Drosophila protein kinase CK2 is rendered temperature-sensitive by mutations of highly conserved residues flanking the activation segment

Abstract

CK2 is a Ser/Thr protein kinase essential for animal development. Although null alleles for CK2 are available in the mouse and Drosophila models, they are lethal when homozygous, thus necessitating conditional alleles for analysis of its developmental roles. We describe the isolation of temperature-sensitive (ts) alleles of Drosophila CK2α (dCK2α). These alleles efficiently rescue lethality of yeast lacking endogenous CK2 at 29°C, but this ability is lost at higher temperatures in an allele-specific manner. These ts-variants exhibit properties akin to the wild type protein, and interact robustly with dCK2β. Modeling of these ts-variants using the crystal structure of human CK2α indicates that the affected residues are in close proximity to the active site. We find that substitution of Asp²¹² elicits potent ts-behavior, an important finding because this residue contributes to stability of the activation segment and is invariant in other Ser/Thr protein kinases.

Introduction

CK2 is a Ser/Thr protein kinase that is ubiquitous and highly conserved (reviewed in [1, 2]. This enzyme has been purified from diverse organisms such as yeast, insects and mammals, and is composed of catalytic α subunits and regulatory β subunits. The CK2 holoenzyme is formed by the association of two catalytic α subunits with a core-dimer of the regulatory β subunits [3-5]. In most organisms, there are two distinct CK2α subunits, α and α′, which are encoded by distinct genes. This, however, is not the case in the fruit fly Drosophila melanogaster and the fission yeast Schizosaccharomyces pombe, which harbor a single CK2α gene [6, 7]. In contrast, budding yeast contain two catalytic subunits, α and α′, which are encoded by the CKA1 and CKA2 genes, respectively. It is important to, however, note that the yeast α/α′ subunits do not functionally correspond to metazoan α/α′ subunits; the nomenclature in yeast reflects the electrophoretic migration of these subunits and the order in which the corresponding genes were isolated and sequenced [8-10]. In contrast, most organisms contain a single isoform of CK2β. However, multiple isoforms of CK2β, which are encoded by distinct genes, have been reported in Saccharomyces cerevisiae and Drosophila [10-14]. Despite these differences in subunit composition, CK2 activity is messenger-independent, and this enzyme preferentially phosphorylates Ser/Thr residues that meet the consensus S/T-E/D-x-E/D [15, 16]. Given the unique micro-acidic determinant for phosphorylation, computational approaches have been employed to identify potential targets of this enzyme [17, 18]. The preponderance of potential CK2 targets, which include proteins involved with transcription, translation, cell cycle, and cell signaling, are being confirmed by proteomic approaches [19], suggest that this enzyme performs higher order functions (reviewed in [20]).

Genetic studies in yeast were the first to demonstrate that CK2 is an essential enzyme. In budding yeast, deletion of both the CKA1 plus CKA2 genes is lethal, and cells depleted of CK2 exhibit a pseudomycelial morphology, reminiscent of cdc mutations [9]. It was subsequently shown (using CKA2-tsalleles) that loss of CK2 activity elicited cell cycle arrest at the G1/S and G2/M transitions [21], thus establishing this enzyme as a regulator of cell proliferation, a role also evidenced for mammalian CK2 [22], and a lymphoproliferative disorder called Theileriosis [23]. Aside from its roles in functions that are cell-autonomous, it is becoming increasingly evident that CK2 activity is important for animal development. For example, CK2 regulates early embryonic development via the phosphorylation of proteins such as Antennapedia, Engrailed, Odd-skipped and Ultrabithorax [24-27]. More recently, the roles of this enzyme have been described in wingless/wnt signaling, and for Notch signaling during neurogenesis [28-31].

To better define the developmental roles of CK2, mutants are necessary. However, null alleles of CK2 elicit lethality early in mouse development due to cell-autonomous defects [32-34]. In this regard, Drosophila is a widely studied and an appropriate model for studying human disease and mammalian development (reviewed in [35]). A null allele of Drosophila CK2α, named Timekeeper (Tik), was identified in a screen for modifiers of the circadian clock [36]. Animals heterozygous for Tik display a pronouncedly lengthened behavioral rhythm, suggesting that this allele acts in a dominant manner [37]. Molecular analysis demonstrated that the Tik allele harbors two missense mutations, which replace Met¹⁶¹ with Lys and Glu¹⁶⁵ with Asp. Of these two residues, Met¹⁶¹ is invariant in all CK2α subunits and lies within the ATP binding pocket [38]. Accordingly, its substitution with Lys impedes ATP-binding and inactivates kinase activity, a finding confirmed for a Timekeeper variant of human CK2α [39]. Because of this inactivity, Tik homozygotes are larval lethal. However, Tik/+ flies do not display any overt developmental abnormalities, and thus embryogenesis, segmentation, and neurogenesis proceed normally [28, 36, 40]. As a result, RNAi approaches have been used to assess its roles in neurogenesis [28], but the perdurance of preformed enzyme has precluded analysis of early embryonic functions or in other developmental programs. A similar situation exists for Drosophila CK2β, for which hypomorphic and null alleles have been identified. The hypomorphic allele andante (replaces Met¹⁶⁶Ile) also affects the circadian clock, but animals homozygous for the andante allele do not display any overt developmental defects [41]. An insertion of a transposable (P) element in CK2β was subsequently identified, and named CK2β^mbuP1 because these flies displayed a ‘mushroom body undersized (mbu)’ phenotype due to hypo-proliferation of cells that comprise this structure. Imprecise excision of this P-element resulted in deletion of the CK2β gene (called CK2β^mbuΔA26), which was embryonic lethal when homozygous [42], as in also the case for mouse CK2β [32]. However, CK2β^mbuΔA26/+ flies do not display any overt developmental defects. Together, these studies suggest that the circadian clock is highly sensitive to CK2 dosage, whereas other developmental programs are perhaps more buffered.

