A longitudinal study of children identifies antibody Fc-mediated functions and antigen targets of immunity to Plasmodium vivax malaria

SUMMARY

Plasmodium vivax is the most widespread cause of malaria with a high burden of disease. Progress in reducing the global malaria burden has stalled with no vaccines available partly due to a limited knowledge of targets and mechanisms of protective immunity. We developed a platform to quantify antibody functions to multiple P. vivax antigens and dissect immunity in a longitudinal cohort of children from Papua New Guinea at risk of P. vivax malaria. We identified antigens targeted by multiple functional antibodies, including interactions with Fcγ receptors, which mediate different cellular effector functions, and complement-fixation, advancing our understanding of P. vivax immunity. We identified specific antigens targeted by antibodies associated with protection from P. vivax malaria. Evaluating thousands of possible combinations, we identified subsets of antigens in the most protective combinations providing leads for developing highly protective multi-antigen P. vivax vaccines eliciting multi-functional antibody responses to achieve and sustain elimination.

Graphical Abstract

eTOC Blurb

Understanding immunity against Plasmodium vivax, a leading cause of malaria, will accelerate development of vaccines, which are currently lagging. Opi et al. identify important mechanisms and target antigens of protective P. vivax immunity in children, including combinations of responses that afford greater protection, revealing pathways for vaccine development.

INTRODUCTION

Progress in combating malaria has stalled with no reduction in disease burden in recent years, highlighting the need for new interventions. Malaria remains a leading cause of morbidity and mortality with an estimated 250 million cases and 608,000 deaths reported in 2022¹. Combating malaria is a key target of the UN Sustainable Development Goals². Two species, Plasmodium falciparum and Plasmodium vivax, cause most cases of malaria globally. P. vivax is the most geographically widespread species and a major cause of malaria outside Africa, including Asia-Pacific and South America³. Further, while P. falciparum contributes to significant morbidity and mortality in Africa, P. vivax disease burden has also become increasingly evident in Africa in recent years⁴. P. vivax presents a significant barrier to malaria elimination as currently available tools largely prioritise P. falciparum and are less effective against P. vivax, which can be transmitted at very low parasitaemia (below detection limits of standard diagnostic tests) even before symptoms appear. Moreover, P. vivax has a dormant liver stage, known as hypnozoites, which causes infection relapses weeks to years following the initial mosquito bite, contributing to disease and transmission⁵. Further, anti-malarial drug and insecticide resistance are increasing in areas where P. vivax is endemic³. In countries and regions where the disease burden of P. vivax has increased in recent years, this burden is likely to be underestimated due to difficulty in detection of low-level and asymptomatic infections and failure to account for the impact of chronic infections on overall health⁶.

Given the disease burden and challenges, a P. vivax vaccine could be transformative for control and elimination by preventing disease, including relapses from hypnozoites, and transmission⁷. The WHO and major global organisations have proposed the development of a malaria vaccine with at least 75% efficacy by 2030 for P. falciparum and P. vivax⁸. Unfortunately, no P. vivax vaccine has progressed to efficacy trials in endemic settings and limited candidates are under development^7,9. Two vaccines evaluated in human challenge models showed no efficacy¹⁰, although recent data suggest that targeting merozoites might be a promising strategy¹¹. Merozoite and sporozoite antigens of P. vivax are strong targets for vaccine development¹². Sporozoite forms are inoculated by a feeding mosquito and establish infection in the liver, while the merozoite stage of the parasite invades and replicates in reticulocytes, leading to clinical symptoms of malaria, and possible severe or fatal disease. Therefore, identifying specific antigens on merozoites and sporozoites that are targets of P. vivax protective immunity is a priority for developing an efficacious P. vivax vaccine. However, a major barrier for vaccine development has been a limited understanding of the targets and mechanisms of action of immunity against P. vivax to enable candidate selection and prioritisation. The lack of a long-term in vitro culture system and limited animal models means that new approaches are needed to understand immunity and identify potential vaccine candidates.

Naturally acquired immunity to malaria develops following repeated exposure to disease-causing Plasmodium species¹³. Antibodies are a key component of acquired protective immunity and play a key role in the two recently licensed vaccines for P. falciparum¹⁰. Antibodies can mediate protection through multiple mechanisms, including by direct inhibition of host-cell invasion and by mediation of functional activities through the Fc region of immunoglobulin molecules. Blood-stage growth and sporozoite motility and invasion inhibition assays, relying on the direct effect of antibodies on their own, have been used to evaluate some merozoite and sporozoite candidate antigens, respectively, predominantly for P. falciparum. However, correlations with protection from clinical malaria for these antibody mechanisms have been inconsistent for P. falciparum and unknown for P. vivax^10,14. Studies with P. falciparum have found that antibodies can mediate complement-fixation and activation against merozoites and sporozoites, which is initiated by IgG or IgM binding to C1q to active the classical complement cascade^15–18. This leads to parasite killing and inhibition of host-cell invasion^15–18. Further, IgG can interact with Fcγ-receptors on different immune cells to mediate subsequent cellular effector functions including opsonic phagocytosis by neutrophils, monocytes and macrophages, and antibody-dependent cellular cytotoxicity by natural killer cells^19–22. In humans, three types of activating Fcγ-receptors (FcγRI, FcγRIIa and FcγRIIIa/b) are important for these mechanisms and their expression varies across immune cell types. Binding of IgG to FcγRs is influenced by the IgG subclass, epitope specificity and glycosylation²³. FcγRIIa and FcγRIIIa/b are low affinity receptors and therefore require clustering of IgG for binding²³. Emerging evidence has found that these Fc-mediated mechanisms, involving IgG binding to complement or FcγRs, correlate with protection against P. falciparum malaria; their induction by vaccines may be crucial for achieving high efficacy^15,16,22,24. The role of these functions in protection against P. vivax malaria has not been defined, nor have the key antigens that are the targets of these functional antibodies²⁵. P. vivax has substantial biological differences from P. falciparum; therefore, findings from P. falciparum cannot be directly translated to P. vivax. Nearly all key antigens differ between the two species, and immunity to P. falciparum is not effective against P. vivax^26,27. The two recently implemented vaccines for P. falciparum, RTS,S and R21, are not effective against P. vivax and the target epitopes of protective antibodies for these vaccines are not present in P. vivax. The leading blood-stage vaccine antigen for P. falciparum, Rh5, is not present in P. vivax. Therefore, new knowledge on immune targets for P. vivax are needed to enable vaccine development.

Here, we aimed to identify major targets and mechanisms of protective immunity to P. vivax to support leads and approaches for vaccine development. To do this, we established a platform to quantify multiple functional activities of antibodies to an array of P. vivax antigens, including the ability of antibodies to fix complement and interact with different FcγR types that is the essential first step for cellular effector functions. In a longitudinal cohort study of young children at risk of P. vivax malaria, we identified targets of functional antibodies and quantified their acquisition and identified antigen-specific antibodies that were associated with protection from clinical P. vivax malaria. Using statistical models, we evaluated protective associations for thousands of combinations of antigen-specific responses for each functional activity to identify antigens and combinations that give the strongest potential protective efficacy. Further, we evaluated whether multiple functional activities are associated with greater protection, and we identified antigens that are targeted by multi-functional antibodies. Our findings provide crucial knowledge and data to enable a step-change in vaccine development for P. vivax.

