Components of a Calmodulin-dependent Protein Kinase Cascade

Ca2+/calmodulin-dependent protein kinases I and IV (CaMKI and CaMKIV, respectively) require phosphorylation on an equivalent single Thr in the activation loop of subdomain VIII for maximal activity. Two distinct CaMKI/IV kinases, CaMKKα and CaMKKβ, were purified from rat brain and partially sequenced (Edelman, A. M., Mitchelhill, K., Selbert, M. A., Anderson, K. A., Hook, S. S., Stapleton, D., Goldstein, E. G., Means, A. R., and Kemp, B. E. (1996)J. Biol. Chem. 271, 10806–10810). We report here the cloning and sequencing of cDNAs for human and rat CaMKKβ, tissue and regional brain localization of CaMKKβ protein, and mRNA and functional characterization of recombinant CaMKKβ in vitro and in Jurkat T cells. The sequences of human and rat CaMKKβ demonstrate 65% identity and 80% similarity with CaMKKα and 30–40% identity with CaMKI and CaMKIV themselves. CaMKKβ is broadly distributed among rat tissues with highest levels in CaMKIV-expressing tissues such as brain, thymus, spleen, and testis. In brain, CaMKKβ tracks more closely with CaMKIV than does CaMKKα. Bacterially expressed CaMKKβ undergoes intramolecular autophosphorylation, is regulated by Ca2+/CaM, and phosphorylates CaMKI and CaMKIV on Thr177 and Thr200, respectively. CaMKKβ activates both CaMKI and CaMKIV when coexpressed in Jurkat T cells as judged by phosphorylated cAMP response element-binding protein-dependent reporter gene expression. CaMKKβ activity is enhanced by elevation of intracellular Ca2+, although substantial activity is observed at the resting Ca2+ concentration. The strict Ca2+ requirement of CaMKIV-dependent phosphorylation of cAMP response element-binding protein, is therefore controlled at the level of CaMKIV rather than CaMKK.

recently discovered enzyme (1-3) that enhances the activities of the multifunctional calmodulin-dependent protein kinases types I and IV (CaMKI and CaMKIV, respectively) by ϳ20 -50fold. Activation is accomplished through selective phosphorylation of an equivalently positioned Thr in the activation loop of CaMKI and CaMKIV, Thr 177 and Thr 196 /Thr 200 , 2 respectively (4,5). To date, CaMKI and CaMKIV are the only CaM kinases known to require phosphorylation by an activating CaMKK, in addition to Ca 2ϩ /CaM, for maximal activity. However, in light of the differing subcellular and tissue distributions (4, 6 -13) and substrate specificities (14) of CaM kinases I and IV, the ultimate physiological targets of these Ca 2ϩ -dependent protein kinase cascades are likely to be divergent.
CaMKIV is expressed primarily in the nuclei of cells in thymus, testes, and subanatomical regions of brain, where it is thought to play a role in mediating Ca 2ϩ -regulated transcription (reviewed in Ref. 15). It has been shown to stimulate CREB, ATF-1, and SRF-mediated transcription in transient transfection systems (16 -20). Activation by CaMKIV occurs via direct phosphorylation of the activating serines of these transcription factors, Ser 133 (CREB), Ser 63 (ATF-1), and Ser 103 (SRF), respectively (17)(18)(19)21). CaMKIV has been suggested to be the physiologically relevant CREB kinase required for the delayed phase of long term potentiation activated in response to synaptic stimulation in hippocampal neurons (22). Evidence that CaMKIV can function as the CREB kinase required for T cell activation has also been described (23). CaMKIV has been shown to induce transcription from a CREB/AP1-like binding element of the viral BZLF1 promoter in Epstein-Barr virusinfected B cells, to promote transition from latency to the lytic phase (24). In addition, CaMKIV has been suggested to stimulate transcription by mediating Ca 2ϩ -dependent activation of the mitogen-activated protein kinase pathway in NG108 cells (25).
Specific cellular roles for CaMKI are less clear. CaMKI is expressed throughout the brain and in most nonneuronal tissues (4, 6 -8) and will phosphorylate in vitro multiple peptides (26) as well as proteins including synapsins I and II (27)(28)(29)(30), cystic fibrosis transmembrane conductance regulator (31), CREB (19,32), and ATF-1 (19). CaMKI is capable of stimulating transcription by phosphorylating the activating serines of CREB and ATF-1 in transient transfection systems (19), although endogenous CaMKI has been detected only cytoplasmically where access to these nuclear transcription factors would not be expected. The expression pattern and relatively broad substrate specificity of CaMKI suggest that this enzyme is likely to have multiple cellular roles.
By analogy to the mitogen-activated protein kinase pathway, in which transient phosphorylation and activation of one kinase by an upstream kinase kinase has been shown to be the physiological mechanism of signal transduction, it has been proposed that there are CaM kinase cascades in cells. Consistent with this, a rapid, transient phosphorylation and activation of endogenous CaMKIV in response to stimuli that elevate intracellular Ca 2ϩ has been demonstrated in Jurkat cells (33,34). Evidence for the involvement of CaMKK was obtained from transient transfection studies in which overexpressed wild type CaMKIV, but not the CaMKIV T200A mutant, could be activated similarly to the endogenous enzyme (35). That CaMKK␣ can mediate the activation of CaMKIV in vivo has been demonstrated in COS and NG108 cells overexpressing this isoform (25,36). Rapid CaMKK-catalyzed phosphorylation and activation of CaMKI in PC12 cells in response to increased intracellular Ca 2ϩ induced by depolarization or ionomycin was reported (37). It has not yet been determined whether signals other than Ca 2ϩ participate in the CaM kinase cascade, although it was recently reported that cAMP is capable of downregulating this pathway (38).
