Crystal Structures of the Carboxyl cGMP Binding Domain of the Plasmodium falciparum cGMP-dependent Protein Kinase Reveal a Novel Capping Triad Crucial for Merozoite Egress

The Plasmodium falciparum cGMP-dependent protein kinase (PfPKG) is a key regulator across the malaria parasite life cycle. Little is known about PfPKG’s activation mechanism. Here we report that the carboxyl cyclic nucleotide binding domain functions as a “gatekeeper” for activation by providing the highest cGMP affinity and selectivity. To understand the mechanism, we have solved its crystal structures with and without cGMP at 2.0 and 1.9 Å, respectively. These structures revealed a PfPKG-specific capping triad that forms upon cGMP binding, and disrupting the triad reduces kinase activity by 90%. Furthermore, mutating these residues in the parasite prevents blood stage merozoite egress, confirming the essential nature of the triad in the parasite. We propose a mechanism of activation where cGMP binding allosterically triggers the conformational change at the αC-helix, which bridges the regulatory and catalytic domains, causing the capping triad to form and stabilize the active conformation.


Introduction
The malaria parasite Plasmodium falciparum has a complex life cycle comprising phases in both a human host and a mosquito vector [1][2][3]. Once the sporozoites are injected into a person by a mosquito during a blood meal, they must quickly relocate to the liver through the bloodstream and invade hepatocytes, where they then differentiate and generate thousands of liver stage merozoites within a single schizont. Following egress, the merozoites must then invade red blood cells (RBCs) and again multiply by asexual replication. A blood stage schizont is formed within 48 hours of red blood cell invasion and releases up to 32 new merozoites into the blood stream to invade fresh red cells. Proliferation of these asexual blood stage parasites leads to pathology, and furthermore, a small proportion of the asexual blood stage parasites commits to sexual development whereby distinct male and female gametocytes are formed, which are required for transmission to mosquitoes.
The complex malaria parasite life cycle is a highly regulated process, but the molecular details are poorly understood. Several biochemical and genetic studies indicate that cyclic nucleotides play a critical role at a number of stages in the Plasmodium life cycle [4,5]. Recent studies discovered that the P. falciparum cGMP-dependent protein kinase (PfPKG) is involved in both asexual and sexual development [6][7][8] and specific inhibition of PfPKG blocked progression of the life cycle at multiple stages [6,7,9].
P. falciparum expresses a single PKG that has similar functional domains to mammalian PKG (mPKG), but exhibits some fundamental differences in domain organization. PfPKG contains an N-terminal regulatory (R) domain including a possible auto-inhibitory (AI) sequence and a C-terminal catalytic (C) domain [10]. Both the R-and C-domains have approximately 30-40% sequence identity to their counterparts in mPKG. The major differences are that PfPKG is larger than mPKG and does not contain an N-terminal dimerization domain, thus functioning as a monomer (Fig. 1A). Moreover, the R-domain of PfPKG contains four consensus cyclic nucleotide-binding domains (PfCNB-A, B, C and D), whereas mPKG has only two (mCNB-A and B) [11]. A previous study indicated that PfCNB-C is degenerate since cGMP binding could not be detected [12]. The CNB domains commonly include an eight-stranded -barrel and a variable number of -helices [13]. Embedded within the -barrel, a key structural motif, the Phosphate Binding Cassette (PBC), binds the sugar-ribose portion of cGMP and consists of a short helix (P-helix) followed by a loop.
Cyclic GMP binding to PfCNB-D has been shown to have the greatest effect on kinase activation [12]. In this study, each domain was disabled by mutation of a putative cGMP-binding residue and its effect on PfPKG activation was analyzed. Disabling PfCNB-D resulted in a 10fold increase in the activation constant for cGMP (Ka :cGMP ) and as much as a 50% reduction in maximal activity. In contrast, disabling PfCNB-A, B, or C showed little to no effect. Since little  structural information is available for mPKG and none for PfPKG, the current model of activation is based on solution-based studies of mPKG [14,15]. The current model suggests that, in the absence of cGMP, the R-domain binds to the C-domain with high affinity, with the AI sequence bound within the catalytic cleft and preventing kinase activity. Cyclic GMP binding alters the conformation of the R-domain and releases the C-domain, causing activation. However, the details of the regulation and activation mechanisms are poorly understood. Here we report high-resolution crystal structures of PfCNB-D in the presence and absence of cGMP. The data reveals not only the structural basis of cGMP selectivity, but also a PfPKG-specific capping triad that allosterically triggers kinase activation. Mutagenesis of the capping triad residues impairs kinase activation and abrogates P. falciparum blood stage merozoite egress.

