme and provide new targets for drug design. A number of factors could be responsible for the possible regulation of AK at the post-translational level, including protein stability, subcellular localization, regulatory binding partners, and post-translational modifications such as protein phosphorylation. In the present study, we report that in vitro AK does not serve as an efficient substrate for representatives of several major classes of protein serine/ threonine kinases. Although CaMKII, CK1, and Cdk5 were found to phosphorylate AK weakly, the maximal stoichiometry achieved in these reactions remained below 0.01 molÆmol)1. These low levels of phosphorylation effect- ively preclude further biochemical characterization, such as the identification of phosphorylation sites or the assessment of a possible effect of AK phosphorylation on AK activity. Furthermore, they strongly suggest that these reactions are unlikely to occur in vivo or otherwise be physiologically relevant. Taken together, our findings indicate that AK is unlikely to be regulated by any of the protein kinases investigated here.

Recent immunohistochemical studies have shed light on the pattern of brain AK expression, with a roughly homogenous distribution reported in astrocytes throughout the brain, in addition to pockets of high neuronal expression in the olfactory bulb, striatum, and brainstem [24]. In agreement, the immunoblot analysis shown here indicates that AK levels are roughly equivalent in most brain regions, with midbrain and brainstem structures showing somewhat lower levels than the cerebellum and various components of the forebrain. Furthermore, one or the other AK isoform predominates in most tissues, including the brain, where the short isoform is prevalent. The functional significance of this isoform preference at the level of tissues and whole organs remains unknown. Although no difference was observed in enzymatic activity between recombinant mAK-L and mAK-S, it is possible that in vivo the two molecules are functionally distinct in some other important respect, such as transcriptional and/or translational regula- tion, rate of turnover, subcellular localization, or association with as yet undefined regulatory factors.

The most abundant nucleoside kinase in mammals, AK has emerged as a key enzyme in the regulation of interstitial On the other hand, it is important to note that our screen was by no means exhaustive, and although the protein kinases tested in this study represent most of the principal classes of protein serine/threonine kinases, the possibility remains that an untested, perhaps unidentified, protein kinase phosphorylates AK. Future studies utilizing more broad-based strategies, such as immunoprecipitation of AK from radiolabeled cells or tissue preparations, may reveal AK-specific regulatory pathways of this nature.

3554 B. Sahin et al. (Eur. J. Biochem. 271)

(cid:1) FEBS 2004

9. Latini, S. & Pedata, F. (2001) Adenosine in the central nervous system: release mechanisms and extracellular concentrations. J. Neurochem. 79, 463–484.

10. Brundege, J.M. & Dunwiddie, T.V.

(1996) Modulation of excitatory synaptic transmission by adenosine released from single hippocampal pyramidal neurons. J. Neurosci. 16, 5603–5612. 11. Kaczmarek, L.K., Jennings, K.R., Strumwasser, F., Nairn, A.C., Walter, U., Wilson, F.D. & Greengard, P. (1980) Microinjection of catalytic subunit of cyclic AMP-dependent protein kinase en- hances calcium action potentials of bag cell neurons in cell culture. Proc. Natl Acad. Sci. USA 77, 7487–7491.

12. Woodgett, J.R. & Hunter, T. (1987) Isolation and characterization of two distinct forms of protein kinase C. J. Biol. Chem. 262, 4836–4843.

In addition to these central observations, our studies have produced several findings of technical significance. The results reveal a potentially important hazard in the use of the pET vector system for the recombinant expression of putative PKC substrates, and perhaps substrates of other protein kinases. With regard to the analysis of phospho- proteins by thin-layer chromatography, it should be noted that the novel use of polyethyleneimine-impregnated cellu- lose TLC plates was essential to the generation of good phosphopeptide maps using AK. These observations may be of interest to other investigators studying AK and PKC.

Acknowledgements

13. Bibb, J.A., Nishi, A., O’Callaghan, J.P., Snyder, G.L., Horiuchi, A.M.L., Ule, J., Pelech, S.L., Meijer, L., Saito, T., Hisanaga, T., Czernik, A.J., Nairn, A.C. & Greengard, P. (2001) Phosphoryla- tion of protein phosphatase inhibitor-1 by Cdk5. J. Biol. Chem. 276, 14490–14497.

14. Bibb, J.A., Snyder, G.L., Nishi, A., Yan, Z., Meijer, L., Fienberg, A.A., Tsai, L.H., Kwon, Y.T., Girault, J.A., Czernik, A.J., Huganir, R.L., Hemmings, H.C. Jr, Nairn, A.C. & Greengard, P. (1999) Phosphorylation of DARPP-32 by Cdk5 modulates dopamine signalling in neurons. Nature 402, 669–671.

