Regulation of Protein Phosphatase 2A by Direct Interaction with Casein Kinase 2α

See allHide authors and affiliations

Science  09 May 1997:
Vol. 276, Issue 5314, pp. 952-955
DOI: 10.1126/science.276.5314.952


Timely deactivation of kinase cascades is crucial to the normal control of cell signaling and is partly accomplished by protein phosphatase 2A (PP2A). The catalytic (α) subunit of the serine-threonine kinase casein kinase 2 (CK2) bound to PP2A in vitro and in mitogen-starved cells; binding required the integrity of a sequence motif common to CK2α and SV40 small t antigen. Overexpression of CK2α resulted in deactivation of mitogen-activated protein kinase kinase (MEK) and suppression of cell growth. Moreover, CK2α inhibited the transforming activity of oncogenic Ras, but not that of constitutively activated MEK. Thus, CK2α may regulate the deactivation of the mitogen-activated protein kinase pathway.

Down-regulation of the mitogen-activated protein kinase (MAPK) cascade is crucial to normal growth control. PP2A plays an important role in this process by dephosphorylating the activating site in MAPK as well as in the enzyme that activates MAPK, MEK (MAPK or extracellular signal–regulated kinase kinase) (1). The core PP2A enzyme is a dimer of one catalytic (PP2Ac) and one regulatory (PR65/A) subunit; an additional, variable regulatory (B) subunit binds to PR65 and confers substrate specificity to the dephosphorylating activity (2). The SV40 virus–encoded small t antigen substitutes for one type of B subunit, resulting in a decrease in phosphatase activity toward MEK and an abnormal activation of the mitogenic MAPK cascade (3).

CK2 is a widely expressed, conserved serine-threonine kinase, the signaling function of which is obscure (4). Holoenzymic CK2 is a constitutively active tetramer of catalytic (CK2α) and regulatory (CK2β) subunits; a CK2β-free pool of CK2α also exists (5, 6). The region of small t antigen required for binding of PP2A (3) contains a sequence motif (HENRKL) that is also found between subdomains VIB and VII of the CK2α kinase domain, on what corresponds to a noncatalytic, solvent-exposed loop connecting β strands 7 and 8 in the known kinase structures (7) (Fig.1A). The motif is conserved in CK2α chains fromDrosophila to humans, but not in other kinases. The sequence was mutated as indicated in Fig. 1A, and glutathione-S-transferase (GST) fusion proteins were made with both wild-type and mutant CK2α and tested for binding to purified core PP2A in vitro. PP2A specifically bound to GST-CK2α or to the catalytically inactive mutant GST-CK2αK, but not to the mutant with an altered binding domain, GST-CK2αBD (Fig. 1B). The binding-deficient mutant had the same kinase activity as wild type (8), ruling out a major denaturing effect of the mutation on the structure of the CK2α molecule. Although the region of CK2α required for binding to the CK2β subunit is distinct from the HENRKL motif (9), binding of recombinant CK2β to GST-CK2α prevented subsequent binding of PP2A (Fig. 1C). Thus, PP2A associates with free CK2α, but not with holoenzymic CK2.

Figure 1

(A) Local similarity between vertebrate CK2α and SV40 t antigen. In the binding-deficient CK2α mutant (CK2αBD), residues labeled with asterisks were mutated to alanine (17). Abbreviations for the amino acid residues are as follows: D, Asp; E, Glu; H, His; I, Ile; K, Lys; L, Leu; N, Asn; R, Arg; and Y, Tyr. (B) The indicated GST-CK2αHA fusions were incubated with purified dimeric PP2A, then adsorbed onto glutathione-Sepharose beads (18). After washing, the beads were analyzed by protein immunoblotting with anti-PP2Ac (19). (C) GST-CK2α was incubated with MBP-CK2β (20), then mixed with PP2A at the indicated MBP-CK2β:PP2A molar ratio. After adsorption to beads, the amounts of MBP-CK2β and PP2Ac bound to GST-CK2α were analyzed by immunoblotting with anti-CK2β and anti-PP2Ac. (D) Purified core PP2A was incubated with catalytically inactive or wild-type (WT) GST-CK2α, in the presence of [γ-32P]ATP and okadaic acid (1 μM) (15). After boiling in 1% SDS, PP2Ac was immunoprecipitated and analyzed by SDS-PAGE and autoradiography. (E) The indicated mixtures were preincubated, then assayed for dephosphorylation of Raf-phosphorylated, 32P-labeled His-MEK1 (21). Shown is the mean loss (±SD) in MEK1 phosphorylation, relative to input (n = 3) (22).

