Adhesion of Plasmodium falciparum infected erythrocytes in ex vivo perfused placental tissue: a novel model of placental malaria

Pregnancy-associated malaria is a major cause of morbidity among pregnant women and their offspring in Plasmodium falciparum endemic areas. Plasmodium falciparum infected erythrocytes sequester in the intervillous space, causing placental malaria. Pregnancy-associated malaria is associated with placental intervillositis, maternal anemia and low birth-weight [1]. Current measures to protect pregnant women from pregnancy-associated malaria are insecticide-treated bed nets, intermittent preventive treatment in pregnancy and treatment of infections [2]. However, pregnancy-associated malaria is often asymptomatic and may occur before the first antenatal visit [3]. Increasing resistance to anti-parasite and anti-mosquito drugs along with changed vector behaviour is reducing efficacy of current protective measure for pregnant women. Parasites express P. falciparum erythrocyte membrane protein 1 (PfEMP1) on the surface of infected erythrocytes, mediating cytoadhesion to endothelial cells, platelets, erythrocytes and syncytiotrophoblast, thereby evading circulation and destruction in the spleen. VAR2CSA, a unique member of the PfEMP1 protein family, was discovered in 2003 [4], since then a large base of evidence supports the causal relationship between VAR2CSA and placental malaria [5‐12]. Parasites infecting pregnant women bind to chondroitin sulfate A (CSA) [13] and recombinant VAR2CSA bind with high affinity to CSA [14‐16]. However, binding to immunoglobulin and hyaluronic acid have also been associated with placental malaria [17‐19]. Furthermore, it is not known whether parasites binding to receptors other than CSA can accumulate in the placenta as such parasites are restricted by immunity, since women in malaria endemic regions develop protective antibodies during childhood. Interaction with multiple receptors may have implications for how the pathology manifests during infections, but also for the development of a vaccine to induce antibodies that inhibit the binding of infected erythrocytes to placental tissue. This is an important question in areas of reduced malaria prevalence, as less exposure to malaria in childhood may affect development of protective immunity, leaving women more susceptible to infection when they reach reproductive age. Currently, adhesion blocking capacity of antibodies has largely been tested in assays where only one receptor, namely CSA, is present [20‐22], however the efficacy of such antibodies may be limited if sequestration occurs by other pathways. Recent work have shown that human placental and cancer cells express a distinct form of chondroitin sulfate, that is not present in other normal human tissue [23]. Interpretation of binding assays using bovine CSA is, therefore, a major concern, as VAR2CSA-expressing infected erythrocytes are likely to bind with higher affinity to placental CSA. Studies of the mechanisms of placental sequestration have used placental tissue cryosections, however, these studies are contradictory, as they have demonstrated both exclusive CSA dependence and involvement of immunoglobulin binding [18, 24, 25]. Some of these differences may have been incurred by the fixation of tissues that can damage secondary protein structure resulting in alteration of important epitopes and/or receptors.

There are established models in which adhesion of infected erythrocytes is studied under homogenous flow conditions [20, 24]. Although important knowledge of parasite adhesion can be derived from these models, they do not simulate the complex flow through the villous tree in the intervillous space. Modelling of the intervillous blood flow suggests that there is a pressure gradient depending on the spacing between spiral artery inflow and venous outflow, vessel caliber, and maternal blood flow rate [26].

Establishment of in vivo murine models is difficult as murine malaria parasites do not possess PfEMP1 s [27] or knob structures [28] although it has previously been shown that Plasmodium berghei infected erythrocytes accumulate in the placental tissue of BALB/c mice due to low blood flow [29]. Furthermore, a var2csa transgenic P. berghei murine placental model has recently been developed [30]. Plasmodium falciparum infection models in Aotus monkeys are currently being developed but they are also associated with considerable costs and difficulties. Buffet et al. described an ex vivo perfusion model utilising human spleen to investigate clearance of drug treated P. falciparum infected erythrocytes [31]. In this paper, adaptation of a human placental perfusion model that replicates natural infected erythrocyte binding to intact human placental tissue under flow conditions close to the physiological placental flow is described. The dual ex vivo placental perfusion model, where the maternal and fetal circulation is re-established in a human placental cotyledon, has been used extensivley for many years to study placental physiology, such as transfer of nutrients and IgG across the placental barrier [32, 33]. Placental perfusion has shown the role of IgG in mediating merozoite surface protein-1 antigen transfer from mother to fetus [34]. The model has also been used to study placental infections with enterovirus [35] and cytomegalovirus [36].

A series of 15 perfusions were performed and several steps were taken to optimize the model for perfusion with P. falciparum infected erythrocytes (Fig. 1).

The existing knowledge of parasite sequestration in placental malaria originates from field studies and laboratory models showing that parasites from pregnant women bind to CSA [13], that sera from pregnant women previously exposed to placental malaria inhibit parasite binding to CSA [55], that anti-VAR2CSA antibodies are predominantly acquired by women exposed to malaria during pregnancy and rarely found at high concentration in malaria exposed men or children. Moreover, high levels of anti-VAR2CSA IgG in pregnant women are associated with favourable birth outcomes [8]. The occurrence of placental malaria in pregnant women in malaria-endemic areas that have developed protective antibodies towards a wide array of blood stage antigens during childhood underlines the importance of acquiring antibodies, most likely anti-adhesive, towards PfEMP1 molecules. Furthermore, binding to receptors other than CSA has been associated with placental malaria [17‐19]. This could have important implications for the etiology of pathology and for the development of vaccines that aim to reduce parasite accumulation in placental tissue. Therefore, an ex vivo placental perfusion model was implemented to study parasite adhesion in intact placental tissue.

