Optimization of an Imidazopyridazine Series of Inhibitors of Plasmodium falciparum Calcium-Dependent Protein Kinase 1 (PfCDPK1)

A structure-guided design approach using a homology model of Plasmodium falciparum calcium-dependent protein kinase 1 (PfCDPK1) was used to improve the potency of a series of imidazopyridazine inhibitors as potential antimalarial agents. This resulted in high affinity compounds with PfCDPK1 enzyme IC50 values less than 10 nM and in vitroP. falciparum antiparasite EC50 values down to 12 nM, although these compounds did not have suitable ADME properties to show in vivo efficacy in a mouse model. Structural modifications designed to address the ADME issues, in particular permeability, were initially accompanied by losses in antiparasite potency, but further optimization allowed a good balance in the compound profile to be achieved. Upon testing in vivo in a murine model of efficacy against malaria, high levels of compound exposure relative to their in vitro activities were achieved, and the modest efficacy that resulted raises questions about the level of effect that is achievable through the targeting of PfCDPK1.


■ INTRODUCTION
Malaria is one of the most prevalent infectious diseases of the developing world. In excess of 3 billion people are at risk, and it currently leads to the deaths of around 655,000 people each year, with the majority of these occurring in sub-Saharan Africa among children under five years of age. 1 Resistance to existing antimalarial drugs is widespread, 2 and therefore, new therapeutic approaches are urgently needed. Calcium-dependent protein kinases (CDPKs) are directly regulated by Ca 2+ and are found in plants and organisms in the alveolate lineage, 3 but they are absent in humans. They are present in Apicomplexan parasites including Plasmodium falciparum, the causative agent of the most severe form of malaria. CDPKs in Plasmodium are present as a multigene family containing at least five members, 4 and different CDPKs are proposed to be functional at different stages of the parasite life cycle. P. falciparum calcium-dependent protein kinase 1 (Pf CDPK1), first identified by Zhao et al., 5 is expressed in the asexual blood stages of the parasite responsible for disease pathology. It has been shown to be encoded by an important gene, 6,7 and it is implicated in parasite motility and host cell invasion, where it is able to phosphorylate components of the molecular motor that drives parasite invasion of red blood cells. 8,9 The prevention of this invasion process could break the parasite lifecycle, causing the parasites to die and the infection to be cleared. Pf CDPK1 therefore represents a novel target for the potential treatment of malaria and offers promise for achieving selectivity over the kinases of the human host. More recently, its role in translational regulation of motor complex transcripts in gametocytes 10 and in schizont development 11 has also been reported. There has been interest in CDPKs as drug targets, 12 although relatively few inhibitors have been reported in the literature: Kato et al. 7 and Lemercier et al. 13 have reported inhibitors of Pf CDPK1, and inhibitors of the CDPK1 enzymes from the related Apicomplexan protozoa Toxoplasma gondii and Cryptosporidium parvum have also been described. 14−16 A high throughput screen of our compound collection against the isolated recombinant Pf CDPK1 enzyme identified a hit series containing an imidazopyridazine core as the primary series of interest, and the initial development of the structure− activity relationship (SAR) in this series has been described previously. 17 Compounds with potent enzyme inhibitory activity had been generated, which also showed good kinase selectivity against a human kinase panel and promising in vitro ADME profiles. In particular, compound 1 ( Figure 1) represented an early lead, with low nanomolar inhibitory potency against Pf CDPK1, sub-500 nM antiparasite activity, and modest in vivo efficacy in a P. berghei mouse model of malaria.
In order to advance this series, improvements were sought in the in vitro antiparasite activity and pharmacokinetic profile of the series while maintaining a good selectivity profile against human kinases to generate compounds with the potential to show improved in vivo efficacy.

■ RESULTS AND DISCUSSION
A structure-guided design approach using a homology model of Pf CDPK1 (based on TgCDPK1, PDB ref: 3I7C) 18 was used in attempting to gain increased binding affinity against the target and correspondingly increase the cellular potency of the inhibitors. The homology model had proved effective in explaining the SAR up to this point, and it was therefore used as a key component in considering how additional potency could be gained. In particular, it suggested that the binding pocket occupied by the isopentyl chain of compound 1 was not optimally filled and that there was potential to gain additional beneficial interactions with the enzyme in this region. Virtual libraries were enumerated with a diverse range of groups at this position and examined through docking using Glide SP. 19 For enumeration purposes, the basic amine group was set to be either N-methylpiperidine or 1,4-diaminocyclohexane, which had been previously demonstrated to be optimal for potency, 17 and the heteroaryl linker ring was set as either pyridine or pyrimidine. Analysis of the docking results suggested that replacement of the isopentyl group with an aromatic ring containing a suitably positioned hydrogen-bond acceptor (for example, a 2-pyridyl group, as in compound 2; Figure 2) could give increased binding affinity.
