Of membranes and malaria: phospholipid asymmetry in Plasmodium falciparum-infected red blood cells

Abstract

Malaria is a vector-borne parasitic disease with a vast impact on human history, and according to the World Health Organisation, Plasmodium parasites still infect over 200 million people per year. Plasmodium falciparum, the deadliest parasite species, has a remarkable ability to undermine the host immune system and cause life-threatening disease during blood infection. The parasite’s host cells, red blood cells (RBCs), generally maintain an asymmetric distribution of phospholipids in the two leaflets of the plasma membrane bilayer. Alterations to this asymmetry, particularly the exposure of phosphatidylserine (PS) in the outer leaflet, can be recognised by phagocytes. Because of the importance of innate immune defence numerous studies have investigated PS exposure in RBCs infected with P. falciparum, but have reached different conclusions. Here we review recent advancements in our understanding of the molecular mechanisms which regulate asymmetry in RBCs, and whether infection with the P. falciparum parasite results in changes to PS exposure. On the balance of evidence, it is likely that membrane asymmetry is disrupted in parasitised RBCs, though some methodological issues need addressing. We discuss the potential causes and consequences of altered asymmetry in parasitised RBCs, particularly for in vivo interactions with the immune system, and the role of host-parasite co-evolution. We also examine the potential asymmetric state of parasite membranes and summarise current knowledge on the parasite proteins, which could regulate asymmetry in these membranes. Finally, we highlight unresolved questions at this time and the need for interdisciplinary approaches to uncover the machinery which enables P. falciparum parasites to hide in mature erythrocytes.

Supplementary information

The online version contains supplementary material available at 10.1007/s00018-021-03799-6.

Introduction

Cell membranes are the semi-permeable divide between the inside and out, allowing cells to maintain differences between intracellular conditions and the extracellular environment. From viruses to bacteria to parasites, many pathogens have evolved to breach this barrier and propagate inside other cells. Though not without its challenges, this lifestyle provides access to nutrients, allows them to hijack cellular machinery, and provides protection from the host immune system.

The malaria parasite, Plasmodium spp., develops and reproduces inside red blood cells (RBCs) during part of its lifecycle. Aside from being a particularly nutrient-rich environment, living in the blood also ensures that the parasites can be ingested when mosquitoes take a bloodmeal from the human host, hence allowing Plasmodium to progress to the next stage of its lifecycle inside this insect vector. The parasite has a sustained history of co-evolution with its human host, and the most common human mutations are found in RBCs, particularly in haemoglobin, and were most likely driven by the enormous selection pressure of malaria infections prior to reproductive age [1]. One particular advantage to colonising RBCs is that these terminally differentiated cells do not express Major Histocompatibility Complexes (MHC) on their cell surface, which would otherwise reveal the parasite’s presence to cytotoxic T cells. Furthermore, Plasmodium parasites, and particularly the most virulent species Plasmodium falciparum, express highly variable surface antigens, increasing the difficulty of generating a protective humoral immune response. While these mechanisms of immune evasion provide a degree of protection to the parasite, RBCs have other ways of signalling to the immune system that something is wrong.

This review will concentrate on the concept of membrane asymmetry in RBCs infected with the human malaria parasite, Plasmodium falciparum, and its interaction with the immune system during malaria infection.

The role of membrane asymmetry in eukaryotic cells

Healthy eukaryotic cells maintain a difference in the lipids present in the outside and in the inside layer of the plasma membrane lipid bilayer, termed membrane asymmetry [2, 3]. In particular, the lipids phosphatidylcholine (PC) and sphingomyelin (SM) are enriched in the outer layer, with the headgroups facing the extracellular environment. The aminophospholipids phosphatidylserine (PS) and phosphatidylethanolamine (PE) are sequestered in the inner layer, facing the cytoplasm, along with phosphatidylinositol (PI) and its phosphorylated forms.

The imbalance between the two membrane leaflets is important for a number of cellular functions. Asymmetry gives each of the leaflets different biophysical and biochemical properties. PS and PI have an overall negative charge (anionic), whereas PE, PC, and SM are zwitterionic at physiological pH. The functional roles of PS and PI in the inner layer include interacting with proteins in the membrane and cytoplasm, such as protein kinase C or cytoskeleton components; these interactions can affect enzyme activity, cell signalling, and provide stability to the membrane [4–10].

