Introduction

Malaria is a major global health problem with > 240 million cases and > 600,000 deaths reported annually [1]. Despite initial gains in reducing malaria in the early 2000s, progress has stalled since 2015 and the burden is now increasing. Furthermore, the impacts of climate change are expected to substantially exacerbate malaria in coming years [2]. As such, new interventions including novel vaccines and therapeutics are urgently needed. While Plasmodium falciparum accounts for the majority of malaria-induced morbidity and mortality worldwide [1], around 3 billion people are also at risk of P. vivax malaria globally and there is a growing recognition that it is also a major cause of severe malaria and chronic illness [3,4,5,6]. Even in the absence of exposure to infectious mosquitoes, dormant liver stage hypnozoites can release infectious P. vivax merozoites (which infect erythrocytes) into the bloodstream resulting in repeated relapses over time and chronic illness. P. knowlesi is closely related to P. vivax and a significant cause of human malaria in Southeast Asia, occurring as a zoonotic infection from macaque monkeys [7]. The leading vaccine against P. falciparum, RTS,S vaccine, has modest efficacy (30–50%) against P. falciparum only [8], and progress on P. vivax vaccines has been very limited, with extremely few candidates in development or clinical trials [9, 10]. The development of vaccines or novel therapeutics that protect against the blood stage of P. falciparum and P. vivax is a priority because this is the life stage during which clinical disease occurs. Targeting merozoites, which infect erythrocytes, prevents blood-stage replication and disease, and specifically targeting merozoites would prevent relapses of P. vivax [11].

P. falciparum and P. vivax are thought to have independently evolved from primate to human hosts [12, 13] and have adopted alternative erythrocyte invasion mechanisms, ligands and receptors [10, 11]. Because P. knowlesi is phylogenetically closely related to P. vivax, it is a valuable model of P. vivax, a species that cannot be maintained in long-term culture [14]. Apical membrane antigen 1 (AMA1) is one of few invasion proteins shared between P. vivax and P. falciparum and has been shown to play an essential role in infection of host erythrocytes by merozoites of P. falciparum [15]. AMA1 is also a prominent target of acquired immunity and has been a leading vaccine candidate [11, 16,17,18,19]. Furthermore, AMA1 is present in all mammalian and avian Plasmodium species, even though most other merozoite invasion ligands and surface proteins are different between species [11]. The three-domain structure of AMA1 is conserved among species, but the amino acid sequence varies [20,21,22,23]. Immediately before merozoite invasion, AMA1 binds to rhoptry neck protein 2 (RON2), an event that initiates formation of the tight junction, essential for host cell invasion [23, 24].

RON2 is a parasite-derived protein that is released by merozoites, and it translocates into the host erythrocyte membrane (forming a complex with RON4, 5 and 8) [25, 26]. A short β-hairpin loop in the C-terminal domain of RON2 is exposed on the erythrocyte surface and interacts with the hydrophobic groove of AMA1 [23, 27]. This molecular event has been primarily established for P. falciparum, but less clearly for P. vivax. Synthetic peptides corresponding to this loop of PfRON2 bind to PfAMA1 and inhibit P. falciparum invasion, leading researchers to conclude that this interaction is essential for invasion [15, 23, 26,27,28,29]. Agents that inhibit the AMA1–RON2 interaction in P. falciparum effectively disrupt erythrocyte invasion [30], making it an attractive target for prophylactic and therapeutic interventions.

We recently reported that PvAMA1 can functionally complement PfAMA1 in P. falciparum parasites. P. falciparum parasites with genetic replacement of PfAMA1 with PvAMA1 had no apparent changes in erythrocyte invasion dynamics or replication rate [31]. This observation indicated conservation of function between PfAMA1 and PvAMA1, and raised the question of whether PvAMA1 might have the ability to bind PfRON2 protein during erythrocyte invasion since AMA1-RON2 binding was considered essential for invasion, or that there may be additional AMA1 binding interactions during invasion. The establishment of P. falciparum parasites expressing PvAMA1 and techniques to mutate PfAMA1 and PvAMA1 sequences [31, 32] provided a powerful platform to investigate the function of AMA1 and RON2 in invasion. Furthermore, this model provides a rare opportunity to use transgenic approaches to introduce targeted mutations to understand the function of an invasion ligand of P. vivax. Currently, knowledge of specific differences or common features in the function and interactions of P. falciparum and P. vivax invasion ligands is extremely limited, even for leading vaccine candidates where there are often major differences between these two most important malaria species.

In this study, we investigated the functional conservation and divergence of AMA1 and RON2 with P. falciparum, P. vivax and P. knowlesi to gain insights for vaccine and therapeutic development for these three major causes of human malaria. P. vivax cannot be readily propagated or genetically manipulated in vitro; therefore, we used a chimeric approach of expressing PvAMA1 in P. falciparum parasites, which we have previously validated [31]. P. knowlesi is closely related to P. vivax and is also used as a model of P. vivax [33]. We used complementary approaches to advance our understanding of AMA1 in erythrocyte invasion by P. falciparum, P. vivax and P. knowlesi including, (i) evaluating invasion-inhibitory RON2 peptides with P. falciparum, P. vivax and P. knowlesi parasites, (ii) generating transgenic parasites with mutation of the RON2-binding regions (Loop1E) of AMA1 and evaluating their invasion function and (iii) evaluating the impact of mutations in AMA1 on susceptibility to invasion-inhibitory antibodies.

Methods

Ethics statement

For generating antibodies, animal ethics approval was provided by the Animal Ethics Committee of the Walter and Eliza Hall Institute (Australia), or the SVL Animal Care and Use Committee (for Quadvax, Fig. 7). No human samples or data were used.

Role of funding sources

Funders had no role in study design; collection, analysis and interpretation of data; writing of manuscript; and in the decision to submit the paper for publication.

Sequence analysis

Amino acid sequences were aligned in Geneious (https://www.geneious.com), and a distance matrix constructed in R using DECIPHER [34]. A neighbour-joining dendrogram was created using default settings, and a heatmap constructed with SPIDER [35]. The dendrogram was rendered using FigTree (http://tree.bio.ed.ac.uk/software/figtree), and images were further edited with Inkscape (http://www.inkscape.org).

Parasite culture and genotyping

P. falciparum asexual stage parasites were maintained in culture in human erythrocytes (blood group type O +) at a haematocrit of 4% in RPMI-HEPES supplemented with 0.25% (w/v) Albumax™ (Invitrogen) and 5% (v/v) heat inactivated human serum. P. falciparum were synchronised using sorbitol and heparin treatments as described previously [36, 37]. P. knowlesi [14] was maintained in RPMI-HEPES with 0.5% v/v Albumax II (Gibco) and synchronised using Percoll. Transgenic W2mef P. falciparum strains expressing PfAMA1 W2mef allele (W2-W2), PfAMA1 3D7 allele (W2-3D7) and PvAMA1 Palo Alto allele (W2-PvAMA1) were generated in previous studies [31, 32, 38].