RESULTS

FcγR-binding and complement-fixing antibodies are acquired to multiple P. vivax antigens among young children

To quantify target-specific functional antibodies relevant to protective immunity, we developed a high-throughput bead-based multiplex assay using the Luminex xMAP technology, allowing us to measure FcγR-binding and complement-fixing antibodies to multiple P. vivax antigens simultaneously. P. vivax recombinant antigens were coupled to magnetic microspheres at optimised concentrations. For this study, we selected a panel of 30 antigens (Supplementary Table S1) based on the following criteria: i) proteins previously defined as merozoite and sporozoite proteins, and other proteins predicted to be expressed on the parasite surface, but not yet fully confirmed; ii) immunogenic proteins shown to be targets of acquired immunity (IgG) in prior studies^28,29. These include some proteins with known or likely functions in host-cell infection by merozoites or sporozoites. However, there is a severe lack of data on the functions of P. vivax proteins. We tested the antigens in our panel for antibodies with functions relevant to protective immunity; the ability to interact with and bind activating FcγR types involved in cellular effector functions: i.e. FcγRI, FcγRIIa and FcγRIIIa. The latter two FcγR types are low-affinity receptors that bind to antigen–antibody complexes³⁰. We also tested the ability of antibodies targeting these P. vivax antigens to fix serum complement component C1q, which is the first step in the antibody-mediated classical activation pathway of the complement system, and C1q-fixation correlates with formation of the membrane attack complex¹⁶.

We quantified antibody-mediated FcγRI-, FcγRIIa- and FcγRIIIa-binding using antibodies in plasma samples obtained from a longitudinal cohort of young children from Papua New Guinea (PNG) exposed to P. vivax (n=188; samples collected at enrolment) (Figure 1A; STAR Methods, Supplementary Table S2). These were compared to the FcγR-binding activity of samples from malaria-naïve donors (n=20) and data presented as fold-change in mean fluorescent intensity (MFI) of PNG antibody samples relative to malaria-naïve donors. We observed significant FcγR-binding with antibodies from PNG children for most antigens. Antibodies from children demonstrated significantly higher binding to FcγRI for all P. vivax antigens tested except Pv-fam-a (PVX_088820) compared to antibodies from malaria-naïve control donors (p<0.01) (Figure 1B and Supplementary Figure S1A). Antibodies from PNG children had significantly higher binding to FcγRIIa and FcγRIIIa compared to malaria-naïve controls for 20/30 and 17/30 P. vivax antigens respectively (Figure 1B; Supplementary Figure S1B and S1C). FcγRI-binding activity had the highest seroprevalence in the PNG child cohort: for 29/30 antigens the seroprevalence was >30% (range 15–88%). For FcγRIIa-binding, 11/30 antigens had a seroprevalence >30% (range 10–76%) and for FcγRIIIa-binding only 5/30 antigens had a seroprevalence >30% (range 3–80%). The antigen with the highest seroprevalence was RBP2b for all FcγR types (88%, 76% and 80% respectively). We quantified C1q-fixation by antibodies in the same children (n=188), in comparison to malaria-naïve donors (n=20). Complement-fixing antibodies were very common in PNG children, with a seroprevalence >90% for most antigens (27/30), and the magnitude of C1q-fixation was statistically significantly higher than in malaria-naïve individuals for all 30 antigens (Figure 1B; and Supplementary Figure S2).

Figure 1: FcγR-binding and complement-fixing antibodies are acquired against P. vivax antigens.

(A) Overview of the study: 188 children from PNG aged 1–3 years were recruited and followed up by active and passive surveillance for 16 months for presentation with P. vivax clinical malaria. Plasma samples collected at enrolment were measured for antibody magnitude including IgG, IgG subclasses and IgM, and antibody functions including FcγR (I, IIa, IIIa) binding and complement-fixation to a panel of 30 P. vivax antigens using a high-throughput bead-based multiplex assay we developed. We then identified major antigen targets, including combinations, of P. vivax functional antibodies and those associated with protection from prospective risk of clinical P. vivax malaria. (B) Functional antibody acquisition was analysed as fold-change in mean fluorescent intensity of PNG antibody samples (n=188) relative to malaria-naïve Melbourne donor samples (n=20) for each antigen and antibody function. MFI fold change for each antigen and function was then classified as high, medium or low (based on tertiles), or no significant increase. Differences in FcγR-binding and complement-fixation activity between PNG children and malaria-naïve controls were tested using the Wilcoxon rank-sum test and p-values are indicated with asterisks; *p= 0.05 to 0.01, **p<0.01 to 0.001, ***p<0.001 and >0.05 = ns. (C) The most prominent targets of each individual functional antibody type (FcγR-binding and complement-fixation) were identified and are shown ordered by MFI fold change (highest to lowest). Antigens that were the most prominent targets of multiple antibody functions are shown in the multifunction column (calculated as the sum of responder categories across the four functions). See also Supplementary Table S1 and S2, Figures S1 and S2 and STAR Methods.

From these analyses we identified the most prominent target antigens for each functional antibody type (high, medium, low, or no significant fold-change), as well as the antigens that were the most prominent targets of multiple functions (Figure 1C). The three most prominent targets included MSP1–19, MSP8 and RBP2b for FcγRI-binding, RBP2b, MSP1–19 and RAMA for FcγRIIa-binding, RBP2b, MSP1–19 and RON2 for FcγRIIIa-binding and MSP5, PVX_101530 and Pv-fam-a (PVX_112670) for C1q-fixation. When looking at all functions (FcγR-binding and C1q-fixation combined) RBP2b, MSP1–19, RON2, RAMA, and MSP8 were the five most prominent targets of multifunctional antibody responses (Figure 1C). These results show that FcγR-binding and complement-fixing functional antibody responses are acquired to multiple P. vivax antigens with a subset of antigens more prominently targeted with multiple antibody functions.

Specific targets of antibodies with FcγR-binding activity are associated with protection against P. vivax malaria

Using our longitudinal study of children, we examined whether antibody-mediated FcγR-binding was prospectively associated with a reduced risk of P. vivax clinical malaria during follow-up (defined as fever plus parasite density >500/μL) for each antigen³¹ (using the causal framework for analysis shown in Supplementary Figure S3). Individuals were classified as either high, intermediate or low responders based on tertiles for FcγR-binding antibodies to each P. vivax antigen tested. Using a negative binomial generalised estimating equation (GEE) model, we determined the adjusted incidence rate ratios (aIRRs) for P. vivax clinical malaria among the 188 children followed over a 16-month period. We compared high and intermediate responders to low responders for FcγRI-, FcγRIIa- and FcγRIIIa-binding antibodies for each of the 30 P. vivax antigens. The model accounted for confounding variables of age, season, village of residence and exposure (estimated using molecular force-of-infection)³².

For all FcγR types, there were multiple antigens for which high FcγR binding was associated with reduced risk of P. vivax malaria, suggesting a role for FcγR-mediated mechanisms in P. vivax immunity. High FcγRI-binding antibodies were associated with a significantly lower risk of P. vivax clinical malaria for 19/30 antigens (Figure 2A, Supplementary Table S3). Adjusted incidence rate ratios (aIRRs) ranged from 0.45 to 1.04, and for 29/30 antigens there was ≥20% reduction in P. vivax clinical malaria risk (aIRR ≤0.8). The 5 antigens with the strongest protective associations were (in order) Pv-fam-a (PVX_112670) (55% reduced risk; aIRR 0.45; 95% confidence interval (CI) 0.30–0.67; p<0.001), RON2 (0.48; 0.30–0.77; p=0.002), MSP3α (0.52; 0.35–0.78; p=0.002), RBP2b (0.53; 0.36–0.80; p=0.002) and MSP7B (0.54; 0.36–0.82; p=0.004).

Figure 2: FcγR-binding antibodies are associated with protection from P. vivax malaria.

