Inhibitory Cross-talk by cAMP Kinase on the Calmodulin-dependent Protein Kinase Cascade*

The calmodulin-dependent kinase (CaM-K) cascade, a Ca2+-triggered system involving phosphorylation and activation of CaM-KI and CaM-KIV by CaM kinase kinase (CaM-KK), regulates transcription through direct phosphorylation of transcription factors such as cAMP response element-binding protein. We have shown previously that activated CaM-KIV can activate the mitogen-activated protein kinases (Enslen, H., Tokumitsu, H., Stork, P. J. S., Davis, R. J., and Soderling, T. R. (1996) Proc. Natl. Acad. Sci. U. S. A. 93, 10803–10808), and the present paper describes a novel regulatory cross-talk between cAMP kinase (PKA) and CaM-KK. PKA gave rapid phosphorylation in vitro and in cells of recombinant CaM-KK, resulting in 50–75% inhibition of CaM-KK activity, part of which was due to suppression of CaM-binding by phosphorylation of Ser458 in the CaM-binding domain. However, the Ser458 → Ala mutant, or a truncation mutant in which the CaM-binding and autoinhibitory domains were deleted, was still partially suppressed by PKA-mediated phosphorylation. The second inhibitory site was identified as Thr108 by site-specific mutagenesis. Treatments of COS-7, PC12, hippocampal, or Jurkat cells with the PKA activators forskolin or isoproterenol gave 30–90% inhibition of either endogenous or transfected CaM-KK and/or CaM-KIV activities. These results demonstrate that the CaM kinase cascade is negatively regulated in cells by the cAMP/PKA pathway.


MATERIALS AND METHODS
Recombinant CaM-KIV from Sf9 cells was purified on CaM-Sepharose (5,6). CaM-KK was either expressed in Escherichia coli and purified on CaM-Sepharose or in COS cells as a FLAG-tagged construct and purified by immunoprecipitation with the FLAG antibody. Standard in vitro CaM-KK phosphorylation reactions contained 5-25 nM PKA, the indicated concentrations of CaM-KK (or CaM-KIV), either 1 mM CaCl 2 , 5 M CaM or 1 mM EGTA, and the other standard protein kinase assay constituents (5). For assays of CaM-KK activity, the reactions contained either 40 g/ml His-CaMKIV 178 -246 (14) or 5 M CaM-KIV (0.08 M was indicated) and 5 M CaM (1 M in Fig. 2C). Reactions were terminated by either addition of SDS sample buffer and resolution on SDS-PAGE (for His CaM-KIV 178 -246 ) or by dilution of reaction aliquots, which were then assayed for CaM-KIV activity using syntide-2 as substrate in the presence of EGTA (5,6). In vitro phosphatase treatments of CaM-KK were performed as described (8). PP1C was a kind gift from Dr. Anna DePaoli-Roach (Indiana University), and human red blood cell PP2A catalytic subunit was from Upstate Biotechnology, Inc. Site-directed mutants of CaM-KK were made using a site-specific plasmid DNA mutagenesis kit (5 Prime 3 3 Prime, Inc., Boulder, CO). 32 P incorporation and CaM binding to CaM-KK were quantitated by densitometry using a computer program "Image" (National Institutes of Health, Research Service Branch).
For studies in intact cells, COS-7 cells were transfected with plasmid (Mock) or CaM-KK (6,14), and PC12 cells (10), primary cultures of hippocampal neurons (15), or Jurkat cells (8) were transfected and/or cultured as referenced. Cells were lysed in buffer A (14) which contains the protein phosphatase inhibitors pyrophosphate (10 mM), NaF (50 * This work was supported by National Institutes of Health Grants GM41292 (to T. R. S.) and Training Grant DK07674 (to G. W.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. mM), sodium vanadate (1 mM), microcystin-LR (1 M), and the transfected or endogenous CaM-KK activities in the 10,000 ϫ g supernatant were assayed. Transfected His-CaM-KIV in COS-7 and PC12 cells was purified on Ni 2ϩ -Sepharose, endogenous Jurkat CaM-KIV was immunoprecipitated (8), and their Ca 2ϩ -independent activities were assayed (5,6).