In marked contrast, the presence of PS in the outer membrane layer of nucleated eukaryotic cells is a hallmark of apoptosis [11], and acts as an ‘eat me’ signal to the immune system that the cell is unhealthy. Macrophages and monocytes then bind to exposed PS and engulf the cell [12–16]. Importantly, cell death with PS exposure generally results in a non-inflammatory immune response, and is thus considered a ‘gentle’ form of cell removal; this differs from necrosis, which leads to a local inflammatory response [17, 18]. PS exposure also occurs in erythrocytes/red blood cells (RBCs), where the process is referred to as eryptosis [19, 20].

Regulation of asymmetry

Membrane asymmetry is regulated by three main classes of integral membrane proteins—flippases, floppases, and scramblases [21–23]. Asymmetry is established and maintained by flippases and floppases (Fig. 1). Flippases are responsible for translocating lipids from the outer layer to the inner layer, while floppases function in the opposite direction. Most flippases are highly specific for the aminophospholipids (such as PS and PE), and exhibit an approximately tenfold higher affinity for PS than for PE [24–27]. This selectivity ensures that these lipids, and particularly PS, are sequestered in the inner layer (Fig. 1). Floppases are specific for PC and SM, though may also externalise PS at a much slower rate. Both flippase and floppase rely on the hydrolysis of adenosine triphosphate (ATP) as an energy source to move lipids between the two leaflets [23] (Fig. 1).

Fig. 1.

Membrane asymmetry in RBCs is regulated by flippase, floppase, and scramblase enzymes. Flippase and floppase hydrolyse ATP to move PS and PE to the inner layer (flippase) and PC to the outer layer (floppase). Scramblase moves lipids indiscriminately when activated by cytoplasmic Ca²⁺. Text indicates the molecular identity of the proteins in RBCs, with number of each protein per cell based on proteomic data [45]. Blue question mark uncertain functional assignment; purple question mark uncertain presence in RBCs. PS phosphatidylserine, PE phosphatidylethanolamine, PC phosphatidylcholine, ATP adenosine triphosphate, ADP adenosine diphosphate, ABC ATP binding cassette. UniProt: ATP11C (Q8NB49); ATP11A (P98196), ATP11B (Q9Y2G3); ABCC1 (P33527); ABCA7 (Q8IZY2); ABCG2 (Q9UNQ0); ABCC5 (O15440); ABCB4 (P21439); PLSCR1 (O15162); PLSCR4 (Q9NRQ2); TMEM16F (Q4KMQ2)

To counteract the functions of flippase and floppase, scramblases are responsible for the disruption of asymmetry [23] (Fig. 1). They are not specific for particular lipids and can translocate PC, PE, and PS in either direction. Unlike flippases and floppases, the activity of plasma membrane scramblases is repressed in healthy cells [23]. Accordingly, scramblases only become activated in old or damaged cells, when the intracellular Ca²⁺ concentration rises. Scramblase can also be activated by platelet activating factors, or by caspase-mediated cleavage during apoptosis [28–32]. Scramblases function is independent of ATP hydrolysis and quickly results in a breakdown of asymmetry, leading to PS exposure on the cell surface. The cell can then be recognised by the immune system and removed by phagocytes. Some studies have indicated that scramblase activation alone is not sufficient to fully collapse membrane asymmetry and that concurrent repression of flippase activity is required as well [23, 33]. Several studies have also suggested that membrane cholesterol acts as a scramblase inhibitor in healthy cells; the addition of cholesterol can repress scramblase activity, while cholesterol depletion results in scramblase activation even at low Ca²⁺ concentrations [34–36].

Together, the action of flippases and floppases, and the repression of scramblases, results in steady-state maintenance of asymmetry. The activation of scramblases, and potentially the inactivation of flippases and floppases, results in the breakdown of asymmetry and exposure of PS on the cell surface.

