Inhibitors of Plasmodium falciparum methionine aminopeptidase 1b possess antimalarial activity

Abstract

With >1 million deaths annually, mostly among children in sub-Saharan Africa, malaria poses one of the most critical challenges in medicine today. Although introduction of the artemisinin class of antimalarial drugs has offered a temporary solution to the problem of drug resistance, new antimalarial drugs are needed to ensure effective control of the disease in the future. Herein, we have investigated members of the methionine aminopeptidase family as potential antimalarial targets. The Plasmodium falciparum methionine aminopeptidase 1b (PfMetAP1b), one of four MetAP proteins encoded in the P. falciparum genome, was cloned, overexpressed, purified, and used to screen a 175,000-compound library for inhibitors. A family of structurally related inhibitors containing a 2-(2-pyridinyl)-pyrimidine core was identified. Structure/activity studies led to the identification of a potent PfMetAP1b inhibitor, XC11, with an IC₅₀ of 112 nM. XC11 was highly selective for PfMetAP1b and did not exhibit significant cytotoxicity against primary human fibroblasts. Most importantly, XC11 inhibited the proliferation of P. falciparum strains 3D7 [chloroquine (CQ)-sensitive] and Dd2 (multidrug-resistant) in vitro and is active in mouse malaria models for both CQ-sensitive and CQ-resistant strains. These results suggest that PfMetAP1b is a promising target and XC11 is an important lead compound for the development of novel antimalarial drugs.

Methionine aminopeptidases (MetAPs) are metalloproteases that catalyze the removal of the N-terminal initiator methionine during protein synthesis (1). MetAPs are evolutionarily highly conserved from prokaryotes to eukaryotes (2). Although there is typically a single gene encoding this enzyme in prokaryotes, at least two distinct genes are known for this enzyme family in eukaryotes, which differ by an insertion of a polypeptide within the catalytic domain in the type 2 (MetAP2) enzyme in comparison with MetAP1. Genetic and biochemical evidence exists in support of the notion that this family of proteins is essential, because knockout of the genes encoding MetAPs in both bacteria and yeast is lethal (3, 4). In addition, inhibition of a single MetAP protein by chemical inhibitors was sufficient to cause cell cycle arrest of certain mammalian cells. For example, the fumagillin family of natural products inhibits endothelial cell proliferation and, hence, angiogenesis by specifically blocking the enzymatic activity of MetAP2 (5, 6). Indeed, MetAP inhibitors are now of great interest as potential anticancer, antiinfective and antiarthritis agents (7–9). More recently, fumagillin and its congeners have been reported to inhibit the growth of P. falciparum in culture (10), likely through inhibition of the malaria MetAP2 enzyme. Together, these observations raised the possibility that inhibition of other MetAP isoforms may be sufficient to block malaria growth.

In this study, we cloned all four isoforms of P. falciparum MetAP cDNA and obtained purified enzymes with enzymatic activity for PfMetAP1a, b, and c, but not PfMetAP2. Using PfMetAP1b as a target, a high-throughput screen of a large chemical library led to a previously undescribed structural class of inhibitors for the enzyme. Structure/activity studies identified a potent inhibitor, XC11, that was highly selective for PfMetAP1b among the four malaria MetAP enzymes. XC11 and some other analogs blocked P. falciparum growth in cell culture. Importantly, XC11 also inhibited both CQ-sensitive and -resistant mouse malaria strains, dramatically prolonging the survival of malaria-infected animals. These results suggest that selective targeting of PfMetAP1b is a promising strategy for the development of novel antimalarial drugs.

Results

Identification of PfMetAP1b as an Active Methionine Aminopeptidase Encoded in the P. falciparum Genome.

