Phosphorylation of Ser640 in Muscle Glycogen Synthase by DYRK Family Protein Kinases*

Glycogen synthase, a key enzyme in the regulation of glycogen synthesis by insulin, is controlled by multisite phosphorylation. Glycogen synthase kinase-3 (GSK-3) phosphorylates four serine residues in the COOH terminus of glycogen synthase. Phosphorylation of one of these residues, Ser640 (site 3a), causes strong inactivation of glycogen synthase. In previous work, we demonstrated in cell models that site 3a can be phosphorylated by an as yet unidentified protein kinase (3a-kinase) distinct from GSK-3. In the present study, we purified the 3a-kinase from rabbit skeletal muscle and identified one constituent polypeptide as HAN11, a WD40 domain protein with unknown function. Another polypeptide was identified as DYRK1A, a member of the dual-specificity tyrosine phosphorylated and regulated protein kinase (DYRK) family. Two isoforms of DYRK, DYRK1A and DYRK1B, co-immunoprecipitate with HAN11 when coexpressed in COS cells indicating that the proteins interact in mammalian cells. Co-expression of DYRK1A, DYRK1B, or DYRK2 with a series of glycogen synthase mutants with Ser/Ala substitutions at the phosphorylation sites in COS cells revealed that protein kinases cause phosphorylation of site 3a in glycogen synthase. To confirm that DYRKs directly phosphorylate glycogen synthase, recombinant DYRK1A, DYRK2, and glycogen synthase were produced in bacterial cells. In the presence of Mg-ATP, both DYRKs inactivated glycogen synthase by more than 10-fold. The inactivation correlated with phosphorylation of site 3a in glycogen synthase. These results indicate that protein kinase(s) from the DYRK family may be involved in a new mechanism for the regulation of glycogen synthesis.

Glycogen is the main storage form of glucose in mammals and plays an important role in whole body glucose metabolism. For example, a substantial proportion of ingested glucose is converted to muscle glycogen (1,2). The rate of muscle glycogen synthesis is determined by its entry into muscle and the phosphorylation state of glycogen synthase, both of which processes are controlled by insulin (3,4). The muscle isoform of glycogen synthase is phosphorylated at nine or more sites by multiple protein kinases (for reviews see Refs. 4 -6). Phosphorylation leads to inactivation of glycogen synthase, but activity can be restored by the allosteric activator glucose 6-phosphate. In rabbit skeletal muscle glycogen synthase, the critical phosphorylation sites that control the enzyme activity are NH 2 -terminal residues Ser 7 (site 2) and Ser 10 (site 2a) and COOH-terminal residues Ser 640 (site 3a) and Ser 644 (site 3b) (7)(8)(9)(10). Of these, site 3a is probably the most important. In vitro, glycogen synthase kinase-3 (GSK-3) 1 phosphorylates sequentially sites 4, 3c, 3b, and 3a, where recognition of site 4 by GSK-3 requires that glycogen synthase has been previously phosphorylated at site 5 by casein kinase II (11)(12)(13). It was proposed that phosphate serves as part of the recognition determinant for GSK-3 in the sequence motif -S-X-X-X-S(P)- (14), a hypothesis that has been reinforced by the recent solution of the crystal structure of GSK-3 (15,16). This mechanism of phosphorylation has been termed "hierarchal" (6). However, we also demonstrated that, for glycogen synthase expressed in COS cells and Rat1 fibroblasts, disruption of the recognition sequence for GSK-3 by Ser/Ala substitution at sites 3c, 4, and/or 5 did not preclude inactivation of the enzyme (9,10,17). Our work has shown also that sites 3a and 3b can be directly phosphorylated by as yet unidentified protein kinases (10,18). In the present study, we purified a protein kinase that specifically phosphorylates site 3a in glycogen synthase (3a-kinase). We identified the 3akinase catalytic subunit as a member of the DYRK (dual specificity tyrosine phosphorylated and regulated kinase) family of protein kinases. The protein kinase isolated from muscle is oligomeric and also contains a WD domain protein called Han11. Control of glycogen synthase by DYRK represents a novel mechanism, and a potentially novel pathway, for the regulation of glycogen synthesis.

Construction of Expression Vectors and Site-directed Mutagenesis-
The cDNA for rabbit muscle glycogen synthase was subcloned into the pCMV-4 expression vector as described previously (19) and was designated pCMV-GS. Construction of the vector containing the mutation in the cDNA for glycogen synthase that changes amino acid 7 (site 2) from serine to alanine was described earlier (9) and was designated pCMV-GS(site 2). To construct the vectors for expression of rabbit skeletal muscle glycogen synthase in Escherichia coli a novel NdeI site was introduced at the initiator ATG by site-directed mutagenesis using PCR with pCMV-GS and pCMV-GS(site 2) as templates. The PCR-generated fragments of wild type and mutated (site 2) glycogen synthase were digested with both NdeI and AsuII, then combined with the AsuII/ BamHI fragment from pCMV-GS and ligated into NdeI/BamHI-digested pET-23a vector to generate pET-GSBam and pET-GS(site 2)Bam, respectively. In these vectors, the cDNA for glycogen synthase encodes amino acid residues from Met 1 to Arg 303 . To make a truncated protein, the GAC codon for Asp 683 was replaced by the TGA stop codon in pCMV-GS vector as described previously (10). The cDNA for truncated glycogen synthase was digested with BamHI, and the 1.1-kb fragment was ligated into BamHI-cut pET-GSBam to produce pET-GS⌬ vector. The pCMV-4-based constructs for expression of truncated glycogen synthase, which additionally contains phosphorylation site mutations, were described previously (10). Three glycogen synthase mutants, S640A,S644A,S648A,S652A,S656A⌬682 (AAAAA⌬682), S644A,S-648A,S652A,S656A⌬682 (SAAAA⌬682), and S640A,S648A,S652A,S-656A⌬682 (ASAAA⌬682) (see Ref. 10), in which the indicated Ser residues were changed into Ala, were digested with BamHI and the 1.1-kb fragments were ligated into BamHI-cut pET-GS(site 2)Bam to produce pET-2,AA⌬, pET-2,SA⌬, and pET-2,AS⌬, respectively (the mutants structure is shown in Fig. 1).
