Background
Plasmodium falciparum the causative agent of tropical malaria still kills over one million people per year, mostly children under five years in sub-Saharan Africa [
1]. One major virulence factor of this parasite is the highly variant
Plasmodium falciparum erythrocyte membrane protein-1 (PfEMP-1) family [
2]. Members of this family appear at the surface infected red blood cell, during the trophozoite stage. Their large, highly variable ectodomains, consist of several subdomains (termed Duffy binding-like, DBL, and cystein rich interdomain regions, CIDR), which interact with a number of cellular host receptors such as CD36, ICAM-1, CSA, PECAM, E-Selectin, VCAM and others (reviewed by [
3]). This interaction causes the retention of mature blood stage forms in the tissue which express the cited receptors and is believed to result in pathogenic processes associated with malaria [
4]. Each infected red blood cell (IRBC) displays only one PfEMP-1 allele at its surface and this selective expression is controlled at the transcriptional level [
5] by allelic exclusion [
6,
7]. It is believed that only one of the approximately fifty
var promoters is active and concomitantly localized to a specific compartment at the nuclear periphery [
8‐
10]. The transcriptional activity of
var promoters is apparently determined by histone modifications [
11‐
14], which may then permit or inhibit the recruitment of yet largely unknown transcription factors. The pattern of
var transcription is inherited through several generations [
15,
16] despite the absence of DNA methylation [
17] and switches in the transcription result in altered antigenic phenotypes.
Several groups tried to correlate disease outcomes with adhesive IRBC phenotypes, with sometimes contradicting outcomes for ICAM-1 and/or CD36-mediated adhesion [
18,
19]. The excessive CSA-mediated adhesion of IRBC was clearly correlated to severe malaria outcomes in primigravids [
20] and the PfEMP1
csa-encoding var2csa gene, upregulated in CSA-adhesive parasites, was analysed in detail [
21] and the structure of its crucial DBLγ domain mapped [
22].
So far, few studies associated the
var genes to adhesive phenotypes in cultivated field isolates. One Indian isolate was obtained after enrichment of an ICAM-1-adhesive phenotype [
23]. Parasites obtained from the placenta of primigravid women frequently transcribe var2csa and adhere to CSA found in abundance on syncithiotrophoblasts [
24]. The primary
var transcript sequence can be of importance for the design of specific PfEMP-1 domains that may be used as a vaccine inducing antibodies, which then inhibit or decrease cytoadherence of IRBCs, as shown for CD36-binding CIDR [
25] and CSA-binding DBLγ domains [
26]. Also, for the elucidation of the epigenetic control mechanisms that orchestrate
var gene transcription it may be of interest to reproducibly obtain parasites with a defined
var locus switched on or off in its native, unmodified genomic context. The enrichment of adhesive phenotypes is achieved by the panning procedure, where cytoadherent IRBCs are incubated either on isolated receptor molecules [
15] or on receptor-expressing cells, such as amelanotic melanoma cells expressing CD36 [
27], Saimiri brain endothelial cells (SBEC) expressing CSA and/or CD36 and ICAM-1 [
28], human lung endothelial cells (HLEC) expressing CD36 and ICAM-1 [
29], human choriocarcinoma cells expressing CSA and CD36 [
30] or Chinese hamster ovary cells (CHO) transfected with the respective receptors [
31]. After several rounds of panning and amplification of the parasites a sufficiently high homogeneity of the cytoadherent phenotype is obtained which then permits functional or transcript analysis. The CHO cell lineages are especially convenient since they express the receptors in a constitutive fashion without the need of external stimulation as it is the case for SBEC or HLEC. However, the CHO cell lineage has the disadvantage that there is a – to date unknown – receptor which is recognized by equally unknown parasite encoded PfEMP1 or other molecules which may mask the desired receptor-ligand interaction [
32]. In order to estimate the viability of the transfected CHO cell lines in the enrichment of relevant adhesive phenotypes and later transcript analysis,
var gene transcription was monitored of different
P. falciparum 3D7 parasite lines panned on receptors expressed on stably transfected CHO-cell lines [
31,
33].
Var transcripts were then analysed by reverse transcription followed by quantitative real time PCR (RT-qPCR). In addition, we also evaluated if the frequently used approach of reverse transcription followed by PCR, fragment cloning and sequence read counting (RT-PCRcsc), is reliable to estimate dominant
var transcripts in
P. falciparum. This issue is of major importance since field isolates can currently be tested for their
var transcripts only by this approach.