Expression of AMA1 proteins in HEK293F cells

Sequence selection and modification

The protein sequences for each of the antigens were selected: PfAMA1 (XP_001348015.1; 3D7 genotype), PvAMA1 (ACB42433.1; Palo Alto genotype) and PkAMA1 (XP_002259339.1; Strain H) (Supplementary data Table S1). At the N-terminus of the sequences, the native signal peptide was removed and replaced with a signal peptide for tissue plasminogen activator, followed by a 6-histidine tag. At the C-terminus of all sequences, the sequence was truncated to remove the transmembrane domain and cytoplasmic tail. All sequences were assessed for potential glycosylation sites (http://www.cbs.dtu.dk/services/NetNGlyc/). The AMA1 sequences for each species were then modified to prevent potential glycosylation (PfAMA1, six changes; PvAMA1, three changes; PkAMA1, seven changes) (Supplementary data Table S2).

Generation of protein expression vectors

Each AMA1 protein sequence was used to generate DNA sequences that were codon-optimised for mammalian expression and then synthesised (GeneArt). The final protein sequences are provided (Figure S7). Synthetic genes were supplied in a pUC vector, and then cloned into pcDNA 3.1(+) using Xho1 and BamH1 restriction sites. The final plasmids were quantified and used to transfect HEK293F cells.

HEK293F cell culture and transfection

HEK293 Freestyle™ cells (Thermo Fisher Scientific) were cultured following the manufacturer’s protocols and used for expression of recombinant proteins as described [39, 40]. In brief, cells were cultured in Erlenmeyer shaker flasks (125 mL, Corning) with FreeStyle™ 293 Expression Medium (Thermo Fisher Scientific) at 37 °C, 8% CO2 at 135 rpm on an orbital shaker. Cells were counted using the trypan blue (0.4%; Thermo Fisher Scientific) cell exclusion method using Countess™ Cell Counting Chamber Slides (Thermo Fisher Scientific) and the Countess™ automated cell counter (Thermo Fisher Scientific). HEK293F cells were transfected for protein expression following the manufacturer’s protocol (Thermo Fisher Scientific) with minor alterations. On the day of transfection, cells were centrifuged (700 g at 4 °C, 10 min) and resuspended in HEK293F expression media with 1:100 antibiotic/anti-mycotic solution (Thermo Fisher Scientific) to a final density of 1 × 106 cells/mL. For a 30 mL transfection, 90 µL of polyethylenimine transfection reagent (25 kDa linear; Polysciences; stock 1 mg/mL) was added to 0.6 mL of OptiPro™ Serum Free Medium (Thermo Fisher Scientific) and incubated for 5 min. This solution was then added to the DNA solution (30 µg of purified plasmids and 0.6 mL of OptiPro™ Serum Free Medium (Thermo Fisher Scientific)). After 10 min incubation at room temperature, this final solution was added to the cells, and then returned to the orbital shaker and incubator. The next day, lupin (1:40 of 20% w/v, Biotech Solabia) and pluronic acid F-68 (1:100 of 10% w/v, Thermo Fisher Scientific) were added. The expressed proteins were harvested 6 days post-transfection by centrifuging cells (700 g, 10 min) to collect supernatant that was then filtered (0.2 µm membrane) and stored at 4 °C until purification. Protein expression was tested using SDS-PAGE gels and western blot.

Protein purification and dialysis

The harvested media containing the expressed proteins was passed over nickel resin columns (Life Technologies), washed with 20 mM imidazole/phosphate-buffered saline (PBS) (Sigma-Aldrich), and bound proteins eluted in several 1 mL fractions in 500 mM imidazole/PBS. Eluted fractions were tested for the presence of protein by spectroscopy and SDS-PAGE. Fractions containing protein were pooled, filter-sterilised, dialysed into sterile PBS, and adjusted to a concentration of 1 mg/mL by centrifugation using 10 000 MW cut-off filters.

SDS-PAGE, colloidal staining and western blot analysis of recombinant proteins

The folding and purity of expressed proteins were tested by colloidal blue staining and western blot. For colloidal blue stained gels, 1–2 µg of protein was run on an SDS-PAGE gel under either native or denaturing conditions (+ TCEP, to break conformation-specific disulfide bonds), and then stained with the colloidal Coomassie stain (GelCode Blue Stain Reagent, Thermo Fisher Scientific). For the western blot, 50 ng samples were run on an SDS-PAGE gel under either native or denaturing conditions (+ TCEP, to break conformation-specific disulfide bonds) and then transferred to nitrocellulose membrane (iBlot Transfer Stack nitrocellulose, Thermo Fisher Scientific). Membranes were then probed with antigen-specific antibodies: 4E11 mAb against PfAMA1 and rabbit polyclonal antibodies against PvAMA1.

Peptides

We used two sets of RON2 peptides: (i) peptides of 39 amino acids, based on the PfRON2sp1 peptide originally described by [22]: PfRON2 (Asp-2021 to Ser-2059), PvRON2 (Asp-2050 to Thr2088) and PkRON2 (Asp-1975 to Thr-2013); (ii) peptides of 48 amino acids, based on the PfRON2L peptide described by [29] (Asp-2021 to Lys-2067) but with the addition of an N-terminal Met-2020: PfRON2L (Met-2020 to Lys-2067) and PvRON2L (Met-2049 to Lys-2096). All RON2 peptides were synthesised with cyclised cysteine residues to create the RON2 loop. We found that the shorter peptides were water soluble and more suitable for testing in GIA. The longer peptides based on the PfRON2L peptide were used to dissect the functional roles of the N- and C-terminal amino acids flanking the β-hairpin loop core. However, these peptides needed to be solubilised in DMSO, making them less suitable for use in invasion and growth inhibition assays. As a negative control we used a small extracellular sequence of PfRON3 protein (Glu-574 to Lys-603) (ERYGVLKQCPLDIVKNLNQQCEYVSFEIKK), synthesised as a peptide of 30 amino acids with cyclised cysteine residues. All peptides were synthesised by LifeTien (Somerset, NJ, USA).

Antibodies and RON2L binding assays

The generation of rabbit antibodies against recombinant ectodomain PfAMA1 (3D7 allele) and PvAMA1 (Palo Alto allele) proteins has been described previously [32, 41]. mAbs were isolated from mice vaccinated with PfAMA1, as reported previously [41]. Standard ELISA was used to confirm antibody specificity using established methods [16, 42]. RON2L binding to recombinant AMA1 proteins was measured by a plate-based assay. Briefly, recombinant AMA1 proteins or the control Pfs230 protein were coated onto 96-well ELISA plates (Maxisorb) at a concentration of 10 μg/mL, overnight at 4ºC. Plates were washed and incubated with 10 μg/mL biotin-tagged PvRON2L peptide at 37ºC for 1 h. Plates were then washed and incubated with 2 μg/mL streptavidin-conjugated HRP (Millipore) for 1 h at 37 °C. After washing, plates were developed by the addition of ABTS.