HAN11 was amplified by PCR from the EST clone AA442821 using two primers (5Ј-TGAGAATTCCATATGTCCCTGCACGGCAAACGGA-AGGAGATC-3Ј and 5Ј-AGTGTCGACTCGAGCTACACTCTGAGTATC-TCCAGGCAGTTGTTG-3Ј). To construct pGBDU-HAN11, the 1-kb PCR fragment was digested with EcoRI and SalI and cloned into Eco-RI/SalI-cut pGBDU-C2 vector (20). To construct pET32-HAN11, pGD-BU-HAN11 was digested with EcoRI and SalI, and the 1-kb fragment encoding HAN11 was ligated into pET32a (Novagen), which had been previously cut with EcoRI and SalI. The resulting vector encodes a fusion protein that contains, starting from the NH 2 terminus, a Trx tag (thioredoxin), a His 6 tag, an S-tag and HAN11. To construct a fusion protein containing an NH 2 -terminal His 6 tag, an S-tag followed by HAN11, we used PCR with two primers (5Ј-TCCCGGGAAAAATGTC-TCACCATCATCATCATCATTC-3Ј and 5Ј-AGTGTCGACTCGAGCTAC-ACTCTGAGTATCTCCAGGCAGTTGTTG-3Ј) and pET32-HAN11 as template. This PCR product was digested with SmaI and XhoI and ligated into the pcDNA vector (Invitrogen), which had been cut with HindIII, blunt-ended, and cut with XhoI. The resulting plasmid was designated pcDNA-S-HAN11. The SCAN11 gene (chromosome XVI open reading frame YPL247C) was cloned from the genome of yeast strain PJ69-4A by amplification of a 1.6-kb fragment using primers 5Ј-GAATTCCATATGGATCCCTTTCACAATGGCAAT-3Ј and 5Ј-AGAT-CTGTCGACTCAAAGGACGCGGACGTTTTGAAA-3Ј. This product was cut with EcoRI and SalI and ligated into EcoRI/SalI-cut pGBDU-C2 vector to generate pGBDU-SCAN11.
Human DYRK2 was amplified by PCR from the EST clone AW665190 using two primers (5Ј-GCGGCCGCAGAATTCATGAATGA-TCACCTGCATGTCGG-3Ј and 5Ј-AGATCTGTCGACTCAGCTAACAA-GTTTTGGCAACACTG-3Ј). To construct pFLAG-DYRK2, the 1.8-kb PCR product was digested with NotI and SalI and inserted into NotI/-SalI-cut pFLAG-CMV-2 (Sigma). To construct pGEX-DYRK2, the 1.8-kb PCR fragment was digested with EcoRI and SalI and inserted into pGEX-4T1 (Pharmacia), which had been cut with same restriction endonucleases. The 1.9-kb fragment containing the cDNA for splice variant b of human DYRK1B (Ref. 21, GenBank TM accession number NP_006474) was excised from EST clone BE907785 by StuI and ligated into the pFLAG-CMV-2 vector that had been cut with HindIII and blunt-ended and into the pGBDU-C3 vector that had been cut with EcoRI and blunt-ended. The resulting constructs, which both contain an additional 31-bp sequence from the 5Ј-untranslated region and 121-bp sequence from 3Ј-untranslated region, were designated pFLAG-DYRK-1Bb or pGBDU-DYRK1Bb. The 1.5-kb fragment containing 3Ј end of splice variant a (21, GenBank TM accession number NP_004705) of DYRK1B was generated by digestion of EST clone BG832206 with SalI and BglII. This fragment was used to substitute the 1-kb SalI/BamHI sequence in pFLAG-DYRK1Bb or the 1-kb SalI/BglII sequence in pGBDU-DYRK1Bb to produce pFLAG-DYRK1Ba or pGBDU-DYRK-1Ba, respectively. To construct pGEX-DYRK1Ba, pGBDU-DYRK1Ba was digested with BglII, blunt-ended, and cut with EcoRI. Then the 2.4-kb fragment was ligated into pGEX-4T1, which had been digested with SalI, blunt-ended, and cut with EcoRI. Mouse DYRK1A was amplified by PCR from EST clone BE372353 using two primers (5Ј-G-AATTCGCGGCCGCGATGCATACAGGAGGAGAG-3Ј and 5Ј-AGATCT-GTCGACTCACGAGCTAGCTACAGGACTC-3Ј). To construct pFLAG-DYRK1A, the 2.3-kb PCR product was digested with NotI and SalI and ligated into NotI/SalI-cut pFLAG-CMV-2. To construct pGEX-DYRK1A, the 2.3-kb PCR product was digested with EcoRI and SalI and ligated into EcoRI/SalI-cut pGEX-4T1.
Expression and Purification of Recombinant Proteins-E. coli cells BL21/DE3, transformed with pET vectors that contained the cDNA for mutated glycogen synthases, were grown at 37°C until the A 600 was ϳ0.7, then induced with 0.1 mM isopropyl ␤-D-thiogalactopyranoside and grown at 30°C for an additional 16 h. Purification of the recombinant proteins was by the method of Zhang et al. (13). Vectors encoding DYRK1A and DYRK2 as glutathione S-transferase (GST) fusion proteins (pGEX-DYRK1A and pGEX-DYRK2) were used to transform E. coli cell BL21/DE3. Transformants were isolated, induced with isopropyl ␤-D-thiogalactopyranoside for 16 h at 37°C, and the recombinant proteins were purified over glutathione-agarose. Bound proteins were eluted from the resin with 20 mM glutathione. The purified proteins were dialyzed against a buffer comprising 50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 15 mM ␤-mercaptoethanol, and 30% glycerol to remove residual glutathione, and were stored at Ϫ80°C. The concentrations of protein kinases were determined by measuring optical density of the bands in Coomassie-stained gels using bovine serum albumin as a standard.