Discussion
In the present study, a practical and simple system for the analysis and selection of
P. falciparum adhesive phenotypes and their
var transcripts was evaluated. This system was so far used only for the diagnosis of cytoadherent
P. falciparum isolates [
31,
44], but not for the deliberate selection and enrichment of adhesive phenotypes for posterior ligand-receptor interaction or other analyses. Most of the herein used receptors are also commercially available as purified molecules, however, either these molecules are so expensive that they turn the use in quantitative selection experiments unviable or they are not displayed properly for recognition upon coating of supports such as Petri dishes or culture flasks. Also, in the case of ICAM-1 and CD36, free ICAM-1 or CD36 molecules do not detach IRBC interacting with these molecules on the cell surface, although interaction may be abrogated by specific antibodies against ICAM-1 or CD36 [
48]. Therefore, the selection of parasites by specific detachment of adhered parasites, as it is feasible for the interaction of IRBC with CSA expressed on
Saimiri brain endothelial cells [
49], proved impossible.
A major handicap of the herein used system is the yet unknown receptor present on CHO-745 cells which is readily recognized by 3D7 parasites as previously described [
32]. These authors demonstrated a phenotype which could be inhibited by anti-CD36 monoclonal antibody, indicating that CHO-745 may carry a CD36-like receptor on its surface. They also showed that protein A pretreated parasites were significantly cytoadherence-inhibited (although adhesion was not totally abrogated) indicating also an interference with opsonizing IgM antibodies. Since the binding ligand of 3D7 IRBC selected on CHO-745 was trypsin sensitive, the hypothesis was that the interaction IRBC to CHO-745 may at least partly be caused by PfEMP1 and anticipated the detection of specific
var transcripts, which encode the PfEMP1 responsible for this unknown receptor. Considering the existence of this unknown receptor and supposedly uniform transgene-receptor expression on CHO-CD36, CHO-VCAM, CHO-ICAM-1, CHO-Selectin and CHO-K1 (not specifically tested for herein), the model will only be able to select for strong binding ligands, since IRBC will otherwise be selected for the unknown receptor. Its appearance in all but one selected 3D7 parasite line suggests the possibility that the unknown receptor is recognized by PfEMP1 encoded by PFD0995c/PFD1000c.
Upon analysis of the transcription profile of 3D7 selected on CSA-expressing CHO-K1 cells, no specific increased transcript abundance of var2csa was observed. In contrast to the FCR3 strain, the 3D7 parasite line was only poorly selectable on purified fixed CSA (data not shown and [
21]). In a panning experiment using purified CSA, the 3D7 strain also showed decreased affinity for this receptor in comparison to other strains [
50]. This possibly explains why no var2csa upregulation was observed, and the
var transcription profile was somehow similar to the CHO-745-panned parasites. Other authors obtained the var2csa transcribing phenotype by the selection of NF54 which is isogenic with 3D7 [
21,
51]. A similar effect was observed for the CHO-VCAM phenotype, although it is unknown if 3D7 encodes a competent ligand for this molecule. To date, no
var gene in any strain was associated to VCAM-adhesion, although this receptor seems recognizable by PfEMP1 [
52]. Strong ligands with high binding affinities are expected to override adhesion to the unknown receptor. In the case of 3D7, CD36 seems to be readily recognized and the CHO-CD36 adherent parasites contained very low levels of the PFD0995c/PFD1000c transcript. On the other hand, 3D7-CD36 parasites also adhered to CHO-745 cells, as seen already by other groups [
32], without transcribing PFD0995c/PFD1000c. The most abundant
var transcript in this parasite line was PFD0615c. Although never functionally examined, this gene contains a CIDR domain which is grouping with other CIDR domains that are in fact CD36 binder [
43]. This indicates that PFD0615c encodes a stronger binding CIDR domain than PFD0995c/PFD1000c, considered as a moderate CIDR-binder [
43]. The cytoadherence of 3D7-CD36 to each other cell line including CHO-745 may be explained by the assumption that the recognized although unknown receptor on CHO-745 is CD36-like as proposed by Andrews and colleagues [
32].