The purification and partial amino acid sequencing of two distinct CaMKI/IV kinases (CaMKK␣ and CaMKK␤) from rat brain demonstrated that CaMKK comprised a family of activating kinases (39). The existence of multiple CaMKKs with multiple substrates suggests a complexity of Ca 2ϩ /CaM-regulated signaling cascades. Determining the functional uniqueness of the different activating kinases represents an important question to be addressed, although as of yet, in vitro comparison of CaMKK␣ and CaMKK␤ purified from brain has revealed no differences with respect to specific activity, substrate specificity, or regulation by Ca 2ϩ /CaM (39). The partial amino acid sequence data for rat brain CaMKK␣ (39) indicated identity with the deduced sequence of a cloned cDNA for a rat brain CaMKK (40). We report here the cloning of cDNAs for rat and human CaMKK␤ and differential tissue and brain regional distributions of CaMKK␣ and CaMKK␤. We further characterize the ability of recombinant CaMKK␤ to activate CaMKI and CaMKIV through phosphorylation of Thr 177 and Thr 200 , respectively, as well as investigate the basis of control by Ca 2ϩ / CaM of this phosphorylation event, both in vitro and in Jurkat T cells.

EXPERIMENTAL PROCEDURES
cDNA Cloning-A human partial cDNA clone isolated by the IMAGE Consortium (expressed sequence tag accession number R56818), was previously shown to be homologous to the C-terminal portion of purified rat brain CaMKK␤ (39). Oligonucleotides corresponding to R56818 bp 61-80 (5Ј-TGCTGGACAAGAACCCCGAG-3Ј) and 391-368 (5Ј-CGT-TACCTTGAAGCTCAAAAC-3Ј) were used as primers to amplify by PCR a 330-bp product from first strand cDNA that had been synthesized from human brain total RNA. The PCR product was confirmed by sequencing to encode CaMKK␤ and radiolabeled with [ 32 P]dCTP using Klenow DNA polymerase by random priming (41). The radiolabeled, amplified fragment was used as a probe to screen under low stringency conditions, a ZAPII directional rat hippocampus/cerebral cortex cDNA library constructed and provided by N. Nakanishi (Harvard Medical School). Out of 1 ϫ 10 6 plaques, 15 positive clones were identified. A 1.2-kilobase pair clone, R15, was found to encode residue 302 to the end of the coding region and included 380 bp of 3Ј noncoding sequence. A 1.0-kilobase pair clone, R9, encoded residues 144 -485. The remaining 13 clones were redundant with R15 and R9. The missing 5Ј sequence of recombinant CaMKK␤ was obtained using the Life Technologies, Inc. system of rapid amplification of 5Ј cDNA ends, according to the manufacturer's instructions. Briefly, rat brain total RNA isolated by the Ultraspec (Biotecx) method was reverse transcribed with a gene-specific primer encoding recombinant CaMKK␤ residues EIAILKKL (bp 724 -704; 5Ј-GCTTCTTGAGGATGGCAATTT-3Ј). After RNA digestion with RNase H, tailing of the first strand cDNA 3Ј-end was carried out with terminal deoxynucleotidyl transferase and dCTP. Polymerase chain reaction was then performed with poly(dC)-tailed first stand cDNA template, a nested primer corresponding to recombinant CaMKK␤ residues SKKKLIR (bp 604 -584; 5Ј-GGATCAAGCTTCTTTTTGGAC-3Ј) and an anchor primer (5Ј-CUACUACUACUAGGCCACGCGTCGACT-AGTACGGGIIGGGIIGGGIIG-3Ј). This yielded a 522-bp product coding for recombinant CaMKK␤ residues 27-201. A second round of 5Ј rapid amplification of 5Ј cDNA ends was carried out following the procedure described above, except that the first stand synthesis primer corresponded to recombinant CaMKKb residues HVSITGF (bp 464 -444; 5Ј-AAACCCGTGATGGAGACGTGG-3Ј), and the nested primer corresponded to residues MNGRCI (bp 362-345; 5Ј-ATGCAGC-GTCCATTCATG-3Ј). This yielded a 454-bp product that encoded residues 1-121 and 92 bp of 5Ј noncoding sequence. A human cortical brain cDNA library provided by A. Roses (Glaxo Wellcome, Research Triangle Park, NC) (42) was also screened. A 2.2-kilobase pair clone, 9A, was found to encode residue 173 to the end of the coding region and 887 bp of 3Ј noncoding sequence including a poly(A)-tail. The 0.7-kilobase clone, 6.3, encoded residues 1-185 and 113 bp of 5Ј noncoding sequence.
DNA Sequencing-Chain termination sequencing was performed with the Sequenase version 2.0 DNA Sequencing kit from U.S. Biochemical Corp. and with the Thermo Sequenase Radiolabeled Terminator Cycle Sequencing kit from Amersham Life Science.