Assay of Protein Kinase Activity-Protein kinase activity was determined by measuring the incorporation of [ 32 P]phosphate into glycogen synthase mutants WT⌬; 2,AA⌬; 2,SA⌬; or 2,AS⌬ (Fig. 1). Unless otherwise stated, assays contained, in a final volume of 20 l: 0.2 g of glycogen synthase, 10 M ATP, 0.5-1 Ci of [␥-32 P]ATP, a portion of enzyme sample, and kinase assay buffer (10 mM Tris-HCl, pH 7.4, 10 mM MgCl 2 , 1 mM dithiothreitol). After incubation at 37°C for 20 min, reactions were terminated by addition of SDS-PAGE sample buffer and boiling for 5 min. Polypeptides were separated by SDS-PAGE, gels were stained with Coomassie Blue and dried. The phospholabeled proteins were detected by autoradiography, then excised from gels and the radioactivity incorporated was quantitated by scintillation counting.
Purification of 3a-Kinase from Rabbit Skeletal Muscle-Approximately 6 kg of rabbit skeletal muscle was homogenized in 12 liters of buffer A (50 mM Tris-HCl, pH 7.4, 100 mM NaCl, 1 mM EDTA, 0.1% 2-mercaptoethanol, 0.1 mM phenylmethylsulfonyl fluoride (PMSF), 0.5 mg/ml benzamidine) and centrifuged for 40 min at 5000 ϫ g. The supernatant was filtered through glass wool and solid ammonium sulfate was added to bring the final concentration to 55% saturation. After stirring for 60 min, the suspension was centrifuged for 30 min at 5000 ϫ g, the supernatant was discarded, and the pellet was resuspended in 1.2 liters of buffer A. The solution was centrifuged for 60 min at 100,000 ϫ g, and the supernatant was loaded onto a 600-ml column of phenyl-Sepharose CL-4B (Sigma) equilibrated in buffer A. After subsequent washings with buffer A and buffer A with 25% ethylene glycol, 3akinase was eluted with buffer A containing 50% ethylene glycol and dialyzed against 50 mM Tris-HCl, pH 7.4, 50 mM NaCl, 1 mM EDTA, 0.1% 2-mercaptoethanol, 0.1 mM PMSF (buffer B). The dialyzed protein was chromatographed on a 200-ml column of SP-Sepharose HP (Pharmacia) equilibrated in buffer B. After washing with buffer B, 3a-kinase was eluted with a 400-ml linear gradient of 50 -400 mM NaCl. The majority of 3a-kinase activity was eluted at ϳ200 mM NaCl. The active fractions were pooled, dialyzed against 20 mM Tris-HCl, pH 8.0, 50 mM NaCl, 1 mM EDTA, 0.1% 2-mercaptoethanol, 0.1 mM PMSF (buffer C), and loaded onto a 40-ml column of Q-Sepharose (Pharmacia) equilibrated in buffer C. The column was washed with buffer C, followed by a 200-ml linear gradient of 50 -400 mM NaCl in buffer C. The 3a-kinase activity was eluted at ϳ130 mM NaCl. The most active fractions were pooled, dialyzed against 50 mM Tris-HCl, pH 7.4, 50 mM NaCl, 1 mM EDTA, 0.1% 2-mercaptoethanol, 0.1 mM PMSF, 0.01% Brij 35 (buffer D) and applied to a 5-ml column of heparin-Sepharose HT (Pharmacia) equilibrated in buffer D. The column was developed with a 100-ml linear gradient of 50 -400 mM NaCl in buffer D and 3a-kinase was eluted as one major peak at 350 mM NaCl. The active fractions were pooled, dialyzed against 10 mM potassium phosphate, pH 7.4, 50 mM NaCl, 2 mM dithiothreitol, 0.01% Brij 35 (buffer E), and loaded on a 1-ml column of ceramic hydroxyapatite (Bio-Rad) equilibrated in buffer E. The column was developed with a 40-ml linear gradient of 10 -300 mM potassium phosphate. The fractions containing 3a-kinase were pooled, dialyzed against buffer E, and stored at Ϫ80°C.
Tryptic Digestion of Purified Proteins and Amino Acid Sequencing-Purified material from the heparin-Sepharose step was concentrated and subjected to SDS-PAGE. Proteins were transferred to a polyvinylidene difluoride membrane and stained with Coomassie Blue. The membrane pieces containing protein bands were excised, destained in 70% acetonitrile, and dried. The membrane pieces were placed in 50 l of digestion mixture containing 100 mM Tris-HCl, pH 8.0, 10% acetonitrile, 0.1% reduced Triton X-100, and 0.2 g of modified porcine trypsin (Promega, sequencing grade). After incubation for 24 h at 37°C, samples were sonicated and supernatants were removed. Membranes were washed two times with 0.1% trifluoroacetic acid and the supernatants were pooled with the previous one. The combined supernatants were chromatographed on a Vydac C 18 column (250 ϫ 2.1 mm) connected to an Applied Biosystems Microbore high performance liquid chromatography system equilibrated in 0.1% trifluoroacetic acid. The column was developed with a linear acetonitrile gradient in 0.1% trifluoroacetic acid. Peptides detected by absorbance at 214 nm were collected and analyzed by Edman sequencing.