In comparison with the isopentyl group in this pocket, the 2pyridyl nitrogen atom could potentially form an additional hydrogen-bond interaction with the backbone N-H of Asp-212 (depicted in Figure 3B), and it also appeared to be an excellent spatial and electrostatic fit into the rest of this pocket. Compound 2 was predicted to be able to retain the other key interactions made by compound 1: a hydrogen bond between the imidazopyridazine core and the backbone N-H of Tyr-148 at the hinge region, a hydrogen bond with the sidechain carboxylic acid of Asp-212, and an additional hydrogen bond between the basic amine group and Glu-152 as it points out toward the solvent.
The top scoring compounds identified from the docking studies were then synthesized according to the route in Scheme 1. Intermediate 3 was functionalized through nucleophilic substitution on the chloro-at the 6-position to install the BOCprotected amine side chains in intermediates 4 and 17. These were elaborated through Suzuki coupling with the appropriate boronate reagents to give the 5-aminopyrimidine products 5 and 18, the 5-thiomethylpyrimidine 13 or 5-aminopyridine 20. The final compounds were prepared by either a Buchwald coupling with an aryl halide or through oxidation of the thiomethyl pyrimidine followed by nucleophilic displacement with the appropriate amine.
The SAR of the synthesized analogues is shown in Table 1. As predicted by the docking studies, variants containing a 2pyridyl or phenyl group at this position both showed increased binding affinity to the enzyme versus compound 1 (Table 1, examples 2 and 6). As the potency of the most potent compounds was now below the limit of detection of our primary Kinase Glo enzyme assay, a thermal denaturation assay was used to quantify the differences and define a rank order in binding affinity between the most potent compounds. 20 This revealed that the 2-pyridyl compound 2 resulted in a larger shift in the thermal denaturation temperature of the protein (ΔT m ) than its phenyl counterpart 6, and both showed a higher thermal shift than compound 1 in this assay, which displayed a ΔT m of 15.7 K. Gratifyingly, this difference in enzyme affinity was reflected in the potency of the compounds against the P. falciparum parasite, with compound 2 showing an EC 50 of 80 nM compared with 180 nM for compound 6. Alternative heteroaryl groups were then explored: 2-pyrazine 7 showed good potency, albeit weaker than those of 2 and 6, but 3-pyridyl 8 and 2-pyrimidyl 9 lost potency against both the enzyme and parasite. The addition of substituents to the pyridyl ring was investigated: 3-fluoropyridyl gave a boost in potency against both the enzyme and the parasite, with compound 10 displaying a high thermal shift of 28.0 K and excellent EC 50 of  12 nM against the parasite. The introduction of 5-position substituents to the pyridine ring such as trifluoromethyl (11) and methyl (12) led to excellent enzyme affinity and increased thermal shift values relative to 10, although their antiparasite potency decreased. When a CH 2 spacer group was introduced, the 3-pyridyl variant 14 was relatively weak against the enzyme, whereas the 2-pyridyl variant 15 and the 3-pyrazole 16 showed good enzyme inhibitory potency. This was again consistent with the predictions of the homology model, which suggested that 15 could form an H-bond with Asp-212, whereas 14 could not. However, all of these variants were weak against the parasite. Switching to the N-methylpiperidine basic side chain (19) gave good potency against the enzyme and the parasite, while changing the pyrimidine linker ring to a pyridyl linker ring (21) retained high enzyme potency but led to a loss in potency against the parasite. Scheme 1. Synthetic Route to Heteroaryl and Aryl Variants a a Reagents and conditions: (a) trans-cyclohexane-1,4-diamine, NMP, microwave, 180°C, then di-tert-butyl dicarbonate, DMAP, Et 3 N, THF, reflux; (b) 2-aminopyrimidine-5-boronic acid pinacol ester, Pd(dppf)Cl 2 , Cs 2 CO 3 , dioxane/water, 90°C; (c) aryl/heteroaryl halide, Pd(OAc) 2 , CyPF-t Bu or Xantphos, NaO t Bu or Cs 2 CO 3 , DME or dioxane, 80°C; (d) 4 M HCl/dioxane, MeOH; (e) 2-(thiomethyl)pyrimidine-5-boronic acid, Pd(dppf)Cl 2 , Cs 2 CO 3 , dioxane/water, 90°C; (f) m-chloroperoxybenzoic