Introduction

Two hundred million cases of malaria occur annually resulting in half-a-million deaths of which most are children from sub-saharian Africa1. Malaria is caused by infection with protozoan parasites of the genus Plasmodium with Plasmodium falciparum and Plasmodium vivax accountable for the most virulent forms of the disease. The emergence of parasite resistance to most of the available antimalarial treatments, including artemisinins, has already been observed, making the discovery of alternative therapeutic targets a pressing need1. Eukaryotic protein kinases (ePKs) form a large family of enzymes with major roles in the regulation of many cellular processes2. The therapeutic relevance and druggability of ePKs is emphasized by the growing number of kinase inhibitors that are used in the clinic or have entered clinical trials to treat various conditions such as cancer3.

Not surprisingly, protein kinases are also pivotal to regulate cellular processes in pathogenic parasites such as Leishmania4 or Plasmodium5,6,7. The genome of Plasmodium falciparum includes more than 80 putative protein kinases8, several of which have been shown to be indispensable for the parasite survival5,9,10. The relevance of the Plasmodium kinome in the intrincated life cycle of the parasite, is apparent from the fact that 50% of its proteome becomes phosphorylated during the erythrocytic stage, with a vast majority of proteins displaying multiple phosphorylation sites11. In this context, a central regulatory role has been suggested for the Plasmodium falciparum protein kinase CK2 (PfCK2)11, as an enzyme regulating the activity of several other important kinases. Moreover, reverse genetics experiments showed that the catalytic α subunit α of PfCK2 is essential for pathogen survival, during its replicative cycle whithin the host red blood cells5,6, making of PfCK2 a promising drug target. In contrast to many kinases that only become activated upon receiving specific regulatory signals, CK2 is a pleiotropic and constitutively active Ser/Thr kinase that phosphorylates and regulates the activity of a suprisingly large number of proteins, including transcriptional activators12. Unlike the human enzyme (hCK2), PfCK2 undergoes autophophorylation at Thr63 but not at Tyr186, a residue located within its activation loop13. Here, using mass spectrometric analysis, we show that PfCK2 also displays Tyrosine kinase activity, leading to autophosphorylation at Tyr30. The human hCK2 ortholog is built up of either homo- or hetero-tetramers formed by combinations of two catalytic (α and α′) subunits and two regulatory (β) subunits14. In contrast, PfCK2 achieves its quaternary structure using one single catalytic (α) and two regulatory (β1 and β2) subunits. The cytoplasmic and nuclear localization of PfCK2 are consistent with pull-down experiments linking PfCK2 with the regulation of nucleosome assembly, suggesting a role for this kinase during mitotic cell division6. The three-dimensional structure of hCK2 has already been well characterized in its free form, in complex with the non-hydrolysable ATP analogue AMPPNP, and in complex with several potent small molecule inhibitors targeting the enzyme ATP binding site14,15,16,17,18. However an experimental PfCK2 structure is still unavailable. The 65% level of amino acid sequence identity between the catalytic PfCK2 and hCK2 α subunits, indicates that both enzymes share a well conserved overall structure. However, determination of the Plasmodium enzyme 3D structure would allow a precise comparative structural analysis with its human ortholog that could lead to the identification of enzyme regions, including at its ATP binding site, amenable to specific inhibition. Here, as a first step towards this goal, we report the bacterial expression, purification and crystallization of an enzymatically inactive mutant of PfCK2α having the mutation D179S and its crystal structure bound to ATP at 3.0 Å resolution. Using the wild-type protein expressed in the same conditions, we provide experimental evidence from mass spectrometry of a dual Serine/Threonine and Tyrosine kinase activity of PfCK2. Using a peptide substrate, we also study the impact of single and double mutations (to Ala or Asp) targeting phosphorylable residues Tyr30 or Thr63 on PfCK2 kinase activity. Finally, we show that CX4945, a potent hCK2 inhibitor18, inhibits PfCK2 with an IC50 of 13.2 nM.