Generation of transgenic parasite lines

Genes encoding PfAMA1 (3D7 allele) and PvAMA1 (Palo Alto allele) with mutated RON2-binding residues (AMA1-M) were generated by Spliced by Overlapping End (SOE) PCR, using a previously described codon-optimised PfAMA1 and PvAMA1 DNA as templates and synthetic oligonucleotides incorporating relevant DNA mutations (Geneworks Australia) [31, 32, 38]. All mutated PfAMA1 and PvAMA1 alleles were inserted into the plasmid vector pCC1AMA1TP.1 and transfected into W2mef parental P. falciparum as previously described [31, 32, 38]. P. falciparum populations confirmed to have plasmid integration into the AMA1 target after positive selection with WR99210 were cloned by limiting dilution, and positive populations were confirmed by integration-specific PCR.

Live-cell microscopy

Highly synchronous parasite cultures at 4% haematocrit were diluted 1/25 in media and allowed to settle to produce a monolayer of cells in a 35 mm Fluorodish (World Precision Instruments). All live-cell experiments were performed at 37 °C on a Zeiss AxioObserver Z1 fluorescence microscope equipped with humidified gas chamber (94% N2, 1% O2 and 5% CO2) [24]. Time-lapse videos were recorded at four frames per second with a high-resolution AxioCam MRm camera. ImageJ and Prism (GraphPad) were used to perform image and statistical analyses. For data sets with normal distribution an unpaired t test was used. For data sets without normal distribution the Mann–Whitney U test was used. For comparison of deformation scores between groups a Chi-square analysis was performed. A value of p ≤ 0.05 was used as the determinant of statistical significance for all tests.

Quantifying invasion inhibition

Growth-inhibition assays were performed over two cycles of replication and measured by flow cytometry as described in detail elsewhere [31, 32, 43, 44]. Synchronised early-trophozoite stage P. falciparum were grown with rabbit or human IgG at a concentration specified in each assay. PfRON2 and PvRON2 peptides were solubilised in PBS and tested against PBS controls. PfRON2L, PvRON2L, PfPvRON2L and PvPfRON2L were solubilised in neat DMSO, then diluted to a final concentration of 2 mg/mL peptide in 10% DMSO in PBS; in these experiments an equal volume of 10% DMSO in PBS was used as the control. After two invasion cycles, early-trophozoite stage P. falciparum were stained, and parasitaemia measured by flow cytometry. Each sample was run in either duplicate or triplicate in at least two separate assays for each AMA1 transfectant line. Percentage invasion inhibition was calculated as:

 = (1-(parasitaemia in test well ÷ mean parasitaemia in non-inhibitory control wells)) × 100.

Non-immunised rabbit IgG or KLH-immunised rabbit IgG were tested as controls in invasion inhibition assays when evaluating rabbit IgG. PfRON3 peptides were used as controls when evaluating testing peptides. Non-immunised mouse IgG was used as controls in assays evaluating mouse monoclonal antibodies. RON2 peptides and antibodies to AMA1 have been previously shown to specifically inhibit merozoites invasion of erythrocytes [24, 37]; therefore, inhibition of growth by RON2 peptides and anti-AMA1 antibodies was interpreted as inhibition of invasion.

Polyacrylamide gel electrophoresis (SDS-PAGE) and western blot analysis

SDS-PAGE and immunoblot analysis of parasite proteins was performed as previously described [31]. Synchronised schizont-stage parasite cultures were lysed with saponin, washed in PBS and resuspended in non-reducing SDS sample buffer (Invitrogen). Samples were sonicated and heated to 100 °C for 5 min prior to SDS-PAGE. Proteins were separated on 3–8% tris–acetate gels (Invitrogen) and transferred onto nitrocellulose using the iBlot system (Invitrogen) according to standard protocols. Blots were probed with anti-3D7, anti-PvAMA1 [31] or QuadVax rabbit IgG (20 μg/mL IgG) or 2.5 μg/mL 1F9, 4E11, 4E8, 1B10, 5A6 or 1E10 monoclonal antibodies [41]. Blots were also probed with a mouse monoclonal antibody generated against the P. falciparum HSP-70 protein (1:2000) as a loading control. Horseradish peroxidase-coupled (HRP) goat anti-rabbit Ig (1:2000) or sheep anti-mouse Ig (1:2000; Millipore) were used as secondary antibodies.

Parasite multiplication assays

The multiplication rate of transgenic parasites expressing either wild-type (WT) 3D7 AMA1 (W2-3D7) or mutated forms of 3D7 AMA1 (W2-3D7 PvLoop1b + 1E and W2-3D7 PvLoop1E) was calculated by flow cytometry over a single invasion cycle (n = 5 wells per line) [45]. Briefly, synchronised ring stage parasites were adjusted to 0.5% parasitaemia and 4% haematocrit and aliquoted into three replicate 96-well plates termed Cycle 0, Cycle 1 Static and Cycle 1 Suspension (n = 5 100 μL wells per plate, per parasite line). Cycle 0 (C = 0) parasites were allowed to develop into trophozoites for 12 h at 37 °C. Cycle 1 (C = 1) parasites were allowed to develop through a single cycle of erythrocyte invasion for 60 h at 37 °C under either static culture conditions or with gentle agitation to maintain erythrocytes in suspension. C = 0, C = 1 Static and C = 1 Suspension trophozoite-stage parasites were fixed for 1 h at room temperature by the addition of glutaraldehyde (ProSciTech) to a final concentration of 0.25% (v/v). After fixation, parasites were washed in HTPBS, stained with 10 × SYBR Green dye (Invitrogen) and 5 × 105 erythrocytes counted per well using a BD FACSCantoII flow cytometer. FACS counts were analysed using FlowJo™ (V6.4.7) software (Treestar). Parasite multiplication was calculated by dividing C = 1 parasitaemia by C = 0 parasitaemia for each parasite line.

Statistics

For continuous data, p-values were determined for comparisons between two groups using Wilcoxon’s rank test, or Wilcoxon’s signed-rank test for paired data. Analyses were conducted using GraphPad Prism.

Results

Divergence in the RON2 sequence of P. falciparum from P. vivax and P. knowlesi

To understand divergence of RON2 and AMA1 between species, we analysed sequence similarity between P. falciparum, P. vivax and P. knowlesi, as well as other representative Plasmodium spp. from the Laverania and Vinckeia subgenera (Fig. 1; Supplementary Figures S1 and S2). Neighbour-joining dendrograms for the RON2-loop sequence and AMA1 ectodomain show that P. vivax and P. knowlesi group together, whereas P. falciparum grouped separately with P. reichenowi (Fig. 1). Distance matrices support this grouping. Notably, sequences from P. vivax and P. knowlesi are separated by a very low distance (0.10 and 0.15 for RON2-loop and AMA1, respectively), whereas the distance between P. falciparum and P. vivax or P. knowlesi was substantially higher (0.46–0.5 for RON2-loop and 0.42 for AMA1).

