The Regulatory b Subunit of Protein Kinase CK2 Mediates Formation of Tetrameric CK2 Complexes*

Protein kinase CK2 is a tetrameric enzyme composed of two catalytic (a and/or a*) subunits and two regulatory (b) subunits. Because CK2b is synthesized in excess of CK2a, we hypothesized that formation of CK2b homodimers precedes the incorporation of the catalytic subunits of CK2 into complexes. To test this hypothesis, we cotransfected cells with two epitope-tagged variants of CK2b. The results of these cotransfection studies demonstrate that interactions between two CK2b subunits take place in the absence of CK2a. Together with results from previous biosynthetic labeling studies, these results suggest that formation of CK2b homodimers occurs before incorporation of catalytic subunits of CK2 into CK2 complexes. We also cotransfected Cos-7 cells with a deletion fragment of CK2b (i.e. Myc-b1–166) together with full-length hemagglutinin (HA)-tagged CK2b and/or CK2a*. Although complexes between Mycb1–166 and HA-b were readily detected, we obtained no evidence of direct interactions between Myc-b1–166 and HA-CK2a*. These results suggest that residues within the N-terminal 166 amino acids of CK2b are sufficient for interactions between CK2b subunits, whereas the C-terminal domain of CK2b is required for complex formation with the catalytic subunits of CK2. Finally, we observed that expression of full-length HA-b promotes phosphorylation of Myc-b1–166 by HA-CK2a*.

Protein kinase CK2 1 (formerly known as casein kinase II) is a protein serine/threonine kinase involved in various aspects of cellular regulation (1)(2)(3)(4). The enzyme is ubiquitously distributed in eukaryotic organisms and is essential for viability (5)(6)(7). There have been a number of reports that CK2 is overexpressed in tumors or in leukemic cells (8 -11). Furthermore, the dysregulated expression of CK2 in the lymphocytes of transgenic mice results in lymphocyte transformation. In these transgenic mice, CK2 exhibited cooperation with c-Myc or with Tal-1 in transformation (12,13). Collectively, these results suggest that CK2 is a regulatory component of the protein kinase networks that regulate the growth and division of cells.
There is a significant body of evidence demonstrating that CK2 phosphorylates and regulates the activity of numerous proteins involved in the control of various aspects of cellular function (1)(2)(3)(4). However, the regulation of CK2 in intact cells remains an area of considerable controversy (14). In general, there is a consensus that the majority of CK2 in cells exists as a tetrameric complex composed of two catalytic (designated ␣ and/or ␣Ј) and two noncatalytic ␤ subunits (1)(2)(3)(4). The catalytic subunits contain all of the conserved consensus motifs for members of the protein kinase family. By comparison, in mammals, the ␤ subunit of CK2 does not share extensive homology with any known proteins. It is noteworthy that CK2␤ exhibits remarkable conservation between species. In fact, the deduced amino acid sequences of human and chicken CK2␤ are identical. The deduced sequences of Xenopus CK2␤ and zebrafish CK2␤ differ from those of human and chicken CK2␤ by only one and two amino acids, respectively (4). This high level of evolutionary conservation suggests that CK2␤ has important cellular functions. By itself, CK2␤ has no known catalytic activity, but it appears to regulate the activity of CK2␣ in a number of ways. In particular, CK2␤ stabilizes CK2␣ and modulates the ability of CK2␣ to interact with and phosphorylate substrate proteins (15)(16)(17)(18). Furthermore, CK2␤ appears to mediate the activating effects of compounds such as polyamines that may have a role in regulating CK2 in cells (19,20). Overall, these results suggest that CK2␤ is a critical mediator of the cellular functions of CK2.
To fully define the cellular functions of CK2 and to understand its regulation, a thorough understanding of the functions of each of its subunits is critical. In recent studies using the yeast two-hybrid system, we and others demonstrated that CK2␤ has the ability to interact with CK2␣ or with CK2␤, whereas CK2␣ is only able to interact with CK2␤ (21)(22)(23). These results suggested that interactions between two ␤ subunits are responsible for bringing two ␣:␤ dimers into active tetrameric complexes. However, because yeast contain endogenous CK2, it was not possible to exclude the possibility that indirect interactions between two CK2␤ subunits were being mediated by endogenous CK2␣. From biosynthetic labeling studies, we had also demonstrated that CK2␤ is synthesized in excess of CK2␣ (24). Furthermore, these studies demonstrated that newly synthesized CK2␣ is rapidly incorporated into complexes with CK2␤, whereas newly synthesized CK2␤ associates more slowly into complexes with CK2␣. Consequently, we hypothesized that formation of complexes between CK2␤ subunits precedes the incorporation of catalytic subunits into tetrameric complexes. In the present study, we provide evidence from transfection studies using two epitope-tagged versions of CK2␤ (i.e. CK2␤ with an N-terminal HA epitope designated HA-CK2␤ and CK2␤ with an N-terminal Myc epitope designated Myc-CK2␤) to support this hypothesis. Furthermore, toward the objective of controlling the activity of CK2 within cells by controlling its ability to form tetrameric complexes, we * This work was supported by grants from the National Cancer Institute of Canada with funds from the Canadian Cancer Society and the Terry Fox Foundation. 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.