PfCNB-D binds cGMP with the highest affinity and selectivity
While all PfPKG CNB domains have been shown to influence kinase activation to varying degrees [12], their individual affinities for either cGMP or cAMP have never been investigated. To assess their affinities, we purified each domain and measured their affinities for cGMP and cAMP using a competition fluorescence polarization (FP) assay. We expressed the CNB domains in E. coli Δcya TP2000, which lacks adenylyl cyclase activity [16], hence avoiding bacterial cAMP contamination. Competition FP measurements show that the isolated PfCNB-D domain binds cGMP with a half maximal effective concentration (EC 50 ) of 38 nM, while the A and B domains bind cGMP with EC 50 values ranging from 1 to 1.2 μM (Fig. 1B). In contrast, PfCNB-D binds cAMP with an EC 50 of 7.4 μM whereas other domains show values ranging from 12 to 77 μM. Thus, PfCNB-D binds cGMP with the highest affinity and is over 190-fold selective for cGMP. PfCNB-C shows very weak affinity for either 8-Fluo-cGMP or 8-Fluo-cAMP. This is expected because PfCNB-C lacks the key cGMP interacting residues in its cGMP binding pocket (S1 Fig.) [10,12].
Structure determination of PfCNB-D in the apo and cGMP bound states and overall structures of PfCNB-D To understand the structural basis of its high affinity and selectivity for cGMP required for activation, we solved the structures of PfCNB-D, both in the apo and cGMP bound states (Figs. 1C and 1D). The crystal structure of the apo protein was solved at 1.89 Å using the structure of CNB-B of human PKG Iα (residues 205-324) as a search model for molecular replacement (MR) (PDB code: 3SHR) [17]. The final model containing residues 403-526 shows four ordered sulfate ions, with one of the sulfates occupying the cyclic phosphate site due to a high sulfate concentration in the crystallizing condition. (Fig. 1C). The N-terminal portion of the αC-helix (residues 520-526) forms a loop, providing crystal contacts with the PBC of the neighboring molecule in the crystal lattice (S2 Fig.). The rest of the αC-helix (residues 527-542) is not modeled due to the lack of electron density (Fig. 1C). The structure of the PfCNB-D:cGMP complex was solved at 2.0 Å by MR using its apo structure as a search model. The final model of the complex includes residues 401-542. The entire structure shows robust electron density including the αC-helix (residues 520-542) (Fig. 1D). Additionally, the cGMP pocket shows clear electron density for cGMP bound in a syn conformation (S3 Fig.). While the overall structure of PfCNB-D is similar to that of the highly cGMP selective human PKG I, the conformation of the C-terminal helix and the capping interaction at the cGMP pocket are strikingly different [18,19]. The structures of PfCNB-D show a typical cyclic nucleotide binding domain fold consisting of an 8 stranded -barrel flanked by α helices at both termini [13,20]. Structural comparison of the apo and cGMP bound states also shows that the helical regions undergo major structural changes whereas the -barrel does not. The αC-helix is not ordered in the absence of cGMP, but becomes ordered upon binding cGMP and shields the cGMP pocket (Fig. 1). Statistics for crystallographic data and structural refinement are summarized in Table 1.
The PfCNB-D:cGMP complex shares many structural features with human PKG I, but shows a distinct capping mechanism.
The crystal structure of the PfCNB-D:cGMP complex shows that, while the overall structure of its cGMP pocket is similar to that of the human PKG I CNB-B, its capping mechanism is completely different and the pocket is highly shielded from solvent due to the fully ordered C-helix ( Fig. 2A) [18,19]. The cGMP pocket can be broken into four sites, and interactions at the first three sites are similar to those seen in the human PKG I CNB-B:cGMP complex  Figs. 2A and 2B). These include E483 and A485, which capture the sugar-phosphate of cGMP through hydrogen bonds (Site 1), T493, which bridges the O2A of cyclic phosphate and the N2 of guanine though hydrogen bonding interactions (Site 2), and L472 and R473 at strand 5, which specifically recognize the guanine moiety through Van der Waals (VDW) interaction and hydrogen bonds (Site 3). Site 4 is distinct from mPKG and consists of R484 at the PBC and Q532/D533 on the αChelix, which form a unique capping interaction with cGMP. As seen in Fig. 2B, the side chains of R484 and Q532 form a hydrogen bond and stack against the guanine moiety of cGMP, shielding the entire cGMP through VDW contacts. In particular, the hydrophobic arm of R484 shields the sugar-phosphate whereas its guanidinium group along with the side chain of Q532 shields the guanine moiety, providing a hydrophobic cap for cGMP. Additionally, R484 forms a salt bridge with D533. The guanine moiety of cGMP becomes sandwiched between the R484/Q532 and I465 at strand 4 ( Fig. 2B). Moreover, an ordered water molecule captured between E483, R484, and Q532 interlinks these residues, further stabilizing the capping interaction (Fig. 2B). R484 of PfPKG corresponds to K308 in human PKG I ( Figs. 2A and 2B). However, K308 forms no interaction with the αC-helix or the bound cGMP, but instead interacts with a neighboring Q311 through a hydrogen bond ( Fig. 2A). Despite little sequence similarity at the αC helix, the side chain of Q532 in PfPKG aligns structurally with that of the aromatic side chain of Y351, the conserved capping residue in human PKG Iβ, further confirming its role as a capping residue for cGMP (Figs. 1A and 2A). Finally, an additional αChelix residue, V536, interacts with L472 at strand 5 through a VDW contact and shields the guanine moiety from solvent ( Fig. 2A).