We searched for PfMetAP genes that were homologous to the catalytic domains of human and yeast MetAP genes in the P. falciparum 3D7 genome database (http://plasmodb.org). Among the four putative PfMetAP genes, one was identified as PfMetAP2 (Gene ID: PF14_0327) based on the presence of the unique 64-aa insertion toward the C terminus of the catalytic domain. The remaining three showed high homology to MetAP1 from both human and yeast (Fig. 1) and were tentatively named PfMetAP1a, 1b, and 1c (Gene ID: PFE1360c, PF10_0150, and MAL8P1.140, respectively). Similar to human and yeast MetAP1, all three putative PfMetAP1 proteins contained five highly conserved residues, two Asp, one His, and two Glu, that coordinate two metal ions to form the binuclear active sites of all MetAP enzymes known to date (Fig. 1). Of the three putative PfMetAP1 proteins, PfMetAP1b was most closely related to the human and yeast MetAP1 based on the zinc-finger motif present in its N-terminal extension, suggesting that PfMetAP1b may play an important role in malaria growth and survival.

Fig. 1.

Protein sequence multialignment (ClustalW; www.ebi.ac.uk) for PfMetAP1a, PfMetAP1b, PfMetAP1c, Human MetAP1 (HuMetAP1), and Yeast MetAP1 (ScMetAP1). Their C-terminal catalytic domains were highly conserved, including the 5 metal-chelating residues (2×Asp, 2×Glu, 1×His, highlighted in pink) coordinating two adjacent divalent metal ions. PfMetAP1a has no N-terminal extension. PfMetAP1b has zinc finger motif (highlighted in green) followed by a linker to C-terminal catalytic domain. PfMetAP1c has a signal peptide (highlighted in blue) followed by a transit peptide domain (highlighted in red) targeting the apicoplast (11) (as predicted by PlasmoAP, an online software from http://plasmodb.org) at the N-terminal. The C-terminal catalytic domain has an insertion (highlighted in brown) of ≈210 aa inside, which is common in malaria protein (12).

The full-length PfMetAP1b cDNA, amplified from a P. falciparum cDNA library, was subcloned into the pGEX-6P-2 vector, overexpressed in Escherichia coli, and purified to near homogeneity as judged by SDS/PAGE (Fig. 2A). Recombinant PfMetAP1b protein was repurified after proteolytic cleavage and removal by glutathione Sepharose of the GST fusion protein (Fig. 2A, lane 2). The purified recombinant PfMetAP1b protein was fully active as a methionine aminopeptidase toward a methionine-containing oligopeptide substrate in vitro with a K_M of 327.3 μM and k_cat of 13.9 min⁻¹, comparable with recombinant MetAP1 enzymes from other organisms (13, 14). The availability of large quantities of active recombinant PfMetAP1b protein and the convenient spectrophotometric enzymatic assay (13) enabled a high-throughput screening of library of compounds for PfMetAP1b inhibitors.

Fig. 2.

Isolation of recombinant PfMetAP1b, structure of XC11, and assessment of its selectivity for PfMetAP1b among four putative PfMetAPs. (A) Purification of recombinant PfMetAP1b. Lane 1: molecular mass (kilodaltons, Benchmarker ladder; Invitrogen); lane 2: GST tagged PfMetAP1b; lane 3: PfMetAP1b. (B) Chemical structure of XC11, a PfMetAP1b hit. (C) Effect of XC11 on the activity of PfMetAP1a (▵), PfMetAP1b (□), and PfMetAP1c (Δ1–143) (▾). (D) Effect of XC11 (▵) on the interaction between Dexamethasone-fumagillin dimer and PfMetAP2-Gal4AD fusion protein in a mammalian three-hybrid system compared with TNP-470 (■).

Identification of a Previously Uncharacterized Class of Selective PfMetAP1b Inhibitors Containing the Pyridinyl-Pyrimidine Core by High-Throughput Screening.

A total of 175,000 compounds were screened at 5 μM final concentrations in 384-well plates. False positive hits due to inhibition of the coupling enzyme, prolyl aminopeptidase, were eliminated through a secondary screen by using the coupling enzyme alone. A number of hits belonging to different structural classes were obtained, and their relative potency for inhibition of PfMetAP1b was determined further by measuring the IC₅₀ values in the enzymatic assay. The most potent hits of each structural class then were tested for their selectivity for PfMetAP1b among the four PfMetAP proteins as well as human MetAP enzymes and their ability to inhibit the growth of P. falciparum in erythrocyte culture. From these follow-up analyses, one structural class, that which contained a 2-(2-pyridinyl)-pyrimidine core, stood out as the most promising inhibitors of PfMetAP1b and P. falciparum in vitro.

