Antibody responses to merozoite antigens after natural Plasmodium falciparum infection: kinetics and longevity in absence of re-exposure

Background Antibodies against merozoite antigens are key components of malaria immunity. The naturally acquired antibody response to these antigens is generally considered short-lived; however, the underlying mechanisms remain unclear. Prospective studies of travellers with different levels of prior exposure, returning to malaria-free countries with Plasmodium infection, offer a unique opportunity to investigate the kinetics and composition of the antibody response after natural infection. Methods Adults diagnosed with P. falciparum malaria in Stockholm, Sweden (20 likely malaria naïve and 41 with repeated previous exposure during residency in sub-Saharan Africa) were sampled at diagnosis and 10 days and 1, 3, 6, and 12 months after treatment. Total and subclass-specific IgG responses to P. falciparum merozoite antigens (AMA-1, MSP-119, MSP-2, MSP-3, and RH5) and tetanus toxoid were measured by multiplex bead-based immunoassays and ELISA. Mathematical modelling was used to estimate the exposure-dependent longevity of antibodies and antibody-secreting cells (ASCs). Results A majority of individuals mounted detectable antibody responses towards P. falciparum merozoite antigens at diagnosis; however, the magnitude and breadth were greater in individuals with prior exposure. In both exposure groups, antibody levels increased rapidly for 2 weeks and decayed thereafter. Previously exposed individuals maintained two- to ninefold greater antibody levels throughout the 1-year follow-up. The half-lives of malaria-specific long-lived ASCs, responsible for maintaining circulating antibodies, ranged from 1.8 to 3.7 years for merozoite antigens and were considerably short compared to tetanus-specific ASCs. Primary infected individuals did acquire a long-lived component of the antibody response; however, the total proportion of long-lived ASCs generated in response to infection was estimated not to exceed 10%. In contrast, previously exposed individuals maintained substantially larger numbers of long-lived ASCs (10–56% of total ASCs). Conclusion The short-lived nature of the naturally acquired antibody response, to all tested merozoite antigens, following primary malaria infection can be attributed to a combination of a poor acquisition and short half-life of long-lived ASCs. Greater longevity is acquired with repeated infections and can be explained by the maintenance of larger numbers of long-lived ASCs. These insights advance our understanding of naturally acquired malaria immunity and will guide strategies for further development of both vaccines and serological tools to monitor exposure. Electronic supplementary material The online version of this article (10.1186/s12916-019-1255-3) contains supplementary material, which is available to authorized users.

Positive and negative controls (pooled plasma from adult highly malaria exposed Kenyan donors and malaria unexposed Swedish donors, respectively) and a standard calibrator of serially diluted purified IgG from malaria immune donors were run on each plate in the malaria antigen multiplex assay. In the TTd assay the positive control consisted of the same highly reactive plasma pool from Kenyan donors. A serially diluted "in-house" reference sera from highly TTd reactive volunteers that had been recently booster immunised using diTeBooster® (AJ Vaccines, Denmark) was used as a standard calibrator.
For each antigen and each IgG subclass, the assay optical density (OD) or median fluorescent intensity (MFI) was converted to a relative concentration in arbitrary units (AU/mL) by interpolation from the standard calibrator curve using a five-parameter sigmoidal curve fitting. The highest concentration standard calibrator in each assay was assigned an arbitrary value of 960 AU/mL irrespective of its signal intensity. The interpolated concentration values were scaled by multiplying by the relative sample dilution factor (e.g. 10 for total IgG, 5 for IgG 1 and IgG 3 , 3 for IgG 2 , and 1 for IgG 4 ). The arbitrary antibody units are directly comparable only within each assay i.e. within antigen and IgG subclass. The assay lower limit of quantification (LLoQ) was established as the lowest concentration standard calibrator sample that displayed highly repeatable signal intensity measurements above the background noise across all experiments (defined as a CV% of all replicates of less than 20%). All samples and controls were assayed in duplicate and sample replicates with reactivity within the measuring range of the assay and a coefficient of variation (CV%) of above 20% were repeated. Samples with reactivity above the assay range were repeated at a higher dilution. A threshold of seropositivity was defined as the mean reactivity of 20 malaria unexposed controls plus three standard deviations.

Antibody kinetics model
Following first exposure to blood-stage P. falciparum infection, we assume that the proliferation and differentiation of B cells leads to a boost in antibody secreting cells (ASC) of size β. A proportion ρ of these ASCs are assumed to be long-lived with half-life d l , with a proportion 1 -ρ being short-lived with half-life d s . It is assumed that all ASCs secrete IgG molecules which decay with a half-life d a . The antibody level of an individual at time t after their first exposure to malaria at time τ 0 is given by: where r a is the rate of decay of IgG molecules, related to the half-life of IgG molecules according to r s and r l are similarly defined.
3 Individuals with a past history of exposure to P. falciparum will have some pre-existing level of antibodies (A 0 ) at the time of infection (τ 0 ). The rate of decay of this pre-existing antibody response is determined by the rate of decay of long-lived ASCs r l . The antibody level of an individual with previous exposure at time t after their most recent exposure to malaria at time τ 0 is given by: A schematic overview of the model is shown in Figure 1 (main text).

Fitting the model to data
The model was fitted to longitudinal antibody level measurements from all participants. Mixed effects methods were Individuals can be either from the 'previously unexposed' or 'previously exposed' groups. We assume that for individuals from both groups, the parameters n s d , n l d and n a d are drawn from the same population-level distribution.
That is we assume that the average duration of short-lived and long-lived ASCs and IgG molecules is the same in both groups. We assume that the two exposure groups differ in terms of the magnitude of boosting to ASCs ( As the proportion of the ASCs that are long-lived must be bounded by 0 and 1, the local parameters n ρ are assumed to be drawn from a Beta distribution.
And for individuals in the previously exposed group, the mixed effects likelihood is: Note that for individuals in the previously exposed group, the pre-existing antibody level 0 n A will be variable, depending on a large number of covariates such as time since previous residence in an endemic area. We therefore do not attempt to constrain pre-existing antibody levels using mixed effects.  convergence. Such large numbers of iterations were needed because of the large number of parameters to be estimated. The effective number of iterations was calculated using the effective Size routine in the R library coda and the effective size was checked to be > 1,000 for the global parameters. The MCMC fitting process was repeated multiple times to ensure consistent results and test for lack of convergence.

Total model likelihood
The model was fitted separately to data from all antigens (total IgG and IgG subclasses) with the parameter estimates presented in Supplementary Table S1.      previously naïve (magenta) and in the previously exposed (black). The estimated population level parameters with corresponding variance parameters are also available in supplementary Table S1.