The unique capping interaction is essential in cGMP binding and kinase activation
To test the role of the capping triad residues in cGMP binding, we mutated these residues to alanine both in PfCNB-D and in the full length R-domain and measured their affinities for cGMP and cAMP using a competition FP assay ( Table 2). Our measurements showed that mutating any of the capping triad residues drastically reduced its affinity for cGMP, but only showed a slight reduction in its cAMP affinity. Specifically, in the truncated PfCNB-D, mutating the capping triad increases its EC 50 values for cGMP from 38 nM to 7 μM whereas the same mutations increase the values for cAMP from 7.4 μM to 65 μM. In the full length R-domain containing all four CNBs, the same set of mutations increases its EC 50 values for cGMP from 79 nM to as much as 1.4 μM, compared with the cAMP values changing from 9.6 μM to 21 μM further demonstrating the important role of the capping triad in high affinity cGMP binding.
Next, we generated the same set of mutations in full-length PfPKG (residues 1-853) (S6 Fig.) and measured its effect on kinase activation using a microfluidic mobility-shift assay (Figs. 3A and 3B). Singly mutating R484 or Q532 to alanine increased the activation constants for cGMP (K a:cGMP ) from 66 nM to 2-3 μM, whereas mutating D533 showed only a slight increase (to 330 nM) (S1 and S2 Tables). In addition, mutating all three residues increased its K a:cGMP over 28-fold (to >1.8 μM) and reduced its maximum activity by 90% (Figs. 3A and 3B). In summary, our data demonstrate that the capping triad is crucial not only for cGMP binding, but also for cGMP-dependent activation of PfPKG.
Mutating the capping triad residues disables merozoite egress in P. falciparum To determine the role of the capping triad on malaria parasite development, we utilized a chemical genetic approach employing a selective PKG inhibitor (compound 2, an imidazopyridine) and a transgenic parasite line expressing an engineered PKG allele (incorporating a threonine to glutamine substitution, T618Q) that confers inhibitor resistance [6][7][8]21]. While compound 2 inhibits endogenous PfPKG by occupying a hydrophobic pocket adjoining the ATP-binding domain, the T618Q mutant has a bulkier side chain that prevents inhibitor binding and is no longer sensitive to compound 2 retaining kinase activity. . Key residues that stabilize the C-helix including the capping residues are shown with transparent surface in the following color theme: side chain carbon, black; oxygen, red; nitrogen. A water molecule captured between E483, R484, and Q532 is shown as a blue sphere. The C atoms of glycine residues located between at the αB and αC helices are shown as black spheres. Hydrogen bonds are shown as dotted lines. (B) Detailed interactions between PfCNB-D and cGMP. Zoomed in views for each cGMP binding site are shown on either side. The backbone amide of A485 is marked with a blue dot. The individual cGMP interacting residues are shown with the following color theme: side chain carbon, black; oxygen, red; nitrogen, blue. The residues binds cGMP with VDW contacts including the capping residues are shown with transparent surface. Hydrogen bonds are shown as dotted lines with their distances in Å units. In this study, we generated three new P. falciparum transgenic lines containing an ectopic copy of the full length HA-tagged pkg gene harboring the T618Q mutation, but also containing either the R484A mutation (PKG T618Q/R484A -3xHA), the R484A/Q532A/D533A triple mutation (PKG T618Q/R484A/Q532A/D533A -3xHA) or no additional mutation (PKG T618Q -3xHA) as a control. The HA-tagged pkg genes were integrated as a single copy into a modified pseudogene locus via Bxb1 integrase mediated attB/attP recombination (S7 Fig.) [22]. Expression of the ectopic copy is driven by the parasite-derived ama1 promoter displaying a schizont-specific activity profile similar to the endogenous pkg promoter. In