Antibody Production and Purification-Oligopeptides (custom synthesized by Research Genetics, Inc.) were based on regions that were non-conserved between CaMKK␤ and CaMKK␣, an N-terminal region of CaMKK␤ (amino acid residues 28 -49), and the C terminus of CaMKK␣ (residues 488 -505). Keyhole limpet hemocyanin-coupled peptides were injected into rabbits, and antisera were harvested by standard immunological techniques. Antibodies were affinity-purified as follows. Non-keyhole limpet hemocyanin versions of each of the respective immunizing peptides were coupled, using 1 mg of peptide/ml resin, to activated CH-Sepharose 4B (Amersham Pharmacia Biotech) as per the manufacturer's instructions. The resin was packed into a 1.5 ϫ 1.6-cm column equilibrated in 10 mM Tris, pH 7.5, 50 mM NaCl (column buffer). Antibodies were precipitated with 50% (NH 4 ) 2 SO 4 and then resuspended in and dialyzed against column buffer and chromatographed on the affinity column. The column was washed with column buffer and then with 10 mM Tris, pH 7.5, 500 mM NaCl (to remove nonspecifically bound proteins). The antibodies were eluted with 100 mM glycine, pH 2.5, immediately neutralized with 1 M Tris, pH 8.0, and dialyzed against 10 mM Tris, pH 7.5, 100 mM NaCl. In the case of CaMKK␤, in addition to the affinity chromatography step, microtubule-Sepharose was employed as a negative chromatographic step to remove a small subpopulation of contaminating anti-tubulin antibodies. To evaluate the specificities of the respective CaMKK␣ and CaMKK␤ antibodies, the following control experiment was performed. CaMKK␣ and CaMKK␤ were purified from rat brain by the procedures described in detail by Edelman, et al. (39). The purified enzymes were subjected to immunoblotting as described below using both sets of affinity-purified antibodies. Both anti-CaMKK␣ and anti-CaMKK␤ were found to be completely specific for their respective isoforms. 3 Immunoblotting-Freshly dissected rat tissues were frozen on dry ice and homogenized in a buffer containing 50 mM Tris, pH 8.0, 150 mM NaCl, 0.1 mM EGTA, 1% Nonidet P40 (Nonidet P-40), 1 mM phenylmethylsulfonyl fluoride, 10 mg/liter aprotinin, and 10 mg/liter leupeptin. Homogenates were centrifuged at 49,000 rpm (100,000 ϫ g) for 30 min in a Beckman Optima TLX ultracentrifuge. The supernatants (100 g of protein/lane) were electrophoresed on a 7.5% acrylamide SDS gel and electrophoretically transferred at 100 V for 1.5 h to a polyvinylidene difluoride membrane. The membrane was blocked overnight at 4°C in 5% nonfat dry milk/PBS, rinsed with deionized water, and incubated for 2 h at room temperature in 1% bovine serum albumin/PBS containing primary antibodies, either anti-CaMKK␤ at a dilution of 1:40 or anti-CaMKK␣ at a dilution of 1:4000. The membrane was rinsed three times with PBS, incubated in PBS for 5 min, and then rinsed an additional two or three times with PBS. It was then incubated for 90 min at room temperature in 5% bovine serum albumin/PBS containing 0.6 Ci/ml 125 I-labeled anti-rabbit IgG. Following washing 3 ϫ 20 min with 0.05% Triton X-100/PBS, the membrane was air-dried and exposed to x-ray film.
Expression of Recombinant Kinases-The overlapping recombinant CaMKK␤ clones were incorporated into one full-length coding sequence in the plasmid pBK by digestion and religation of unique restriction sites within the overlap regions (EcoRI at bp 1413, XhoI at bp 635, and DraIII at bp 162). The entire coding region was then amplified by polymerase chain reaction using a forward primer (5Ј-GAAGTTTCTA-GACACACCATG-3Ј) that created an XbaI restriction site 11 bases upstream of the ATG initiation codon and a reverse primer (5Ј-CCCTT-TAGTGAGGGTTAATT-3Ј) complementary to the T3 promoter present in the multiple cloning site of pBK. A 172-bp XbaI/DraIII fragment from the polymerase chain reaction product containing the ATG initiation codon was substituted for the corresponding region of the original clone. DNA sequencing indicated that no mutations had been generated by the procedure. An XbaI fragment containing the full-length coding sequence without the amino-terminal untranslated sequence was then subcloned into the maltose-binding protein (MBP) fusion protein, Esch-erichia coli expression plasmid, pMAL (New England Biolabs). Following induction of MBP-CaMKK␤ protein expression in E. coli by isopropyl-1-thio-␤-D-galactopyranoside, the fusion protein was isolated by passing the cell extract over an amylose-Sepharose column and was eluted from the column with maltose. Recombinant CaMKIV, CaMKIV T200A , CaMKIV 1-317 , CaMKIV 1-317 T200A , CaMKI, and CaMKI T177A were expressed in E. coli as glutathione S-transferase fusion proteins, which were purified on glutathione-Sepharose columns and eluted with glutathione as described previously (4,5).