Membrane asymmetry in red blood cells (RBCs): molecular identity of the regulators

Many researchers have used RBCs as a model system for studying membrane asymmetry, because they are (i) relatively simple cells with no internal membranes, (ii) are readily available, and (iii) easy to separate from other blood components (e.g. [3, 9, 37–40]). Since RBCs cannot undergo the same apoptotic mechanisms as nucleated cells, such as DNA fragmentation and mitochondrial depolarisation, PS exposure as a mechanism of cell clearance may be particularly important for erythrocytes, along with other markers of senescence such as complement deposition [41].

While it is possible to study the general characteristics of flippase, floppase, and scramblase activity without knowing the specific proteins involved, identifying the proteins responsible for lipid translocation provides mechanistic insights into the molecular processes which regulate and disrupt membrane asymmetry. Fourteen functionally characterised or potential flippase proteins have been identified in the human genome [42, 43], and proteomic data suggest that three—ATP11A, ATP11B, and ATP11C—are present in human RBCs [44, 45] (Fig. 1). Of these, ATP11C is the most abundant, accounting for around 80–90% of the flippase molecules which are present [44, 45]. Mutations in ATP11C result in a lowered ability of RBCs to translocate exogenously-introduced PS to the inner leaflet of the membrane [46, 47]. A human patient with this mutation exhibited congenital haemolytic anaemia, likely because the ability to retain PS in the inner layer was compromised, leading to premature senescence and accelerated RBC clearing [46]. ATP11C mutations also resulted in accelerated senescence and clearance of RBCs in mice, and altered development of other blood cells such as B cells [47–49].

Floppase activity has been described for several ABC transporter proteins, including some members of the ABCB multidrug resistance (MDR) and ABCC multidrug resistance-associated protein (MRP) families and related proteins [50–58]. Of the family members that may be implicated in lipid floppase activity, ABCC1, ABCA7, ABCG2, ABCC5, and ABCB4 (the latter only in trace amounts) have been identified in mature human RBCs [45, 59] (Fig. 1).

There is still considerable debate over the molecular identity of proteins responsible for scramblase activity. A putative scramblase protein, phospholipid scramblase protein one (PLSCR1), was first identified in RBCs [60, 61] (Fig. 1). Several other proteins, including transmembrane protein 16F (TMEM16F, also called anoctamin-6) and the XK-related protein 8 (Xkr8), have since been identified and categorised as candidate scramblases, along with other family members of these proteins [28, 62–65]. Scramblase activity could come from any or all of these proteins, or possibly hitherto unrecognised proteins, depending on cell type and function.

The classification of PLSCRs as scramblases later came under scrutiny, because a murine knockout of the PLSCR1 homologue had no observable defect in PS scrambling in RBCs [66], leading some to conclude that PLSCR1 had been misidentified [67]. However, this confusion may be an error of conflating human and murine RBCs, since PLSCR1 is reportedly mostly absent from mouse RBCs, though its homologue PLSCR3 is highly abundant, whereas the opposite was found in human RBCs [35]. A PLSCR1 knockout is therefore unlikely to affect the scramblase activity of murine erythrocytes. Classification of PLSCR1 as a bona fide scramblase was further questioned because no mutations were present in the PLSCR1 gene of patients with Scott’s Syndrome, a bleeding disorder linked to impaired scramblase activity where platelets can no longer activate PS-dependent coagulation cascades [66, 67]. In good agreement with this finding, no defects in scrambling were detected when PLSCR1 from Scott’s Syndrome patients were reconstituted into proteoliposomes [68]. Scott’s Syndrome has since been linked to mutations in TMEM16F [63], which is highly expressed in platelets but lowly abundant in RBCs [45, 69], and only expressed in trace amounts in erythroblasts [35]. Xkr8 has been linked to apoptotic PS exposure in nucleated cells, as it is activated by caspase-mediated cleavage [28]. Xkr8 was not found in human RBCs proteomes [45]. Therefore, it remains to be determined if scramblase activity in human RBCs can be assigned to PLSCR1, its less abundant homolog PLSCR4, the low amounts of TMEM16F, or another unrecognised protein [35, 45] (Fig. 1).

