Linking EPCR-Binding PfEMP1 to Brain Swelling in Pediatric Cerebral Malaria

Abstract

Brain swelling is a major predictor of mortality in pediatric cerebral malaria (CM). However, the mechanisms leading to swelling remain poorly defined. Here, we combined neuroimaging, parasite transcript profiling, and laboratory blood profiles to develop machine-learning models of malarial retinopathy and brain swelling. We found that parasite var transcripts encoding endothelial protein C receptor (EPCR)-binding domains, in combination with high parasite biomass and low platelet levels, are strong indicators of CM cases with malarial retinopathy. Swelling cases presented low platelet levels and increased transcript abundance of parasite PfEMP1 DC8 and group A EPCR-binding domains. Remarkably, the dominant transcript in 50% of swelling cases encoded PfEMP1 group A CIDRα1.7 domains. Furthermore, a recombinant CIDRα1.7 domain from a pediatric CM brain autopsy inhibited the barrier-protective properties of EPCR in human brain endothelial cells in vitro. Together, these findings suggest a detrimental role for EPCR-binding CIDRα1 domains in brain swelling.

Keywords: EPCR; PfEMP1; Plasmodium falciparum; brain swelling; cerebral malaria; var.

Figure 1. Case definitions of P. falciparum severe malaria syndromes in children

(A) Cases underwent blood and parasite profiling (UM, CM), funduscopic examination (CM), and MRI for brain swelling evaluation (CM). Signs of malarial retinopathy are indicated. (B) Ret+CM and Ret−CM cases have overlapping but differential malaria-associated disease symptoms, including the presence and severity of brain swelling. (C) Ret+CM, brain swelling, and fatality are inversely associated with circulating platelet levels. Median and IQR are represented by horizontal lines. (D) Ret+CM and fatal cases are characterized by higher levels of Pfhrp2. Median and IQR are represented by horizontal lines. See also Figure S1 and Tables S1 and S2.

Figure 2. Specific PfEMP1 domains/variants are overexpressed in Ret+CM cases

(A) PfEMP1 schematic illustrating binding domains and corresponding phenotypes. (B–E) Scatter dot plot showing var transcript levels across patient groups, Medians and IQRs are represented by horizontal lines. See also Figure S2 and Table S3 and S4.

Figure 3. DC8 CIDRα1.1 and group A CIDRα1.7 EPCR-binding transcripts are enriched in CM cases with severe swelling and fatal outcomes

(A) Schematic diagrams of EPCR-binding head structures by CIDRα1 sub-classification and the domain arrangement for dual EPCR and ICAM-1-binding (CIDRα1-DBLβ1/3) sequences. (B) Heat map of median values of CIDRα1 subtype and ICAM-1-binding motif expression (dual EPCR and ICAM-1 binders). (C) Bar graphs showing transcript levels of CIDRα1 subtypes and ICAM-1-binding motif. Median and IQR are indicated. (D) Spearman correlations of transcript levels for individual EPCR-binding CIDRα1 subtypes and ICAM-1-binding motif amongst CM cases. Rho and statistical significance (correcting for multiple comparisons) are indicated. See also Figure S2 and Tables S3–5.

Figure 4. Machine learning models of retinopathy and swelling

(A) Host and parasite features that discriminate Ret+CM patients from UM cases after Pfhrp2 filtration. Other retinopathy and swelling models are found in Figure S3. (B) Receiver operating characteristic curve of the RF models. Area under the curve (95% confidence intervals). (C) Bubble graphs that summarize RF models. The area of the bubble represents the MDCA. Significant features are represented by outlined bubbles. (D) Primers targeting different var domains were grouped according to domain classification or predicted binding phenotype to create set enrichments. Statistical significance and false discovery rate (FDR) are indicated. See also Table S6.

Figure 5. Diversity and similarity in CIDRα1.7 sequences from swelling cases

(A) Schematic illustrating the position of the cloned DBLα-CIDRα1.7 fragment and the DBLα tag covered by Illumina sequencing. (B) Representation of the proportion of unique DBLα tags from Illumina sequencing that made up 90% of the var repertoire in each patient. DBLα tags that matched those from the cloned DBLα-CIDRα1.7 fragment are highlighted (purple). (See Figure S4; Tables S7–S9). (C) A neighbor-joining tree of CIDRα1.7 sequences amplified from children with Ret+CM in Malawi compared to published sequences from Tanzania (Jespersen et al., 2016) and representative CIDRα1 sequences (Rask et al., 2010), which were included to provide outgroup comparisons. Clusters of identical or highly related sequences are highlighted in grey shading and labeled 1–6. Only patient 3373 did not have a matching DBLα tag-CIDRα1.7. †: Fatal cases, CM: Cerebral malaria, RD: Respiratory distress, SA: Severe anemia.

Figure 6. Inhibition of the APC-EPCR interaction by a common brain-specific 62B1-1-CIDRα1.7 domain isolated from a fatal pediatric CM case

(A) Schematic of the expressed recombinant proteins, size exclusion chromatography trace and SDS-PAGE. (B) 62B1-1-CIDRα1.7 binds to EPCR with picomolar affinity. A representative graph from n = 3 independent experiments. (C) 62B1-1-CIDRα1.7 blocks binding of APC to EPCR. Left: representative flow cytometry histograms show binding of APC to CHO745-EPCR. Right: percentage of APC binding to CHO745-EPCR cells in the presence of varying concentrations of the CIDR domains, relative to APC binding alone. Mean ± SD of n = 3–5 independent experiments. Ab: Antibody. See Figure S5. (D) 62B1-1-CIDRα1.7 partially blocks APC-mediated protection from thrombin-induced barrier disruption of primary human brain endothelial cells (HBMECs). Left: graph showing the cell index over time. Right: graph showing the level of APC protection from thrombin-induced barrier disruption in the presence of 62B1-1-CIDRα1.7 and IT4var31-CIDRα4, relative to APC alone. Mean ± SD of n = 4 independent experiments **p<0.01, ***p<0.001 by the Student’s unpaired, two-tailed t-test.