A total of 31 analogs that differ in substitution at different positions of the pyrimidine core were used for structure/activity relationship study (Table 1). A substitution of 2-pyridinyl group at the R1 position appeared to be essential for activity, because none of the other substitutions, including phenyl (XC8–10), 3-pyridinyl (XC21 and XC23), and 4-pyridinyl (XC24), at this position yielded analogs with activity toward PfMetAP1b. In contrast to R1, the R2 position appeared to tolerate a wide range of substitutions without a significant loss in activity with the exception of a hydroxyl group at this position (XC2), which is inactive. For this group of analogs, the substituents at R3 and R4 positions were changed in tandem. Of the three combinations of substitutions at R3 and R4, the highest potency was achieved when R3 is chlorine and R4 is a methyl group. For example, XC15 and XC7, which share the same R1 and R2 substituents, differ in IC₅₀ for PfMetAP1b inhibition by >30-fold. Of all of the analogs tested, XC11 emerged as the most potent inhibitor of PfMetAP1b, which was selected for further characterization (Fig. 2B).

Table 1.

Inhibition of pyrimidine derivatives on the activity of PfMetAP1b and proliferation of human foreskin fibroblasts

Compound	R1	R2	R3	R4	PfMetAP1b	Fibroblast
XC1	2-pyridinyl	N,N-dimethyl	OCH3	H	1.17 ± 0.065	9.52 ± 0.002
XC2	2-pyridinyl	OH	OCH3	H	>100	>100
XC3	2-pyridinyl	(4-methoxyphenyl)thio	OCH3	H	1.13 ± 0.071	10.2 ± 0.002
XC4	2-pyridinyl	(2,4-dichlorophenyl)thio	OCH3	H	1.04 ± 0.066	>100
XC5	2-pyridinyl	N-methyl-N-phenyl	OCH3	H	2.17 ± 0.001	27.9 ± 0.003
XC6	2-pyridinyl	1-pyrrolidinyl	OCH3	H	7.91 ± 0.003	>100
XC7	2-pyridinyl	4-morpholinyl	OCH3	H	5.13 ± 0.001	61.3 ± 0.003
XC8	Phenyl	N-methyl	OCH3	H	>100	ND*
XC9	Phenyl	1-pyrrolidinyl	OCH3	H	>100	ND
XC10	Phenyl	4-morpholinyl	OCH3	H	>100	ND
XC11	2-pyridinyl	N-(2-phenylethyl)	Cl	CH3	0.112 ± 0.007	75.2 ± 0.96
XC12	2-pyridinyl	N-methyl(phenylmethyl)	Cl	CH3	0.158 ± 0.014	73.4 ± 1.4
XC13	2-pyridinyl	1-pyrrolidinyl	Cl	CH3	0.521 ± 0.034	ND
XC14	2-pyridinyl	N-methyl	Cl	CH3	0.843 ± 0.066	>100
XC15	2-pyridinyl	4-morpholinyl	Cl	CH3	0.152 ± 0.012	>100
XC16	2-pyridinyl	(Phenylmethyl)thio	Cl	CH3	0.938 ± 0.059	>100
XC17	2-pyridinyl	Ethanone, 1-(4-methoxyphenyl)-2-thio	Cl	CH3	1.02 ± 0.076	>100
XC18	2-pyridinyl	Ethanone, 1-(4-fluorophenyl)-2-thio	Cl	CH3	1.24 ± 0.081	>100
XC19	2-pyridinyl	Ethanone, 1-phenyl-2-thio	Cl	CH3	1.54 ± 0.10	>100
XC20	2-pyridinyl	Ethanone, 1-(4-morpholinyl)-2-thio	Cl	CH3	3.12 ± 0.28	>100
XC21	3-pyridinyl	N-(phenylmethyl)	Cl	CH3	>100	>100
XC22	2-pyridinyl	N-[1,2-Ethanediamine,	Cl	CH3	0.199 ± 0.002	ND
		N′-[5-(trifluoromethyl)-2-pyridinyl]]
XC23	3-pyridinyl	N-[1,2-Ethanediamine,	Cl	CH3	>100	ND
		N′-[5-(trifluoromethyl)-2-pyridinyl]]
XC24	4-pyridinyl	N-[1,2-Ethanediamine,	Cl	CH3	>100	ND
		N′-[5-(trifluoromethyl)-2-pyridinyl]]
XC25	2-pyridinyl	N-(phenylmethyl)	H	CF3	1.06 ± 0.095	>100
XC26	2-pyridinyl	N-[(2-chlorophenyl)methyl]	H	CF3	1.06 ± 0.093	>100
XC27	2-pyridinyl	N-[[2-(trifluoromethyl)phenyl]methyl]	H	CF3	1.07 ± 0.074	>100
XC28	2-pyridinyl	N-(2-thienylmethyl)	H	CF3	1.03 ± 0.072	>100
XC29	2-pyridinyl	Phenylthio	H	CF3	1.11 ± 0.055	>100
XC30	2-pyridinyl	(4-methoxyphenyl)thio	H	CF3	1.23 ± 0.086	>100
XC31	2-pyridinyl	2-propynylthio	H	CF3	1.18 ± 0.089	>100