Asymmetry in disease

Many links between diseases and disruptions to membrane asymmetry regulation in a variety of cell types have been described [70]. Some diseases result from mutations to the regulatory enzymes, including the putative scramblase TMEM16F and flippase ATP11C. Of note, RBCs from sickle cell patients have higher PS exposure, which may be responsible for some symptoms of sickle cell disease, such as accelerated senescence and clearance of RBCs [15, 71–74]. This observation indicates that PS exposure in the outer leaflet can be a consequence of altered erythrocyte physiology rather than direct defects in the lipid transport proteins. A better understanding of the underlying molecular mechanisms in inherited disorders of membrane asymmetry might ultimately lead to tailored interventions.

The key role of phospholipid asymmetry in disease progression is also increasingly recognized in infectious diseases. Some intracellular pathogens appear to deliberately expose PS to facilitate invasion into host cells. This strategy is termed ‘apoptotic mimicry,’ and is employed by viruses and parasites [75–78]. A prime example is the intracellular parasite Leishmania, the causative agent of leishmaniasis. Leishmania parasites expose PS to exploit the ability of their host cells, macrophages, to recognise the ‘eat me’ signal. Once inside, the parasite seizes control of the cell. Of particular interest is the investigation of membrane asymmetry in malaria, caused by Plasmodium parasites, due to its high global mortality and morbidity.

The case of Plasmodium falciparum-infected RBCs (iRBCs)

The malaria parasite, Plasmodium, lives inside RBCs for part of its lifecycle. As the parasite develops, it induces significant alterations to the host cell, including changes to membrane composition. Many studies have attempted to explore whether these modifications include PS exposure, but have reached different conclusions. Much of the debate in the literature can be attributed to differences in methodology, and consensus on experimental protocols is warranted to integrate results into a consistent view of the roles of phospholipid asymmetry in host-parasite interactions. A fascinating aspect of blood infection by human-infecting Plasmodium species is the particular preference for a subpopulation of RBCs; Plasmodium vivax and Plasmodium ovale invade immature RBCs, termed reticulocytes; P. falciparum and Plasmodium knowlesi can employ mature RBCS as host cells, and Plasmodium malariae has a preference for older erythrocytes [79–81]. Accordingly, PS exposure could vary considerably between infections from different Plasmodium species.

Plasmodium falciparum has received the most attention in the literature, both because of its high global mortality since P. falciparum causes the vast majority of malaria-related deaths, as well as the ease of culturing P. falciparum parasites in vitro. The following section analyses studies (listed in Supplementary Table 1) which investigated membrane asymmetry, and particularly PS exposure, in RBCs infected with P. falciparum (iRBCs).

Early methods for investigating asymmetry

Membrane asymmetry in iRBCs was first investigated by modifying the lipids in the outer leaflet, either chemically or enzymatically, to determine the ratio of lipids in the outer leaflet (modified) to lipids in the inner leaflet (unmodified). Studies using this assay came to different conclusions about whether or not asymmetry was disrupted in iRBCs [82–85]. The major limitation of these assays is the difficulty in accounting for parasite membranes inside the RBC, which add to the ‘unmodified’ lipid pool. It was also necessary to ensure that the enzyme or chemical agent did not gain access to the internal lipids, even though modifying external lipids can compromise membrane integrity. A procoagulation assay, based on the PS-dependent conversion of prothrombin to thrombin, also resulted in opposing conclusions [83, 85]. Another early method was to stain RBCS with Merocyanide 540, a dye which preferentially fluoresces when bound to membranes with high fluidity, which is a feature of symmetrical membranes [84]. However, membrane fluidity is also impacted by other factors, such as cholesterol concentration [86, 87], and thus this method did not exclusively indicate altered asymmetry.