In order to bypass the lethal effects associated with a complete loss of CK2 functions, two complementary approaches have been employed. These involve the identification of substrates of this enzyme, and the role of phosphorylation is then confirmed by analyses of phenotypes following expression of variants that are non-phosphorylatable or mimic the constitutively phosphorylated substrate. This approach has been successfully employed to assess CK2 functions in embryogenesis and neurogenesis [24-26, 29]. Alternatively, RNAi-based approaches have also been used [28]. However, these approaches are not applicable to all developmental programs, and thus necessitate conditional (ts) alleles.

We describe here the isolation and characterization of ts-variants of dCK2α using a yeast complementation system. We show that these variants exhibit significant similarity to the biological properties of wild type CK2α, but are rendered non-functional at higher temperatures in an allele-specific manner. Importantly, we show that replacement of Asp²¹² with Asn elicits potent ts-behavior. This finding is of general relevance, because Asp²¹² is invariant in the Ser/Thr protein kinase family, and plays a critical role in stabilizing the activation segment.

Materials and Methods

Construction of Ts alleles

The complete open reading frame of wild-type dCK2α encompassing the region from Met¹-Gln³³⁶ (Genbank accession number, M16534, [6]) was amplified by PCR to introduce BamH1 and Xho1 sites at the 5′ and 3′ ends, respectively. The PCR product was subcloned into pBSII(KS+) (Stratagene) and completely sequenced to confirm the absence of any PCR errors. The dCK2α cassette was then subjected to PCR-based site-directed mutagenesis to generate the putative ts-variants, dCK2α-A¹⁷⁷T, dCK2α-A¹⁷⁷V, dCK2α-D²¹²N, and dCK2α-G⁸⁹D. The selection of these sites was based on extant ts-alleles of yeast CKA2 (see Fig. 1A). Sequences of primers used for mutagenesis are available upon request. The resulting PCR products were subcloned into pBSII(KS+) and verified by sequencing.

Figure 1. Yeast CKA2-ts alleles.

A) The substitutions in five ts-alleles of yeast CKA2 are shown along with the corresponding residues in dCK2α. Residues shown in bold are invariant in dCK2α. B) Alignment of relevant segments of CK2α subunits showing conservation of Gly⁸⁹, Ala¹⁷⁷, and Asp²¹². Sc-CKA1p has an insertion of 38 amino acids that is denoted as ‘^’. Abbreviations are: Hs, Homo sapiens; Xl, Xenopus laevis; Sc, Saccharomyces cerevisiae, and Dm, Drosophila melanogaster. C) Schematic of yeast screen.

Rescue of yeast harboring null alleles of yeast CKA1 plus CKA2

Constructs encoding The WT/mutant dCK2α variants were subcloned into the BamH1 and Xho1 sites of the vector pESC-LEU (Stratagene). This vector utilizes a divergent GAL1/10 promoter that enables inducible expression of heterologous proteins in yeast. Similarly, dCK2α-WT was cloned into the BamH1 and Xho1 sites of the vector pESC-URA (Stratagene). In both cases (pESC-LEU or pESC-URA), expression of dCK2α-WT or its site-specific variants is under control of the GAL1 promoter.

The yeast strain RPG41-1a was utilized as a host for these studies [9]. This strain is deleted for both the CKA1 and CKA2 genes and is rescued by a single copy CEN/ARS-URA3-marked plasmid harboring a wild-type copy of the CKA2 gene. Strain RPG41-1a was transformed with the pESC-LEU constructs containing the WT or putative ts-alleles of dCK2α using lithium acetate. Transformants harboring both plasmids were isolated on synthetic minimal dextrose (SM-Dex) media containing adenine and lysine, and the resulting strains were named YKP1 through YKP5 (Table 1). These strains were grown non-selectively in rich galactose media (YPG) to drive expression of dCK2α constructs. Finally, ~300 cells were plated on synthetic minimal galactose (SM-Gal) media supplemented with adenine, lysine, uracil and 0.75 mg/ml 5-Fluoroorotic acid (5 FOA, Zymogen). 5FOA is toxic to cells containing a wild type URA3 gene. Survivors of this selection have dropped the (URA3-marked) CKA2-encoding plasmid, and their inviability is thus rescued by dCK2α-constructs contained in the vector pESC-LEU. The resulting strains, named YKP6-10, were reconfirmed for uracil auxotrophy. Growth was assessed by spotting ~250 cells of each strain followed by incubation at temperatures ranging from 25°C to 37°C.

