Identification of the autophosphorylation sites of the Xenopus laevis Pim-1 proto-oncogene-encoded protein kinase.

Pim-1 is an oncogene-encoded serine/threonine kinase expressed primarily in cells of the hematopoietic and germ line lineages. Previously identified only in mammals, pim-1 cDNA was cloned and sequenced from the African clawed frog Xenopus laevis. The coding region of Xenopus pim-1 encoded a protein of 324 residues, which exhibited 64% amino acid identity with the full-length human cognate. Xenopus Pim-1 was expressed in bacteria as a glutathione S-transferase (GST) fusion protein and in COS cells. Phosphoamino acid analysis revealed that recombinant Pim-1 autophosphorylated on serine and threonine and to a more limited extent on tyrosine. Electrospray ionization mass spectroscopy was undertaken to locate these phosphorylation sites, and the primary autophosphorylation site of GST-Pim-1 was identified as Ser-190 with Thr-205 and Ser-4 being minor sites. Ser-190, which immediately follows the high conserved Asp-Phe-Gly motif in catalytic subdomain VII, is also featured in more than 20 other protein kinases. To evaluate the importance of the Ser-190 site on the phosphotransferase activity of Pim-1, Ser-190 was mutated to either alanine or glutamic acid, and the constructs were expressed in bacteria as GST fusion proteins and in COS cells. These mutants confirmed that Ser-190 is a major autophosphorylation site of Pim-1 and indicated that phosphorylation of Pim-1 on the Ser-190 residue may serve to activate this kinase.

Pim-1 (1) is a member of the small family of known protooncogene-encoded serine/threonine protein kinases. The pim-1 gene was first discovered as a preferential proviral integration site in murine leukemia virus-induced T and B cell leukemias (2,3). Extensive transgenic studies have confirmed that pim-1 is an oncogene in mice (4 -6), but the only evidence that Pim-1 contributes to human cancers is that the protein and mRNA levels for this kinase are elevated in some human tumors and cell lines (7)(8)(9).
Although the normal functions of Pim-1 have been obscure, its mRNA and protein expression patterns indicate that it is involved in hematopoietic signal transduction and development of male germ cells (1,3,7,10,11). Expression and stability of pim-1 mRNA and protein is tightly regulated at different levels, implying that the kinase is functionally potent (7,9,(12)(13)(14). Pim-1 is induced by mitogens and growth factors, including GM-CSF, 1 G-CSF, IL-2, IL-3, IL-5, IL-6, IL-7, concanavalin A, phorbol esters, interferon-␥, Steel factor, and in response to T cell receptor cross-linking, and it may act as a cytoplasmic mediator in a signal transduction pathway (9,12,13,(15)(16)(17)(18). Up-regulation of Pim-1 by signaling through the GM-CSF receptor family requires the presence of the GM-CSF receptor membrane proximal domain and may involve the JAK2 tyrosine kinase (19 -24). STAT1 sites within the pim-1 gene support this hypothesis (18). During GM-CSF stimulation, Pim-1 is involved in a DNA synthetic pathway rather than an apoptotic inhibition pathway (24).
Relatively little is known concerning the post-translational regulation of Pim-1, even though it was identified as a protein kinase more than a decade ago (3). There has been no evidence (e.g. band shifts of Pim-1 during SDS-polyacrylamide gel electrophoresis) to support the phosphorylation of Pim-1 during signal transduction. Nor has the phosphotransferase activity of the kinase been shown to be affected by phosphorylation. Pim-1 can autophosphorylate, but the sites of autophosphorylation and the functional consequences of this event have not yet been determined (14,(25)(26)(27).
As Pim-1 phosphotransferase activity has not yet been shown to be regulated by upstream kinases, it is possible that an autophosphorylation event may modulate the activity of the kinase. One of the goals of this study was to identify the major autophosphorylation site in Pim-1 and to assess its role. Pim- 1 has not yet been examined from non-mammalian systems, nor has it been purified to homogeneity. In this study, we have cloned Pim-1 from Xenopus laevis and characterized a purified glutathione S-transferase (GST) fusion form of this kinase that was expressed in Escherichia coli. The recombinant GST-Pim-1 was highly active and exhibited strong autophosphorylating activity in vitro. Immobilized metal affinity chromatography (IMAC) and high pressure liquid chromatography (HPLC)-elec-trospray ionization-mass spectrometry (ESI-MS) were employed to identify the main autophosphorylation site as . Mutational analysis confirmed that Ser-190 was indeed the major site of autophosphorylation and indicated that phosphorylation of this residue serves to activate this kinase.