The association between FcγR-binding antibodies measured at enrolment to a panel of 30 P. vivax recombinant antigens and prospective risk of P. vivax clinical malaria was evaluated in a longitudinal cohort of 188 PNG children for (A) FcγRI-binding, (B) FcγRIIa-binding and (C) FcγRIIIa-binding. FcγR-binding antibodies were classified into tertiles of high, intermediate and low (reference) responders. Adjusted incidence rate ratios were calculated using negative binomial regression with adjustment for age, season, village of residence and individual differences in P. vivax exposure. Infection count data were tested for overdispersion. Forest plots show adjusted incidence rate ratios and 95% confidence intervals for high responders relative to low responders represented by the dotted line (See also Supplementary Table S3). The top 3 antigens combination is shown as a red bar, which represents the high responders for the three most protective antigens compared to those who were low responders for all 3 antigens (See Supplementary Tables S3 and Figures S3 and S4A–C).

For FcγRIIa, there were 12 antigens for which high activity was associated with a significantly lower risk of P. vivax clinical malaria, and high responders had ≥20% reduction in risk of P. vivax malaria for 20/30 antigens. The 5 antigens most strongly associated with protection were RON2 (aIRR 0.44; 95% CI 0.29–0.68; p<0.001), MSP3β (0.54; 0.37–0.77; p=0.001), PTEX150 (0.55; 0.37–0.80; p=0.002), PVX_091710 (0.58; 0.37–0.90; p=0.016) and Pv-fam-a (PVX_125728) (0.58; 0.40–0.84; p=0.004) (Figure 2B, Supplementary Table S3).

A smaller proportion of antigens showed significant associations with protection for FcγRIIIa (7/30). High FcγRIIIa-binding to 8 antigens was associated with ≥20% reduced risk of P. vivax clinical malaria, and the 5 antigens with the strongest associations were RON2 (aIRR 0.52; 95% CI 0.35–0.77; p=0.001), MSP3α (0.61; 0.40–0.92; p=0.020), Pv-fam-a (PVX_090265) (0.61; 0.41–0.91; p=0.017), SIAP2 (0.62; 0.40–0.98; p=0.040) and PTEX150 (0.62; 0.44–0.95; p=0.029) (Figure 2C, Supplementary Table S3).

We hypothesised that higher protective associations may be observed with antigen combinations. To explore this, we selected the 3 antigens that were most strongly associated with protection for each FcγR type and compared individuals who were high responders to those who were low responders for all 3 antigens. The strength of association among high responders to all the top 3 antigen combinations tended to be higher than that observed for individual antigens (Figure 2A–C, Supplementary Figure S4A–C, Supplementary Table S4). These results suggest that FcγR-binding by antibodies play a role in immunity to P. vivax malaria and identifies specific targets of these antibodies in protective immunity.

Complement-fixation by antigen-specific antibodies is associated with protection against P. vivax malaria

We identified specific antigens for which antibody-mediated complement-fixation was prospectively associated with a reduced risk of P. vivax clinical malaria. For 23/30 antigens, high C1q-fixation activity was associated with ≥20% reduction in the risk of P. vivax clinical malaria compared to low activity; however, not all would be regarded as statistically significant. The antigens with the 5 strongest protective associations were MSP5 (aIRR 0.58; 95% CI 0.40–0.86; p=0.006), RON2 (0.61; 0.41–0.91; p=0.017), RBP2b (0.62; 0.43–0.91; p=0.015), Pv-fam-a (PVX_092990) (0.62; 0.42–0.93; p=0.021) and Pv-fam-a (PVX_125728) (0.63; 0.42–0.94; p=0.023) (Figure 3A, Supplementary Table S3). To further understand the potential role of complement-fixing antibodies in protective immunity, we examined the protective association for a combination of antigen-specific responses using the 3 antigens with the strongest individual protective association. High responders for C1q-fixing antibodies to a combination of the top 3 antigens (MSP5, RON2, and RBP2b; all are merozoite antigens) had substantially stronger protective associations than that seen for the most protective individual antigens (combined response aIRR 0.23; MSP5 aIRR 0.58, RON2 aIRR 0.61 and RBP2b aIRR 0.62) (Figure 3A, Supplementary Figure S4D, Supplementary Table S4). These results support a potential role for complement-fixing antibodies in protective immunity and identifies antigens targeted by protective antibodies.

Figure 3: Complement-fixing antibodies are associated with protection from P. vivax clinical malaria, and IgG1 and IgG3 subclasses are major drivers of functional antibody responses.

(A) Association between complement (C1q)-fixing antibodies measured at enrolment to a panel of 30 P. vivax recombinant antigens and prospective risk of P. vivax clinical malaria was tested in a longitudinal cohort study of 188 PNG children. C1q-fixing antibodies were classified into tertiles of high, intermediate and low (reference) responders. Adjusted incidence rate ratios were calculated using negative binomial regression with adjustment for age, season, village of residence and individual differences in P. vivax exposure. Infection count data were tested for overdispersion. Forest plots represent adjusted incidence rate ratios and 95% confidence intervals for high responders relative to low responders represented by the dotted red line (See also Supplementary Tables S3 and Figure S3). The top 3 antigens combination is shown as a red bar, which represents the high responders for the three most protective antigens compared to those who were low responders for all 3 antigens (See also Supplementary Figures S3 and S4D). (B) IgG1 and IgG3 antibody acquisition for the 5 most protective antigens for each antibody function type of FcγRI, FcγRIIa, FcγRIIIa-binding and complement-fixation are shown (some antigens were in the top 5 most protective group for more than one functional activity). Data are presented as fold-change in MFI (compared to malaria-naïve controls) with the boxplots represent the median and interquartile ranges. Differences in MFI fold change between IgG1 and IgG3 were assessed using the Wilcoxon rank-sum test and p-values are shown (See also Supplementary Figure S4E for comparisons with malaria naïve individuals). (C) Relationships between IgG subclasses and functional activities by linear regression analysis including IgG1 (blue) and IgG3 (yellow) for each of the top 5 antigens. Regression coefficients are shown, and significant associations are indicated by asterisks: *p= 0.05 to 0.01, **p<0.01 to 0.001 and ***p<0.001.

Antigen-specific IgG and IgG subclasses are major drivers of functional antibody responses

We investigated antibody types that are the major drivers of FcγR binding and complement fixation activity for those antigens that had the strongest protective associations. FcγR-binding interactions are only mediated by IgG, whereas complement-fixation can be mediated by IgG and IgM. Among the IgG subclasses, IgG1 and IgG3 are the major mediators of complement-fixation and FcγR interactions³³. Linear regression models were used to analyse relationships between antibody types and functional activities; our analysis focused on the five antigens that were most strongly associated with protection for each antibody function.

We first quantified IgG subclass responses which revealed that IgG1 was significantly higher than IgG3 among children for all antigens examined (Figure 3B, Supplementary Figure S4E). For FcγRI-binding activity, IgG1 and IgG3 were significant determinants of activity for all 5 of the most protective antigens, except RBP2b which was not significant for IgG3 (Figure 3C). The strength of the associations was also higher for IgG1 for all antigens. For FcγRIIa- and FcγRIIIa-binding, both IgG1 and IgG3 were significant determinants of activity, except IgG3 was not significant for the two Pv-fam-a antigens for FcγRIII. For complement-fixing antibodies, IgG was positively correlated with activity for the top 5 antigens (statistically significant only for RBP2b and RON2) whereas IgM was not a major determinant of complement-fixing activity for any of these antigens; 4 of 5 coefficients had negative values for IgM (Supplementary Table S5). IgG1 was positively correlated for 4 antigens and IgG3 was positively correlated for 4 antigens, but only clearly statistically significant for 3 and 2 antigens respectively (Figure 3C). These results suggest IgG is the major driver of complement-fixation with both subclasses playing a role.