Results

Expression and Purification of PfCK2α and its mutants

We used a multi-construct approach in E. coli in order to obtain protein yields suitable for X-ray crystallography. We screened more than 40 protein constructs having either N- or C-terminal truncations with the tag located at the N- or C-terminal end of the protein19. We found that N-terminal truncations of PfCK2α usually yielded insoluble proteins, while C-terminal truncations improved protein solubility and stability (Fig. 1A). Moreover, we observed a significant improvement in protein stability when the hexa-histidine tag used for affinity purification was displaced from the N- to the C-terminus of the protein. A protein fragment spanning residues Ile12 to Ser335 of PfCK2α was used for large-scale expression, purification and enzymatic activity measurement. For crystallization and structure determination, the “Aspartate 179 to Serine” single mutant of the same fragment (PfCK2αD179S) was designed (Fig. 1B). Non-phosphorylatable mutants, namely Thr63 to Alanine (PfCK2αT63A) and Tyr30 to Alanine (PfCK2αY30A), as well as double mutant PfCK2αT63AY30A, were generated to examine the effect of autophosphorylation at Thr63 and Tyr30 on kinase activity. Likewise, phospho-mimetic mutants, namely Thr63 to Aspartate (PfCK2αT63D) and Tyr30 to Aspartate (PfCK2αY30D) as well as double mutant PfCK2αT63DY30D, were also obtained.

Figure 1
figure 1

Protein Expression. (A) Selected constructs from the series generated for identifying the protein boundaries yielding the most soluble and stable PfCK2α enzyme when expressed in E. coli. Red, yellow and green rectangles represent insoluble, marginally soluble, and soluble constructs respectively. The position of the hexahistidine tag is indicated as “His6” and theTEV cleavage sites, and amino- and carboxy terminal residues labelled. (B) 12% SDS-PAGE analysis of the purified PfCK2α constructs (wild-type and single mutant) used in this study spanning residues Ile12-Ser335. (C) Needle-shaped PfCK2αD179S crystals obtained after optimization using micro-seeding (see methods).

Full size image

PfCK2α is an active Ser/Thr kinase

The activity of the protein was assessed using [γ32P]-ATP based radiograms (see methods). To unequivocally ascertain phosphorylation to intrinsic PfCK2α activity, an inactive kinase PfCK2αD179S single mutant was used as negative control (the expression of an alanine mutant at position 179 proved to be insoluble). The “179-DWG-181” motif in the catalytic loop is an absolutely conserved amino acid sequence motif responsible for ATP and Mg2+ binding by the kinase and whose mutation typically causes bona fide loss of kinase activity. The results of an enzymatic assay displayed in Fig. 2A,B unequivocally demonstrate that the wild-type PfCK2α enzyme is catalytically active and confirmed PfCK2α autophosphorylation activity.

Figure 2
figure 2

Autophosphorylation activity of PfCK2. (A) Autoradiograph kinase activity assay was carried out using 2 μg of wild type (lane 1) or D179S mutant PfCK2 (lane 2) with Mg [γ32P]-ATP. (B) The proteins were resolved by SDS-PAGE, stained with Coomassie blue and their activity analysed by autoradiography. (C) Annotated MS/MS spectra showing identification of Y30 phosphorylation using the Mascot protein database search software. The Tyr30 (labeled Y16 in the spectrum) is unambiguously identified as it is sandwiched by a series of y-ions [y16,y17,y18(phosphorylation site),y19,y20]. y13 ion was trimmed 10 times to show the weaker ions in the spectrum.

Full size image

Identification of autophosphorylation sites

Mass spectrometry (MS) analysis of the active PfCK2α identified four autophosphorylation sites: two presenting as major at Tyr30 and Thr63, and two as minor phosphorylation sites at Tyr61 and Ser55 (Fig. 2C and Table 1), representing approximately 90% and 10% of the identified phosphorylated peptides respectively. Moreover, the same MS analysis conducted using the catalytically inactive PfCK2αD179S protein proved that the post-translational modifications observed were indeed the result of catalytic autophosphorylation and were not caused by E. coli enzymes (Fig. 2A). The identified phosphorylation sites agree with the acidic nature of the CK2 peptide substrates recognition sites12, and confirm the catalytic specificity that was previously attributed to PfCK213. We could corroborate Thr63 as one of the two major phosphorylation sites, together with Tyr30 that unexpectedly also appeared to undergo phosphorylation (Fig. 2C and supplementary data). This dual autophosphorylation pattern of PfCK2 differs from the single phosphorylation pattern described previously13,20, but is in agreement with the Tyrosine specificity found in the autophosphorylation pattern of the human enzyme13,20. Unlike Tyr186 (equivalent to Tyr182 from hCK2 in the kinase activation loop)13, Tyr30 does not belong to a conserved autophosphorylation site. A structure-based alignement of the amino-acid sequces of hCK2, PfCK2 and PvCK2 is presented in Fig. 3. In the 3D structure, the hydroxyl moiety of Tyr30 is within hydrogen bond distance from Glu184 and in the vicinity of the putative peptide substrate binding site delinated by the basic patch encompassing residues Lys78-Arg84. The functional role of these two autophosphorylation sites was explored by site-directed mutagenesis followed by kinase activity (see below).