EXPERIMENTAL PROCEDURES
Materials-The Pim-1-CT and the Tel anti-Pim-1 antibodies were kind gifts from Dr. Michael Lilly (Seattle VA Hospital) and Drs. R. Amson and A. Telerman (CEPH, Paris, France), respectively. The anti-GST antibody was obtained from Molecular Probes, and the anti-phosphotyrosine monoclonal antibodies PY-20 and 4G10 were from Santa Cruz Biotechnology and Upstate Biotechnology Inc., respectively. The blotting grade affinity-purified goat anti-rabbit IgG (HϩL) alkaline phosphatase conjugate was from Bio-Rad, and the electroimmunoassay grade affinity-purified goat anti-mouse IgG (HϩL) alkaline phosphatase conjugate was obtained from Calbiochem. Pim-1 substrate peptide (AKRRRLSA) was produced on an Applied Biosystems 430A peptide synthesizer and purified in the laboratory of Dr. Ian Clark-Lewis (Biomedical Research Center, University of British Columbia). Phosphatase HPTP␤ was a kind gift from Ken Harder (Biomedical Research Center). The pGEX-2T vector was from Pharmacia Biotech Inc. The E. coli bacterial strain UT5600 as well as restriction enzymes BamHI and HindIII were from New England Biolabs. The isopropyl ␤-D-thiogalactopyranoside was from Fisher. The cAMP-dependent protein kinase inhibitor peptide, soybean trypsin inhibitor, and other reagents were from Sigma. All oligonucleotides were synthesized on an Applied Biosystems 392 DNA synthesizer.
Cloning of X. laevis Pim-1-Stage VI immature oocytes were isolated from the ovaries of mature female X. laevis (28) and washed with 1 ϫ Xenopus oocyte medium (29). Total RNA was isolated from these oocytes and used as a template to make cDNA (30). A Xenopus pim-1 probe was constructed using oligonucleotides based on the sequences of human and murine pim-1 sequences (31). Sense and antisense primers (5Ј-CTG ACC CGG GCT CGA GGC ICC IGG IAA (G/A)GA (G/A)AA (G/ A)GA (G/A)CC-3Ј and 5Ј-CTG ACC CGG GCT CGA GAT (C/T)TC (C/ T)TC (G/A)TC (G/A)TG TTC (G/A)AA IGG-3Ј, respectively) were expected to yield a 669-base pair product. PCRs were carried out for 29 cycles with an initiation temperature of 94°C for 35 s, annealing at 50°C for 90 s, and elongation for 73°C for 90 s. PCR products were subcloned into pBluescript, amplified, examined by restriction digest analysis, and the identity of the PCR clones was confirmed by sequencing, using a double template protocol as dictated by the Pharmacia kit. A Xenopus oocyte cDNA library was kindly supplied by Dr. Leonard Zon (Children's Hospital, Boston) and was screened with the amplified Xenopus PCR fragment by standard methods (30). Positive clones were amplified and sequenced using the standard dideoxy chain termination method (32). Sequence searches were performed with a software package developed by Dr. Allen Delaney (Biomedical Research Center).
Construction of Mutant Pim-1 Bacterial Expression Vectors Using PCR-The coding region of the X. laevis pim-1 was amplified from a cDNA clone by PCR using the pim-1 based oligonucleotides, 5Ј-CGA T-GG ATC CAT GCT TCT CTC TAA ATT CGG-3Ј and 5Ј-GAT CGA AT-T CCA GAC TCT CGT TGC TTG A-3Ј, which allowed directional, inframe insertion into the pGEX-2T vector. To create the kinase-inactive (K69A) Pim-1 mutant, PCR was used with the 5Ј sense primer and the antisense primer K1, 5Ј-CTC CTT AGC TAC GTG GCG CAC AGC GAC CGG CTG-3Ј (corresponding to amino acids 64 -74, with the lysine codon, TCC (nucleotides 205-207), changed to an alanine codon, GCG). To create Ser-190 mutants of Pim-1, sense primer PM1 (5Ј-CTG ATC GAT TTT GGC GCC GGG GCG CTA CTC-3Ј) was used to make a serine to alanine mutant (S190A) and the sense primer PM2 (5Ј-CTG ATC GAT TTT GGC GAA GGG GCG CTA CTC-3Ј) was used to make a serine to glutamic acid mutant (S190E) of Pim-1. Both sense primers were used with PM3 (5Ј-GAT CGA ATT CCA GAC TCT CGT TGC TTG A-3Ј) to amplify a 415-base pair fragment that was used to replace the ClaI/EcoRI fragment of WT Xenopus pim-1. PCR products were purified, digested, and ligated into appropriate restriction sites of pGEX-2T. The ligations were used to transform DH5a high competence cells (Life Technologies, Inc.), and positive clones were selected on the basis of ampicillin resistance. All clones were sequenced to confirm that the mutations were as expected.