acid, CH 2 Cl 2 ; (g) 3-(aminomethyl)pyridine or 2-(aminomethyl)pyridine or (1-methyl-1H-pyrazol-3-yl)methylamine, dioxane, reflux; (h) tert-butyl 4-aminopiperidine-1-carboxylate, DIPEA, NMP, 130°C; (i) 2-chloro-3fluoropyridine, Pd(OAc) 2 , Xantphos, Cs 2 CO 3 , dioxane, microwave, 130°C or thermal, 90°C; (j) formaldehyde, AcOH, sodium triacetoxyborohydride, THF; (k) 2-aminopyridine-5-boronic acid pinacol ester, Pd(dppf)Cl 2 , Cs 2 CO 3 , dioxane/water, 90°C. Leading compounds were profiled in in vitro ADME assays, and selected data are shown in Table 2. In general, the compounds had low measured log D values and displayed good stability in both mouse and human microsomes but poor PAMPA permeability. Kinase selectivity screening against a human kinase panel revealed that they showed good selectivity, and the selectivity profile of compound 10 is shown in Figure 4, in comparison with that of compound 1. Compound 10 also showed IC 50 > 25 μM against CYP-P450 isoforms 1A, 2C9, 2C19, 2D6, and 3A4. However, when 10 was tested for in vivo efficacy in the 4-day Peters test 21 (P. berghei murine model of malaria) with a 50 mg/kg once daily oral dosing regimen, it showed no significant reduction in parasitemia levels (4% reduction). This was thought to be a consequence of low plasma exposure, consistent with poor absorption in accordance with its low permeability.
Although the introduction of the 2-pyridyl group gave improved enzyme and antiparasite potency, poor permeability was seemingly limiting the bioavailability of the compounds when dosed in vivo. Efforts were therefore directed at increasing the permeability to allow sufficient exposure while retaining high potency. In the first case, the pyrimidine linker ring and distal 3-fluoro-2-aminopyridyl group present in compound 10 were retained, and the basic amine portion was modified. The two amino substituents giving the highest potency up to this point are both highly basic (calculated pK a values for the conjugate acids of 10 and 19 are >9 in both cases 22 ), and therefore, attenuating this basicity represents one approach to potentially gaining improved permeability. Compound 10 also contains four hydrogen bond donors (HBDs) which may be leading to poor permeability. Variations at this portion of the molecule were explored, and selected variants are shown in Table 3. The requirement for a basic amine side chain in order to achieve cell-based activity against the parasite was demonstrated by example 22, in which the basic side chain was replaced by a methyl group. Although this compound is still relatively potent against the enzyme with an IC 50 of 44 nM, there is complete loss of potency against the parasite; this effect is consistent with the observations of Lemercier et al. 13 The high permeability of 22 also confirms that the basic center is making a significant contribution to the poor permeability of the compounds. Piperazine 23 showed weaker enzyme, and antiparasite potency and no improvement in permeability and although the N-methylpiperazine 24 (with just one HBD and a calculated pK a of 7.1) showed improved permeability, it displayed weak antiparasite activity. The pyrrolidines 25 and 26 showed good enzyme affinity but low permeability. The more flexible variants 27−29 showed lower enzyme affinity, and although the weakly basic morpholine 27 showed improved permeability, it was inactive against the parasite, whereas piperazine 28 and pyrrolidine 29 both showed low permeability. As the diaminocyclohexane basic group seemed optimal for antiparasite activity, cyclization of the free −NH 2 into a ring was explored as a way to potentially improve permeability, through the attenuation of pK a and/or the reduction in the number of HBDs. The cyclized variants 30 and 31 showed good potency against both the enzyme and parasite; however, a modest improvement in permeability was only achieved with morpholine variant 30. Compounds with oxygen-linked basic amine groups such as piperidine 32 and pyrrolidine 33 were potent against the enzyme and showed improved permeability, but they were inactive against the parasite. Carbon-linked piperidine variant 34 showed good enzyme inhibitory potency but a weak effect against the parasite, coupled with low permeability.