Table 1 Phosphorylation sites (bold and underlined) identified in the recombinant PfCK2α derived from this work.
Full size table
Figure 3
figure 3

Structure-based sequence alignment of CK2 sequences from two species of the Plasmodium parasite and of human CK2. Amino-acid sequence alignment of alpha subunits of CK2 proteins from human (P68400), Plasmodium vivax (A0A0J9V3K6) and Plasmodium falciparum (Q8IIR9). The figure was prepared using http://espript.ibcp.fr45. The secondary structures elements for PfCK2 (this work) are displayed above the alignment, with the helices depicted as springs and strands as arrows. The hCK2 C-terminal tail (ARMGSSSMPGGSTPVSSANMMSGISSVPTPSPLGPLAGSPVIAAANPLGMPVPAAAGAQQ) was removed from the alignment because no equivalent residue is present in PfCK2α.

Full size image

Catalytic activity measurements

To explore the functional relevance of the autophosphorylation sites, the enzymatic activity of Tyr30 and Thr63 PfCK2 mutants were assayed in a peptide-based phosphorylation assay (Table 2). Wild type and mutant D179S defined the catalytic range of PfCk2 with respectively 1498.74 ± 43.22 and 1.48 ± 3.32 units of activity (ua) per mg of purified protein. Single mutants Tyr30Ala and Thr63Ala had reduced activity of 877.52 and 360.68 ua/mg, respectively, demonstrating a direct impact of autophosphorylation in regulating PfCK2 kinase activity. Taking into account that these measurements were made only with the catalytic α subunit, these changes indicate that both Tyr30 and Thr63 play a direct role in activation that might be further modulated in the presence of the regulatory subunits. PfCK2 proteins having single mutation of Tyr30 or Thr63 to aspartate had further reductions in catalytic activities compared to the corresponding Ala mutants (Table 2). Moreover, PfCK2 having double Tyr30 and Thr63 Aspartate mutant had negligible kinase activity. This possibly suggests that sequential autophosphorylation is needed for activation of PfCK2., an event that could not be mimicked by the introduction of a shorter carboxylic side chain as presented in the form of an Asp residue.

Table 2 PfCK2α variants activity measurements using the peptide [RRRDDDSDDD] as phosphate acceptor for the incorporation of radiolabelled [γ32P]-ATP.
Full size table

Enzyme inhibition

The human CK2 protein has been thoroughly characterized as a potential target for therapeutical treatment against cancer21,22. To explore whether CX4945 an ATP competitive inhibitor targeting hCK2 can be repurposed to target PfCK2 for the treatment of malaria, we measured its IC50 for both enzymes. Our measurements show that both proteins can be inhibited (Fig. 4) with a three fold difference in IC50 value: Our result for hCK2 (IC50 of 4.7 nM) is in agreement with previous measurements of the CX4945 potency (1 nM IC50)23, while we find that CX4945 inhibits PfCK2 with an IC50 value of 13.2 nM. The difference in IC50 between the two homologous enzymes, although small, indicates that their ATP binding sites are slightly different, which is consistent with the crystallographic structure determination reported below. Whether these structural differences could be exploited to modify CX4945 to specifically inhibit the parasite enzyme requires further medicinal chemistry studies informed by the present structural information. Interestingly, similar inhibitory responses were characterized in the case of quinalizarin13, suggesting that inhibitors bind the human and the Plasmodium proteins slightly differently, despite the close similarity of their active sites.

Figure 4
figure 4

Effect of compound CX4945 on CK2 activity. The compound CX4945 displays concentration-dependent inhibition on both CK2 proteins used here, human (circles, hCK2) and Plasmodium falciparum (squares, PfCK2). The IC50 value of CX4945 for each protein is indicated. Data were analysed using the GraphPad Prism software using a Log [inhibitor] vs normalized response fit.