Production of GST Fusion Proteins-The pGEX-2T vectors containing Xenopus WT and mutant pim-1 inserts were used to produce GST-Pim-1 fusion protein. In brief, UT5600 E. coli (New England Biolabs) was transfected with the pim-1-containing pGEX-2T vectors and grown in 2 ϫ YT medium (30) in the presence of 75 g/ml ampicillin. The expression of the fusion protein was induced with 200 M isopropyl ␤-D-thiogalactopyranoside for 3 h. The bacterial were pelleted by centrifugation at 2,000 ϫ g for 15 min, the pellet was washed in 25 ml of phosphate-buffered saline and resuspended in 10 ml of phosphatebuffered saline containing 1 mM EDTA, 0.1 mM EGTA, 0.2 mM phenylmethylsulfonyl fluoride, 1% Triton X-100, and protease inhibitors (10 g/ml) aprotinin, leupeptin, and soybean trypsin inhibitor. Lysozyme (200 g/ml) and DNase (50 mg/ml) were added to each pellet, followed by incubation on ice for 30 min. The lysate was centrifuged (12,000 ϫ g for 10 min at 4°C), and the supernatant was removed and added to glutathione-agarose beads and mixed gently on a rotator at 4°C for 30 min. The beads were then poured into a 1-cm-diameter column and washed with 10 volumes of STE buffer (50 mM Tris-HCl, pH 8.0, 4 mM EDTA, 150 mM NaCl, 0.1% Triton X-100), followed by 10 volumes of STEC buffer (50 mM Tris-HCl, pH 8.0, 200 mM NaCl, 6 mM CaCl 2 ). The fusion protein was eluted from the beads with 10 mM glutathione in STEC buffer and immediately frozen at Ϫ70°C.
Production of Anti-Pim-1 Antibodies-The affinity-purified rabbit polyclonal Pim-1 anti-peptide antibodies were prepared as described by Sanghera et al. (33) and are commercially available from Upstate Biotechnology Inc. Pim-1-III was based on the kinase subdomain III region of the human Pim-1 (amino acid residues 70 -91). Pim-1-NT was patterned after the amino terminus of human Pim-1 (amino acid residues 1-37). Anti-Pim-1-XI was raised against a peptide modeled on the catalytic subdomain XI of xPim-1 (residues 275-292). Rabbit polyclonal antiserum was produced against the full-length GST-xPim-1 bacterial fusion protein eluted from the glutathione beads after concentration. Rabbits were first injected with 250 g of GST-xPim-1 in incomplete Freund's adjuvant; 200 g was used for the second and third injections, and 100 g of the fusion protein was used for each successive injection. The rabbits were bled and the serum purified by protein G purification following the manufacturer's protocol (Pharmacia). The protein G-purified serum (1.5 ml) was incubated with rotation for 30 min at 4°C with GST beads (500 l) to remove any antibodies specific for GST. The beads were centrifuged and washed once with 500 l of phosphatebuffered saline (4°C), and the supernatants were combined, aliquoted, and frozen.
Creation of Pim-PEF1 Expression Vectors, Transfection, and COS Expression-The coding region of Xenopus WT pim-1 was amplified using PCR oligomutagenesis using oligonucleotides 5H6PM (5Ј-(P) AT-G GAA GAG GAA GAG GAA GAG CTT CTC TCT AAA TTC GGA TC-G -3Ј) and PM3Ј (5Ј-CAA AGC TTT ACA GAC TCT CGT TGC TTG A-GC-3Ј), which added a 5Ј tag of six glutamic acid residues on the Pim-1 protein. The PCR fragment was subcloned into the SmaI site of the PEF1 expression, vector and then the Pim-PEF1 vectors were transformed into DH5a/P3 bacteria. Transformed bacteria were grown on plates containing both ampicillin (100 g/ml) and tetracycline (10 g/ml). Clones were selected for the presence of an insert by digestion with SalI, the orientation of the inserts was determined by BglII digestion, and the ends of the WT and the mutants were sequenced to confirm that the ligation was as expected.
Positive clones were transfected into the COS cells and grown 72 h before harvesting. Cells were homogenized in buffer containing 10 mM MES, pH 6.0, 50 mM ␤-glycerophosphate, 2 mM MgCl 2 , 2 mM phenylmethylsulfonyl fluoride, 1 mM EGTA, 1 mM dithiothreitol, 100 M Na 3 VO 4 , 0.5% Triton X-100, and 10 g/ml each of leupeptin and aprotinin. The homogenates were sonicated then centrifuged at 12,000 ϫ g to separate the supernatant and pellets, which were frozen immediately. Samples of the homogenates, supernatants, and pellets were analyzed by Western blotting.