Infected erythrocytes expressing VAR2CSA, of both the FCR3 and NF54 genotype, accumulated in the perfused tissue, whereas an EPCR binding (FCR3-IT4VAR19) and a transgenic parasite where PfEMP1 expression had been silenced (DC-J) did not accumulate in the placenta. There was a small loss of It4var19 and DC-J infected erythrocytes, which may be explained by late stage erythrocyte rupture during the experiment.

These results show that the ex vivo perfusion model is suitable to measure placental accumulation of parasites and confirm that infected erythrocyte accumulation in placental tissue is VAR2CSA specific. Furthermore, as the EPCR-binding parasite did not accumulate in placental tissue, in spite of expression of EPCR in human placenta [56] and on the surface of the syncytiotrophoblast [57], it is likely that infected erythrocytes binding to receptors other that CSA will not accumulate in placental tissue. The exclusive interaction with CSA is possibly conferred by shielding of other receptors by the glycocalyx [58]. This suggest a minor role for binding to immunoglobulin and hyaluronic acid [19, 59, 60], but further studies elucidating the involvement of these receptors in the placental perfusion model could shed further light on these observations. Taken together with the finding that disruption of the var2csa gene impairs adherence to syncytiotrophoblast in vitro [11] the findings in this study imply that a vaccine against placental malaria can be limited to target only the CSA binding epitope of VAR2CSA.

An advantage of the perfusion model is the opportunity to study parasite adhesion to intact placental tissue under physiologically relevant flow conditions. The details of the flow dynamics of a placental cotyledon still remain to be elucidated, but different methods can be used to estimate the intervillous flow (see Additional file 3 for a discussion). An intervillous flow of 8–10 ml/min divided on three cannulas to optimize the oxygen delivery to the cotyledon is estimated to be within the physiological range. Methodological studies investigating the effect of increasing the number of maternal cannulas and the flow rate during ex vivo placental perfusion have been performed [61, 62]. Increased spiral artery velocity may impair maternal-fetal exchange and damage the villous tissue [63] and was associated with morphological damage of the syncytiotrophoblast in the perfusion model [62]. Although replicating the physiological flow in the perfusion model is a challenge, a flow is created where the maternal perfusate percolates the villous tree at a pressure gradient from the maternal cannulas to the outflow through the basal plate thus simulating the intervillous blood flow. This is an advantage compared to placental malaria models using a capillary or chamber flow system without the villous tree.

Another advantage of the perfusion model is that the intervillous space of the placenta is intact and unfixed so that adherence of infected erythrocytes, not only to the syncytiotrophoblast, but also in the intervillous space, can be studied. It has been suggested that infected erythrocytes bind to a loose matrix of chondroitin sulphate proteoglycan chains in the intervillous space [64]. Histological examination of the perfused tissue showed parasite accumulation both in the intervillous space and to a lesser degree on the syncytiotrophoblast, as seen in women with placental malaria [65]. Relatively few erythrocytes are seen in the fetal vessels and the intervillous space, in perfused tissue compared to non-perfused tissue, as fetal and maternal blood have been washed out [66] and due to the low haematocrit of 1 % in the perfusion experiments. Transmission electron microscopy showed electron dense filaments connecting knobs on the infected erythrocyte with syncytiotrophoblast microvilli as previously shown in tissue from women with placental malaria [67], not limited to denudated areas as recently suggested [68]. Moreover, scanning electron microscopy showed adherence of infected erythrocytes to intact syncytiotrophoblast.

This proof of concept study could form the basis of studying placental pathology in native tissue. However, close collaboration with a labor ward is necessary, as placentas must be accessed immediately after delivery and brought to the laboratory for processing within a short period of time. Moreover, it is imperative that stable and well-characterized PfEMP1 phenotypes are established prior to generating the amounts required for the perfusion experiments. Still, the perfusion group in Greifswald performed pilot experiments, suggesting that the model is indeed transferrable between different centres.

Adhesion of VAR2CSA expressing parasites was inhibited, but not blocked completely by soluble CSA. It has recently been shown that placental chondroitin sulfate shares a specific signature with chondroitin sulfate expressed on cancer cells, which is not found in other healthy human tissues [23]. The incomplete binding inhibition may be caused by differences in parasite tropism for bovine CSA and the human oncofetal CSA found in the placenta.

Antibodies against the full-length VAR2CSA protein inhibited parasite binding in placental tissue in a concentration dependent manner. Some variation was seen between experiments and the binding inhibition appeared greater when the experiments where performed with increasing concentrations of serum compared to the reversed order. One explanation could be that infected erythrocytes that had adhered in the placenta during previous phases were released when a higher concentration of serum was perfused through the tissue. This is a weakness of the model, which needs to be investigated further as it affects the ability to quantify the inhibition capacity if experiments are run in several phases.

This proof of concept study show that VAR2CSA expressing infected erythrocytes accumulate in the ex vivo placental perfusion model with a distribution similar to that seen in natural placental infection. Parasite accumulation is inhibited by soluble CSA and anti-VAR2CSA antibodies. In conclusion, the perfusion model bestows an opportunity to study parasite binding biology in intact human tissue and the functional properties of antibodies from both pre-clinical and clinical trials.