Full size image

Crystal structure determination

The wild type PfCK2 was subjected to extensive crystallization trials. Initially, only severely twinned needle shaped crystals giving diffraction at best to about 3.4 Å were obtained with the wild type enzyme, and the resulting electron density maps were of poor quality. A significant improvement in crystals quality was obtained when the inactive mutant PfCK2αD179S was used. The difference possibly reflects sample hereogeneity provoked by autophosphorylation when using the wild type enzyme. The structure of the complex between the catalytically inactive mutant PfCK2αD179S with ATP was prepared by cocrystallization, transferring crystals seeds initially grown in the presence of protein, ATP and Mg2+ into a solution devoid of Mg2+ (see methods). Data collection and refinement statistics for PfCK2αD179S bound to ATP are presented in Table 3. The structure was refined to Rwork /Rfree = 0.192/0.231 at a resolution of 3.0 Å. Clear electron density was visible for all three molecules (named A, B, C) present in the asymmetric unit (Fig. S1A), allowing unequivocal assignement for of all main chain and a majority of side chain atoms for the polypeptide chain up to residue 328. Residues 329–335 at the C-terminal end of the protein are disordered. The structures of all three monomers were tightly restrained during refinement, resulting in r.m.s. deviations values for their main chain atoms between 0.21 Å and 0.24 Å after superposition (for a total of 316 α-carbon atoms)24. A structure-based alignement of the sequences of PfCK2, PvCK2 and hCK2 is presented in Fig. 3.

Table 3 Data collection and structure refinement statistics.
Full size table

Analysis of crystal contacts reveals interfaces between the three monomers extending over an area of approximately 700 Å2. Residues brought in extensive intermolecular contact originate from the C-lobe of molecule C and the N terminal arm preceeding the N-Lobe of molecule A. Interestingly, some crystal contacts involve residues from the P-loop, and phosphorylation site around Thr63 whose structure might be affected by crystal packing forces.

A residual difference electron density map at a level of 5 σ allowed unambiguous building of the bound ATP for each of the three molecules present in the asymmetric unit (Fig. 5). Here, the ATP moiety is bound in the active site in the absence of divalent metal ions (Mg2+) that are seen to coordinate the phosphate groups in nucleotide binding proteins, indicating that the energy contributions of active site residues are sufficient for substrate binding. Using isothermal titration microcalorimetry, we compared the binding parameters of PfCK2αD179S (the protein used for crystallization) for ATP in the presence and absence of Mg2+. These measurements returned a K D value of 22.3 µM in the absence of Mg2+ and 16.3 µM in the presence of a concentration of 1 mM of Mg2+, confirming that the mutant protein can bind ATP in the absence of Mg2+ with the same 1:1 stoichiometry. The microcalorimetry data reveal a negative enthalpy change of ΔH indicating a binding mechanism involving polar interactions. This is in agreement with what is observed in the crystal structure of the complex. Consistent with the overall high level of amino-acid sequence identity of 65% (Fig. 3), PfCK2α shares with its human ortholog the canonical kinase bi-lobular tertiary structure with a N-terminal domain arranged around a central β-sheet and a bundle of α-helices forming the C-terminal domain (Fig. 5A). The active site is located at the interface between both sub-domains forming a groove that penetrates deeply inside the protein (Fig. 5B). A structural comparison of PfCK2αD179S (monomer A) with hCK2 in its active (PDB codes 3NSZ and 2PVR)18,25 and inactive forms (PDB code 3FWQ)26 returns r.m.s.d. values of 0.85 Å, 1.22 Å and 1.56 Å (after superposing α-carbon atoms from residues 12 to 312), respectively.

Figure 5
figure 5

(A) Side-by side representation of the PfCk2αD179S structure (left and hCK2 (right) shown in the same orientation. The N-terminal tail, the N-lobe and C lobe are displayed in different colors. Structural elements discussed in this paper, namely P-loop, hinge region and C-helix are indicated. The ATP (PfCK2) and AMPPNP (hCK2) ligands bound in the active site are displayed as sticks. (B) Close up view of the bound ATP (sticks) in the PfCK2 active site with an omit residual electron density map superimposed. (C) Superposition of how the substrate is bound by the PfCK2αD179S and human enzyme: residues contributing to the active site of PfCk2αD179S are displayed using the same color code as in panel A. The hCK2-AMPPNP structure (from PDB code: 2PVR) is shown in blue. ATP and AMPPNP substrates are shown as sticks and Mg ions as green spheres (from PDB code: 2PVR). Residues are labelled in black for PfCk2αD179S and in blue for hCK2. (D,E) Flat representation of the atomic interactions (generated with Ligplot45) between PfCK2αD179S and ATP (panel D, this work) and hCK2 (PDB code: 2PVR) and AMPPNP (panel E).