Glutamic acid-tagged Pim-1 proteins were partially purified by fractionation on Resource Q column chromatography. COS extract (2 mg) was applied to a 2-ml Resource Q column and eluted with a 0 -0.8 M NaCl gradient over 10 ml, in buffer containing 10 mM MOPS, pH 7.2, 25 mM ␤-glycerophosphate, 2 mM EDTA, 5 mM EGTA, 2 mM Na 3 VO 4 . Fractions of 0.25 ml were collected and analyzed for the presence of Pim-1 by assays with Pim-1 substrate peptide and Western blotting.
In Vitro Autophosphorylation of Expressed GST-xPim-1-Phosphorylations were performed in a final volume of 50 l, with 1-25 g of GST-xPim-1 in 5 l of STEC buffer, 250 M [␥-32 P]ATP (6 ϫ 10 3 cpm/pmol), 500 nM cAMP-dependent protein kinase inhibitor, and assay dilution buffer containing 10 mM MOPS, pH 7.2, 10 mM MgCl 2 , and 1.25 mM MnCl 2 . Preincubations were performed over ice, and the kinase reaction at 30°C for 30 min was started with the inclusion of the ATP and terminated by the addition of 50 l of 2 ϫ SDS-sample buffer. The sample was boiled for 5 min and clarified by centrifugation and the supernatant electrophoresed on a 10% SDS-polyacrylamide gel. In some experiments, the gels were stained to ensure that the amount of protein loaded per lane was consistent. The gels were dried, and radioactive bands were cut from the gel and counted by scintillation counting. Stoichiometry of autophosphorylation was determined by measuring the mol of ATP incorporated per mol of GST-xPim-1. In other experiments, the GST-xPim-1 was transferred from the SDS-polyacrylamide gel onto polyvinylidene difluoride membrane for Western blotting and autoradiography.
Phosphorylation of Synthetic Peptide Substrate by Expressed xPim-1-The peptide (AKRRRLSA) was determined previously to be an optimal substrate for Pim-1. 2 Pim-1 substrate peptide phosphorylations were performed in a final volume of 25 l, with 1 mg/ml peptide, 0.5 g of GST-xPim-1 in 5 l of STEC buffer, 50 M [␥-32 P]ATP (2,000 cpm/ pmol), 1.5 mM MnCl 2 or 25 mM MgCl 2 , 0.5 M cAMP-dependent protein kinase inhibitor peptide, and assay dilution buffer. All preincubations were performed on ice, and the kinase reactions at 30°C for 10 min were started with the addition of the ATP and terminated by application of 20 l of the reaction mixture onto a 1.5-cm 2 piece of Whatman P81 phosphocellulose paper. Filter papers were washed in phosphoric acid (1% v/v) and the radioactivity quantitated in an LKB Wallac 1410 scintillation counter.
Phosphoamino Acid Analysis and Tryptic Phosphopeptide Mapping of in Vitro Autophosphorylated xPim-1-32 P-Labeled GST-xPim-1 (by autophosphorylation) on a polyvinylidene difluoride membrane was visualized by autoradiography, excised from membrane, and chopped into 0.5-mm 2 pieces. For phosphoamino acid analysis, membrane-bound protein was digested in 300 l of constant boiling HCl at 105°C for 1.5 h. The acid was removed, and the membrane was washed briefly with dH 2 O to remove residual amino acids. Water and acid were removed by evaporation in a vacuum centrifuge, and amino acids were sequentially washed and vacuum dried. Amino acids were redissolved in water/acetic acid/pyridine buffer (94.5/5/0.5 v/v/v) containing 1 mg/ml each of phosphoserine, phosphothreonine, and phosphotyrosine standards. Phenol red (0.5% in buffer (w/v)) was spotted on the origin as a control. Approximately 2,000 cpm were spotted 2 cm from the bottom of a cellulose sheet (Kodak chromogram) and electrophoresed for 1.5 h at 750 volts until the phenol red spot had migrated 7-8 cm from the origin. The plate was air dried and sprayed with ninhydrin solution (0.25% in ethanol) and developed to visualize standards either by heating with a hairdryer or by baking in an oven at 90°C for 5 min. X-ray film was exposed to the cellulose sheet for 18 h and subsequently developed.
The full-length radiolabeled xPim-1 fusion protein was detected by autoradiography and excised from the gel for tryptic two-dimensional phosphopeptide mapping (34). Phosphoamino acid analysis was performed on resolved phosphopeptides that were extracted from the cellulose plate with two washes of 200 l of 20% acetonitrile (0.1% trifluoroacetic acid) and once with 200 l of 60% acetonitrile, with sonication in a water bath sonicator for 5 min between extractions to break up the cellulose.
