Allosteric activation of apicomplexan calcium-dependent protein kinases

Abstract

Calcium-dependent protein kinases (CDPKs) comprise the major group of Ca2+-regulated kinases in plants and protists. It has long been assumed that CDPKs are activated, like other Ca2+-regulated kinases, by derepression of the kinase domain (KD). However, we found that removal of the autoinhibitory domain from Toxoplasma gondii CDPK1 is not sufficient for kinase activation. From a library of heavy chain-only antibody fragments (VHHs), we isolated an antibody (1B7) that binds TgCDPK1 in a conformation-dependent manner and potently inhibits it. We uncovered the molecular basis for this inhibition by solving the crystal structure of the complex and simulating, through molecular dynamics, the effects of 1B7-kinase interactions. In contrast to other Ca2+-regulated kinases, the regulatory domain of TgCDPK1 plays a dual role, inhibiting or activating the kinase in response to changes in Ca2+ concentrations. We propose that the regulatory domain of TgCDPK1 acts as a molecular splint to stabilize the otherwise inactive KD. This dependence on allosteric stabilization reveals a novel susceptibility in this important class of parasite enzymes.

Keywords: VHH; calcium-dependent protein kinase; kinase activation.

Fig. S1.

TgCDPK1 truncations lacking the regulatory domain are catalytically inactive. (A) Two constructs were expressed and assayed for kinase activity: Truncation I lacked the CAD completely, whereas truncation II retained the autoinhibitory α-helix. The constructs’ sequence coverage of WT-enzyme amino acids is indicated in parentheses. Values are normalized to maximal activity. Mean ± SD, representative experiment. (B) The isolated TgCDPK1 KD cannot be activated by the CAD in trans. The purified KD yielded no measurable activity even at the 5 µM tested. Addition of purified CAD to the KD did not lead to any measurable kinase activity. Values are normalized to maximum WT activity. Mean ± SEM, n = 3 independent experiments.

Fig. 1.

TgCDPK1 is not activated by removal of its regulatory domain. (A) Introduction of a 3C protease cleavage site between the CAD and KD of TgCDPK1, in the kinase referred to as 3C, enables separation of the two domains upon protease treatment. The wild-type (WT) enzyme was unaffected by the protease treatment. The CAD and KD could be further isolated from the digested 3C. (B) Kinase activity of WT and 3C, before (–) and after (+) protease treatment. Mean ± SEM, n = 3 independent experiments.

Fig. 2.

1B7 recognizes a calcium-dependent conformation of TgCDPK1. (A) IRDye800-labeled 1B7 was used to probe decreasing concentrations of total T. gondii lysate or recombinant kinase (Top). Samples were separately probed for parasite actin (ACT1; red) or the 6xHis-tag on recombinant proteins (HIS; green). (B) Incubation of parasite lysates with immobilized 1B7 depletes TgCDPK1/TgCDPK3 in the presence of CaCl₂ but not in EGTA. The immunoprecipitated material could be eluted from the beads by subsequent incubation with EGTA. Blot was probed with 1B7 (green), and ACT1 (red) is included as a control. (C) SEC of 1B7-TAMRA and TgCDPK1, alone or incubated at equimolar concentrations, in buffer containing 1 mM CaCl₂ or EGTA. Absorbance was recorded at 280 nm for total protein (blue) and 550 nm for 1B7-TAMRA (red). (D) IP of recombinant kinases by 1B7 in the presence of 1 mM CaCl₂ or EGTA. The domain structure of each kinase, indicating the KD and each of the EF hands (I–IV), is depicted for wild-type (WT), D368A/D415A (N-lobe mutant), and D451A/D485A (C-lobe mutant). Means ± SEM, n = 3 independent experiments. Insert shows the results from a representative experiment. (E) Calcium dependency of TgCDPK1 activation or 1B7 binding. Values are normalized to maximal activity or binding. Means ± SEM, n = 3 independent experiments.

Fig. S2.

Identification of heavy chain-only antibodies to study the structure of TgCDPK1. (A) Individual or pooled serum from alpacas was tested for reactivity against T. gondii lysate by immunoblot. The secondary antibody showed no reactivity on its own. (B) Panning of a combined VHH phage display library from alpacas 1–4 and 8 against immobilized TgCDPK1 enriched for a group of nearly identical sequences from which 1B7 was selected. Sequences from 96 clones before enrichment are shown for comparison. (C) Sequence of 1B7 highlighting the variable loops (red), the sortase substrate motif (green), and the 6xHis-tag (blue). (D) Expression and purification of 1B7. Following the induction of 1B7 expression by IPTG, a lysate was prepared. The lysate was collected on a Ni-NTA column, and a sample of the flow-through was collected. The column was washed and 1B7 was eluted with buffer containing 500 mM imidazole. The eluted material was further enriched by SEC, and peak fractions were collected and concentrated. Samples collected during the purification were resolved by SDS/PAGE and stained with Coomassie for total protein.