Full size image

PfCK2αD179S adopts an active conformation similar to hCK2

We compared the structure of PfCK2αD179S α with both hCK2 and protein kinase A (PKA, PDB accession code 1ATP), the first member of the kinase family whose structure was thoroughly characterized27. PKA is frequently used for ascribing an active or inactive form to a novel kinase structure. Out of the eleven kinase motifs classified by Hanks and Hunter28, the following structural elements: αC-helix, hinge region and the P-loop are particularly relevant to define a kinase active or inactive conformation. These three motifs cooperate to build an active enzymatic form, with the P-loop, which is located between the first two antiparallel β-strands in the N lobe main β-sheet, showing the most significant conformational flexibility between various kinase structures29. Based on the conformations adopted by these three elements, the present crystal structure reveals that PfCK2αD179S most resembles an active conformation26,30,31. Superposition of PfCK2αD179S with hCK2 reveals that the N-lobe displays larger differences with r.m.s. deviations of respectively 0.81 (PDB code: 3NSZ), 1.41 (3WAR), 1.46 (2PVR) and 1.83 Å (3FWQ) for α-carbon atoms of residues 12–121. Interestingly, the RMSD in the C lobe are similar (~0.8 Å) regardless of the exact hCK2 structure used for comparison, over the 131–331 amino-acid range, indicating that the PfCK2αD179S structure displays a good degree agreement with all of them. The PfCK2αD179S structure displays a well aligned R-spine32, involving residues Leu101, Leu89, Trp180 and His158 and its C-spine, formed by residues Val57, Ala70, Phe125, Ile166, Met167 and Ile168, Met225 and Met229 are characteristic of an active state (Fig. S1B), despite the fact that Phe125 departs from the position occupied by Met128 in PKA32,33, and conforms more to its counterpart Phe121 in hCK225. We also note that residues Ala70 and Val57 that are involved in the formation of the C-spine are not conserved compared to the residues found in hCK2: Val53 and Val66. Since their conformation is directly influenced by the adenine base upon substrate binding, this difference might be of relevance as small changes in the active site of CK2 have already been proven significant for determining the strength of inhibitor binding34.

Likewise, the hinge region encompassing strand β5 that links the N and C Lobes, including the polypeptide backbone between Phe117 and Phe125, adopts a conformation similar to the active forms of hCK2 (with RMSD for Cα atoms for residues 117–125 of 0.26 Å compared to the active hCK2 structure (3NSZ). Like in hCK2, the PfCK2αD179S hinge region accomodates the adenine moiety within the ATP binding site. Replacement of residue Val116 (hCK2) by Ile120 in PfCK2α leads to a small alteration of the shape of the ATP binding pocket. This amino-acid substitution has an impact on ATP binding as the main chain amino group of Val116 (of hCK2) makes a direct contact with the N1 amine of the adenine base and allows the carbonyl group of Glu114 to form a hydrogen bond with the N6 of the adenine base. These contacts are absent in the PfCK2-ATP complex (compare Fig. 5D,E)

The P-loop which includes residues Gly49–Tyr54 of PfCK2αD179S adopt an extended conformation similar to the one found in the active forms of CK2 (with RMSD 0.55 Å, 0.48 Å and 0.54 Å following superpositions with structures 3NSZ, 3WAR (the highest resolution structure available of an active CK2) and 2PVR respectively and it markedly differs from the inactive “collapsed” inactive form of CK2 (PDB code: 3FWQ), with a RMSD 2.35 Å. Nonetheless, several residues preceding the P-loop such as Met46 and Arg51, adopt a conformation distinct from either the active (RMSD 0.52 Å and 0.65 Å for structures 3NSZ and 2PVR respectively) or inactive hCK2 structures (RMSD 1.23 Å with PDB: 3FWQ). Interestingly, the β1 and β2 strands of PfCK2αD179S do not overlap with their counterpart structures in the active human structure (PDB code: 2PVR). Residue Gly50 is 3.28 Å away from the position adopted by the equivalent residue Gly46 in hCK2, which is due to a shift of the entire beta-sheet between the human and Plasmodium proteins Such displacements are induced by ATP binding (see below) and propagate across the beta blade with the Cα of residue Val109 of PfCK2 displaced by 5.83 Å from the position adopted by Val105 in the active human enzyme.