IMAC-HPLC-ESI-MS Analysis of xPim-1 Tryptic Peptides-Two samples of expressed GST-xPim-1 were prepared, with an estimated 4.2 mg of fusion protein bound to 5 mg of glutathione beads/sample. One sample was autophosphorylated in vitro with cold ATP, and the other was assumed to be autophosphorylated or phosphorylated in vivo. Both samples were washed extensively with thrombin buffer and digested with 10 g of thrombin for 1 h at room temperature. After thrombinization, the slurries were poured into columns and the flow-through and eluate collected. The columns were washed with 2 volumes of thrombin buffer and the flow-through and eluate were concentrated in Centricon-10 tubes by sequential centrifugation and washes with 50 mM ammonium bicarbonate buffer. Once the volume of the Pim-1 was reduced to 100 l, 10 l of trypsin was added, and the sample was allowed to digest for 24 h at 37°C, with constant agitation.
Following tryptic digestion of the GST-xPim-1, the resultant phosphopeptides were analyzed by IMAC-HPLC-ESI-MS. Micro-IMAC-HPLC-ESI-MS was performed using instrumentation and protocols as detailed elsewhere (35).
Samples were subjected to ESI-MS analysis on a PE Sciex (Thornhill, Ontario) APIIII triple quadruple mass spectrometer equipped with a pneumatically assisted ESI source (ion spray). The mass to charge ratio (m/z) was scanned repetitively over the range of 300 -2,000. The output data from the mass spectrometer were analyzed with computer software provided with the machine. Mass spectra were displayed for observed peaks of ion detection events, and mass ranges from the data set could be extracted. Peptide masses were calculated based on com-puter matching of observed signals with the predicted m/z values for the various possible charge states of the same peptide and the theoretical fragmentation of the peptide sequence listing all possible fragments along with predicted charged mass values.

RESULTS
Cloning of Xenopus Pim-1-Degenerate oligonucleotides based on the human pim-1 sequence were successfully used to amplify part of the pim-1 coding region from X. laevis cDNA. This pim-1 PCR fragment was used as a probe to detect and identify the amphibian cognate from an Xenopus oocyte cDNA library. The full-length Xenopus pim-1 cDNA clones were approximately 2.7 kilobases in length, similar in size to that found in other species (25,26,31,36). A 969-base pair open reading frame was included in the 1,348 nucleotides sequenced, which specified a 323-amino acid protein with a molecular mass of 36,970 daltons (Fig. 1).
Xenopus Pim-1 (xPim-1) shares a high degree of amino acid sequence similarity with its mammalian counterparts (Fig. 1). The xPim-1 amino acid sequence has 86% overall homology and 65% overall identity with human Pim-1, with the regions of highest homology in the catalytic domain (71% identity). Although the xPim-1 NH 2 terminus shares 49% identity with human Pim-1, the two proteins only exhibit 15% identity within their COOH termini, and the amphibian Pim-1 is eight amino acids longer.
Bacterially Expressed Pim-1 Autophosphorylates in Vitro-xPim-1 was expressed as a bacterial fusion protein. Analysis by Western blotting demonstrated that the expected 549-amino acid, 63.3-kDa fusion protein was produced ( Fig. 2A). Of the seven different Pim-1 antibodies tested, those that were developed against xPim-1 amino acid sequences gave the strongest signals on Western blots. Bacterial expression of GST-xPim-1 yielded a relatively pure preparation with few degradation products as shown by Amido Black staining (Fig. 2B, lane 5), and the full-length fusion protein was the major species radiolabeled in an in vitro autophosphorylation reaction of GST-xPim-1 (Fig. 2B, lane 7). However, the stoichiometry of autophosphorylation of GST-xPim-1 was less than 0.01 mol/mol; as discussed later this reflected substantial phosphorylation of the GST-xPim-1 in E. coli prior to cell lysis. Only the full-length GST-xPim-1 was analyzed in detail because thrombin cleavage yielded poor recoveries of phosphotransferase activity (Ͼ3%) and xPim-1 protein (data not shown).
Phosphoamino acid analysis revealed that GST-xPim-1 autophosphorylates primarily on serine and threonine residues and to a minor extent on tyrosine residues (Fig. 2C). In vitro autophosphorylation of GST-Pim-1 on tyrosine was also demonstrated by detection by anti-phosphotyrosine antibodies on a Western blot (Fig. 2, lanes 9 and 10). Likewise, we observed that human GST-Pim-1 also autophosphorylates on phosphotyrosine by phosphoamino acid analysis and Western blotting with anti-phosphotyrosine antibodies (data not shown). The autophosphorylation of Pim-1 on tyrosine was unexpected because the most recent studies have reported that the Pim-1 autophosphorylated strictly on serine and threonine residues (14,26,27,37).