Data are expressed as mean IC₅₀ (micromolar) ± SD. ND, not determined.

The selectivity of XC11 for PfMetAP1b among the four PfMetAP enzymes was investigated. Using purified recombinant PfMetAP1a and PfMetAP1c lacking its N-terminal 143 noncatalytic domain (to increase the solubility of the recombinant protein) in the same coupled enzymatic assay (13), XC11 was found to be >100-fold selective toward PfMetAP1b over the other two related PfMetAP1 enzymes (Fig. 2C). Because recombinant PfMetAP2 does not possess appreciable enzymatic activity when expressed in E. coli or insect cells (X.C. and J.O.L., unpublished data), we used a mammalian three-hybrid assay similar to the yeast three-hybrid system (15) to determine the potential interaction between XC11 and PfMetAP2. Two fusion proteins were coexpressed in 293T cells, one being the full-length PfMetAP2 fused with the VP16 transactivation domain and the other being the hormone-binding domain of rat glucocorticoid receptor fused to the Gal4 DNA-binding domain. In the presence of a dexamethasone-fumagillin dimer, the two fusion proteins are brought together to the enhancer region of a luciferase reporter gene under the control of multimerized Gal4-binding sites. Whereas TNP-470, which, like fumagillin, bound to PfMetAP2 and inhibited the activation of the reporter gene mediated by dexamethasone-fumagillin in a dose-dependent manner, XC11 only weakly inhibited the interaction between dexamethasone-fumagillin and PfMetAP2-VP16AD at 100 μM (Fig. 2D). We note that the apparent IC₅₀ value for TNP-470 in the mammalian three-hybrid assay is much higher than that for its inhibition of endothelial cell proliferation due, in part, to the overexpression of the PfMetAP2-VP16-AD fusion protein. Nevertheless, the difference in the apparent IC₅₀ values between TNP-470 and XC11 for competition against the Dexamethasone-fumagillin dimer clearly indicated that XC11 interacts with PfMetAP2 much more weakly than TNP-470 in 293T cells. Furthermore, the IC₅₀ values for TNP-470 against PfMetAP1a, 1b, and 1c were >100 μM (data not shown). Together, these results indicate that XC11 is quite selective for PfMetAP1b among the four PfMetAP enzymes. To further assess the selectivity and potential toxicity of XC11, we also determined its effects on the enzymatic activity of human MetAP1 and MetAP2 as well as the proliferation of primary human foreskin fibroblasts. XC11 has a moderate selectivity for PfMetAP1b over its human counterparts, with IC₅₀ values of 0.7 and 7 μM for HuMetAP1 and HuMetAP2, respectively. Importantly, XC11 showed minor toxicity toward the primary human fibroblasts with an IC₅₀ of 75 μM in a cell proliferation assay (Table 1).