In addition to binding, GST-CK2α could partially phosphorylate the PP2A dimer on the PP2Ac subunit (Fig. 1D). However, the stoichiometry of phosphorylation [up to 0.1 mol of inorganic phosphate (Pi) per mole of PP2Ac] varied among different batches of PP2A, apparently because of interference from contaminating proteins (10). In the presence of adenosine triphosphate (ATP), GST-CK2α, but not GST-CK2αK, stimulated PP2A activity by 30 to 50%, when Raf-phosphorylated MEK1 was used as a PP2A substrate (Fig. 1E). Similar results were obtained withp-nitrophenylphosphate as phosphatase substrate (10). Therefore, CK2α-catalyzed phosphorylation appeared to enhance PP2A activity. Because PP2A rapidly dephosphorylated itself, under these conditions its activity could not be quantitatively correlated with its phosphorylation stoichiometry, and the degree of activation may be underestimated.

To determine whether CK2α binds PP2A in vivo, hemagglutinin (HA) epitope–tagged CK2α or CK2αBD was transiently expressed in NIH 3T3 cells, then immunoprecipitated with an antibody to HA (anti-HA), and analyzed by immunoblotting. Coprecipitating PP2A was associated with the wild-type, but not the mutant, CK2α protein (Fig.2A). By 32P labeling the CK2α immunoprecipitate in an autophosphorylation reaction, a band of similar size to PP2Ac was also detected, which could be reimmunoprecipitated with anti-PP2Ac (Fig. 2B). Immunoprecipitates of overexpressed wild-type CK2α contained activity that dephosphorylated MEK1 at a rate significantly higher than activity from CK2αBD immunoprecipitates (Fig. 2C). The phosphatase activity component specific to wild-type CK2α was dose-dependent (Fig. 2D) and was abolished by 10 nM okadaic acid, a specific inhibitor of PP2A (Fig.2E). The MEK1 phosphatase activity associated with catalytically inactive CK2α was lower than that in wild type (Fig. 2E), possibly because CK2α kinase activates PP2A (Fig. 1E). To determine if endogenous CK2α also exists in a complex with PP2A, we immunoprecipitated PP2Ac from resting or growth factor–stimulated NIH 3T3 cells. CK2α was detected in PP2Ac immunoprecipitates from quiescent cells but not from cells that had been treated for 10 min with platelet-derived growth factor (PDGF) (Fig. 3). Thus, CK2α might be a signal-responsive regulator of PP2A, raising the possibility that CK2α might affect MEK1 activity.

Figure 2

Coprecipitation of endogenous PP2A with transiently expressed CK2α. (A) Immunoprecipitates of the indicated HA-tagged protein were analyzed by immunoblotting with anti-PP2Ac (14). Lane 1: Untagged CK2α. Wild-type and mutant CK2α were expressed at similar levels (10). (B) Immunoprecipitates similar to those in (A) were incubated with [γ-32P]ATP (23), then either directly analyzed by SDS-PAGE and autoradiography (lanes 1 to 3) or eluted and reimmunoprecipitated with anti-PP2Ac (lanes 4 and 6) or control antiserum (lane 5). pp37: major autophosphorylation product. Lanes 4′ to 6′: fourfold more intense exposure of lanes 4 to 6. (C) Dephosphorylation of MEK1 by wild-type and binding-deficient CK2α-HA immunoprecipitates. Independent immunoprecipitates were incubated for the indicated times with32P-labeled His-MEK1 (16). His-MEK1 dephosphorylation was quantitated as in Fig. 1E. (D) The indicated numbers of cells were transfected with wild-type or control plasmids, immunoprecipitated with anti-HA, and the CK2α-HA–associated phosphatase activity was measured as the excess MEK1 dephosphorylation obtained with wild-type versus control precipitate (n = 3) (24). (E)32P-labeled His-MEK1 was incubated for 5 min with the indicated anti-HA immunoprecipitates (16), and MEK1 dephosphorylation was measured as above and normalized to the relative amount of immunoprecipitated CK2α-HA (n = 3). OA: 10 nM okadaic acid.