‡ have also examined the ability of individual domains of CK2␤ to interact with full-length CK2␣ and/or CK2␤ in cells. In this regard, we have demonstrated that an N-terminal fragment of CK2␤ encoding residues 1-166 retains the ability to interact with CK2␤ in cells but fails to directly form stable complexes with CK2␣Ј. These results indicate that residues required for stable interactions between two CK2␤ subunits in cells are localized within the N-terminal 166 residues of CK2␤. By comparison, stable interactions between CK2␤ and the catalytic subunits of CK2 require additional residues. Importantly, our results suggest that deletion mutants of CK2␤ such as CK2␤1-166 offer new strategies for controlling the functions of CK2 in cells by altering the subunit composition of CK2 in cells in a manner that prevents formation of intact tetrameric CK2 complexes.
Plasmids Constructs-Full-length HA-CK2␣Ј and Myc-CK2␤, containing N-terminal HA or Myc epitope tags, respectively, were expressed using pRc/CMV as described previously (18). Full-length CK2␤ containing an N-terminal HA tag was constructed using a strategy similar to that used for construction of HA-CK2␣Ј. Briefly, a NotI site was introduced at the 5Ј end of the coding region of the CK2␤ cDNA using the polymerase chain reaction with the following primers: sense (ATA AGA ATG CGG CCG CAG CAG CTC AGA GGA G) and antisense (CTT GGG GCA GTA GAG CTT). The amplified sequence was subcloned into the TA vector (Invitrogen) and verified by sequencing (29). Myc epitope-tagged CK2␤ 1-166 was amplified from full-length Myctagged CK2␤ in plasmid pZW12 using the sense primer (CTG GCG GCC GCT CTA GAA CTA G) and the antisense primer (GGG GCC CCT ACA TGT GAG GGA AAC CAG) to introduce a stop codon followed by an ApaI site. The construct was verified by sequencing and cloned into pRc/CMV as a NotI/ApaI fragment. The vector pRc/CMV is suitable for transfection of mammalian cells using the CMV promoter to direct protein expression.
Transfection and Biosynthetic Labeling of Cos-7 Cells-Cos-7 cells were maintained in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% fetal calf serum (Intergen) at 37°C in an atmosphere of 5% CO 2 . Cells were transfected using the calciumphosphate coprecipitation method as described previously (18, 34) using 25-50 g of total DNA/10-cm plate. Unless otherwise indicated, cotransfections were performed by adding equivalent amounts of each plasmid. Cells were washed thoroughly with phosphate-buffered saline 12-16 h after addition of the calcium phosphate precipitate and were harvested 24 -48 h after removal of the precipitate. For biosynthetic labeling, transfected cells were washed twice with phosphate-buffered saline and were then incubated in prewarmed methionine and cysteinefree medium (ICN) for 10 min. Cells were then radiolabeled by the addition of 35 S-EasyTAG TM (NEN Life Science Products) protein labeling mix (0.25 mCi/10-cm dish) where indicated.
Immunoprecipitation of Transfected Proteins-For immunoprecipitation of expressed proteins, cells were lysed with Nonidet P-40 lysis buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 1% Nonidet P-40) for immune complex kinase assays or with L buffer (phosphate-buffered saline with 1% Nonidet P-40, 1% deoxycholic acid) for analysis of co-immunoprecipitating proteins (18). For cell extraction, lysis buffers were supplemented with protease inhibitors (1% aprotinin, 0.1 mM phenylmethylsulfonyl fluoride, 10 g/ml leupeptin). Monoclonal antibodies (12CA5 or 9E10) were covalently attached to protein A-Sepharose as described (30) and incubated with lysates for 1 h with rocking on ice. For peptide competition experiments, 20 g of the appropriate antigenic peptide was added for each l of antibody. The Sepharose beads were then washed with Nonidet P-40 lysis buffer and then three times with kinase buffer (50 mM Tris-Cl, pH 7.4, 10 mM MgCl 2 , 1 mM dithiothreitol). After resuspending the washed immune complexes in kinase buffer, CK2 activity was measured using the synthetic peptide RRRDDDSDDD as a CK2 substrate (18). Kinase reactions contained 50 mM Tris-Cl, pH 7.4, 150 mM NaCl, 10 mM MgCl 2 , 0.1 mM [␥-32 P]ATP (specific activity, 200 -500 cpm/pmol) and synthetic peptide (0.1 mM). Reactions were initiated by the addition of immune complex and were terminated after 20 min at 30°C by spotting an aliquot of the reaction mix on phosphocellulose paper (P81, Whatman) as described (35). Immune complexes that were prepared using L buffer for the analysis of co-immunoprecipitating proteins were washed three times with L buffer, and bound proteins were then solubilized with Laemmli sample buffer.