A kinase-inactive mutant of Xenopus GST-Pim-1 was created as a negative control to ensure that contaminating or copurifying proteins did not contribute to kinase activity. The kinaseinactive Pim-1 mutant was created by changing Lys-69, a residue required for ATP binding, to an alanine residue. The kinase-inactive Pim-1 mutant failed to immunoreact with antiphosphotyrosine antibodies (Fig. 2B, lane 4). Furthermore, it was not radiolabeled in an autophosphorylation assay, even though a faintly radiolabeled 70-kDa protein was present in the kinase-inactive Pim-1 mutant preparation (Fig. 2B, lane 8); as this protein was also detected by the anti-GST-xPim-1 antibody, it was probably part of the original immunizing preparation. Although this 70-kDa protein possessed autokinase activity, it did not phosphorylate the kinase-inactive GST-Pim-1. This may be the product of the bacterial dnaK gene, a close homolog of the eukaryotic heat-shock protein family that increases in response to stress and may bind preferentially to abnormal (expressed) proteins along with other heat-shock proteins (grpE, La) enhancing susceptibility to cellular proteases (38 -40).
Two-dimensional Phosphopeptide Analysis of xPim-1 Autophosphorylation Sites-Two-dimensional separation of tryptic phosphopeptides produced from in vitro autophosphorylated GST-xPim-1 yielded two strongly phosphorylated peptides (spots 1 and 2), as well as a number of moderate and weakly phosphorylated peptides (Fig. 3A). Each spot was numbered (Fig. 3B) for reference. The addition of increasing amounts of trypsin caused the relative strength of some spots to change, indicating that some of these peptides were the result of incomplete tryptic cleavage. None of the minor spots disappeared completely even when higher amounts of trypsin were used. Dephosphorylation of GST-xPim-1 with phosphatases prior in vitro autophosphorylation did not result in any significant changes in the relative intensities of the different spots.
After extraction from the TLC plate, peptides identified by two-dimensional phosphopeptide analysis were subject to phosphoamino acid analysis. The results of this experiment are summarized in Fig. 3B. Some samples did not yield enough radioactivity to warrant further analysis by IMAC-HPLC-ESI-MS.
Identification of Isolated Tryptic Phosphopeptides by IMAC-

Pim-1 Kinase Autophosphorylation
HPLC-ESI-MS-Phosphorylated peptides were selectively retained by the IMAC column and eluted from the matrix with sodium phosphate buffer and were subjected to subsequent analysis by HPLC-ESI-MS. A total ion chromatogram for sample 7 is shown in Fig. 4, which reveals the presence of two major charged species. Similar analyses were performed on the rest of the peptides. Although each sample should ideally contain only one peptide, additional peptides may be present because of comigration with the phosphopeptide of interest on the total ion chromatogram or may also be present because of noncovalent interaction with the phosphopeptide of interest. Peaks that corresponded to tryptic peptides of GST-Pim-1 were sequenced by the sequential cleavage of the NH 2 terminus during MS. Samples not yielding interpretable results are not described.
A third site of xPim-1 autophosphorylation was indicated as Ser-4, in a tryptic peptide corresponding to the last two residues of GST and the first 5 residues of Pim-1 (GSMLLSK) (data not shown). This sample also contained a peptide and peptide fragments corresponding to amino acids 56 -60 of GST, PYYID. Although this second peptide was not phosphorylated, it may have been retained on the IMAC column by interaction with the other bound peptides.
An additional phosphorylated peptide was identified in the GST portion of the fusion protein. The mass data analysis of this pinpointed Thr-17 as being the phosphorylated residue in that peptide (IKGLVQPTRLL). This phosphorylation site was unexpected, as preliminary experiments demonstrated that free GST was not phosphorylated by GST-xPim-1 (data not shown). The identified autophosphorylation sites of xPim-1 are shown in Fig. 1 as boxed residues. GST-xPim-1 was analyzed after extraction and purification from the bacteria without in vitro autophosphorylation. GST-xPim-1 was thrombinized to remove the GST, trypsinized, then the entire sample was loaded onto the IMAC column and ana- lyzed by liquid phase chromatography mass spectrometry. Phosphopeptides containing the phosphorylated Ser-190 site (m/z 1213) as well as the dephosphorylated form (m/z 1133) were identified. This indicated that the GST-xPim-1 was already substantially autophosphorylated in the bacteria. The exact amount of phosphorylation by the time of purification could not be established as the liquid phase chromatography mass spectrometry treatment causes partial removal of the phosphate group, but more than 50% of GST-xPim-1 was estimated to be phosphorylated on Ser-190 (data not shown).