E-Selectin was detected as a receptor for cytoadherence [
52], yet, no isolate or strain was specifically selected for this specific phenotype. Also, cytoadherence to this receptor seems to be a rare event [
53] and was never clearly identified as the cause of severe malaria, although E-Selectin in its soluble form is increased in acute phase malaria [
54]. It is also unknown which PfEMP1-domain mediates cytoadherence to E-Selectin. In our studies, the
var transcript PFD1015 was clearly associated to CHO-E-Selectin adherence. In addition the
var gene PFD1000c representing the supposed ligand for the unknown receptor of CHO-745 cells was also present, explaining the cross-adherence to CHO-745 and other CHO cells. The cytoadherence to CHO-ICAM-1 showed that transcript PFD0625c was found in elevated levels. Since ICAM-1, as well as CD36, is a relative abundant receptor in man, a number of competent PfEMP1 molecules are expected to be present in the 3D7 parasite line. Indeed, a couple o
var genes were found transcribed together with the CHO-745 specific transcripts PFD0995c/PFD1000c. The
var gene PFD0625c, however, contains no DBLβ/C2 domain important for ICAM-1-binding [
55,
56], indicating that this
var gene may not encode the ICAM-1-recognizing PfEMP1. Interestingly, other authors found a comparable result when analysing
var transcripts by microarray analysis. Upon multiple pannings on purified ICAM-1 the most transcribed
var gene detected did not contain a DBL-β/C2 domain [
50]. In another study, still another
var gene than PFD0625c was found in 3D7 parasites selected for ICAM-1 adhesion [
57]. This indicates that it is not clear whether the CHO-ICAM-1 adhesive phenotype really binds to human ICAM-1.
The relative transcript quantities against their corresponding upstream regions were then clustered [
58]. Most of the transcripts in selected parasites were from upsC and upsB promoters and only in minor quantities transcripts from subtelomeric locations were observed, even when discarding the dominantly found PFD0995c/1000c transcript. It is possible that strong binding PfEMP1-coding
var genes are found in centromeric localizations, while weaker binding PfEMP1 molecules are encoded by subtelomeric
var genes, prone to ectopic recombination and rearrangements [
59]. Clearly, the unambiguous identification of ligands and receptors is necessary and may confirm or refute this hypothesis.
The analysis of transcripts by the to-date unique applicable system of transcript analysis in field isolates showed a significant discordance to the results obtained in RT-qPCR. Of the two main transcripts that were highly represented in the CHO-Selectin binding parasites, only the PFD1000c transcript was detected by clone analysis. A similar result was shown before using even another primer pair binding upstream of the universal
var oligonucleotides [
47]. In a similarly designed study, Gatton and coworkers also postulated that the PFD1015c DBLα fragment may not be amplified by the universal
var oligos [
60]. It is puzzling why the PFD1015 DBL1α tag was not even cloned once, since it has perfect target site for the universal oligonucleotide pair used herein. In more than 500 sequence reads from diverse 3D7 cDNAs, the PFD1015
var DBL1α tag was encountered only once. Notably, in other studies amplifying either genomic or transcribed targets [
41,
57,
61] the PFD1015c tag was also never observed. The discrepancy of the two approaches was also observed in other adhesive phenotypes, however, the cDNAs used in these experiments were not from the same lot as the ones tested herein by qRT-PCR (Table
2). Assuming that RT-qPCR using the described primer set is the most reliable experimental procedure to study
var gene transcription in 3D7, the RT-PCR/cloning approach is highly misleading and not reflecting true transcript quantities, at least in the conditions used herein. This implies that data from field samples analysed by this method may also not display the true transcript levels in the corresponding parasites. There is no simple explanation for these differences between the two methods. Either, certain quantities of 3'-truncated transcripts are present, which can be reverse transcribed and amplified by DBL1α specific primers. However, two groups recently demonstrated that there is no "relaxed" transcription in ring stage parasites [
5,
62] and thus no significant accumulation of 3'-truncated transcripts. On the other hand, it may be speculated that an existing primer bias is exacerbated upon amplification of cDNA due to secondary structures interfering with the processivity of the Taq polymerase. Taken together, our data suggest caution when trying to correlate
var transcripts detected by RT-PCRcsc from field isolates to clinical outcomes. The selection of parasites on transfected CHO-cells revealed reproducibly different patterns of
var transcription in dependence on which CHO cell line was used which may be used in further analyses of receptor-ligand studies or, for example, for the elucidation of transcriptional activation or silencing of
var loci without the need for transfection of artificial
var promoters.
Table 2
Sequence read distribution in cDNA produced from 3D7 parasites panned on CHO-745, CHO-K1, CHO-CD36 and CHO-VCAM.
PF07_0051 | | | 1/29 | 1/17 |
PF08_0107 | 3/30 | | | |
PF10_001 | | | 1/29 | |
PFD0615c | 16/30 | 7/16 | 25/29 | 1/17 |
PFD0635c | 1/30 | | | |
PFD0995c | | | | 3/17 |
PFD1000c | 3/30 | 1/16 | 1/29 | |
PFD1015c | | 1/16 | | |
PFF0845C | 6/30 | 7/16 | | 1/17 |
PFF1580c | | | 1/29 | |
PFL0020w | 1/30 | | | 1/17 |
PFL1970w | | | | 10/17 |
Authors' contributions
UG and LA carried out the panning of parasites, the RT-qPCR and transcription analysis. UG, LA and GW conceived the design of the study, wrote and approved the final manuscript.