Size Exclusion Chromatography-Briefly, transfected and nontransfected Cos-7 cells were lysed in low salt lysis buffer (20 mM Tris, pH 7.5, 150 mM NaCl, 1 mM dithiothreitol, 0.2% Nonidet P-40) and sonicated twice for 10 s on ice, and cell debris was cleared by centrifugation for 20 min at 55,000 rpm. Cell lysate (200 l) was separated using a Superose 12 column on an fast protein liquid chromatography system (Amersham Pharmacia Biotech). Fractions were analyzed by Western blotting using anti-HA, anti-Myc, anti-␣, and anti-␤ antibodies.
Other Procedures-SDS-PAGE (31) and immunoblotting (32) were performed as described previously (18) with the following exceptions. For immunoblots, proteins were transferred to polyvinylidene difluoride membranes (Bio-Rad), and immune complexes were detected using secondary antibodies conjugated to horseradish peroxidase for detection by enhanced chemiluminescence using Supersignal (Pierce) as substrate or with secondary antibodies conjugated to alkaline phosphatase for color development with bromochloroindoyl phosphate and nitro blue tetrazolium as substrates. Protein determinations were performed according to the method of Bradford using bovine serum albumin as standard (33).

Complexes between Regulatory CK2␤ Subunits Form in the
Absence of CK2␣ in Cells-Based on previous observations using the yeast two-hybrid system (21)(22)(23) and with support from cross-linking studies (37,38), it had been demonstrated that the regulatory subunit of CK2 (i.e. CK2␤) interacts with CK2␣ and with CK2␤. To determine whether complexes between two CK2␤ subunits form within cells prior to the formation of complexes with CK2␣, we prepared constructs encoding two different epitope-tagged constructs of CK2␤, HA-CK2␤, and Myc-CK2␤ ( To verify that HA-CK2␤ and Myc-CK2␤ are functional with respect to their capacity to interact with CK2␣, we demonstrated that either HA-CK2␤ (Fig. 2B) or Myc-CK2␤ (Fig. 2C) can be detected in specific anti-CK2␣ 379 -391 immunoprecipitates. Moreover, we also demonstrated that the HA-CK2␤ and Myc-CK2␤ signals are specific to complexes containing endogenous CK2␣ by performing immunoprecipitations in the presence of antigenic ␣ 376 -391 peptide (Fig. 2, A-C). As previously noted (18), the HA-CK2␤ or Myc-CK2␤ bands of different electrophoretic mobility most likely reflect different phosphorylated forms of the proteins. Based on this assumption, it ap-pears that a significant proportion of HA-CK2␤ and Myc-CK2␤ are phosphorylated in these complexes that are isolated by immunoprecipitation with anti-CK2␣ antibodies. The observation that kinase activity toward a CK2-specific synthetic peptide substrate is detected in the appropriate anti-HA (Fig. 2D) or anti-Myc (Fig. 2E) immunoprecipitates is also indicative of complex formation between the transfected proteins and endogenous catalytic subunits of CK2. We do not have a precise explanation for the significant differences observed for the kinase activities of anti-HA (Fig. 2D) or the anti-Myc (Fig. 2E) immunoprecipitates. However, the differences may in part reflect differences in immunoprecipitation efficiency or perhaps conformational differences between the two antibodies.
