The autonomous activity of calcium/calmodulin-dependent protein kinase IV is required for its role in transcription.

Calcium/calmodulin-dependent kinase IV (CaMKIV) is a multifunctional serine/threonine kinase that is positively regulated by two main events. The first is the binding of calcium/calmodulin (Ca(2+)/CaM), which relieves intramolecular autoinhibition of the enzyme and leads to basal kinase activity. The second is activation by the upstream kinase, Ca(2+)/calmodulin-dependent kinase kinase. Phosphorylation of Ca(2+)/CaM-bound CaMKIV on its activation loop threonine (residue Thr(200) in human CaMKIV) by Ca(2+)/calmodulin-dependent kinase kinase leads to increased CaMKIV kinase activity. It has also been repeatedly noted that activation of CaMKIV is accompanied by the generation of Ca(2+)/CaM-independent or autonomous activity, although the significance of this event has been unclear. Here we demonstrate the importance of autonomous activity to CaMKIV biological function. We show that phosphorylation of CaMKIV on Thr(200) leads to the generation of a fully Ca(2+)/CaM-independent enzyme. By analyzing the behavior of wild-type and mutant CaMKIV proteins in biochemical experiments and cellular transcriptional assays, we demonstrate that CaMKIV autonomous activity is necessary and sufficient for CaMKIV-mediated transcription. The ability of wild-type CaMKIV to drive cAMP response element-binding protein-mediated transcription is strictly dependent upon an initiating Ca(2+) stimulus, which leads to kinase activation and development of autonomous activity in cells. Mutant CaMKIV proteins that are incapable of developing autonomous activity within a cellular context fail to drive transcription, whereas certain CaMKIV mutants that possess constitutive autonomous activity drive transcription in the absence of a Ca(2+) stimulus and independent of Ca(2+)/CaM binding or Thr(200) phosphorylation.

Similar to other calcium/calmodulin-dependent kinases, CaMKIV is inhibited by autoinhibitory interactions, which are relieved upon calcium/calmodulin (Ca 2ϩ /CaM) binding, leading to basal CaMKIV activity (17). Ca 2ϩ /CaM binding also exposes the CaMKIV activation loop, which can then be phosphorylated on a specific threonine residue (Thr 200 in human CaMKIV; Thr 196 in mouse CaMKIV) by Ca 2ϩ /calmodulin-dependent kinase kinase (CaMKK) (18,19). This phosphorylation event is associated with a marked increase in the total activity of CaMKIV and the development of some Ca 2ϩ /CaM-independent activity (20,21). Mutation of the activation loop threonine of CaMKIV to a nonphosphorylatable alanine does not affect the basal activity of the kinase but precludes its activation by CaMKK as well as abrogating the ability of the protein to drive transcription in cellular assays (18,20). This suggests that Thr 200 phosphorylation is necessary for CaMKIV transcriptional function. Recent work (22) further suggests that the most important role of Thr 200 phosphorylation is the generation of autonomous activity and implies that autonomous activity is necessary for CaMKIV-mediated transcriptional activity (22). CaMKIV transcriptional function may be negatively regulated by the protein phosphatase PP2A. PP2A can exist in a stable complex with CaMKIV and dephosphorylate CaMKIV on residue Thr 200 ; loss of the CaMKIV/PP2A interaction in cells leads to enhanced CaMKIV transcriptional activity (22,23).
In this study, we further examined the relationship between CaMKIV activation, autonomous activity, and transcriptional function. Here we report that CaMKIV is transformed from a Ca 2ϩ /CaM-dependent enzyme to a fully Ca 2ϩ /CaM-independent or autonomous enzyme after phosphorylation of its activation loop threonine, residue Thr 200 , by CaMKK. By analyzing the behavior of wild-type and mutant CaMKIV proteins in biochemical experiments and cellular transcriptional assays, we demonstrate that CaMKIV autonomous activity is necessary and sufficient for CaMKIV-mediated transcription, even in the absence of Ca 2ϩ /CaM binding or Thr 200 phosphorylation.

EXPERIMENTAL PROCEDURES
Cells and Cell Culture-The human embryonic kidney cell line, QBI-293A (293A), from Quantum Biotechnologies was used for all cell culture experiments and was cultured as recommended by the supplier.
Constructs-FLAG-CaMKIV WT and FLAG-CaMKIV substitution 320 -321 (called FLAG-CaMKIV FNDD in this paper), cloned into the mammalian expression vector pSG5 (Stratagene), were generated as previously described (22). The other CaMKIV mutants were generated by site-directed mutagenesis using single-stranded pSG5-FLAG CaMKIV as template and using the protocol and reagents provided with the Muta-Gene M13 kit (Bio-Rad). The single-stranded DNA was synthesized in CJ236 Escherichia coli that had been transformed with pSG5-CaMKIV and infected with M13KO7 helper phage (Invitrogen). The sense oligonucleotides used were as follows: CaMKIV  , and FLAG-CaMKIV FNDD T200A/pSG5 were created by ligating an EcoRI-or BamHIexcised fragment containing T200A from pSG5-FLAG-CaMKIV T200A/ pSG5 into the respective FLAG-CaMKIV/pSG5 mutant in which the corresponding fragment was removed. An in-frame FLAG tag sequence was introduced at the 5Ј-end of CaMKK␤ cloned into pSG5 (24) by creating an NheI site at the ATG start codon and inserting a linker sequence. The NheI site was obtained by site-directed mutagenesis using PCR with CaMKK␤/pSG5 as template. The following PCR primers were used: sense primer, 5Ј-GGG CGA ATT CCC CTA GAC ACG  CTA GCT CAT TAT GTG TCT CTA GC-3Ј; antisense primer, 5Ј-GCT  AGA GAC ACA TAA TGA GCT AGC GTG TCT AGG GGA ATT CGC  CC-3Ј. The double-stranded linker sequence containing FLAG and NheI  overhangs was created by annealing the following primers: sense  primer, 5Ј-CTA GCA TGG ACT ACA AGG ACG ACG ATG ACA AAG-3Ј;  antisense primer, 5Ј-CTA GCT TTG TCA TCG TCG TCC TTG TAG  TCC ATG-3Ј. Nucleotide sequences of cloning sites and entire cDNA of each mutant was confirmed by automated sequencing (Duke University DNA Sequencing Facility).