Network analysis of all FcγR-binding and complement fixation activities, IgG, and IgM, to all 30 antigens, revealed that antigens typically clustered together for a given antibody type (Figure 4A). This was particularly prominent for C1q, FcγRI, IgG and IgM responses. For FcγRIIa and FcγRIIIa, responses to different antigens were less strongly clustered. Further, FcγR responses clustered more closely with each other and with IgG than with C1q-fixation or IgM. Across all antigens, the correlations between different FcγR-binding activities were significant and generally higher than seen between complement-fixation and FcγR-binding activities (Figure 4B); however, the strength of correlations did vary substantially across the panel of antigens indicating differences in the co-acquisition of different FcγR binding activities between antigens. FcγRI-binding for the different antigens clustered closely with IgG (Figure 4A), reflected in the generally high correlation coefficients for FcγRI versus IgG (Figure 4C). Correlation coefficients were moderate-to-high for IgG versus FcγRIIa and FcγRIIIa and showed more variation in the strength of correlations than seen for FcγRI. Overall, results suggest that the different FcγR-binding activities are generally more strongly co-acquired with each other than they are with complement-fixation, reflective of different mechanisms of protection.

Figure 4: Antibody responses to P. vivax antigens are differentially correlated and co-acquired.

(A) Network plots showing the correlation between the different antibody functions (IgG, IgM, FcγRI, FcγRIIa, FcγRIIIa and complement C1q) to each of the 30 antigens tested were generated using Spearman’s correlation coefficients with the qgraph package in R. Each antigen variable is represented as a node (numbered as per Supplementary Table S1) and the connecting lines represent the correlation between different antigens with only lines for positive correlation coefficients of Spearman’s rho >0.5 displayed. Increasing strength in correlations is seen with closely clustered nodes and vice versa. (B) Correlations coefficients between different functional antibody responses (FcγRI, IIa and IIIa and complement-fixation) against the 30. P. vivax antigens. (C) Correlations coefficients between either IgG or IgM and different functional antibody responses (FcγRI, IIa and IIIa and complement-fixation) against the 30. P. vivax antigens. Correlation coefficients were generated using the Spearman’s rank correlation coefficient and considered as weak (r=0.0–0.3), moderate (r=0.4–0.6) or strong (r=0.7–1.0) correlations with strong correlations indicated in white numbered coefficients. Results are shown as a heatmap based on the strength of correlations. See also Supplementary Figures S5 and S6.

Further analyses reveal the extent of co-acquisition of antibodies to different antigens. Across the 30 antigens, we observed weak to strong correlations between antigens for each FcγR type: Spearman’s rho range 0.17–0.80 for FcγRI (Supplementary Figure S5A), r=0.23–0.76 for FcγRIIa (Supplementary Figure S5B) and r=0.25–0.70 for FcγRIIIa (Supplementary Figure S6A). Complement-fixation between the different antigens was moderately to very strongly correlated (r=0.51–0.91) (Supplementary Figure S6B). The variability in correlations between parameters may reflect variations in the rate at which responses are acquired to different antigens. Highly correlated responses may reflect antigens having similar presentation to the immune system.

These results show that IgG1 and IgG3 are major drivers of functional antibody activity and antigen-specific responses clustered by antibody type, with FcγR-binding activities more strongly correlated with each other than with complement-fixation, suggesting different protective mechanisms.

Antibody responses to specific antigen combinations are associated with protection from P. vivax malaria

To determine whether maximal protection results from responses to specific combinations of antigens we examined protective associations for combined responses to multiple antigens for each functional antibody parameter, FcγR binding and complement fixation. We aimed to identify antigens in combinations that had the strongest protective associations, providing insights that may inform vaccine development or the development of biomarker tools to quantify immunity in populations. To address this, antigen-specific responses were grouped into all possible combinations of responses of 2, 3 and 4 antigens (435, 4,061 and 27,415 combinations, respectively) for each functional antibody type and aIRRs were calculated for each combination. We reasoned that combinations of up to 3 or 4 antigens are feasible in vaccine development.

The most protective associations were observed for combinations of specific antigens and were higher than individual antigens and protective associations generally increased with increasing combination size up to combinations of 3 antigens (Figure 5). For FcγRI, the medians (and ranges) of aIRRs for the top 50 protective combinations of 2, 3 and 4 antigens were 0.51 (0.39–0.54), 0.25 (0.14–0.28) and 0.24 (0.03–0.26) respectively. It was notable that specific combinations of 3 or 4 antigens had very high protective associations (up to 97%). In contrast, the median of the aIRRs for the 30 antigens assessed individually was lower than for combinations (0.64 [0.45–1.04]) (Figure 5). Similar patterns were seen for FcγRIIa and FcγRIIIa whereby protective associations were higher with the top 50 protective combinations of 2, 3 or 4 antigens than for the 30 individual antigens. The medians (and ranges) of aIRRs for the top 50 combinations for FcγRIIa were 0.62 (0.50–0.69), 0.43 (0.33–0.46) and 0.41 (0.32–0.43) for combinations of 2, 3 and 4 antigens, respectively, versus 0.71 (0.45–0.96) for single antigens (Figure 5). For FcγRIIIa, medians (and ranges) of aIRRs for combinations of 2, 3 and 4 antigens were 0.58 (0.56–0.64), 0.50 (0.15–0.62) and 0.45 (0.21–0.50), respectively, versus 0.97 (0.52–1.30) for single antigens (Figure 5). While there was a substantial increase in the protective associations for combinations of 3 antigens compared to 2 antigens for each of the four functional types, there was minimal gain in protective associations for combinations of 4 antigens compared to 3 antigens.

Figure 5: Antigen-specific functional antibody responses to different antigen combinations are associated with protection from P. vivax malaria.

Comparisons in the strength of association with protection from P. vivax clinical malaria between different antigen combinations of each functional antibody are shown. Plots show incidence rate ratios and interquartile ranges for the 30 P. vivax antigens tested individually and the top 50 most protective (with IRRs upper 95% confidence intervals less than 1) 2-antigen, 3-antigen or 4-antigen combinations for FcγRI-, FcγRIIa-, FcγRIIIa-binding and complement (C1q) fixation. Differences in incidence rate ratios between the one-antigen group and the 2-, 3- or 4-antigen combination groups were assessed using the Wilcoxon rank-sum test and p-values are shown in asterisks *p= 0.05 to 0.01, **p<0.01 to 0.001 and ***p<0.001. For some combinations, less than 50 combinations met the criteria for inclusion (IRRs upper 95% confidence interval of less than 1) including FcγRIIa (2-antigen combinations (n=32), FcγRIIIa (2-antigen combinations (n=3), 3-antigen combinations (n=37)).

Similarly, for C1q-fixing antibodies, the protective association for the top 50 combinations of 2 antigens (median aIRR 0.62; range 0.52–0.65), 3 antigens 0.38 (0.30–0.41) and 4 antigens 0.39 (0.36–0.41) were substantially greater than for the individual antigens (median aIRR 0.76; range 0.59 – 0.98), and there was little difference between combinations of 3 and 4 antigens (Figure 5). These analyses also show that protective associations for combinations of 3 and 4 antigens were strongest for FcγRI, followed by C1q, then FcγRIIa and FcγRIIIa.

To identify specific antigens in combinations that may contribute most to protective associations, we ranked combinations based on their aIRRs. We then determined the frequency at which a specific antigen was present in the most protective combinations for each combination size; this was performed for each functional type. This was conducted with combinations of 2 or 3 antigens since protective associations were greater than individual antigens, and combinations of 4 antigens were not clearly superior to 3 antigens.

