A Potent Anti-Malarial Human Monoclonal Antibody Targets Circumsporozoite Protein Minor Repeats and Neutralizes Sporozoites in the Liver

Abstract

Discovering potent human monoclonal antibodies (mAbs) targeting the Plasmodium falciparum circumsporozoite protein (PfCSP) on sporozoites (SPZ) and elucidating their mechanisms of neutralization will facilitate translation for passive prophylaxis and aid next-generation vaccine development. Here, we isolated a neutralizing human mAb, L9 that preferentially bound NVDP minor repeats of PfCSP with high affinity while cross-reacting with NANP major repeats. L9 was more potent than six published neutralizing human PfCSP mAbs at mediating protection against mosquito bite challenge in mice. Isothermal titration calorimetry and multiphoton microscopy showed that L9 and the other most protective mAbs bound PfCSP with two binding events and mediated protection by killing SPZ in the liver and by preventing their egress from sinusoids and traversal of hepatocytes. This study defines the subdominant PfCSP minor repeats as neutralizing epitopes, identifies an in vitro biophysical correlate of SPZ neutralization, and demonstrates that the liver is an important site for antibodies to prevent malaria.

Keywords: Monoclonal antibodies; NVDP minor repeats; Plasmodium falciparum; circumsporozoite protein; malaria vaccines; passive transfer; sporozoites.

Figure 1.. L9 is a neutralizing human mAb that binds the NPNV motif of PfCSP

(A) Left: liver burden reduction (bioluminescence; total flux, photons/sec) in B6-albino mice 40 hrs post-infection (hpi; n=15/group; data pooled from three independent experiments) mediated by 100 µg CIS43 or L9 administered 2 hours before IV challenge with 2,000 Pb-PfCSP-GFP/Luc-SPZ. P-values were determined by comparing L9 to CIS43 and untreated control using the Kruskal-Wallis test with Dunn’s post-hoc correction. Right: serum mAb titers 2 hours after administration of 100 µg CIS43 or L9 in separate mice (n=5/group) determined through ELISA. Differences between CIS43 and L9 were determined using the two-tailed Mann–Whitney test. (B) Left: liver burden reduction (Pf 18S rRNA normalized to number of human hepatocytes) in FRG-huHep mice (n=5/group) administered 50 or 10 µg L9 and challenged IV with 100,000 PfSPZ. Right: serum mAb titers in challenged FRG-huHep mice. Differences between VRC01 (anti-HIV-1 isotype control IgG) and L9 (50 vs. 10 µg) were determined using the two-tailed Mann–Whitney test. (C) Schematic of PfCSP_3D7 depicting the N-terminus, repeat region (with color-coded overlapping 15mer peptides 20–61), and C-terminus. Every NPNV motif in each peptide is underlined. (D) ELISA (MFI, median fluorescence intensity) of CIS43 and L9 binding to peptides 20–61. (E) Competition ELISA (AUC, area under the curve) of L9 binding to rPfCSP_FL in the presence of varying concentrations of peptide 22 (WT, leftmost bar) or variant peptides (subsequent bars) where the indicated residue was mutated to alanine or serine. A-B: lines represent geometric mean. See also Figure S1.

Figure 2.. mAb two-step binding to rPfCSP is associated with high junctional affinity

(A) ITC analyses of L9 IgG binding to wild-type rPfCSP (rPfCSP_WT or FL) and rPfCSP with all four NVDP mutated to NANP (rPfCSP_∆ABCD); N- and C-termini in schematics were omitted. Top, dQ/dt (heat flow, Q, as a function of time). Bottom, the integrated heat associated with each IgG injection shown as a function of the molar ratio between IgG antigen binding sites and rPfCSP in the calorimetric cell. The red line represents the result from best nonlinear least squares fit of the data. Dissociation constant (K_D) and stoichiometry (N) of binding are shown for binding events 1 and 2. ITC data was fit to a two-step binding model if the IgG titrant bound to two sets of sites with different affinity values. The first set of high-affinity sites is saturated at lower IgG concentrations before the second set of lower-affinity sites. (B) ITC analyses of CIS43, 311, 317, 1210, mAb10, and MGU12 IgG binding to rPfCSP_WT (or FL). (C) Schematics of rPfCSP_FL and rPfCSP mutants with truncated repeat regions (23/4, 19/3, 5/3 NANP/NVDP repeats) and identical N- and C-termini. (D) Aggregate stoichiometry (binding events 1+2; no. of antigen binding sites) of mAb binding to 5/3, 19/3, 23/4, and rPfCSP_FL determined through ITC. (E) Affinity (binding events 1 vs. 2; K_D, nM) of mAb binding to 5/3, 19/3, 23/4, and rPfCSP_FL determined through ITC. All ITC plots are representative of 2–3 independent experiments; D-E reflect an average of these experiments. See also Figures S1–2; Tables S1–2, S4.