Figure 3

Mitogen-sensitive association between endogenous CK2α and PP2A. PP2Ac was immunoprecipitated from NIH 3T3 cells that had been serum-deprived for 20 hours, then mock-treated (Aand B) or stimulated with PDGF B-B (50 ng/ml) (R&D) (C). The immunoprecipitation in (B) was blocked with excess immunogenic peptide. The immunoprecipitates were resolved on two-dimensional gels, as described (25), and analyzed by protein immunoblotting with anti-CK2α (26). All blots were treated together with identical development of the luminescent reaction. Longer exposure of the blots showed a reduced amount of CK2α in the PDGF-treated condition (10). Arrows: CK2α isoforms.

Recombinant CK2α did not inhibit MEK1 directly (10). However, overexpressed wild-type, but not mutant, CK2α inhibited the serum-stimulated activity of either cotransfected His-tagged MEK1 (Fig.4A) or cotransfected HA-tagged MAPK (10). (The basal activity of MEK1 or MAPK was too low for evaluating its sensitivity to CK2α.) These results suggest that binding of kinase-active CK2α to PP2A may enhance PP2A activity toward MEK1 in vivo.

Figure 4

Association between CK2α and PP2A correlates with inhibition of MEK1 and suppression of cell growth. (A) His-MEK1 plasmid was cotransfected with the indicated CK2α allele. The transfected cells were serum-starved for 24 hours, then restimulated for 10 min with fetal calf serum (20%). The His-MEK1 was purified (27), immunoprecipitated, assayed with catalytically inactive GST-ERK as substrate (11), and quantitated by immunoblotting with either a monoclonal antibody to MEK1 (lanes 1 to 4) or a polyclonal antibody (lanes 5 to 8). (Asterisk: antibody heavy chain). His-MEK1 specific activity was expressed relative to that of the control cotransfection (right). (B) Triplicate plates of 0.5 × 106 cells were transfected with 10 μg of the indicated plasmid together with 1 μg of pGK-Hygro and subjected for 10 to 14 days to hygromycin selection. Surviving clones were stained with Giemsa and counted. (C) Triplicate plates were transfected with 1 μg of either pLXSN-CK2αHA (containing linked CK2αHA and Neor genes; columns 3 and 4) or empty pLXSN (columns 1 and 2), with or without 10 μg of His-MEK1 plasmid. After G418 selection, clones were counted as in (C). (D) Triplicate plates of 1.5 × 106 cells were transfected with 50 ng of control, RasV12, or MEKD218,D222 vector, in the presence of control or CK2α plasmid as indicated and subjected to focus-formation assay. Foci were scored after 2 weeks.

We selected cells that stably overexpressed CK2α. Expression of wild-type CK2α reduced cloning efficiency (Fig. 4B). This effect was partially reverted by cotransfection with wild-type MEK1 (Fig. 4C). In focus-formation assays, the outgrowth of cells transformed with RasV12 [which uses activation of endogenous MEK as an effector (11)] was reduced by about 60% upon cotransfection of CK2α (Fig. 4D). The comparatively weak transforming activity of a constitutively activated MEK1D218,222 mutant (11) was insensitive to CK2α. These results are consistent with the hypothesis that negative regulation of MEK1 is instrumental to the effect of CK2α on growth.

CK2 is required for cell proliferation (12). In transgenic mice, CK2α cooperates with the Myc and Scl oncogenes for the development of lymphomas (13). From our results, however, it appears that CK2α can negatively regulate cell proliferation. One possible explanation for this apparent paradox is that the form of CK2α that binds PP2A, and that inhibits growth, is likely free of CK2β (Fig. 1C) (6) and may thus differ from the growth-promoting form (holoenzymic CK2) (12). Further, because the physiological function of MEK varies with cellular context, the net proliferative effect of CK2α may also differ according to cell type.

  • * Present address: Institut Suisse de Recherche sur le Cancer, 155 chemin des Boveresses, CH-1066 Epalinges, Switzerland.


View Abstract

Navigate This Article