CaMKK␤ Kinase Assays-CaMKK␤ activity was determined by measuring its ability to activate CaMKI or CaMKIV. CaMKI or CaMKIV was preincubated in the presence or absence of CaMKK␤ at 30°C for varying lengths of time in 50 l of preincubation buffer containing 20 mM Tris-HCl, pH 7.5, 0.5 mM DTT, 0.1% Tween 20, 8 mM MgCl 2 , 400 M ATP, 1 mM CaCl 2 , and 1 M CaM. The CaM kinase synthetic peptide substrate GS-10 (100 M) and 1 Ci [␥-32 P]ATP were then added, and the reaction was allowed to continue for an additional 4 -6 min at 30°C. For reactions performed in the absence of CaM, the CaCl 2 and CaM were omitted, and 1 mM EGTA was included. Phosphate incorporation into GS-10 peptide was quantitated by its binding to P81 phosphocellulose paper as described previously (35). Phosphorylation of CaMKI or CaMKIV by CaMKK␤ was assessed following incubation with CaMKK␤ in phosphorylation buffer containing 20 mM Tris-HCl, pH 7.5, 0.5 mM dithiothreitol, 0.1% Tween 20, 8 mM MgCl 2 , 400 M ATP, 1 mM CaCl 2 , 1 M CaM, and 1 Ci of [␥-32 P]ATP. After 10 min at 30°C, the reaction was stopped by the addition of 10ϫ SDSpolyacrylamide gel electrophoresis sample buffer, resulting in final concentrations of 0.1 M Tris-HCl, pH 6.8, 2% SDS, 2% ␤-mercaptoethanol, and 10% glycerol. Samples were heated 5 min at 95°C and electrophoresed through SDS-10% polyacrylamide gels, which were then dried and exposed to x-ray film. roacetic acid and dried. Phosphate incorporation into CaMKK␤ was quantitated by liquid scintillation counting of the filters.
CaM Overlay-CaMKK␤ (0.5 g) was electrophoresed through an SDS-10% polyacrylamide gel and then transferred to an Immobilon-P (Millipore Corp.) transfer membrane for 2 h at 75 V in buffer containing 25 mM Tris-HCl, 0.2 M glycine, pH 8.2, 0.37% SDS, and 20% methanol. The membrane was rocked in 0.1 M imidazole, pH 7.0, for 10 min followed by 40 min in CaM overlay buffer containing 20 mM imidazole, pH 7.0, 0.2 M KCl, 0.1% bovine serum albumin, and either 1 mM CaCl 2 or 1 mM EGTA, at which point 125 I-labeled recombinant chicken CaM was added at a concentration of 0.2 nM and specific activity of 1 ϫ 10 6 cpm/ml. After 2 h at room temperature, the membrane was washed two times for 30 min in CaM overlay buffer, air-dried, and exposed to x-ray film. Iodination of CaM was carried out using Bolton-Hunter reagent from Amersham Pharmacia Biotech as described by Chafouleas et al. (44).
Transient Transfection Assay-Tag Jurkat cells were maintained in RPMI medium supplemented with 10% fetal calf serum. Cells at log growth phase were pelleted, resuspended in medium, mixed with DNA, and transfected by electroporation at 250 V and 960 microfarads. All transfections included 5 g of 5ϫ GAL4-luciferase reporter plasmid and 5 g of GAL4-CREB expression plasmid together with pSG5 (Stratagene) expression plasmid containing CaMKIV (3 g), CaMKIV T200A (3 g), CaMKIV 1-317 (0.1 g), CaMKIV 1-317T200A (0.1 g), and/or CaMKK␤ (3 g), as indicated. Transfection of CaMKI was achieved using SR␣/ CaMKI (3 g) expression plasmid. In control transfections, empty vectors were included with the reporter and GAL4-CREB plasmids. Luciferase activity was determined using beetle luciferin (Promega) according to the manufacturer. All luciferase values were normalized for protein levels and transfection efficiency.
Protein Determination-Protein concentrations were determined by the method of Bradford (45) using ␥-globulin as standard or by the method of Lowry as described previously (29) using bovine serum albumin as a standard. Fig. 1 is an alignment of the deduced amino acid sequences of human and rat brain CaMKK␤ with the rat CaMKK␤ peptide data (39), sequences of CaMKK␣ (40), and the putative Caenorhabditis elegans CaMKK homologue CELCO5H8.1 (GenBank TM accession number U11029). Human and rat brain CaMKK␤ cDNAs encode 588 and 587 amino acid residue proteins, respectively, with 65 and 56% sequence identity with CaMKK␣ and CELCO5H8.1, respectively. Subsequent to obtaining the sequences of rat and human CaMKK␤ cDNAs, the sequence of a rat CaMKK cDNA identical to the rat CaMKK␤ sequence shown in Fig. 1 was reported by Kitani et al. (46).