RESULTS AND DISCUSSION
We first determined whether PKA could phosphorylate in vitro recombinant CaM-KIV or CaM-KK and found that CaM-KK was rapidly phosphorylated by PKA in the absence (Fig. 1A, lane 7 versus 5) or presence (lane 8 versus 6) of Ca 2ϩ /CaM (Fig. 1, A and B) (see "Note Added in Proof"). In the presence of Ca 2ϩ /CaM, CaM-KK also exhibited strong autophosphorylation (Fig. 1A, lane 6 versus 5), which appeared to be additive with PKA-mediated phosphorylation of CaM-KK (Fig.  1, A and B). Tryptic peptide mapping revealed several 32 Plabeled peptides, suggesting that PKA phosphorylated several sites in CaM-KK (not shown). In contrast to a previous report using purified rat cerebellar CaM-KIV (16), we were unable to obtain phosphorylation of recombinant mouse CaM-KIV by PKA (Fig. 1A, lanes 1-4). This is analogous to the previous claim that cerebellar CaM-KIV could be activated by "autophosphorylation" (17), whereas the recombinant CaM-KIV could not (18 -20). Subsequent experiments showed that recombinant CaM-KIV is activated by recombinant CaM-KK (6). Thus, it is possible in the previous report (16) that PKA phosphorylated the contaminating CaM-KK in the rat brain CaM-KIV, and this suppressed the ability of this CaM-KK to subsequently activate the CaM-KIV (see below). Such an effect could easily be interpreted as PKA-mediated inhibition of CaM-KIV.
To investigate whether the PKA-mediated phosphorylation of CaM-KK was regulatory, we determined its effect on CaM-KK activity using two different assays. CaM-KK is usually assayed by its ability to increase total (assayed in the presence of Ca 2ϩ /CaM) and Ca 2ϩ -independent (assayed in the presence of EGTA) activities of CaM-KIV through phosphorylation of the "activation loop" site (Thr 196 ) in CaM-KIV (21). We have previously documented that CaM-KK specifically phosphorylates Thr 196 in a His-tagged construct containing residues 178 -246 of CaM-KIV (14), and Fig. 1C shows that PKA phosphorylation of CaM-KK inhibited its ability to phosphorylate this substrate by approximately 50%. PKA phosphorylation of CaM-KK also inhibited by 45 Ϯ 3.4% the ability of CaM-KK to activate recombinant CaM-KIV, and this inhibition was reversed by subsequent treatment of the CaM-KK with a mixture of protein phosphatases 1 and 2A (Fig. 1D). We were unable to detect any effect of CaM-KK autophosphorylation on its ability to activate CaM-KIV (data not shown).
The biochemical mechanisms involved in PKA-mediated inhibition of CaM-KK are complex and not easily characterized kinetically. Phosphorylation of CaM-KK by PKA in the absence of Ca 2ϩ /CaM (see ''Note Added in Proof'') strongly suppressed binding of CaM to CaM-KK ( Fig. 2A), and this effect was lost when Ser 458 , a consensus PKA phosphorylation site in the CaM-binding domain, was mutated to Ala (Fig. 2B). It should be noted that the biotinylated CaM-overlay technique used in Fig. 2, A and B, is only qualitative, and PKA phosphorylation of CaM-KK does not completely block its activation by Ca 2ϩ /CaM. For example, in the experiment of Fig. 1D where 5 M CaM was used, PKA gave a 45% inhibition of CaM-KK activity, whereas in Fig. 2C inhibition was about 80% with 1 M CaM in the assay. It is well know that the activities of several enzymes can be inhibited due to phosphorylation of their CaM-binding domains, but this mechanism cannot account for all of the CaM-KK inhibition, since PKA still gave significant inhibition of the Ser 458 3 Ala mutant (Fig. 2C). Furthermore, CaM-KK truncated at residue 434, which generates a constitutively active kinase by removing the CaM-binding and autoinhibitory domains (14), was also inhibited 27.2 Ϯ 5.4% by PKA phosphorylation (not shown). This residual inhibition of CaM-KK 1-434 was greater (55% inhibition) at a low substrate concentration of CaM-KIV (0.08 M) than at the standard 5.4 M concentration (28% inhibition), suggesting a K m effect.
We next identified this second site of inhibitory regulation. CaM-KK has an extended NH 2 terminus containing several PKA consensus phosphorylation sites (e.g. Ser 52 and Ser 74 ), but when we expressed CaM-KK 99 -434 , its activity was still inhibited 30 -35% by PKA (not shown), indicating that the second site was in the catalytic domain. We therefore individually mutated the two remaining candidate PKA phosphorylation sites, Thr 108 and Ser 179 , in CaM-KK 1-434 . Although the Ser 179 3 Ala mutant was still inhibited by PKA, the Thr 108 3 Gly mutant was not inhibited (not shown). These two mutations were then made singularly or as the double mutant (T108G/ S458A) in the full-length CaM-KK, and the effects of PKA on their inhibition (Fig. 2C) and phosphorylation (Fig. 2D) were consistent with the conclusion that PKA-mediated inhibition of CaM-KK due to the combined phosphorylations of Thr 108 in the catalytic domain and Ser 458 in the CaM-binding domain. PKA can apparently phosphorylate additional sites, since the double mutant still exhibited some 32 P incorporation (Fig. 2D), but this residual phosphorylation of the double mutant had no apparent effect on CaM-KK activity (Fig. 2C).