the absence of compound 2, these lines all express two functional copies of the PfPKG enzyme and expression of HA-tagged PKG is comparable between the three lines ( Fig. 3C). All transgenic lines grow at equivalent rates to the parental line (3D7/attB) and an additional untagged control line (PfPKG T618Q ), in which the gatekeeper mutation was previously introduced by allelic replacement (S8 Fig.) [6]. These controls confirm that over-expression of PfPKG has no effect on growth rate and that expression of the capping triad mutants has no dominant negative effect (S8 Fig.). Upon addition of compound 2, the endogenous PfPKG is fully inhibited, allowing us to measure the effects on parasite development of the mutant PfPKGs and to test whether they could rescue the chemically inhibited endogenous enzyme (S9 Fig.). Growth of the three transgenic lines in the presence of compound 2 (1 μM) was compared to that of the parental line and the original untagged T618Q allelic replacement control line. PfPKG T618Q/R484A -3xHA and PfPKG T618Q/R484A/Q532A/ D533A -3xHA transgenics did not proliferate for the duration of the experiment (7 days) whereas the PfPKG T618Q -3xHA control was able to grow at a similar rate to the allelic replacement control line of PfPKG T618Q (Fig. 3D). Examination of Giemsa-stained parasites in blood films showed that transgenic lines with the capping triad mutations, accumulated at the schizont stage and did not progress to ring stage parasites that result from merozoite egress and invasion of new erythrocytes in the presence of compound 2 (Fig. 3E). By contrast, high levels of ring stage parasites were observed in untreated cultures, the PfPKG T618Q replacement line and the PfPKG T618Q -3xHA control line. This block in parasite development was examined in more detail by flow cytometry and the results confirmed the requirement of an inhibitor-resistant  (7) 60.8 ± 1.9 μM (7) D533A 760 ± 40 nM (7) 37.3 ± 1.8 μM (7) R484A/Q532A/D533A 7.2 ± 0.3 μM (7) 65.1 ± 3.2 μM (7) PfPKG full regulatory domain  Wild type 79.0 ± 4.0 nM (10) 9.6 ± 0.8 μM (7) R484A 960 ± 20 nM (7) 19 (9) 21.0 ± 1.4 μM (7) * FP measurements were at least in duplicate. EC 50 is the half-maximal effective concentration, and SEM is standard error of mean.  PfPKG to progress from schizont to ring stage parasites (Fig. 3F). Taken together, the results demonstrate that mutating the capping residues impairs blood stage egress and reinvasion. Moreover, the results are consistent with our previous findings that PfPKG activity is essential for merozoite egress and subsequent invasion of erythrocytes and confirms that the capping triad is required for PfPKG activity in vivo.

Discussion
PfCNB-D functions as a "gatekeeper" for activation of PfPKG Our study of PfCNB-D has demonstrated its role as a "gatekeeper" domain for cGMPdependent activation. Since cGMP binding is the first step in activation, it is essential to understand how cGMP binds the R-domain and what structural changes are associated with this binding event (see below). While there have been several studies reporting binding affinities of mPKG [18,23,24], the binding affinities for PfPKG prior to this study were unknown. Previous mutagenesis studies of PfPKG and its orthologue from Eimeria tenella (EtPKG) showed that, while all CNB domains are required for maximal activation, disabling the CNB-D domain had the largest effect on activation [12,25]. Consistent with these studies, we demonstrate that the largest effect on activation is due to PfCNB-D having the highest affinity and selectivity for cGMP (Fig. 1B). Additionally, since cGMP binding to all three functional CNB domains is required for maximum kinase activity [12,25], the high selectivity of PfCNB-D for cGMP would limit activation by cAMP, and facilitate full activation selectively by cGMP. Therefore, we suggest that PfCNB-D serves as "gatekeeper" for PfPKG activation.
Stepwise transition from the apo to the cGMP bound state.