Protein-ATP interactions

The atomic interactions between PfCK2αD179S and ATP are depicted in Fig. 5B,C. An analysis35 of the substrate-protein interface in PfCK2αD179S and hCK2 (PDB code: 2PVR) reveals that the Plasmodium protein contributes an interface area of 353 Å2 compared to 309 Å2 for hCK2 bound to an ATP analog. An important substitution, Asp179Ser, was engineered for the catalytic characterization of the enzyme that resulted actually in improved protein crystallizability. The loss of the catalytic activity due to the mutation of Asp179 appears to have drastic consequence not only in the catalytic activity of the protein (Fig. 2A) but remarkably also on the positioning of the phosphate groups of the ATP molecule compared to hCK2 (PDB code: 2PVR). However we note that this substitution has no effect on the conformation of the catalytic loop. The loss of the interaction between the α-phosphate and Asp149 results in the triphosphate moiety being buried more deeply in the PfCK2 catalytic cleft, making ionic contact (distance of 3 Å) with Lys81 (Fig. 5D). As noted above, this closer contact formed with the “IGRG” motif (residues 49–52 of PfCK2) by the ATP molecule displaces the β1 strand from the conformation it adopts in the active human enzyme. The main consequence of this conformational change is an active site whose hydrophobic character appears to be enhanced (Fig. 5C), with residues Gly50, Gly52, Val57, Ala70, Lys72, Ile49, Ile120 Glu118, Met167 and Ile178 participating in binding ATP (Fig. 5). Compared to hCK2, (Fig. 5B) several residues that bind ATP are conserved and superpose well in both structures (PfCK2αD179S and 2PVR), including Lys72 (using PfCK2α numbering) which coordinates the α and β-phosphate groups of ATP and Glu85 that makes a salt bridge with Lys72. Besides the catalytically active residues, several other residues involved in ATP binding are also conserved, such as P-loop residues Gly50 to Tyr54 (Fig. 5A). However, as pointed above, in PfCK2αD179S, Gly50 and Gly52 Cα are significantly displaced by distances of 3.28 Å and 1.85 Å respectively, compared to hCK2 (PDB code: 2PVR) (Fig. 6).

Figure 6
figure 6

Conformational changes between the structure adopted by the N-Lobe of PfCK2αD179S (pink) and hCK2 (PDB code: 2PVR, yellow). A superposition of PfCK2αD179S (this work, colored in pink) and the active form of hCK2 (PDB code: 2PVR) shown in yellow. The three arrows show the main displacements described in the text at the level of the p-loop, Hinge and c-helix.

Full size image

Differences between PfCK2αD179S and hCK2

Six amino-acid substitutions are observed at the ATP binding site between the human (Arg43, Leu45, Glu55, Val66, His115, Val116) and Plasmodium enzyme (Lys47, Ile49, Asn59, Ala70, Tyr119, Ile120), giving rise to subtle alterations of the shape of the ATP binding pocket (Figs 3 and 5). As seen above, additional changes of the ATP binding pocket derive from a movement of strand β1. While C-lobe residues contributing to the ATP active site do not show significant variations compared to the active form of the human enzyme (PDB code: 2PVR), several changes arise from the N-lobe, with the Cα atoms of Gly50, Ile49 and Ala70 (located in the first and third strands of the N-Lobe) the most displaced. The main chain carbonyl atom of Ile49 comes to hydrogen bond distance with the O2 hydroxyl group of the ATP ribose (Fig. 5D). Likewise, the side chain CG1 atom of Val57 moves by 1.2 Å closer to the C4 of the base compared to hCK2. The aromatic side-chaisn of Tyr119 and Ile120 create a more hydrophobic interface compared to the human enzyme. Togetehr with Val57 and Ala70 they represent hydrophobic residues substitutions affecting the interaction with the adenine ring. In both cases Cα displacement (1.88 Å and 2.68 Å for Ala70 and Val57 respectively) allow the CG1 atom of Val57 to interact with the N9 amine of ATP. How substitutions of ATP binding residues in hCK2 significantly alter enzyme specificity was reported previously36 and thus, the alterations found in the PfCK2 ATP binding pocket described here and at other locations at the surface of the enzyme (Fig. 6) might be relevant for designing specific inhibitors of the Plasmodium enzyme.

