Functional insights into Plasmodium actin-depolymerizing factor interactions with phosphoinositides

Abstract

Malaria is caused by protozoan parasites, Plasmodium spp., that belong to the phylum Apicomplexa. The life cycle of these parasites depends on two different hosts; the definitive host, or vector, is a mosquito, and the intermediate host is a vertebrate, such as human. Malaria parasites use a unique form of substrate-dependent motility for host cell invasion and egress, which is dependent on an actomyosin motor complex called the glideosome. Apicomplexa have a small set of actin regulators, which are poorly conserved compared to their equivalents in higher eukaryotes. Actin-depolymerizing factors (ADFs) are key regulators responsible for accelerating actin turnover in eukaryotic cells. The activity of ADFs is regulated by membrane phosphoinositides. Malaria parasites express two ADF isoforms at different life stages. ADF1 differs substantially from canonical ADF/cofilins and from Plasmodium ADF2 in terms of both structure and function. Here, we studied the interaction of both Plasmodium ADFs with phosphoinositides using biochemical and biophysical methods and mapped their binding sites on ADF1. Both Plasmodium ADFs bind to different phosphoinositides, and binding in vitro requires the formation of vesicles or micelles. Interaction with phosphoinositides increases the α-helical content of the parasite ADFs, and the affinities are in the micromolar range. The binding site for phosphatidylinositol 4,5-bisphosphate in PfADF1 involves a small, positively charged surface patch.

Actin-depolymerizing factor (ADF)/cofilins are among the most central proteins that regulate cell proliferation, migration, polarity, and dynamic regulation of organ morphology (1, 2). ADF/cofilins control actin dynamics by accelerating actin polymerization and depolymerization via their severing activity as well as by nucleation (3, 4, 5). Upon binding to ADP-G-actin, ADF/cofilin proteins inhibit nucleotide exchange to ATP-G-actin (6). ADF/cofilins are regulated by a plethora of mechanisms, including phosphorylation/dephosphorylation, pH, and interactions with other proteins and phosphoinositides. Phosphorylation of a conserved serine residue at the N terminus of ADF/cofilins inhibits their binding to F-actin (7, 8, 9). The severing and depolymerization properties are regulated by pH, and ADF/cofilins typically are most active at high pH (10, 11, 12).

Phosphatidylinositol and its phosphorylated derivatives, phosphoinositides, are multifunctional lipids involved in the modulation of many cellular events, such as signal transduction, regulation of membrane traffic, cytoskeleton, and the permeability and transport functions of membranes (13, 14). The inositol ring of phosphatidylinositols can be reversibly phosphorylated at positions 3, 4, and 5, which results in the formation of seven possible phosphoinositide species. Phosphatidylinositol 4,5-bisphosphate [PI(4,5)P₂] affects the actin cytoskeleton by interacting directly with several actin-binding proteins (15, 16, 17). PI(4,5)P₂ binding inhibits proteins involved in actin filament disassembly and activates nucleation or polymerization-promoting proteins (15, 16, 18, 19). The interaction of ADF/cofilins with phosphoinositides is a regulatory mechanism that mainly occurs at the plasma membrane. ADF/cofilins are among the few actin-binding proteins present at the leading edge of migrating cells, highlighting the importance of ADF/cofilin–membrane interactions. Both PI(4,5)P₂ and phosphatidylinositol 3,4,5-triphosphate [PI(3,4,5)P₃] interact with ADF/cofilins with a relatively high affinity compared to other phosphoinositides and have been suggested to act as phosphoinositide density sensors (20, 21). The binding of phosphoinositides to ADF/cofilins occurs by electrostatic interactions through a large, positively charged surface area and is mutually exclusive with actin binding (20, 22, 23).

Apicomplexa, a diverse group of largely obligate intracellular parasites, including Plasmodium spp., comprise significant pathogens of animals, including humans. They display a unique form of substrate-dependent motility to cross nonpermissive biological barriers (migration), invade the target host (invasion), and exit from infected cells (egress). All these depend on an actomyosin motor (21, 22). Apicomplexan parasites have a limited number of actin-binding proteins controlling actin filament turnover (24, 25).

Unlike other apicomplexan parasites, Plasmodium spp. have two ADF isoforms: ADF1 and ADF2. ADF1 is expressed during all lifecycle stages and implicated in cell motility (26, 27). ADF2, on the other hand, is expressed in the sexual stages of the parasite lifecycle (28). Both ADF1 and ADF2 share the core fold with canonical ADF/cofilins. However, there are marked differences between the Plasmodium ADFs compared to each other and to canonical ADF/cofilins. The largest difference in the Plasmodium ADFs appears in the C-terminal half, which is involved in interactions with G- and F-actin. A hairpin loop (called F-loop) connecting β-strands 4 and 5 is shorter in ADF1 than in ADF2 and in other ADF/cofilins. β-strand 6 connecting α-helix 3 to the C-terminal helix is missing in ADF1 and has a shorter α-helix 4. In addition, hydrophobic residues implicated in G-actin binding are missing in ADF1 (27, 29). All in all, ADF2 is more similar to canonical ADF/cofilins than ADF1.