One major difference between these studies was whether they looked at samples containing a high number of uRBCs, which would dilute differences measured between uRBCs and iRBCs, or samples where iRBCs have been enriched by removing uRBCs, termed high parasitaemia (expressed as % iRBCs out of total RBCs). The results of some studies may also be affected by long incubations in glucose-free buffer, which could induce artefactual PS exposure due to starvation. Significantly higher energy demand of iRBCs compared to uRBCs results in swift depletion of ATP in glucose-free media, which also causes Ca²⁺ influx through an otherwise inactive channel [88–93]. Without ATP, RBCs are unable to actively regulate membrane asymmetry, since flippase enzymes cannot internalise PS (Fig. 1). Moreover, in the absence of ATP, RBCs cannot extrude Ca²⁺ through the plasma membrane Ca²⁺ ATPase (PMCA), further increasing PS exposure through scramblase activation [94, 95].

In addition to their procoagulation and lipid modification data, Maguire et al. [83] also found that anti-PS antibodies bound only to iRBCs and not uRBCs. This was perhaps the best representation of the level of PS exposure in physiological culture conditions at this point, but specificity of the antibody was not demonstrated [96, 97]. Due to these methodological shortcomings, these early studies have not provided the convincing and consistent answer that the malaria research field was seeking.

Annexin V staining

The development of Annexin V, a protein which binds PS in a calcium-dependent manner, has provided a more reliable tool to measure PS exposure [38, 98–100]. Annexin V binding assays have become the preferred method for investigating PS exposure in a variety of cell types, due to its highly specific binding to PS and the commercial availability of fluorophore-conjugated recombinant protein. Early uses of this method in iRBCs also lead to opposing conclusions about whether or not PS was exposed in these cells, possibly because studies looked at samples with different parasitaemias [97, 101–104].

The Annexin V staining method was subsequently improved by the introduction of two-colour staining to differentiate uRBCs and iRBCs within a single culture sample [105]. Fluorescent DNA stains bind to parasite nucleic acid in iRBCs, leaving the enucleated uRBCs uncoloured. This effectively allows complete assessment of all iRBCs and uRBCs without separation steps or maintaining cultures at unphysiologically high parasitemia, which may add stress to the iRBCs and artificially increase PS exposure. This method has become the most popular way of assessing membrane asymmetry in iRBCs (Supplementary Table 1). Of 24 studies that compared uRBCs and iRBCs with this method, at least 22 found a significant difference between uRBCs and iRBCs. However, the extent differed vastly between studies, ranging from only 1.5% up to 71% of iRBCs, and 0.2 to 20% for uRBCs. Figure 2 summarises studies which included data on the percentage of iRBCs and uRBCs using Annexin V staining with either DNA stain differentiation or other enrichment methods. This variation was not simply a result of different cut-off thresholds for determining PS exposure, as there was also variability in the ratio of PS exposing iRBCs to uRBCs in each study, ranging from around 2:1 to more than 20:1. Many of these studies also evaluated the effects of heat stress, cytokines, drugs, iRBC population density, Natural Killer cells, or thalassemia on iRBC membrane asymmetry, and/or its effects on cytoadherence (Supplementary Table 1).

Fig. 2.

Percentages of uninfected and infected RBCs with PS exposure, measured by Annexin V binding. Percentages were calculated from tables or graphs using ImageJ when not specified in the text. Only results from RBCs not treated with a modifying agent (drug, peptide, heat stress etc.) are shown. Where multiple percentages (e.g. multiple stages) were available, the highest and lowest have been included. Studies are grouped by major methodological considerations. Error bars are shown where available in the original publications. * = used iRBCs enriched by density gradient centrifugation rather than fluorescent DNA stain

Since many of these were not solely attempting to examine the difference between uRBCs and iRBCs under standard culturing conditions, most either included 20–50 min incubations in glucose-free buffers, or 24–48 h in Ringer solution (which contains glucose, but no serum or bovine serum albumin substitute) prior to sample analysis. These conditions may artefactually increase PS exposure [96, 106]. Strikingly, the studies which appear to take samples directly from culture conditions and keep them in glucose-containing media throughout preparation and analysis report among the lowest percentage of iRBCs exposing PS (Fig. 2). Thus, the wide variety between the studies might be attributed to different levels of ATP depletion, along with parasite development stage, parasitaemia in the culture, and incubation time and conditions. PS exposure can also be dependent on RBC age since older RBCs, particularly those that have been stored ex vivo, have lower flippase activity, lower capacity to extrude calcium, and higher scramblase activity [95, 107–113]. We propose that future studies on PS exposure in iRBCs should either ensure a constant supply of glucose or empirically demonstrate that exclusion of glucose has no effect on the extent of PS exposure.

