Background
PfEMP1 is the immunodominant antigen of the malaria parasite
Plasmodium falciparum expressed on the surface of the infected erythrocyte (IE). Adherence of this molecule to host receptors expressed on endothelial cells, uninfected erythrocytes and placental syncytiotrophoblasts facilitates sequestration of IE in vascular tissues, avoiding destruction in the spleen [
1‐
3]. PfEMP1 molecules are encoded by the
var multigene family [
1‐
3]. Individual parasites have approximately 60
var gene variants and switching between single, transcribed
var genes leads to changes in cytoadhesive phenotype as well as clonal antigenic variation and immune escape.
var gene repertoires differ among isolates [
4] and immunity to malaria is dependent on acquisition of antibodies to a range of PfEMP1 variants [
5‐
8]. Immunity to both cerebral malaria [
9] and non-cerebral, severe malaria [
10] is acquired much more rapidly than immunity to uncomplicated malaria. Parasites that cause severe disease appear to express a conserved subset of variant antigens that are encountered earlier in life and that are thus more widely recognized by sera from semi-immune children than parasites causing uncomplicated disease [
11,
12].
PfEMP1s contain combinations of Duffy binding-like domains (DBLα, β, γ, δ, ε, ζ and x) and cysteine rich inter-domain regions (CIDRα, β, γ and δ) [
13]. Some DBL and CIDR domain subtypes mediate adhesion to different host receptors (reviewed in [
14,
15]), and some are organized in semi-conserved domain cassettes (DC) that are present in most parasites [
4].
var genes are also classified using their upstream sequence into groups A, B, C [
16,
17] which comprise 20, 60 and 20 % respectively of the
var gene repertoire [
4]; the unique
var gene called
var2csa has a different upstream sequence (ups E) and is only involved in malaria during pregnancy [
18].
The expression of particular subtypes of DBLα domains in severe malaria suggests severe disease may be preferentially caused by a restricted subset of
var genes [
19,
20]. Increased expression of group A and B
var genes has been associated with clinical, but not specifically severe malaria in Papua New Guinea (PNG) [
21,
22] and with severe malaria in Africa [
23]. Cerebral malaria in Africa was associated with increased expression of group A [
20,
24,
25] or group B [
26]
var genes.
Consistent with its having a role in severe malaria, PfEMP1s encoded by group A and B
var genes appear to be widely expressed by parasites that infect non- or semi-immune individuals. Antibodies from older children preferentially recognized PfEMP1s encoded by Group A
var genes, indicating previous exposure [
27]. Group A and B
var genes dominated infection of a naive individual [
28], and more individuals develop antibodies to group A PfEMP1s than group B or C, and do so at a younger age [
29].
Group A and B
var genes also encode adhesion phenotypes associated with severe disease. In Africa the adhesion phenotype of rosetting is associated with severe malaria [
14] and increased expression of group A
var genes [
19,
21,
25]. Some group A and B PfEMP1s can bind to intercellular adhesion molecule 1 (ICAM-1) [
30,
31], and ICAM-1 expression was up-regulated in brain endothelium and co-localized with sequestered IEs in cerebral malaria patients [
32]. IE adhesion to ICAM-1 has variously been associated with cerebral malaria [
33], clinical but not severe malaria [
34] or inversely correlated with severe disease [
35].
Another phenotype associated with severe disease is adhesion to endothelial protein C receptor, EPCR [
36,
37]. Parasite isolates from African children with severe malaria bound EPCR and expressed DC8 or DC13
var genes [
36,
38]. DC8 and DC13 PfEMP1s are primarily group B and A, respectively [
4], and contain members of the subset of CIDRα1 domain types, which bind EPCR [
36,
37]. Sera from African children with uncomplicated malaria recognize PfEMP1s containing DC8 and DC13 at higher levels than PfEMP1s without DC8 or 13, but it is unclear whether severe malaria specifically induces antibodies to DC8 and DC13 [
39,
40].
DC5 PfEMP1s, which are nearly all group A, are recognized by sera from semi-immune children in a similar manner to other severe malaria associated isolates [
27], and DC5 expression increased markedly during infection of a naive volunteer [
41]. African children acquired antibodies to DC5 more rapidly than to other PfEMP1 domains which is consistent with widespread expression of DC5 in non-immune individuals [
29]. Antibodies reactive with DC5 also correlated with protection from malaria episodes [
42]; however, the evidence directly linking DC5 PfEMP1 expression and adhesion phenotype to severe malaria is less clear. High levels of DC5 sequence expression have been detected in severe malaria, but only together with expression of either DC8 or 13 sequences [
38] and it is conceivable that these DC5 and DC8 or DC13 sequences were present on the same
var genes. DC5 PfEMP1s bind platelet-endothelial cell adhesion molecule 1 (PECAM1) [
43] and IE adhesion to PECAM1 has been implicated in cerebral malaria [
44,
45]. However, IE adhesion to PECAM1 is also commonly found in samples from patients with uncomplicated malaria [
46] and DC5 PfEMP1s were not expressed by parasites selected for adhesion to brain endothelium [
39,
40]. A minority of
var genes containing DC5 do encode EPCR-binding CIDRα1.5 domains although these domains are not part of the DC5 cassette [
4].
Thus several promising candidates have emerged as members of the restricted population of PfEMP1s responsible for severe malaria, but the relative contributions of group A and B, and of DCs 8, 13 and 5 remain unclear. In particular very little is known about PfEMP1s in severe disease in the Asia Pacific region. Determining whether conserved PfEMP1sequences elicit protection from severe malaria disease globally is a priority for vaccine research. In this study plasmas from Papuan patients with severe or uncomplicated malaria were analysed for their reactivity with PfEMP1 polypeptides representative of the different groups and DCs including several expressed by parasites causing cerebral malaria.