Table 1.

S. cerevisiae plasmids and strains

Strain	Relevant chromosomal genotype	Plasmid genotype (URA3)	Plasmid genotype (LEU2)	Reference
RPG41–1a	MATa cka1-Δ1::HIS3 cka2-Δ1::TRP1	CEN6/ARSH4 URA3 CKA2	----------	[17]
YDH7	MATa cka1-Δ1::HIS3 cka2-Δ1::TRP1	----------	CEN6/ARSH4 LEU2 cka2-7	[18]
YDH8	MATa cka1-Δ1::HIS3 cka2-Δ1::TRP1	----------	CEN6/ARSH4 LEU2 cka2-8	[18]
YDH11	MATa cka1-Δ1::HIS3 cka2-Δ1::TRP1	----------	CEN6/ARSH4 LEU2 cka2-11	[18]
YDH13	MATa cka1-Δ1::HIS3 cka2-Δ1::TRP1	----------	CEN6/ARSH4 LEU2 cka2-13	[18]
YKP1	MATa cka1-Δ1::HIS3 cka2-Δ1::TRP1	CEN6/ARSH4 URA3 CKA2	2μ LEU2 GAL1P-dCK2α-WT	This study
YKP2	MATa cka1-Δ1::HIS3 cka2-Δ1::TRP1	CEN6/ARSH4 URA3 CKA2	2μ LEU2 GAL1P-dCK2α-A177T	This study
YKP3	MATa cka1-Δ1::HIS3 cka2-Δ1::TRP1	CEN6/ARSH4 URA3 CKA2	2μ LEU2 GAL1P-dCK2α-A177V	This study
YKP4	MATa cka1-Δ1::HIS3 cka2-Δ1::TRP1	CEN6/ARSH4 URA3 CKA2	2μ LEU2 GAL1P-dCK2α-D212N	This study
YKP5	MATa cka1-Δ1::HIS3 cka2-Δ1::TRP1	CEN6/ARSH4 URA3 CKA2	2μ LEU2 GAL1P-dCK2α-G89D	This study
YKP6	MATa cka1-Δ1::HIS3 cka2-Δ1::TRP1	----------	2μ LEU2 GAL1P-dCK2α-WT	This study
YKP7	MATa cka1-Δ1::HIS3 cka2-Δ1::TRP1	----------	2μ LEU2 GAL1P-dCK2α-A177T	This study
YKP8	MATa cka1-Δ1::HIS3 cka2-Δ1::TRP1	----------	2μ LEU2 GAL1P-dCK2α-A177V	This study
YKP9	MATa cka1-Δ1::HIS3 cka2-Δ1::TRP1	----------	2μ LEU2 GAL1P-dCK2α-D212N	This study
YKP10	MATa cka1-Δ1::HIS3 cka2-Δ1::TRP1	----------	2μ LEU2 GAL1P-dCK2α-G89D	This study
YKP11	MATa cka1-Δ1::HIS3 cka2-Δ1::TRP1	2μ URA3 GAL1P-dCK2α-WT	2μ LEU2 GAL1P-dCK2α-WT	This study
YKP12	MATa cka1-Δ1::HIS3 cka2-Δ1::TRP1	2μ URA3 GAL1P-dCK2α-WT	2μ LEU2 GAL1P-dCK2α-A177T	This study
YKP13	MATa cka1-Δ1::HIS3 cka2-Δ1::TRP1	2μ URA3 GAL1P-dCK2α-WT	2μ LEU2 GAL1P-dCK2α-A177V	This study
YKP14	MATa cka1-Δ1::HIS3 cka2-Δ1::TRP1	2μ URA3 GAL1P-dCK2α-WT	2μ LEU2 GAL1P-dCK2α-D212N	This study
YKP15	MATa cka1-Δ1::HIS3 cka2-Δ1::TRP1	2μ URA3 GAL1P-dCK2α-WT	2μ LEU2 GAL1P-dCK2α-G89D	This study

All strains are derived from RPG41-1a, which carries the additional mutations ade2-101^ochre his3-Δ200 leu2-Δ1 lys2-801^amber trp1-Δ1 ura3-52. The dCK2α-WT/mutant variants were expressed under control of the GAL1 promoter (GAL1^P) in the 2μ expression vectors pESC-URA or pESC-LEU (Stratagene).

Yeast phenotypic analyses

Strains were inoculated in YPG at a starting density of 2-5 × 10⁵ cells/ml. Cultures were grown at the permissive temperature (29°C) with shaking at 300 RPM. Aliquots were removed and fixed with formaldehyde at a final concentration of 3.7%. Cell numbers were determined using a hemocytometer, and the mean generation times were determined from semi-log plots.