Expression and Purification of Recombinant Kinases-293A cells (ϳ95-99% confluence) growing in P150 plates were transfected with DNA constructs using Lipofectamine 2000 reagent (Invitrogen) according to the manufacturer's instructions. Cells were harvested 16 -24 h later. For most CaMKIV preparations, the cells were harvested by washing twice with Hanks' balanced salt solution without Ca 2ϩ , prior to being scraped from the dish into 0.5 ml of lysis buffer containing 25 mM Tris-HCl, pH 7.5, 50 mM NaCl, 0.5% Nonidet P-40, 25 mM NaH 2 PO 4 , 2 mM EGTA, 2 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 1 g/ml aprotinin, 1 g/ml leupeptin, and 1 g/ml pepstatin. The lysate was centrifuged at 14,000 ϫ g for 30 min at 4°C, and the resulting supernatant was incubated for 2 h at 4°C with 20 l of anti-FLAG M2agarose (A2220; Sigma). The resin was then washed three times in phosphatase buffer containing 50 mM Tris-HCl, pH 7.5, 0.1 mM Na 2 EDTA, 5 mM dithiothreitol, 2 mM MgCl 2 , and 0.01% Brij-35. The resin-bound proteins were then treated for 30 min at 30°C with 800 units of protein phosphatase (P0753S New England Biolabs) in 30 l of phosphatase buffer. The resin was then washed three times with lysis buffer and once with TBS (25 mM Tris, pH 7.5, 0.14 M NaCl, 2.7 mM KCl), prior to elution of the FLAG-tagged CaMKIV proteins in 50 l of TBS containing 300 ng/l FLAG peptide (F3290; Sigma) for 1 h at 4°C. The eluted CaMKIV was recovered and passed through filter disks (catalog number. 420162-0000 Kontes/Kimble) to remove any FLAG resin traces. CaMKK␤ protein was isolated similarly except eliminating the -phosphatase treatment steps. For isolation of CaMKIV proteins for examination of T200 phosphorylation and activation status in cultured cells, the recombinant CaMKIV proteins were isolated from either unstimulated transfected cells or transfected cells that had been stimulated with 2.5 M ionomycin (catalog number 407952; Calbiochem) for 5 min. These cells were washed twice with either Hanks' balanced salt solution without Ca 2ϩ or Hanks' balanced salt solution with Ca 2ϩ respectively, prior to being scraped from the dish into 0.5 ml of lysis buffer described above but also containing phosphatase inhibitors, 1 mM Na 3 VO 4 and 1 mM NaF. The CaMKIV proteins were isolated as above but eliminating the -phosphatase treatment steps. Protein concentrations were determined by running samples on SDS-polyacrylamide gels along with ␥-globulin standards and quantifying Coomassie Blue-stained protein bands. All kinases were stored at Ϫ80°C in 27% glycerol.
Calmodulin Purification-Chicken calmodulin was produced by bacterial expression as described previously (25).
Western Blotting-Samples were fractionated by SDS-PAGE and transferred to Immobilon-P membranes (Millipore Corp.). For detection of PP2A and CaMKIV, all blocking and antibody incubation steps were performed in TBS containing 5% nonfat milk. Washes were in TBS with 0.5% Tween 20. Anti-CaMKIV (catalog number 610276; BD Biosciences) was used at a 1:2000 dilution, whereas Anti-PP2Ac (catalog number 610555; BD Biosciences) was used at a 1:1000 dilution. Horseradish peroxidase-conjugated secondary antibodies were from Jackson ImmunoResearch Laboratories, Inc. and were used at a 1:5000 dilution. Immunoreactivity was detected with the ECL kit provided by Amersham Biosciences. Detection of CaMKIV T200P was performed as above except that the primary antibody incubations were in TBS containing 5% bovine serum albumin. The T200P antibody (Exalpha Biologicals Inc., Waterson, MA) was used at a 1:1000 dilution.
In Vitro Activation of CaMKIV by CaMKK␤ for Western Blotting-Purified recombinant CaMKIV (1.7 g/ml final concentration) was incubated with purified recombinant CaMKK␤ (1.7 g/ml final concentration) at 30°C for the indicated periods in buffer containing 41 mM HEPES, pH 7.5, 0.8 mM dithiothreitol, 16.7 mM MgCl 2 , 0.3 mM ATP, and either 1.7 mM CaCl 2 and 1.7 M calmodulin or 3.3 mM EGTA. The reactions were terminated by the addition of 10ϫ SDS loading dye and subsequent heating.
In Vitro Kinase Assays-For CaMKIV kinase assays involving activation by CaMKK␤, CaMKIV was incubated at 30°C, either alone or with CaMKK␤, for 10 min in the buffer described above containing Ca 2ϩ and calmodulin. The incubation period was ended by the addition of either EGTA or water, and the kinase assays were initiated by the addition of GS-10 substrate (PLRRTLSVAA) and [␥-32 P]ATP. Final kinase conditions were 25 mM HEPES, pH 7.5, 0.5 mM dithiothreitol, 0.1% Tween 20, 10 mM MgCl 2 , 0.2 mM ATP, 0.2 mM GS-10, 1 Ci of [␥-32 P]ATP, 1 mM CaCl 2 , and 1 M calmodulin and with or without 4 mM EGTA. The kinase reactions were terminated after 1 min by spotting aliquots of the reaction onto phosphocellulose paper (P-81; Whatman) and washing the filters extensively with 75 mM phosphoric acid. Phosphate incorporation was determined by liquid scintillation counting of the filters. For CaMKIV kinase assays performed in the absence of activation by CaMKK␤, the incubation period prior to the kinase assays was eliminated, and the final kinase conditions were 25 mM HEPES, pH 7.5, 0.5 mM dithiothreitol, 0.1% Tween 20, 10 mM MgCl 2 , 0.2 mM ATP, 0.2 mM GS-10, 1 Ci of [␥-32 P]ATP, and either 1 mM CaCl 2 and 1 M calmodulin or 2 mM EGTA.
