Plasmodium falciparum malaria is a life-threatening parasitic disease that kills some half a million people each year, mostly children living in sub-Saharan Africa. The outcome of the infection, which may be asymptomatic, symptomatic, or life-threatening, is known to be influenced by parasite, host, environmental and socioeconomic factors [1]. In Gabon, respiratory distress and cerebral malaria represent, respectively 31% and 24% of severe falciparum in children [2]. However, the molecular mechanisms underlying this variable pathogenicity are unclear. A combination of the parasitized red blood cell (pRBC) binding (cytoadhesion, agglutination and rosetting), the host environmental and inflammatory response, endothelial cell (EC) activation, and altered hemostasis can lead to blood pulmonary/brain barrier impairment [3]. Cerebral malaria accounts for a significant proportion of malaria mortality and is associated with sequestration of pRBC in brain microvessels, especially pRBC expressing PfEMP-1 domain cassettes 8 and 13 [4, 5]. It has been also demonstrated that pRBC-EC adherence induces caspase 8, 9 and 3 activation, EC apoptosis, and modulates EC expression of TNF-α superfamily genes (Fas, Fas L, DR-6) and apoptosis-related genes (Bad, Bax, Caspase-3, SARP 2, DFF45/ICAD, IFN-γ Receptor 2, Bcl-w, Bik and iNOS) [6]. pRBC-EC adherence also leads to microvessels blockage, hypoxia, proinflammatory cytokine secretion [6–9], EC junction modifications, and EC barrier permeabilization [8, 10, 11].

pRBC-triggered EC apoptosis in the brain, lungs and kidneys has been observed in patients with fatal malaria [12, 13]. Using cocultured human lung EC (HLEC) and pRBC field isolates from Gabon, we showed that although almost all isolates provoked cytoadherence, only a few triggered apoptosis (via cytoadherence or diffusible factors), and that these isolates tended to have been drawn from patients with neurological disorders [9]. The potential of pRBC to trigger HLEC apoptosis varies according to the isolate, and is associated to the expression of Plasmodium apoptosis-linked pathogenicity factors (PALPFs) [9, 14, 15]. Human endothelial cells in different tissues may not all express the same surface receptors [16], and this could explain some functional differences. For example, CD36 is expressed by HLEC [17], but not by human cerebral microvascular endothelial cells-D3 (hCMEC/D3) (N’dilimabaka et al. unpublished data). Parasite ligands that specifically recognize CD36 are able to stimulate HLEC. ICAM-1 is expressed on the surface of both HLEC and hCMEC/D3 [17, 18], and P. falciparum isolates that express surface ligands specifically recognizing ICAM-1 stimulate both cell types. P. falciparum/EC cocultures are widely used to study the pathophysiology of malaria [6, 8, 9, 11, 15, 19–21]. Several types of EC have been used in these models, including lung [17], brain EC [18, 21]. These coculture models can be used: a) to study parasite sequestration (cytoadhesion, agglutination, rosetting, etc.), b) to detect damage (apoptosis) caused to EC, c) to categorize and quantify pRBC-EC interactions (cytoadhesion with or without apoptosis), and d) to study the molecules involved in these interactions. We postulated that different EC types might interact differently with pRBC.

Recently, it was shown that pRBC-induced apoptosis of lung and cerebral EC varies with the in vitro-adapted P. falciparum line [21]. Selected parasite characteristics are modified after adaptation to in vitro culture, and the behavior of field isolates in the same conditions must be determined. We performed ex vivo coculture experiments with P. falciparum-pRBC isolated from malaria patients, and assessed the capacity of the parasites to cytoadhere and to induce apoptosis of both HLEC and hCMEC/D3.

P. falciparum isolates were collected from 25 patients with uncomplicated malaria and 2 patients with severe malaria (1 with convulsions, 1 with prostration). Median age was 48 months (range 8 mouths to 20 years), and mean age was 62.0 ± 55.8 months. Parasite load ranged from 5316 to 315,000 parasites per microliter. All 27 isolates were matured in vitro and enriched by gelatin flotation. pRBC specimens were almost exclusively at the schizont stage.