Statistical analyses
The association between disease severity with age and parasitaemia was assessed using a Mann–Whitney U-test and with gender using a Fisher’s exact test. RU values for individual proteins were compared by Mann–Whitney U-test. To compare between patients with severe and uncomplicated malaria for antibody responses to proteins belonging to a single domain cassette or PfEMP1 group, patients were categorized according to whether their plasma sample lay above or below the median concentration of RU for that antigen: those above or equal scoring 1, or 0 if below. To derive a single quantitative score for each plasma for all the proteins belonging to a single DC or PfEMP1 group, the plasma’s scores for each antigen in that DC or group that were determined by either Luminex or ELISA were summed.
Any individual plasma with a score of 1 for any protein within a DC or PfEMP1 group was classified as a responder to that DC or PfEMP1. Individual plasma samples with a combined score of 0 for all proteins within a DC or group were classified as non-responders.
Differences in the proportions of severe and uncomplicated malaria patients whose plasma responded to a DC or PfEMP1 group were compared by contingency tables using Fisher’s exact test. Differences between severe and uncomplicated malaria patients in the number of proteins within a DC or PfEMP1 group to which patients responded were compared only for patients who responded to at least one protein in the DC or PfEMP1 group using Mann–Whitney
U- tests. This indicated differences in the breadth of the response to PfEMP1s within that group. The patients that did not respond to any protein within the group, i.e. had a score of zero were not included to remove any biases associated with large frequencies of zero values in non-parametric comparisons [
57].
Discussion
Sequencing the transcriptome of parasites causing CM in a single patient enabled assembly of a snapshot of the transcribed
var repertoire in human malaria. Although entire genes could not be assembled, advances in the phylogenetics of
var sequences [
4] allowed the
var contigs to be separated into useful classifications. Previous studies that implicated parasites expressing group A PfEMP1s [
20,
23‐
25] and group B PfEMP1s [
26] in severe disease provided no, or minimal, sequence data and were essentially restricted to classifying sequences to the groups defined by the
var gene upstream sequences or by sequencing short DBLα tags.
The RNAseq of parasites causing CM in this PNG patient was consistent with previous studies of
var genes in pathogenesis and abundantly expressed
var genes identified included DC5 and possibly DC8, the DBLβ3 domain subtype and individual domains including CIDRα1.5 and DBLβ3. Recombinant proteins derived from the latter two domain subtypes have been shown to bind EPCR and ICAM-1 respectively [
30,
36]. Other abundantly transcribed domain subtypes had not previously been identified in severe malaria. This limited study of a single patient indicates that RNAseq will be useful for identifying quantitative differences between transcribed
var genes in severe disease in future studies.
The diversity of the transcribed
var repertoire was consistent with a previous report of cerebral malaria in Africa [
65]. However, 45 % of the reads that assembled into
var transcripts were in the 20 most abundant
var contigs that between them represented 27 domains. Thus the quantitative nature of RNAseq revealed a hierarchy of
var transcript abundance in this patient’s peripheral blood that would be difficult to detect using the non-quantitative, nested RT-PCR approaches available to this previous study [
65]. This suggests that the dominant
var transcripts expressed by parasites causing cerebral malaria in a single patient are probably restricted in number.
Patients with uncomplicated malaria more commonly had antibodies to PfEMP1s that were from Group C or were not from DC5 nor DC8 nor DC13 than patients with severe malaria (Fig.
3a). In contrast, similar proportions of patients with severe and uncomplicated malaria had developed antibodies to the severe malaria associated PfEMP1s (group A and B, DC5, 8 and 13) (Fig.
3a), but the breadth of the response to group A and B PfEMP1s was greater in patients with uncomplicated than severe disease (Fig.
3b). Thus susceptibility to severe disease was associated with recognition of a narrower range of group A and B PfEMP1s and to an overall lack of antibodies to group C PfEMP1s and to PfEMP1s that were not DC5 nor DC8 nor DC13.
Overall, the serology findings in this Papuan adult population are consistent with existing models of infection in African children where parasites expressing severe malaria-associated group A and B PfEMP1s infect naive individuals and elicit antibodies [
20,
66]. Susceptible, semi-immune individuals have antibody to some group A and B PfEMP1s [
27], but protective immunity correlates with acquisition of antibodies recognizing a broader range of PfEMP1s [
5‐
8]. Parasites expressing uncomplicated disease associated group C PfEMP1s, or PfEMP1s that were not DC5, 8 nor 13, would only dominate infections after parasites expressing severe disease associated PfEMP1s were controlled by acquisition of a broad antibody response. The alternative explanation is that group C PfEMP1s and PfEMP1s that were not DC5, 8 nor 13 were abundantly expressed by parasites causing acute, severe malaria but had not yet elicited antibodies. Although this cannot be excluded it is inconsistent with previous studies of
var gene expression in both Africa and PNG [
20‐
26].
Authors’ contributions
MFD conceived the study, performed molecular, serological and bioinformatic analyses and drafted the manuscript. RN helped design the study, designed and organized sample collection and performed molecular studies. TT and ET expressed recombinant proteins. BSF and LT organized sample collection and performed molecular studies. JM, DAL, RNP and NMA helped design the study and organized sample collection. FS performed molecular studies. LT, TL and TG helped design the study, expressed proteins and designed serological analyses. FJIF performed statistical analyses. ZPF performed bioinformatic analyses. SJR and GVB helped design the study and draft the manuscript. PS helped organize sample collection. ATP helped design the study and performed bioinformatic analyses. All authors read and approved the final manuscript.