Transcription Assays-Six-well dishes were seeded with 0.2 ϫ 10 6 293A cells/well. The next day, cells were transfected with 0.02 g of ␤-galactosidase, 0.16 g of Gal4 CREB, 0.4 g of 5ϫ Gal4 luciferase reporter, and 0.4 g of FLAG-CaMKIV/pSG5 DNA per well, using Lipofectamine reagent (Invitrogen) according to the manufacturer's recommendations. Control transfections were also performed substituting the empty pSG5 vector (Stratagene) in place of FLAG-CaMKIV/ pSG5. The following day, cells were treated with fresh medium containing either Me 2 SO vehicle control or ionomycin to a final concentration of 2.5 M. Cells were harvested ϳ16 h later. Luciferase and ␤-galactosidase activities in cell extracts were determined as previously described (10). Transfection efficiency was normalized using ␤-galactosidase activity.
Calmodulin Overlays-Calmodulin was labeled with 125 I using Bolton and Hunter reagent from Amersham Biosciences (IM 5861) following the provided instructions. CaMKIV proteins (100 ng each) were separated by SDS-PAGE, and the gels were then equilibrated for 30 min in transfer buffer (25 mM Tris, 192 mM glycine, 20% methanol, 0.1 M NaCl, pH 8.2) prior to transfer of proteins to Immobilon-P membranes. The membranes were then incubated in 0.1 M imidazole, pH 7.0, for 10 min and then in Solution G (20 mM imidazole, pH 7.0, 0.2 M KCl, 0.1% bovine serum albumin, 0.05% Tween 20) containing either 1 mM CaCl 2 or 1 mM EGTA for 40 min. The membranes were then incubated for 2 h with Solution G containing either CaCl 2 or EGTA along with [ 125 I]CaM (1 ϫ 10 6 cpm/ml). The membranes were then washed with Solution G before being dried and subjected to autoradiography.
Determination of K CaM Values-The calmodulin concentrations required for half-maximal Ca 2ϩ /CaM-dependent kinase activity (K CaM ) were determined for each CaMKIV protein by performing kinase assays under standard assay conditions in the presence of either 2 mM EGTA or 1 mM Ca 2ϩ and with varying calmodulin concentrations. The activity of each kinase in EGTA was subtracted from its activity in Ca 2ϩ /CaM to determine its Ca 2ϩ /CaM-dependent activity. The K CaM for each kinase was determined by analyzing the resulting substrate-velocity curve with GraphPad Prism software.

Phosphorylation of Its Activation Loop Threonine Transforms CaMKIV from a Ca 2ϩ /CaM-dependent Kinase to a Fully
Ca 2ϩ /CaM-independent or Autonomous Kinase-It has been previously observed that activation of CaMKIV by CaMKK leads to the generation of some CaMKIV autonomous activity (20,21). However, the extent of the autonomous activity and its significance had not been entirely understood. Here we further evaluated the generation and importance of CaMKIV autonomous activity. We first examined the phosphorylation of Thr 200 of CaMKIV by CaMKK␤ and the effect of CaMKK␤ on CaMKIV kinase activity. CaMKIV was incubated in the absence or presence of Ca 2ϩ /CaM, either alone or with CaMKK␤. Phosphorylation of CaMKIV on the activation loop residue Thr 200 was assessed by immunoblotting with an anti-T200P antibody (20,21). The specificity of the anti-T200P antibody is demonstrated by the inclusion of the CaMKIV T200A mutant in the experiment. In vitro, WT CaMKIV is phosphorylated by CaMKK␤ on residue Thr 200 only in the presence of Ca 2ϩ /CaM (Fig. 1A). Although CaMKK␤ is a Ca 2ϩ /CaM-responsive enzyme, it possesses significant Ca 2ϩ /CaM-independent activity (unlike CaMKK␣) (24,26). However, WT CaMKIV is not phosphorylated by CaMKK␤ in the absence of Ca 2ϩ /CaM, indicating a requirement for Ca 2ϩ /CaM binding to WT CaMKIV for Thr 200 phosphorylation, as has been previously reported (19). We next examined the effect of CaMKK on the kinase activity of CaMKIV. CaMKIV was incubated in buffer with Ca 2ϩ /CaM, either alone or with CaMKK␤, for 10 min at 30°C. The activity of CaMKIV in the presence of EGTA or Ca 2ϩ /CaM was then assessed in kinase assays. Unactivated CaMKIV is completely Ca 2ϩ /CaM-dependent and has no kinase activity in the presence of EGTA (Fig. 1B). CaMKK␤ induces an ϳ2-fold increase in the activity of WT CaMKIV over its basal activity in Ca 2ϩ / CaM but has no effect on the activity of CaMKIV T200A (Fig.  1B). Surprisingly, the activation of WT CaMKIV by CaMKK␤ was accompanied by the development of complete Ca 2ϩ /CaMindependence; activated WT CaMKIV was as active in EGTA as in Ca 2ϩ /CaM (Fig. 1B). Phosphorylation of Thr 200 is critical for the generation of autonomy as the CaMKIV T200A mutant fails to develop autonomous activity after incubation with CaMKK␤ ( Fig. 1B), and WT CaMKIV also fails to develop autonomous activity if incubated with CaMKK␤ in the absence of ATP (data not shown).