In summary, although lowering the pK a and HBD count resulted in improved permeability in some cases, this tended to be accompanied by a significant reduction of potency against the parasite. The best combination was achieved with compound 30, although it still showed only a weak effect (22% reduction in parasitemia) when tested for in vivo efficacy in the P. berghei model under the same dosing regimen as employed previously. Despite complying with property criteria that may normally be expected to be sufficient to allow permeability and oral bioavailability, structure−property relationships suggested that there were stricter requirements for this series and that the desired balance in profile could not be obtained from modifying the basic group alone.
It had been observed that the pyridine linker ring had given higher permeability than the pyrimidine (comparison of examples 19 and 21), so further variations at this linker ring were then explored. Employing a phenyl ring in this position (35; see Table 4) gave a large increase in permeability versus its pyrimidine analogue (19), although unfortunately this was accompanied by a significant loss in antiparasite activity, with an EC 50 of approximately 400 nM. In order to maintain the phenyl ring but mimic the dipole moments of the pyridyl and pyrimidyl rings, fluorophenyl and difluorophenyl variants 36 and 37 were investigated, although this approach failed to restore potent antiparasite activity. Cyclization of the −NH 2 of the diaminocyclohexane side chain into a pyrrolidine ring with pyridyl or phenyl linker rings gave examples 38 and 39; these compounds showed good permeability and marginally improved antiparasite activity relative to the N-methylpiperidines. Reverting to the diaminocyclohexane basic group with a free −NH 2 (40) gave improved antiparasite activity but reduced permeability, and variants with a pyrazole linker ring were also investigated, and these have been described previously. 23 In summary, although modifications of this linker ring led to significant and steep improvements in PAMPA permeability, the variations that gave improved permeability versus compound 10 while maintaining good enzyme inhibitory potency were also accompanied by a significant loss in antiparasite activity. Although the desired balance in compound profile had not yet been achieved through modification of this group, the higher permeability of these compounds warranted pharmacokinetic studies to confirm that they could achieve good in vivo exposure.
Pharmacokinetics. Pharmacokinetic profiling in rats revealed that there was a good correlation between in vitro ADME and in vivo PK, with compound 35 showing the best PK profile of those tested (data shown in Table 5).
Compounds 35, 38, and 39 were then advanced to testing in the P. berghei in vivo mouse model, and despite their lower antiparasite potency compared to that of earlier examples such as compound 10, they showed superior efficacy. The observed efficacy was still only modest, with best results of 44% and 46% reduction in parasitemia achieved with 35 and 38, respectively, at a 50 mg/kg oral dose (Table 6).
In order to evaluate the exposure in the efficacy model, blood samples were taken for compound 35, which revealed that it had achieved a good exposure with total plasma levels of 1660 ng/mL at 1 h (10-fold over the antiparasite EC 50 ) and 1460 ng/mL at 4 h (8.5-fold over the antiparasite EC 50 ).
Addressing Species Differences between P. falciparum and P. berghei. As the in vivo efficacy model is performed with the rodent parasite P. berghei rather than P. falciparum, there remained a possibility that the low in vivo efficacy achieved might be due to species differences. Significant efficacy differences across the parasite species have been observed in the development of other antimalarial compounds, and therefore, it was desirable to address this possibility. First, an in silico assessment of Pf CDPK1 and PbCDPK1 revealed that across their entire protein sequences PbCDPK1 and Pf CDPK1 have high homology, with 88% sequence identity and 93% similarity.
Furthermore, based on the residues within 10 Å of any atom of ATP in the PbCDPK1 crystal structure (PDB ID: 3Q5I), there is 100% identity. Docking of the inhibitors into the ATPbinding site of PbCDPK1 predicted that they would bind with an affinity similar to that for Pf CDPK1. This prediction was confirmed through the expression of the recombinant PbCDPK1 enzyme and measurement of the IC 50 values, which showed that key compounds bound with similar affinity between the two enzymes (representative compounds 1 and 35 showed PbCDPK1 IC 50 values of 22 and 15 nM, respectively). Finally, in order to probe in vivo differences, selected compounds were tested in a P. falciparum murine model, in which a severe combined immunodeficient (SCID) mouse can be injected with human erythrocytes infected with a suitable strain of P. falciparum parasites. 24 Compounds 1 and 35 were tested in this efficacy model with the same dosing regimen as in the P. berghei model (50 mg/kg, oral, once daily), but disappointingly, no efficacy was observed despite good compound exposure.