Fig. 1
figure 1

Similarities and differences in RON2 and AMA1 amino acid sequences between Plasmodium spp. Neighbour-joining dendrograms (left) showing similarities between the amino acid sequences of Plasmodium spp. For RON2L (A) and the AMA1 ectodomain (B). In both cases, species group as expected by subgenera; the rodent Vinckeia species group together, as do species of the Plasmodium subgenus (including P. vivax and P. knowlesi), and the Laverania subgenus (P. falciparum and P. reichenowi). The distance matrices (right) also support this grouping. Notably, sequences from P. vivax and P. knowlesi are separated by a very low distance, with 0.10 and 0.15 for RON2L and the AMA1 ectodomain, respectively

Full size image

Conservation of AMA1–RON2 binding between P. vivax and P. knowlesi, but divergence from P. falciparum

We investigated whether there is conservation or species specificity in the RON2-loop for binding to AMA1. A variety of PfRON2 peptides have previously been described. The PfRON2L peptide of 47 amino acids [29] spans Asp-2021 to Lys-2067. The PfRON2sp1 peptide of 39 amino acids, originally reported by [23], spans Asp-2021 to Ser-2059. A peptide of 39 amino acids, also called PfRON2sp1, was subsequently described by [22] and spans Met-2020 to Leu-2058. The sequences of the PfRON2, PvRON2 and PkRON2 C-terminal regions have in common two centrally located cysteine residues that enable the formation of the disulfide-bonded β-hairpin loop structure [23] (Fig. 2A). In PfRON2, Arg-2041 located at the tip of the β-hairpin loop forms a critical interaction with the hydrophobic cleft of PfAMA1. In PvRON2 and PkRON2, the corresponding amino acid is substituted for Thr-2055 [22].

Fig. 2
figure 2

Conserved and divergent functions of RON2-loop among P. falciparum, P. vivax, and P. knowlesi. A Alignment of the RON2L peptide sequences from P. falciparum (Pf), P. vivax (Pv) and P. knowlesi (Pk) orthologues. Amino acids conserved between PvRON2 and PkRON2 are in red. Amino acids that form the disulfide-bonded β-hairpin loop are shaded in orange B Invasion inhibition of transgenic P. falciparum expressing PfAMA1 (W2-3D7) or PvAMA1 (W2-PvAMA1) or P. knowlesi (strain H) by PfRON2, PvRON2, PkRON2 and the control PfRON3 peptides. C Binding of PvRON2L peptide to recombinant PfAMA1, PvAMA1, PkAMA1 or Pfs230 (negative control). Data from 2 or more experiments in duplicate and show mean ± SD

Full size image

In this study we used two sets of RON2 peptides: (1) three peptides of 39 amino acids: Asp-2021 to Ser-2059 of PfRON2, Asp-2050 to Thr2088 of PvRON2 and Asp-1975 to Thr-2013 of PkRON2 (these are based on the PfRON2sp1 peptide originally described by [22]) and are termed RON2 peptides in this study. (2) two peptides of 48 amino acids: Met-2020 to Lys-2067 of PfRON2L and Met-2049 to Lys-2096 of PvRON2L. These are based on the PfRON2L peptide described by [29] (Asp-2021 to Lys-2067) but with the addition of an N-terminal Met-2020, and are termed RON2L peptides in this study. The shorter RON2 peptides were generally used as inhibitors in assays and had good solubility in aqueous solutions. The longer peptides based on the PfRON2L peptide were used to dissect the functional roles of the N- and C-terminal amino acids flanking the β-hairpin loop core; they had lower solubility in aqueous solutions and needed to be solubilised in 10% DMSO.

The short RON2 peptides of P. falciparum, P. vivax and P. knowlesi (Fig. 2A) were tested for inhibition of invasion against established transgenic P. falciparum parasites expressing either P. falciparum AMA1 3D7 strain (denoted W2-3D7) and P. vivax AMA1 Palo Alto strain (denoted W2-PvAMA1) [31], and against wild-type P. knowlesi strain H. The W2-PvAMA1 parasites retain the endogenous PfRON2 protein and have an invasion rate and kinetics that did not differ from P. falciparum parasites expressing PfAMA1 [31]. As a negative control peptide, we used a PfRON3 cyclised peptide (Glu-574 to Lys-603).

PfRON2 peptide strongly inhibited the invasion of transgenic W2-3D7 parasites expressing PfAMA1 parasites (Fig. 2B). In contrast, PvRON2 and PkRON2 peptides had little or no inhibition of parasites expressing PfAMA1. PvRON2 and PkRON2 peptides both effectively inhibited parasites expressing PvAMA1 (W2-PvAMA1 parasites), whereas PfRON2 showed only minor inhibition at the highest concentration tested (200 µg/mL), but no inhibition at lower concentrations. Similarly, both PvRON2 and PkRON2 peptides effectively inhibited P. knowlesi parasites, whereas PfRON2 gave no significant inhibition. No inhibition was observed for the PfRON3 control peptide when tested against any of the parasite strains, supporting specificity of the assays (Fig. 2B). These findings led us to conclude that while the pairing of PfAMA1 with PfRON2 is highly species-specific, there is a level of functional conservation in the AMA1–RON2 pairing between P. vivax and P. knowlesi.

Direct binding of PvRON2L peptide to recombinant AMA1 corresponding to the three species was studied to provide supporting evidence for conclusions made from invasion inhibition studies (ectodomains of PfAMA1 (3D7 allele), PvAMA1 (Palo Alto allele) and PkAMA1 (strain H allele) were expressed; Supplementary figure S3). When tested in a plate-based binding assay, PvRON2L demonstrated high binding to PvAMA1 and higher binding to PkAMA1 (Fig. 2C). However, there was little binding to PfAMA1, and no binding to the Pfs230 negative control protein. This pattern of binding is consistent with invasion inhibition data indicating cross-inhibition of P. vivax and P. knowlesi by PvRON2 and PkRON2 peptides, but lack of inhibition of P. falciparum by these same peptides (Fig. 2B).

Sequences flanking the β-hairpin loop of RON2L are not required for species-specific AMA1 binding

To evaluate whether regions adjacent to the RON2 β-hairpin loop influence AMA1 binding, we tested the longer PfRON2L (Met-2020 to Lys-2067) and PvRON2L (Met-2049 to Lys-2096) peptides. Additionally, we synthesised hybrid RON2L peptides in which the N- and C-terminal amino acids (either side of the β-hairpin loop) were replaced with those from P. vivax. The hybrid peptide PvPfRON2L contains the central β-hairpin loop of PfRON2, flanked by the N- and C-terminal amino acids of PvRON2L. The PfPvRON2L hybrid peptide contained the central β-hairpin loop of PvRON2, flanked by the N- and C-terminal amino acids of PfRON2L. These peptides required solubilisation in 10% DMSO; therefore, relative inhibitory activity of peptides was measured against DMSO controls of the same concentration.