Molecular Cloning of CaMKK␤-
Tissue Distribution of CaMKK␤-To obtain information about possible functional differences between CaMKK␣ and CaMKK␤, their tissue distributions at the protein and mRNA levels were examined using immunoblotting and RNase protection, respectively. As shown in Fig. 2A, CaMKK␤ was strongly expressed in brain but was also found in the other CaMKIV-expressing tissues (9, 10, 12), thymus, testis, and spleen. Other tissues such as lung also expressed detectable CaMKK␤, although for some, such as liver, kidney, intestine, and heart, only trace levels were observed. Interestingly, in brain, anti-CaMKK␤ antibodies detect two closely spaced bands, roughly equivalent in immunoreactivity. In contrast, in the other tissues examined, only a single immunoreactive band was observed, which electrophoresed similarly to the lower band of the brain doublet. While the reason for this discrepancy is not entirely clear, several lines of evidence strongly suggest that brain CaMKK␤ may be subject to tissue-specific posttranslational or post-transcriptional processing. First, in some purified preparations of CaMKK␤ from rat brain, a doublet pattern is observed on SDS-polyacrylamide gel electrophoresis, with both bands migrating more slowly than that of the single CaMKK␣ band. Second, in addition to the antibodies to the unique sequence in its N-terminal region, we have also prepared anti-peptide antibodies to a unique region representing the C terminus of rat CaMKK␤: amino acids 571-587 (SPPRT-PPQQPEEAMEPE). In apparently identical fashion, the Cterminally directed CaMKK␤ antibodies recognized a doublet in brain at the same position as the doublet recognized by the N-terminally directed antibodies. In testis, a single CaMKK␤ band was detected, migrating similarly to the faster of the two bands of the brain doublet. In similar fashion to its protein distribution, CaMKK␤ mRNA levels were high in brain, thymus, and spleen and could be detected in testis, although the latter was more variable (data not shown). By contrast, CaMKK␣ expression was more restricted, being detected strongly in brain and below the level of detection in other tissues (Fig. 2B), although message could also be detected in thymus and spleen (data not shown). Thus, as noted above for CaMKK␤, further studies will be required to assess whether other isoforms of CaMKK␣ (perhaps undetectable using the C-terminally directed antipeptide antibodies) are present in nonneural tissues. Fig. 3A depicts the regional distributions in rat brain of CaMKK␤, CaMKIV, and CaMKK␣, by in situ hybridization using antisense riboprobes, and with the sense strand of CaMKK␤ as negative control. Both CaMKKs showed a strong signal in parietal cortex but otherwise exhibited strikingly different patterns of expression. The CaMKK␤ signal was most abundant in cerebellum and outer cortex, moderate in inner cortex and caudate-putamen, and low in hippocampus, olfac- , and CaMKK␣ (B) was analyzed by immunoblotting using affinity-purified, isoform-specific antibodies in the following rat tissues: liver (Liv), kidney (Kid), intestine (Int), brain (Br), thymus (Thy), testis (Tes), lung (Lun), spleen (Spl), and heart (Hrt). tory bulb, brainstem, mesencephalon, and thalamus. This pattern is almost identical to that exhibited by CaMKIV (Fig. 3A, right panel) and originally by Ohmstede et al. using immunohistochemistry (9). In contrast to CaMKK␤, CaMKK␣ expression was most prominent in hippocampus and outer cortex; moderate in inner cortex, brain stem, mesencephalon, thalamus, and olfactory bulb; and low in caudate-putamen and cerebellum. It is of note that, similarly to CaMKK␣, CaMKI expression in brain has also been reported to be highest in hippocampus, cortex, brain stem, and olfactory bulb (8,48).
The expression pattern in cerebellum was further investigated by probing sections of cerebellar cortex for CaMKK␤, CaMKIV, and CaMKK␣ (Fig. 3B). The CaMKK␤ and CaMKIV signals were again abundant relative to CaMKK␣. Higher magnification revealed co-localization of CaMKK␤ and CaMKIV to the granule cell layer with no signal detected above background in the Purkinje cell or molecular layers. In contrast, only a very weak CaMKK␣ signal was observed in the granule cell layer (data not shown). Similarly, CaMKI expression was reported to be low in the granule cell layer relative to the Purkinje cell and molecular layers (8,48). Thus, in brain CaMKK␤ expression tracks more closely with CaMKIV, and CaMKK␣ expression tracks more closely with CaMKI.
Characterization of Recombinant CaMKK␤-We next examined properties of recombinant CaMKK␤. Rat CaMKK␤ was overexpressed and purified from E. coli as an MBP fusion protein, and its ability to activate and phosphorylate CaMKIV and CaMKI was assessed. CaMKK␤ does not directly phosphorylate the CaM kinase substrate peptide GS-10 (Fig. 4A, last  two bars). Thus, its ability to activate CaMKIV could be determined by measuring the latter's peptide kinase activity after a 10-min preincubation in the presence or absence of CaMKK␤. Fig. 4A shows that CaMKIV wild type is strongly activated by CaMKK␤ and that activation is a Ca 2ϩ /CaM-dependent process. The activating phosphorylation site is identified as Thr 200 based on a lack of activation following its replacement with a nonphosphorylatable Ala residue. In similar fashion to activation, CaMKIV was phosphorylated by CaMKK␤ in a Ca 2ϩ /CaMdependent manner with the extent of phosphorylation relative to basal autophosphorylation (10-fold) paralleling the degree of activation (Fig. 4B). Consistent with its lack of activation, phosphorylation of the CaMKIV T200A mutant was unaffected by CaMKK␤ (lanes 5-8). The ability of CaMKK␤ to phosphorylate and activate CaMKI was examined next. CaMKK␤ induced, in a Ca 2ϩ /CaM-dependent fashion, a 25-fold increase in CaMKI activity but was without effect on CaMKI T177A (Fig.  4C). Similarly, substantial phosphorylation of CaMKI by CaMKK␤ was observed, and this phosphorylation was abolished by replacement of Thr 177 with Ala (Fig. 4D). As also shown in Fig. 4, B and D, autophosphorylation of CaMKK␤ was observed, although enhancement by Ca 2ϩ /CaM tended to be variable, and some degree of Ca 2ϩ /CaM-independent autophosphorylation was consistently observed.