Site-directed Mutagenesis of xPim-1 Autophosphorylation Site and Analysis of Mutants-In vitro autophosphorylation experiments following phosphatase treatment were performed to determine if differences in phosphotransferase activity toward exogenous substrates existed between the dephosphorylated and autophosphorylated forms of GST-xPim-1. The phosphatases tested included broad specificity acid and alkaline phosphatases and a broad specificity, tyrosine-specific phosphatase corresponding to the intracellular domain of HPTP␤ (data not shown). However, the phosphotransferase activities of all samples of phosphatase-treated and autophosphorylated GST-xPim-1 toward the Pim-1 substrate peptide AKRRRLSA were similar. It was possible that the various phosphatases that were used may not have dephosphorylated residues that affected the phosphotransferase activity of Pim-1. In partial support of this, no band shifts in size of GST-xPim-1 were observed on an SDS-polyacrylamide gel after phosphatase treatment (data not shown).
To evaluate directly the importance of the major autophosphorylation site in xPim-1, PCR mutagenesis was used to construct two xPim-1 mutants by changing the Ser-190 residue to alanine (S190A) and to glutamic acid (S190E). Alanine, a nonphosphorylatable residue, mimicked the dephosphorylated state of the Pim-1, and glutamic acid, a charged acidic residue, was used in an attempt to imitate phosphorylation at this site. The mutants were expressed as bacterial fusion proteins, and products of the expected size were immunodetected with Xenopus Pim-1 antibodies by Western blotting analysis (Fig. 6A). An autoradiogram of in vitro autophosphorylated WT and mutant GST-xPim-1 species (Fig. 6B) demonstrated that the autophosphorylation of both mutants was reduced very significantly compared with the WT Pim-1. Although phosphoamino acid analysis was attempted repeatedly to determine if a proportional amount of phosphoserine was reduced compared with phosphothreonine, autophosphorylation of the mutants did not allow enough isotope to be incorporated into the protein to allow analysis by this method.
The specific activity of WT GST-xPim-1 autophosphorylation was 107 pmol/min/mg, whereas the specific activities for autophosphorylation of the S190A and S190E mutants were estimated to be 4.9 and 5.2 pmol/min/mg, respectively. As the degree of autophosphorylation of the mutants was greatly re-duced in these mutants, Ser-190 was confirmed as a major site of GST-xPim-1 autophosphorylation.
GST-xPim-1 Autophosphorylation at the Ser-190 Site Appears to Be Activating-To assess whether the exogenous phosphotransferase activities of the Ser-190 mutants were different from the WT GST-xPim-1, kinase assays were performed using the Pim-1 substrate peptide AKRRRLSA. To ensure that the effects observed did not reflect the bacterial expression system, the WT xPim-1 and S190E mutant were also expressed in COS cells. The K m and V max values of the mutants for AKRRRLSA peptide phosphorylation are shown in Table I. The apparent K m values were similar in the S190E and WT forms of GST-Pim-1, indicating that the replacement of a serine to a glutamic acid did not change the affinity of the enzyme for the substrate peptide. The K m value for the S190A mutant was lower than with the WT, indicating that this mutant had a higher affinity for the substrate peptide. However, the V max values for both S190A and S190E were similar and 7-fold lower than the V max value of the WT. Since Pim-1 is highly phosphorylated at Ser-190 in vivo, these findings indicated that phosphorylation at this site may activate Pim-1. DISCUSSION xPim-1 represents the first non-mammalian Pim-1 homolog to be characterized. In addition to high homology with the mammalian cognates, the full-length bacterially expressed frog GST-xPim-1 fusion protein was catalytically more active and was produced in much higher amounts with less degradation products than with human GST-Pim-1 constructs. 3 The identification of the main sites of autophosphorylation of xPim-1 was complicated by the fact that the enzyme was substantially phosphorylated as it was obtained from the bacteria. However, IMAC-HPLC-ESI-MS permitted assignment of Ser-190 as a principal site of autophosphorylation in vivo. Sitedirected mutagenesis confirmed these findings and indicated that phosphorylation of Ser-190 may serve to further activate the kinase. The introduction of a glutamic acid residue in place of Ser-190 was unable to mimic the effect of phosphorylation. It could be argued that a free hydroxyl group as in Ser-190 might be important for the catalytic activity of Pim-1. This might account for why the S190A and S190E mutants were 7-fold less active as kinases. However, in most GST-Pim-1 molecules from the bacteria, where the enzyme was highly active, it was already phosphorylated.