Having verified functional expression of HA-CK2␤ and Myc-CK2␤ and specificity of isolation and detection, we subsequently cotransfected the two constructs into cells (Fig. 3). We were primarily interested in determining whether the two versions of CK2␤ are capable of interacting with each other in cells in the absence of CK2␣. Consequently, two preliminary rounds of immunoprecipitations were carried out using a mixture of anti-CK2␣ and anti-CK2␣Ј antibodies to deplete extracts of tetrameric CK2 complexes. These depleted extracts were subsequently immunoprecipitated with either anti-HA or anti-Myc antibodies (Fig. 3). Immunoprecipitations were also performed with the mixture of anti-CK2␣ and anti-CK2␣Ј to verify that extracts had indeed been depleted of tetrameric complexes and with anti-CK2␤ antibodies to verify that the tagged proteins do FIG. 1. Immunoprecipitation of Myc-CK2␤ and HA-CK2␤. A, constructs encoding CK2␤ with an N-terminal Myc epitope that is reactive with 9E10 monoclonal antibodies (i.e. Myc-␤) and with an N-terminal HA epitope that is reactive with 12CA5 monoclonal antibodies (i.e. HA-␤) are illustrated. The epitope at the C terminus of CK2␤ that is recognized by anti-CK2␤ 198 -215 antibodies is also indicated. HA-CK2␤ is a protein of 254 amino acids including amino acids 2-215 of CK2␤. The 40-amino acid tag at the N terminus of this protein contains a triple repeat of the YPYDVPDY epitope sequence derived from influenza virus hemagglutinin that is recognized by 12CA5 monoclonal antibodies. By comparison, Myc-CK2␤ is a protein of 241 amino acids that includes a 26-amino acid N-terminal sequence with one copy of the EQKISEEDL epitope sequence derived from the c-Myc protein that is recognized by 9E10 monoclonal antibodies as well as amino acids 1-215 of CK2␤. A schematic representation of endogenous CK2␤ is also illustrated. B, Cos-7 cells were transfected with either pRc/CMV encoding HA-␤ (lanes 1-3) or Myc-␤ (lanes 4 -6) and were then labeled with [ 35 S]methionine/cysteine for 20 min as described under "Materials and Methods." Extracts were prepared and immunoprecipitations performed with 12CA5 antibodies (lanes 1 and 4), with 9E10 antibodies (lanes 2 and 5) or with anti-CK2␤ 198 -215 antibodies (lanes 3 and 6). HA-␤ and Myc-␤, which exhibit similar electrophoretic mobilities, are indicated. Endogenous CK2␤ is also indicated. C, cells were transfected and immunoprecipitated as in B without radiolabeling. Immunoprecipitates were analyzed on immunoblots with 12CA5 antibodies. D, cells were transfected and immunoprecipitates prepared and analyzed on immunoblots using 9E10 antibodies. The positions of HA-CK2␤ or Myc-CK2␤ are indicated in C and D, respectively. The position of the L chain of IgG is also indicated. Immunoblots were analyzed using 12CA5 antibodies. The nonphosphorylated and phosphorylated forms of HA-␤ are indicated. C, cell extracts were prepared and immunoprecipitated as in A. Immunoblots were analyzed using 9E10 antibodies. The nonphosphorylated and phosphorylated forms of Myc-␤ are indicated. D, cell extracts were prepared as in A and immunoprecipitated with 12CA5 antibodies. Immune complex kinase assays were performed as described under "Materials and Methods." E, cell extracts were prepared and immunoprecipitates were performed using 9E10 antibodies in preparation for immune complex kinase assays.

FIG. 3. Co-immunoprecipitation of Myc-CK2␤ and HA-CK2␤.
Cos-7 cells were transfected with pRc/CMV (50 g/10-cm dish) or with mixtures of pRc/CMV constructs encoding Myc-CK2␤ and HA-CK2␤ in the amounts indicated (g/10-cm dish). Extracts from transfected cells were subjected to two successive rounds of immunoprecipitation with a mixture of anti-CK2␣ 376 -391 and anti-CK2␣Ј 333-350 antibodies to deplete extracts of tetrameric CK2 complexes. The depleted extracts were then immunoprecipitated with 9E10 antibodies (first three lanes) or with 12CA5 antibodies (last three lanes) as indicated. Immunoprecipitates were electrophoresed on replicate gels and were then transferred to polyvinylidene difluoride membrane for immunoblotting with 12CA5 indeed encode CK2␤ (data not shown). Examination of Fig. 3 clearly demonstrates that HA-CK2␤ can be detected in anti-Myc immunoprecipitates (Fig. 3A) and vice versa (Fig. 3B). Again, bands of different electrophoretic mobility most likely reflect differences in phosphorylation state as previously noted (18). However, it is notable that, unlike Fig. 2 where a significant proportion of the HA-CK2␤ and Myc-CK2␤ exhibits retarded electrophoretic mobility characteristic of phosphorylation, the majority of the HA-CK2␤ and Myc-CK2␤ observed in Fig. 3 appears to be unphosphorylated. Small amounts of phosphorylated CK2␤, most evident for HA-␤, could reflect the isolation of small quantities of tetrameric CK2 that was not removed during the two rounds of immunodepletion with anti-CK2␣ and anti-CK2␣Ј antibodies. The amount of Myc-CK2␤ that is found in anti-HA immunoprecipitates and vice versa is much greater than the amount of Myc-CK2␤ or HA-CK2␤ that is detected in anti-CK2␣/CK2␣Ј immunoprecipitates (data not shown). Collectively, these data suggest that the complexes between Myc-CK2␤ and HA-CK2␤ have formed in the absence of a catalytic CK2 subunit.