MALDI-TOF Mass Spectrometry Analysis-Protein bands were excised from gels, cut into small pieces, destained with 50% acetonitrile, 50 mM ammonium bicarbonate, reduced with 10 mM dithiothreitol (Sigma), and alkylated with 55 mM iodoacetamide (Sigma). After alkylation, the gel slices were digested with trypsin (Promega) in 50 mM ammonium bicarbonate (enzyme substrate ratio 1:50 to 1:100) overnight at 37°C. The peptides were extracted from the gel with 0.1% trifluoroacetic acid in water and 0.1% trifluoroacetic acid in 60% acetonitrile for 30 min. One microliter of the extracted protein solution was spotted on a MALDI plate with 1 l of ␣-cyano-4-hydroxycinnamic acid (Sigma) in a 50% acetonitrile, 50% methanol matrix solution. Mass spectra were recorded using the positive reflection mode of a MALDI-TOF mass spectrometer (Micromass, Manchester, UK). The time of flight was measured using the following parameters: 3,400 V pulse voltage, 15,000 V source voltage, 500 V reflectron voltage, 1,950 V MCP voltage, and low mass gate of 400 Da. For high accuracy mass measurement, the instrument was tuned to a resolution just over 15,000 full width at peak half-height. The MALDI-MS data obtained were used for data base searches.
Antibodies-HAN11 antibodies were generated against a peptide having sequence corresponding to residues 45-51 in HAN11. Rabbits were immunized with this peptide (NKVQLVGLDEESSEFIC) coupled to Imject® maleimide-activated mariculture keyhole limpet hemocyanin (Pierce). After the third booster injection, serum was collected and incubated with an affinity resin prepared by coupling a peptide to Sulfo-Link beads (Pierce). After exhaustively washing the resin, the anti-peptide antibodies were eluted with 0.3 M glycine-HCl, pH 2.7, neutralized, and dialyzed against phosphate-buffered saline. Antibodies to FLAG (M2) were from Sigma. Antibodies to Dyrk1A were from BD Biosciences. Antibodies p-GS(S641), which specifically recognize phosphorylated site 3a in glycogen synthase, were from Cell Signaling Technology.
Northern Blot Analysis-A human multiple tissue Northern blot (Clontech) was probed with a 32 P-labeled 1.6-kb HAN11 fragment excised from EST clone AA442821 by EcoRI and NotI digestion. The procedure was performed according to the manufacturer's protocol.
Electrophoresis and Immunoblotting-To analyze protein expression, mouse tissues were homogenized in a buffer containing 50 mM Tris-HCl, pH 7.4, 100 mM NaCl, 1 mM EDTA, 0.1% Triton X-100, 0.1 mM N ␣ -p-tosyl-L-lysine chloromethyl ketone (TLCK), 2 mM benzamidine, 0.5 mM PMSF, and 0.01 mg/ml leupeptin. The homogenates were centrifuged for 20 min at 17,000 ϫ g and the supernatants were retained for analysis. Approximately 50 g of protein from each tissue was separated by SDS-PAGE followed by transfer to nitrocellulose, which was probed with affinity purified anti-HAN11 antibodies.
Transient Transfection and Analysis of Expressed Proteins-COS-M9 cells were transiently transfected by using LipofectAMINE (Invitrogen). Briefly, 0.5-1 g of plasmid DNA per 6 l of LipofectAMINE was used to transfect cells in a 35-mm dish. Cells were grown for 2 days and frozen in liquid nitrogen in 0.2 ml of buffer A, containing 50 mM Tris-HCl, pH 7.8, 10 mM EDTA, 2 mM EGTA, 100 mM NaF, 1 mM dithiothreitol, 1 mM PMSF, 0.1 mM TLCK, or in buffer B that contained 50 mM Tris-HCl, pH 7.8, 100 mM NaCl, 1 mM dithiothreitol, and protease inhibitors. Cells frozen in buffer A or buffer B were used for glycogen synthase assay or to analyze for interactions between expressed proteins, respectively. After thawing, cells were homogenized and centrifuged at 14,000 ϫ g for 10 min. To analyze the interaction between S-tag HAN11 and FLAG-tagged DYRK isoforms, the supernatants were incubated with S protein-agarose (Novagen) in the presence of 0.1% Triton X-100. The precipitates were washed three times with buffer B containing 0.1% Triton X-100 and proteins were eluted with SDS-PAGE loading buffer. The supernatants, the pellet fractions, and the S protein-agarose precipitates were separated by SDS-PAGE and analyzed by Western blot.
Glycogen Synthase Assay-Glycogen synthase activity was determined in the pellet fractions of COS cell homogenates by measuring the incorporation of [ 14 C]glucose from UDP-[U-14 C]glucose into glycogen (22) as described previously (9). The Ϫ/ϩ glucose-6-P activity ratio was calculated after subtraction of endogenous glycogen synthase activity in COS cells (9).
Activities of recombinant GST-DYRK1A and GST-DYRK2 were measured by the inactivation of added recombinant glycogen synthase protein as monitored by the decrease in the Ϫ/ϩ glucose-6-P activity ratio as a function of time. Reaction mixtures (total volume 0.1 ml) were composed of 20 mM Tris-HCl, pH 7.4, 10 mM MgCl 2 , 1 mM dithiothreitol, 0.1 mM ATP, and 3 g of glycogen synthase. Reactions were started by adding 20 l of GST-DYRK1a, GST-DYRK2, or protein kinase dilution buffer (20 mM Tris-HCl, pH 7.4, 0.1 M NaCl, 1 mM dithiothreitol, 30% glycerol). After incubation at 30°C for the indicated times the 10-l aliquots were removed and mixed with 80 l of glycogen synthase assay buffer containing 20 mM EDTA (9) to measure glycogen synthase activity.