At 200 μg/mL, both the longer PfRON2L and hybrid PvPfRON2L (sequences shown in Fig. 3A) effectively inhibited the invasion of P. falciparum expressing PfAMA1 (W2-3D7). In contrast there was little inhibition of parasites expressing PvAMA1 (Fig. 3B,C). Titrating the PfRON2L and hybrid PvPfRON2L peptides against parasites expressing PfAMA1 (W2-3D7) showed that the PvPfRON2L hybrid peptide had lower inhibitory activity compared to the PfRON2L peptide (Fig. 3D). Therefore, the central β-hairpin loop appears to mediate species-specific AMA1 binding by PfRON2L peptides, and the N- and C-terminal amino acids appear to play a role in enhancing binding to AMA1. This is consistent with a previous study reporting that a longer PfRON2L peptide (amino acids 2021–2059) had higher affinity binding to recombinant PfAMA1 than a shorter PfRON2 peptide (amino acids 2027–2055) [23, 29]. Crystal structures of AMA1-RON2L complexes identified amino acids either side of the central β-hairpin loop that make contact with AMA1 and these amino acids differ between PfRON2L and PvRON2L [22]. Therefore, these additional contact points would be absent with the PvPfRON2L peptide interaction with PfAMA1 and likely explains their lower inhibitory effect against parasites expressing PfAMA1. The longer PvRON2L peptide (including the additional N- and C-terminal amino acids) was poorly soluble, even in DMSO. However, the hybrid PfPvRON2L peptide (Fig. 3E) could be evaluated and was found to strongly inhibit invasion of parasites expressing PvAMA1 (W2-PvAMA1), but did not inhibit invasion of P. falciparum expressing PfAMA1 (W2-3D7) (Fig. 3F). This further supports the importance of the central β-hairpin loop for defining species-specific AMA1-binding activity.

Fig. 3
figure 3

The central β-hairpin loop of RON2L peptides mediate species-specific inhibition of invasion. A Alignment of P. falciparum (Pf) RON2L and the chimeric P. vivax flank/P. falciparum central β-hairpin loop hybrid RON2L (PvPfRON2L) peptides. Amino acids that differ between PfRON2L and PvRON2L are in red. Amino acids that form the β-hairpin loop are shaded in orange. B Invasion inhibition of P. falciparum expressing PfAMA1 (W2-3D7) or PvAMA1 (W2-PvAMA1) by 200 μg/mL PfRON2L peptides. C Inhibition of transgenic P. falciparum expressing PfAMA1 (W2-3D7) or PvAMA1 (W2-PvAMA1) by 200 μg/mL PvPfRON2L. D Inhibition of transgenic P. falciparum expressing PfAMA1 (W2-3D7) by PfRON2L and PvPfRON2L peptides. E Alignment of PvRON2L (PvRON2L) and the chimeric P. falciparum flank/P. vivax central β-hairpin loop hybrid RON2L (PfPvRON2L) peptides. Conserved amino acids are in red. Amino acids that form the β-hairpin loop are shaded in orange. F Inhibition of transgenic P. falciparum expressing PfAMA1 (W2-3D7) or PvAMA1 (W2-PvAMA1) by 200 μg/mL PfPvRON2L

Full size image

Identification of key amino acids in PvAMA1 Loop1E for species-specific RON2 binding

To further understand the function of AMA1–RON2 binding in erythrocyte invasion, we mutated key RON2-binding residues in PvAMA1 expressed in our transgenic P. falciparum model. Previous studies suggest that the PfRON2L peptide interacts with eight key amino acids of PfAMA1 [23]. Of these, amino acids Phe-183 and Tyr-251 are conserved across PfAMA1, PvAMA1 and PkAMA1 (Fig. 4A). The other six are located adjacent to, or form part of, the N-terminal face of Loop1E of PfAMA1. PvAMA1 differs from PfAMA1 at four out of six amino acids in Loop1E, whereas PkAMA1 differs from PvAMA1 for only one of the six amino acids. A high level of sequence conservation across these six amino acids of AMA1 Loop1E is observed between P. vivax, P. ovale, P. malariae, P. knowlesi, P. coatneyi and P. cynomolgi (Supplementary figure S4). We mutated the Loop1E sequence of PvAMA1, replacing it with the sequence from PfAMA1 (3D7 allele). To be conservative, we did not mutate the entire PvAMA1-Loop1E sequence; rather only seven amino acids in the N-terminal face were replaced (equivalent to Gly-222 to Asn-228 in PfAMA1)(Fig. 4B). While Ser-232 and Tyr-234 of PfAMA1 have also been reported to form part of the binding pocket, they are conserved between PvAMA1 and PfAMA1 and so were not changed in PvAMA1.

Fig. 4
figure 4

Differences between Plasmodium species in AMA1 amino acids that interact with RON2-loop, and generation of a hybrid PvAMA1-PfLoop1E mutant parasite line. A Key P. falciparum (Pf) AMA1 amino acid residues that have been identified to play a role in binding to PfRON2 aligned against those found at the same relative position in PvAMA1 and PkAMA1 (residue numbers are PfAMA1 specific). Amino acids conserved across all three species are in red; except amino acid 232, which is the same in PfAMA1 and PvAMA1, but differs in PkAMA1. Amino acids conserved between PvAMA1 and PkAMA1 only are in green. Amino acids located in the Loop1E domain (222–234) are shaded in orange. B Alignment of PfAMA1 and PvAMA1 Loop1E amino acid sequences. Proposed PfRON2 interacting residues are in red. PvAMA1 (Palo Alto) residues that differ from PfAMA1 (3D7) are shown in magenta and an alternative amino acid present in other PvAMA1 alleles is shown in orange. The hybrid PvAMA1 incorporating PfAMA1 Loop1E is shown, with PfAMA1-specific amino acid residues in blue. C Transgenic P. falciparum expressing a chimeric PvAMA1 with the PfAMA1-Loop1E sequence (denoted PvAMA1PfLoop1E) was generated (see Supplementary Material, Figure S5). Analysis of PvAMA1PfLoop1E protein expression was investigated by western blot. Schizont protein extracts were prepared from transgenic P. falciparum lines expressing either PfAMA1 (W2-3D7), PvAMA1 (W2-PvAMA1) or PvAMA1PfLoop1E (W2-PvAMA1PfLoop1E) and run in non-reducing conditions (3–8% acrylamide gradient gel). Western blots were probed with anti-3D7 PfAMA1 (3D7 allele) or anti-PvAMA1 polyclonal antibodies (20 μg/mL IgG) raised in rabbits. Antibodies labelled the expected ~ 83 kDa and ~ 66 kDa species of PfAMA1 protein, whereas a smaller ~ 81 kDa band is seen in transgenic parasites expressing PvAMA1 and PvAMA1PfLoop1E mutant proteins. Equal loading of schizont samples was confirmed by probing blots with anti-PfRON4.2 mAb (5 μg/mL IgG)

Full size image

P. falciparum parasites (W2mef strain) were transfected with a plasmid containing the PvAMA1PfLoop1E gene (Figure S5A). Complete allelic exchange in the cloned parasite line (W2-3D7 PvLoop1E) was confirmed by PCR (Figure S5B). Western blots of schizont protein extracts probed with anti-AMA1 antibodies detected the expected PfAMA1 bands at ~ 83 kDa and ~ 66 kDa processed bands (Fig. 4C), that have been previously described [46]. In contrast, western blots of transgenic P. falciparum W2-PvAMA1 or W2-PvAMA1PfLoop1E schizont protein extracts detected a band with lower molecular weight (predicted to be ~ 81 kDa) and a less abundant band at ~ 63 kDa band, similar to previous findings of labelling PvAMA1 in western blots [31]. The size shift supports the analysis by PCR indicating that replacement of PfAMA1 with PvAMA1 genes was successful (Figure S5B), and is consistent with the predicted molecular weights of the PvAMA1 [31] and PvAMA1PfLoop1E proteins and confirmed the loss of endogenous PfAMA1 protein expression in W2-PvAMA1PfLoop1E parasites (Fig. 4C).