Activation of CaMKIV in Cells Is Accompanied by the Development of Autonomous Activity-Having established that activation of CaMKIV in vitro leads to the generation of a fully autonomous enzyme, we examined whether this phenomenon occurred in cells. CaMKIV proteins, isolated from unstimulated 293A cells, or from 293A cells that had been stimulated with the calcium ionophore ionomycin, were assayed for Thr 200 phosphorylation and for kinase activity. Ionomycin treatment leads to increases in intracellular Ca 2ϩ and Ca 2ϩ /CaM levels, which influence both CaMKIV and CaMKK in cells (19,21). Only a fraction of the CaMKIV in the cell became activated after the ionomycin stimulus, as demonstrated by our observation that only a fraction of CaMKIV was phosphorylated on residue Thr 200 ( Fig. 2A); correspondingly, autonomous activity generated was only a fraction of total CaMKIV activity (Fig.  2B). This partial activation of the cellular pool of CaMKIV led to an ϳ5-fold increase in autonomous activity but had no significant effect on total CaMKIV activity (assayed with Ca 2ϩ / CaM). This result suggested that the physiologically relevant outcome of CaMKIV activation in cells might be the generation of autonomous activity rather than the increase in the total activity of CaMKIV. Another observation that supports the hypothesis that CaMKIV autonomous activity is biologically important is the fact that whereas WT CaMKIV drives transcription in a Ca 2ϩ stimulus-dependent manner, the CaMKIV T200A mutant, which is incapable of developing autonomous activity, is transcriptionally incompetent (Fig. 2C) (24). Thus, we hypothesized that autonomous activity is important for CaMKIV function in transcription.
In order to pursue this hypothesis, we created a series of mutant CaMKIV proteins for use in biochemical experiments and cellular transcription assays. The regulatory domain was targeted because numerous published reports on CaMKIV and related Ca 2ϩ /CaM-dependent kinases have indicated that mutations in this region would lead to relief of CaMKIV autoinhi- The blot was stripped and reprobed for CaMKIV as a loading control. B, WT or T200A CaMKIV was incubated in buffer with Ca 2ϩ /CaM, either alone or with CaMKK␤, for 10 min at 30°C, prior to the initiation of kinase assays. For kinase assays in the presence of EGTA, EGTA and then GS-10 substrate was added after the incubation period. For kinase assays in Ca 2ϩ /CaM, GS-10 substrate alone was added after the incubation period. Data shown are from one experiment that is representative of five independent experiments. bition and constitutive activity (27)(28)(29)(30). A series of sequential 6-amino acid NAAIRS or NAAIRN substitution mutants was made spanning residues 300 -352 (Fig. 3) and named CaMKIV N300, N306, N312, N318, N324, N330, N336, N341, and N347. (The NAAIRS and NAAIRN sequences are flexible substitutions that do not generally disrupt overall protein structure (31).) The numbering of mutants indicates the most N-terminal residue mutated. A double mutant, named FNDD, where residues Phe 320 and Asn 321 were changed to aspartic acid residues was also generated, since mutations of these residues have previously been shown to lead to CaMKIV autonomous activity in the absence of Ca 2ϩ /CaM (29). Ca 2ϩ /CaM binding and PP2A binding of the mutant CaMKIV proteins were assessed, since these properties are important for the regulation of WT CaMKIV. The Ca 2ϩ /CaM-dependent and -independent kinase activities of unactivated CaMKIV mutants were also assessed to identify mutants with constitutive autonomous activity.
CaMKIV Residues 324 -341 Are Important for Calmodulin Binding-CaMKIV mutants FNDD, N318, N324, N330, and N336 bound [ 125 I]calmodulin poorly compared with WT CaMKIV (Fig. 4) in calmodulin overlay assays performed in the presence of Ca 2ϩ , implicating residues between 318 and 341 as important for calmodulin binding. (None of the CaMKIV pro-teins bound [ 125 I]calmodulin in the absence of Ca 2ϩ (data not shown).) To further validate these results, K CaM values were determined for WT and mutant CaMKIV. The K CaM value was defined as the calmodulin concentration at which each kinase achieved half-maximal Ca 2ϩ /CaM-dependent kinase activity and was determined by analyzing the substrate-velocity curve with GraphPad Prism software (results summarized in Table  I). Two of the mutants, N312 and FNDD, displayed very little or no Ca 2ϩ /CaM-dependent activity, and K CaM values could not be determined for these kinases using the above approach. WT CaMKIV was found to have a K CaM value of ϳ77 nM, which is consistent with values reported in the literature (17,19). Of the kinases tested, three mutants had K CaM values much greater than WT CaMKIV. N324 and N330 had K CaM values over 100 M each (more than a 1000-fold greater than WT), and N336 had an average K CaM of 329 nM, an ϳ4-fold increase over WT CaMKIV. These data suggest that residues between 324 and 341 are important for Ca 2ϩ /CaM binding, with residues between 324 and 335 being most important. There was an excellent correlation between the calmodulin overlay data and the K CaM data with the exception of one mutant, N318. This mutant had a K CaM value of 66 nM, close to the K CaM of 77 nM for WT CaMKIV (Table I), yet it bound [ 125 I]calmodulin poorly compared with WT CaMKIV in the overlay assay (Fig. 4). Therefore, we could not make a definitive judgment on the calmodulin binding properties of N318, although we suspected that it might respond normally to Ca 2ϩ /CaM in cells based on its K CaM value.
CaMKIV Residues between 306 and 323 Are Important for PP2A Binding-PP2A and CaMKIV have been shown to exist in a stable complex, with PP2A apparently acting as a negative Other mutants made were CaMKIV T200A, which contains a single point mutation that changes threonine 200 to an alanine, and CaMKIV FNDD, which contains a double mutation that changes phenylalanine 320 and asparagine 321 to aspartic acid residues. Regions of CaMKIV important for autoinhibition, calmodulin binding, and PP2A binding were determined through analysis of these mutants.