Analysis of Phosphorylation of Site 3a in Glycogen Synthase-Glycogen synthase was incubated in the presence of GST-DYRK1A and GST-DYRK2 under the same conditions as described for time course inactivation of glycogen synthase. At the indicated times, 18-l aliquots were removed, mixed with SDS-PAGE gel loading buffer, and analyzed by Western blot using phospho-specific antibodies that recognize phosphorylated site 3a (p-GS(S641) antibodies).
Activity Assay of Recombinant DYRK1A and DYRK2-Protein kinase activity was determined by measuring the incorporation of [ 32 P]phosphate into protein substrates. The substrates, which included purified rabbit skeletal muscle glycogen synthase, glycogen phosphorylase b, phosphorylase kinase, protein phosphatase inhibitor-2, and recombinant targeting subunit (R GL /GM) of the muscle-specific glycogen-associated type 1 protein phosphatase (exon 1), were kindly provided by Dr. To estimate endogenous phosphorylation in protein substrates, GST-DYRK1A and GST-DYRK2 were excluded from the reaction mixture. After incubation at 37°C for 10 min, reactions were terminated by addition of SDS-PAGE sample buffer and boiling for 5 min. Histones H1 and H2B were obtained from Roche. Individual caseins, ␣ s1 -casein, ␤-casein, and -casein, obtained from Dr. P. J. Roach, Indiana University, were originally a gift from Dr. Elizabeth W. Bingham, USDA Eastern Regional Research Center, Philadelphia, PA. Polypeptides were separated by SDS-PAGE, gels were stained with Coomassie Blue and dried. The phospholabeled proteins were detected by autoradiography, then excised from gels and the radioactivity incorporated was quantitated by scintillation counting.
Expression of HAN11 in Rat1 Fibroblasts-Stable clones of Rat1 fibroblasts expressing StagHAN11 were generated by transfection of cells with pcDNA-S-HAN11 using LipofectAMINE and selection of the clones with 0.5 mg/ml Geneticin (Invitrogen).

Design of Substrates for Purification of Glycogen Synthase
Kinases-To monitor 3a-kinase activity during purification, mutant forms of glycogen synthase were designed, expressed in E. coli, and purified. All the glycogen synthases are truncated at Lys 682 to avoid phosphorylation at sites 1a and 1b and additionally contain Ser/Ala substitutions at sites 2, 3c, 4, and 5. Enzyme truncated at Lys 682 but containing intact phosphorylation sites serves as ''wild type'' (Fig. 1). As shown previ- The amino acid sequence of the region containing phosphorylation sites (underlined) from 3a to 5 in the wild type enzyme is given. For the mutants, the amino acids in the positions corresponding to each phosphorylation site are indicated. Truncated wild type and the mutants of glycogen synthase are designated by the symbol ⌬. All mutants additionally contain Ser/Ala substitution at site 2, which is marked x. Truncated wild type glycogen synthase has site 2 intact. ously, truncation did not have any effect on the activity, expression in COS cells, or the susceptibility of the enzyme to proteolysis indicating that the mutation does not significantly affect the overall conformation of glycogen synthase (10). The only difference between the mutants 2,AA⌬, 2,SA⌬, and 2,AS⌬ is the availability of phosphorylation sites 3a and 3b (Fig. 1). The mutant 2,SA⌬ served for detection of 3a-kinase and mutants 2,AA⌬ and 2,AS⌬ served as negative controls.
Purification and Characterization of 3a-Kinase from Rabbit Skeletal Muscle-Initial experiments were designed to identify protein kinase activity in skeletal muscle capable of phosphorylating the glycogen synthase mutant 2,SA⌬ and not capable of phosphorylating the mutants 2,AA⌬ and 2,AS⌬. This protein kinase was designated 3a-kinase. Although 3a-kinase activity cannot be assayed in skeletal muscle extracts, activity toward 2,SA⌬ was detectable after initial ammonium sulfate precipitation and chromatography on phenyl-Sepharose. Proteins eluted from phenyl-Sepharose were chromatographed sequentially over multiple columns. Column fractions were collected and assayed for 3a-kinase activity, active fractions were pooled and prepared for loading in subsequent chromatographic steps. Proteins in pools and in column fractions were also analyzed by SDS-PAGE followed by silver staining. A typical purification is summarized in Table I. The 3a-kinase was purified about 3000fold with a recovery of 1% from the phenyl-Sepharose eluate. Purified 3a-kinase is very active toward 2,SA⌬ and wild type glycogen synthase but is incapable of phosphorylating 2,AA⌬ (Fig. 2). A trace phosphorylation of 2,AS⌬ was also observed. These data indicate a strong preference for the phosphorylation of site 3a. Incubation of the mutant 2,SA⌬ with 3a-kinase in the presence of [␥-32 P]ATP resulted in incorporation of ϳ0.7 mol of [ 32 P]phosphate/mol of glycogen synthase (not shown).
The final purified preparations contained two major species with apparent molecular weights of 39,000 and 68,000 (Fig. 3) on SDS-PAGE as well as several minor bands. Comparison of the protein composition with protein kinase activity in individual fractions of chromatography profiles suggested that the 54and 39-kDa polypeptides best correlated with the 3a-kinase activity. The apparent M r of 3a-kinase as determined by gelfiltration chromatography on Superose 12 was 138,000 (not shown). The distribution of 3a-kinase activity in the gel-filtration profile also correlated with the polypeptides of M r ϳ 54,000 and ϳ39,000. These results may indicate that the 3a-kinase is an oligomeric protein.