Structural comparison of the apo and cGMP bound states reveals two distinct sets of hydrophobic networks that link the PBC to the rest of the CNB domain. Careful analysis suggests that these networks transmit cGMP-induced changes at the PBC to the rest of the helical regions and allow PfCNB-D to transition from one state to the other (Figs. 1, 4 and 5). In the absence of cGMP, the PBC is open with its αP-helix positioned away from the -barrel. The αB-helix, which interacts directly with the PBC, is tilted away from the -barrel ( Figs. 4 and 5A). The α3 10 -helix of the N3A motif interacts directly with the PBC (Figs. 5B and S4A). Three phenylalanines from different regions, F419 (α3 10 -helix), F434 (αA-helix) and F455 (3), together with the hinge residues, L486 (αP-helix) and F513 (αB-helix) as previously defined in the CNB domain of Epac2, form a hydrophobic core that stabilizes the apo conformation [26]. Upon cGMP binding, the structural changes occur at the helical regions in a highly coordinated manner. The first step is the αP-helix adopting a compact conformation when cGMP binds to the PBC (Figs. 5A and S4B). This structural change is driven by the hydrogen bond formed between O1A of the cyclic phosphate and the backbone amide of A485, which is seen a previously established line where the endogenous PfPKG gene has been HA-tagged. Blots were incubated with anti-HA, anti-human PKG, and anti-PfGAPDH as a loading control. Note that the anti-human PKG antibody does not react with HA-tagged PKG (lane 2) because the free carboxyl terminus is crucial for antibody binding. This allows the differentiation between endogenous and HA-tagged PKG. (D) Growth of the transgenic lines in the presence of 1 μM compound 2 over 8 days. The established compound 2-resistant line 3D7/T618Q was included as a positive control. (E) Late trophozoites/early schizonts of the three transgenics were cultured for 12 hours in the presence of 1 μM compound 2 (lower panels) or DMSO (upper panels) and parasite development examined on Giemsa-stained blood smears. (F) Quantification of (E). >1000 cells were counted for each culture and condition and ring stage parasitaemia determined. Data represent the mean of three experiments (error bars = SD). Dark grey bars are DMSO controls, light grey bars 1 μM compound 2 treated samples. For each sample, parasitaemia was adjusted to make the DMSO control 10% to eliminate variability from differences in parasitaemia between experiments. (G) Sequence alignment of apicomplexan PKGs. Conserved residues are shaded in yellow (identical), in green (functionally similar), and in gray (identical in >66%). The capping triad residues are typed in red and marked with arrows. Toxoplasma gondii PKG, TgPKG; Theileria orientalis PKG, ToPKG; Eimeria tenella PKG, EtPKG; Babesia bovis PKG, BbPKG, and Cryptosporidium hominis, ChPKG. in other CNBs (Fig. S4B) [20]. This causes the αP-helix to rotate slightly clock-wise and tilt approximately 20°toward the -barrel (Figs. 5A and S4B). The second step involves L486 and L487 on the αP-helix, which together with F513 on the αB-helix, undergo a cogwheel-like motion causing the αB-helix to move toward the PBC (Fig. 5B). The third step is the N3A motif moving away from the PBC, pushed away by the αB-helix moving toward the PBC (Fig. 5C). These changes occur because the hydrophobic tip of the α3 10 -helix (I418 and F419) no longer  Novel Capping Triad at CNB-D of PfPKG Crucial for Merozoite Egress interacts with L486 at the PBC, but slides along strand β3 and forms new interactions with I457. Finally, the αC-helix moves close to the cGMP pocket and forms the capping triad shielding the bound cGMP (Fig. 5D). One side of the αC helix displays several residues that form a Velcro-like interaction with the rest of the domain and the bound cGMP ( Fig. 2A). An extensive network formed between two interlinking loops including captured waters (the loops between the α3 10 -αA helices and the αB-αC helices) helps form this Velcro-like interaction (S5 Fig.). In particular, a hydrophobic interaction between P520 (at αC) and Y421 (at the α3 10 and αA loop) stabilizes the flexible αB-αC loop with G519 and orients the N-terminal region of the αC-helix for the Velcro-like interaction (Figs. 5D and S5).

Structural basis of cGMP selectivity
Structural comparison of PfCNB-D with the cGMP selective human PKG Iβ CNB-B combined with the sequence comparison with the other PfCNB domains confirms the amino acids required for cGMP selectivity and explains the different cyclic nucleotide affinity values measured for PfCNB domains. Both structures show highly similar interactions with cGMP including charged interactions at the PBC and β5 that specifically recognize the guanine moiety (Sites 1-3) and hydrophobic interactions that shield the bound cGMP, although the nature of the latter interactions differs ( Fig. 2A). Due to the capping triad and V546-L472 interactions at Site 4, the cGMP pocket of PfCNB-D is more shielded compared to the human PKG Iβ CNB-B. Since these are not charged interactions specific for cGMP, the more shielded pocket could result in higher affinity for both cGMP and cAMP. Consistent with the structure, our measurements on PfCNB-D show smaller EC 50 values for both cGMP and cAMP compared to these of human PKG Iβ CNB-B (EC 50 cGMP = 215 nM and EC 50 cAMP = 52 μM) [18]. Sequence alignment shows that PfCNB-A and B have either the same or chemically similar residues that can recognize cGMP at the first three sites whereas PfCNB-C lacks most of these residues (S1 Fig.). These observations are also consistent with our measurements showing clear cGMP binding in PfCNB-A and B, and little to no binding in PfCNB-C. Sequence alignment also shows that PfCNB-B has an arginine analogous to the capping triad forming R484 of PfCNB-D, but lacks the analogous Q532/D533 residues at the corresponding positions in the C-terminus suggesting that the capping triad is unlikely to form in PfCNB-B. Consistent with these observations, mutating the capping triad residues to alanine in PfCNB-D reduces its cGMP affinity to a similar level as in PfCNB-A and B.

