Sulfonylpiperazine compounds prevent Plasmodium falciparum invasion of red blood cells through interference with actin-1/profilin dynamics

Abstract

With emerging resistance to frontline treatments, it is vital that new antimalarial drugs are identified to target Plasmodium falciparum. We have recently described a compound, MMV020291, as a specific inhibitor of red blood cell (RBC) invasion, and have generated analogues with improved potency. Here, we generated resistance to MMV020291 and performed whole genome sequencing of 3 MMV020291-resistant populations. This revealed 3 nonsynonymous single nucleotide polymorphisms in 2 genes; 2 in profilin (N154Y, K124N) and a third one in actin-1 (M356L). Using CRISPR-Cas9, we engineered these mutations into wild-type parasites, which rendered them resistant to MMV020291. We demonstrate that MMV020291 reduces actin polymerisation that is required by the merozoite stage parasites to invade RBCs. Additionally, the series inhibits the actin-1-dependent process of apicoplast segregation, leading to a delayed death phenotype. In vitro cosedimentation experiments using recombinant P. falciparum proteins indicate that potent MMV020291 analogues disrupt the formation of filamentous actin in the presence of profilin. Altogether, this study identifies the first compound series interfering with the actin-1/profilin interaction in P. falciparum and paves the way for future antimalarial development against the highly dynamic process of actin polymerisation.

Fig 1. Resistance selection and whole genome sequencing reveal actin-1 and profilin as candidate proteins involved in the MoA of MMV291.

(A) Chemical structure of MMV291. (B) i Drug cycling on and off for 3 cycles and subsequent cloning out of parental lines resulted in 2 clones from 3 populations (Pop B, C, and D) that maintained stable resistance to MMV291 in a 72-hour growth assay. Growth has been normalised to that of parasites grown in 0.1% DMSO with error bars representing the standard deviation of 3 biological replicates. ii EC₅₀ values derived from nonlinear regression curves in GraphPad Prism with 95% confidence intervals shown in brackets. und = undefined. Source data can be found in S1 Data. (C) Genome sequencing of the MMV291-resistant parasites revealed a different nonsynonymous single nucleotide polymorphism (SNP) shared by the clonal lines across 2 related proteins; Populations D and B contained K124N and N154Y mutations in profilin (PF3D7_0932200), respectively, while Population C contained a M356L mutation in actin-1 (PF3D7_1246200). Scale bar indicates 100 base pairs. (D) The positions of the resistance mutations were mapped onto the X-ray structures of P. falciparum actin-1 (purple) (PDB: 6I4E) (42) and P. falciparum profilin (pink) (PDB: 2JKG) (36), which revealed PFN(N154Y) and ACT1(M356L) lie on either side of the proteins’ binding interfaces. PFN(K124N) resides on the opposing side of profilin. In this case, the X-ray structures of Oryctolagus cuniculus actin and human profilin (PDB: 2PBD) (56) were utilised as a template to spatially align the 2 parasite proteins.

Fig 2. Introduction of the SNPs in profilin and actin-1 into 3D7 parasites mediates resistance to MMV291.

(A) i Strategy to create the donor plasmid to introduce PFN(N154Y), PFN(K124N), and ACT1(M356L) SNPs into 3D7 parasites. Homology regions (HRs) were designed to the 5′ flank (HR1) and 3′ flank (HR2) whereby HR1 was made up of the endogenous genes’ sequence (HR1A) and recodonised fragments (HR1B), encompassing the resistant mutation alleles. A synthetic guide RNA (gRNA) was designed for either profilin or actin-1 to direct Cas9 to the cleavage site and induce double crossover homologous recombination. WR99210 was used to select for integrated parasites via the human hydrofolate reductase (hDHFR). ii Integration into the profilin or actin-1 locus was validated whereby a 5′ UTR primer (i/v) was used in combination with a primer located in the glmS region (k). B) i Integrated parasites were tested in a 72-hour LDH growth assay, which revealed the resistant mutations conferred resistance against MMV291 and confirmed the profilin and actin-1 proteins as involved in the MoA of the compound. Growth has been normalised to that of parasites grown in 0.1% DMSO, and error bars indicate the standard deviation of 3 biological replicates. Source data can be found in S1 Data. ii EC₅₀ values derived from nonlinear regression curves in GraphPad Prism with 95% confidence intervals shown in brackets. und = undefined.

Fig 3. MMV291-resistant parasites demonstrate varying resistance to 4 analogues of MMV291.

Two clones from 3 independently derived MMV291-resistant parasite lines were tested in 72-hour LDH growth assays. Varying degrees of resistance to S-W827 (A), S-W936 (B), S-W414 (C), and S-W415 (D) was observed, with Population C clones demonstrating the greatest resistance and Population B clones retaining the most sensitivity to the 4 molecules. Values were normalised to parasite growth in 0.1% DMSO, with error bars representing the mean of 3 biological replicates. Dose response curves were generated in GraphPad Prism using nonlinear regression to derive mean EC₅₀ values, which are stated in the table. S.D indicates the standard deviation calculated from EC₅₀ values across 3 biological experiments. Heat map indicates degree of resistance from 3D7 control lines, with yellow and red indicating the lowest and highest degree of resistance, respectively. Source data can be found in S1 Data.

Fig 4. MMV291 treatment prevents F-actin formation in merozoites.