In parallel with in vivo testing, efforts had also been focused on achieving an improved compound profile with respect to the combination of antiparasite potency and pharmacokinetics. Although modification of the linker ring had led to improved pharmacokinetics, this had been achieved at the expense of antiparasite potency, so attention then turned to the distal pyridyl ring as a point of modification to potentially restore potency. The introduction of substituents around this ring had previously produced compounds with the highest enzyme affinities as determined by thermal denaturation experiments ( Table 1, compounds 11 and 12), so this approach offered the potential to yield increased antiparasite potency. Molecular   modeling suggested that there was sufficient space in the binding pocket for small substituents to be accommodated, so a number of variants were synthesized, and the resulting SAR is shown in Table 7. With N-methylpiperidine as the basic group and a phenyl linker ring in place, the addition of 5-chloro (41), 6-methyl (42), or 5-fluoro (43) substituents was well tolerated, with sub-100 nM antiparasite EC 50 values for 41 and 42 and acceptable in vitro ADME. The 5-and 6-trifluoromethyl substituted pyridines 44 and 45 also showed excellent antiparasite EC 50 values of 50 and 30 nM, respectively, with good in vitro ADME characteristics. Compounds 46−48, with the diaminocyclohexane basic group, also showed a good balance between antiparasite activity and ADME properties; in particular, compound 48 demonstrated the best profile with an antiparasite EC 50 of 80 nM combined with high stability in mouse and human microsomes and acceptable permeability.
Compounds 41 and 48 were advanced to the P. berghei in vivo mouse model; however, they did not show significantly improved efficacy compared to that of previous examples (Table 8), and a maximum of 51% reduction in parasitemia was achieved with compound 41. Blood samples revealed that compound 41 had achieved a good exposure (total plasma concentration 45-fold over in vitro antiparasite EC 50 at 4 h and 35-fold at 24 h), whereas 48 showed lower exposure, consistent with its lower observed in vivo efficacy.

■ CONCLUSIONS
An imidazopyridazine series of Pf CDPK1 inhibitors was optimized employing a structure-guided design approach with  a homology model of Pf CDPK1. Initial 2-pyridyl variants showed excellent potency in assays against the Pf CDPK1 enzyme and P. falciparum parasite but did not possess sufficient permeability to be effective in vivo. Modifications of the basic side chain and aryl linker rings allowed permeability to be improved, but this led to a reduction in antiparasite potency.
Finally, further optimization of the linker ring and distal pyridyl ring led to molecules possessing well-balanced profiles with respect to potency and in vitro ADME and suitable for in vivo dosing, with high resulting compound exposures following oral administration. However, despite the improvements made to the compound profiles compared with those of the early lead compound 1, the in vivo reduction in parasitemia in a murine P. berghei model of malaria remained modest, with a maximum reduction of 51% with an oral dose of 50 mg/kg for compound 41. The reason for this limited effect remains unclear: although the Pf CDPK1 enzyme has been shown to be encoded by an important gene in the malaria parasite and high levels of in vivo compound exposure were achieved relative to in vitro antiparasite EC 50 , it may be that a higher multiple of this EC 50 is required for a longer period of time to produce higher efficacy or that low levels of residual CDPK1 enzyme activity may be sufficient for parasite survival. There may be a level of redundancy allowing a CDPK1-based inhibitory effect to be circumvented, and indeed, a recent report has suggested that CDPK1 may not in fact be required for host cell invasion in the erythrocytic stage in Plasmodium berghei but rather that it plays an essential role in the mosquito sexual stages. 25 Differences in parasite biology may also explain the limited effect of the inhibitors in vivo: for example, P. berghei has only a 24-h multiplication cycle and develops in reticulocytes in the mouse compared with a 48-h cycle in mature red cells for P. falciparum, and differences in synchrony of parasite development may also have contributed to the reduced efficacy of the compounds. Further studies on these compounds including understanding their mechanism of action, the identification of additional CDPK1 substrates, and the development of better read-outs of activity are ongoing, and these will be reported in future publications.