Among 2-antigen combinations, the most featured (≥20% frequency) antigens in the highly protective combinations for FcγRI included RBP2b (32%) and Pv-fam-a (PVX_112670) (20%) (Figure 6A). For FcγRIIa, the most frequent antigens were RON2 (48%) and Pv-fam-a (PVX_090265) (30%), and for FcγRIIIa they were RBP2b (39%), RON2 (23%), Pv-fam-a (PVX_090265) (23%) and PTEX150 (23%). For combinations of C1q-fixing antibodies, RBP2b (36%) and MSP7L (20%) were the most frequent antigens in the most protective combinations (Figure 6A). In the 3-antigen combinations, antigens in the most protective combinations included RBP2b (38%), RON2 (26%) and MSP7B (20%) for FcγRI (Figure 6B); PVX_097715 (28%), Pv-fam-a (PVX_090265) (28%), RON2 (26%) and RBP2b (21%) for FcγRIIa; RBP2b (35%), Pv-fam-a (PVX_090265) (26%), RON2 (22%) and PTEX150 (22%) for FcγRIIIa; and RBP2b (43%) for C1q (Figure 6B).

Figure 6: Specific P. vivax antigens are more frequent in the most protective antigen combinations for FcγR-binding and complement-fixation.

The frequency with which each antigen was present in the top 10% most protective antigen combinations of (A) 2 antigens and (B) 3 antigens, for each functional antibody parameter (FcγRI, FcγRIIa, FcγRIIIa and C1q) are shown. This was determined by calculating the incidence rate ratios for all combinations of 2 or 3 antigens and then ranking them from the most to least protective combinations (from lowest to highest ratio). The frequency of each antigen in the top 10% most protective combinations for each of the 30 antigens tested are shown in the circles; p<0.0001 (Chi square contingency table) for differences in the frequency with which antigens appeared in protective combinations, for all functional parameter analyses. (C) Antigens that were most frequent in the most protective associations considering all functional parameters are shown. The top 7 antigens are shown ordered from highest to lowest. For each functional parameter, the top quartile (n=7) of antigens was identified based on frequencies shown in A or B (combinations of 2 or 3 antigens, respectively) and ranked according to their frequency. For each antigen, a sum-of-ranks was calculated from the four functional parameters FcγRI, IIa, IIIa and complement-fixation).

We extended this analysis by identifying which antigens were most frequent in protective combinations when considering all functional antibody measures together. We ranked antigens based on how frequently they occurred in the most protective antigen combinations for multiple antibody functions (FcγRs and C1q combined). For 2 antigen combinations, RBP2b, RON2, MSP3α, Pv-fam-a (PVX_112670), Pv-fam-a (PVX_090265), MSP5 and Pv-fam-a (PVX_096995) were most frequent in protective combinations across multiple functions (Figure 6C). For 3 antigen combinations, RBP2b, RON2, Pv-fam-a (PVX_090265), Pv-fam-a (PVX_112670), MSP3α, PVX_091710) and MSP5 were the most common antigens (Figure 6C).

Our findings suggest that the most highly protective associations occur with combinations of responses to specific antigens. An alternative hypothesis was that higher protective associations may be observed for children with a broad repertoire of functional antibodies targeting multiple antigens compared to children with a narrow repertoire of functional antibodies. Therefore, we calculated a repertoire score for each individual child considering responses to all 30 antigens tested, and grouped children into categories of high, medium or low repertoire scores based on tertiles. This was done for each functional antibody type (the 3 FcγR types and C1q fixation). Using this analysis, we did not observe stronger associations in children with a broad repertoire of functional antibodies when compared to children with a narrow repertoire for FcγRI (aIRR 0.99; 95% CI 0.99–1.00), FcγRIIa (1.00; 95% CI 0.99–1.00), FcγRIIIa (1.00; 95% CI 0.99–1.00) and C1q (1.00; 95% CI 0.99–1.00). These results show that responses to specific combinations of antigens result in stronger protective associations rather than simply having a broad repertoire of antibody responses and identifies a subset of antigens that occur frequently in the most protective combinations.

Multiple antibody functions are associated with stronger protective immunity

We investigated the extent to which multiple functional antibody responses to an individual antigen are associated with protection from P. vivax malaria. For each antigen, we assessed protective associations for combinations of FcγR-binding activities and combinations of FcγR-binding and C1q-fixation activity (Figure 7A). For most antigens (20/30), we observed stronger associations with protection for high responders to combinations of multiple functions (FcγRs combined and/or FcγRs and C1q combined) than with individual functions. Notably, for some antigens specific FcγR binding or complement fixation activity was strongly protective and a combination of responses did not improve protection (Figure 7A). Protective associations for FcγRs combined were stronger than individual FcγR associations for 16 proteins. For these 16 proteins, the median aIRR was 0.54 (46% reduction in risk; 95% CI 0.49–0.59) for a combined responses to all 3 FcγR types versus medians of 0.65, 0.67 and 0.89 for FcγRI, FcγRIIa and FcγRIIIa respectively (p<0.05) when assessed as individual responses (Figure 7B). Protective associations for FcγR types and C1q combined were stronger than individual functional responses for 18 proteins. For these proteins the median aIRR for combined responses was 0.39 (95%CI 0.33–0.48) versus a median of 0.72 (95%CI 0.68–0.79) for C1q only (p<0.0001) (Figure 7C). Further, protective associations appeared stronger for FcγR types and C1q combined versus FcγR types combined only, median aIRR 0.39 versus 0.54 (Figure 7B–C). These results suggest that some antigens effectively induce polyfunctional antibody responses associated with protection and a combination of functional activities might be needed for maximal protection for some antigens.

Figure 7: Combinations of antigen-specific functional antibody responses are associated with protection from P. vivax malaria.

(A) Comparisons in the strength of association with reduced risk to clinical malaria between individual and combinations of functional antibody responses for each of the 30 P. vivax antigens tested are shown. For each antigen, incidence rate ratios were generated for high versus low responders for each individual functional antibody type, or for high versus low responders to combinations of all FcγR-binding antibodies (FcγRI, FcγRIIa and FcγRIIIa) or combinations of all FcγR-binding and complement (C1q)-fixing antibodies. Forest plots show the strength of associations with protection from P. vivax clinical malaria in antigen-specific functional antibody responses (FcγRI, FcγRIIa, FcγRIIIa and C1q) assessed individually or in combination. Plots represent adjusted incidence rate ratios and 95% confidence intervals for high responders relative to low responders represented by the dotted red line. Comparisons of the strength of associations with protection for P. vivax antigens when quantifying (B) combinations of FcγR antibody responses compared to individual FcγR responses and (C) combinations of FcγR and complement-fixing antibody responses compared to individual complement-fixing antibody responses are shown.

DISCUSSION

Identifying potential vaccine candidates for P. vivax malaria has been challenging due to a limited understanding of the mechanisms of immunity and specific targets of protective responses. In addition, there has been a lack of approaches and platforms that enable antigen-specific antibodies to be evaluated for multiple antibody functions. We addressed these roadblocks by developing a high-throughput multi-antigen assay platform to rapidly quantify and assess antigen-specific functional antibody responses, complement-fixation and FcγR-binding, to multiple P. vivax antigens simultaneously. Our analysis of antibodies among a longitudinal cohort of children exposed to P. vivax malaria, targeting 30 different P. vivax antigens, and with analyses of >125,000 antibody combinations, identified specific antigens targeted by antibodies that were strongly associated with protection from P. vivax malaria. It was notable that combinations of three antigens were sufficient for maximal protective associations and we identified antigens that featured in the most protective combinations that have promise for inclusion in multi-antigen vaccines. We provided data on antibody FcγR binding and complement fixation to multiple antigens correlated with protection from P. vivax malaria and evaluated multiple antibody functions and their association with protection. These findings provide leads for developing multi-antigen P. vivax vaccines with the aim of generating multi-functional antibody responses for potent immunity to protect against disease and accelerate elimination.