Antimalarial Effects of XC11 in Vitro.

XC11 inhibited the P. falciparum proliferation of both the CQ-sensitive 3D7 and the CQ-resistant Dd2 isolates in a dose-dependent manner with IC₅₀ values of 0.90 and 3.1 μM, respectively (Fig. 3A). In contrast, those compounds (i.e., XC2, XC8–10, XC21, and XC23–24) that were inactive in inhibiting PfMetAP1b enzyme activity (Table 1) were also inactive when tested in P. falciparum 3D7 strain in culture with IC₅₀ values >100 μM (data not shown), which are >100 times higher than that for XC11 (0.90 μM). The lack of effects of those analogs against either PfMetAP1b and malaria parasites further supports the notion that PfMetAP1b is the relevant pharmacological target of XC11. We then determined the morphologic effect of XC11 on synchronized ring-stage P. falciparum (ITG strain, CQ-resistant) parasites. Although control parasites proceeded to schizont-stage in 24 h, parasites were arrested at the ring stage in the continued presence of XC11 (Fig. 3B).

Fig. 3.

Effects of XC11 on malaria parasite growth in vitro. (A) Inhibitory effect of XC11 on the proliferation of P. falciparum 3D7 strain (CQ-sensitive) (□) and Dd2 strain (multidrug-resistant) (▴). (B) Blood smear of initially synchronized ring-stage P. falciparum ITG strain (CQ-resistant) 24 h after continuous exposure to either carrier control (B1) or compound XC11 (B2).

Antimalarial Activity of XC11 in Vivo.

Having shown that XC11 is active against P. falciparum in erythrocyte culture, we next determined its antimalarial activity in vivo in both CQ-sensitive (Plasmodium berghei) and -resistant (Plasmodium yoelii) mouse models. XC11 reduced P. berghei K173 parasitemias in a dose-dependent manner with a standard 4-day parasite suppression test (ref. 16; Fig. 4A). Importantly, XC11 also extended the mean survival time (MST) of infected animals. Dosing of XC11 at 20 mg/kg twice daily cured 60% of mice (Fig. 4B). In mice infected with P. yoelii 17X lethal strain, which is naturally CQ-resistant (16), XC11 was found to significantly inhibit parasitemia on day 4 (Fig. 4C) and extend MST in a dose-dependent manner (Fig. 4D). XC11 alone at 20 mg/kg extended MST by 733% and cured 80% of mice. A synergistic effect also was observed when XC11 was used in combination with CQ. When 20 mg/kg XC11 was given together with 5 mg/kg CQ, all of the mice were cured (Fig. 4D), as opposed to a 40% cure rate with a 5 mg/kg CQ regimen.

Fig. 4.

Effect of XC11 in C57BL/6 mice infected with P. berghei (CQ-sensitive) (A and B) or P. yoelii (CQ-resistant) (C and D). (A) The level of parasitemia on day 5. ∗∗, P < 0.01 vs. control group. (B) The total survival time. Control (■); 2.5 mg/kg XC11 (○); 5.0 mg/kg XC11 (▵); 10 mg/kg XC11 (×); 20 mg/kg XC11 (□); 1.0 mg/kg CQ (▴); 5.0 mg/kg CQ (●). n = 5 in all groups. (C) The level of parasitemia on day 4. ∗, P < 0.05; ∗∗, P < 0.01 vs. control group (no drug treatment); #, P < 0.05 and ##, P < 0.01 vs. 5.0 mg/kg CQ-treated group. (D) The total survival time. Control (■); 5.0 mg/kg XC11 (○); 20.0 mg/kgXC11 (□); 5.0 mg/kgCQ (▵); XC11 5.0 mg/kg + 5.0 mg/kg CQ 5 (×); 20.0 mg/kg XC11 + 5.0 mg/kg CQ (●). n = 5 in all groups.