FIG. 4. Calmodulin overlay implicates CaMKIV residues between 318 and 341 as important for Ca 2؉ /CaM binding. WT and mutant CaMKIV proteins and bovine serum albumin (BSA) used as a negative control that had been transferred onto Immobilon-P membranes were subjected to a calmodulin overlay using 125 I-labeled calmodulin in buffer containing Ca 2ϩ . The membranes were subjected to autoradiography to detect bound [ 125 I]calmodulin. Shown is an autoradiogram from one experiment that is representative of three independent experiments. In the absence of Ca 2ϩ , no signal was detected for any of the CaMKIV proteins (data not shown).
regulator of CaMKIV activity (23). We recently demonstrated that residues within the CaMKIV regulatory region are important for binding of CaMKIV to PP2A and that inhibition of the WT CaMKIV/PP2A interaction in cells could result in increased CaMKIV transcriptional activity, which correlated with increased levels of Thr 200 phosphorylation (22). Using the additional CaMKIV mutants generated for this study, we expanded upon these findings by examining the ability of each CaMKIV protein to co-immunoprecipitate PP2A. Most of the CaMKIV mutants co-immunoprecipitated similar amounts of PP2A as compared with WT CaMKIV (Fig. 5). However, N306, N312, FNDD, and N318 co-immunoprecipitated very low amounts of PP2A, suggesting that CaMKIV residues between 306 and 323 are important for PP2A interaction. Based on these results, we anticipated that the activity of N306, N312, FNDD, and N318 might be misregulated in the cell, since PP2A normally acts to negatively regulate WT CaMKIV activity.
Residues between 306 and 329 Are Involved in Autoinhibition of CaMKIV-In order to assess whether any of the mutations resulted in constitutive autonomous activity, we analyzed the kinase activities of unactivated wild-type and mutant CaMKIV proteins (either in 2 mM EGTA or in 1 mM Ca 2ϩ and 1 M CaM). WT CaMKIV has no significant autonomous activity in its unactivated state and is fully dependent on Ca 2ϩ /CaM for its activity (Fig. 6, A and B). In contrast, unactivated CaMKIV mutants N306, N312, FNDD, N318, and N324 each had activity in EGTA in the absence of activation (Thr 200 phosphorylation) with specific activities of 17, 21, 22, 4, and 16 nmol/min/ mg, respectively, compared with less than 1 nmol/min/mg for WT CaMKIV (Fig. 6B). These data indicate that residues between 306 and 329 are important for maintaining CaMKIV in an inactive state in the absence of bound Ca 2ϩ /CaM or Thr 200 phosphorylation. Conversely, several of the mutants had significantly lower specific activities than WT CaMKIV in the presence of 1 mM Ca 2ϩ and 1 M CaM (Fig. 6A). For N324, N330, and N336, this was due to their poor Ca 2ϩ /CaM binding properties; in the presence of higher concentrations of calmodulin, these kinases achieved specific activities approaching that of WT CaMKIV (data not shown). The low activities of N300, N312, FNDD, and N318 were not improved by increased concentrations of Ca 2ϩ /CaM (data not shown). It is likely that the low activity of N300 is due to impairment of the catalytic function of the enzyme, since the mutated residue Val 300 is a hydrophobic residue conserved among kinase catalytic domains. However, for N312, it seems probable that the same mutations that partially relieve autoinhibition also prevent the full relief of autoinhibition upon Ca 2ϩ /CaM binding, whereas FNDD and N318 may exhibit both this defect and impaired Ca 2ϩ /CaM binding.
CaMKIV Autonomous Activity in Cells Correlates with CaMKIV Transcriptional Activity-Having evaluated some of the regulatory properties of wild-type and mutant CaMKIV proteins in vitro, we next examined the function of each kinase in cells. We first looked at indicators of activation for each CaMKIV protein isolated from unstimulated cells or from ionomycin-stimulated cells. Wild-type and mutant CaMKIV proteins were immunoprecipitated from unstimulated 293A cells or from 293A cells that had been stimulated for 5 min with 2.5 M ionomycin. The CaMKIV proteins were then analyzed for Thr 200 phosphorylation status or assayed for kinase activity in the presence of EGTA to assess autonomous activity. We also examined the ability of each CaMKIV protein to drive transcription in CREB-luciferase assays in the absence of or in response to a Ca 2ϩ stimulus (2.5 M ionomycin). As observed previously, WT CaMKIV becomes phosphorylated on Thr 200  after an ionomycin stimulus (Fig. 7A), and this is accompanied by the development of autonomous activity (Fig. 7B). Correspondingly, WT CaMKIV drives transcription only after an ionomycin stimulus (Fig. 7C), consistent with the hypothesis that CaMKIV autonomous activity is required for CaMKIV transcriptional activity. The behaviors of the CaMKIV mutants are also consistent with this hypothesis. Several of the mutants (N300, N341, and N347) behave similarly to WT CaMKIV, becoming phosphorylated on residue Thr 200 (Fig. 7A), developing greatly increased levels of autonomous activity (Fig. 7B), and driving transcription only after an ionomycin stimulus (Fig. 7C). The behavior of N300 is particularly noteworthy, since in its unactivated state, this mutant has much lower Ca 2ϩ /CaM-dependent activity than WT, N341, or N347 CaMKIV proteins (Fig. 6). Nevertheless, in cells N300 develops similar levels of autonomous activity as WT, N341, and N347 CaMKIV proteins upon a Ca 2ϩ stimulus and drives transcrip-tion to a similar degree. Noticeably, a fraction of the cellular pool of N300 appears to be phosphorylated on residue Thr 200 in the absence of a Ca 2ϩ stimulus (Fig. 7A). This may be due to an increased susceptibility of this mutant to Thr 200 phosphorylation by CaMKK. Indeed, even at ambient intracellular Ca 2ϩ concentrations, N300 interacts with CaMKK much more avidly than WT CaMKIV in co-immunoprecipitation experiments (supplemental Fig. S1). However, Thr 200 phosphorylation of N300 in unstimulated cells is accompanied by only low levels of autonomous activity. It is tempting to speculate that the relatively low levels of autonomous activity of WT and N300 CaMKIV proteins isolated from unstimulated cells (which are not associated with transcriptional activity) may represent a snapshot of rapidly fluctuating levels of activity in the cell rather than sustained levels of autonomous activity.
