Characterization of a novel inhibitory human monoclonal antibody directed against Plasmodium falciparum Apical Membrane Antigen 1

Abstract

Malaria remains a major challenge to global health causing extensive morbidity and mortality. Yet, there is no efficient vaccine and the immune response remains incompletely understood. Apical Membrane Antigen 1 (AMA1), a leading vaccine candidate, plays a key role during merozoite invasion into erythrocytes by interacting with Rhoptry Neck Protein 2 (RON2). We generated a human anti-AMA1-antibody (humAbAMA1) by EBV-transformation of sorted B-lymphocytes from a Ghanaian donor and subsequent rescue of antibody variable regions. The antibody was expressed in Nicotiana benthamiana and in HEK239-6E, characterized for binding specificity and epitope, and analyzed for its inhibitory effect on Plasmodium falciparum. The generated humAbAMA1 shows an affinity of 106-135 pM. It inhibits the parasite strain 3D7A growth in vitro with an expression system-independent IC₅₀-value of 35 μg/ml (95% confidence interval: 33 μg/ml-37 μg/ml), which is three to eight times lower than the IC₅₀-values of inhibitory antibodies 4G2 and 1F9. The epitope was mapped to the close proximity of the RON2-peptide binding groove. Competition for binding between the RON2-peptide and humAbAMA1 was confirmed by surface plasmon resonance spectroscopy measurements. The particularly advantageous inhibitory activity of this fully human antibody might provide a basis for future therapeutic applications.

Figure 1. Plasma IgG reactivities against AMA1.

Using samples from 31 Ghanaian blood donors (G01–31) as well as a European control pool of two donors which had never been exposed to Malaria (NIP), ELISA was performed to estimate the reactivity against 3D7 AMA1. As positive control, a pool of four highly reactive Ghanaian donors was used. Values reflect relative reactivity in comparison to this positive control’s reactivity, which was set to 1 (A). The reactivity of the samples to AMA1 (3D7) and AMA1 (DiCo 1-3) (previously published in Feller et al.) were compared and the degree of correlation was estimated using the Spearman rank sum test (B). The reactivity of the plasma samples against both the native antigen and the reduced/alkylated antigen was estimated by ELISA, and shown as a box plot (C). The p-values were estimated using the Wilcoxon matched pairs test.

Figure 2. Cloning, expression and binding specificity of humAbAMA1.

HumAbAMA1 was cloned as full length expression construct in both vectors pTRAkt and pTT5 for transient expression in plants and in HEK293-6E cells, respectively (A). The expression and purity of the full length antibodies expressed both in plants (N. benthamiana) and in mammalian cells (HEK293) was controlled by SDS-PAGE. Both non-reduced and reduced samples of the antibodies were loaded on the gel (B). The binding of humAbAMA1to AMA1 (3D7) was tested by indirect ELISA. Dilutions of the antigen (AMA1 (3D7)) were coated onto ELISA plates, then human antibodies were applied followed by goat anti-human antibody coupled to alkaline phosphatase (Dianova, Hamburg, Germany). Black diamonds: Semi-immune (premune) plasma pool; green circles: humAbAMA1 (N. benthamiana); Red open squares: humAbAMA1 (HEK293); orange inverted triangles: mock control antibody (2G12).

Figure 3. Sensorgrams of surface plasmon resonance (SPR) measurement for the determination of humAbAMA1 affinity to AMA1 (3D7).

The affinity of the antibody humAbAMA1 was estimated by SPR spectroscopy. For each cycle recombinant and purified humAbAMA1 either produced in N. benthamiana (A, green sensogramms) or produced in HEK293-6E cells (B, red sensograms) was captured on a Protein A surface (65 RU for N. benthamiana produced or 32 RU for HEK produced humAbAMA1). Subsequently, the antigen AMA1 (3D7) was injected at concentrations of 500 nM, 250 nM, 125 nM, 62, 5 nM, 31.25 nM, 15.6 nM, 7.6 nM, 3.9 nM, 1.95 nM and 0.975 nM for 180 s, followed by a buffer injection step for 420 sec. To determine kinetic constants, the data was fitted using a 1:1 Langmuir interaction model using the software BIAEval 4.0. Fit curves are shown in light grey.

Figure 4. Immunofluorescence imaging of humAbAMA1.