Several phosphoinositides, including phosphoinositide 3-phosphate (PI3P), phosphoinositide 4-phosphate (PI4P), PI(4,5)P₂ as well as low levels of phosphoinositide 3,4-bisphosphate [PI(3,4)P₂] and PI(3,4,5)P₃, have been detected in Plasmodium falciparum infected erythrocytes (30). A previous study using recombinant PfADF1 has shown a low micromolar affinity for phosphatidylinositol derivatives (27). However, little is known about how Plasmodium ADFs interact with phosphoinositides and their specificities. Here, we used biochemical and biophysical techniques to study the interactions of Plasmodium ADFs with different phosphoinositides.

Results

Plasmodium ADFs bind phosphoinositides specifically

The interaction of Plasmodium ADFs with different phosphoinositides was assessed by incorporating 10% phosphoinositides into 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphatidylcholine (POPC) vesicles. For historical reasons, we have used ADF1 from P. falciparum (referred to as PfADF1) and ADF2 from Plasmodium berghei (PbADF2) in all our studies. The sequence identities between the ADFs of these two malaria parasite species are 86% for ADF1 and 75% for ADF2. Cosedimentation assays for both Plasmodium ADFs showed hardly any binding to POPC vesicles without phosphoinositides, but in the presence of phosphoinositides, the binding was significantly enhanced (Fig. 1, A and B). 18 to 25% of PfADF1 sedimented in the presence of all different phosphoinositides (Fig. 1, A and C), suggesting either a lack of specificity or a limitation of this method. For PbADF2, the lowest binding was observed for PI(4,5)P₂ (Fig. 1D); only approximately 10% of PbADF2 sedimented with PI(4,5)P₂, compared to 13 to 28% with the other phosphoinositides (Fig. 1, B and D).

Figure 1.

Plasmodium ADFs interact with different phosphoinositides. A representative SDS-PAGE analysis of the vesicle cosedimentation assay of PfADF1 (A) and PbADF2 (B), with various 10% phosphoinositides, POPC, and DMPC:DMPG vesicles. S and P denote the supernatant and pellet, respectively. Scatter dot plot showing the proportion of PfADF1 (n = 4 except for DMPC:DMPG vesicles where n = 3) (C) and PbADF2 (n = 3) (D) in the pellet fractions quantitated from Figure 1, A and B, respectively. The final protein and lipid concentrations were 8 and 500 μM, respectively. Data were plotted as mean ± SD. For both, C and D, asterisks represent statistical significances determined with unpaired two-tailed t-tests against the POPC control. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001. The exact p values are shown in Table S2. E, molecular structures of POPC, PI(4,5)P₂, DMPC, and DMPG. ADF, actin-depolymerizing factor; DMPC, 1,2-dimyristoyl-sn-glycero-3phosphocholine; DMPG, 1,2-dimyristoyl-sn-glycero-3-phospho-glycerol; PbADF, ADF from Plasmodium berghei; PfADF, ADF from Plasmodium falciparum; POPC, 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphatidylcholine; PI(4,5)P₂, phosphatidylinositol 4,5-bisphosphate.

As phosphoinositides have negatively charged head groups (Fig. 1E), it was investigated, whether ADFs also bind to other negatively charged vesicles. For this purpose, negatively charged 1,2-dimyristoyl-sn-glycero-3phosphocholine (DMPC):1,2-dimyristoyl-sn-glycero-3-phospho-glycerol (DMPG) (1:1) (Fig. 1E) vesicles were used. Interestingly, neither of the Plasmodium ADFs bound to DMPC:DMPG vesicles more than to POPC (Fig. 1, C and D), suggesting that binding of the Plasmodium ADFs to phosphoinositides is specific to the head groups.

