A Signaling Complex of Ca2+-Calmodulin-Dependent Protein Kinase IV and Protein Phosphatase 2A

See allHide authors and affiliations

Science  22 May 1998:
Vol. 280, Issue 5367, pp. 1258-1261
DOI: 10.1126/science.280.5367.1258


Stimulation of T lymphocytes results in a rapid increase in intracellular calcium concentration ([Ca2+]i) that parallels the activation of Ca2+-calmodulin–dependent protein kinase IV (CaMKIV), a nuclear enzyme that can phosphorylate and activate the cyclic adenosine monophosphate (cAMP) response element–binding protein (CREB). However, inactivation of CaMKIV occurs despite the sustained increase in [Ca2+]i that is required for T cell activation. A stable and stoichiometric complex of CaMKIV with protein serine-threonine phosphatase 2A (PP2A) was identified in which PP2A dephosphorylates CaMKIV and functions as a negative regulator of CaMKIV signaling. In Jurkat T cells, inhibition of PP2A activity by small t antigen enhanced activation of CREB-mediated transcription by CaMKIV. These findings reveal an intracellular signaling mechanism whereby a protein serine-threonine kinase (CaMKIV) is regulated by a tightly associated protein serine-threonine phosphatase (PP2A).

Cellular responses to external signals require coordinated control of protein kinases and phosphatases; multiple complexes containing both intracellular signaling enzymes are likely to be important for the regulation and specificity of signal transduction pathways. The targeting of protein kinases and phosphatases to specific subcellular compartments, through association with scaffold proteins such as A-kinase anchoring proteins, may contribute to the specificity of cellular signaling (1). However, the enzymes are retained by the anchoring protein in their inactive state (2), and potential regulatory interactions within these multiprotein complexes remain unknown. Preexisting complexes containing active protein kinases and phosphatases provide an alternative and conceptually attractive mechanism by which the appropriate phosphorylation state of intracellular substrates is maintained. The α subunit of casein kinase II (CKII), but not the CKII holoenzyme (containing α and β subunits), associates with the catalytic (C) subunit of PP2A (3). This association was proposed as a mechanism for growth suppression in which CKIIα-stimulated phosphorylation of the C subunit of PP2A would increase the activity of the phosphatase. However, PP2A exists in cells as a heterotrimeric holoenzyme rather than as a free catalytic subunit (4), and phosphorylation of the C subunit of PP2A on Ser-Thr residues in vivo has not been shown.

CaMKIV is important for T cell activation (5). Stimulation of the T cell receptor (TCR) causes increases in [Ca2+]i and activation of CaMKIV. Because CaMKIV phosphorylates the nuclear protein CREB on Ser133, it appears to contribute to increased transcription of immediate early genes containing CRE sequences. The immediate early genes are required for increased expression of the interleukin-2 (IL-2) gene (6). Activation of CaMKIV in T cells is transient, rising to 15 times the background amount by 1 min after TCR stimulation and returning to basal activity within 5 min (7-9). This decline in CaMKIV activity occurs despite a maintained [Ca2+]i in excess of that needed for activation of the enzyme. After binding of Ca2+-calmodulin, CaMKIV is phosphorylated on a single Thr residue in the activation loop by a Ca2+-calmodulin–dependent kinase kinase (CaMKK) (9, 10) and, once activated, CaMKIV activity becomes independent of Ca2+ and calmodulin (8, 9, 11, 12). Because phosphorylation of CaMKIV is required to generate autonomous activity, we reasoned that a protein phosphatase might inactivate CaMKIV even in the presence of increased [Ca2+]i.

Phosphorylated CaMKIV (8) [as well as CREB phosphorylated on Ser133(13)] is an in vitro substrate for PP2A. To determine whether these enzymes exist as a complex in cells, we immunoprecipitated CaMKIV from Jurkat T cells and analyzed the immune complex by immunoblotting with an antibody specific for the C subunit of PP2A. As shown in Fig. 1A, PP2A coimmunoprecipitated with CaMKIV (7). Because both CaMKIV and PP2A proteins are abundant in the brain, we used this tissue to further test whether CaMKIV and PP2A exist as a multiprotein complex. Rat brain soluble extracts were incubated with a glutathione S-transferase–CaMKIV (GST-CaMKIV) fusion protein (9). Both the C and Bα regulatory subunits of PP2A were isolated with wild-type GST-CaMKIV (Fig. 1B). Because the B subunit is variable among holoenzyme preparations of PP2A, these results suggested the presence of a CaMKIV-PP2A holoenzyme complex.