In a previous study (18), we had demonstrated that HA-CK2␣Ј and Myc-CK2␤ form complexes that can be isolated by co-immunoprecipitation assays. Furthermore, those results indicated that Myc-CK2␤ is readily autophosphorylated upon formation of complexes with HA-CK2␣Ј, whereas in the absence of complex formation, Myc-CK2␤ remained unphosphorylated (18). Therefore, to demonstrate that complexes between HA-CK2␤ and Myc-CK2␤ are capable of forming tetrameric CK2 complexes containing catalytic subunits, we next performed transfections of HA-CK2␤ and Myc-CK2␤ in the presence or absence of HA-CK2␣Ј (Fig. 4). Myc-CK2␤ was isolated by immunoprecipitation with 9E10 antibodies, and immunoprecipitates were examined by immunoblots using 12CA5 antibodies (Fig. 4) to detect the presence of co-precipitating HA-CK2␣Ј or HA-CK2␤. As expected, either HA-CK2␣Ј (Fig. 4, lane  1) or HA-CK2␤ (Fig. 4, lane 2) can be isolated in complexes with Myc-CK2␤. Furthermore, both HA-CK2␤ and HA-CK2␣Ј are detected in anti-Myc immunoprecipitates when cells are cotransfected with plasmids encoding Myc-CK2␤ together with HA-CK2␣Ј and HA-CK2␤ (Fig. 4, lane 3). Importantly, when isolated in the latter complexes, HA-CK2␤ exhibits a profile similar to that observed in Fig. 2B and appears to be phosphorylated. Because autophosphorylation is an intramolecular process, this result indicates that HA-CK2␤ is part of a complex that contains HA-CK2␣Ј, which provides kinase activity, and with Myc-CK2␤, which provides the Myc epitope that is used for isolation of the complex. By comparison, in complexes lacking HA-CK2␣Ј (Fig. 4, lane 2), HA-CK2␤ does not appear to be phosphorylated. In cells that were transfected solely with plasmid encoding Myc-CK2␤ (Fig. 4, lane 4), neither HA-CK2␣Ј nor HA-CK2␤ are detected. An identical blot probed with anti-Myc confirmed the presence of Myc-CK2␤ in all lanes (data not shown). Overall, these observations provide further support that the epitope-tagged CK2 subunits are competent for the formation of intact CK2 complexes.
Interactions between CK2␤1-166 and CK2␣Ј or CK2␤ in Transfected Cos-7 Cells-Results from a number of laboratories including our own have led to the elucidation of functional domains of CK2␤ (15, 19, 22, 23, 38 -40). These results were obtained primarily by in vitro reconstitution of bacterially expressed proteins or through analyses using the yeast twohybrid system. Overall, these result suggest that CK2␤ forms stable interactions with CK2␣ through residues within its Cterminal domain and that CK2␤ interacts with another CK2␤ subunit through residues with its N-terminal domain. However, there are discrepancies regarding the precise domain(s) of CK2␤ that are responsible for stable interactions with another CK2␤. From the studies of Kusk et al. (23), the region of CK2␤ that was required for homodimerization was mapped to residues 1-145 of CK2␤, whereas the data of Boldyreff et al. (22) suggested that residues 155-165 were required for homodimerization of CK2␤. Our own studies with the yeast two-hybrid system were inconclusive (39). To extend the results of our own studies, we were interested in further examining the interaction domains of CK2␤ in intact mammalian cells. Importantly, we sought to examine these interactions using deletion products with small Myc epitope tags instead of the relatively large fusions that are used in the two-hybrid system. One objective of these studies is to develop strategies for preventing the formation of tetrameric complexes or for altering the subunit composition of CK2 complexes. In initial transfection experiments, we assessed the feasibility of expressing, in Cos-7 cells, a deletion construct of CK2␤ (designated Myc-CK2␤1-166) that contains the elements defined by both Kusk et al. (23) and Boldyreff et al. (22) that are required for interactions between two CK2␤ subunits (Fig. 5). As seen in Fig. 5, Myc-CK2␤1-166 (lane 2) was expressed at a level comparable with that of full-length Myc-CK2␤ (lane 1). We also transfected cells with constructs encoding were transfected into Cos-7 cells using the vector pRc/CMV. A control transfection using pRc/CMV without insert was also performed (lane V). Following transfection, cells were biosynthetically labeled with [ 35 S]methionine/cysteine as described under "Materials and Methods" for 6 h. Cell extracts were prepared, and immunoprecipitates were performed with 9E10 antibodies. Immunoprecipitates were subjected to SDS-PAGE on 15% gels, and immunoprecipitated proteins were visualized with a PhosphorImager.