Phosphoinositide binding induces conformational changes in Plasmodium ADFs

We used synchrotron radiation CD (SRCD) spectroscopy to study whether the conformation of Plasmodium ADFs is affected by interaction with phosphoinositides. All phosphoinositide vesicles, except PI(3,4,5)P₃ and PI(3,4)P_2, increased the α-helical content of PfADF1, while the POPC control vesicles did not, as indicated by an increase in the positive absorption peak at 195 nm and the two negative absorption peaks at 208 nm and 222 nm (Fig. 2, A and B, Table 1). The largest changes occurred at 195 nm. Deconvolution of the SRCD spectra between 180 and 250 nm using the beta structure selection server (31) showed an increase of α-helix content up to 1.5-fold seen upon interaction with PI3P (Table 1). All phosphoinositides, except PI(3,4)P₂, increased the α-helical content of PbADF2, PI(3,5)P₂ having the most prominent effect. To investigate whether the micellar structure of the phosphoinositides in solution is important for the ADF binding, the effect of the soluble form SPI(4,5)P₂, that is, dibutanoyl phosphatidylinositol 4,5-bisphosphate, on the secondary structure of the PfADF1 was also tested. SPI(4,5)P₂ did not affect the protein conformation (Fig. 2B). In addition, we did not detect any interaction between SPI(4,5)P₂ and PfADF1 using isothermal titration calorimetry (Fig. S3). Thus, the vesicle environment seems to be required for the binding of PfADF1 to phosphoinositides.

Figure 2.

SRCD spectra of Plasmodium ADFs in the presence of various phosphoinositides. SRCD spectra of PfADF1 with and without different (A) monophosphoinositide vesicles and (B) diphosphoinositide and triphosphoinositide vesicles. SRCD spectra of PbADF2 with and without different (C) monophosphoinositide vesicles, and (D) diphosphoinositide and triphosphoinositide vesicles with final protein and lipid concentrations of 0.12 mg/ml and 875 μM, respectively. ADF, actin-depolymerizing factor; PbADF, ADF from Plasmodium berghei; PfADF, ADF from Plasmodium falciparum; SRCD, synchrotron radiation CD.

Table 1.

Deconvolution of SRCD spectra of Plasmodium ADFs with different phosphoinositides

Protein ± phosphoinositide vesicles	Secondary structure content (%)
	α-Helix	β-Strand	Turn	Other
PfADF1	24	22	11	43
PfADF1+POPC	24	20	11	46
PfADF1+PI3P	41	13	7	39
PfADF1+PI4P	38	12	10	40
PfADF1+PI5P	40	8	10	42
PfADF1+PI(3,4)P2	27	21	12	40
PfADF1+PI(3,5)P2	36	17	7	40
PfADF1+PI(4,5)P2	39	13	10	38
PfADF1+PI(3,4,5)P3	27	22	10	41
PfADF1+SPI(4,5)P2	22	21	12	46
PfADF1 (PDB ID: 2XF1)	32	26	5	38
PbADF2	19	36	10	36
PbADF2+POPC	16	36	10	38
PbADF2+PI3P	42	17	3	38
PbADF2+PI4P	37	18	5	40
PbADF2+PI5P	36	19	5	40
PbADF2+PI(3,4)P2	29	28	5	38
PbADF2+PI(3,5)P2	41	16	4	39
PbADF2+PI(4,5)P2	40	15	7	38
PbADF2+PI(3,4,5)P3	31	25	6	38
PbADF2 (PDB ID: 2XFA)	31	27	9	33

ADF, actin-depolymerizing factor; PbADF, ADF from Plasmodium berghei; PfADF, ADF from Plasmodium falciparum; PI3P, phosphoinositide 3-phosphate; PI4P, phosphoinositide 4-phosphate; PI(3,4)P₂, phosphoinositide 3,4-bisphosphate; PI(3,4,5)P₃, phosphatidylinositol 3,4,5-triphosphate; PI(4,5)P₂, phosphatidylinositol 4,5-bisphosphate.

Both PfADF1 and PbADF2 bind to phosphoinositides with a micromolar affinity

Both Plasmodium ADFs contain single tryptophan residues (Trp-26 in PfADF1 and Trp-92 in PbADF2). The Trp-92 in PbADF2 is at positions similar to tryptophans seen in conventional ADF/cofilins (Fig. 3). Thus, intrinsic tryptophan fluorescence spectroscopy was used to determine the binding affinity of the Plasmodium ADFs to different phosphoinositides (Fig. 4). The background fluorescence from the phosphoinositide vesicles alone was monitored by phosphoinositide vesicle titrations (Fig. S1). The fluorescence intensity of these vesicle titrations was subtracted from the protein phosphoinositide titration, and the changes in tryptophan fluorescence versus phosphoinositide concentration were plotted.

Figure 3.

