Severe malaria, and cerebral malaria in particular, remains one of the most dreaded manifestations of Plasmodium falciparum infection. During its development within human red blood cells, the parasite exports to the surface of the infected cell proteins that mediate its adhesion to the host endothelium, causing hypoxia, vascular occlusion and endothelial activation. In cerebral malaria, the sequestration of infected erythrocytes in cerebral capillaries is thought to be responsible for coma and cerebral edema. The PfEMP1 protein plays a central role in this process: encoded by the var multigene family (approximately 60 copies per genome), it exhibits a nearly unlimited variability that enables the parasite to evade the immune system. Certain arrangements of its domains (Duffy-binding-like and cysteine-rich interdomain region domains) are associated with severe forms, notably the DC8 and DC13 cassettes. This extreme diversity, however, makes it difficult to identify the variants actually expressed in patients, as public databases do not reflect this wealth of sequences.
To address this challenge, the authors deployed a multi-omics approach on samples from patients in Cotonou (Benin) presenting with uncomplicated or severe malaria. By combining whole-genome sequencing and RNA sequencing, they reconstructed novel var sequences and assembled a custom protein database of 295,601 sequences, including 87,489 associated with P. falciparum. This database was then used for protein identification by tandem mass spectrometry (LC-MS/MS).
Differential analysis of the RNA sequencing data revealed increased expression of transcripts for the DBLα1.7, DBLα1.1, DBLα2 and DBLβ12 domains in cerebral malaria cases compared with uncomplicated malaria; these domains belong to the DC8 and DC13 cassettes implicated in cerebral pathogenesis, and the expression profile of patients with severe anemia was distinct from that of cerebral cases. At the protein level, 57 PfEMP1 proteins were identified, of which only 10 corresponded to sequences known from public databases, the remainder deriving from the reconstructed var genes. The two predominant architectures were NTS-DBLα-CIDRα-DBLβ and NTS-DBLα-CIDRα-DBLδ, with the CIDRα-DBLβ tandem being associated with possible dual binding to the ICAM-1 and EPCR receptors. Moreover, the proportion of peptides corresponding to the DBLβ domain was higher in patients with severe forms.
This work is among the first to report the direct protein identification of whole PfEMP1 proteins from field samples, confirming the validity of the approach in the face of the extreme variability of these proteins. The authors nonetheless highlight the limitations of the bottom-up LC-MS/MS strategy, which does not allow optimal sequence coverage, as well as the strongly recombinogenic nature of var genes, which calls for caution when inferring the presence of domains not directly detected. These findings should contribute to vaccine development strategies based on PfEMP1 proteins.