(A) Synchronised schizonts from a P. falciparum parasite line expressing an F-actin-binding chromobody were incubated with DMSO, Cytochalasin D (CytD), MMV291, and analogues S-W936 and R-W936, for 20 minutes at 37°C to allow merozoite egress. Merozoites were then imaged to detect either a normal punctate apical F-actin fluorescence signal or uniform signal, indicative of the inhibition of F-actin formation. Arrow heads depict punctate F-actin signal, and scale bar indicates 1 μm. (B) The proportion of merozoites with a punctate or uniform signal were scored with >550 merozoites counted for each treatment. Merozoites treated with the lower concentrations of the less active R-W936 had equal proportions of punctate and uniform fluorescence signals, like the DMSO control. In contrast, CytD, MMV291, and the active S-W936 compounds all greatly inhibited the formation of a punctate F-actin signal. Error bars represent the standard deviation of 2 biological replicates, each made up of 3 technical replicates from 3 individual counters. Statistical analysis was performed using a one-way ANOVA, comparing the mean of CytD punctate proportions with the mean of other treatments. *** indicates P < 0.001; no bar indicates not significant. DMSO and CytD were used at concentrations of 0.1% and 1 μM, respectively. Source data can be found in S1 Data.

Fig 5. MMV291 analogues interfere with actin polymerisation in the presence of profilin in vitro.

PfACT1 (4 μM) under polymerizing conditions was quantified in the supernatant and pellet fractions in the presence of the MMV291 analogues (25 μM) or DMSO and upon addition of PfPFN (16 μM). (A) In the absence of PfPFN, 80 ± 4% of PfACT1 sedimented to the pellet fraction with the vehicle DMSO treatment. S-W936 decreased the amount of actin in the pellet to 68 ± 7%, while the remaining compounds had no significant effects on actin sedimentation. (B) Upon addition of PfPFN, actin sedimentation decreased to 21 ± 1% with DMSO treatment. All MMV291 analogues, S-MMV291, R-MMV291, S-W936, R-W936, S-W414, and S-W827, decreased the amount of actin in the pellet further to 11 ± 1%, 15 ± 2%, 8 ± 2%, 10 ± 4%, 9 ± 4%, and 5 ± 2%, respectively. Results are plotted as mean ± standard deviation of the relative amounts of actin in the pellet fraction. The data are based on at least 3 independent assays each performed in triplicate. Statistical significances were determined using an unpaired two-tailed t test, where ** P ≤ 0.01 and *** P ≤ 0.001, and **** ≤ 0.0001. No bar indicates not significant. Source data can be found in S1 Data.

Fig 6. MMV291 disrupts actin-dependent apicoplast segregation and induces a partial delayed death phenotype.

(A) i Representative panels from live cell imaging of apicoplast targeted acyl carrier protein (ACP) tagged with GFP revealed that treatment of trophozoites for 24 hours with both 5 μM and 10 μM MMV291 (10 and 20 × EC₅₀) disrupted apicoplast segregation, resulting in an increase in abnormal apicoplast clumping at schizonts. Scale bar indicates 5 μm. ii These images were quantified by 3 independent blind scorers, which showed a significant decrease in the normal segregation of apicoplasts between the 10 μM MMV291 and DMSO treatments, while the 5 μM MMV291 was not significant (ns). The number of cells analysed were 60, 48, and 47 for DMSO, 5 μM MMV291, and 10 μM MMV291, respectively, which were captured over 3 biological replicates. The error bars represent the standard deviation of 3 independent blind scoring. Statistics were performed via a two-way ANOVA using GraphPad Prism between the DMSO segregated panel and the other treatments. *** indicates P < 0.001. (B) Schematic of the delayed death assay set-up. At 0–4 hours post invasion (hpi) ring-stage parasites expressing nanoluciferase (Nluc) were exposed to titrations of compounds for approximately 40 hours before compounds were washed out. Each cycle for 3 cycles, samples were collected for evaluation of Nluc activity to quantify parasitemia. (C) Azithromycin (i), chloroquine (ii), and MMV291 (iii) were evaluated in the delayed death assay where it was found that unlike azithromycin, MMV291 did not display characteristics of a delayed death inhibitor but had partial reduction in parasite growth at the highest concentration used (10 μM) in the second and third cycles. In contrast, the fast-acting antimalarial chloroquine exhibited killing activity in the first cycle. Growth was normalised to that of parasites grown in 0.1% DMSO and EC₅₀ values (iv) were derived from dose–response curves plotted from nonlinear regressions in GraphPad Prism with 95% confidence intervals of these values specified in brackets. Source data can be found in S1 Data.

Fig 7. Proposed model for MMV291 interference in profilin-mediated filamentous actin polymerisation.

(A) Treadmilling model of profilin’s role in sequestering G-actin and stimulating the exchange of ADP for ATP before delivering the subunits to the barbed end of the growing filament. Here, formin initiates the polymerisation process to form F-actin. Hydrolysis of the G-actin-ATP occurs at this end to produce G-actin-ADP and inorganic phosphate (P_i), to stabilise the filament. The slow release of P_i at the pointed end induces filament instability and proteins such as ADF1 bind to G-actin-ADP to aid in the release of the subunits, thereby severing the filaments. (B) A potential mechanism for MMV291’s inhibitory activity could be through the stabilisation of the G-actin/profilin dimer therefore inhibiting the formation of F-actin and preventing the generation of force required for invasion. ADF1, actin depolymerising factor 1; F-actin, filamentous actin; G-actin, globular actin.