In contrast to the behaviors of WT, N300, N341, and N347 CaMKIV proteins, the T200A CaMKIV mutant, which cannot be phosphorylated on the activation loop threonine (Fig. 7A), fails to develop autonomous activity in cells (Fig. 7B) and is transcriptionally inactive (Fig. 7C). Two other mutants, N330 and N336, also display poor transcriptional activity in these assays. N330 fails to become phosphorylated on residue Thr 200 in response to an ionomycin stimulus (Fig. 7A), fails to develop autonomous activity (Fig. 7B), and correspondingly fails to drive transcription (Fig. 7C). N336, on the other hand, is poorly phosphorylated on Thr 200 in response to an ionomycin stimulus (Fig. 7A), develops only low levels of autonomous activity (Fig.  7B), and correspondingly drives transcription to a low degree (Fig. 7C). The failure of N330 and N336 to develop autonomous activity in a cellular context can be attributed to their poor calmodulin binding properties, which preclude them from binding Ca 2ϩ /CaM and from being activated at physiological Ca 2ϩ / CaM concentrations. (Both N330 and N336 are phosphorylated on Thr 200 and develop substantial levels of autonomous activity in vitro after incubation with CaMKK and high levels of Ca 2ϩ / CaM (supplemental Figs. S2 and S3). For the T200A, N330, and N336 mutants, their inability to drive transcription correlates with their failure to develop autonomous activity in cells, supporting the hypothesis that autonomous activity is necessary for CaMKIV-mediated transcription. Finally, several of the mutants, N306, N312, FNDD, N318, and N324, are capable of driving transcription in the absence of an ionomycin stimulus (Fig. 7C). Each of these kinases has significant levels of autonomous activity when isolated from unstimulated cells (Fig. 7B). This is not surprising, because each of these kinases has autonomous activity even in the absence of Thr 200 phosphorylation (Fig. 6B). Additionally, N306, N312, FNDD, and N318 are phosphorylated on Thr 200 to varying degrees when isolated from unstimulated cells. This is probably a consequence of their relatively poor ability to bind PP2A (Fig. 5), since these kinases would not be expected to be efficiently dephosphorylated by PP2A after activation resulting from normal cellular events such as transient Ca 2ϩ fluxes. The transcriptional activities of N306, N318, and N324 are significantly increased after an ionomycin stimulus (Fig. 7C), correlating with increased levels of Thr 200 phosphorylation (Fig. 7A) and autonomous activity in cells (Fig. 7B). 2 These experiments illustrate a clear relationship between autonomous activity of CaMKIV and CaMKIV transcriptional function and suggest 2 N324, which appears incapable of binding Ca 2ϩ /CaM (Fig. 4 and Table I), is able to be activated in cells upon a Ca 2ϩ stimulus. This is because, unlike WT CaMKIV, N324 does not require Ca 2ϩ /CaM for Thr 200 phosphorylation by CaMKK; N324 can be phosphorylated by CaMKK␤ in the absence of Ca 2ϩ /CaM and in the presence of EGTA (supplemental data, Fig. S2). This is probably because the same mutations that partially relieve autoinhibition of the enzyme also expose residue Thr 200 of N324, even in the absence of bound Ca 2ϩ /CaM.

FIG. 7. Behavior of WT and mutant CaMKIV proteins in cells.
A, WT and mutant CaMKIV proteins were isolated from 293A cells that had not been stimulated (U) or from 293A cells that had been stimulated with 2.5 M ionomycin for 5 min (S). Phosphorylation of CaMKIV on residue Thr 200 was assessed by immunoblotting, whereas CaMKIV levels were assessed by Coomassie Blue protein staining. Shown is one experiment that is representative of four independent experiments. B, WT and mutant CaMKIV proteins isolated from unstimulated (U) or from ionomycin-stimulated (S), 293A cells were assayed for kinase activity in the presence of EGTA. Shown are means and S.E. from four independent experiments. C, 293A cells were transfected with empty pSG5 vector (control), with WT CaMKIV, or with mutant forms of CaMKIV, together with Gal4-CREB and 5ϫ Gal4 luciferase reporter. The cells were either unstimulated (Me 2 SO; DMSO) or were stimulated with 2.5 M ionomycin (ION), and after ϳ16 h, CREB transcription was assessed by determining luciferase activity in cell lysates. Data shown are from one experiment that is representative of four independent experiments. that autonomous activity is the primary requirement for CaMKIV-mediated transcription.
Autonomous Activity, in the Absence of Thr 200 Phosphorylation, Is Sufficient to Drive Transcription-For WT CaMKIV (and for some of the mutants), autonomous activity is tightly linked to Thr 200 phosphorylation. In order to determine whether CaMKIV autonomous activity, in the absence of Thr 200 phosphorylation, would be sufficient for CaMKIV transcriptional function, we made mutations of residue Thr 200 to the nonphosphorylatable alanine in three of the mutants that have autonomous activity in vitro in the absence of Thr 200 phosphorylation. We then looked at levels of Thr 200 phosphorylation and autonomous activity of these three additional mutants, N312 T200A, FNDD T200A, and N318 T200A, when isolated from unstimulated or ionomycin-stimulated cells, and we also tested their ability to drive transcription in CREB-luciferase assays. N312 T200A, FNDD T200A, and N318 T200A were not phosphorylated on Thr 200 when isolated from either unstimulated 293A cells or from 293A cells stimulated with ionomycin (Fig. 8A). Each of these three mutants had autonomous activity in cells that was not increased upon an ionomycin stimulus (Fig. 8B), and each drove transcription in a Ca 2ϩ stimulus-independent manner (Fig. 8C). These results demonstrate that, at least in the context of some CaMKIV mutants, autonomous activity, even in the absence of Thr 200 phosphorylation, is sufficient for CaMKIV transcriptional function. DISCUSSION Numerous studies have reported that activation of CaMKIV is accompanied by the development of some Ca 2ϩ / CaM-independent activity (20,21). However, here we show that CaMKIV is actually capable of becoming fully autonomous after activation by CaMKK. The most likely reason why this observation has not been reported previously is that the experimental conditions under which CaMKIV activation has been studied were not optimal for observing the generation of complete CaMKIV autonomy.