Multiple sequence alignment of Plasmodium ADFs and selected other ADF/cofilins. The amino acid sequences of Plasmodium ADFs were aligned with other ADF/cofilin family members using ClustalW2 (51). Strictly conserved residues are shown in red boxes, and regions of residues with similar properties are indicated with blue boxes. The secondary structure elements of PfADF1 and PbADF2 are shown above the alignment in black and green, respectively. G-actin–binding sites identified in yeast cofilin by mutagenesis (21) and synchrotron footprinting (53) are marked with black triangles and circles, respectively. Residues involved in the F-actin–binding site are marked with underlined black triangles. The phosphoinositide-binding sites identified by mutagenesis in yeast cofilin (21) and mouse cofilin-1 (20) are marked with green and red diamonds, respectively. The PfADF1 residues mutated in this study are indicated by asterisks above the sequences. The sequences include those of P. falciparum ADF1 (PfADF1), P. berghei ADF2 (PbADF2), T. gondii ADF (TgADF), S. cerevisiae cofilin (ScCof), A. thaliana ADF1 (AtADF1), A. castellanii actophorin (AcAct), M. musculus cofilin-1 (MmCof), and H. sapiens cofilin (HsCof). ADF, actin-depolymerizing factor.

Figure 4.

Determination of Plasmodium ADF phosphoinositide interactions using tryptophan fluorescence. In each sample, the phosphoinositide concentration versus change in fluorescence intensity at 360 nm wavelength was plotted and analyzed using a one-site specific binding model to obtain the binding affinities of 10 μM PfADF1 for (A) PI4P, (B) PI5P, (C) PI(3,4)P₂, (D) PI(4,5)P₂, (E) PI(3,4,5)P3 as well as (F) 10 μM PbADF2 PI(4,5)P₂. Data from three independent experiments were plotted as mean ± SEM. The total lipid concentration was 10 times the phosphoinositide concentration. ADF, actin-depolymerizing factor; PI3P, phosphoinositide 3-phosphate; PI4P, phosphoinositide 4-phosphate; PI(3,4)P₂, phosphoinositide 3,4-bisphosphate; PI(3,4,5)P₃, phosphatidylinositol 3,4,5-triphosphate; PI(4,5)P₂, phosphatidylinositol 4,5-bisphosphate; PbADF, ADF from Plasmodium berghei; PfADF, ADF from Plasmodium falciparum.

When aqueous solutions of PfADF1 were excited at 295 nm, emission spectra with maxima at 334 nm were obtained (Fig. S4). The titration of PfADF1 with phosphoinositides resulted in 5 to 7 nm shifts of the fluorescence emission maxima to longer wavelengths (red shift) in a concentration-dependent manner. A red shift typically results from a tryptophan becoming more exposed to solvent. In addition, titration of PfADF1 with phosphoinositides also increased the peak intensity. A POPC vesicle titration showed a small increase of tryptophan fluorescence intensity without any shifts in the maxima (Fig. S4A). It was not possible to determine K_d for the interaction between PfADF1 and POPC, suggesting either no binding or a very weak interaction (Fig. S5A). All the phosphoinositides, except for PI3P and PI(3,5)P₂, had K_ds in the range of 47 to 200 μM (Fig. 4, A–E). The K_d for PI(3,5)P₂ could not be determined, and PI3P showed an affinity in the millimolar range (Fig. S5). Of the different phosphoinositide vesicles, PI4P had the highest affinity (Fig. 4).

When PbADF2 was excited at 295 nm, emission spectra with maxima at 328 nm were obtained (Fig. S4G). The titration of PbADF2 with phosphoinositides resulted in red shifts of around 6 nm and a concentration-dependent increase in fluorescence intensity. PbADF2 titration with POPC vesicles alone resulted in an increase in fluorescence intensity but no shifts of the maxima were observed, and the binding affinity could not be determined (Fig. S5D). The K_d of PbADF2 binding to PI(4,5)P₂ vesicles was 66 ± 23 μM (Fig. 4F), which indicates a higher affinity than observed for PfADF1 to PI(4,5)P₂ (Fig. 4). We could not determine the binding affinity for PI5P vesicles (Fig. S5E), although we observed a similar shift in the fluorescence maximum as seen for PI(4,5)P₂, suggesting weak binding.

Mapping the PI(4,5)P₂-binding site on PfADF1

Previous studies on chicken and yeast cofilins have been conducted using NMR and native gel electrophoresis to map the phosphoinositide-binding sites. These techniques have limitations, as NMR chemical shift experiments were carried out using water-soluble di-C8 forms of PI(4,5)P₂, and native gel electrophoresis assays were performed under nonphysiological conditions (21, 32). Thus, we combined mutagenesis of PfADF1 with CD and fluorometric assays to shed light on the phosphoinositide binding mode. The previous results suggested that phosphoinositide binding with PfADF1 requires formation of vesicles or micelles, which would be most consistent with binding via electrostatic interactions. Positively charged residues in four clusters important for phosphoinositide binding in other ADF/cofilins were mutated to alanine or glutamine, guided by a multiple sequence alignment (Fig. 3). These mutations are located on the surface of PfADF1 (Fig. 5B). In addition, we mutated Ser-3, a phosphorylation target, to either glutamate to mimic phosphorylation at this site, or an alanine to remove the polar side chain. All mutants were properly folded, as indicated by SRCD (Fig. S2).