Role of the αC-helix and capping triad in activation
While similar conformational changes were observed in the helical subdomains of the human PKG Iβ CNB-B, these changes at the αC-helix and the capping mechanism are very different between the two structures [18,19]. The αC-helix in the human PKG Iβ CNB-B:cGMP complex shows a single turn of helix followed by a short loop and only shields the bound cGMP through the capping interaction, whereas the fully helical αC helix of PfCNB-D is more ordered and shields the entire base of the β barrel through a Velcro-like interaction. Moreover, while a single aromatic residue at the αC-helix provides the capping interaction in human PKG Iβ, the triad formed by R484/Q532/D533 provides the capping interaction in PfPKG. Although these structures only represent a small portion of the R-domain and may suffer from artifacts associated with using truncated domains for crystallization and having different crystal packing environments, the dramatic conformational changes at the αC-helix are observed in both structures suggesting its highly dynamic nature. Since the activity of PKG is allosterically regulated by the structural rearrangement within its R-domain upon cGMP binding and the αC-helix is located between the R-and C-domains [14,20], we hypothesized that the conformational change of the αC-helix is a key step in disrupting the high affinity R-C interaction and releasing the C-domain and that the capping triad is a key structural feature that stabilizes the active conformation.
Our binding and activation results strongly support this hypothesis. Reductions in cGMP binding affinity were seen for all individual mutants, R484A, Q532A, or D533A of PfCNB-D and the full-length R domain, with D533A having the least reduction. The same trend is observed for kinase activation where major reductions were seen in R484A or Q532A as expected, but only a slight reduction in D533A. Mutating all three showed the highest reduction both in cGMP affinity and in activity compared to wild type. These results suggest that the interaction between R484 and Q532 plays a bigger role in cGMP binding and activation than the R484-D533 interaction. Possibly, this is because Q532 is located on the inner side of the αC-helix and is directly involved in the capping interaction with cGMP. In contrast, D533 is located on the outer surface of the αC-helix with its side chain exposed to solvent contributing much less to the capping interaction.
Lastly, we show that mutation of the capping triad in the parasite itself blocks blood stage development. PfPKG plays an essential role in egress of blood stage merozoites from erythrocytes [7], and inhibition of PfPKG prevents the release of proteins from apical organelles required for egress and reinvasion [8]. It has also been suggested that PfPKG is the universal regulator of calcium release across the malaria parasite life cycle, which is achieved by phosphorylation of phospholipid kinases and generation of IP 3 [21]. Disruption of PKG function, therefore, interferes with a number of essential downstream signaling processes. Here we used a chemical-genetic approach to demonstrate that disruption of the capping triad within the highly cGMP selective PfCNB-D domain, abrogates PKG's ability to promote blood-stage egress, reinvasion, and signaling through downstream effectors, further strengthening the conclusion that the capping triad formation is an integral part of the activation mechanism. Thus, we propose that the αC-helix serves as an allosteric switch that triggers activation upon cGMP binding. Additionally, the αC-helix, strategically located between the R and C domains functioning as a molecular switch for activation, is likely to be a common feature of all PKGs, though the molecular details of the capping may differ due to the highly divergent amino acid sequence in this region. Amino acid sequence comparisons with other apicomplexan parasitic PKGs strongly suggest that the capping triad is a shared feature that could be targeted for developing broad-spectrum inhibitors against this family of important parasites (Fig. 3G) affecting humans and animals, that includes Toxoplasma, Cryptosporidium, and Eimeria. [10,27]. We envision such compounds that specifically bind the PfCNB-D pocket and disrupt the formation of the capping triad will be potent inhibitors of these parasites.

Construct design, protein expression and purification
A manually designed construct consisting of PfPKG amino acids 1-853 with codons optimized for expression in E.coli was synthesized at Life Technologies. All PfPKG constructs were ligated into the His-tagged bacterial expression vector, pQTEV [28]. In designing the expression constructs of each CNB, we used the sequence of the cGMP selective CNB-B that we recently reported [18]. The domain boundaries for the individual constructs are: residues 1-173 for CNB-A, 158-294 for CNB-B, 275-428 for CNB-C and 401-542 for CNB-D. All PfPKG proteins were expressed in TP2000 E. coli. The cells were grown at 37˚C to an OD 600nm of 0.6, then induced with 0.5 mM IPTG, and grown for an additional 10 hours at 25˚C. Cells were harvested by centrifugation, then resuspended in 50 mM potassium phosphate (pH 7.5), 500 mM NaCl and 1 mM β-mercaptoethanol, and lysed using a cell disruptor (Constant Systems).