Analysis by live microscopy showed that merozoites of the W2-PvAMA1PfLoop1E parasites invaded erythrocytes in vitro with similar kinetics to PvAMA1 parasites, suggesting the functions of PvAMA1 and PvAMA1PfLoop1E are biologically equivalent. There was no significant difference between the parasite lines for all parameters examined, including the time of first merozoite–erythrocyte contact to the start and end of deformation, or initiation and completion of invasion (Fig. 5A). Mutation of Loop1E of PvAMA1 greatly reduced the ability of PvRON2 peptide or PkRON2 peptide to inhibit invasion compared to P. falciparum expressing unmodified PvAMA1 (Fig. 5B), suggesting binding of PvRON2 and PkRON2 to PvAMA1 is largely ablated by mutation of Loop1E. Interestingly, there was no significant gain in inhibition by the PfRON2 peptide (Fig. 5B). This suggests the cluster of six amino acids in PvAMA1 Loop1E plays a major role in determining species-specific binding to PvRON2 and PkRON2. These findings, in which RON2-loop-binding activity of PvAMA1 has been greatly reduced, suggest that the binding of PvAMA1 to the RON2-loop is not essential for invasion. The ability of PvAMA1 to complement the function of PfAMA1 in P. falciparum invasion is not explained by an ability of PvAMA1 to bind PfRON2.

Fig. 5
figure 5

Parasites expressing PvAMA1 incorporating PfAMA1-Loop1E amino acids invade erythrocytes normally but display reduced inhibition by PvRON2 peptide. A Merozoite invasion kinetics of transgenic P. falciparum (Pf) lines expressing P. vivax (Pv) AMA1 (W2-PvAMA1) or PvAMA1PfLoop1E (W2-PvAMA1PfLoop1E) as measured by live-cell imaging at six consecutive stages of erythrocyte invasion. N = number of events filmed per stage of invasion. Statistical analysis by unpaired t test (invasions per schizont rupture) or Mann–Whitney U test (all other panels). The bar graphs are box-and-whisker type, showing the median (line), inter-quartile range (box) and range (whiskers). B Differential inhibition of transgenic W2-PvAMA1 and W2-PvAMA1PfLoop1E parasite lines by titrated PfRON2, PvRON2, PkRON2 or the negative control PfRON3 peptide. Mean and standard deviation of two assays performed in duplicate wells. Statistical analysis by paired t test for each individual peptide concentration. P-value score for all peptide concentrations ranges defined by the lower bar

Full size image

Mutations in Loop1E of PfAMA1 reduce PfRON2 binding, but do not affect parasite invasion

To further understand the interaction between PfAMA1 and RON2-loop, we generated parasites with mutation of key amino acids in PfAMA1. Seven amino acids of the N-terminal face of Loop1E of PfAMA1 (Gly-222 to Asn-228) were substituted with those from PvAMA1 (Fig. 6A). This was the reciprocal of the mutations generated in PvAMA1-PfLoop1E parasites (Fig. 4B). Parental W2mef P. falciparum was transfected with the plasmid containing the PfAMA1PvLoop1E gene (Figure S6A). Complete allelic exchange in the cloned parasite line, W2-3D7-PvLoop1E, was confirmed by PCR (Figure S6B). Expression of the PfAMA1PvLoop1E mutant protein in transgenic P. falciparum was confirmed by probing schizont protein extracts with polyclonal antibodies and a mAb to PfAMA1 (Fig. 6B).

Fig. 6
figure 6

Parasites expressing PfAMA1 incorporating PvAMA1-Loop1E amino acids replicate in erythrocytes normally but display reduced invasion inhibition by PfRON2 peptide. A Alignment of PfAMA1 and PvAMA1 Loop1E amino acid sequences. PfRON2 interacting residues are in red. PvAMA1 (3D7 strain) residues that differ from PfAMA1 are in magenta, and an alternative amino acid used by other PvAMA1 alleles is in orange. The hybrid PfAMA1 incorporating PvAMA1-Loop1E is shown, with PvAMA1-specific amino acid residues in blue. B Western blots of schizont material expressing PfAMA1 (W2-3D7) or PfAMA1 with PvLoop1E (W2-3D7 PvLoop1E) probed with 2.5 μg/mL 1F9 mAb, 20 μg/mL polyclonal rabbit anti-3D7 AMA1 IgG or 1:500 diluted polyclonal rabbit anti-HSP70 antisera. C P. falciparum multiplication rates were measured for W2-3D7 and W2-3D7 PvLoop1E transgenic P. falciparum. Rates were calculated for parasite replication in static erythrocyte culture and suspended erythrocyte culture conditions. Mean and standard deviation of each condition was tested over five wells. Statistical analysis was done by paired t test. D, E Differential inhibition of parasites expressing PfAMA1 and PfAMA1-PvLoop1E by 200 μg/mL PfRON2 peptide (D) or 200 μg/mL PvRON2 peptide (E). Mean and standard deviation of two assays was performed in triplicate wells. Statistical analysis was by paired t test

Full size image

The insertion of PvLoop1E amino acids into PfAMA1 had no detrimental effect on P. falciparum invasion and growth in vitro. In fact, the multiplication rate of PfAMA1PvLoop1E parasites was slightly higher than that of the control PfAMA1 parasites (15% and 6% greater under static and suspension culture conditions, respectively) (Fig. 6C). Mutation of the amino acids in Loop1E of PfAMA1 greatly reduced the ability of PfRON2 peptide to inhibit invasion (Fig. 6D). However, the insertion of PvAMA1 Loop1E amino acids did not lead to any significant gain in inhibition by PvRON2 peptide (Fig. 6E). This is consistent with results obtained for the PvAMA1 Loop1E mutant line and supports the concept that species-specific binding of the RON2L peptide to AMA1 is mediated by Loop1E, and the RON2L–AMA1 interaction is not essential for invasion.