The other autophosphorylation sites of Pim-1 identified using IMAC-HPLC-ESI-MS analysis of tryptic phosphopeptides were Ser-4 and Thr-205. These phosphorylation sites are also conserved in all other known Pim-1 homologs, but it is unclear if phosphorylation of these sites has any physiological importance. In contrast to GST-xPim-1, the majority of autophospho-

Pim-1 Kinase Autophosphorylation
rylation of human GST-Pim-1 is located on threonine residues by phosphoamino acid analysis, so it is possible that the Thr-205 site may be more important in the human enzyme (data not shown). Both the Ser-190 and the Thr-205 residues are located within the "activation segment" of the kinase, so phosphorylation of these resides might be expected to induce conformational changes in the protein that may serve to modulate its kinase activity (for review, see Ref. 41). The Ser-190 site is easy to identify in other unrelated protein kinases, as it immediately follows the conserved DFG motif in catalytic subdomain VII (42). Sequence comparisons of catalytic domain VII indicate that in general, most other kinases contain hydrophobic methionine or leucine residues at this site. However, about 20 known kinases feature a serine residue at this location including ASFV, kinase from African swine fever virus (43); DM, myotonic dystrophy kinase (44); GSK3␣/␤, rat glycogen synthase kinase-3 (45); sgg/zw 3 , kinase encoded by Drosophila segment polarity genes (46); MCK1, kinase encoded by a Saccharomyces cerevisiae meiotic induction gene (47); MDS1, a GSK3 homolog encoded by a S. cerevisiae gene that acts as a suppressor of mck1 mutants (48); ASK␣/␥, Arabidopsis thaliana GSK3 homologs (49); KNS1, a nonessential S. cerevisiae protein kinase (50); YAK1, a S. cerevisiae kinase downstream to the Ras/cAMP pathway (51); PST-K1, a distant relative of c-Mos (accession number L05668); Elm-1, a S. cerevisiae kinase involved in a differentiation pathway induced by nitrogen starvation (52); and Clk, a human kinase with homology to cdc2 (53). These kinases conserve the following (L/I/V/A)K(L/I/V)X-DFGS(A/C) phosphorylation site motif, where X generally corresponds to a cysteine or isoleucine residue. This site has not previously been shown to be phosphorylated in protein kinases, and it represents a new location for regulation.
The Thr-205 site is a more familiar site of phosphorylation in protein kinases and is located seven residues upstream of the conserved APE motif in catalytic subdomain VIII. Other kinases that possess activating phosphorylation sites just before this location include cyclin-dependent kinases, mitogen-activated kinases, MAPKAPK-2, glycogen synthase kinase-3, Src, Rsk, and cAMP-dependent protein kinase (for review, see Ref. 54). Although many of these sites may be autophosphorylated, they are generally targeted by activating protein kinases that operate upstream in signaling pathways.
Although the immunodetection data and phosphoamino acid analysis indicated that GST-Pim-1 autophosphorylated on tyrosine residues, we were unable to identify unequivocally a tyrosine phosphorylation site by IMAC-HPLC-ESI-MS. The presence of tyrosine autophosphorylation sites was examined using the novel approach of identifying peaks that were shifted in mass and retention time after treatment with the tyrosinespecific phosphatase HPTP␤. Although we were able to observe peptides that shifted in liquid phase chromatography mass spectrometry retention time after HPTP␤ treatment, these peptides were difficult to relate to the xPim-1. A tentative site, Tyr-133, was suggested by this method, but we did not confirm this result by site-directed mutagenesis as this residue was not well conserved in Pim-1 from other species.
The tyrosine at amino acid residue 198 in xPim-1 corresponded to Tyr-416 in Src, which is a conserved phosphorylation site in all tyrosine kinases (41,55,56) and led to the initial classification of Pim-1 as a tyrosine kinase (25,56). In the search for tyrosine phosphorylation sites, this residue was not identified as being phosphorylated in GST-Pim-1. The lack of detectable autophosphorylation at Tyr-198 does not preclude the possibility that it may be subject to phosphorylation and regulation by other protein kinases.
It is very likely that phosphorylation is not the sole method of post-transcriptional regulation of Pim-1 function. Many other kinases in the cell are part of dynamic, multiprotein complexes. Pim-1 is a small 35-kDa protein that consists of a catalytic domain without obvious autoregulatory or association domains, but it may still complex with other proteins. An imperfect amphipathic ␣-helix motif in the COOH terminus of Pim-1, similar to that of cAMP-dependent protein kinase (57) and as of yet without a defined function, could be involved in regulatory protein-protein interactions. Alternatively, phosphorylation of Pim-1 at its NH 2 terminus may also regulate the kinase indirectly. For example, autophosphorylation of the Mos oncogene-encoded serine/threonine protein kinase near its NH 2 terminus at Ser-3 apparently stabilizes the kinase by preventing its ubiquitination and degradation (58,59).