Myc-CK2␤1-131, Myc-CK2␤132-215, Myc-CK2␤132-200, and Myc-CK2␤152-215, but all of these constructs were expressed at levels dramatically lower than that observed with Myc-CK2␤1-166 (data not shown). We do not have any data concerning the low level of expression of the other deletion products and did not further examine them in mammalian cells.
Because Myc-CK2␤1-166 was expressed at levels close to those seen with wild-type CK2␤, we cotransfected this construct with either HA-CK2␤ or with HA-CK2␣Ј to examine its ability to form complexes with the other subunits of CK2 in cells. We opted to use HA-CK2␣Ј for these experiments rather than HA-CK2␣ because we had previously observed higher levels of expression with HA-CK2␣Ј (18). Furthermore, CK2␣ and CK2␣Ј are very closely related, and there are no obvious indications that they exhibit any differences in their ability to interact with the regulatory CK2␤ subunit (21). Following cotransfection, Myc-CK2␤1-166 was isolated from cell extracts by immunoprecipitation using 9E10 antibodies, and immunoprecipitates were subsequently analyzed for the presence of co-precipitating HA-CK2␣Ј or HA-CK2␤. As shown in Fig. 6A To further study the formation of CK2 complexes, transfections were performed with Myc-CK2␤ or Myc-CK2␤1-166 and both HA-CK2␤ and HA-CK2␣Ј. Anti-myc immunoprecipitates were performed and analyzed by immunodetection with anti-HA antibodies. To avoid the appearance of the immunoglobulin from immune complexes (as seen in Fig. 6A), these blots were developed using biotinylated anti-HA antibodies that were detected directly with peroxidase-conjugated anti-biotin antibodies. Using these biotinylated antibodies, interactions . Consistent with what is observed in lane 7, the ratio of immunoreactive HA-CK2␣Ј to immunoreactive HA-CK2␤ is expected to be 2:1 in these complexes. Furthermore, in accordance with the results of Fig. 4, a significant proportion of the HA-CK2␤ in these complexes appears to be phosphorylated.
Anti-myc immunoprecipitates were also performed from ex- CK2␤ (i.e. expected to be two catalytic subunits per complex). Alternatively, it is possible that two intact CK2␤ subunits are necessary for the formation of complexes with catalytic CK2 subunits that are stable under the conditions of immunoprecipitation used in this study.
In the experiments shown in Fig. 6, the detection of Myc-CK2␤1-166 in anti-Myc immunoprecipitates was confounded by the light chain of IgG (data not shown). Therefore, as an alternative to immunoblots, we also performed biosynthetic labeling experiments. As shown in Fig. 7, with full-length Myc-CK2␤ (Fig. 7A), autophosphorylation is evident when it is coexpressed with either HA-CK2␣Ј (lane 2) or with a combination of HA-CK2␣Ј and HA-CK2␤ (lane 4). Autophosphorylation of Myc-CK2␤ is not evident when Myc-CK2␤ is cotransfected with the control vector (lane 1) or with HA-CK2␤ (lane 3), indicating that the catalytic subunits of CK2 are required for autophosphorylation and that it does not serve as a substrate for endogenous CK2. With Myc-CK2␤1-166 (Fig. 7B), autophosphorylation is not evident when it is cotransfected by itself or with either HA-CK2␤ or HA-CK2␣Ј (lanes 1-3). However, autophosphorylation of Myc-CK2␤1-166 is evident when co-expressed with HA-CK2␤ and HA-CK2␣Ј, demonstrating that HA-CK2␤ promotes autophosphorylation of Myc-CK2␤1-166. The shift in electrophoretic mobility that was observed with Myc-CK2␤1-166 when co-expressed with HA-CK2␣Ј and HA-CK2␤ was not observed when a kinase-inactive mutant of HA-CK2␣Ј was utilized (data not shown). This result confirms that the shift in mobility of Myc-CK2␤1-166 is the result of phosphorylation. DISCUSSION Through the examination of complexes that are formed between transfected CK2 subunits in Cos-7 cells, we have obtained evidence that complexes between two CK2␤ subunits can form in cells in the absence of CK2␣. These results are consistent with the interpretation that complexes between CK2␤ subunits precede incorporation of catalytic CK2 subunits into tetrameric complexes. Our previous biosynthetic labeling studies are also consistent with this interpretation (24) because in that study we demonstrated that CK2␤ is synthesized in excess of CK2␣. Furthermore, by following the fate of newly synthesized CK2 subunits (24), we demonstrated that CK2␣ is rapidly incorporated into complexes with CK2␤. By comparison, incorporation of CK2␤ into complexes with CK2␣ occurs more slowly. Under some circumstances (i.e. if CK2␣ were more abundant than CK2␤) it may be possible that the assembly of tetrameric CK2 complexes forms by a different mechanism (i.e. CK2␣:CK2␤ heterodimers precede tetramer assembly). However, because our biosynthetic labeling studies suggest that CK2␤ is synthesized in excess of CK2␣ (24) and because this study demonstrates that complexes between CK2␤ subunits form in the absence of catalytic subunits, we believe that the most likely interpretation of the data is that formation of CK2␤ complexes precedes incorporation of catalytic CK2 subunits into tetrameric complexes. Importantly, these results suggest that the recent crystal structure of a dimeric form of the regulatory CK2␤ subunit may indeed represent a physiologically relevant structure (41) because it does not appear from the data presented here that catalytic CK2 subunits are required for the formation of complexes between two CK2␤ subunits.