Fig. S3.

Extended dilution series of recombinant TgCDPKs. (A) Western blots using either IRdye800-labeled 1B7 or an antibody recognizing the 6xHis-tag of the recombinant proteins. (B) Densitometric analysis of Western blots showing similar recognition of both kinases by 1B7.

Fig. S4.

Ca²⁺ dependency of 1B7 binding. (A) 1B7 binds TgCDPK1 reversibly. Precipitation of recombinant TgCDPK1 by covalently immobilized 1B7 is shown. The experimental setup is diagrammed on top. Kinase was incubated with 1B7 beads, and the free [Ca²⁺] was modulated by adding EGTA or CaCl₂, as indicated. Samples were taken after each free [Ca²⁺] change to determine whether complex formation had occurred, as measured by the presence of TgCDPK1 in the eluate. (B) Immunoblot comparing the detection of different kinases by IRDye800-labeled 1B7 (green). The domain structure of each kinase, indicating the KD and each of the EF hands (I–IV), is depicted for wild type (WT), D368A/D415A (N-lobe mutant), and D451A/D485A (C-lobe mutant). Detection of recombinant kinases by their His-tags (red) is included as a loading control.

Fig. 3.

1B7 inhibits TgCDPK1 and related kinases. (A) In vitro kinase assay with increasing concentrations of 1B7. Mean ± SEM, n = 3 independent experiments. (B) TgCDPK1-dependent (6-Fu-ATPγS) and total (ATPγS) thiophosphorylation in parasite lysates incubated with varying 1B7 concentrations. Thiophosphorylation is visualized by Western blot with rabbit mAb 51-8 (red); 1B7 staining is shown as a loading control (green). (C) Immunoblot of P. falciparum lysate from strain 3D7 or W2mef probed with 1B7-IRdye800 (green) or an antibody recognizing parasite actin (ACT1; red).

Fig. 4.

1B7 stabilizes a novel conformation of TgCDPK1. (A) Cartoon depicting the structure of the complex with the catalytic KD (gray), the CaM-like CAD (orange), bound calcium ions (green spheres), and 1B7 (blue). The position of the active site pocket is indicated (red arrow). (B) Specific interactions within the CAD:1B7 binding interface. Hydrogen bonds and salt bridges are depicted as black dotted lines. Interacting residues are shown as sticks for the CAD and 1B7. Cys51, in the beta strand immediately preceding CDR2, and Cys104, in CDR3, form a stabilizing, intrachain disulfide in 1B7. CDR1 on 1B7 makes no contacts with TgCDPK1.

Fig. S5.

Details of the Ca²⁺/1B7-bound TgCDPK1 structure. (A) CDRs of 1B7. CDR1 (yellow), CDR2 (orange), and CDR3 (red) of 1B7 are colored. The orientation of 1B7 shown is the same as in Fig. 4A. (B) Ca²⁺ occupancy of the TgCDPK1 EF hands in the presence of 1B7. Stick representation of the four EF hands that comprise the CAD is shown. The built Ca²⁺ ions are depicted as green spheres. The electron density shown is from a Sigma-A weighted difference (mFo-DFc) omit map for Ca²⁺ contoured at 6 σ. (C–E) Representative images of the final 2mFo-DFc electron density map contoured at 1 σ. 1B7 (teal) and the CAD (orange) are shown as sticks, and calcium ions (green) are depicted as dots. (C) View of the hydrophobic core of the CAD C lobe. (D) Interaction site between CDR3 of 1B7 and EF hand II of the CAD. (E) Interaction site between CDR3 and the N-lobe of the CAD.

Fig. S6.

Sequence comparison of Apicomplexan CDPKs. (A) Multiple sequence alignment of the 1B7-binding region of the CAD of 30 CDPKs from T. gondii, P. falciparum, and C. parvum. Residues known to interact with 1B7 at the six sequence-specific CAD positions are highlighted in red and numbered according to their position in TgCDPK1. Conservative mutations expected to still interact with 1B7 are highlighted in green. The remaining residues are highlighted white to blue according to sequence identity. Nonconserved insertions are omitted for clarity. Phe364 and Glu426 are conserved in nearly all CDPKs. His365 is only present in TgCDPK1. Glu423 or the conservative aspartate mutant are present in most CDPKs. Positions 350 and 361 are considered the important sequence determinants. Although non–1B7-binding CDPKs may have a conservative mutation (i.e., K350R or T361S) at an individual position, none of these homologs have the correct identity at both of the sites. (B) Phylogenetic tree of TgCDPK1 homologs. (C) Table of sequences used.