It has been previously demonstrated that the activating Thr residues of CaMKIV and CaMKI are accessible for phosphorylation by CaMKK only after Ca 2ϩ /CaM has bound to the respective CaMK (4,36,47). The CaMKKs, however, are also Ca 2ϩ / CaM-binding proteins whose activity is enhanced by Ca 2ϩ /CaM (4,36,39,47). The CaM overlay shown in Fig. 5A demonstrates Ca 2ϩ -dependent binding of 125 I-labeled CaM to recombinant CaMKK␤. To assess whether CaMKK␤ is regulated by Ca 2ϩ / CaM, an activation assay was performed using the truncated form of CaMKIV, CaMKIV 1-317 , as substrate for CaMKK␤. Since CaMKIV 1-317 has its autoinhibitory and CaM-binding domains deleted, any observed effects of Ca 2ϩ /CaM are attributable to an interaction of Ca 2ϩ /CaM with CaMKK␤. As shown in Fig. 5B, Ca 2ϩ /CaM enhances CaMKIV 1-317 activity by direct activation of CaMKK␤.
Taken together, the data of Figs. 4 and 5 indicate that CaMKK␤ activates CaMKIV and CaMKI by phosphorylating their activation loop Thr residues, Thr 200 and Thr 177 , respectively, in a Ca 2ϩ /CaM-dependent fashion and that at least a portion of this CaM dependence may be due to direct Ca 2ϩ /CaM stimulation of CaMKK␤ activity. We also observed that compared with CaMKKs purified from brain (39), the potency of recombinant CaMKK␤ to activate CaMKI or CaMKIV was ϳ40-fold lower (data not shown), suggesting that additional regulatory signals may impinge on CaMKK␤. CaMKK␤ Autophosphorylation-Incubation in the presence of [␥-32 P]ATP leads to phosphorylation of CaMKK␤ (Fig. 4, B  and D). The extent of 32 P incorporation is unaffected by the presence or absence of CaMKIV or CaMKI (data not shown), indicative of an autocatalytic mechanism. Approximately 0.6 mol of phosphate/mol of enzyme is maximally incorporated (Fig. 6A). Phosphorylation followed by factor Xa cleavage to separate the fusion partner, MBP, from CaMKK␤ indicated that all of the label was incorporated into the latter (data not shown). The degrees of autophosphorylation and CaMKK␤ concentration, over a range of 100 -2000 nM, were linearly related, indicative of an intramolecular reaction (Fig. 6B). This linear relationship was observed in both the presence and absence of Ca 2ϩ /CaM. Of particular interest was the question of whether CaMKK␤ autophosphorylation regulated its ability to activate CaMKIV. To address this question, CaMKK␤ was preincubated in autophosphorylation buffer containing Ca 2ϩ /CaM for 2 h in either the presence (circles) or absence (triangles) of MgATP prior to its incubation with CaMKIV 1-317 . As shown in Fig. 6C, autophosphorylated CaMKK␤ activated CaMKIV with the same rate and to the same extent as did unphosphorylated enzyme both in the presence (open symbols) or absence (filled symbols) of Ca 2ϩ /CaM. A similar result was obtained with CaMKK␤ autophosphorylated in the presence of EGTA (data FIG. 4. Phosphorylation and activation of CaMKIV and CaMKI by CaMKK␤. A, activation of CaMKIV. Following a 10-min preincubation in the presence or absence of CaMKK␤ (2 g/ml) and Ca 2ϩ /CaM as indicated, CaMKIV (wild type, WT) (4 g/ml) or CaMKIV T200A (4 g/ml) activity was determined in a 6-min assay using GS-10 peptide substrate. Values represent the mean Ϯ S.E. (n ϭ 3). B, phosphorylation of CaMKIV. CaMKIV (4 g/ml) or CaMKIV T200A (4 g/ml) was incubated in the presence or absence of CaMKK␤ (2 g/ml) and Ca 2ϩ /CaM as indicated for 10 min in phosphorylation buffer and electrophoresed through SDS-10% polyacrylamide gels, which were exposed to x-ray film. The positions of the glutathione S-transferase-CaMKIV and MBP-CaMKK␤ fusion proteins are indicated by arrows. C. Activation of CaMKI. Following a 10 min preincubation in the presence or absence of CaMKK␤ (2 g/ml) and Ca 2ϩ /CaM as indicated, CaMKI (4 g/ml) or CaMKI T177A (4 g/ml) activity was determined in a 6 min assay. Values represent the mean Ϯ S.E. (n ϭ 3). D. Phosphorylation of CaMKI. CaMKI (4 g/ml) or CaMKI T177A (4 g/ml) was incubated in the presence or absence of CaMKK␤ (2 g/ml) and Ca 2ϩ /CaM for 10 min in phosphorylation buffer and electrophoresed through SDS-10% polyacrylamide gels, which were exposed to x-ray film. The positions of the glutathione S-transferase-CaMKI and MBP-CaMKK␤ fusion proteins are indicated by arrows. not shown). Thus, neither the total activity of CaMKK␤ nor its regulation by Ca 2ϩ /CaM appears to be altered by autophosphorylation.
