| Description |
- Background:
Plasmodium falciparum antigenic diversity compromises the development of lasting immunity. Currently available vaccines for malaria contain a single sequence of P. falciparum circumsporozoite protein (CSP), and protection following vaccination is variant-specific. Multi-variant vaccines may help to overcome antigenic diversity, but identifying a minimum set of variants that provide protection against a broad range of CSP variants remains a challenge.
Methods:
We recently reported an epidemiologic signature of natural immunity to P. falciparum where time to reinfection with parasites bearing homologous CSP T-cell epitopes was delayed following symptomatic vs asymptomatic infections. We hypothesized that this delayed reinfection would extend to cross-protective ‘epitope types’ which may be physiochemically alike. Thus, we applied quantitative metrics of amino acid physiochemical properties (PCPs) to group CSP Th2R and Th3R epitope types via hierarchical clustering. Using pfcsp sequences (344 infections, 155 unique haplotypes) from a 14-month longitudinal cohort in Western Kenya, we evaluated PCP-based epitope type clusters by assessing the phenotype of delayed time to reinfection with parasites bearing physiochemically-similar epitope types after symptomatic vs asymptomatic infections.
Results:
Ten PCP metrics were selected from the literature and grouped into those describing <5 features, summative measures of >5 features, and experimentally derived values. At Th2R, clustering epitope types by any of the PCP metrics reproduced the phenotype with >20 groups at p = 0.05, but the empiric substitution measures (Dayhoff and Yampolsky) show the phenotype in as few as 10 groups; no PCP metric reliably displayed the phenotype with <10 groups. Clustering the Th3R epitope types displayed the phenotype with 8 groups using the Yampolsky metric. In addition, when comparing the clusters made by each PCP metric, we discovered that increasing the number of clusters created more similar, but never identical, clusters for Th2R whereas some PCP metrics created identical clusters for the Th3R epitope types. For the empiric PCP metrics, we found that the range of amino acid pair distances was smaller but the normalized values were higher than the other metrics, which may give insight into their effectiveness in clustering.
Conclusions:
Overall, our study demonstrates that amino acid properties can identify immunologically-similar CSP epitopes that share recognition by naturally-acquired immune responses. We found that the PCP metrics that can best identify this similarity for the infections in our cohort were empirically derived values. This may offer a path forward to exploiting parasite diversity and reducing the search space for variant-transcendent responses.
|
|---|