Fig. 5.

1B7 prevents reorganization of the CAD to its activation-associated conformation. (A) Comparison of 1B7-bound TgCDPK1 to the inactive (3KU2) and active (3HX4) conformations. The three TgCDPK1 conformations are aligned by their KDs (gray), depicting the CAD (orange), 1B7 (blue), bound calcium (green), and unknown cations (black). ANP (phosphoaminophosphonic acid-adenylate ester) is built as yellow sticks to indicate the active site. The first helix of the CAD (CH1) is labeled with “1.” (B) Aligned carbon trace of the CADs from the active (blue) and 1B7-bound structures (orange). (C) Superposition of 1B7 and the bound CAD as ribbon diagrams over a space-filling model of the active structure.

Fig. S7.

A disulfide bond stabilizes the structure of Ca²⁺/1B7-bound TgCDPK1. (A) Disulfide bond (yellow) between Cys247 of the KD (gray) and Cys505 of the CAD (orange) rendered as sticks in the Ca²⁺/1B7-bound TgCDPK1 structure. (B) 1B7 inhibition of TgCDPK1 is not affected by reducing conditions. Addition of up to 10 mM DTT did not affect kinase activity or inhibition by 1B7. Kinase and 1B7 were assayed at 100 nM and 150 nM, respectively. Values are normalized to maximal WT activity. Mean ± SD, representative experiment.

Fig. 6.

The KD of TgCDPK1 is intrinsically inactive and stabilized by the Ca²⁺-bound CAD. (A) Superposition of the active conformation of the TgCDPK1 KD with an inactive conformation obtained from a simulation without the CAD, in which the KD readily departed from the (starting) active conformation. The superposition is obtained using the C-lobe backbone atoms. (B) The rmsd trace of the αC helix relative to the (starting) active conformation in simulations of the TgCDPK1 KD with or without the CAD. Conformations with an rmsd less than 3 Å are considered active. (C) The relative occupancy of the catalytically active conformation in simulations of the TgCDPK1 KD with or without the CAD. As shown, the presence of the CAD stabilizes the active conformation. (D) Structure of the KD highlighting the integrity of the R-spine (red) in the Ca²⁺-bound (active) conformation, and its disruption in the Ca²⁺-depleted (inactive), the simulation without the CAD (–CAD), and the Ca²⁺/1B7-bound (1B7-bound) conformations. (E) Kinase activity of TgCDPK1 mutants carrying different flexible linkers between the KD and CAD domains. (F) Kinase activity of different N-terminal truncations removing the first 35 (Δ35) or 44 (Δ44) amino acids preceding the KD. (G) Kinase activity of point mutants in the N-terminal extension, alone or in combination. All kinase experiments plotted as means ± SEM, n = 3 independent experiments.

Fig. S8.

Superposition of the Ca²⁺/1B7-bound and Ca²⁺-bound structures. (A) The two structures are shown individually and superimposed. For clarity, the KD of the Ca²⁺/1B7-bound structure is not shown in the superimposed image. (B) Detailed view of the predicted clash between the N-terminal extension (red) in the active, Ca²⁺-bound structure and CDR3 of the VHH (blue). The KD (gray) and CAD (yellow) from the active structure are shown.

Fig. 7.

Allosteric activation is required for TgCDPK1 activity in vivo. (A) Strategy depicting the TgCDPK1 cKO carrying an HA9-tagged allele that can be shut down by addition of ATc. The complemented strains additionally carry a constitutive Myc-tagged allele (WT or F39A), expressed under the endogenous TgCDPK1 promoter. (B) Immunoblot of regulatable (HA9) or constitutive (Myc) TgCDPK1 alleles in the different strains grown in the presence or absence of ATc for 48 h. Parasite actin (ACT) is included as a loading control. (C) TgCDPK1-dependent thiophosphorylation (51-8; red) in lysates from the various parasite strains following growth in the presence of ATc. Reactions were performed in the presence or absence of Ca²⁺. Tubulin is included as a loading control. (D) Plaque formation by the various strains in the presence or absence of ATc. (E) Model for the activation of TgCDPK1 and its inhibition by 1B7. The KD (gray) is intrinsically inactive and its catalytic site is occluded by the CAD (orange). At ∼300 nM Ca²⁺, the CAD is partially occupied by Ca²⁺ and can bind 1B7 (blue). Higher Ca²⁺ concentrations are required for full Ca²⁺ occupancy of the CAD and stabilization of the KD, which leads to activation. The region of the CAD competitively bound by 1B7 and the N-terminal extension is circled and highlighted in the Inset.