Two analytic approaches provided important insights into the most prominent targets of antibodies associated with protective immunity. The first included analysis of associations with protection for each individual antigen and function. From this, antigens including RON2, MSP3α, MSP7, RBP2b, MSP3β, MSP5, SIAP2 and Pv-fam-a (PVX-112670) were identified. It was notable that some antigens were strongly associated with protection for two or more functional antibody types (e.g. RBP2b and RON2). A second approach used statistical models to estimate protective associations for all possible combinations of 2, 3 and 4 antigens (>125,000 combinations) for each antibody function. Subsequently, our analysis identified the frequency with which antigens appeared in the most protective associations. Antigens that occurred with high frequency (≥20%) in the most protective combinations of 2 and 3 antigens included RON2, MSP7B, MSP7L, RBP2b, Pv-fam-a (PVX_090265), PVX_097715, Pv-fam-a (PVX_112670) and PTEX150. Some antigens were not significantly associated with protection when analysis was performed individually (Figures 2 and 3A) but were found in highly protective combinations for some functions (Figure 6) (e.g. MSP7L for C1q-fixation). For some antigens, antibodies with combined multi-functional activity were associated with higher protection than found with analysis of a single functional activity. However, overall, the greatest association with protection was observed for combined responses to multiple antigens rather than multiple functions to a single antigen.

Antibodies to sporozoite antigens were generally not prominent targets in protective associations, whereas several merozoite antigens were strongly associated with protection. This may be because most clinical episodes are due to relapses (arising from hypnozoite reactivation) rather than new infections from mosquito inoculations^5,34,35. This would suggest that merozoite antigens need to be included in vaccines for P. vivax. Further studies are needed to investigate the potential role of antibodies to sporozoite antigens in protective immunity. Achieving highly protective immunity and vaccine efficacy are likely to involve antibodies to multiple antigens. Therefore, analyses of protective combinations are highly informative for vaccine development pathways. Notably, it was not simply the presence of antibodies to a higher repertoire of antigens that was associated with higher protective immunity; rather, it was combinations of FcγR binding and complement fixing antibodies to specific antigens that were most protective. Our approach may be a suitable framework that could be applied to future studies of protective immunity to further inform the selection of antigens for vaccine development. The lack of functional correlates of protection for malaria has been a significant constraint for selecting and advancing candidates in vaccine development. Therefore, our data here address a major knowledge gap. Previous studies on P. vivax immunity have been few in number and have largely focused on the role of a limited number of antigens and/or only quantified IgG responses and their associations with protective immunity from P. vivax malaria and associations with protection have been identified for IgG responses to some antigens³⁶.

Considering vaccine design, our data suggest that specific combinations of three antigens that generate potent Fc-mediated functional activity may be sufficient to achieve high-level protection against clinical P. vivax malaria. Combinations of this size would be feasible in vaccine development, and several existing childhood vaccines contain multiple antigens³⁷. Our findings are also valuable for enabling the development and application of biomarkers of immunity to monitor populations, quantify the impacts of malaria interventions and differences between populations, and identifying populations at risk of malaria resurgence due to waning or insufficient immunity. Our data suggest that quantifying functional antibodies to 3 key antigens (e.g. RBP2b, RON2 and Pv-fam-a (PVX_090265)) for one functional activity may be a good biomarker of immunity.

Our findings advance important concepts in protective immunity to P. vivax, providing evidence that Fc-mediated functional activities of antibodies (FcγR-binding and complement-fixation), targeting specific antigens, are linked with protection. This was evident in several ways. For each of FcγR-binding and complement-fixation activity, antibodies to a subset of antigens correlated with protection. Further, combinations of 2 or 3 antigens were strongly associated with protection for each functional antibody type. Protective associations also tended to be higher for antigens with strong multi-functional activity. However, there were some notable exceptions to this. For example, MSP1₁₉ was a prominent target of functional antibodies, but these responses were not associated with protection and MSP1₁₉ did not appear in the most protective combinations. Prior studies using P. falciparum have shown that antibody interactions with FcγR types on immune cells promote phagocytosis by neutrophils and monocytes and antibody-dependent cellular cytotoxicity by natural killer cells against merozoites and sporozoites^{19,20,22,24,38}. Antibody binding of C1q activates the classical complement cascade leading to killing of merozoites and sporozoites, and inhibition of host-cell invasion^15–18,39. Studies in mouse models using rodent Plasmodium spp also provide evidence that Fc-mediated functions are involved in malaria immunity^40–42. The assay approach we used measures the initial and essential step for Fc-mediated immune responses, FcγR engagement and complement binding by antibodies, C1q fixation. Cellular assays, such as antibody dependent phagocytosis and cellular cytotoxicity measure distal, cell-dependent outcomes. Both are functional measurements at different points of larger biological events. Future studies using cell-based assays in response to opsonized targets are needed to define the roles of different cell types and functions and evaluate the influence of cellular and host factors that can influence outcomes.

The prevalence of FcγR-binding activity by antibodies was highest for FcγRI, with most antigens having a prevalence of FcγRI-binding antibodies >30%. The prevalence was lower for FcγRIIa-binding activity and lowest for FcγRIIIa. This may be due to FcγRI being a high-affinity receptor that binds to monomeric IgG, whereas FcγRIIa and FcγRIIIa/b are low-affinity receptors that bind antigen–antibody complexes⁴³. FcγRIIIa and FcγRIIIb are expressed on neutrophils and some monocytes, whereas only FcγRIIIa is expressed in natural killer cells⁴⁴. FcγRI and FcγRIIa are expressed by neutrophils, monocytes and macrophages. IgG1 responses were higher than IgG3 for all antigens examined, and both IgG1 and IgG3 were correlated with all functional activities, although their relative importance varied among antigens, which is consistent with their known biological activities. IgG was the driver of complement-fixation, with no clear role for IgM.

In conclusion, we identified specific antigenic targets and combinations of responses that are strongly linked with protective immunity, including antigens that feature in highly protective combinations. Our findings identify roles for antibody Fc-mediated functional mechanisms, FcγR binding and complement fixation, in immunity to P. vivax malaria, marking a major change in our knowledge. These findings open multiple avenues for vaccine development to overcome existing roadblocks. Future studies should evaluate combinations of selected antigens, identified here, as vaccine candidates for generating multi-functional antibodies for maximal protective immunity. A highly protective vaccine will greatly facilitate achieving and sustaining malaria elimination in the future.

Limitations of the study

Our study has some limitations to consider. We only evaluated children in one malaria-endemic country; our findings encourage future studies in additional geographic and transmission settings. Further knowledge of the role of antibodies to specific antigens could also be obtained from animal models, testing candidate vaccines for immunogenicity and efficacy. Considering the limited knowledge of P. vivax biology and immunity, it is possible that other antigens and antibody functions not tested here may be playing a role in immunity. A strength of our study is the scale of antigens and parameters quantified in a longitudinal study of children with P. vivax. The cohort included young children at high risk of malaria who were acquiring P. vivax immunity, and our analysis was strengthened by the inclusion of molecular force-of-infection to account for variance in exposure. Our findings highlight several emerging vaccine candidates and less-studied antigens as immune targets, and our findings strongly encourage future research on specific antigens to build knowledge on their structure and function to support vaccine design.

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by lead contact, Prof. James Beeson (james.beeson@burnet.edu.au)

Materials availability

Biological reagents generated for the purpose of this study can be requested from the lead contact.

Data and code availability

• Data reported in this paper will be shared by the lead contact for non-commercial purposes upon request and may depend on clearance from ethics or regulatory committees.