By examining the formation of complexes between deletion products of CK2␤ and full-length CK2␣Ј or CK2␤ in cells, we have identified a deletion product of CK2␤ (i.e. CK2␤1-166) that retains the ability to interact with CK2␤ but fails to stably interact directly with CK2␣Ј. The former results are consistent with the recent crystal structure of a dimeric form of CK2␤ (41). From that study, it has been demonstrated that interactions between CK2␤ subunits are mediated by a zinc finger that is composed of Cys 109 , Cys 114 , Cys 137 , and Cys 140 . Moreover, our results may also have broader implications regarding the assembly of CK2 in cells. Although definitive conclusions may not be possible until detailed three-dimensional structural information for tetrameric CK2 is available, our results indicate that a single intact CK2␤ subunit is sufficient to recruit a catalytic subunit into a multi-subunit CK2 complex that allows intramolecular autophosphorylation of CK2␤.
It is intriguing that full-length HA-CK2␤ promotes phosphorylation of Myc-CK2␤1-166 (Fig. 7). One possible explanation for this observation is that the intramolecular autophosphorylation of CK2␤ that occurs involves phosphorylation by the catalytic subunit that is associated with the other CK2␤ subunit (i.e. trans-phosphorylation). This suggestion is based on the observation that Myc-CK2␤1-166 does not directly interact with HA-CK2␣Ј. When Myc-CK2␤1-166 forms a complex with full-length CK2␤, a catalytic CK2 subunit may then be recruited to the complex with a resultant autophosphorylation of Myc-CK2␤1-166. However, we have not excluded other possibilities. For example, we do not know the nature of the complexes that form between Myc-CK2␤1-166: HA-CK2␤ and HA-CK2␣Ј. One might expect that these complexes would be trimeric (i.e. containing only one catalytic subunit) because Myc-CK2␤1-166 does not directly interact with HA-CK2␣Ј. However, because we observe that full-length HA-CK2␤ is also phosphorylated in complexes containing Myc-CK2␤1-166, it is possible that the one catalytic subunit has sufficient mobility to autophosphorylate both Myc-CK2␤1-166 and HA-CK2␤ or alternatively, it is possible that the complexes do in fact contain two catalytic subunits. Our attempts to resolve this issue by gel filtration on an fast protein liquid chromatography were inconclusive because we observed that complexes that formed between catalytic and regulatory subunits of CK2 migrated as complexes with apparent molecular weights greatly exceeding that expected for trimeric or tetrameric CK2 complexes (data not shown). As recently noted by Guerra and Issinger (42), this observation likely reflects the fact that CK2 interacts with a variety of cellular proteins to form high molecular weight complexes.
Overall, although there are unresolved issues regarding the nature of the complexes that are formed with Myc-CK2␤1-166, our data suggest that a single full-length CK2␤ subunit is sufficient to recruit catalytic subunits into complexes that allow autophosphorylation. It is also interesting to note that a transphosphorylation reaction is responsible for the autophos- phorylation reactions that are observed upon dimerization of a number of receptor tyrosine kinases (including the insulin, epidermal growth factor, and nerve growth factor receptors) (43)(44)(45)(46). Consequently, although we have not excluded other possibilities, it is plausible that the autophosphorylation of a protein serine/threonine kinase such as CK2 is mechanistically similar to that of receptor tyrosine kinases.
As yet, the functional significance of the autophosphorylation of CK2␤ remains undefined. However, on the basis that kinase-inactive CK2 is expressed in cells at lower levels than enzymatically active CK2, we speculated that autophosphorylation may enhance the stability of CK2 (18). In this regard, it is interesting that we observe with Myc-CK2␤ or Myc-CK2␤1-166 that apparent levels of expression are higher under situations where they are autophosphorylated than under situations where they are not phosphorylated (Fig. 7A, compare lanes 2  and 4 with lanes 1 and 3; Fig. 7B, compare lane 4 with lanes  1-3). However, because autophosphorylation is an apparent marker of complex assembly, our results may simply indicate that complexed subunits are more stable than noncomplexed subunits.