15. Spychala, J. & Mitchell, B.S. (2002) Cyclosporin A and FK506 decrease adenosine kinase activity and adenosine uptake in T-lymphocytes. J. Lab. Clin. Med. 140, 84–91.

16. Huang, F.L. & Glinsmann, W.H. (1976) Separation and char- acterization of two phosphorylase phosphatase inhibitors from rabbit skeletal muscle. Eur. J. Biochem. 70, 419–426.

The authors would like to thank Lisa Monteggia at UT Southwestern Medical Center for providing the mouse brain cDNA library used in these experiments, Paul Fitzpatrick and Colette Daubner at Texas A&M University for providing recombinant tyrosine hydroxylase for use in CaMKII phosphorylation reactions, and Donna Hanson of Bodman Industries for TLC materials and technical assistance regarding TLC of AK phosphopeptide. This work was supported by funding from the Medical Scientist Training Program at UT Southwestern Medical Center (BS), the National Cancer Institute (JS), the National Institute of Drug Abuse (JAB), the National Alliance for Research on Schizophrenia and Depression (JAB), the National Institute of Mental Health (ACN and RWG), the Department of Defense (JAB and AAF), the Department of Veterans Affairs (RWG), and the Ella McFadden Charitable Trust Fund at the Southwestern Medical Foundation (JAB).

References

17. Schatzman, R.C., Raynor, R.L., Fritz, R.B. & Kuo, J.F. (1983) Purification to homogeneity, characterization and monoclonal antibodies of phospholipid-sensitive Ca2+-dependent protein kinase from spleen. Biochem. J. 209, 435–443.

1. Fredholm, B.B. (2003) Adenosine receptors as targets for drug

development. Drug News Perspect. 16, 283–289.

18. de Jonge, H.R. & Rosen, O.M. (1977) Self-phosphorylation of cyclic guanosine 3¢:5¢-monophosphate-dependent protein kinase from bovine lung: effect of cyclic adenosine 3¢:5¢-monophosphate, cyclic guanosine 3¢: 5¢-monophosphate and histone. J. Biol. Chem. 252, 2780–2783.

19. Campbell, D.G., Hardie, D.G. & Vulliet, P.R. (1986) Identifica- tion of four phosphorylation sites in the N-terminal region of tyrosine hydroxylase. J. Biol. Chem. 261, 10489–10492.

2. Strecker, R.E., Morairty, S., Thakkar, M.M., Porkka-Heiskanen, T.,Basheer,R.,Dauphin,L.J.,Rainnie,D.G.,Portas,C.M.,Greene, R.W. & McCarley, R.W. (2000) Adenosinergic modulation of basal forebrain and preoptic/anterior hypothalamic neuronal activity in the control of behavioral state. Behav. Brain Res. 115, 183–204.

20. Desdouits, F., Siciliano, J.C., Greengard, P. & Girault, J.A. (1995) Dopamine- and cAMP-regulated phosphoprotein DARPP-32: phosphorylation of Ser-137 by casein kinase I inhibits dephos- phorylation of Thr-34 by calcineurin. Proc. Natl Acad. Sci. USA 92, 2682–2685.

3. Ledent, C., Vaugeois, J.M., Schiffmann, S.N., Pedrazzini, T., El Yacoubi, M., Vanderhaeghen, J.J., Constentin, J., Heath, J.K., Vassart, G. & Parmentier, M. (1997) Aggressiveness, hypoalgesia and high blood pressure in mice lacking the adenosine A2a receptor. Nature 388, 674–678.

21. Girault, J.A., Hemmings, H.C. Jr, Williams, K.R., Nairn, A.C. & Greengard, P. (1989) Phosphorylation of DARPP-32, a dopa- mine- and cAMP-regulated phosphoprotein, by casein kinase II. J. Biol. Chem. 264, 21748–21759.

4. Fredholm, B.B., Battig, K., Holmen, J., Nehlig, A. & Zvartau, E.E. (1999) Actions of caffeine in the brain with special reference to factors that contribute to its widespread use. Pharmacol. Rev. 51, 83–133.

5. Cass, C.E., Young, J.D. & Baldwin, S.A. (1998) Recent advances in the molecular biology of nucleoside transporters of mammalian cells. Biochem. Cell Biol. 76, 761–770.

6. Arch., J.R. & Newsholme, E.A. (1978) Activities and some properties of 5¢-nucleotidase, adenosine kinase and adenosine deaminase in tissues of vertebrates and invertebrates in relation to the control of the concentration and the physiological role of adenosine. Biochem. J. 174, 965–977.