Activation of CaMKIV and CaMKI by CaMKK␤ in Jurkat T Cells-The ability of CaMKK␤ to activate CaMKs was examined in Jurkat cells that express CaMKI, CaMKIV, and CaMKK endogenously. Since either CaMKIV or CaMKI can phosphorylate and activate CREB when co-expressed with GAL4-CREB and a GAL4-luciferase reporter construct, this system was used as an indicator of CaMK activity in a Jurkat cell transient transfection assay. Following cotransfection of GAL4-CREB and luciferase reporter with expression plasmid for CaMKIV or CaMKI and/or CaMKK␤, the cells were left untreated or were stimulated for 5 h with 1 M ionomycin. As shown in Fig. 7A, transfection of full-length CaMKIV produces no increase in luciferase activity in unstimulated cells but a 5-fold increase in ionomycin-treated cells, indicating that CaMKIV activity is tightly controlled by intracellular Ca 2ϩ levels. The ionomycin-induced enhancement of activity increased to 12-fold in cells overexpressing both CaMKIV and CaMKK␤. In contrast, the CaMKIV T200A mutant was not activated by ionomycin or CaMKK␤. Thus, activation of full-length CaMKIV intracellularly requires Ca 2ϩ elevation and CaMKKcatalyzed phosphorylation of Thr 200 . In contrast to the fulllength enzyme, the Ca 2ϩ /CaM-independent CaMKIV 1-317 mutant enhanced luciferase activity in unstimulated cells (3-fold), and the addition of ionomycin only increased activity an additional 2-fold, consistent with a significant loss of Ca 2ϩ regulation (Fig. 7B). Co-expression of CaMKK␤ along with CaMKIV 1-317 led to an additional 2-fold increase in luciferase activity, but there was only a modest (1.5-fold) further increase after ionomycin addition. Thus, whereas CaMKK␤ is stimulated by ionomycin-induced elevation of intracellular Ca 2ϩ , consistent with Ca 2ϩ /CaM regulation in vitro, most of the Ca 2ϩ control in vivo is exerted at the level of CaMKIV rather than CaMKK␤. As expected, the CaMKIV 1-317 T200A mutant was unresponsive to ionomycin treatment and/or transfection of CaMKK␤.
As demonstrated in Fig. 7C, expression of CaMKI or coexpression of CaMKI and CaMKK␤ increases luciferase activity in unstimulated Jurkat cells with ionomycin inducing further increases of 4-and 3-fold, respectively. Similarly, Sun et al. (19) observed that expression of CaMKI in GH 3 cells stimulated, to some extent, CREB-mediated transcription in the absence of depolarization-induced Ca 2ϩ elevation, although the effect of the latter on transcriptional activation was potentiated by CaMKI. Since CaMKI is fully Ca 2ϩ -dependent in vitro, the difference between CaMKI and CaMKIV (Fig. 7, compare A with C) suggests that in vivo a proportion of the former may be capable of activation at resting Ca 2ϩ levels. DISCUSSION CaMKK␣ and CaMKK␤ are members of a recently identified group of kinases that selectively phosphorylate and activate CaMKI and CaMKIV (Fig. 1). Both CaMKK␣ and CaMKK␤ are also CaM-binding proteins whose activity can be enhanced by Ca 2ϩ /CaM (Fig. 6 We demonstrate here that CaMKK␣ and CaMKK␤ are distributed differently among rat tissues with the distribution of CaMKK␤ correlating with that of CaMKIV. Furthermore, both between brain regions (Fig. 3A) and in cerebellar cellular layers (Fig. 3B), CaMKK␤ and CaMKIV display similar distributions. For example, within the cerebellum, the intense CaMKK␤ and CaMKIV signals co-localized to the granule cell layer, where relatively little CaMKK␣ was detected. Taken together, these results suggest that CaMKK␤ is involved in the acute regulation of CaMKIV activity. Similarly, based on regional brain localizations, CaMKK␣ may have an important role in the regulation of CaMKI. On the other hand, since CaMKI displays a broad and essentially ubiquitous tissue distribution (4, 6 -8), it is unclear which CaMKK isoform is responsible for its regulation in nonneural tissues. Resolution of this latter issue may require investigation into the possible existence of additional members of the CaMKK family.
We have characterized recombinant CaMKK␤ produced in bacteria. The bacterially expressed enzyme provides the opportunity to examine CaMKK␤ in the absence of co-purifying CaMKK␣ or potential contaminating regulatory enzymes. Recombinant CaMKK␤ activates CaMKI (25-fold) and CaMKIV (12-fold) by phosphorylating Thr 177 and Thr 200 , respectively (Fig. 4), and binds to and exhibits activity enhanced by Ca 2ϩ / CaM (Figs. 5 and 6). These properties are similar to those described previously for recombinant CaMKK␣ (35,48). Therefore, to date, functional differences between CaMKK␣ and CaMKK␤ have yet to be identified. Given the high amino acid sequence identity between CaMKK␣ and CaMKK␤ throughout the catalytic and CaM-binding domains (Fig. 1) (65% identical, 80% similar), it is not surprising that the two enzymes function similarly in vitro.