We next wanted to ascertain whether activation of PKA in cultured cells would result in inhibition of CaM-KK. COS-7 cells were transfected with wild-type CaM-KK or the double mutant, and the effect of PKA activation was determined upon forskolin treatment for 20 min. Cells were then lysed in the presence of phosphatase inhibitors, and CaM-KK activity in the lysate was assayed for its ability to increase the Ca 2ϩindependent activity of recombinant CaM-KIV (Fig. 3A, left  panel). Extracts from transfected cells exhibited robust CaM-KK activity, which was decreased by forskolin treatment by about 45% for wild-type CaM-KK but not for the double mutant. When cells were 32 P-labeled, forskolin treatment strongly enhanced phosphorylation of wild-type CaM-KK, but forskolin-induced labeling of the double mutant CaM-KK was strongly suppressed compared with wild-type (Fig. 3A, right  panel). Similar activation of CaM-KK was obtained on the endogenous CaM-KK in PC12 cells where the inhibitory effect of forskolin (41%) was mimicked by the cAMP analog cBimps (35%) but not by the inactive forskolin analog 1,9-dideoxyforskolin (Fig. 3B). Strong suppression of endogenous CaM-KK activity was also observed for forskolin treatment of cultured hippocampal neurons (65.9 Ϯ 10.3%, Fig. 3C) and Jurkat cells (52.0 Ϯ 5.4%, Fig. 3D), where isoproterenol produced a rapid (t1 ⁄2 ϭ 30 s), partial suppression (36.0 Ϯ 2.4%) of CaM-KK activity (Fig. 3E). These results clearly demonstrate in a variety of cells that either transfected CaM-KK or endogenous CaM-KK activities could be partially inhibited by agonists known to activate PKA. The selective PKA inhibitor H89 blocked the inhibition of CaM-KK by forskolin treatment in PC12 and Jurkat cells (not shown).
To confirm that the PKA suppression of CaM-KK activity produced a decrease in activation of its downstream target, CaM-KIV, we determined the effect of forskolin treatment on the Ca 2ϩ -dependent activation of either transfected or endogenous CaM-KIV. We have shown previously that COS-7 cells transfected with CaM-KK and CaM-KIV exhibit a rapid Ca 2ϩdependent activation of the CaM-KIV upon treatment with ionomycin (14). This activation is due to phosphorylation of CaM-KIV by CaM-KK, because transfections with either the Thr 196 3 Ala mutant of CaM-KIV plus CaM-KK or by CaM-KIV alone do not exhibit Ca 2ϩ -dependent activation. We used ionomycin in the present study because agonist-dependent Ca 2ϩ mobilization can be modulated by activated PKA, and this would complicate interpretation of results. Fig. 4A shows that the Ca 2ϩ -dependent activation of expressed CaM-KIV in COS-7 cells was blocked about 40% by pretreatment with forskolin (left bars), whereas forskolin had no effect when the CaM-KK double mutant was the catalyst (right bars). With PC12 cells we used the endogenous CaM-KK, but transfected with His-tagged CaM-KIV, because PC12 cells do not contain detectable CaM-KIV. Forskolin pretreatment of the transfected PC12 cells gave a 62.4 Ϯ 0.7% inhibition of the ionomycinstimulated activation of CaM-KIV (Fig. 4B). Last, we used the same treatment protocol on Jurkat cells as they contain both endogenous CaM-KK and CaM-KIV, which can be rapidly activated by stimulation of the CD3 receptor or by ionomycin (8). Ionomycin treatment alone gave a 5-fold activation of CaM-KIV, and this activation was largely blocked by forskolin pretreatment (Fig. 4C).
Based on the results reported herein, we conclude that activation of PKA partially suppresses the activity of CaM-KK, and this inhibition of CaM-KK is transmitted to its downstream target CaM-KIV. These results are consistent with a pattern of extensive cross-talk between the cAMP/PKA and calcium intracellular signaling pathways. For example, both the synthesis and degradation of cAMP are highly regulated by Ca 2ϩ as several adenylate cyclases can be either activated or inhibited by Ca 2ϩ or Ca 2ϩ /CaM (22), and type I phosphodiesterases are stimulated by Ca 2ϩ /CaM (23). Furthermore, CaM-KII can inhibit type III adenylate cyclases (24) and type Ib phosphodiesterase (25), whereas CaM-KIV can inhibit type I adenylate cyclases (26). Conversely, PKA can phosphorylate several CaMdependent enzymes, such as type Ia phosphodiesterase (27) and myosin light chain kinase (28), thereby inhibiting their CaM-binding and enzyme activations, and this same mecha-nism appears to account in part for the inhibition of CaM-KK by PKA reported here. Thus, it is clear that these major signaling systems in cells exhibit extensive positive and inhibitory interactions at numerous sites along their pathways. Because PKA and several of the CaM kinases can phosphorylate numerous substrates, this cross-talk allows fine-tune controls for these complex physiological responses.