22. Singh, B., Hao, W., Wu, Z., Eigl, B. & Gupta, R.S. (1996) Cloning and characterization of cDNA for adenosine kinase from mam- malian (Chinese hamster, mouse, human and rat) species: high frequency mutants of Chinese hamster ovary cells involve struc- tural alterations in the gene. Eur. J. Biochem. 241, 564–571. 23. Kishimoto, A., Nishiyama, K., Nakanishi, H., Uratsuji, Y., Nomura, H., Takeyama, Y. & Nishizuka, Y. (1985) Studies on the phosphorylation of myelin basic protein by protein kinase C and adenosine 3¢: 5¢-monophosphate-dependent protein kinase. J. Biol. Chem. 260, 12492–12499.

7. Mathews, I.I., Erion, M.D. & Ealick, S.E. (1998) Structure of resolution. Biochemistry 37,

human adenosine kinase at 1.5 A˚ 15607–15620.

24. Gouder, N., Scheurer, L., Fritschy, J.M. & Boison, D. (2004) Overexpression of adenosine kinase in epileptic hippocampus contributes to epileptogenesis. J. Neurosci. 24, 692–701.

25. Boison, D., Scheurer, L., Zumsteg, V., Rulicke, T., Litynski, P., Fowler, B., Brandner, S. & Mohler, H. (2002) Neonatal hepatic

8. Spychala, J., Datta, N.S., Takabayashi, K., Datta, M., Fox, I.H., Gribbin, T. & Mitchell, B.S. (1996) Cloning of human adenosine kinase cDNA: sequence similarity to microbial ribokinases and fructokinases. Proc. Natl Acad. Sci. USA 93, 1232–1237.

Recombinant mouse AK as a protein phosphorylation target (Eur. J. Biochem. 271) 3555

(cid:1) FEBS 2004

steatosis by disruption of the adenosine kinase gene. Proc. Natl Acad. Sci. USA 99, 6985–6990.

30. Wiesner, J.B., Ugarkar, B.G., Castellino, A.J., Barankiewicz, J., Dumas, D.P., Gruber, H.E., Foster, A.C. & Erion, M.D. (1999) Adenosine kinase inhibitors as a novel approach to anticonvulsant therapy. J. Pharmacol. Exp. Ther. 289, 1669–1677.

31. Poon, A. & Sawynok, J.

26. Sakowicz, M., Grden, M. & Pawelczyk, T. (2001) Expression level of adenosine kinase in rat tissues. Lack of phosphate effect on the enzyme activity. Acta Biochim. Pol. 48, 745–754.

27. McNally, T., Helfrich, R.J., Cowart, M., Dorwin, S.A., Meuth, J.L., Idler, K.B., Klute, K.A., Simmer, R.L., Kowaluk, E.A. & Halbert, D.N. (1997) Cloning and expression of the adenosine kinase gene from rat and human tissues. Biochem. Biophys. Res. Commun. 231, 645–650.

28. Pak, A.A., Haas, H.L., Decking, U.K. & Schrader, J. (1994) Inhibition of adenosine kinase increases endogenous adenosine and depresses neuronal activity in hippocampal slices. Neuro- pharmacology 33, 1049–1053.

(1999) Antinociceptive and anti- inflammatory properties of an adenosine kinase inhibitor and an adenosine deaminase inhibitor. Eur. J. Pharmacol. 384, 123–138. 32. Pawelczyk, T., Sakowicz, M., Podgorska, M. & Szczepanska- Konkel, M. (2003) Insulin induces expression of adenosine kinase gene in rat lymphocytes by signaling through the mitogen-acti- vated protein kinase pathway. Exp. Cell Res. 286, 152–163. 33. Mackiewicz, M., Nikonova, E.V., Zimmerman, J.E., Galante, R.J., Zhang, L., Cater, J.R., Geiger, J.D. & Pack, A.I. (2003) Enzymes of adenosine metabolism in the brain: diurnal rhythm and the effect of sleep deprivation. J. Neurochem. 85, 348–357. 34. Alanko, L., Heiskanen, S., Stenberg, D. & Porkka-Heiskanen, T. (2003) Adenosine kinase and 5¢-nucleotidase activity after pro- longed wakefulness in the cortex and the basal forebrain of rat. Neurochem. Int. 42, 449–454.

29. Miller, L.P., Jelovich, L.A., Yao, L., DaRe, J., Ugarkar, B. & Foster, A.C. (1996) Pre- and peristroke treatment with the ade- nosine kinase inhibitor, 5¢-deoxyiodotubercidin, significantly reduces infarct volume after temporary occlusion of the middle cerebral artery in rats. Neurosci. Lett. 220, 73–76.