Background
Severe life threatening
P. falciparum malaria is a major cause of mortality and morbidity in young children in sub-Saharan Africa. In endemic areas, severe malaria is most prevalent in children under the age of five years, before they acquire immunity to severe disease as a result of repeated exposure. Severe malaria manifests in children in three partly overlapping syndromes; impaired consciousness (IC), respiratory distress (RD), and severe malarial anemia (SMA) [
1]. Of these, IC and RD were found to be the key indicators of life threatening malaria in a hospital setting in sub-Saharan Africa [
1].
The pathophysiological process underlying each of these syndromes is still not understood. Disturbed microcirculation is thought to play a major role [
2]. In cerebral malaria (CM) (severely impaired consciousness), sequestration of parasite infected erythrocytes (IE) in the microvasculature of the brain is thought to be important [
3‐
6].
PfEMP1, a parasite encoded protein expressed on the surface of the IE, interacts with host receptors on the microvascular endothelia as well as unparasitized erythrocytes (a phenotype referred to as rosetting) leading to sequestration of the IE in organs. PfEMP1 is therefore thought to play a central role in parasite virulence. PfEMP1 is encoded by about 60 var genes per parasite genome and undergoes antigenic variation. Switches in the expression of the repertoire of var genes results in a high degree of plasticity in the antigenic and adhesive properties of the infecting parasite population.
Epidemiological studies have shown a subset of
var genes preferentially expressed in young and non-immune children to be associated with severe malaria especially in children with IC [
7‐
10]. This is consistent with 1) the hypothesis that some PfEMP1 variants have growth advantage in immunologically naïve children as the result of exhibiting a superior ability to sequester [
11‐
18] and 2) the observed relationship between the density of sequestration in vital organs such as the brain and fatal malaria [
4,
5,
18].
Beside sequestration of IE, severe malaria is characterised by systemic endothelial activation and widespread release of activation markers such as von Willebrand factor (vWF) [
19], soluble ICAM-1(sICAM-1) [
20] and angiopoietin-2 (ang-2) [
21,
22]. As sequestration occurs in the endothelial cells (ECs) of the microvasculature, it is believed that PfEMP1 mediated adhesion of parasitized red blood cells to the host microvasculature induces endothelial activation compromising the vascular integrity [
23,
24]. Recently, endothelial activation markers such as ang-2, soluble Tie-2 receptor, vWF have been shown to be associated with severe malaria [
22,
25]. Furthermore ang-2 is associated with retinopathy [
25], a feature identified as a surrogate marker for cerebral sequestration [
5,
26] and a recent study found fibrin deposition in the brain to be associated with sequestration of IE [
27]. If there is a connection between parasite
var expression patterns and disease severity, through mechanisms involving sequestration and endothelial activation we would expect to observe a relationship between the expression of the
var subset associated with severe malaria and markers of endothelial activation.
Previously, we showed through
var expression profiling of 217 clinical
P. falciparum isolates, that expression of a specific subset of “group A-like” PfEMP1 types is associated with severe malaria [
8]. This was based on PCR amplification of a region within the DBLα domain of PfEMP1 and sequencing. Moreover, we showed that expression of these group A-like PfEMP1 types is associated primarily with the severe syndromes of IC, while the parasite rosette phenotype was associated primarily with RD [
8,
10]. In previous post-mortem studies of Thai adults [
3,
4] and African children [
5], the severity of IC was shown to correlate with sequestration of IE by direct binding to the vascular endothelia [
18]. The subgroup of
vars expressed by the infecting parasites on the surface of IE may therefore determine the level of endothelial activation as suggested by [
28].
In this study, we explored whether a relationship exists between widespread endothelial activation (represented by plasma ang-2 levels), parasite VSA expression and severe malaria. Specifically, we investigated whether widespread endothelial activation could provide a causal link between the expression of the group A-like
vars and IC [
10] on the one hand and rosette frequency and RD [
10] on the other. To this end we measured ang-2 levels in the plasma of children from Kilifi, Kenya, presenting with either severe or non-severe malaria.
Discussion
In this study we tested whether a marker of widespread endothelial activation, plasma ang-2 levels throw light on our previously reported associations between expression of group A-like
var genes and malaria with IC on the one hand and rosetting frequency with RD on the other [
10]. Previously, a link between in vivo endothelial dysfunction and severe malaria has been established [
21]. In addition, a further study in Malawi revealed that plasma ang-2 is higher in children with retinopathy positive cerebral malaria [
25]. As retinopathy is a surrogate marker of cerebral sequestration [
5,
26,
42], this supports the idea that parasite sequestration causes endothelial activation. The recent identification of endothelial protein C receptor (EPCR) as a receptor for severe disease-associated PfEMP1 and the hypothesis that this interaction may drive inflammation [
27,
28] gives further support to this notion. Following these observations, we explored whether group A-like PfEMP1 may cause higher levels of activation of these cells and release of ang-2, exacerbating inflammation and leading to coma.
Though group A-like
var subgroup showed a significant but weak association with ang-2 this association dropped out when severe and non-severe cases are considered separately (Figure
2). Moreover, group A-like
var expression was independently associated with IC when adjusted for ang-2 in a logistic regression model (Figure
3, and Table
1). This is consistent with a model in which 1) IE expressing group A-like PfEMP1 can sequester by binding to ECs in the absence of widespread endothelial activation and inflammation and 2) expression of group A-like
var contribute to IC in a pathway at least partly independent of widespread ang-2 release.
