The Antimalarial Natural Product Salinipostin A Identifies Essential α/β Serine Hydrolases Involved in Lipid Metabolism in P. falciparum Parasites

Abstract

Salinipostin A (Sal A) is a potent antiplasmodial marine natural product with an undefined mechanism of action. Using a Sal A-derived activity-based probe, we identify its targets in the Plasmodium falciparum parasite. All of the identified proteins contain α/β serine hydrolase domains and several are essential for parasite growth. One of the essential targets displays a high degree of homology to human monoacylglycerol lipase (MAGL) and is able to process lipid esters including a MAGL acylglyceride substrate. This Sal A target is inhibited by the anti-obesity drug Orlistat, which disrupts lipid metabolism. Resistance selections yielded parasites that showed only minor reductions in sensitivity and that acquired mutations in a PRELI domain-containing protein linked to drug resistance in Toxoplasma gondii. This inability to evolve efficient resistance mechanisms combined with the non-essentiality of human homologs makes the serine hydrolases identified here promising antimalarial targets.

Keywords: Plasmodium falciparum; Salinipostin A; activity-based probes; chemical proteomics; lipid metabolism; malaria; natural products; serine hydrolases.

Figure 1.. Sal A Covalently Binds and Inhibits Serine Hydrolases

(A) Chemical structure of salinipostin A (Sal A) and palmitic acid. (B–D) Competition assays with the serine hydrolase-reactive fluorophosphonate-rhodamine (FP-Rho) probe identify multiple serine hydrolases as potential targets of Sal A in (B) Plasmodium falciparum parasites, (C) Toxoplasma gondii Δku80 and ΔPPT1 parasites, and (D) HEK293T cells with or without preincubation with Sal A.

Figure 2.. Sal Alkyne Identifies Multiple Target Proteins in P. falciparum

(A) Proposed mode of action of Sal A. This natural product can form a covalent adduct with the nucleophilic serine catalytic residues at the active site of α/β hydrolase enzymes, rendering them inactive. (B) Semi-synthesis of Sal A alkyne (Sal alk). DCM, dichloromethane; TBAI, tetrabutylammonium iodide (see Figure S1). (C) Sal alk labeling of proteins in P. falciparum with or without Sal A preincubation. After the Sal alk probe labeling, lysates of saponin-isolated parasites were prepared and labeled proteins were conjugated with TAMRA-azide using a CLICK reaction (see Figure S2). (D) Volcano plots of Sal alk targets in P. falciparum. The x axis shows the logarithm values of difference in spectral counts of protein between Sal alk-treated and DMSO vehicle-treated parasites (left) and Sal A-pretreated (competition) and DMSO vehicle-treated parasites (right). The statistical significance was determined without correcting for multiple comparisons in triplicates and the y value in the volcano plot is minus one times the logarithm of the p value. The proteins with high abundance and competition by the parent compound Sal A are highlighted in red (see Tables 1 and S1 for identified proteins in red).

Figure 3.. PfMAGLLP Is Structurally and Functionally Similar to Human MAGL

(A) Sequence alignment of PfMAGLLP and hMAGL using Clustal W2. Identical or highly similar residues are indicated with (*) or (:), respectively. Sequence identity is 17% and the fold and function assignment system score is −74.8. The conserved G-X-S-X-G motif and residues of the catalytic triad are highlighted in yellow and cyan, respectively. The cysteine adjacent to the active site is highlighted in green (see Figure S3). (B) Predicted structure of PfMAGLLP (green) superimposed with human MAGL structure (PDB: 3jw8, cyan) (Bertrand et al., 2010). Catalytic triads are shown in yellow for PfMAGLLP and purple for hMAGL (see Figure S4A). (C) Processing of 4-methylumbelliferyl (4-MU)-based fluorogenic substrates by recombinant PfMAGLLP. Substrates (10 μM) were incubated with 5 nM of rPfMAGLLP and initial cleavage rates (relative fluorescence units/s) were measured. Data were derived from three independent experiments performed in duplicate and were calculated as non-linear regressions using GraphPad Prism with means ± SD (see Figure S4B). (D) Hydrolysis of arachidonyl-1-thio-glycerol by rPfMAGLLP. Hydrolysis of the thioester bond by the enzyme generates a free thiol that reacts with 5,5′-dithiobis-(2-nitrobenzoic acid) resulting in a yellow product thionitrobenzoic acid with an absorbance of 412 nm (blue) (Ellman, 1959). In the absence of either enzyme or substrate, there was no absorbance of TNB.

Figure 4.. PfMAGLLP Inhibitor Screening

(A) Inhibition of rPfMAGLLP as determined by initial cleavage rates (relative fluorescence units/s) by known human MAGL inhibitors (structures shown below). Reactions contained 5 nM rPfMAGLLP and 10 μM 4-methylumbelliferyl octanoate substrate. Data were derived from three independent experiments performed in duplicate and were calculated as non-linear regressions using GraphPad Prism. IC₅₀ values are listed (mean ± SD, n = 3). (B) Inhibition of rPfMAGLLP with a small library of triazole urea-containing compounds (mean ± SD, n = 3, see Figures S5B–S5D). (C) EC₅₀ values of PfMAGLLP inhibitors in 72 h treatment of P. falciparum W2 parasites, beginning as synchronized ring stages, with error bars, SD. (n = 6 parasite cultures from two independent experiments with triplicates). Structures of representative compounds are shown (see Figure S6A).

Figure 5.. Effect of Orlistat in Comparison to Sal A

(A) Chemical structure of Orlistat. (B) IC₅₀ values of Sal A and Orlistat for inhibition of 4-methylumbelliferyl octanoate processing by rPfMAGLLP (mean ± SD, n = 3). (C) EC₅₀ values of Sal A and Orlistat in 72 h treatment of P. falciparum W2 parasites, beginning as synchronized ring stages with error bars, SD (n = 6 parasite cultures from three independent experiments with duplicates). (D) Giemsa-stained parasites. Synchronized trophozoite-stage parasites were treated at 24–30 h post-invasion with DMSO (vehicle control), 2 μM Sal A, or 6 μM Orlistat and imaged 18 h later (see Figure S7).

Figure 6.. Sal A-Selected Parasites

(A and B) Dose-response curves for Sal A-selected parasite clones and the corresponding parental line (Dd2_Polδ) tested against (A) Sal alk or (B) the lipid metabolism inhibitor Orlistat (see Table S2). N,n = 5,2. Data are shown as means ± SEM.