■ EXPERIMENTAL SECTION Chemistry. All commercial reagents and solvents were used without further purification. Silica gel chromatography was carried out using a Biotage SP4 or Isolera MPLC system with prepacked silica gel cartridges. Preparative HPLC was carried out using an apparatus made by Agilent. The apparatus is constructed such that the chromatography (column: either a 19 × 100 mm (5 μm) C-18 Waters Xbridge or a 19 × 100 mm (5 μm) C-6Ph Waters Xbridge column, both at a flow rate of 40 mL/min) is monitored by a multiwavelength UV detector (G1365B manufactured by Agilent) and an MM-ES + APCI mass spectrometer (G-1956A, manufactured by Agilent) connected in series, and if the appropriate criteria are met, the sample is collected by an automated fraction collector (G1364B manufactured by Agilent). Collection was triggered by a combination of UV or mass spectrometry or based on time. Typical conditions for the separation process are as follows: the gradient was run over a 7 min period (gradient at the start, 10% methanol and 90% water; and gradient at the finish, 100% methanol and 0% water; as buffer, 0.1% formic acid, 0.1% ammonium hydroxide, or 0.1% trifluoroacetic acid was added to the water). Purity of all final derivatives for biological testing was confirmed to be >95% as determined using the following conditions: an Agilent HPLC instrument with a C-18 Xbridge column (3.5 μm, 4.6 × 30 mm, gradient at the start, 10% acetonitrile and 90% water; gradient at the finish, 100% acetonitrile and 0% water; as buffer, either 0.1% ammonium hydroxide or 0.1% trifluoroacetic acid was added to the water). A flow rate of 3 mL/min was used with UV detection at 254 and 210 nm. The structure of the intermediates and final products was confirmed by 1 H NMR spectroscopy and mass spectrometry. 1 H nuclear magnetic resonance (NMR) spectroscopy was carried out using a JEOL ECX400 spectrometer in the stated solvent at around room temperature unless otherwise stated. Characteristic chemical shifts (δ) are given in parts-per-million using conventional abbreviations for the designation of major peaks: e.g., s, singlet; d, doublet; t, triplet; q, quartet; dd, doublet of doublets; and br, broad. Analytical mass spectra were recorded using a MM-ES + APCI or ES mass spectrometer (G-1956A or G-6120B, manufactured by Agilent).

N-(3-Fluoropyridin-2-yl)-5-[6-(4-methylpiperazin-1-yl)imidazo-[1,2-b]pyridazin-3-yl]pyrimidin-2-amine
Thermal Denaturation Assay. Purified recombinant Pf CDPK1 was diluted to 1 μM in 10 mM HEPES buffer at pH 7.5 containing 150 mM NaCl, 1 mM Ca 2+ , and 1/1000 SYPRO Orange dye (Invitrogen). Test compounds were prediluted to 400 μM in 40% v/v DMSO in water, and 1 μL of diluted compound was added to 39 μL of enzyme/ dye mix in white 96-well quantitative PCR plates (Thermo Scientific) to give a final compound concentration of 10 μM and preincubated for 30 min at rt. Reference melting temperatures were obtained in parallel by the addition of 1 μL of 40% v/v DMSO to 39 μL of diluted Pf CDPK1. The plates were sealed with transparent adhesive covers (Biorad) and subjected to a temperature gradient from 25 to 95 K at a rate of 1 K/min using a quantitative PCR machine (MX3005P, Stratagene). Fluorescence data were acquired at 1 min intervals using the FAM/ROX filter set (Ex 492 nm/Em 610 nm) and the raw data exported to Excel (Microsoft) for analysis. Data were processed to identify the fluorescence maxima and minima, and the midpoints of melting curves were determined by fitting to the Boltzmann equation (XLfit add-in, IDBS software). Results were expressed as ΔT m values relative to DMSO controls where ΔT m = T m (inhibitor) − T m (DMSO controls). P. falciparum in Vitro Parasite Assay. P. falciparum EC 50 values were measured using an in vitro model of malaria parasite growth which measures merozoite invasion of red blood cells. Test cultures were set up at 0.5% parasitemia and 2% hematocrit from a synchronized stock culture of 3D7 P. falciparum. Compounds were diluted into 2% DMSO and added to parasites 24 h postinvasion using a 95 μL parasite culture in a 96-well plate and incubated under static conditions. Compounds were tested at least in duplicate. Cells were recovered 48 h later and processed for FACS analysis: 50 μL of parasite culture was transferred into a FACS tube and mixed with 500 μL of 500 μg/mL hydroethidine in PBS to stain parasite DNA. The parasites were incubated for 20 min at 37°C, then diluted with 1 mL of PBS, and stored on ice prior to FACS analysis. The data were acquired using CellQuest Pro software on a FACSCalibur (Becton Dickinson). Growth inhibition was calculated using the following formula: % growth inhibition = (1 − [parasitemia of culture/ parasitemia of control culture]) × 100.