On the balance of evidence, it is very likely that PS is more exposed in iRBCs than uRBCs in vitro, but caution should be taken when drawing comparisons to in vivo iRBCs, since differences in the environment could lead to different results. PS exposure further increases when iRBCs are incubated at febrile temperatures in vitro [114–116]. Accordingly, it is possible that the absence of febrile temperature intervals in standard in vitro cultures may actually lead to an underestimation of the PS exposure level which would occur in vivo, rather than generating artificially high levels of PS exposure. It is also possible that the varied conditions encountered by the parasite within the human body (e.g. location of the parasite, host genetics, nutrition, and immune status) may influence the degree of PS exposure in vivo. In addition, culturing parasites to high parasitaemia increases the level of PS exposure [117–119]. Given all these potential variables, it will be important to systematically investigate whether PS exposure differs between malaria patients with different parasitaemia and disease severity.

Scramblase activation: the trigger of altered asymmetry in iRBCs?

Studying PS exposure in iRBCs would not be complete without exploring the underlying molecular mechanisms. The obvious primary candidate for altered membrane asymmetry is activation of scramblase in iRBCs, which has recently been experimentally demonstrated [120]. Since scramblase is activated by cytoplasmic Ca²⁺, previous reports have hypothesised that Ca²⁺ influx into iRBCs could be responsible [70, 83, 115, 120, 121]. However, Ca²⁺ regulation in iRBCs is in itself a contentious field with conflicting studies; these discrepancies may once again result from differences in the methodology and experimental conditions, such as glucose availability. While it is firmly established that iRBCs have a higher total concentration of Ca²⁺ than uRBCs, the majority of this Ca²⁺ is sequestered to the intracellular parasite and its parasitophorous vacuole, where it is unable to activate RBC membrane scramblases [120–124]. Reports on the Ca²⁺ concentration in the RBC cytoplasm have reached different conclusions, with some finding it can be elevated compared to uRBCs [120, 125], and others finding that the Ca²⁺ concentration was unchanged or even slightly lower than uRBCs [124, 126].

Oxidative stress which is produced by metabolic by-products (e.g. as a result of haemoglobin digestion by the parasite) can activate scramblase activity, by increasing Ca²⁺ influx to RBCs [70, 83, 105, 115, 127]. Within an in vitro culture, both iRBCs and their neighbouring uRBCs are subject to oxidative damage, and several studies have demonstrated that PS exposure in both cell types increases with the parasitaemia of the culture [117–119]. This observation, therefore, supports the hypothesis that parasite-induced oxidative stress contributes to PS exposure in these cells. This is corroborated by clinical studies observing that oxidative damage to RBCs can also occur during in vivo malaria infection [128–131]. However, the clinical consequences of in vivo PS exposure have not yet been systematically investigated. The influence of parasite-induced oxidative stress beyond the iRBCs is also supported by the in vivo observation that plasma from patients with P. falciparum infection can also induce PS exposure in healthy RBCs in vitro [132]. This could be related to oxidising compounds in the serum. Interestingly, plasma from P. vivax patients had no effect on PS exposure despite similar parasitaemia in these patients, highlighting the need for further clinical studies.

Furthermore, scramblase can also be activated by bioactive lipids: ceramide for example, probably sensitises scramblase to the effects of Ca²⁺, rather than directly affecting Ca²⁺ influx [133–135].

Parasite invasion is accompanied by a temporary spike in intracellular Ca²⁺ [136]. Therefore, it may be possible that the parasite invasion process causes scramblase activation and PS exposure. Furthermore, P. falciparum can utilise an invasion pathway involving the ligation of the glycoprotein Glycophorin C on the RBC surface [137, 138]. It has also been demonstrated that PS exposure is induced after antibody binding to Glycophorin C on the RBCs surface, a process which could also contribute to PS exposure during parasite invasion [139]. However, it is not yet clear if this PS exposure would persist, therefore contributing to PS exposure in later stage iRBCs, or if the exposed PS would be reinternalized by the action of flippase proteins, and therefore independent from the PS exposure observed later in parasite development. In the case of the former, exposed PS would leave ring-stage iRBCs vulnerable to phagocytosis while still in circulation through the host bloodstream. Furthermore, most studies demonstrate lower PS exposure in early-stage parasites than late-stage parasites, and it therefore seems very unlikely that parasite invasion is solely responsible for the PS exposure in iRBCs, although it may be the beginning of the process.