Percent flocculation was determined as described [14]. Briefly, yeast strains were grown to mid-log phase at 29°C with shaking at 300 RPM. Cultures were allowed to stand at room temperature for 5 min, and free cells in the supernatant were determined based on absorbance at 600 nm (A₆₀₀). EDTA was added to the remaining culture at a final concentration of 20 mM to disperse the flocculated cells and to estimate total cells by A₆₀₀. Percent flocculation was calculated as 100×[(A₆₀₀ total-A₆₀₀ free)/A₆₀₀ total].

Terminal phenotypes were determined for strains rescued by dCK2α constructs. Following growth to mid log phase at 29°C, cultures were transferred to the non-permissive temperature (37°C). Samples were collected prior to, and 22 hr after the shift to non-permissive temperature, fixed with formaldehyde at a final concentration of 3.7% and photographed under DIC at a magnification of 400X.

Expression of dCK2α-ts variants

Strains were grown in YPG for ~30 hr at the permissive temperature (29°C). An equal number (~10⁷) of cells of each strain were harvested and washed in 1X Phosphate Buffered Saline (PBS). Cells were solubilized with sample buffer [43] and boiled for 10 min. Cell extracts were normalized to total protein and separated by SDS-polyacrylamide gel electrophoresis (SDS-PAGE). One gel was stained with Coomassie, while the second was electrophoretically transferred to nitrocellulose. Western blot analysis was conducted using a polyclonal (rabbit) antibody raised against Drosophila CK2 holoenzyme [44] at a dilution of 1:1000 and secondary antibody (affinity purified goat-anti-rabbit IgG) coupled to alkaline phosphatase (Bio-Rad) at a dilution of 1:5000. Immunoblots were visualized using nitroblue tetrazolium (460 μg/ml) and 5-bromo-4-chloro-3-indolyl phosphate (16.5 μg/ml), and reactions were terminated by the addition of EDTA to a final concentration of 50 mM.

Yeast 2 Hybrid (Y2H) interactions

WT/ts alleles of dCK2α were subcloned into the BamH1 and Xho1 sites of plasmid pGBT9, in which proteins are expressed as C-terminal fusions with the DNA-binding domain of yeast Gal4 [45]. dCK2β/β′ were expressed as C-terminal fusions with the activation domain of yeast Gal4, and were generated in plasmid pACT [46] as described previously [47, 48]. Y2H assays were conducted in yeast strains PJ69-4a and PJ69-4α [49] using a mating assay. Strain pJ69-4a was transformed with all pGBT9 constructs, whereas strain pJ69-4α was transformed with the pACT constructs. Transformants were mated, selected on SM-Dex containing histidine, adenine, uracil and lysine. Expression of the reporter genes HIS3, ADE2, and LacZ was assayed on SM-Dex medium lacking histidine or adenine as previously described [48].

Molecular modeling of dCK2α^Ts structures

Wild type and mutant proteins were modeled based on the known crystal coordinates of CK2. The Swiss PDB Viewer (www.expasy.org) and PyMOL (www.pymol.org) software packages were used to model the protein structure with the atomic coordinates of the human CK2 holoenzyme [50]. The corresponding residues of CK2α were ‘mutated’ to identify vicinal residues affected by each mutation. The coordinates were obtained from the Protein Data Bank (ID code 1JWH).

Results

Selection of dCK2α residues for site-specific mutagenesis

Four ts-alleles of CKA2 have been described [21]. These are cka2-7, cka2-8, cka2-11, cka2-12 and cka2-13 (Fig. 1A). Based on the high conservation of CK2α subunits, we identified residues that are conserved in dCK2α, as candidates for mutagenesis. While four of these alleles involved substitutions of two residues, one allele involved a single amino acid substitution. For example, cka2-7 harbors the substitutions Ala¹⁹⁰Thr and Thr³³⁶Met, whereas cka2-12 harbors the substitutions Ala¹⁹⁰Val and His²⁹⁴Tyr. Sequence alignments illustrated that Ala¹⁹⁰ is invariant in CK2α subunits (Fig. 1B), whereas Thr³³⁶ or His²⁹⁴ are not conserved (not shown). For example, metazoan CK2α subunits harbor a Pro in place of Thr³³⁶, which is located in vicinity of the C-terminus. In the crystal structure, Pro is not involved in structural stability of CK2α or its ability to interact with CK2β [38, 50]. We therefore hypothesized that the Thr³³⁶Met substitution might not contribute to the ts-behavior. Similarly, His²⁹⁴ is unique to yeast CKA2p, and the corresponding position in other CK2α subunits is either Arg (in CK2α) or Asn (in CK2α′). Since substitution of Ala¹⁹⁰ elicited ts behavior in two ts-alleles of CKA2, we targeted the corresponding residue, Ala¹⁷⁷ of dCK2α Φιγ. 1Β, and generated two variants, dCK2α-A¹⁷⁷T and dCK2α-A¹⁷⁷V.