Discussion

The cotranslational processing of N-terminal methionine is a highly conserved process during evolution; it is essential for the survival and proliferation of both prokaryotes and eukaryotes. Although yeast expresses two and mammals contain three MetAP enzymes, P. falciparum possesses four MetAP enzymes. It has been shown that the two yeast MetAP-encoding genes are functionally redundant. Deletion of either MetAP genes in yeast causes a slow-growth phenotype, and lethality is seen only when both MetAP-encoding genes are knocked out. Unlike yeast, evidence exists supporting the notion that each individual MetAP gene may be crucial in higher eukaryotes. For example, knockout of MetAP2 in the worm caused cessation of germ cell development (17), whereas mutation of its ortholog in the fly led to severe developmental abnormalities (18). Moreover, inhibition of MetAP2 alone by the highly specific small molecule inhibitors fumagillin or ovalicin is sufficient to block angiogenesis and immune response in mammals (19). These observations suggest that inactivation of one member of the MetAP enzyme family could have a dramatic effect on the survival or growth of higher eukaryotic organisms, raising the question of whether inhibition of one of the four PfMetAP proteins is sufficient to cause a blockade of malaria growth.

We cloned and expressed all four PfMetAP proteins and determined their enzymatic activity. Whereas all three MetAP1 homologues in their recombinant form exhibited full methionine aminopeptidase activity, PfMetAP2 is inactive even upon expression and purification from insect cells. It is known that mammalian MetAP2 undergoes posttranslational glycosylation, which may be required for activity (20). It is possible that PfMetAP2 undergoes a unique posttranslational modification that is not recapitulated in either bacteria or insect cells. It is also possible that PfMetAP2 is active only in a unique cellular environment within the malaria parasites yet to be reconstituted in vitro. But the availability of fully active recombinant PfMetAP1 enzymes made it possible to identify their inhibitors via high-throughput screening.

We chose PfMetAP1b over its two homologues as our first screening target for its close resemblance to the human and yeast MetAP1 with a conserved N-terminal zinc finger motif. Upon screening 175,000 compounds, several distinct structural classes of small molecule inhibitors were identified for PfMetAP1b. However, only one structural class containing pyridinyl-pyrimidine core showed antimalarial activity upon testing in malaria culture. This result is not surprising, because inhibitors that were active in vitro have to penetrate into the parasites and remain sufficiently stable to exert its effects on PfMetAP1b protein in vivo. A limited structure/activity relationship study with 31 structurally related analogs revealed key structural elements required for activity. The most critical structural feature is the 2-(2-pyridinyl)-pyrimidine core, as alteration of relative configuration of the substitution from 2-pyridinyl to either 3- or 4-pyridinyl groups yielded analogs that are inactive. 2-(2-pyridinyl)-pyrimidine is an excellent chelator of metal ions. Other inhibitors of MetAP enzymes with metal-chelating capability have been shown to bring a third metal ion into the active site for the binding and inhibition of the enzyme (21, 22). Whether a similar mode of inhibition is operative for XC11 against PfMetAP1b remains to be seen.

The selectivity exhibited by XC11 among the four PfMetAP enzymes is remarkable. Whereas the selectivity of XC11 among the three PfMetAP1 enzymes was straightforward to determine, the effect of XC11 toward PfMetAP2 was difficult, because it has defied efforts to express and purify it in an active form. To circumvent the problem, we resorted to the mammalian version of the yeast three-hybrid system that is used for detection of the interaction between small molecules and their protein targets (15). The three-hybrid system has the advantages of expressing protein targets in a cellular environment and being independent of the intrinsic enzymatic activity of a protein. All it detects is the physical interaction between a ligand and its binding protein. We thus were able to establish a three-hybrid system in mammalian cells and demonstrate the activation of a reporter gene mediated through the three-hybrid interaction between the dexamethasone-fumagillin and PfMetAP2 fused to VP16 activation domain. The specificity of this three-hybrid system was underscored by the dose-dependent inhibition of the three-hybrid interaction by TNP-470. In contrast to TNP-470, XC11 only marginally affected the three-hybrid reporter system at the highest concentration (100 μM) used, suggesting that XC11 has low affinity for PfMetAP2. Although it is not surprising that XC11 is more selective for PfMetAP1b over PfMetAP2, it also exhibited >100-fold selectivity for PfMetAP1b over its closely related homologues PfMetAP1a and PfMetAP1c. Although PfMetAP1b differs from both PfMetAP1a and PfMetAP1c in certain residues in the active site (Fig. 1), the precise molecular basis of this extraordinary selectivity of XC11 for PfMetAP1b remains to be determined.