Figure 1

Identification of a CaMKIV-PP2A complex. (A) Coimmunoprecipitation of PP2A with CaMKIV. Immunoprecipitations from Jurkat T cell extracts were performed as described (7) with a rabbit polyclonal antibody raised against the COOH-terminal 17 residues of human CaM KIV (Imm.) or preimmune serum (Preimm.). Immune complexes were subjected to immunoblot analysis with a monoclonal antibody to the C subunit of PP2A (15). (B) Isolation of PP2A with a GST-CaMKIV fusion protein (9). Rat brain soluble extracts were incubated with buffer or wild-type GST-CaMKIV fusion protein (10 μg) for 3 hours at 4°C and then purified with glutathione-Sepharose (17). The resin was extensively washed and bound proteins were eluted with 20 mM glutathione, resolved by SDS-PAGE, and subjected to immunoblot analysis with antibodies to the C and Bα regulatory subunits of PP2A. (C) Binding to calmodulin-Sepharose. Fractions containing proteins of large molecular mass (250 to 700 kD) from gel filtration of rat brain soluble extracts were incubated with calmodulin-Sepharose in the presence (+Ca2+) or absence (–Ca2+) of 5 mM CaCl2 (36). Bound proteins were eluted with 15 mM EGTA, resolved by SDS-PAGE, and subjected to protein immunoblotting with antibodies to the indicated PP2A subunits. (D) Elution of CaMKIV and PP2A from a Superdex-200 gel filtration column after sequential fractionation of brain extracts on phenyl-Sepharose, calmodulin-Sepharose, and Mono Q columns (14). Presented is an immunoblot of fractions from the final gel filtration column showing CaMKIV and the C and Bα subunits of PP2A. Although not shown, the A subunit of PP2A was also present. (E) Calmodulin overlay was performed on fraction 23 of the gel filtration column in the presence of 1 mM Ca2+ or 1 mM EGTA (15).

To independently assess the presence of a complex, we tested whether PP2A holoenzyme bound calmodulin-Sepharose (an affinity resin for CaMKIV). A small fraction (less than 10%) of PP2A holoenzyme (ABαC) from rat brain soluble extracts bound calmodulin-Sepharose in a Ca2+-dependent manner (Fig. 1C). We attempted to copurify CaMKIV and PP2A from rat brain soluble extract by sequential purification on phenyl-Sepharose, calmodulin-Sepharose, Mono Q, and Superdex-200 gel filtration columns (14). Immunoblot analysis of the gel filtration fractions demonstrated that CaMKIV remained associated with the PP2A holoenzyme (Fig. 1D). Calmodulin overlay of the peak gel filtration fraction revealed the presence of a single Ca2+-calmodulin–binding protein that comigrated with CaMKIV (Fig. 1E). The CaMKIV-PP2A complex had an apparent molecular mass of 232 kD (from gel filtration chromatography), and both its size and protein immunoblot analysis (15) indicated that the PP2A heterotrimer and kinase were present in a 1:1 ratio. Furthermore, immune complexes obtained from the peak gel filtration fraction with an antibody specific for CaMKIV contained CaMKIV and PP2A in a 1:1 stoichiometry (7,16).

To determine whether the kinase domain of CaMKIV is sufficient to interact with PP2A, and to test whether kinase activity was necessary for this interaction, we analyzed a series of GST-CaMKIV deletion and point mutants for their ability to interact with PP2A (9, 17). Wild-type and several mutant GST-CaMKIV fusion proteins associated with PP2A, but GST alone (Fig.2A) or glutathione-Sepharose beads did not (18). As evidenced by the interaction of the deletion mutant that contained the entire catalytic domain (residues 1 to 317), association with PP2A did not require the autoinhibitory or calmodulin-binding domains of CaMKIV (Fig. 2A). In contrast, an NH2-terminal deletion mutant of CaMKIV that contained amino acid residues 306 to 474 did not interact with PP2A. Catalytically inactive GST-CaMKIV mutants and a mutant GST-CaMKIV (T200A) not activated by CaMKK (9) associated with PP2A; this demonstrated that neither CaMKIV activity nor its phosphorylation by CaMKK is required for interaction with PP2A.