Identification of 3a-Kinase-Pooled fractions of 3a-kinase from the heparin-Sepharose purification step were concentrated and proteins were separated by SDS-PAGE followed by transfer to a polyvinylidene difluoride membrane. The Coomassie-stained band corresponding to the 39-kDa protein was excised and incubated with trypsin. The products of digestion were separated by high performance liquid chromatography and sequence was obtained from two tryptic fragments. Searching the protein sequence database with the two sequences GVYPDLLATSGDYLR and HLEHSTIIYEDPQHHPLIR revealed that both fragments belong to a recently identified pro-tein called HAN11 (23). The physiological role and function of this protein had not been determined. An interesting structural characteristic of HAN11 is the presence of WD40 repeats. However, there was no homology between HAN11 and the catalytic domain of protein kinases. Therefore, we hypothesized that HAN11 is a regulatory or targeting protein that associates with the catalytic subunit of 3a-kinase.
Further identification of polypeptides associated with 3akinase was performed by mass spectrometry. Pooled fractions of partially purified 3a-kinase from hydroxyapatite chromatography (lane 4, Fig. 3) were concentrated and subjected to SDS-PAGE. Coomassie Blue staining revealed 10 proteins of apparent M r from 39,000 (Fig. 4, band 10) to 175,000 (Fig. 4, band 1). Individual polypeptides were excised from the gel, digested  with trypsin, and analyzed by MALDI-TOF mass spectrometry. The experimentally determined tryptic peptide masses were then subjected to analysis using ProFound, and the identity of protein band 9 (Fig. 4) was confirmed as DYRK1A (probability ϭ 1.00 and Z score ϭ 2.27). The protein of band 10 was identified as HAN11 (probability ϭ 1.00 and Z score ϭ 2.43) confirming the results of protein sequencing by automated Edman degradation. The protein of band 9 corresponds to a polypeptide with apparent M r ϳ 54,000, whereas DYRK1A has M r ϳ 90,000. Therefore, the DYRK1A that was purified could be one of the alternatively spliced forms of DYRK1A that lacks COOH-terminal histidine repeats and a serine/threonine domain (24) or it might represent a proteolytic fragment of DYRK1A. Some other bands were identified as co-chaperone TPR2 (bands 6 and 7) and AMP-deaminase-1 (band 5).
Tissue Distribution of HAN11-The tissue distribution of HAN11 expression was analyzed by Northern blotting of poly(A) ϩ RNA isolated from several different tissues. Two distinct transcripts for HAN11 (2 and 8.2 kb) were detected in all tissues (Fig. 5A). The 2-kb transcript was particularly abundant in heart, placenta, skeletal muscle, and pancreas. The significance of the difference in transcript size is not yet known. Analysis of mouse proteins revealed relatively high levels of HAN11 expression in brain, ovary, and testis (Fig. 5B). Lower levels were detected in placenta, liver, skeletal muscle, kidney, and pancreas.
Protein Kinases from DYRK Family Interact with HAN11-To determine whether HAN11 interacts with DYRK1A and other members of the DYRK family (25), we co-expressed NH 2terminal FLAG-tagged mammalian DYRK isoforms, DYRK1A, DYRK1B, and DYRK2, with NH 2 -terminal S-tagged HAN11 in COSM9 cells. DYRK1A, DYRK1B, and DYRK2 were detected as polypeptides with M r ϳ 90,000, 78,000, and 61,000, respectively, which is consistent with the predicted sizes of the protein kinases (Fig. 6A). After incubation of the lysates of COS cells with S protein-agarose, the precipitated proteins were detected with anti-HAN11 and anti-FLAG antibodies. As shown in Fig. 6B, S-tag HAN11 protein was detected in the S protein-agarose precipitates from cells transfected with S-tag HAN11 or cells co-transfected with both S-tag HAN11 and the different isoforms of DYRK. With anti-FLAG antibody, only DYRK1A and DYRK1B, but not DYRK2, were detected in precipitates from cells co-transfected with S-tag HAN11 and DYRK (Fig. 6C). DYRK2 did not co-precipitate with S-tag HAN11 even with a higher level of expression than shown in Fig. 6A (not shown). These data demonstrate that DYRK1A and DYRK1B interact with HAN11 and that the interaction can occur in mammalian cells. To identify proteins that interact with HAN11, stable clones of Rat1 fibroblasts expressing HAN11, NH 2 -terminal fused with the S-tag polypeptide, were generated. One of the clones expresses an ϳ2.5-fold higher level of HAN11 fusion protein as compared with the endogenous HAN11 (Fig. 7A). Pull-down of the fusion protein using S protein-agarose followed by incubation of the precipitates with [␥-32 P]ATP demonstrated phosphorylation of the polypeptide with M r ϳ 90,000 (Fig. 7C). A protein of the same size was identified as DYRK1A by immunoblot analysis (Fig. 7B) indicating that endogenous DYRK1A interacts with HAN11 even when only modestly overexpressed. It is likely that the 32 Plabeled polypeptide of M r ϳ 90,000 (Fig. 7B) in the HAN11 precipitates is an autophosphorylated form of DYRK1A.