Together, multiple mechanisms could contribute to the dynamics of PS exposure in iRBCs, although more systematic experimental evidence is still required. Changes in membrane asymmetry can serve as critical signatures of an intracellular infection of an otherwise immunologically silent, terminally differentiated host cell.

In vivo evidence and effects of PS exposure in malaria infection

Underlying the debate over whether or not PS is exposed in iRBCs, is the question of the biological relevance of any observed changes on malaria infection. Clearly, caution should be taken in interpreting the observed extent of PS exposure in iRBC in vitro, which could be caused or exacerbated by culturing conditions that differ fundamentally from a Plasmodium infection in vivo. Some have argued that low levels of PS exposure detected in vitro would unlikely be enough to have an effect in vivo [70, 96, 97]. While these concerns deserve careful consideration, there are several lines of evidence and perceivable ways in which PS exposure, even in small amounts, may be clinically relevant for disease outcomes in vivo, including (1) exposed PS leading to non-opsonic phagocytosis of iRBCs by macrophages and monocytes, (2) the effect of anti-phosphatidylserine antibodies, (3) exposed PS as a mediator of iRBC cytoadherence, and (4) PS exposure as a way of controlling parasite burden in the host. These complementary mechanisms are analysed and interpreted below. Since the potential for increasing PS exposure and other eryptotic mechanisms for host-directed malaria therapy was the subject of an excellent recent review [140], we will focus here on PS exposure which occurs without targeted drug intervention.

Non-opsonic phagocytosis by macrophages and monocytes

Non-opsonic phagocytosis may be particularly important in the control of first-time malaria infections. Histological data indicate that macrophages and monocytes phagocytose iRBCs in vivo [141–144], although from ex vivo samples, the different mechanisms which may be responsible cannot be distinguished e.g. direct PS recognition, other ligand recognition, complement/antibody-mediated opsonisation, or filtration in the spleen. In vitro studies showed that macrophages and monocytes phagocytose iRBCs more than uRBCs, independent of any opsonising factors (i.e. antibody or complement), which suggests the involvement of non-opsonic recognition signals [120, 145–147]; these signals potentially include exposed PS. Early on, Turrini et al. [148] described that monocyte phagocytosis of iRBCs was reduced when PS recognition sites were blocked with PS liposomes, but not PC liposomes, indicating that at least a portion of monocyte recognition of iRBCs is directly influenced by PS. More recently, treatment of iRBCs with Annexin V has been demonstrated to reduce monocyte phagocytosis [120]. Several candidate receptors for the recognition of exposed PS have been recently reviewed [16]. Figure 3 illustrates potential candidates for ligands and recognition signals involved in non-opsonic phagocytosis of iRBCs.

Fig. 3.

Potential candidates for non-opsonic recognition signals between iRBCs and macrophages/monocytes. The scavenger receptor CD36 recognises PfEMP1, exposed PS, and potentially other ligands. Treatment with antibodies against CD36 decreases phagocytosis by 50–60%, while removal of iRBC surface proteins (including PfEMP1) with trypsin decreases phagocytosis by ~ 75–85% [145–147]. Exposed PS can also be recognised by other receptors including BAI-1, STAB-2, TIM-1, and TIM-4, some of which may be present on monocytes and macrophages [16]. List is not exhaustive. PS phosphatidylserine, iRBC infected red blood cell, PV parasitophorous vacuole

An emerging scenario is that iRBCs which are sequestered from circulation, and thus protected from splenic macrophages, may be vulnerable to phagocytosis by circulating monocytes. Candidate cells are non-classical (CD14^dimCD16⁺) and intermediate monocytes (CD14⁺CD16⁺), which can ‘patrol’ along the vascular endothelium and phagocytose apoptotic cells, possibly through PS recognition [149]. Thus, these monocytes may also be able to phagocytose sequestered iRBCs. High portions of non-classical monocytes are associated with better disease outcomes in malaria patients, though this could also be due to their anti-inflammatory effects [150].