The lysate was cleared using ultracentrifugation and the supernatant was loaded onto an IMAC nickel column (Bio-Rad) on ÄKTA purifier (GE Healthcare). The N-terminal his-tagged PfPKGs were eluted by linear gradient with cell suspension buffer containing 300 mM imidazole. To remove the His-tag, the sample was incubated with His-tagged tobacco etch virus (TEV) protease in a 50:1 molar ratio at 4˚C overnight and loaded onto Bio-Rad IMAC column for TEV separation. The sample without his-tag was obtained by collecting flow-through fractions. The sample was loaded onto a HiLoad 16/60 Superdex 75 gel filtration column (GE healthcare) equilibrated with a buffer containing 25 mM Tris (pH 7.5), 150 mM NaCl, and 1 mM TCEP. For the purification of wild type and mutant full-length PfPKG, we performed an additional chromatography step using a Resource Q anion exchange column (GE Healthcare) before the final size exclusion step using a Hiload 16/60 Superdex 200 column (GE Healthcare).
Crystallization, data collection, phasing, model building, and refinement To obtain crystals of the PfCNB-D:cGMP complex, the protein sample was pre-incubated with 5 mM cGMP and concentrated to 50 mg/mL using a 10 kDa cutoff Amicon Ultra (Millipore). Crystals were obtained using the hanging drop method. 2 μL of protein solution was mixed with 2 μL of reservoir solution containing 0.2 M lithium sulfate, 15% ethanol, 0.1 M citrate at pH 5.5 and 10% 1,5-diaminopentane dihydrochloride. After three days incubation at 4˚C, drops produced rod-shape crystals belonging to a P3 1 21 space group that diffracted to 2.0 Å resolution. Apo crystals were obtained at 22˚C in a solution containing 0.2 M lithium sulfate, 25% (w/v) PEG 3350 and 0.1 M Bis-Tris (pH 5.5). The apo crystal belongs to a P2 1 2 1 2 1 space group and diffracted to 1.89 Å resolution. All protein concentrations were measured using Bradford assay. Crystals of the cGMP complex and the apo form were cryoprotected with 25% ethylene glycol and Paratone-N, respectively, before freezing.
Diffraction experiments were performed at the Advanced Photon Source (Argonne, IL, USA). Diffraction data were processed and scaled using iMosflm [29] with satisfactory statistics ( Table 1). Phasing of the apo structure was accomplished using the molecular replacement program Phaser-MR [30] with the previously determined C-terminal cGMP binding domain of PKG Iα structure (PDB code: 3SHR; residues 205-324) as a search model. Manual building of the model was performed using the program Coot [31], followed by rounds of refinement using phenix.refine [32]. Phasing of the PfCNB-D:cGMP structure was accomplished using the molecular replacement program Phaser-MR [30] using a truncated apo structure of PfCNB-D (residues 404-518) as a search model. Manual building of the model was also performed using Coot [31], followed by rounds of refinement using phenix.refine [32]. The final refinement round included restrained TLS refinement parameters [33].

Fluorescence polarization
The direct fluorescence polarization (FP) assay was performed following the procedure from Moll et al [34]. Measurements were performed in 150 mM NaCl, 20 mM MOPS plus 0.005% (w/v) CHAPS pH 7.0 using the Fusion a-FP microtiter plate reader at room temperature in a 384 well microtiterplate (Perkin Elmer, Optiplate, black). The protein concentration was varied while the concentration of 8-Fluo-cAMP/8-Fluo-cGMP (Biolog Life Science Institute (Bremen, Germany)) was fixed at 1-5 nM. The FP signal was detected for 2 seconds at Ex 485 nm and Em 535 nm with a PMT Voltage of 1.100 V. Data were analyzed with GraphPad Prism 5.03 (GraphPad Software, San Diego, CA) by plotting the polarization signal in mPol against the logarithm of the protein concentration. The K D values were calculated from sigmoidal doseresponse curves.
For competition FP experiments, the protein concentration and the concentration of 8-Fluo-cGMP were fixed to give a polarization signal that was 50% of the maximum value obtained from direct FP measurements. The protein/8-Fluo-cGMP mixture was incubated with varying concentrations of unlabeled cGMP or cAMP (Biolog Life Science Institute, Bremen, Germany). FP signals were detected as indicated above. Data were analyzed with GraphPad Prism 5.03 by plotting the polarization signal in mPol against the logarithm of the cyclic nucleotide concentration. The EC 50 values were calculated from sigmoidal dose-response curves. Plasmodium falciparum culture, transfection, and synchronization P. falciparum lines 3D7/attB [22], 3D7/PfPKGT618Q [6], and 3D7/PfPKG-HA [35] were cultured according to standard procedures [36] in human A+ erythrocytes (National Blood Transfusion Service, UK) and RPMI 1640 medium supplemented with 0.5% Albumax type II (Lifetech) and 5 nM WR99210 (Jacobus Pharmaceuticals, New Jersey).