DYRK Family Protein Kinases Phosphorylate and Inactivate Glycogen Synthase-To determine whether DYRKs inactivate glycogen synthase by phosphorylation of site 3a, we expressed mutants of glycogen synthase with the three isoforms of DYRK individually in COS cells (Fig. 8). DYRK1A, DYRK1B, and DYRK2 were unable to inactivate glycogen synthase containing Ser/Ala substitutions at phosphorylation sites 2, 3a, 3b, 3c, 4, and 5 ( Fig. 8, 2,AA), which eliminate all the regulatory serine residues. Reinstating a serine residue at site 3a (2,SA; Fig. 8) resulted in inactivation of glycogen synthase indicating phosphorylation of this residue in COS cells. Expression of the mutant 2,SA with each isoform of DYRK further inactivated glycogen synthase suggesting a higher stoichiometry of phosphorylation at site 3a. Mutation of two serine residues, sites 2 and 3a (Fig. 8, 2,3a), led to a similar activation state of glycogen synthase as the mutant 2,SA. DYRK1A or DYRK1B did not inactivate the mutant 2,3a in COS cells (Fig. 8, A and B), indicating that both protein kinases phosphorylate site 3a but no other sites that affect glycogen synthase activity. However, DYRK2 further inactivated mutant 2,3a (Fig. 8C) presumably by phosphorylation of other COOH-terminal phosphorylation sites. To provide more evidence for phosphorylation of site 3a by DYRKs, two isoforms, DYRK1A and DYRK2, were expressed in E. coli as GST fusion proteins, purified, and incubated with the glycogen synthase mutant 2,SA⌬. Both protein kinases inactivated this form of glycogen synthase ( Fig. 9A) but had no effect on the activation state of the glycogen synthase mutant 2,AA⌬ (not shown). Inactivation of the mutant 2,SA⌬ was accomplished by phosphorylation of site 3a as detected with phosphorylation site-specific antibodies (Fig. 9B). Phosphorylation of glycogen synthase by DYRK1A and DYRK2 was compared with phosphorylation of some other proteins, which are involved in regulation of glycogen synthesis, as well as some commonly used generic substrates of protein kinases. Relative to glycogen synthase, both protein kinases exhibited very low activity with glycogen phosphorylase b, phosphorylase kinase, targeting subunit (R GL /GM) of the muscle-specific glycogen-associated type 1 protein phosphatase (exon 1), protein phosphatase inhibitor-2, histones, and caseins (Table II). Based on these results, glycogen synthase would appear to be a relatively specific substrate for DYRKs. DISCUSSION The activity of glycogen synthase is under the control of hormones, such as insulin, catecholamines, and glucagon (4 -6), and non-hormonal stimuli, such as the blood glucose level, amino acid availability, and exercise. The need to respond to multiple stimuli may necessitate the involvement of multiple signaling pathways to regulate enzyme activity and the rate of glycogen synthesis. Much of the control of glycogen synthase occurs via changes in the phosphorylation state of sites 2, 2a, 3a, and 3b (7-10), whereas phosphorylation of other sites has little or no effect on enzyme activity (9,10). It was first demonstrated that sites 3a and 3b were phosphorylated by the protein kinase GSK-3 (11)(12)(13). Studies with cells and with animals have also shown that known inhibitors of GSK-3 can affect glycogen synthesis (26,27). However, more recent studies indicated that alternative pathway(s) for the phosphorylation of these sites in glycogen synthase may operate (10,18).
Here we report that one alternative pathway for phosphorylation of site 3a may include one or more members of the DYRK family of protein kinases.
Mammalian DYRKs are a subfamily of mitogen-activated protein kinase-related protein kinases and were originally discovered on the basis of homology to the Saccharomyces cerevisiae Yak1 and Drosophila mini-brain kinases (25,28). DYRKs possess Ser/Thr phosphorylation activity as well as autophosphorylation activity on Tyr residue(s). Autophosphorylation at FIG. 6. Interaction of HAN11 and DYRK. COS cells were transiently transfected with 0.5 g of plasmid encoding S-tag HAN11 (S-HAN11) and 0.5 g of plasmid encoding FLAG-DYRK1A (DYRK1A), FLAG-DYRK1B (DYRK1B), or FLAG-DYRK2 (DYRK2). Pull-down was performed with S protein-agarose beads (Novagen). Immunoblot analysis of total cell lysates with FLAG antibody (A) to identify DYRK or with HAN11 antibody (B) to identify both HAN11 and Stag HAN11 was performed. Immunoblot analysis of pulled down proteins was performed with FLAG antibody (C) or HAN11 antibody (D). The numbers to the left indicate the molecular masses (kDa).

FIG. 7.
Interaction between HAN11 and DYRK1A in Rat-1 fibroblasts. Stable clones of Rat-1 cells transfected with empty plasmid (control) or with plasmid encoding S-tag HAN11 were homogenized and proteins were analyzed by immunoblot with HAN11 antibody (A). Proteins from the soluble fractions of the clones were pulled-down with S protein-agarose. One portion of the pulled-down proteins was analyzed by immunoblot with DYRK1A antibody (B). Another portion was incubated in the presence of [␥-32 P]ATP, the proteins were separated by SDS-PAGE, and an autoradiogram was prepared (C). a conserved YXY motif located in the activation loop between consensus kinase subdomains VII and VIII is the mechanism for activation of DYRKs (29). In the DYRK family, DYRK1A has been the best characterized to date. The DYRK1A gene maps to the critical region of the Down's syndrome locus (30) and causes learning and memory defects when overexpressed in transgenic mice (34), consistent with the gene dosage effect in Down's syndrome. Targeted disruption of the DYRK1A gene in mice led to general growth delay and death during midgestation suggesting a non-redundant role of DYRK1A (32). Mice heterozygous for the mutation (DYRK1A ϩ/Ϫ ) had a de-creased brain size in a region-specific manner (32). DYRK1A phosphorylates various substrates in vitro, including the signal transducer and activator of transcription 3 (STAT3) (33), the ⑀ subunit of eukaryotic initiation factor 2B, the microtubuleassociated protein tau (34), the transcription factor of the forkhead family FKHR (35), and dynamin (36) indicating that DYRK1A might participate in several biochemical pathways.  (Fig. 1) was incubated with recombinant DYRK1A (2 g/ml), DYRK2 (0.3 g/ml), or without protein kinase (control) for the indicated times. Samples were analyzed for glycogen synthase activity ratio (A), for phosphorylation of site 3a (B), which was performed by immunoblot with phosphorylation sitespecific antibodies, and for glycogen synthase protein (C), which was performed by immunoblot with antibodies raised against glycogen synthase. Although DYRK1A is formally a dual-specificity protein kinase because of its ability to autophosphorylate on tyrosine residue(s), it phosphorylates other known substrates only on serine/threonine residues, like GSK-3 and mitogen-activated protein kinase. Analysis of the phosphorylation of peptide substrates identified DYRK1A as a proline-directed kinase with a consensus recognition sequence Arg-Pro-Xxx-(Ser/Thr)-Pro, where Xxx is any amino acid (37). Interestingly, replacement of the ϩ1 Pro by Ala almost completely eliminates substrate phosphorylation by DYRKs, but Val here does allow phosphorylation, especially by DYRK2 (38). Site 3a in glycogen synthase is located in the sequence Arg-Pro-Ala-Ser(Site 3a)-Val (Fig. 1), which makes it a good target for DYRK-mediated phosphorylation. Although we have established the specificity for site 3a of the kinase, whether purified from muscle or through expression of DYRKs in cultured mammalian cells, we have not carefully analyzed whether DYRKs phosphorylate other sites in glycogen synthase. DYRK2-mediated inactivation of the glycogen synthase mutant 2,3a, but not 2,AA, in COS cells indicates that DYRK2 might phosphorylate site(s) other than 3a, such as 3b, 3c, 4, and/or 5. Phosphorylation of site 3b would directly inactivate glycogen synthase (9,10). Interestingly, Pro at Ϫ2 in substrates is an important requirement for DYRK1A, but not for DYRK2 (41). Therefore, it is possible that DYRK2 might phosphorylate site 4, which is located in sequence Arg-His-Ser-Ser(site4)-Pro (Fig. 1). Phosphorylation of site 4 would ''prime'' glycogen synthase for GSK-3 action, which in turn would phosphorylate sites 3c and 3b. Alternatively, the phosphorylation of sites other than 3a following expression of DYRK in COS cells could be indirect and mediated by a kinase activated by DYRK.