Figure 2

Independence of association of CaMKIV and PP2A from enzyme activity. (A) Isolation of PP2A with wild-type and mutant GST-CaMKIV fusion proteins (9,17). Rat brain soluble extracts were incubated with the indicated GST fusion proteins (10 μg) and analyzed as described in Fig. 1. Data shown are from a single experiment repeated two to six times with various wild-type and mutant GST-CaMKIV fusion proteins. (B) Isolation of CaMKIV with microcystin-Sepharose. Rat brain or Jurkat T cell soluble extracts were incubated with microcystin-Sepharose in the absence (–) or presence (+) of the PP2A inhibitor microcystin (19). Bound proteins were eluted with SDS-PAGE sample buffer and subjected to immunoblot analysis with antibodies to the C subunit of PP2A and to CaMKIV. The rat brain immunoblot (middle) was stained with Ponceau S (left) to compare proteins eluted from microcystin-Sepharose (15). The locations of the A, Bα, and C subunits of PP2A are marked by arrows. Results are representative of three independent experiments.

To test whether PP2A catalytic activity was required for association with CaMKIV, we used microcystin-Sepharose to purify PP2A from extracts prepared from rat brain and Jurkat T cells (19). Both PP2A and CaMKIV were isolated from the extracts with the resin (Fig. 2B). Because microcystin is an inhibitor of PP2A that binds to the substrate binding site of the phosphatase catalytic subunit (20), PP2A activity appeared not to be required for association with CaMKIV. Association of CaMKIV and PP2A with microcystin-Sepharose was attenuated by pretreatment of the extracts with free microcystin. Hence, the purification of the proteins by the affinity resin occurred in a microcystin-dependent fashion and was not a result of nonspecific association with the beads.

Although association of CaMKIV with PP2A did not require activity of either enzyme, we tested the potential activity of CaMKIV and PP2A toward each other in vitro. The highly enriched complex of CaMKIV-PP2A was incubated in the presence or absence of Ca2+-calmodulin, CaMKK, and the PP2A inhibitor okadaic acid, and the proteins were separated by SDS–polyacrylamide gel electrophoresis (PAGE). The only phosphorylated protein detected by autoradiography appeared to be CaMKIV, as indicated by its size and increased phosphorylation after treatment with CaMKK (Fig. 3). Immunoblot analysis confirmed that the only okadaic acid–sensitive phosphatase present in the preparation was PP2A (18). Thus, the enhanced phosphorylation of CaMKIV observed in the presence of okadaic acid (Fig. 3) indicates that CaMKIV is a substrate for PP2A and suggests that PP2A may also regulate CaMKIV activity in vivo.

Figure 3

Dephosphorylation of CaMKIV by PP2A. A CaMKIV-PP2A preparation (Fig. 1, gel filtration fraction 23) was incubated for 15 min at 30°C in the presence of 10 mM MgCl2, 100 μM ATP, and [γ-32P]ATP (2000 dpm/pmol) and the indicated components (+); CaMKK (0.53 ng/μl) 3 mM CaCl2, 1 μm calmodulin, and 1 μm okadaic acid. EGTA (25 mM) was added, and the incubation was continued an additional 30 min. Reactions were stopped with sample buffer and subjected to SDS-PAGE and autoradiography.

To determine whether PP2A regulates CaMKIV activity in intact cells, we examined the effects of SV40 small t antigen, a specific inhibitor of PP2A activity (21-24), on CaMKIV-mediated activation of CREB-dependent transcription in Jurkat T cells. We used a Gal4-CREB construct together with a 5×(Gal4)-luciferase reporter plasmid because this assay specifically requires phosphorylation of Ser133of the Gal4-CREB molecule. This chimeric protein binds DNA as a monomer and is therefore not influenced by endogenous CREB or other proteins that can heterodimerize with CREB. Previous studies have established that CaMKIV activates transcription by direct phosphorylation of Gal4-CREB on Ser133(25) and that this assay requires phosphorylation of CaMKIV by CaMKK (9, 25, 26). Expression of small t antigen augmented CaMKIV-mediated Gal4-CREB activity that had been elicited by the Ca2+ ionophore ionomycin (Fig. 4) or by the CD3 antibody (16). This effect appeared to be specific for Ca2+-dependent, CaMKIV-mediated activation of Gal4-CREB, as expression of small t antigen had no effect on its own or on Gal4-CREB activity stimulated by constitutively active cAMP-dependent protein kinase, which also specifically and directly phosphorylates Gal4-CREB on Ser133 but in a Ca2+-independent manner (25, 26). These results indicate that the CaMKIV-PP2A complex regulates CaMKIV activity and plays a key role in controlling CRE-mediated gene transcription in T cells.