Endothelial cell susceptibility

Seven of the apoptogenic isolates induced both cerebral and pulmonary EC apoptosis, 8 induced only hCMEC/D3 apoptosis, and 1 induced only HLEC apoptosis. Finally, 15 of the apoptogenic isolates induced hCMEC/D3 apoptosis, and 8 HLEC apoptosis (Fig. 2). Also, the mean OD of these 15 apoptogenic isolates targeting hCMEC/D3 is higher than that of the 8 apoptogenic isolates targeting HLEC apoptosis (6.25 ± 2.22 vs 4.12 ± 1.04 respectively; p < 0.01). Thus, cerebral ECs were more susceptible than lung ECs to pRBC-induced apoptosis.

Half of the apoptogenic isolates triggered only hCMEC/D3 cell death (mean OD = 5.8 ± 2.1). Four of these isolates acted via soluble factors, one via direct contact without cytoadherence, two via both contact without cytoadherence and soluble factors, and the last by cytoadherence. One isolate specifically induced HLEC apoptosis (OD = 3.53) via diffusible soluble factors. All the isolates able to kill HLEC (OD range 3.4 to 6.4) in contact coculture were also able to kill hCMEC/D3 in contact coculture (OD range 3.0 to 11.4). HLEC apoptosis was induced by contact or by diffusible factors, but never by both. For example, F26, F28, F29 and F31 pRBC induced HLEC apoptosis via soluble factors but not by direct contact or cytoadhérence.

Knowledge of the nature and role of parasite ligands and host cell receptors associated with malaria severity is crucial for improving or developing therapeutic strategies that combine both parasite elimination and endothelial cell protection. Here, to limit possible artifacts linked to parasite adaptation to in vitro culture, we used field isolates with a minimum of ex vivo manipulation. Coculture of RBC infected by these isolates with hCMEC/D3 and HLEC was carried out in contact and non contact conditions to assess the capacity of RBC infected by these isolates to bind to ECs and to induce apoptosis, and also to determine the general mechanism of action (contact, cytoadherence and/or soluble factors). RBC parasitized-3D7 strain (in vitro maintained culture) and uninfected RBC were used respectively as positive and negative controls for both adherence and apoptosis assays as previously documented [6, 9, 14, 15, 21, 26]. The use of uninfected RBC as negative control instead of non-adherent P. falciparum lines is more linked to the scarcity of valuable non cytoadherent parasite line such as P. falciparum D10 described by Anders et al. in 1983 [27] and this does not have any influence on our results. pRBC-mediated HLEC apoptosis has been already investigated by means of transmission electron microscopy, annexin V assay, caspase activity assay, and nucleosome release ELISA [6, 21]. The ECs used in this study were first examined for ICAM-1 and CD36 expression. The hCMEC/D3 expressed ICAM but not CD36, while HLEC expressed both receptors.

We found that 59% of the pRBC specimens triggered EC apoptosis, confirming that pRBC clinical isolates can induce lung [9, 15, 21] and brain EC apoptosis [21, 28]. However, in previous studies only about 20% of clinical isolates induced human lung EC apoptosis, compared to 59% of our isolates [9, 15].

The main finding of this study is that P. falciparum induced preferential EC apoptosis: half of the 16 apoptogenic isolates specifically targeted hCMEC/D3 cells, while only one specifically targeted HLEC, and seven targeted both cell types. Whether these cell types differ in their intrinsic susceptibility to apoptogenic stimuli was not assessed and so it is uncertain whether these findings can be generalized to the situation in vivo. However, if a similar difference in susceptibility to P. falciparum-induced apoptosis occurs in vivo, this might be important in the pathogenesis of severe malaria. Overall, 15 (94%) of the 16 apoptogenic isolates triggered brain EC apoptosis, but only 8 (50%) killed HLEC (p < 0.05). Three pRBC phenotypes were observed: the first exclusively killed hCMEC/D3, the second only killed HLEC, and the last killed both ECs (mixed phenotype). Evidence of EC activation and apoptosis leading to altered vascular integrity and blood pulmonary/brain barrier breakdown, has been found during severe malaria [6, 8, 12, 29]. We suspected that patients with severe malaria might be infected with strains preferentially inducing either brain or lung EC apoptosis, but almost all our patients had uncomplicated malaria, meaning we were unable to test this hypothesis.

In contact experiments, 9 isolates induced both HLEC and hCMEC/D3 apoptosis. Of these, 4 isolates acted via cytoadherence. All 4 of these latter isolates (F15, F30, F32 and F40) were able to bind to hCMEC/D3 and HLEC. This is in line with two studies using field isolates in which as many as half of the pRBC samples that triggered EC apoptosis acted through cytoadherence [9, 15].