The possibility of involvement of DYRKs in the regulation of glycogen synthesis in skeletal muscle is consistent with data about tissue distribution of these kinases. DYRK1A is ubiquitously expressed, with higher levels in brain, heart, placenta, and skeletal muscle (24,39,40). The closely related family member, DYRK1B, is expressed in skeletal muscle, testes (21,41), and several types of cancer cells (41). Expression of DYRK2 was detected in many tissues, including skeletal muscle, with highest level in testes (42). DYRK1A and DYRK1B, but not DYRK2, contain a bipartite nuclear localization signal in the NH 2 -terminal domain. Most of a green fluorescent fusion protein of DYRK1A was found to accumulate in the nucleus of transfected COS-7 and HEK293 cells, whereas green fluorescent protein-DYRK2 was predominantly detected in the cytoplasm (21,43). In contrast to green fluorescent protein-DYRK1A, green fluorescent protein-DYRK1B was found in both the nucleus and the cytoplasm of COS cells (21). Endogenous DYRK1B was found predominantly in the cytoplasm of colon carcinoma cells (41). Therefore, it is possible that the localization of DYRKs in cells is determined by cell type and/or the level of expressed protein. An important question raised by this study is of the relative importance of DYRK1A and DYRK2 as physiological glycogen synthase kinases. Cellular localization of course could be an important factor in dictating whether a protein kinase can act on its substrate and hence the cytosolic predominance of DYRK2 might favor this enzyme as a candidate. However, even nuclear localization does not preclude a physiological role as there is a report that glycogen synthase may have a nuclear localization until glucose is available (44). This issue will be an important focus of future work.
We have demonstrated that two isoforms of DYRK, DYRK1A and DYRK1B, interact with the protein HAN11. HAN11 is a human homolog of the plant protein AN11 that is involved in the regulation of transcription of anthocyanin biosynthetic genes in Petunia hybrida (23). Another homolog of AN11, TTG1, regulates several developmental and biochemical pathways in Arabidopsis, including the formation of hairs on leaves, stems, and roots, and the production of seed mucilage and anthocyanin pigment (45). AN11, TTG1, and HAN11 belong to the class of tryptophan and aspartic acid (WD) repeat proteins. This family of proteins has been shown to play a role in numerous cellular functions including signal transduction, mRNA processing, gene regulation, vesicular trafficking, and regulation of the cell cycle (46 -48). The structure of the ␤-subunit of heterotrimeric G-proteins, one of the best characterized WD proteins, revealed that this class of proteins forms a ␤-propeller structure, which apparently creates a stable platform allowing simultaneous interaction with multiple proteins (49 -52). If HAN11 is also able to interact with multiple proteins, it is possible that it targets the DYRK1 proteins to their substrates. In plant cells, AN11 is located in the cytosolic fraction (23). Therefore, HAN11 might target DYRKs to cytosolic locations for regulation of specific cellular functions. Interestingly, Yak1p, the DYRK homolog in S. cerevisiae, rapidly shuttles between the nucleus and the cytoplasm in response to glucose indicating that, in yeast, compartmentalization of Yak1p is regulated (53). It is possible that HAN11 provides targeting of DYRK1A and DYRK1B to cellular locations that are involved in glycogen synthesis and degradation. However, co-expression of HAN11 and glycogen synthase in COS cells did not significantly change the activation state of glycogen synthase (results not shown). These negative results could be explained if COS cells contain significant amounts of endogenous HAN11 (Fig.  6B), which already saturates endogenous DYRK. Alternatively, HAN11 might not be involved in DYRK-mediated phosphorylation of glycogen synthase and, although co-purified with DYRK1A, could be involved in other aspects of DYRK function. Further studies are under way to address this issue.
In summary, the phosphorylation of the functionally important site 3a in glycogen synthase by a novel category of protein kinase may represent a completely new pathway for the regulation of glycogen synthase activity and glycogen synthesis. Little is known of the control of DYRK and it will be of considerable interest to identify what physiological stimuli, hormonal, metabolic or other, lie upstream of DYRKs. Inhibitors of DYRK kinases, like inhibitors of GSK-3, would have the potential to promote glycogen synthesis and perhaps act as hypoglycemic agents.