• This paper does not report original code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

STAR METHODS

Experimental Model And Study Participant Details

Study population

Briefly, between March and September 2006, 264 children aged 1–3 years resident in Maprik District, East Sepik Province of Papua New Guinea (PNG) were recruited into a longitudinal cohort study, described previously³¹. Participants were followed up with active surveillance for clinical episodes of malaria every 2 weeks for 16 months. Additionally, every 8 weeks all participants were examined and blood samples collected twice, 24 hours apart, to improve detection of low-level infections. Malaria rapid diagnostic tests (RDTs) were applied for testing febrile children or those displaying symptoms of a febrile illness and the results were confirmed by light microscopy. The presence of specific human malaria species was determined using a semiquantitative post-PCR and ligase detection reaction fluorescent microsphere assay (LDR-FMA)⁴⁵. All parasitologically confirmed malaria cases (by RDT and light microscopy) were treated with artemether-lumefantrine (Coartem), and those with moderate to severe anaemia (haemoglobin [Hb] level of 7.5 g/dl) received treatment with artemether-lumefantrine and 4 weeks of iron and folate supplementation according to national treatment guidelines. Children were also passively followed up over the entire follow-up period through presentations as outpatients at local health centers. P. vivax clinical malaria episodes were defined as the presence of fever (axillary temperature >37.5°C or history of fever in the preceding 48 hours) with a concurrent P. vivax parasite density > 500 parasites/μL. The current study is restricted to 188 (out of 264) children recruited at the start of the study who completed follow-up and had plasma samples available for antibody assays. Antibody magnitude was then measured in the 188 recruitment plasma samples. Plasma samples collected from malaria naïve donors residing in Melbourne, Australia, (n=20) were used as negative controls in all assays.

Participants in the cohort study were enrolled and managed by the PNG Institute of Medical Research. Ethics approval was provided by the Medical Research Advisory Committee of the National Department of Health, PNG. Written informed consent was provided by the parents or guardians of all children.

General cohort characteristics at enrolment

Age in years, median (IQR)

1.72 (1.34–2.43)

Gender, n (%)

Female – 80 (42.6)

Male – 108 (57.4)

Location, n (%)

East Sepik Province, PNG – 188 (100)

P. vivax prevalence, %

Light microscopy PCR positive – 40%

All participants were of Papua New Guinean ethnicity and resident in East Sepik Province of Papua New Guinea

METHOD DETAILS

Proteins

Antibody assays were performed with a panel of 30 P. vivax recombinant proteins. These were selected based on known location on the surface of merozoites or sporozoites, known or predicted function in host-cell infection by merozoites or sporozoites, reported as targets of acquired human antibodies and/or having shown previous associations with protection from clinical P. vivax malaria for IgG reactivity^29,46. The panel includes proteins expressed during the blood-stage (n=15), pre-erythrocytic stage (n=4), several less-well characterized proteins (n=10) and a sexual-stage antigen. All proteins were expressed using the wheat germ cell-free protein expression system as previously described²⁸ except for PvCSP210, PvCSP247 and PvAMA1 that were expressed using the mammalian HEK293 expression system⁴⁷. Additional details on all proteins are provided in Supplementary Table S1.

Multiplex FcγR-binding assays

Examples of the performance of assays using different antigens with optimised conditions is shown in Supplementary Figure S7. P. vivax recombinant proteins were coupled onto magnetic carboxylated Luminex MicroPlex microspheres (Luminex Corporation) at optimised protein concentrations determined by generation of log-linear standard curves with a PNG positive pooled plasma sample as reported in previous studies²⁹. The PNG pooled sample standard curve was included in all assay runs to ensure stability of coupled proteins, reproducibility of results and testing of plasma samples at the same dilution. All 30 P. vivax recombinant proteins were tested simultaneously in each well. 50 μL of recombinant protein-coupled microspheres were incubated with 50 μL of test plasma samples at a dilution of 1:100 in PBS-0.05% Tween 20, for 30 minutes in 96-well plates. Protein-conjugated microspheres were then incubated with 50 μL of either biotinylated recombinant monomeric FcγRI (0.5 μg/mL), or biotinylated recombinant ectodomain dimers of FcγRIIa H131 (0.2 μg/mL) or FcγRIIIa V158 (0.1 μg/mL) for 30 minutes. This was followed by incubation with 50 μL of phycoerythrin (PE)-conjugated streptavidin at a dilution of 1:250 in PBS-0.05% Tween 20, for 15 minutes. Initial assessment of FcγR-binding showed low signal/reactivity for both FcγRIIa- and FcγRIIIa-binding. Therefore, the assay was further optimised to increase the fluorescent signal. Following addition of PE-conjugated streptavidin, samples were incubated with a biotinylated goat anti-streptavidin antibody with a further second incubation step with PE-conjugated streptavidin. This multi-layered addition of PE-conjugated streptavidin with a biotinylated anti-streptavidin antibody results in the addition of more fluorochromes and resulted in a 10-fold amplification of the FcγR-binding fluorescent signal compared with the single addition of PE-conjugated streptavidin (Supplementary Figure S7C). All incubations were performed at room temperature in the dark with 3 wash steps in between each incubation with PBS-0.05% Tween 20. The microspheres were finally resuspended in 80 μL of PBS-0.05% Tween 20, and FcγR-binding activity was then determined by measuring median fluorescent intensity (MFI) on the MAGPIX reader. Background MFI determined from blank wells containing microspheres but no plasma samples were subtracted from each sample. During assay optimisation duplicate test samples showed near perfect correlation (Supplementary Figure S7D) therefore all test samples were run in singlicate to conserve limited cohort samples. To check for plate-to-plate variation in assay performance each assay plate included a 10-point standard curve from a serial dilution of a PNG positive pool plasma sample and another 3 separate individual PNG positive samples run at a dilution of 1:100.

Multiplex complement-fixation assay

Complement-fixation assays were performed using the same panel of 30 P. vivax recombinant protein-coupled microspheres as the FcγR-binding assays above. Examples of assay performance using different antigens with optimised conditions is shown in Supplementary Figure S7. 50 μL of recombinant protein-coupled microspheres were incubated with 50 μL of test plasma samples at a dilution of 1:100 in PBS-0.05% Tween 20, for 30 minutes in 96-well plates. This was then followed by incubation with 50 μL of 50mg/mL purified human complement factor C1q (Millipore 204876); binding of C1q is the first essential step in the classical complement-activation pathway. The coupled microspheres were then incubated with 50 μL of rabbit anti-C1q antibody⁴⁸ at a dilution of 1/250 in PBS-0.05% Tween 20, for 15 minutes and a further incubation with a PE-conjugated anti-rabbit IgG antibody (Life Technologies P-2771MP) at a dilution of 1/250 in PBS-0.05% Tween 20, for 15 minutes. All incubations were performed at room temperature in the dark with 3 wash steps in between each incubation with PBS-0.05% Tween 20. The microspheres were finally resuspended in 80 μL of PBS-0.05% Tween 20, and C1q-fixing antibody acticity determined by measuring median fluorescent intensity (MFI) on the MAGPIX reader. Background MFI determined from blank wells containing microspheres but no plasma samples were subtracted from each sample. During assay optimisation, while duplicate test samples showed a very strong correlation (Spearman’s rho coefficient 0.96) a few samples showed variability for different antigens (Supplementary Figure S7D) therefore all test samples were run in duplicate, and data are presented as the geometric mean fluorescent intensity.

To check for plate-to-plate variation in assay performance each assay plate included a 10-point standard curve determined from a serial dilution of a PNG positive pool plasma sample and another 3 separate individual PNG positive samples run at a dilution of 1:100.