Changes in invasion-inhibitory activity of monoclonal antibodies against PfAMA1 following mutation of Loop1E

We investigated the invasion-inhibitory activity of mAbs to PfAMA1, including mAbs that have been previously reported to bind PfAMA1-Loop1E (4E11, 4E8 and 1B10) [41], and mAbs to domains 2 and 3 (5A6 and 1E10, respectively) (Fig. 7). The Loop1E mAbs were previously shown to block PfRON2 peptide binding to PfAMA1 [41]. In western blots with schizont protein extracts, Loop1E mAbs labelled PfAMA1 in parasites expressing PfAMA1, but there was a complete loss of reactivity against parasites expressing the PfAMA1PvLoop1E mutant, confirming that these mAbs target Loop1E and that changes in the seven amino acid stretch that differs between PfAMA1 and PvAMA1 is important for mAb epitope recognition. In contrast, polyclonal rabbit IgG raised against PfAMA1 (QuadVax [41]), and mAb 5A6 (domain 2) and mAb 1E10 (domain 3) labelled both AMA1 types. Loop1E binding mAbs significantly inhibited parasites expressing PfAMA1, but did not inhibit PfAMA1PvLoop1E parasites. This confirms that targeted mutation has disrupted the function of PfAMA1 Loop1E, complementing earlier findings evaluating inhibition by RON2 peptides (Fig. 6), and demonstrates that these mutations allow parasites to escape antibody inhibition. mAb 5A6, targeting domain 2, inhibited PfAMA1 and PfAMA1 Loop1E mutant parasite lines at similar levels. Interestingly, mAb 1E10, targeting domain 3, showed greater inhibition of PfAMA1PvLoop1E parasites, suggesting that disruption of AMA1-Loop1E binding to the RON2-loop leads to a greater reliance on domain 3 interactions for invasion.

Fig. 7
figure 7

Inhibitory activity of monoclonal antibodies against PfAMA1 following mutation of Loop1E. A 3D structure of AMA1 [41] with arrows indicating the approximate binding sites of mAbs used in this study (1B10, 4E8, 4E11, 1E10, and 5A6). Other known inhibitory mAbs 1F9 and 4G2 are shown for reference. B Western blots of schizont material from P. falciparum expressing wild-type PfAMA1 (W2-3D7) or PvAMA1 with PvLoop1E (W2-3D7 PvLoop1E) were probed with the Loop1E binding mAbs 4E11, 4E8, 1B10 or the domain 2 binding mAb 5A6 as a loading control. All antibodies were tested at a final concentration of 2.5 µg/mL IgG. C Differential growth inhibition of transgenic W2-3D7 and W2-3D7 PvLoop1E parasite lines by mAbs 4E11, 4E8, 1B10, 5A6, and 1E10 or by PfRON2 peptide. The Y-axis shows % inhibition; mean and standard deviation of two inhibition assays performed in triplicate wells. Statistical analysis was by paired t test; *indicates p < 0.05

Full size image

Discussion

In this study, we investigated interactions between AMA1 and RON2 for erythrocyte invasion by P. falciparum, P. vivax and P. knowlesi to gain new insights to inform the development of novel vaccines and therapeutics against the three species causing malaria. Our findings support several important conclusions, based on studies with different RON2-loop peptides as inhibitors and parasites expressing PfAMA1, PvAMA1 or PkAMA1; mutation studies of key AMA1 residues that bind RON2-binding residues in AMA1; and antibody inhibition assays. A major novel finding was that high-affinity binding of RON2-loop to AMA1 is not essential for invasion, which changes the current model of invasion interactions for AMA1 (Fig. 8). Furthermore, AMA1 can tolerate mutations that enable evasion of inhibitory antibodies or molecules targeting the AMA1–RON2 binding interaction, without impacting on erythrocyte invasion. Together, these findings indicate that targeting of AMA1 with vaccines and therapeutics will need to be broader than just the interaction of AMA1 with the RON2-loop to maximise efficacy and reduce the risk of vaccine escape. An additional new insight is that there is conservation in AMA1–RON2 binding between P. vivax and P. knowlesi, whereas PfAMA1 Loop1E and PfRON2 β-hairpin loop have diverged such that there is little capacity for cross-species binding between this species and P. vivax or P. knowlesi. The minimal β-hairpin loop of RON2 defines species specificity and function, and key residues in Loop1E of AMA1 that are required for binding RON2-loop have diverged between P. falciparum and P. vivax species to result in species specificity of the AMA1–RON2-loop interaction. In contrast, the more closely-related P. vivax and P. knowlesi have not substantially diverged, conserving cross-species binding of the RON2-loop to AMA1. These knowledge of specific interactions, conservation and divergence, will be valuable for informing vaccine design for all three malaria species.

Fig. 8
figure 8

Models of possible AMA1 binding interactions for invasion. AMA1 is present on the merozoite surface and the RON2-loop (labelled as RON2L) is present on the erythrocyte surface. AMA1 is known to bind to the RON2-loop and new data suggest that there is an additional binding interaction of AMA1. In one model (panels A–C), AMA1 binds to Receptor B, which is a proposed molecule on the erythrocyte surface, yet to be identified, that acts a second receptor for AMA1 binding. In the alternative model (panels D–E), the second binding interaction of AMA1 is to a second site on RON2 that is separate to the RON2-loop. A, D. In P. falciparum parasites, PfAMA1 binds to the PfRON2-loop (PfRON2L) and either Receptor B (A) or another region of PfRON2 (D) to mediate invasion. Key amino acids of PfAMA1 Loop1E for binding to RON2 are indicated (G, N, M, and I), as identified in this study. B, E. In genetically modified P. falciparum parasites expressing PvAMA1, PvAMA1 does not bind to the PfRON2-loop, but binds to either Receptor B (B) or another region of PfRON2 (E), allowing invasion to occur. The lack of binding of PvAMA1 to the PfRON2-loop is due to differences in key amino acids of AMA1 Loop1E between P. vivax and P. falciparum (ASFV for PvAMA1 and GNMI for PfAMA1). C, F. In P. vivax and P. knowlesi parasites, PvAMA1 and PkAMA1 can bind to the PvRON2-loop or PkRON2-loop, as well as Receptor B (C) or another region on PvRON2 or PkRON2 (F). There is conservation of AMA1-RON2-loop binding between the two species. The amino acid sequences of the β-hairpin loop of RON2 are identical between P. vivax and P. knowlesi. Of the key RON2-interacting amino acids of AMA1 Loop1E, only one amino acid differs between P. vivax and P. knowlesi. The native conformation and surface orientation of RON2 is not currently fully defined