Methods

Cloning, expression and purification of PfCK2α and its mutants

The cDNA of PfCK2α was cloned into the pNIC28-Bsa4 vector using ligation-independent cloning as described37. In order to obtain a soluble protein suitable for structure determination, we used various N- and C-terminal truncations of the protein and placed the hexa-Histidine tag at either end of the polypeptide chain (Fig. 1A). Protein expression was conducted using BL21(DE3) Rosetta T1R cells with kanamycin and chloramphenicol at 37 °C inoculating 1 liter of Terrific broth with a 2% (v/v) inoculum until an OD600 of 0.6–0.8. The temperature was decreased to 16 °C. After 30 minutes, Isopropyl β-D-1-thiogalactopyranoside (Affymetrix) was added to the culture to a final concentration of 0.5 mM. After overnight incubation, cells were harvested at 4000 × g for 10 minutes and re-suspended in the lysis buffer (100 mM Na Hepes, 500 mM NaCl, 10 mM Imidazole, 10% (v/v) Glycerol and 0.5 mM Tris (2-carboxyethyl) phosphine (TCEP) at pH 8.0. Pellets stored at −80 °C were thawed in an iced-water bath with 25 µl of EDTA-free protease inhibitor cocktail (Calbiochem) and 10 mg lysozyme added to the thawed cell re-suspension. Sonication was applied in cycles of 3 seconds pulses followed by 5 seconds delays at 30% amplitude for 5 minutes (Sonics Vibra-cell). Cells debris were discarded following centrifugation at 47.000 × g for 30 minutes at 4 °C. The supernatant was filtered with a 1.2 µm pore syringe filter and subsequently loaded onto a 1 ml Ni-NTA Superflow column (Qiagen). Proteins were eluted using a stepwise gradient with 8%, 22% and 100% of a 500 mM imidazole solution in a buffer containing 20 mM Na Hepes, 500 mM NaCl, 10% (v/v) Glycerol and 0.5 mM TCEP at pH 7.5. Fractions eluting from the 22% imidazole buffer were pooled and concentrated by ultra-filtration with a 10 kDa cutoff concentrator (Vivascience) and loaded onto a preparative Hiload 16/60 Superdex 200prep column (GE) pre-equilibrated in 20 mM NaHepes, 300 mM NaCl, 10% (v/v) Glycerol and 0.5 mM TCEP at pH 7.5. Elution of PfCK2α occurred as single and symmetric peak with an apparent molecular mass of 41 kDa. Protein identity was confirmed by SDS-PAGE (Fig. 1B) and mass spectrometry (39,665 Da). The protein was concentrated to 7 mg/ml, flash-frozen in liquid nitrogen and stored at −80 °C until use. The PfCK2αD179S mutant was generated using the QuickChange Protocol with forward primer GAAAATAGACAAATTAGATTAATTAGTTGGGGTCTAGCTGAATTTTATC, reverse primer GATAAAATTCAGCTAGACCCCAACTAATTAATCTAATTTGTCTATTTTC and template DNA vc026. Mutation was confirmed by sequencing. Expression and purification of the PfCK2αD179S mutant was performed following the same protocol as for the wild type enzyme. Using the same method, mutants PfCK2αY30A, PfCK2αT63A, PfCK2αT63AY30A, PfCK2αY30D, PfCK2αT63D, PfCK2αT63DY30D were obtained.

Protein kinase autophosphorylation assay

A quantity of 2 µg of either wild type or D179S mutant of the PfCK2α protein were incubated for 30 min at 30 °C in 20 µl of buffer containing 50 mM TrisHCl at pH 7.5, 50 mM MgCl2, [γ32P]-ATP and 8 µg of bovine serum albumin. The reaction was stopped by the addition of SDS sample buffer, boiled and the proteins resolved by SDS-PAGE. The gel was wrapped in a film and subjected to autoradiography with an X-ray film38.

Protein kinase assay

Catalytic activity was quantified by measuring the incorporation of radiolabeled phosphate into a human CK2 substrate RRRDDDSDDD at a final concentration of 300 µM, as described earlier39. Human CK2 (CSNK2 DU813) served as positive control. One unit of activity was defined as the incorporation of 1 nanomole of radioactive phosphate into the substrate per minute. Specificity was calculated by dividing activity units by the amount in milligrams of assayed purified protein and substracting background level. Results were the averages of duplicate measurements.

Inhibition activity assay

Two replicates were performed for hCK2 and PfCK2, respectively. 2.5 μl of incremental concentrations of CX4945 compound was assayed with either enzyme in 50 μl total volume consisting of 50 mM Tris-HCl at pH 7.5, 10 mM DTT, 0.1 mM EGTA, 300 μM peptide substrate, 10 mM magnesium acetate and 100 μM γ33P-ATP. The reaction was incubated at 30 °C for 10 mins. Reactions were stopped by spotting 40 μl out of the 50 μl assay mixture onto 1.5 cm × 1.5 cm square of Whatman P81 paper, which were washed in 75 mM phosphoric acid, followed by acetone, then air dried. 1 ml Microscint 0 was added to tubes containing dried filter papers before counting on a liquid scintillation counter.