In the course of experiments designed to optimize CaMKIV activation by CaMKK in vitro, we found that high amounts of CaMKK␤ (relative to CaMKIV) were required for maximal activation of CaMKIV (i.e. for maximal T200 phosphorylation and for generation of complete CaMKIV autonomy). In all of the experiments presented here examining activation of CaMKIV by CaMKK in vitro, we utilized approximately equal amounts of CaMKIV and CaMKK␤. In contrast, many of the previous studies examining CaMKK activation of CaMKIV have used catalytic amounts of CaMKK relative to CaMKIV (20,32). Furthermore, studies performed prior to the identification of CaMKK as the activating kinase for CaMKIV (29,33,34) and prior to the cloning and characterization of CaMKK␣/␤ (24, 35, 36) used brain extract (which probably contained relatively low levels of CaMKK) to activate CaMKIV in vitro. Under these conditions, only a fraction of CaMKIV appeared to be activated. This led us to suspect that in many previous studies looking at CaMKIV activation in vitro, levels of CaMKK were the limiting factor for complete CaMKIV activation. This also raised the intriguing possibility that the relationship between CaMKIV and CaMKK could be stoichiometric rather than catalytic.
Numerous studies have suggested that activation of CaMKIV by CaMKK is important for CaMKIV transcriptional function. For example, co-transfection of CaMKK along with CaMKIV was found to greatly potentiate Ca 2ϩ -dependent CaMKIV transcription in cells (24,36). (This might indicate that low levels of endogenous CaMKK were limiting for activation of the exogenously expressed CaMKIV protein.) Other studies have shown that the T200A CaMKIV mutant, which is incapable of activation by CaMKK, is transcriptionally inactive (18,24), suggesting that activation by CaMKK is actually critical for CaMKIV function in transcription. In this study, we observe that, in vitro, activation of CaMKIV by CaMKK converts CaMKIV to a fully Ca 2ϩ /CaM-independent enzyme. This finding appears to be physiologically important, since our experiments indicate that the primary consequence of CaMKIV activation in cells is the generation of autonomous activity rather than any significant increase in total activity (Fig. 2B). In addition, the generation of autonomous activity of WT CaMKIV upon an ionomycin stimulus correlates with the ability of WT CaMKIV to drive transcription in cellular assays. Together, these observations led us to hypothesize that CaMKIV autonomous activity is required for transcription. In order to pursue this hypothesis, we created a series of mutants for use in comparative biochemical and cellular transcriptional assays.
We analyzed the regulatory properties of WT and mutant CaMKIV proteins in vitro, assessed their activation in cells, and examined their ability to drive CREB-mediated transcrip- tion in cellular assays. Our results corroborate our hypothesis that CaMKIV autonomous activity is required for transcription. In the case of each CaMKIV protein, transcriptional activity correlates with the generation of or levels of autonomous activity in cells. Additionally, it appears that Ca 2ϩ /CaM binding and Thr 200 phosphorylation are not strictly required for CaMKIV transcription.
The lack of a requirement for Ca 2ϩ /CaM binding of CaMKIV for its role in transcription is exemplified by the N324 mutant. N324 fails to bind detectable levels of [ 125 I]calmodulin in overlay assays and has a K CaM value in excess of 100 M, indicating that this mutant is unlikely to bind Ca 2ϩ /CaM under physiological conditions. Nevertheless, N324, which has constitutive autonomous activity, is capable of driving CREB-mediated transcription (Fig. 7C), suggesting that Ca 2ϩ /CaM-binding is not absolutely required for CaMKIV transcription. 2 We also found that Thr 200 phosphorylation was not strictly required for CaMKIV transcription. In order to determine whether autonomous activity, in the absence of Thr 200 phosphorylation, would be sufficient for CaMKIV transcriptional activity, we made threonine 200 to alanine mutations in N312, FNDD, and N318, three kinases that have autonomous activity even in their unactivated states (Fig. 6B). These three additional mutants (N312 T200A, FNDD T200A, and N318 T200A) had levels of autonomous activity in cells that were not increased upon ionomycin stimulation, and each drove transcription in a Ca 2ϩ stimulus-independent manner. These results demonstrate that autonomous activity, even in the absence of Thr 200 phosphorylation, is sufficient for CaMKIV transcriptional function. Thus, the main role of Thr 200 phosphorylation in regulating CaMKIV transcriptional activity appears to be the generation and/or maintenance of autonomous activity. This is quite similar to the role of Thr 286 phosphorylation in the related kinase CaMKII; phosphorylation of Thr 286 prevents a return of CaMKII to its autoinhibited state upon Ca 2ϩ /CaM removal and confers enzyme autonomy (37)(38)(39).
Previous studies have shown that PP2A negatively regulates CaMKIV transcriptional function through dephosphorylation of CaMKIV residue Thr 200 (22,23). Our study corroborates the importance of PP2A in regulating CaMKIV function. For example, in unstimulated cells, CaMKIV mutants that are impaired in PP2A binding (Fig. 5) generally exhibit higher levels of Thr 200 phosphorylation, autonomous activity, and transcriptional activity (Fig. 7) than CaMKIV proteins that bind PP2A normally. This is presumably because mutants that bind PP2A poorly are less efficiently dephosphorylated in cells after activation resulting even from ambient cellular Ca 2ϩ transients. Thus, a fraction of the cellular pool of these mutants is somewhat activated even in unstimulated cells. The significance of this is exemplified by the behavior of N318 and N324. Although, in vitro, unactivated N318, which binds PP2A poorly (Fig. 5), has less autonomous activity than unactivated N324, which binds PP2A normally (Fig. 5), in cells, N318 exists in a more activated state even in the absence of an ionomycin stimulus (Fig. 7A), has greater autonomous activity (Fig. 7B), and drives transcription to a greater degree than N324 (Fig. 7C).