Recent in vitro studies have shown that IE expressing PfEMP1 subsets containing domain cassette 8 (DC8) and 13 (DC13) can bind to brain endothelial cells (via EPCR) in a manner that is not dependent on the induction of adhesion molecules such as ICAM-1 [
15,
17,
43] that are induced by inflammation. The expression of these PfEMP1 subsets (i.e. DC8 and DC13) in
P. falciparum isolates sampled from children with severe and non-severe malaria have been shown to be associated with severe malaria [
9]. Therefore, in the presence of low host IE surface antibodies, the expression of group A-like PfEMP1 (such as DC13) may give the parasite growth advantage in the initial infection, dominating the sequestered biomass, and causing impaired consciousness before widespread endothelial activation and inflammation occurs.
Our converse observation was that the relationship between ang-2 and IC is not altered by adjustment for group A-like
var expression (Figure
3 and Table
1). This suggests that parasites expressing group A-like PfEMP1 do not form part of an explanatory link between ang-2 and IC. This in turn suggests that endothelial activation does not confer a growth advantage to parasites expressing group A-like PfEMP-1 through cytoadherence to inducible adhesives molecules such as ICAM-1. ICAM-1 mediated IE cytoadhesion is known to be mediated by a subset of PfEMP1 and has been shown to be associated with cerebral malaria [
44‐
46]. It has been proposed that pathogenesis in this case is mediated through a positive feedback whereby cytoadhesion promotes ICAM-1 expression which further promotes cytoadhesion [
47]. If IE adhesion to ICAM-1 or other inducible adhesive molecules is encoded by subsets of both group A and not group A-like PfEMP1 it would provide an explanation for why ang-2 association with IC remains independent of the group A-like
var expression. We are now in a position to refine approaches to measure
var gene expression in clinical isolates and explore their associations with different pathogenic mechanisms.
It is important to note that that sequestration per se does not need to play a direct role in endothelial activation for it to be important in disease pathology. Since all IE are thought to sequester, high parasite burden, whatever the underlying cause could lead to endothelial activation independent of the PfEMP1 cytoadhesive profile. This is consistent with the findings of two in vitro studies that used human brain microvascular endothelial cell lines (HBMEC [
48,
49] and hCMEC/D3 [
50] to test whether cytoaherance to ECs is important for endothelial activation. Although they differ in their conclusion on what may be responsible for EC activation, they both agree that binding of IE to the ECs is not necessary for endothelial activation. The study by zougbede et al. further demonstrate that contact between the IE and ECs is not required for endothelial activation [
50]. Again in a post-mortem study, Silamut et al. [
4] observed, generalised endothelial activation, not confined to sites of parasite sequestration in brain samples from Thai and Vietnamese adults who died from severe malaria, further supporting the possibility of endothelial activation occurring independently of IE binding [
4]. Moreover, sera from patients with falciparum malaria were found to induce increased expression of substance P, an observation that was not made with sera from healthy controls [
51]. These results raise the possibility that widespread endothelial activation is independent of direct cytoadherence of the IE to the vascular endothelia and that parasite derived circulating factors may be responsible for endothelial activation.
Clearly, parasite density (in the context of falciparum malaria) is likely to be an important determinant of endothelial activation. Parasite density may contribute to endothelial activation by increasing the amount of released microparticles, free haemoglobin, and other molecules. Recently, levels of plasma PfHRP2 (thought to represent parasite load [
40,
52], though found not useful as a marker of parasite load in a study in Kilifi [
53]), was found to differentiate between retinopathy positive CM from retinopathy negative CM [
41]. Considering this, the finding that ang-2 is higher in retinopathy positive CM cases compared to retinopathy negative cases [
25] is consistent with parasite load being an important determinant of endothelial activation. Moreover, previous work suggests that high parasite burden would lead to an increase in lactic acid production [
54], causing acidosis and RD. In this context we found ang-2 to be significantly correlated with parasite density (Figure
4E-F) and base-excess (Additional file
1: Figure S2). Parasite density was shown to be associated with respiratory distress (Table
2) which is a manifestation of metabolic acidosis [
30], measured by base excess.
Our
var gene expression measurements and assessment of rosetting phenotype come from circulating parasites rather than sequestered ones. Since sequestration and microvascular obstruction occur in multiple small areas of the microvasculature and not in a uniform manner [
4,
55,
56], we cannot rule out the possibility of group A-like PfEMP1 involvement in localised EC activation in the brain microvasculature which is not reflected in the measured plasma ang-2 levels. The latter possibility has been suggested by a recently published study that found fibrin deposition occurring more commonly in CM cases than fatal encephalopathic controls that is restricted to microvessels with sequestered IE [
27].
Competing interests
The authors declare that they have no competing interest.
Authors’ contributions
PCB, KM: designed the study and provided overall study supervision. GMW, JNM, MO, and PCB: processed samples, performed var sequencing, rosetting and IE surface antibody measurements. AA, EK, and MM performed the plasma Ang-2 ELISA: AA, GF and PCB performed the statistical analysis. AA and PCB wrote the manuscript with input from all authors. All authors read and approved the final manuscript.