The ts allele cka2-8 (Fig. 1A) harbors two mutations Glu⁵¹Lys and Gly¹⁰²Asp (Fig. 1A). The former residue is not conserved in metazoan subunits, which harbor a Gln at this position. In contrast, Gly¹⁰² is invariant in metazoan CK2α subunits (Fig. 1B). We therefore reasoned that Gly¹⁰² might (on its own) elicit ts behavior of cka2-8, and generated the variant dCK2α-G⁸⁹D. The remaining two ts alleles are cka2-11 and cka2-13 (Fig. 1A). The former allele involves two substitutions, Asp²²⁵Asn and Glu²⁹⁹Lys, while the latter is characterized by a single substitution, Asp²²⁵Asn. Importantly, Asp²²⁵ is invariant in the Ser/Thr protein kinase family, and plays a role in the structural stability of the catalytic loop [51, 52]. Since substitution of Asp²²⁵Asn in cka2-13 was, by itself, sufficient to elicit ts behavior, we generated the variant dCK2α-D²¹²N (Fig. 1A,B).

These variants were tested for their ability to complement the lethality of yeast lacking CK2 subunits (Fig. 1C), and employed the yeast strain RPG41-1a, which is deleted for the CKA1 plus CKA2 genes, and is rescued by a URA3 marked plasmid carrying the wild type yeast CKA2 gene. This strain was transformed with pESC-LEU plasmids that harbored the dCK2α variants (see Table 1). Strains harboring both plasmids (URA3-CKA2 plus LEU2-dCK2α) were grown on SM-Gal media, to allow for expression of dCK2α variants. The URA3-marked plasmid harboring CKA2 was then evicted using 5FOA (see Materials and Methods). Plates were incubated at a sub-optimal temperature (25°C) to ensure the survival of the putative dCK2α-ts strains. Under these conditions, survivors that had dropped the URA3 marked plasmid were recovered from yeast strains YKP1-5 (see Fig. 1C, and data not shown). These rescued strains were named YKP6-10 (Fig. 1C and Table 1). As controls, we tested but found that strain RPG41-1a, by itself, or harboring the empty vector pESC-LEU, did not lead to any survivors following 5FOA treatment (data not shown), consistent with the essential role of CK2 for viability [9], and demonstrating that dCK2α-constructs were necessary for rescue. The ability of wild-type dCK2α to rescue yeast is consistent with previous studies [53], and suggests that the pESC-LEU vector is suitable for expression of dCK2α, as also shown for Xenopus CK2α [54]. Strains rescued by dCK2α-variants exhibited a colony size comparable to that of a strain rescued by wild-type dCK2α or yeast CKA2 (data not shown), suggesting that these alleles are not functionally compromised at 25°C.

dCK2α variants exhibit ts-behavior

We next tested strains rescued by dCK2α variants for growth at the permissive temperature (PT, 29°C), and at higher non-permissive temperatures (NPT, 33, 35 and 37°C). As Fig. 2A shows, at 29°C yeast rescued by wild type dCK2α exhibit growth comparable to the control strain RPG41-1a (expressing yeast CKA2). Furthermore, strains rescued by yeast CKA2 or dCK2α-WT displayed equivalent growth at 35 and 37°C; the slight growth deficit at 37°C was similar in both strains (Fig. 2A) and might reflect heat-shock effects of this higher temperature. These results suggest that dCK2α-WT is not intrinsically temperature sensitive in yeast. Similarly, strains YKP7-10 (rescued by dCK2α variants) exhibited robust growth at 29°C (the PT), which appeared qualitatively similar to that of strains rescued by yeast CKA2 or dCK2α-WT (Fig. 2A). In contrast, strains YKP7, YKP8 and YKP9 displayed a complete growth-arrest at 33°C (not shown) or at 35°C (Fig. 2A), indicating that these variants were, indeed, temperature-sensitive. YKP10 (expressing dCK2α-G⁸⁹D) exhibited significant, but incomplete, growth arrest at 37°C (Fig. 2A) indicating that it is the weakest ts-variant. A temperature higher than 37°C (such as 40°C) was not tested for dCK2α-G⁸⁹D, because this temperature is, itself, non-permissive for growth of wild type yeast (data not shown), or those rescued by yeast CKA2 or dCK2α-WT (inset in Fig. 2C). Thus the rank order of temperature sensitivity is: yeast CKA2 = dCK2α-WT > dCK2α-G⁸⁹D > dCK2α-D²¹²N ≅ dCK2α-A¹⁷⁷T > dCK2α-A¹⁷⁷V (see inset in Fig. 2C). As controls, we tested strains cka2-7, cka2-11 and cka2-13, and all exhibited ts-behavior and NPT's as previously described (Fig. 2A, and [21]). We also confirmed that expression of the dCK2α variants was necessary for rescue, as no growth was detected in rich dextrose medium (YPD), which leads to repression of the GAL1/10 promoters [55]. The weak growth of the dCK2α-WT rescued strain in YPD (Fig. 1A) has been observed before [53], and might reflect low levels of a fully active protein.