Microfluidic mobility-shift assay
3D7/attB ring stage parasites were co-transfected with the pPfPKG/attP expression plasmid and the bxb1 integrase plasmid (pINT) to facilitate recombination of the attP plasmid with the attB site integrated into the cg6 pseudogene locus as described [22]. The following drugs were applied 24 hours post transfection for drug selection: G418 (Sigma, 250 μg/ml) for five days to select for the presence of pINT, and blasticidin S HCl (Sigma, 5 μg/ml) to select for the presence of the pPfPKG/attP plasmid. Blasticidin-resistant parasite cultures were established approximately three weeks post transfection. Transfectant cultures were maintained on blasticidin and WR99210 pressure.
Parasite synchronization was achieved by repeated sorbitol treatments. Highly synchronous parasites were obtained by magnet purification of mature schizonts (MACS, Miltenyi Biotech) followed by the addition of fresh erythrocytes and a sorbitol treatment three to five hours later.
Extraction of PKG from P. falciparum schizonts, SDS-PAGE and immuno-blotting Schizont stage parasite cultures were harvested by centrifugation, erythrocytes lysed with 0.15% saponin in PBS and the parasite pellet washed twice in PBS. Parasites were resuspended in ice-cold lysis buffer (10 mM Tris-HCl, 150 mM NaCl, 0.5 mM EDTA, 0.5% NP40, pH 7.5) and incubated on ice for 20 minutes. All buffers were supplemented with complete protease inhibitor cocktail (Roche). The samples were centrifuged at 16000 g for 20 minutes at 4°C and the supernatant assayed by SDS-PAGE/western blot. Proteins were resolved on a NuPAGE Novex 4-12% Bis-Tris gel (Lifetech) and blotted on to nitrocellulose membrane. The membrane was blocked with 5% milk in PBS and probed with the following antibodies: anti-human PKG 1:2500 (rabbit, Enzo Life Sciences); anti-HA 3F10 1:5000 (rat, Roche Diagnostics); anti PfGAPDH 1:15000 (mouse, obtained from Claudia Daubenberger, Swiss Tropical and Public Health Institute). HRP-coupled anti-mouse, rabbit, and rat secondary antibodies (DAKO) were used at 1:6000. Schizont accumulation assay and growth assays P. falciparum cultures were synchronised by repeated sorbitol lysis, then either compound 2 (1μM) or DMSO was added to late trophozoites / early schizonts and cultures harvested 16h later to count newly formed ring stage parasites. Cells were fixed in 4% formaldehyde/0.1% glutaraldehyde in PBS for 30 minutes at room temperature, stained with SYBR green (Lifetech) diluted 1:10000 in PBS for 30 minutes, washed with PBS and analysed on a FACSCalibur cell analyser (Becton Dickinson). FACS data were analysed using FlowJo software. For growth assays, starting parasitaemia was 0.2% ring stages and cultured in presence or absence of compound 2 (1 μM). Samples were taken every 48h and fixed, stained, and analyzed as described above.

Coordinates
Atomic coordinates and structure factors of the PfCNB-D:cGMP complex and apo structures have been deposited in the Protein Data Bank (http://www.pdb.org) under accession numbers 4OFG and 4OFF. The bound cGMP and cGMP pocket residues are shown in stick with electron density. A |Fo|-|Fc| omit map is contoured at σ = 1.0. The cGMP interacting residues and cGMP are labeled and colored by atom type (carbon, white; nitrogen, blue; oxygen, red; phosphorus, orange) except for the carbon atoms. The carbon atoms belonging to the PBC are colored in yellow, the β4-β5 strands in dark gray, αC helix in red, and cGMP in white. Transgenic parasites harbouring an extra gene copy of HA-tagged pfpkg T618Q in the cg6/attB locus express both pkg genes simultaneously. Active wild-type enzyme is produced from the endogenous PfPKG locus while enzyme produced from the cg6/attB locus carries the T618Q substitution conferring resistance to compound 2. Upon addition of compound 2, wild-type enzyme will be inhibited and the parasite depends on the activity of PKG-HA T618Q . This system allows us to assess whether a mutation introduced into the PKG-HA T618Q background interferes with kinase activity by asking whether it can complement endogenous PKG function in presence of compound 2. (EPS) S1