Most of the information regarding the domains of interaction on the subunits of CK2 has been obtained from studies with the yeast two-hybrid system (22,23,39). Given the propensity of the two-hybrid system to yield false positives or negatives (47,48), it is therefore important that results obtained in yeast be confirmed by another means. In this regard, our studies demonstrate that the N-terminal domain of CK2␤ (i.e. CK2␤1-166) interacts with another CK2␤ but fails to interact directly with a catalytic CK2␣Ј subunit. Despite the failure of CK2␤1-166 to interact directly with CK2␣Ј, there are further indications from a number of studies that this region does play a role in the regulation of CK2 and that it may be in close proximity to CK2␣ within the CK2 tetramer. In particular, Krehan et al. (37,38) demonstrated that residues 55-70 of CK2␤ could be crosslinked to CK2␣. Furthermore, acidic residues within the same region have been implicated as the binding sites for polyamines that stimulate CK2 activity (19). These residues may function as an autoinhibitory domain that interacts with the catalytic site of CK2␣. In a similar vein, the biochemical experiments of Marin et al. (15) demonstrated that CK2␤1-77 potently inhibits the activity of CK2␣ toward substrates such as calmodulin. However, this fragment of CK2␤ did not form a stable complex with CK2␣. Collectively, these studies suggest that the Nterminal region of CK2␤ does communicate with CK2␣, but this region by itself is not sufficient for stable interactions with CK2␣. If the intramolecular autophosphorylation of CK2 did indeed occur through transphosphorylation, the N-terminal region of CK2␤ may actually interact with the catalytic subunit of CK2 that is stably attached to the other CK2␤ subunit within the tetramer. Consequently, by mediating the formation of tetrameric CK2 complexes, CK2␤ can exert the dual regulatory functions that have been described by Marin et al. (15). Although the C-terminal domain of CK2␤ stably interacts with one catalytic subunit and influences CK2 by stabilizing and stimulating the kinase activity of that catalytic subunit, the N-terminal domain (i.e. negative regulatory domain) of the same CK2␤ molecule has the potential to interact with and regulate the other catalytic subunit. Stated in another way, a single catalytic subunit may have interactions with both CK2␤ subunits within the CK2 tetramer; stable interactions require the C-terminal domain of one CK2␤, whereas transient interactions involve the N-terminal domain of the other CK2␤ subunit. Tetrameric complexes of CK2 may therefore be subject to more precise regulation than would be afforded in a complex composed of one regulatory subunit and one catalytic subunit.
Further work will be required to obtain experimental support for this model.
Our results also suggest that the C-terminal region of CK2␤ is necessary for the direct formation of stable interactions with CK2␣ because evidence of HA-CK2␣Ј was not detected in Myc-CK2␤1-166 immunoprecipitates unless full-length CK2␤ was present. From two-hybrid studies (23), the smallest fragment of CK2␤ that exhibits the capacity for stable interactions with CK2␤ encodes residues 152-200. The precise boundaries of the CK2␣ interaction domain of CK2␤ and whether a single domain of interaction exists remains to be determined. It is also noteworthy that CK2␤ has been shown to interact with other protein serine/threonine kinases including A-Raf and c-Mos (49 -52). Interestingly, interactions between these enzymes and CK2␤ also involve the C-terminal domain of CK2␤. These results suggest that CK2␣, A-Raf, and c-Mos may have common three-dimensional elements that are involved in interactions with a common domain of CK2␤.
A detailed understanding of the assembly of CK2 subunits into tetrameric complexes may yield insights for the control of CK2 or its functions in cells. In particular, through the overexpression of a fragment of CK2␤ such as CK2␤1-166, which retains the ability to interact with CK2␤ but fails to interact directly with the catalytic subunits of CK2, it may be possible to alter the composition of CK2 complexes within cells. These strategies may be valuable for probing possible functions for CK2␣ and/or CK2␤ that exist outside of tetrameric CK2 complexes. For example, by preventing the formation of tetrameric complexes between CK2␣ and CK2␤, it may be possible to enhance the ability of CK2␤ to perform its other functions. Additionally, because the N-terminal region of CK2␤ has been shown to be important for interactions with CK2 substrates such as Nopp140 (53) and p53 (54), expression of a deletion of CK2␤ such as CK2␤1-166 may offer a strategy for altering the phosphorylation of some cellular CK2 targets.