IgG, IgG subclasses and IgM multiplex assays

IgG and IgM antibody reactivity was measured using the multiplex Luminex assay as described in detail previously⁴⁹. Briefly, P. vivax coupled-microspheres were incubated with 50 μL of test plasma samples at a dilution of 1:100 in PBS-0.05% Tween 20, for 30 minutes in 96-well plates followed by 3 wash steps with PBS-0.05% Tween 20. This was followed by the addition of 100 μL of PE-conjugated donkey anti-human IgG Fc (Jackson ImmunoResearch JI709116098), mouse anti-IgG1 hinge secondary antibody (SouthernBiotech 9052–09), mouse anti-IgG3 hinge secondary antibody (SouthernBiotech 9210–09) or donkey anti-IgM secondary antibody (Jackson ImmunoResearch 709-116-073) at dilutions of 1:100 for IgG and IgG subclass antibodies and 1:400 for anti-IgM antibody, and incubation for 15 minutes. The microspheres were washed three times and then resuspended in 80 μL of PBS-0.05% Tween 20, and total IgG and IgM antibody magnitude determined by measuring median fluorescent intensity (MFI) on the MAGPIX reader.

Quantification and Statistical analysis

We compared differences in FcγR-binding and complement-fixing activity between different groups using the Wilcoxon rank-sum test while correlations between continuous variables were assessed using the Spearman’s rank correlation coefficient (rho) and considered as weak (r=0.0–0.3), moderate (r=0.4–0.6) or strong (r=0.7–1.0) correlation⁵⁰. Linear regression analysis was used to analyse the relationship between antibody types (IgG, IgG1, IgG3, IgM) and functional activities (FcγR-binding and complement fixation). To test the association between functional antibodies and the count of prospective clinical P. vivax malaria infections, we generated incidence rate ratios (IRRs) to quantify infection risk using a negative binomial generalised estimating equation (GEE) model with an exchangeable correlation structure and semi-robust variance estimator as previously described³⁶. A directed acyclic graph (DAG) was used to visualize the causal framework for statistical analyses (Supplementary Figure S3). The model adjusted for age, season and village of residence. We also accounted for individual differences in exposure, that is the molecular force of blood-stage infection, defined as the number of new P. vivax blood-stage clones acquired per year at risk that was square root transformed for better fit^36,51. Infection count data were tested for overdispersion. For the analysis, antibody magnitude was categorised into tertiles of high (H), intermediate (I) and low (L) responders based on activity of respective functional antibodies tested and the low group used as reference in all analysis. Multiple comparison adjustments were not performed, instead adjusted incidence rate ratios (aIRRs) with 95% confidence intervals were calculated and reported⁵², as shown in Figures 2 and 3A. Further, when interpreting associations between specific antibody responses with protection, we considered the strength of associations, whether the 95% confidence intervals crossed 1 and the overall coherency of results^53,54. Seropositive samples were classified as having an OD greater than the mean + 3 standard deviations of malaria non-exposed Australian donors (n=20).

Antibody breadth scores were calculated by assigning a score of 1, 2 and 3, to low, medium and high reactivity for each functional antibody parameter, respectively, and adding up each individual’s scores for each antigen (n=30) to calculate the breadth score.

For the analysis on antigen combinations (Figures 5 and 6), all possible combinations of 2 (435 combinations), 3 (4061 combinations) or 4 (27,415 combinations) antigens were assessed using adjusted negative binomial regression. We assigned a score of 1, 2 and 3 to low, medium and high reactivity for each functional antibody parameter, respectively. These tertile categories were then added together for each combination, and combination scores were then used to create 3 approximately equal sized groups reflecting low, intermediate and high combination responses^16,55. These tertile combination responses were used in the negative binomial regression, as described above to quantify protective association. As such, analysis of protective associations for 2-antigen combinations were performed between individuals who were high for both antigens or high to one and medium to the other antigen (H+H, H+M or M+H) versus individuals who were low to both antigens or low to one antigen and medium to the other antigen (L+L, L+M or M+L). For 3 antigen combinations, comparisons were made between individuals who were (H+H+H or H+H+M) versus those who were (L+L+L or L+L+M). For 4 antigen combinations, comparisons were made between individuals who were (H+H+H+H or H+H+H+M or H+H+M+M) versus those who were (L+L+L+L or L+L+L+M or L+L+M+M). We identified the top 10% protective combinations (based on aIRR, provided the upper 95% confidence interval was less than 1) and calculated the frequency with which each antigen was present in the top 10% most protective antigen combinations of 2 antigens or 3 antigens, for each functional antibody parameter (FcγRI, FcγRIIa, FcγRIIIa and C1q). To identify the most prominent targets of multiple functional antibodies (Figure 1C), we calculated the total for each antigen responder score (high, intermediate, low; from Figure 1B).

Network plots on the correlation between different antigen-specific antibody parameter to all 30 P. vivax antigens tested (Figure 4A) were generated using the “qgraph” package in R using Spearman’s rank correlation coefficient data generated as described above. All analyses were performed using Stata v17.1 (StataCorp, Texas, USA) or R (R Foundation for Statistical Computing, Vienna, Austria).

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Goat anti-Rabbit IgG (H+L) Cross-Adsorbed Secondary Antibody, PE	Life Technologies	P-2771MP
Rabbit anti-C1q antibody	Burnet Institute	N/A
PE-conjugated Donkey F(ab)2 anti-human IgG	Jackson Immunoresearch	JIR 709-116-098
Anti-IgG1 hinge secondary antibody	SouthernBiotech	9052-09
Anti-IgG3 hinge secondary antibody	SouthernBiotech	9210-09
Anti-IgM secondary antibody	Jackson ImmunoResearch	709-116-073
Biological samples
Data and human plasma samples from cohort of Papua New Guinean children	This paper, see Table S2, STAR Methods	N/A
Human plasma samples from malaria-naïve Australian adults	Australian Red Cross, this paper	N/A
Human plasma samples (including reference pool) from immune Papua New Guinean adults	This paper	N/A
Chemicals, peptides, and recombinant proteins
27 P. vivax proteins expressed in wheat germ cell free system	CellFree Sciences	N/A
3 P. vivax proteins expressed in mammalian HEK293 expression system	Burnet Institute	N/A
Magnetic COOH beads	BioRad	171,506(xxx)
sulfo-N-hydroxysuccinimide (S-NHS)	Sigma	56485
N-ethyl-N′-(3-(dimethylamino)propyl)carbodiimide (EDC)	Sigma	3449
Bovine Serum Albumin (BSA)	Sigma	A7906
Recombinant soluble FcγRIIa H131 ectodomain dimers	Burnet Institute	N/A
Recombinant soluble FcγRIII V158 ectodomain dimers	Burnet Institute	N/A
Recombinant human FcγRI protein	Sino Biological	10256-H08H
Purified human C1q	Merck Millipore	204876
Streptavidin, (R-PE)	Thermo Fisher Scientific	SA10041
Goat Anti-Streptavidin, Biotinylated	Vectorlabs	BA-0500-.5
BSA, 100g	Sigma	A7906
Tween	Thermo Fisher Scientific	BP337
Software and algorithms
R Studio	RStudio, Inc.	V 2025.5.1.513
STATA	StataCorp, Texas, USA	V 17.1
Other
MAGPIX Instrument	Luminex	https://www.luminexcorp.com/magpix-system/#overview

Highlights.

• Quantified P. vivax functional antibodies using high-throughput multi-antigen platform

• Identified prominent targets of antibodies that engage Fcγ-receptors and fix complement

• Identified antigen-specific functional antibody responses protective against P. vivax

• Specific antigen and antibody function combinations are linked to high-level protection