PbCDPK1 Enzyme Assay. To establish activity of the compounds against recombinant P. berghei CDPK1 enzyme, ATPase activity was measured using a biosensor sensitive to ADP (rhodamine-labeled ParM, gift of M. Webb, NIMR). The progress of the reactions was monitored by an increase in fluorescence corresponding to the accumulation of ADP using a Pherastar plate reader (BMG Labtech).
Kinase Selectivity. Kinase selectivity profiling was carried out at the National Centre for Protein Kinase Profiling in the MRC Protein Phosphorylation Unit at the University of Dundee. P. berghei Murine in Vivo Efficacy Protocol. Plasmodium berghei ANKA strain expressing GFP 26 was used. Mice were NMRI females (20−22 g). Compounds were solubilized or suspended in a solution consisting of 70% Tween-80 and 30% ethanol, followed by a 10-fold dilution in H 2 O. Chloroquine (Sigma C6628) was used as a control drug. Test procedure: Day 0, from a donor mouse with approximately 30% parasitemia, heparinized blood (containing 50 μL of 200 u/mL Heparin) is taken and diluted in physiological saline to 10 8 parasitized erythrocytes per mL. Of this suspension, 0.2 mL was injected intravenously (i.v.) into experimental groups of 3 mice and a control group of 3 mice. Four hours postinfection, the experimental groups were treated with a single dose of compound by the oral (p.o.) route. Days 1−3: 24, 48, and 72 h postinfection, the experimental groups were treated with a single daily dose of compound p.o. at 50 mg/kg. Day 4: 24 h after the last drug treatment, 1 μL of tail blood was taken and suspended in 1 mL of PBS buffer. Parasitemia was determined with a FACScan (Becton Dickinson) by counting 100′000 red blood cells. The difference of the mean infection rate of the control group (= 100%) to the test group was calculated and expressed as percent reduction. The results are expressed as the reduction of parasitemia on day 4 in % as compared to the untreated control group. As an example, activity determination with a mean of, e.g., 2% parasitemia in treated mice and a mean of, e.g., 40% parasitemia in the control animals is calculated as follows: (40% − 2%)/40% *100 = 95% reduction in parasitemia.
P. falciparum Murine in Vivo Efficacy Protocol. Described in ref 24. In Silico Studies. BLAST was used with the Pf CDPK1 sequence (UniProt: P62344) to search the PDB to identify suitable templates for homology modeling. Sequence alignments were generated and structures inspected using the Schrodinger molecular modeling suite, 2010 version. 19 The structure of TgCDPK1 (PDB: 3I7C) was selected as the most suitable template for a variety of reasons including good crystallographic quality and high sequence identity, particularly around the ATP site; of the 49 residues within 6 Å of the crystal ligand, 69% are identical, and 92% are homologous. Homology modeling was carried out using Prime, 19 and further refinements were not required due to the excellent correspondence between the target and template. The resultant structure was prepared for docking studies involving solvent removal, valence and charge assignment, addition of hydrogen atoms, orientation of ambiguous groups (e.g., amide groups of Asn and Gln), and restrained minimization to relieve residual strain. A docking grid was generated with size and location based on the 3I7C crystal ligand, including optional constraints to prominent H-bonding groups including Tyr148 at the hinge, Glu152 at the entrance to the pocket, and Asp212 of the DFG-loop. Virtual libraries of compounds were enumerated and prepared for docking using LigPrep 19 then docked flexibly using GlideSP 19 with a H-bonding constraint to Tyr148 N-H, outputting up to three diverse poses per ligand. High scoring candidate molecules were prioritized for synthesis following manual inspection and consideration of additional factors such as favorable/unfavorable contacts and ligand strain.