Figure 5.

Effect of mutations on PfADF1 binding to PI(4,5)P2 determined using tryptophan fluorescence.A, the phosphoinositide concentration versus change in fluorescence intensity of 10 μM PfADF1 and its mutants at 360 nm wavelength were plotted and analyzed using a one-site specific binding model to obtain the binding affinity for PI(4,5)P_2. The data were plotted as mean ± SEM from three independent experiments. The total lipid concentration was 10 times the phosphoinositide concentration. B, electrostatic surface potential of PfADF1 [PDB ID: 2XFA, (29)]. The amino acids mutated in this study are indicated. Shown are two different orientations 90° apart (red negative, blue positive, and white, neutral). PfADF, actin-depolymerizing factor from Plasmodium falciparum; PI(4,5)P₂, phosphatidylinositol 4,5-bisphosphate.

The effect of the mutations on phosphoinositide binding was studied using PI(4,5)P₂. Based on the SRCD results, PfADF1 mutations did not have notable effects on binding of PI(4,5)P₂ (Fig. S2). However, in the tryptophan fluorescence assay, residues Arg-86 and Arg-88 (Fig. S5, F and G) and Lys-100 and Lys-101 (Fig. S5, H and I) of α3 helix as well as Lys-19, Arg-21, and Lys-22 (Fig. S5J) seemed to be critical for PI(4,5)P₂ binding, as binding was abolished in the alanine and/or glutamine mutants (Table 2 and Fig. S5). This suggests that these residues may contribute to the PI(4,5)P₂ binding. The S3A mutation decreased phosphoinositide binding. However, substitution of serine with glutamate to mimic phosphorylation did not have any effect on the binding affinity to PI(4,5)P₂ (Table 2). Overall, these results suggest that the PI(4,5)P₂-binding site on PfADF1 may be a small positively charged patch located concentrated around these surface residues (Fig. 5B).

Table 2.

Binding affinities of PfADF1 and several mutants to PI(4,5)P2 determined using the tryptophan fluorescence assay

Protein	Kd (μM)
PfADF1	110 ± 19
S3A	460 ± 290
S3E	110 ± 70
K72A	230 ± 86
R86AR88A	n.d.
R86QR88Q	n.d.
K100AK101A	n.d.
K100QK101Q	n.d.
K19AK21AK22A	n.d.

Data are plotted as mean ± SEM (n = 3) (n.d. = not determined).

ADF, actin-depolymerizing factor; PfADF, ADF from Plasmodium falciparum; PI(4,5)P₂, phosphatidylinositol 4,5-bisphosphate.

Discussion

Unlike other Apicomplexa, Plasmodium spp. express two ADFs: ADF1 and ADF2. They are present at different life stages of the parasite, indicating different functions. While ADF1 is essential for parasite viability and invasion and is present throughout the parasite lifecycle, ADF2 is not expressed during the erythrocytic stages (26). Although phosphoinositides only account for 1% of total cellular lipids in eukaryotic cells, they are critical in various cellular processes, such as signal transduction, cell motility, cytoskeletal organization, and membrane transport. They are not present uniformly in cell membranes but enriched in certain cellular compartments (33). ADF/cofilins are regulated by phosphoinositides via inhibition of actin-binding activity because the phosphoinositide and actin-binding sites partially overlap (15, 20, 21). ADF/cofilins typically bind multiple PI(4,5)P₂ head groups through a large, positively charged protein interface (20). In the case of Plasmodium, the regulation of the ADFs by phosphoinositides has not been studied before.

Here, we studied whether and how the Plasmodium ADFs bind to different phosphoinositides. PI(4,5)P₂ is the most abundant phosphoinositide in the plasma membrane in human and likely to play a dominant regulatory role over other phosphoinositides in the organization of the actin cytoskeleton (33). However, other phosphoinositides also interact with ADF/cofilins, and little is known about membrane compositions in the apicomplexan membranes and during different life cycle stages. Both Plasmodium ADFs bind different phosphoinositides in POPC vesicles (Fig. 2). Interestingly, water-soluble short-chain phosphoinositides did not affect the conformation of the Plasmodium ADFs, indicating that the membrane environment or micelle/vesicle formation is important for binding. The curvature of the membrane and ratio of phosphoinositide to other lipids has been reported to be important for phosphoinositide binding in other cytoskeletal proteins (21, 34). These results are consistent with previous findings in the Toxoplasma gondii ADF (TgADF), which did not show any interaction with dioctanoyl PI(4,5)P₂ in the isothermal titration calorimetry or NMR experiments (35). This might be due to the lack of formation of vesicles or micelles, as a water-soluble phosphoinositide was used.