The specific binding of the recombinant antibody humAbAMA1 to both schizonts (upper panel) and free merozoites (lower panel) was analyzed by immunofluorescence assays. Parasites were co-stained with humAbAMA1 (red, first column), murine anti-MSP4 antibody (green, second column) and the nuclear stain Hoechst 33342 (blue, third column). The last column shows the overlay image. Human and murine antibodies were detected by fluorescently labelled antibodies goat-anti-human IgG (H + L)-Cy3 (Dianova) and goat-anti-mouse IgG (H + L)-Alexa Fluor^® 488 (Life Technologies), respectively. Images were taken at a 630-fold magnification with a Leica TCS SP8 microscope. The right panel shows the overlay of the three fluorescence images. Scale bar 7.5 μm.

Figure 5. In vitro growth inhibition activity of humAbAMA1.

The anti-parasitic effect was estimated using a standardized Growth Inhibition Assay (GIA). Purified antibodies were added to the parasite strain Plasmodium falciparum 3D7A (A), HB3 (B) and FCR3 (C) at schizont stage and co-incubated for 40–46 h. Growth of the parasites was estimated by a pLDH assay. All tests were carried out in technical triplicates; error bars correspond to the standard error of the mean. Antibodies tested were the humAbAMA1 produced in plants (green circles), humAbAMA1 produced in HEK293 (red open squares, only tested in strains 3D7A), chimeric antibody 4G2 (blue closed squares) and chimeric 1F9 (purple triangles).

Figure 6. Synergistic effect of monoclonal antibodies on parasitic growth inhibition.

The combinations of the humAbAMA1 with either the MSP10-specific humAb10.1 (pink diamonds) or the chimeric antibody 4G2 (blue circles), respectively, were tested for synergistic growth inhibition. Antibodies were applied singly as well as in combinations. Antibodies were mixed in different ratios (1:4, 1:2, 1:1, 2:1, 4:1) proportional to their respective estimated IC₅₀ values. Titrations of each of the mixtures were added to the parasites. The IC₅₀ value of each of the antibody mixes was estimated and is depicted as relative inhibition. Curves were fitted using the One phase Decay function of GraphPad Prism 5.0.

Figure 7. Competition measurements of humAbAMA1 with the 4G2 and 1F9.

Using SPR spectroscopy, pairwise epitope mapping was performed. The chip was functionalized with antibody mAb1D7, which binds to domain III of AMA1. This chip was regenerable. Subsequently, in each cycle AMA1 (3D7) (1750 RU) was caught. Then, humAbAMA1, 1F9, or 4G2 were injected singly (cycles 1, 2, 4 and 5) or consecutively (cycles 3 and 6). In cycles 1–3, 1 μg/ml of each antibody was applied, whereas in cycles 4–6, the concentration of each antibody was 10 μg/ml. In case spectral resonance is increasing with the binding of the secondary antibody, no competition is detectable, in case the signal is not increased after binding of the secondary antibody, the epitopes either overlap or binding is compromised by allosteric competition.

Figure 8. Epitope Mapping.

The epitope of humAbAMA1 was mapped using the CLIPS-constrained peptides. Its localization pertains to domain I of AMA1 (yellow) and is highlighted in red. The published epitope of 1F9 within domain I is highlighted in cyan, the region in which the epitopes of humAbAMA1 and 1F9 overlap is highlighted in magenta. The domain II of AMAI is shown in blue. The RON2 peptide (green) is bound to its specific hydrophobic binding groove in domain I. The structural data was obtained from the protein data base PDB, accession number 3ZWZ. The structure in the second image is a 90° rotation of the first image.

Figure 9. Competition of humAbAMA1 and RON2sp1 peptide.

The determination of the epitope region recognized by humAbAMA1 using the peptide array, as well as the binning experiments suggests a possible steric interference between humAbAMA1 and RON2sp1. To investigate the potential competition between RON2sp1 and humAbAMA1 for AMA1-binding AMA1 (3D7) was preincubated with various concentrations of the peptide RON2sp1 for 1 hour. Subsequently, the AMA1 (3D7):RON2sp1 solutions were injected over a surface featuring a defined amount (400 RU) of humAbAMA1 caught on a Protein A-functionalized chip using a BIAcore T200 SPR instrument. For visualization of the results the measured response units were plotted against the concentration of RON2sp1 (μM) used as competitor.