We have observed that although the substrate specificity and CaM regulation of recombinant CaMKK␤ are similar to that reported for CaMKK␤ purified from brain (39), the former enzyme is about 40-fold less potent as a CaMK activator than FIG. 6. CaMKK␤ autophosphorylation. A, time course. CaMKK␤ (0.02 g/l) was incubated in autophosphorylation buffer plus MgATP at 30°C in the presence (open circles) or absence (closed circles) of Ca 2ϩ /CaM. At the indicated times, aliquots were removed for determination of phosphate incorporation. This experiment was repeated three times with similar results. A representative experiment is shown. B, dependence on CaMKK␤ concentration. CaMKK␤ was incubated at the indicated concentrations as described in A. After 20 min at 30°C, phosphate incorporation was determined. This experiment was repeated three times with similar results. A representative experiment is shown. C, effect on activity. CaMKK␤ (0.016 g/l) was incubated for 2 h in autophosphorylation buffer containing Ca 2ϩ /CaM in either the presence (circles) or absence (triangles) of MgATP. The enzyme was then preincubated with CaMKIV 1-317 (0.16 g/l) in either Ca 2ϩ /CaM (open symbols) or EGTA (equivalent to 1 mM free EGTA) (closed symbols) for the indicated times. CaMKIV 1-317 activity toward GS-10 was then determined in a 4-min assay. -Fold activation represents the increase in CaMKIV 1-317 activity over that obtained when the preincubation was carried out in the absence of CaMKK␤. Each point represents the mean of two determinations. the latter. It is possible that the low activity of recombinant CaMKK␤ results from the trivial reason that the enzyme fails to fold properly in E. coli or because of fusion with MBP, although removal of MBP by cleavage with factor Xa protease had no effect on activity. 3 However, an alternative explanation for the lower activity of recombinant CaMKK␤ is that this enzyme requires a post-translational modification for maximal activity that does not occur in E. coli. It may be noted, in this context, that in analogous fashion, the low activity of bacterially expressed CaMKIV or highly purified CaMKI led to the discovery that activation by CaMKK was required for maximal activity (29,30,50). Such a suggestion is consistent with the primary structure of CaMKK␤ (Fig. 1). There are Thr and Ser residues in the activation loop of CaMKK␤ positioned similarly to the residues of kinases that must be phosphorylated for maximal activity, including cyclic AMP-dependent protein kinase cyclin-dependent protein kinase 2, mitogen-activated protein kinase, CaMKI, and CaMKIV. Based on the crystal structures of eight unique kinases and on protein kinase primary sequence alignments, Johnson et al. (51) described structural determinants that can be used to predict whether or not a kinase will require activation loop phosphorylation for activity. Most importantly, all kinases that require phosphorylation have an Arg adjacent to the catalytic Asp that is believed to require neutralization by a negative charge from the activation loop to allow optimal alignment of the active site. The presence in CaMKK␤ of an Arg residue adjacent to the catalytic Asp (Asp 311 ) as well as the absence of a compensatory negatively charged residue at the correct position within the activation loop make it a good candidate as a kinase requiring phosphorylation. Although we have altered the CaMKK␤ activation loop Thr and Ser residues to Ala and Glu by mutagenesis, the mutant proteins were not expressed well in E. coli. Attempts are presently being made to determine whether expression of the mutants will occur in Jurkat cells.
A capacity for autophosphorylation is a property that CaMKK␤ shares with several CaM kinases. In the CaM kinase family, the best characterized autophosphorylation event is that of CaMKII on Thr 286 following binding and activation by Ca 2ϩ /CaM. The incorporated phosphate prevents the autoinhibited conformation of the enzyme from reforming after Ca 2ϩ / CaM dissociation, resulting in "autonomous" activity (reviewed in Refs. 52 and 53). CaMKIV also exhibits Ca 2ϩ /CaM-dependent autophosphorylation on Ser 12 and Ser 13 , which has been suggested to trigger relief from intrasteric inhibition of the enzyme by the N-terminal domain of the protein (35). This represents a process separate from the autoinhibition relieved by Ca 2ϩ /CaM binding and, if prevented by mutagenesis of Ser 12 and Ser 13 to nonphosphorylatable Ala residues, results in an enzyme incapable of being activated by either CaM or CaMKK. Thus, there is precedence for CaM kinase autophosphorylation having regulatory effects on enzyme activity. Recombinant CaMKK␤ autophosphorylation, however, was found not to alter enzyme activity or regulation by CaM. Nonetheless, although these in vitro measures of activity are apparently unaltered after autophosphorylation, it is possible that in vivo, CaMKK␤ autophosphorylation may have functional consequences. It may, for example, alter the turnover rate of the enzyme, direct the enzyme to distinct subcellular compartments, or promote protein-protein interactions with other signaling molecules.
The CaMKKs differ from other CaM kinase family members in exhibiting activity that is enhanced by, but not completely dependent on, Ca 2ϩ /CaM (this report; Refs. 4 and 39). This is true in the cell (Fig. 7C) as well as in vitro (Figs. 5B and 6C) and supports a model in which the Ca 2ϩ dependence of activation comes from a tight requirement for Ca 2ϩ /CaM binding to the CaMK rather than to the CaMKK. It therefore remains to be determined what role autophosphorylation and Ca 2ϩ /CaM binding to the CaMKKs play in a cellular context.