A previous study on yeast and chicken cofilins showed that they have stronger interactions with diphosphorylated and triphosphorylated phosphoinositides than the monophosphorylated forms (20, 32). In contrast, we could not show a clear preference among the various phosphoinositides in the lipid cosedimentation or tryptophan fluorescence assays. However, based on our data, PfADF1 has the highest affinity toward PI4P, and the phosphorylation at position 4 in general seems to be important for binding. Interestingly, CD spectroscopy showed that monophosphorylated phosphoinositides induced a larger conformational change upon binding to the ADFs, than diphosphorylated and triphosphorylated phosphoinositide (Fig. 2, Table 1). This was particularly the case for PI3P, which however, appeared to have a very low binding affinity in the fluorescence assay. Yet, PI3P may have in vivo relevance during parasite infection. The production of phosphoinositides increases after infection of red blood cells with P. falciparum, particularly for PI3P, PI4P, and PI(4,5)P₂. Of the total phosphoinositide monophosphates in infected erythrocytes, 30% are PI3P. PI3P levels do not fluctuate in most eukaryotic cells, but in P. falciparum, there is an approximately 4-fold increase in the proportion of them to total phosphoinositides. It plays a role in hemoglobin uptake, biogenesis of the apicoplast, and artemisinin resistance. Normally, unicellular organisms do not produce PI(3,4,5)P₃, however, the presence of PI(3,4,5)P₃ in the P. falciparum schizonts, and the P. berghei ookinetes has been confirmed (30).

The affinity of PI5P to PfADF1 was 320 μM, while it could not be determined for PbADF2 (Fig. S5), suggesting no binding or a very weak interaction. In line with this, PI5P has been found only in asexual blood stages of malarial parasites (36), where ADF2 is not present. We could not determine the binding affinity for PI(3,5)P₂, which is in line with in vivo findings. PI(3,5)P2 could not be detected in infected red blood cells (30). PI(4,5)P₂ is involved in calcium signaling cascades as a substrate for phospholipase C in sporozoite gliding motility, merozoite egress, and male gametocyte exflagellation (37). An increase in the α-helical content upon binding to phosphoinositides has been previously observed for profilin and gelsolin. CD studies of gelsolin synthetic peptides showed that they undergo coil-to-helix transition upon PI(4,5)P₂ binding (38, 39).

A number of Plasmodium ADF1 mutants, in which a charged residues were mutated to alanine, or glutamine, were constructed to map the PI(4,5)P₂-binding site on the PfADF1. Three of the mutants (K19AR21AK22A, R86AR88A, and K100AK101A) did not strongly interact with PI(4,5)P₂. These sites reside in α1 and the long α3 helix of PfADF1, respectively, suggesting that the PfADF1 might not have a defined phosphoinositide-binding pocket. Therefore, the PfADF1 might simultaneously interact with more than one phosphoinositide molecule or does not contain a specific interaction site. In support of this, the positioning of proteins in membrane (PPM) Webserver (40) predicted that N terminus and α3 helix of both PfADF1 and PbADF2 interact with membranes (Fig. 6). Similar observations have been reported for mouse cofilin-1, yeast cofilin, and human cofilin-1 (20, 21, 41).

Figure 6.

Predicted orientations of PfADF1 and PbADF2 upon membrane binding.A, PPM docking of PfADF1 (A) and PbADF2 (B) on mammalian plasma membrane. In both models, the N termini and α3 are predicted to interact with the membrane. PPM, positioning of proteins in membrane; PbADF, actin-depolymerizing factor from Plasmodium berghei; PfADF, actin-depolymerizing factor from Plasmodium falciparum.

A recent study on different actin-binding proteins showed that the ADF/cofilins bind 3 to 6 phosphoinositide head groups and that the interaction is transient (22). The PfADF1 crystal structure contains four sulphate ions that might mimic phosphoinositide-binding site (29). Interestingly, these sulphates interact with Arg-6, Arg-21, Lys-100, and Lys-101. In this study, mutations to Arg-21, Lys-100, and Lys-101 were found to affect the phosphoinositide binding (Fig. 5A, Table 2). To understand how PfADF1 binds and severs actin filaments, chemical crosslinking and mass spectrometry with protein complex structure reconstruction was performed to build PfADF1 in complex with G- and F-actin (42). In this study, Lys-100 of PfADF1 was found to be crosslinked with SD1 of actin and is thought to be involved in the noncanonical F-actin binding site (42). Thus, phosphoinositide binding to PfADF1 may inhibit actin binding activity. Here, we could only identify small patches of positively charged residues on the parasite ADFs. These residues are all conserved between P. falciparum and P. berghei ADF1s, underlining their functional importance. In contrast to this, mouse cofilin-1 binds to multiple PI(4,5)P₂ headgroups simultaneously via a large positively charged protein interface that overlaps with the G- and F-actin binding sites (20). This was also observed in the affinity for phosphoinositide. PI(4,5)P₂ had a K_d of approximately 110 μM, which is ∼30-fold lower than that of mouse cofilin-1 with a K_d of 4 μM (20). Vesicles containing POPC/phosphatidylcholine/phosphatidylserine (60:20:20) were used in those experiments, whereas POPC vesicles were used in this study. This might also have contributed to the much lower affinity of PfADF1.