Figure 3.. Cytotoxic PfCSP mAbs prevent SPZ from accessing and invading hepatocytes

(A) Inhibition of PfSPZ invasion of HC-04 hepatocytes by 10, 1.0, and 0.1 µg/mL PfCSP mAbs. Data were combined from two independent experiments (triplicate wells/mAb). Bars represent the mean ± SEM; dotted lines indicate the highest mAb-mediated inhibition at each concentration. Each mAb was compared to the corresponding concentration of VRC01 using a two-way ANOVA with Bonferroni’s correction. (B) Liver burden reduction in mice 40 hpi (n=10/group; data pooled from two independent experiments, solid vs. open symbols; line represents geometric mean) mediated by 75 (squares) or 25 (circles) µg/mouse mAb administered 2 hours before IV challenge with 2,000 Pb-PfCSP-GFP/Luc-SPZ. Dotted lines indicate the geometric mean of the background and untreated controls. P-values were determined by comparing each mAb to untreated using the Kruskal-Wallis test with Dunn’s post-hoc correction. (C) Left: schema for intravital liver imaging in mice (N=3/group) sequentially administered 30 µg Alexa405-labeled mAb (blinded), 1 µg rhodamine-labeled dextran, and 100,000 Pb-PfCSP-GFP-SPZ. The number of SPZ (No. of SPZ) observed across the 3 independent experiments per mAb were combined for analysis. The locations of individual Pb-PfCSP-SPZ were noted; traversal was detected by dextran uptake into wounded hepatocytes. Representative images depict discrete Pb-PfCSP-SPZ in the sinusoid, traversing a dextran⁺ hepatocyte, or invading a dextran⁻ hepatocyte. Right: locations of Pb-PfCSP-SPZ in the liver calculated as the percentage of total SPZ observed. (D-F) Left: representative time-lapse images (min:s:ms) of discrete Pb-PfCSP-SPZ (D) exiting a sinusoid and traversing a hepatocyte, (E) shedding a mAb-bound PfCSP tail in a CSPR, and (F) undergoing dotty death. Right: percentages of Pb-PfCSP-SPZ that (D) initiated traversal (i.e., traversed ≥1 hepatocyte), (E) underwent a CSPR, and (F) underwent dotty death. C-F: P-values were determined by comparing each mAb to VRC01, L9, or 317 using the chi-squared test with Bonferroni’s correction. E-F: arrow and star respectively indicate anterior and posterior ends of SPZ. See also Figure S3; Table S4.

Figure 4.. L9 is more potently protective than several published PfCSP mAbs

(A) Survival curves of mice challenged with five infected mosquito bites 3 days after passive transfer of 600, 300, or 100 µg PfCSP mAbs (n=number of mice/group; data were combined from seven independent blinded experiments). P-values were determined by comparing L9 to every other mAb and untreated control using the log-rank test. (B) Serum mAb titers one day prior to challenge in mice from A. P-values were determined by comparing mAbs to each other using the Kruskal-Wallis test with Dunn’s post-hoc correction. (C) Dose-response relationship of infection probability (percent protected) versus mAb dose (µg) and pre-challenge serum titers (µg/mL) for L9, 317, CIS43, and mAb10 estimated by a 2PL regression model. Thick line denotes average relationship across all experiments; lighter lines denote individual experiment relationships predicted from the model. (D) ED₅₀ and EC₅₀ with 95% confidence interval (CI) of L9, 317, CIS43, and mAb10 estimated from the 2PL model. See also Figure S4; Tables S3–4.