AI-designed biomarker improves malaria diagnostics

The malaria parasite Plasmodium vivax can persist in a dormant state, causing relapsed infections and ongoing transmission. To detect possible dormant infections, clinicians use a diagnostic test containing parasite proteins, such as reticulocyte-binding protein 2b, or PvRBP2b, that trigger a host antibody response. Of the biomarkers in this test, a response to PvRBP2b provides the strongest indication of a dormant infection, but PvRBP2b is difficult to produce and has low stability. Jaison D Sa at the Walter and Eliza Hall Institute of Medical Research and the University of Melbourne, Australia, and an international team recently reported stabilized PvRBP2b variants in the Journal of Biological Chemistry.

Illustration of red blood cells infected with malaria (yellow) and Plasmodium pathogens in the bloodstream (green).

Because much of PvRBP2b’s surface binds antibodies, the team had to preserve these sites while boosting stability to keep the protein viable for diagnostic tests. They determined that they needed to mutate residues in the protein core, a challenge to maintaining the overall protein structure. They used computational modeling and an artificial intelligence–based sequence generator to design three PvRBP2b variants. All three purified variants had higher yields and greater thermal stability than the nonmutated protein.

X-ray crystallography and biolayer interferometry, a technique measuring light reflection patterns to sense biomolecule interactions, confirmed that the variants retained the original overall structure and antibody-binding capabilities. Finally, in plasma assays using samples from individuals in malaria-endemic regions, the variants elicited antibody responses comparable to the original PvRBP2b protein.

These variants could improve malaria diagnostic kits and may help solve protein stability issues in other diagnostic tests.