The N terminus of the ADF/cofilin proteins is conserved and shown to be important for the F- and G-actin binding. The ADF/cofilins are negatively regulated by phosphorylation (9, 11). However, the S3E mutation, which mimics phosphorylation at this site, did not affect the PfADF1 binding to PI(4,5)P₂, indicating that phosphorylation does not affect the phosphoinositide binding (Table 2). In line with this, phosphorylation of chicken cofilin did not affect the affinity to phosphoinositide or the mode of phosphoinositide binding (32). The alanine mutation to this residue moderately affected binding, suggesting that the residue is involved in the PI(4,5)P₂-binding site. Many studies have shown that the phosphorylation of the N-terminal serine blocks actin interactions and that the S3E mutation made the mutant completely inactive (11, 43, 44). Many cytoskeletal proteins, such as WASP, bind phosphoinositides by basic or aromatic residues (45). As this study identified some basic residues that are involved in the phosphoinositide binding, there might be other residues that may be involved in the phosphoinositide binding, which require further study.

Experimental procedures

Protein expression and purification

Expression constructs for both PfADF1 and PbADF2 in the pETNKI-his-SUMO3 vector (NKI Protein Facility) with an N-terminal His₆ tag were obtained from Dr Moon Chatterjee. Several point mutants (S3A, S3E, K72A, K72Q), double mutants (R86AR88A, R86QR88Q, K100AK101A, K100QK101Q), and a triple mutant (K19AK21AR22A) of PfADF1 were generated from the PfADF1 plasmid usingPCR with Phusion High-Fidelity DNA polymerase (Thermo Fisher Scientific Inc). In each case, the reaction mixture was incubated with DpnI to remove the methylated template. Plasmid DNA was ligated using T4 DNA ligase, followed by transformation to TOP10 competent cells (Invitrogen), which were then plated on Luria-Bertani agar plates containing 50 μg/ml kanamycin. Plasmids were isolated from single colonies and screened for the presence of the desired mutation by DNA sequencing at the DNA Sequencing Core Facility at Biocenter Oulu.

WT PfADF1 and the PfADF1 mutants were transformed into Escherichia coli BL21CodonPlus (DE3) RIPL (Agilent, Santa Clara), while PbADF2 was transformed into Escherichia coli Rosetta (DE3) (Novagen). Selected transformants were inoculated into Luria-Bertani medium at +37 °C with 50 μg/ml kanamycin and 34 μg/ml chloramphenicol and grown overnight at +37 °C. Expression cultures were grown in ZYM-5052 autoinduction medium (46) at +37 °C for 4 h after inoculation with 1% preculture. The cultures were then cooled to +20 °C and incubated for further 36 h for all constructs. The cells were harvested by centrifugation at 5020g for 45 min, washed with PBS, and stored at −20 °C. PfADF1 and PbADF2 were purified as described before (47). The PfADF1 mutants were purified following the PfADF1 purification protocol.

Lipid vesicle preparation

Phosphatidylinositol phosphates were purchased from Echelon Biosciences. POPC was purchased from Avanti Polar Lipids (Alabester) and DMPC and DMPG from Antrace. Lipid stocks of long-carbon-chain phosphoinositides were prepared by dissolving dry lipid in chloroform:methanol:water (20:13:3 v/v), while short-carbon-chain lipids were dissolved in either buffer or water. DMPC:DMPG (1:1) were dissolved in chloroform:methanol (4:1 v/v). Vesicles were prepared by adding deionized water to dried lipids under vigorous shaking. The suspensions were clarified by sonication (Branson 450 Digital Sonifier, Marshall Scientific LLC) for 2 min at 10% amplitude with 1 s pulses, forming small unilamellar vesicles. A mixture of 2 mM POPC:phosphoinositide (90:10), 2 mM POPC, and 5 mg/ml DMPC:DMPG (1:1) were prepared.