Crystallization

PfCK2αD179S at a concentration of 10 mg.ml−1 was screened for crystallization with commercial kits in presence of 1 mM ATP using a mosquito® crystallization robot (TTP Labtech). Very thin needles were obtained immediately after setting up, in Morpheus A9 conditions: 0.06 M Divalents; 0.1 M buffer system 3 at pH 8.5, 50% (v/v) Precipitant Mix 1, with a protein:precipitant ratio of 2:1. To grow larger crystals, these thin needles were added into protein solution, right before mixing with the precipitant. Crystals suitable for diffraction obtained by microseeding in 0.2 M Lithium Sulfate, 0.1 M BIS-TRIS pH 5.5, 25% (w/v) Polyethylene glycol 3,350 at 20 °C are displayed in Fig. 1A. Prior to flash-freezing in liquid nitrogen, crystals were transferred into a cryoprotectant solution containing the precipitant solution supplemented with 30% (v/v) glycerol.

Data collection, structure determination and refinement

Diffraction data were collected on the PX-III beamline at SLS (Paul Sherrer institute, Villigen, Switzerland) and integrated using XDS40. Solvent content analysis using CCP424 indicated the presence of three monomers in the asymmetric unit. The structure was determined using the molecular replacement program PHASER24 with the hCK2 structure (PDB code: 3NSZ) as a probe. Manual building using COOT41 was combined with refinement using BUSTER-2.1042 with tight NCS restraints between the three molecules related by ncs. Images of the structure were generated using PyMOL (http://www.pymol.org).

Microcalorimetry

Isothermal titration microcalorimetry experiments were performed at 25 °C with a PEAQ-ITC isothermal titration calorimeter (Malvern). Protein concentration in the microcalorimeter cell (0.2 mL) was 50 μM. A total of 19 injections of 2 μl of ATP at a concentration of 500 μM were performed at intervals of 180 s while stirring at 600 rpm. The experimental data were fitted to theoretical titration curves using the manufacturer’s software.

Mass spectrometry analysis

LC-MS/MS was done as previously described with minor modifications43. Briefly, PfCK2 wild type and mutant protein gel bands were excised, reduced with DDT and alkylated with IAA, and then digested with trypsin. Tryptic peptides were separated and analyzed on a Dionex Ultimate 3000 RSLCnano system coupled to a Q-Exactive mass spectrometer (Thermo Electron, San Jose, USA) using a 60 min gradient. A full MS scan (350–1600 m/z range) was acquired at a resolution of 70,000 at m/z 200 and a maximum ion accumulation time of 100 ms. Dynamic exclusion was set as 30 s. Resolution for HCD spectra was set to 17,500 at m/z 200. The AGC setting of full MS scan and MS2 were set as 106 and 2 × 105, respectively. The ten most intense ions above a 1, 000 counts threshold were selected for HCD fragmentation with a maximum ion accumulation time of 100 ms. Isolation width of 2 Th was used for MS2. Single and unassigned charged ions were excluded from MS/MS. For HCD, normalized collision energy was set to 28%. The underfill ratio was defined as 0.1%. The MS raw file was converted into mgf format using ProteomeDiscoverer version 1.4. Protein identification was performed using Mascot server (version 2.4.1, Matrix Science, Boston, MA) against a Plasmodium falciparum protein database including the WT and MT PfCK2α proteins (containing 5,649 sequence and 431,5306 residues). Mascot search was limited to a maximum of two missed trypsin cleavages, #13 C of 2, mass tolerance of 5 ppm for peptide precursors, and 0.02 Da mass tolerance for fragment ions. Fixed modification was carbamidomethyl at Cys residues, while variable modifications included oxidation at methionine residues, deamidation at asparagine and glutamine, and phosphorylation at serine, threonine and tyrosine. The extents of phosphorylation of detected phosphorylation sites were determined by integrating the areas of the extracted ion chromatograms of the unphosphorylated and phosphorylated peptides. The extracted ion chromatograms were extracted from a windows of +/−5 ppm of the precursor ions (see supplementary data).

Data availability statement

Protein Data Bank accession code: The atomic coordinates and structure factors have been deposited in the Protein Data Bank with accession code 5XVU.