Figure 2. dCK2α variants exhibit ts-behavior.

A) Yeast strains YKP6-10 rescued by the indicated dCK2α variants were tested for growth on rich (YP) galactose media at the permissive temperature (29°C) and at higher non-permissive temperatures (35 and 37°C), or on rich (YP) dextrose media at 29°C. In each case, two independent clones were analyzed. In parallel, RPG41-1a (wild type CKA2) and three cka2-ts strains were also tested as controls. Note that the control strains YDH7 (cka2-7), YDH11 (cka2-11) and YDH13 (cka2-13) exhibit robust growth at 29°C, but are arrested at 35°C, a temperature higher than their minimum NPT. B) The indicated dCK2α-variants were tested for ts behavior in the presence of wild type CKA2 or wild type dCK2α. C) Terminal morphologies of yeast cells expressing wild type CKA2 or dCK2α at 29 and 37°C. The phenotypes of strains rescued by dCK2α-ts variants at the NPT (37°C); these cells exhibit morphologies akin to wild type cells at 29°C, but are not shown for simplicity.

We have previously found that expression of a catalytically null dCK2α variant (Tik) in flies wild type for CK2 elicits dominant neural defects akin to loss of Notch signaling [28]. Because CK2 is a multimeric enzyme, we tested whether the dCK2α-ts variants behaved in a dominant manner (in yeast), but found that in each case co-expression of wild type dCK2α or yeast CKA2 completely suppressed the temperature-sensitivity of the relevant strains (Fig. 2B), suggesting that these alleles are recessive. No such suppression was seen in strains YKP6-10 that were transformed with the empty expression vector, pESC-URA3 (data not shown).

It has been previously shown that depletion of CK2 results in ‘pseudomycelial’ cells [9]. Furthermore, ts-alleles of CKA2 display cell-cycle arrest at the G1/S and G2/M transitions [21], whereas those of CKA1 display enlarged cells due to defects in cell polarity [56], a phenotype also described for loss of CK2 in fission yeast [57]. We thus assessed the terminal phenotype of dCK2α-ts rescued strains. Yeast with wild-type (endogenous) CK2 or those rescued by dCK2α-WT display a normal cellular morphology at the PT (29°C) or at 37°C (Fig. 2C). Similar morphologies were observed for strains rescued by dCK2α-ts variants at the PT of 29°C (data not shown). In contrast, when shifted to 37°C (the NPT), strains rescued by dCK2α-A¹⁷⁷T and dCK2α-G⁸⁹D displayed ‘pseudomycelial’ cells (Fig. 2C), a phenotype that recapitulates depletion of endogenous CK2 [9]. In contrast, strains rescued by dCK2α-D²¹²N and dCK2α-A¹⁷⁷V displayed enlarged cells, a phenotype akin to loss of CKA1. Thus, dCK2α complements both, the cell cycle and cell polarity functions of CKA2 and CKA1, respectively, and the terminal phenotype of strains rescued by dCK2α-ts alleles mimic loss of endogenous (yeast) CK2 subunits.

Characterization of dCK2α-ts variants

It has been previously shown that slow-growth upon lowered CK2 levels closely correlates with flocculation, a form of cell-cell aggregation [9, 14]. We thus quantified these phenotypes in strains rescued by dCK2α-ts variants to better assess their rescue efficiencies. At the PT (29°C), the doubling time of strain YKP6 (dCK2α-WT) appear close to that of strain RPG41-1a (endogenous yeast CK2), and similar numbers were obtained for dCK2α-A¹⁷⁷T, dCK2α-D²¹²N and dCK2α-G⁸⁹D (Fig. 3A). The reasons underlying the slightly higher extent of flocculation in the strains rescued by WT/ts-variants of dCK2α is currently unknown, but it is possible that this reflects enzyme activity close to that which is rate limiting for cell proliferation. In contrast, strain YKP8 (rescued by dCK2α-A¹⁷⁷V) exhibited a doubling time and degree of flocculation that were the highest (Fig. 3A, B). Given our observation that this variant does not elicit robust rescue at 29°C (the PT, see Fig. 2B), the possibility is high that dCK2α-A¹⁷⁷V is not fully active, and thus might be structurally compromised. It is important to note that a similar slow growth and flocculation defect has been described for the A¹⁹⁰V variant of yeast CKA2 (cka2-12) even at 29°C [21].

Figure 3. Characteristics of dCK2α-ts variants.

A and B) Mean generation time and percent flocculation were determined for the strains rescued by the indicated dCK2α variants, and for the control strain RPG41-1a (wild type CKA2). C) Immunoblot analysis to assess steady-state expression of dCK2α variants. Abbreviations are: RPG (strain RPG41-1a); WT, dCK2α-WT; A¹⁷⁷T, dCK2α-A¹⁷⁷T; A¹⁷⁷V, dCK2α-A¹⁷⁷V; D²¹²N, dCK2α-D²¹²N; and G⁸⁹D, dCK2α-G⁸⁹D. Note that this antibody does not cross-react with the product of the CKA2 gene (RPG41-1a). D) Yeast two-hybrid interaction of dCK2α-ts variants with dCK2δ. Abbreviations are as in panel B. The reporter genes HIS3 and ADE2 were assayed by growth on minimal media and are depicted as none (−−−) or strong (+++). LacZ values are reported as Miller Units.