Proteases are involved in various physiological processes, some of which are unique for malaria parasites, and they have served as attractive targets for development of antimalarial agents (23, 24). Given the unique role of the MetAP family of enzymes in protein biogenesis, inhibitors of PfMetAP1b including XC11 have a unique mechanism of action that is unrelated to other protease inhibitors and most, if not all, known antimalarial drugs. This unique mode of action is underscored by the inhibition of both CQ-sensitive and -resistant parasitic strains both in vitro (Fig. 3) and in vivo (Fig. 4). As such, inhibitors of PfMetAP enzymes have the potential to circumvent the problem of resistance to known antimalarial drugs, a major hurdle in the effective treatment of the disease.

The identification of XC11 as a potent PfMetAP1b inhibitor and demonstration of its efficacy in blocking malaria growth both in cell culture and in vivo suggest that PfMetAP1b is a promising target for the development of new antimalarial agents, because the inhibition of malaria growth cannot be attributed to the inhibition of any other members of this family of enzymes in light of the specificity of XC11 of PfMetAP1b (Fig. 2 C and D). Moreover, XC11 may inhibit the growth of gametocytes because it has been shown that PfMetAP1b is expressed in gametocytes (http://plasmodb.org), which would extend the efficacy of this family of inhibitors. Improvement of XC11 and its structurally related analogues in its selectivity toward PfMetAP1b over its human counterparts and its potency in vivo thus may lead to the development of a novel class of antimalarial agents.

Materials and Methods

Cloning and Expression of PfMetAP1b.

The gene encoding PfMetAP1b (http://plasmodb.org, Gene ID: PF10_0150) was amplified from P. falciparum 3D7 genomic DNA by using a 5′ primer, 5′-CGCGGATCCGCAAATATTGATGATATA-3′, and 3′ primer, 5′-CGCGAATTCTTAATTATAGTATAGTTCA-3′. The PCR product was cloned into pGEX-6P-2 vector (Amersham, Piscataway, NJ) in frame with an N-terminal GST tag. The DNA insert was confirmed by sequencing. BL21-CodonPlusRIL (Stratagene, La Jolla, CA) was used as the E. coli expression host. The fusion protein was induced with 1 mM (final concentration) IPTG (Sigma, St. Louis, MO) for 24 h at 25°C.

Purification of PfMetAP1b.

The IPTG-induced BL21 bacterial culture was harvested, resuspended in PBS (pH 7.5), and passed through a French press twice. The extract was centrifuged at 17,000 × g for 30 min. The supernatant was loaded onto a Glutathione Sepharose 4B (Amersham) column equilibrated with PBS. The column was washed with 30 bed volumes of the same buffer. An on-column digestion then was performed to remove the GST tag from the fusion protein by loading to the column 40 μl (80 units) of GST-tagged PreScission Protease (Amersham) diluted in 960 μl of cleavage buffer (50 mM Tris·HCl/150 mM NaCl/1 mM EDTA/1 mM DTT, pH 8.0 at 4°C) at 4°C. The digestion was allowed to proceed for 16 h at 4°C. The cleaved PfMetAP1b protein free of the GST tag was eluted from the column, and the eluate was collected, concentrated, and dialyzed against 3 × 100 volumes of PBS containing 10 mM EDTA (pH 8.0 at 4°C) (12 h for each time) to remove metal ions. The concentration of the final product (Apo-PfMetAP1b) was determined by the Bradford assay. SDS/PAGE analysis was performed to determine the size of the protein and the purity of the final product.

High-Throughput Screening.