The transduction signal triggered by pRBC adherence is unclear, but cross-linking of pRBC adhesins on the EC surface is known to induce apoptosis [30]. pRBC binding specifically to CD36 induces EC apoptosis mediated by p59/fyn as well as by caspases [31]. Working with P. falciparum field isolates from Gabon, we have previously demonstrated that the ability of pRBC to trigger human lung EC apoptosis is linked to the expression of genes encoding PALPF [14]. We also demonstrated that selected PALPF transcripts, such as PFD0875c and MAL13P1.206, are involved in parasite cytoadherence [14]. As some isolates can induce hCMEC/D3 apoptosis by cytoadherence, these PALPFs might be involved in hCMEC/D3 apoptosis. This hypothesis is supported by previous work showing that, in brain EC coculture, some PALPF transcripts, particularly PALPF-5 and PALF-2, are up regulated in apoptogenic strains by comparison with non apoptogenic strains [21]. Binding of these PALPFs or other parasite ligands to EC-CD36 and EC-ICAM-1 may activate the Rho kinase signaling pathway that can trigger cell apoptosis. Indeed, CD36 and ICAM-1 are the receptors most commonly used by clinical isolates to survive, even though pRBC adhesion involves many other receptors [32–34].

Our results conflict with those of N’Dilimabaka et al. who found that pRBC binding was not required for EC apoptosis. These authors used P. falciparum field isolates adapted in vitro, whereas we used pRBC freshly collected from malaria patients. Field isolates maintained in vitro undergo clonal phenotypic variations every 48 h, corresponding to intra-erythrocytic development. This phenotypic variation, together with stress generated by in vitro conditions, may influence parasite gene expression and other characteristics.

Among the nine isolates that were apoptogenic in contact experiments, five (F1, F2, F26, F33 and F37) acted by physical contact. Four isolates (F1, F2, F26 and F33) specifically killed hCMEC/D3 while isolate F37 killed both EC types. None of these five last isolates cytoadhere to the EC they killed. This confirms that direct contact between pRBC and EC, without cytoadherence, is sufficient to trigger EC apoptosis [21]. Five of the 16 apoptogenic isolates in our study required direct contact to induce EC apoptosis. This physical contact might induce over-expression of PALPF genes different from those associated with cytoadherence. PALPFs consist mainly of trans-membrane proteins, few of which are involved in cytoadherence [14, 15].

Two-thirds of the apoptogenic pRBC samples killed hCMEC/D3 via soluble factors, compared to only one-quarter for HLEC. Similar results have been obtained with HLEC by Zang Edou et al. 2010. Isolates F29 and F31 were able to kill both ECs. Interestingly, seven apoptogenic isolates killed hCMEC/D3 or HLEC exclusively via soluble factors, and were unable to induce EC apoptosis when placed in direct contact. It appears that both mechanisms (cytoadherence and direct contact) are inhibited. Our results suggest that coculture contact inhibits the expression of genes encoding PALPF soluble antigens in these pRBCs. By contrast, two-thirds of the pRBCs inducing apoptosis by cytoadherence or direct contact were also able to trigger apoptosis via soluble factors. Perhaps all the genes coding for PALPF proteins (adhesins and soluble antigens) were expressed in these latter isolates conferring therefore the capacity of using more than one stimulus.

Finally, we suggest that the three stimuli may trigger the same signaling pathway. PALPF adhesins and soluble antigens expressed by pRBC are able to trigger EC apoptosis via Rho kinase activation. However, the mechanisms by which some pRBCs preferentially target brain versus lung ECs are unknown. The cerebral associated isolate (CAI) (a phenotype killing only hCMEC/D3) express PALPF antigens, the nature and expression kinetics of which are different to those of PALPF expressed by pulmonary associated isolate (PAI) (a phenotype specifically killing HLEC). In this case, the nature and/or affinity of the EC receptors recognized by these PALPF ligands would be also different.

P. falciparum uses at least one of the three stimuli (cytoadherence, direct contact and soluble factors) to induce EC apoptosis. The study indicates that brain EC was more susceptible to apoptosis triggered by P. falciparum compared to the lung EC and may provide new insights into host-parasite interactions.