The distinct NPT's of the Ts variants led us to assess whether this correlated to levels of expression/stability of the mutant proteins. We compared the steady-state expression levels of dCK2 in the Ts strains YKP7-10 to that of dCK2α-WT (YKP6). The strains were grown in YPG to mid-log phase and cell extracts from equivalent number of cells were separated by SDS-PAGE and subjected to immunoblotting with anti dCK2 antibodies (Fig. 3C). This antiserum does not cross-react with the (yeast) CKA2 subunit as evidenced by the absence of a band in extracts of RPG41-1a (Fig. 3C). When tested against rescued yeast, we found that levels of the dCK2α-ts variants were comparable to that of dCK2α-WT (Fig. 3C). Thus, the ts-behavior of dCK2α variants and their rank order of sensitivity do not reflect attenuated expression levels or greater instability in yeast.

It has been previously shown that like the mammalian enzyme, CK2 purified from Drosophila embryos is an α₂β₂ hetero-tetramer [58], and genetic analysis has shown that interaction of dCK2α with dCK2β is necessary for proper in vivo CK2 functions [59]. We therefore assessed for this interaction using the yeast interaction trap, as previously described by others and us [12, 47, 60, 61]. Yeast expressing dCK2α or dCK2β, by themselves, did not elicit expression of the reporter genes HIS3 or ADE2 (as evidenced by their inability to grow in media lacking these amino acids), and the levels of LacZ induced under these conditions were indistinguishable from the ‘baseline’ (<10 Units, Fig. 3D). On the other hand, co-expression of dCK2α plus dCK2β elicited robust induction of all three reporter genes, indicative of a strong interaction, and similar results were obtained for the alternative subunit dCK2β′ (Fig. 3D, and data not shown). When tested with the dCK2α-ts variants, we find that the D²¹²N and G⁸⁹D variants exhibited interaction (with dCK2β or dCK2β′) close to that of the wild type protein, whereas dCK2α-A¹⁷⁷T or dCK2α-A¹⁷⁷V bound more strongly (see Discussion).

Discussion

The studies we report here provide direct evidence that residues altered in yeast CK2 ts-alleles elicit similar conditional destabilization of dCK2α functions and thus constitute the first ts-variants of Drosophila CK2. Furthermore, these studies suggest that all of the yeast CKA2-ts alleles are, in fact, ‘single hits’, and further underscore the remarkable structural/functional conservation of these subunits. These dCK2α variants efficiently rescue yeast lacking endogenous CK2 at the permissive temperature (Fig. 2), but this ability is neutralized at the higher non-permissive temperature. Quantitative analysis of growth rates, flocculation and expression levels (Fig. 3) suggest that these variants are not compromised for function at the lower temperature. Our observation that ts-variants recapitulate the well-described morphological defects that are associated with loss of yeast CKA1p or CKA2p (Fig. 2D) indicates that the dCK2α-variants closely mimic the functions of endogenous (yeast) subunits. In addition, these variants display robust interaction with CK2β (Fig. 3D) suggesting that holoenzyme formation is unlikely to be affected. While we have not directly assessed tetramer formation, we note that a robust two-hybrid interaction closely correlates to formation of the (α₂β₂) holoenzyme in yeast and in vitro [12, 53].

Based on these analyses, it appears that the D²¹²N, A¹⁷⁷T and G⁸⁹D variants of dCK2α exhibit properties closest to the wild type protein, but are rendered non-functional at a higher temperature. This might not be the case for the dCK2α-A¹⁷⁷V variant. Despite exhibiting the greatest temperature sensitivity, i.e., lowest NPT of any variant (32°C, see inset Fig. 2C), dCK2α-A¹⁷⁷V does not appear to fully rescue yeast at lower temperatures (25 and 29°C), and exhibits an interaction with CK2β that is stronger than that with the wild type protein. These differences do not reflect effects of this substitution on stability/expression levels (Fig. 3C). In the case of human CK2 holoenzyme, Ala¹⁷⁹, which corresponds to Ala¹⁷⁷ (in dCK2α) is not involved in the interaction of CK2α with either the body or the tail of CK2β [50], suggesting that substitution of Val at this site should have not influenced this interaction. It is reasonable to speculate that dCK2α- A¹⁷⁷V is structurally compromised. Consistent with this possibility, it has been shown that the cka2-12 allele, which harbors the A¹⁹⁰V substitution (see Fig. 1A), also exhibits the greatest temperature sensitivity [21]. Based on our studies, it thus appears that dCK2α-D²¹²N, dCK2α-A¹⁷⁷T and dCK2α-G⁸⁹D exhibit properties of bona fide ts-alleles.