Compounds were provided by ASDI (Newark, DE) or purchased from Maybridge (Trevillet, UK), Bionet (Cornwall, U.K.), and Sigma, dissolved in 100% DMSO, and stored at −20°C. One hundred seventy-five thousand compounds arrayed in 384-well-plate (Nunc, Roskilde, Denmark) at 5 μM final concentration were screened by using a microplate processor (MAP-C2; Titertek, Huntsville, AL) for dispensing reagents and a microplate reader (BMG, Offenburg, Germany) for data collection. Apo-PfMetAP1b (222 nM final concentration) first was incubated with test compounds in 45 μl of Buffer A for 20 min. The enzymatic reaction was initiated by addition of 5 μl of 6 mM Met-Pro-pNA in Buffer A together with 0.005 units of ProAP. Absorbance at 405 nm was recorded at 30 min. The percentage of net decrease in A₄₀₅ in comparison with control wells with solvent carrier was taken as the percentage of inhibition by a given compound tested. A cutoff of 50% inhibition was used to identify hits. The preliminary hits were rescreened at three different concentrations (30, 10, and 3 μM), and the relative potency was compared and used to rank different hits.

To rule out the possibility that any PfMetAP1b hits found by the above assay may work by inhibiting the coupling enzyme ProAP, the effects of the PfMetAP1b hits on ProAP activity also were determined by using Pro-pNA (Sigma) as substrate. ProAP (0.005 units) first was incubated with the initial PfMetAP1b hits in 45 μl of Buffer A for 20 min. Then, 5 μl of 6 mM Pro-pNA in Buffer A was added. The absorbance at 405 nm was recorded 30 min after initiation of the ProAP catalyzed hydrolysis of Pro-pNA.

Parasite Culture and Screening.

P. falciparum 3D7 (CQ-sensitive) and Dd2 strain (CQ-, quinine-, pyrimethamine-, and sulfadoxine-resistant) were propagated at 37°C in RPMI medium 1640 with 10% human serum at 3–5% hematocrit in a reduced oxygen environment (5% CO₂, 5% O₂, and 90% N₂). Synchronized parasites at ring stage were incubated with test compounds for 24 h (25, 26). [³H]-hypoxanthine (Amersham) was added, and the incubation was continued for an additional 24 h. Cells were frozen at −80°C for 1 h, thawed, and collected by a cell harvester (Harvester96 Mach III, Tomtec, Orange, CT). Radioactivity was counted in a 1450 MicroBeta plate reader (Wallac, Gaithersburg, MD). A 96-well plate with 200 μl of culture medium at 0.2% parasitemia and 2–4% hematocrit gave a radioactive incorporation signal of ≈10,000 cpm at 48 h with background counts <500 cpm. Percentage of control (no drug treatment) was used to determine IC₅₀ values.

Animal Experiments.

All animal experiments were performed under a protocol approved by the Johns Hopkins Animal Care and Use Committee in accordance with institutional standards. C57BL/6 mice (male, 5–6 wk of age) were from the National Cancer Institute. The P. berghei K173 was purchased from American Type Culture Collection (Manassas, VA); the P. yoelii 17X lethal strain was a gift of N. Kumar (Johns Hopkins Malaria Research Institute). Mice were inoculated (i.p.) with 1 × 10⁸ infected erythrocytes on day 0. For a standard 4-day suppressive test, mice were given (i.p.) tested compound and/or CQ or vehicle control 2 h after infection, twice daily for 4 days. Blood was taken from the tail vein on day 4 for P. yoelii and day 5 for P. berghei K173. Giemsa-stained smears were prepared and parasitemias was determined by counting ≈1,000 erythrocytes. For low parasitemias (<1%), up to 3,000 erythrocytes were counted. Data were presented as the mean percentage of parasitemia ± SD. Mice that were alive 30 days after infection with complete clearance of parasitemia and had no recrudescence within the next 30 days were considered cured.

Statistics.

IC₅₀ values for enzymatic assay and cell proliferation assays were determined by using four-parameter logarithmic analysis with GraphPad (San Diego, CA) Prism4 and were presented as mean ± SD for triplicate experiments. For animal tests, P values were determined by using the two-tailed Student’s t test.

Supporting Information.

For additional details, see Supporting Materials and Methods, which is published as supporting information on the PNAS web site.

Abbreviations

CQ

chloroquine

MetAP

methionine aminopeptidase.