Full size image

Several lines of evidence support the conclusion that AMA–RON2-loop binding is not essential for invasion. First, replacement of PfAMA1 with PvAMA1 in P. falciparum parasites did not lead to detectable changes in the invasion rate or kinetics in vitro [31]. PfRON2 is still expressed in these chimeric parasites, yet PvAMA1 has little binding to the PfRON2-loop as indicated by little inhibition of parasites expressing PvAMA1 by PfRON2 peptides (Fig. 2B). Supporting this, prior studies using surface plasmon resonance and recombinant proteins found no detectable binding of PfRON2 to PvAMA1 [22]. Furthermore, mutation of the PfAMA1 Loop1E largely ablated binding to the PfRON2-loop, as indicated by the loss of inhibition by PfRON2 peptides (Fig. 5D). However, invasion still proceeded without any obvious defect in invasion kinetics. Antibodies to PvAMA1 inhibited invasion of parasites expressing PvAMA1, and antibodies to domain 2 and 3 of PfAMA1 could inhibit invasion of the PfAMA1 Loop1E mutant parasites. This indicates that AMA1 still plays an important function in invasion in these modified parasites, suggesting that AMA1 has other important interactions or functions during invasion, aside from binding to the RON2-loop (Fig. 8). It was notable that a mAb targeting domain 3 of PfAMA1 showed greater inhibitory activity against parasites with mutated PfAMA1 Loop1E, which largely ablated RON2-loop binding (Fig. 7). This suggests that during invasion there is a second interaction of AMA1, potentially to a novel receptor on erythrocytes or another region of RON2, which is large protein (2189 amino acids). Currently, the native conformation and surface orientation of RON2 is poorly defined. A prior study reported that recombinant PvAMA1 can bind the surface of erythrocytes, although the magnitude of binding was low and receptors have not been identified [47]. Another study reported that domain 3 of PfAMA1 expressed on the surface of Chinese hamster ovarian cells could bind trypsin-treated erythrocytes [48]. However, full-length AMA1 did not bind erythrocytes, and domain 3 did not bind to untreated erythrocytes, only to trypsin-treated erythrocytes. Therefore, additional binding interactions of AMA1 and the potential role of domain 3 in invasion remains unclear and further research is needed. In addition to protein-binding interactions, evidence suggests that AMA1 has other important roles, and it is possible that a protein binding interaction for AMA1 is not essential for invasion. Prior studies with P. falciparum showed that the cytoplasmic tail of PfAMA1 is phosphorylated during invasion and this appears to be important for efficient invasion of erythrocytes [49, 50]. Exposure of P. falciparum merozoites to a low potassium concentration, as found in blood plasma, can provide a signal for the release of PfAMA1 (and other proteins) from apical organelles to the merozoite surface after schizont rupture [51]. Given the importance of AMA1 in invasion and its potential as a vaccine and therapeutic target, further research is needed to fully define the functions of AMA1.

Taken together, our results suggest there is redundancy in AMA1–RON2-loop binding; while it is involved in invasion, it is not essential, and invasion can be mediated via additional AMA1 interactions. Our data support a role for AMA1–RON2-loop binding in invasion by P. vivax and P. knowlesi, even if the interaction is not essential. Pv or PkRON2L peptides could inhibit invasion by P. knowlesi parasites (Fig. 2B). Surprisingly, PvRON2 peptides could inhibit invasion of parasites expressing PvAMA1 even though the interaction it is not essential for invasion. This might be explained by the binding of exogenous PvRON2 peptides resulting in steric hindrance of other interactions or causing conformational changes in PvAMA1 that impact on other interactions. In P. falciparum expressing PvAMA1, it is possible that PfRON2-loop can bind PvAMA1 with low affinity, which may be sufficient for invasion when combined with a second AMA1 binding interaction. When excess PvRON2 peptide is present in invasion assays, high-affinity binding to PvAMA1 may outcompete the low-affinity PfRON2 binding, leading to inhibition of invasion. However, there was little inhibition of invasion of PvAMA1 parasites by PfRON2 peptides, which would be expected if this model was correct, suggesting that there is not significant binding between PvAMA1 and PfRON2.

Our findings have significant implications for the development of vaccines and therapeutics targeting AMA1. First, our findings show that small changes in sequence can mediate escape of inhibitory antibodies and molecules targeting the AMA1–RON2 interaction. Second, our data indicate that the binding of AMA1 to the RON2 β-hairpin loop is not essential for invasion and that other interactions with AMA1 occur during invasion. Therefore, targeting the AMA1-RON2-loop interaction may not be ideal or sufficient to achieve highly efficacious vaccines, or may be at risk of escape, since key amino acid changes can be tolerated. Revealing other molecular interactions and vaccine strategies that additional AMA1 interactions may enable vaccines with higher protective efficacy. To date, vaccines based on PfAMA1 have not shown substantial clinical efficacy or generated potent invasion-inhibitory antibodies in clinical trials [19, 52, 53]. Our data showing greater inhibition by a mAb targeting domain 3 in parasites with mutation of PfAMA1 Loop1E suggest that domain 3 should be a focus of future studies to understand its function and potential to be targeted by vaccines. Further, a prior study found synergistic inhibition of invasion of P. falciparum when combining AMA1 antibodies to domain 3 and domain 2 [41]. Parasite lines generated here will be useful tools to progress these concepts in AMA1 vaccine design for the three species, as well as evaluating immune responses in clinical trials.

Our studies demonstrated divergence of AMA1–RON2-loop binding specificity between P. falciparum and P. vivax using different RON2 peptide inhibitors of invasion and mutation studies of PfAMA1 and PvAMA1 in live parasites. In contrast, AMA1–RON2-loop binding was conserved between P. vivax and P. knowlesi. Prior crystallography studies of the PvAMA1–PvRON2L homocomplex and PfAMA1–PvRON2L heterocomplex identified amino acids involved in AMA1–RON2 binding interactions [22]. Using invasion assays with live parasites, our studies functionally demonstrated that the disulfide-bonded β-hairpin loop of RON2 mediates the species-specific binding to AMA1 Loop1E using genetically modified parasites and peptide inhibitors. We extended this knowledge by identifying specific amino acids that are critical for the species-specific divergence. Prior studies reported that a PvRON2 peptide could bind recombinant PfAMA1 protein with low affinity [22]. However, we found that PvRON2 peptide could not inhibit the invasion of P. falciparum parasites expressing PfAMA1. Structural analyses have found that RON2 binding causes conformational changes in AMA1. In its unbound state the hydrophobic cleft of AMA1 is partially covered by the disordered domain 2 loop. RON2 binding displaces the domain 2 loop [21, 54]. Some anti-PfAMA1 mAbs, such as 1F9 and 4G2 [29, 55, 56], and those used in this study (4E11, 4E8, and 1B10), or anti-PkAMA1 mAb R31C2 [21], are proposed to inhibit invasion by blocking the RON2 β-hairpin loop from binding the hydrophobic cleft of AMA1. In our study, we experimentally confirmed that three mAbs (4E11, 4E8 and 1B10) inhibit invasion by targeting Loop1E (Fig. 7C).

In conclusion, we used novel approaches to investigate AMA1–RON2 interactions in erythrocyte invasion in P. falciparum and P. vivax, the two major causes of malaria, with comparisons to P. knowlesi, a significant cause of disease in Southeast Asia. We defined species-specific AMA1–RON2 binding for invasive function in P. vivax versus P. falciparum, and cross-species conservation between P. vivax and P. knowlesi. Our data indicate substantial molecular flexibility in AMA1 interactions for invasion, which has interesting evolutionary implications, implying the adaptability of AMA1. High-affinity binding of the RON2 β-hairpin loop to AMA1 does not appear to be essential for invasion. Strategies to target this specific interaction have been a major focus of therapeutic and vaccine development, but strategies will need to be broader to include other AMA1 interactions to maximize efficacy. Our data also suggest that vaccine or therapeutic efficacy could be circumvented by mutations if targeting only the interaction between the RON2-loop and AMA1. The identification of specific residues that define invasion function and species-specific divergence or conservation will inform the design of novel vaccines and therapeutics against malaria.