Figure 4

Regulation of CREB activity in Jurkat T cells. (A) Enhancement of CaMKIV-mediated CREB activity by small t antigen, an inhibitor of PP2A activity. Jurkat cells were transiently cotransfected with 5×(Gal4)-luciferase reporter plasmid (5 μg) and Gal4-CREB expression plasmid (5 μg) together with expression plasmids for CaMKIV (3 μg), small t (1 μg), or constitutively active protein kinase A (PKA, 0.1 μg), as indicated (37). Eighteen hours after transfection, cells were stimulated for 5 hours with buffer (control) or calcium ionophore (1 μM ionomycin) and assayed for luciferase activity. A low concentration of PKA expression plasmid cDNA (0.1 μg) was used in these experiments to achieve a small amount of luciferase activity. Increasing the amount of PKA expression plasmid resulted in a large increase in luciferase activity that also was unaffected by expression of small t (16). Bars represent the relative induction over the value for control plasmid alone, normalized for amount of protein and efficiency of transfection. The values are means ± SE (n = 4). (B) A model for Ca2+-mediated phosphorylation of CREB in T lymphocytes.

In T cells, the Ca2+-calmodulin–dependent dephosphorylation of nuclear factor of activated T cells (NFATc) by calcineurin allows its translocation from the cytosol to the nucleus, where it participates in the induction of IL-2 gene transcription (27-29). Sustained increases in [Ca2+]i during T cell activation maintain NFATc in the nucleus long enough to induce IL-2 transcription; once [Ca2+]i returns to basal amounts, NFAT is rapidly redistributed out of the nucleus to the cytosol (30, 31). Our experiments show how another Ca2+-initiated regulatory event (CaMKIV-stimulated gene transcription) may be inhibited without a decrease in [Ca2+]i. The CaMKIV-PP2A complex would permit rapid dephosphorylation and inactivation of CaMKIV.

The complex of CaMKIV and PP2A exists in resting T cells and is not altered as a function of time after activation (16). How, then, would phosphorylation of CaMKIV and CREB happen at all? Several possible mechanisms may explain the inactivation of CaMKIV in the face of a sustained elevation of [Ca2+]i. First, it is plausible that PP2A is constitutively active toward CaMKIV but that CaMKK transiently outpaces the phosphatase (before becoming inactive itself) to ensure that CaMKIV remains active long enough to phosphorylate CREB. After phosphorylation, a coactivator, phospho-CREB binding protein (CBP), is recruited to the transcription apparatus and CBP binding occurs through the region of CREB that includes Ser133 (32). A second possibility is that PP2A could be inhibited by a Ca2+-independent posttranslational modification such as phosphorylation (33) or by an additional interacting protein (34). Finally, it seems possible that the state of phosphorylation of CaMKIV could, by itself, regulate PP2A activity. Once CaMKIV binds calmodulin and is phosphorylated on Thr200 by CaMKK, it undergoes autophosphorylation on a number of Ser residues in the extreme NH2-terminus (9, 11). The function of some of these modifications is unknown, but could be to regulate activity of PP2A either directly or by recruitment of another protein to the complex.

Regardless of the additional mechanisms involved, our results may reveal a way in which the duration of immediate early gene expression can be regulated independently of [Ca2+]i. Our findings that CaMKIV-PP2A complexes also occur in the brain suggest that similar regulatory mechanisms may contribute to synaptic plasticity and its role in the molecular basis for memory, where the phosphorylation state of CREB is regulated in a Ca2+-dependent manner by CaMKIV (35).

  • * These authors contributed equally to this report.

  • Present address: Howard Hughes Medical Institute, Vollum Institute, Oregon Health Sciences University L-474, 3181 SW Sam Jackson Park Road, Portland, OR 97201, USA.


View Abstract

Navigate This Article