Together, our data suggest a model for CaMKIV function in transcription where autonomous activity is the primary requirement for CaMKIV transcriptional activity. CaMKIV mutants with constitutive autonomous activity can bypass the requirements for Ca 2ϩ /CaM and Thr 200 phosphorylation, indicating that autonomous activity alone is sufficient for transcriptional activity. WT CaMKIV, however, achieves autonomous activity (and transcriptional competence) only after Ca 2ϩ /CaM binding and subsequent Thr 200 phosphorylation, whereas PP2A negatively regulates CaMKIV transcriptional function through dephosphorylation of residue Thr 200 and termination of the autonomous activity of CaMKIV (Fig. 9).
The primary function of PP2A and CaMKK in the regulation of CaMKIV appears to be fulfilled through their opposing effects on CaMKIV Thr 200 phosphorylation status. Our data (Fig.  5), along with previous studies (22,23), indicate that PP2A and CaMKIV may interact in a complex in cells. We explored the possibility that CaMKK might also share this ability to complex with CaMKIV, examining the ability of CaMKK to interact with CaMKIV in co-immunoprecipitation experiments. We found that WT CaMKIV can stably associate with CaMKK (supplemental Fig. S1), as has been previously reported (40,41). However, the transcriptionally inactive T200A CaMKIV mutant also binds CaMKK, similarly to WT CaMKIV, when immunoprecipitated from either unstimulated or ionomycinstimulated cells (supplemental Fig. S1). These data indicate that the interaction between CaMKK and CaMKIV is insufficient to confer CaMKIV transcriptional capability. This hypothesis, that the interaction between CaMKIV and CaMKK is not the primary determinant of CaMKIV transcriptional function, is supported by the behavior of other CaMKIV mutants. For example, N330 also binds CaMKK similarly to WT CaMKIV (supplemental Fig. S1) but is transcriptionally incompetent (Fig. 7C). Thus, CaMKK binding does not preclude the requirement for CaMKIV autonomous activity in CaMKIVmediated transcription. It is possible that the main role of a tight interaction between CaMKIV and CaMKK is to facilitate efficient Thr 200 phosphorylation upon a Ca 2ϩ stimulus. Indeed, it has been reported that an RP-domain mutant of CaMKK, which fails to bind CaMKIV but is not catalytically impaired, is incapable of activating CaMKIV via phosphorylation of the Thr 200 residue, although it phosphorylates other substrates (such as protein kinase B) similarly to wild-type CaMKK (40).
Previous work studying the relationship between PP2A and CaMKIV has indicated that a fraction of the cellular pool of CaMKIV interacts with PP2A in a complex, independent of CaMKIV activation status (23). We have also observed that only a small fraction of CaMKIV in cells appears to interact with CaMKK, based on the relative amounts of these two proteins that co-immunoprecipitate (data not shown), and this interaction occurs even in the absence of a Ca 2ϩ stimulus. Thus, it is fascinating to speculate that some portion of CaMKIV in cells ordinarily exists as a complex with one or both of its regulators. The autonomous activity of CaMKIV within such a complex could thus be very tightly regulated by the opposing activities of bound CaMKK and PP2A.
CaMKIV is an important effector of Ca 2ϩ signaling events, and there are a number of reasons why autonomous activity might be important for its function in transcription. It has become increasingly clear that the amplitude, frequency, duration, source, and localization of Ca 2ϩ signals are critical for determining how increases in intracellular Ca 2ϩ are interpreted by the cell and which Ca 2ϩ -sensitive pathways are activated (42)(43)(44)(45). There has also been a tremendous amount of interest in understanding how short lasting Ca 2ϩ signals lead to long term changes in cellular environment (46,47). Many adaptive changes in response to Ca 2ϩ signals involve gene transcription, and thus transcriptional transducers of Ca 2ϩ signaling are key cellular regulators. Here we show that CaMKIV, once activated, is no longer dependent on Ca 2ϩ /CaM for its activity. This has important implications, since it means that brief elevations in intracellular Ca 2ϩ can lead to an activation of CaMKIV that outlasts the duration of the Ca 2ϩ signal. In this way, CaMKIV may mediate transcription even after the fleeting Ca 2ϩ signal has subsided. Under these circumstances, negative regulation by PP2A may be the primary mechanism for termination of Ca 2ϩ -dependent CaMKIV transcription.
CaMKIV autonomy may have other implications as well. In cells, CaMKIV can be activated by CaMKK after intracellular Ca 2ϩ elevations resulting from treatment with agents such as ionomycin or resulting from physiological stimuli such as CD3mediated activation of T lymphocytes (21). However, it has been unclear where in the cell CaMKIV becomes activated (CaMKIV is primarily nuclear, and CaMKK is primarily cytoplasmic). One possibility is that CaMKIV (unaccompanied or within a complex), cycling in and out of the nucleus, becomes activated in the cytoplasm upon a Ca 2ϩ signal. Activated CaMKIV could then translocate back into the nucleus to potentiate transcription. Because activated CaMKIV would no longer require elevated Ca 2ϩ levels for its activity, the implication is that cytoplasmic Ca 2ϩ signals, even if they never crossed into the nucleus, could lead to activation of CaMKIV and nuclear CaMKIV functions. These possibilities underscore the complexities of Ca 2ϩ signaling and highlight the roles of multiple, converging regulatory mechanisms (Ca 2ϩ /CaM binding, CaMKK activation, and PP2A dephosphorylation) in regulating the Ca 2ϩ /CaM effector, CaMKIV.