It has previously been shown that dCK2α-WT purified from rescued yeast exhibits substrate specificity and kinetic constants that are comparable to protein purified from Drosophila embryos or from bacterial/insect cell expression systems [53, 62-65]. However, a similar analysis on ts-variants has been difficult, since they loose activity upon partial purification from yeast (Bidwai, unpublished), perhaps, reflecting loss of chaperones such as HSP90 and/or CDC37, whose activities are required for maintaining CK2 (and other protein kinases) in a fully active state in vivo [66-70].

To better assess the impact of these substitutions on the ts-behavior, we used the atomic coordinates of human CK2α [38, 50], which is remarkably similar to dCK2α [6, 71]. In the crystal structure, human CK2α consists of two domains, a β-strand rich N-terminal lobe and a C-terminal lobe rich in α-helices; the active site is located in the interfacial region (Fig. 4B). Importantly, the activation segment in human CK2α is represented by sequences from Ile¹⁷⁴->Glu²⁰¹, (green box in schematic in Fig. 4A). Unlike cyclin-dependent kinases, such as CDK2, where the activation segment is conformationally stabilized upon binding of Cyclins, this segment is optimally positioned for catalytic activity of CK2 [50]. We find that Ala¹⁷⁷, which is located in the region between β-strands 8 and 9 (nomenclature from [50], Fig. 4A), itself, lies within the activation segment. The possibility is high that substitution of Ala¹⁷⁷ might impair the conformation of this critical region, and lead to a hypoactive kinase. Consistent with this prediction, replacement of Ala¹⁷⁷ with the bulkier Val residue generates a variant with the weakest complementation efficiency, whereas its replacement with Thr does not impair activity in a measurable manner (Fig. 2B). In contrast, neither Asp²¹² nor Gly⁸⁹ lie within the activation segment, and accordingly, variants of these sites exhibit normal rescue of yeast CK2 mutants, and mean generation times and interactions with CK2β that closely mimic wild type dCK2α (Fig. 3). Interestingly, the only null allele of dCK2α identified to date (Tik) involves two substitutions (M¹⁶¹K and E¹⁶⁵D), which lie immediately outside the activation segment as well (Fig. 4A).

Figure 4. Modeling of substitutions in dCK2α ts-variants.

A) The locations of the substitutions eliciting ts-behavior are shown relative to the activation segment (AS, green box), and the relevant secondary structure motifs as described [50]. The M¹⁶¹ and E¹⁶⁵ substitutions are contained in the CK2α null allele, Tik. B) Molecular model of CK2α-D²¹²N. Note that the corresponding residue in human CK2α is Asp²¹⁴ (right panel). C) Sequence alignments showing conservation of Ala¹⁷⁷ and Asp²¹² in human (Hs) and Drosophila (Dm) CK2α, CDK2, and PKA. The numbers above the alignment refer to residues of dCK2α, and green box depicts the ‘Activation Segment’. D) Residues affected by the D²¹²N substitution in dCK2α are in red, and the ‘Motif Logo’ shown above the alignment depicts conservation of these amino acids in CDK2 and PKA.

Molecular modeling confirms that despite their locations, Gly⁸⁹, Ala¹⁷⁷ and Asp²¹² lie in close proximity of the active site (Fig. 4B and data not shown). Given the evolutionary conservation of the catalytic core of Ser/Thr protein kinases [72, 73], and that CK2 is closely related to the CDK's, we sought to determine whether the dCK2-ts alleles affect conserved residues. This is, indeed, the case, and we find that Ala¹⁷⁷ and Asp²¹² are invariant in Drosophila/human CDK2 as well as in the catalytic subunit of cAMP-dependent protein kinase (PKA, Fig. 4C). The behavior of Asp²¹² is of particular interest because this residue, which is located on α-helix F (nomenclature from [50], see Fig. 4A and B), is also conserved in human and Drosophila CDK2 and PKA, and this acidic residue is important for stability of the activation segment of PKA [51, 74]. These findings suggest that substitution of Asp²¹² with Asn might engender collisions (of the amide) with vicinal residues. In the molecular model, amino acids within 6 angstroms of Asp²¹² are His¹⁴⁶, Arg¹⁵³ and Asp¹⁵⁴, and remarkably, all three residues are invariant in CDK2 and PKA. Thus, Asp²¹² represents a residue of CK2α that is particularly sensitive for eliciting ts-behavior without perturbing its overall structure. Our studies suggest that the approach we describe should be of general applicability to other members of this family of enzymes.

The availability of these ts-alleles of dCK2α will now permit genetic approaches in Drosophila by the replacement of the endogenous allele with these engineered variants. Given the applicability of Drosophila as a relevant model to study human diseases and developmental biology [35, 75, 76], such a resource will facilitate analysis of the roles of this important and conserved protein kinase. Finally, the conserved substitutions that elicit ts behavior should enable the targeting of similar residues in other protein kinases for which ts-alleles have not been forthcoming.