Vesicle cosedimentation assay

A vesicle cosedimentation assay was performed to determine the binding ability of Plasmodium ADFs to different phosphoinositide vesicles. Five hundred micromolars small unilamellar vesicles were mixed with 8 μM protein in 20 mM Hepes pH 7.0, 50 mM NaCl in a total volume of 50 μl and incubated for 30 min at room temperature. The samples were then centrifuged at 434,500g for 1 h at +20 °C using an Optima TL-10 benchtop ultracentrifuge (Beckman Coulter). The supernatant was transferred to a new tube, and the pellet was resuspended in 50 μl of 20 mM Hepes pH 7.0, and 50 mM NaCl. Both supernatant and pellet samples were mixed with 12.5 μl of 5× SDS-PAGE sample buffer (250 mM Tris pH 6.8, 10% sodium dodecyl sulphate, 50% glycerol, 0.02% Bromophenol Blue and 1.43 M β-mercaptoethanol). The samples were incubated for 5 min at +95 °C, and 10 μl of each sample were analyzed on 4% to 20% SDS-PAGE gels. The protein bands were visualized using PageBlue staining (Thermo Fisher Scientific Inc.), and the gels were imaged using a ChemiDoc XRS + system (Bio-Rad). Protein band intensities were quantified using the ImageJ software (https://imagej.net) (48). For each supernatant and pellet pair, the total intensity of ADF was set to 100%, and the relative amounts of ADF in the pellets were presented as percentages. The assay was repeated three times.

Tryptophan fluorescence assay

Tryptophan fluorescence was used to study the interaction between Plasmodium ADFs and different phosphoinositide vesicles. All tryptophan fluorescence spectroscopy experiments were performed in a quartz cuvette with pathlength of 3 mm, using a Fluoromax-4 spectrofluorometer (Horiba Scientific). The excitation wavelength was fixed to 295 nm, and the emission spectra were collected by averaging 10 spectra from 300 to 450 nm at +25 °C. An aliquot containing 10 μM ADF was titrated with different concentrations of phosphoinositide vesicles from 0 to 1.5 mM. Spectra of buffer and vesicles alone were subtracted from the ADF alone and ADF-phosphoinositide vesicles sample, respectively. The binding of the ADFs to POPC was also measured as a control. Data from three independent experiments were analyzed using nonlinear regression with “one site-specific binding” model in GraphPad Prism 8 (GraphPad Software; www.graphpad.com) using.

$$Y=B_{\max \ast \mathrm{X}/(\text{Kd}+\mathrm{X})}$$

where X is the ligand concentration, Y is the fluorescence intensity, B_max is the maximum specific binding, and K_d is the equilibrium dissociation constant.

Synchrotron radiation CD

Secondary structure compositions of ADFs in the presence and absence of lipids were determined using SRCD. Lipid vesicles were mixed with protein at a 1:100 protein-to-lipid (P:L) molar ratio prior to the measurement. Protein samples at 0.12 mg/ml were recorded three times in water between 170 to 280 nm in a Hellma cylindrical absorption cuvette (Suprasil quartz, Hellma GmbH & Co. KG) with a path length of 0.5 to 1 mm at the AU-CD beamline at the ASTRID2 synchrotron (ISA) at +25 °C. Buffer spectra were subtracted, and CD units converted to Δε (M^-1 cm^-1) using CDtoolX (49). Secondary structure deconvolutions were done using beta structure selection (31) or DICHROWEB (50).

Sequence alignment of Plasmodium ADFs with other ADF/cofilin proteins

A multiple sequence alignment of Plasmodium ADFs and other ADF/cofilin proteins was generated with ClustalW2 (51) and visualized using ESPript (52). The UniProtKB accession numbers are as follows: AtADF1, Acanthamoeba thaliana ADF1 (Q39250); AcAct, Acanthamoeba castellanii actophorin (P37167); ScCof, S. cerevisiae cofilin (Q03048); MmCof, Mus musculus cofilin-1 (P18760); HsCof, Homo sapiens cofilin (P23528); PfADF1, P. falciparum ADF1 (Q8I467); PbADF2, P. berghei ADF2 (Q3YPH0); and TgADF, T. gondii ADF (B9Q2C8).

Protein–membrane interaction study

The orientation of the Plasmodium ADFs [PDB IDs 2XF1 and 2XFA (29)] on the plasma membrane was estimated using the PPM server version 3.0 (40). A mammalian plasma membrane was used for the calculations, and heteroatoms, water, and detergents were excluded from the calculations. Both Plasmodium ADFs were treated as peripheral proteins by the PPM program.

Data availability

All the plasmids and relevant data used to support the findings of this study are available upon request from the corresponding authors.

Supporting information

This article contains supporting information.

Conflict of interest

The authors declare that they have no conflicts of interest with the contents of this article.

Reviewed by members of the JBC Editorial Board. Edited by Enrique De La Cruz