Effect of anti-phospholipid antibodies

High levels of anti-phospholipid antibodies, and particularly antibodies against PS, have been found in patients with P. falciparum or P. vivax malaria infections [151–155]. In addition, PS exposure on RBCs correlated with malaria severity in patients [144]. This suggests that there could be a higher quantity of exposed PS in patient blood. Some of it may be from PS-exposing iRBCs [105], but a likely source of exposed PS is also from cell lysis, which occurs after parasite egress and rupture of infected erythrocytes.

Regardless of their causes, reports have demonstrated that these anti-PS antibodies bind to iRBCs and uRBCs exposing PS, and activate complement-mediated lysis of uRBCs artificially induced to expose PS [153, 154]. This mechanism may be especially important for the destruction of sequestered iRBCs, which do not encounter splenic macrophages. Accordingly, the presence of these antibodies has been associated with protective responses and better disease outcomes in placental malaria, where iRBCs sequester by binding to syncytiotrophoblasts [152]. Two other studies in human patients found no association between anti-PS antibody titre and peripheral blood parasitaemia (i.e. the number of parasites circulating in the bloodstream), though these measurements do not account for sequestered parasites [153, 155]. Using the murine model parasite P. yoelii, injection of anti-PS antibodies or Annexin V did not significantly alter the speed of parasite clearance from blood circulation, perhaps indicating that the role of anti-PS antibodies on circulating iRBCs is minor in this system [153]. Importantly, anti-PS antibodies may also exacerbate anaemia by collateral destruction of uRBCs. A study of both human malaria patients and murine models of malaria infection reported that the presence of anti-PS antibodies correlated with the severity of anaemia which continued to persist after parasite clearance, due to a destructive effect on uRBCs [153]. When PS exposure was blocked with Annexin V, anaemia was reduced in mice [153].

More studies are required to determine whether the potential beneficial effects of anti-PS antibodies against iRBCs would outweigh the negative effects of exacerbating anaemia.

Cytoadherence

A unique feature of P. falciparum-infected RBCs is cytoadherence: the propensity to bind endothelial cells, thereby enabling the parasite to sequester out of the circulating blood into the vascular periphery and capillaries and avoiding splenic clearance. Exposed PS may contribute to iRBC cytoadherence to the vascular endothelium, which is mediated at least in part by the membrane glycoprotein CD36. This was demonstrated by Eda and Sherman [105] when cytoadherence was diminished by blocking exposed PS on iRBCs in a concentration-dependent manner not only with Annexin V, but also with the PS interacting ligands CD36 or thrombospondin. Exposing CD36 to PS liposomes or soluble PS prior to adding iRBCs also decreased cytoadherence, further demonstrating a role of PS in cytoadherence [105]. Zhang et al. [114] also used blocking of PS with Annexin V to demonstrate an increase in PS-dependent cytoadhesion upon exposure of iRBCs to febrile temperatures. Together, these data provide a convincing basis for a direct link between PS exposure and cytoadherence of iRBCs.

P. falciparum is the only human malaria species that uses PfEMP1-mediated cytoadherence, and hence PS-mediated cytoadhesion might be a more universal mechanism and particularly important in the sequestration of the Plasmodium species which do not contain orthologs of PfEMP1 [156, 157]. However, experimental data is so far incomplete. Anti-PS antibodies were also detected at high levels in P. vivax patients [151]; additional studies are warranted to investigate whether there is a role for PS exposure in P. vivax sequestration [158, 159]. Studies using non-human primates and murine malaria models have also indicated a potential role for PS exposure in cytoadherence, primarily through the PS-binding ligand CD36, although further mechanistic studies are needed to establish a causal link [160–163].

Controlling parasite burden

Although it is tempting to envisage the malaria parasite as a malevolent villain, it is important to remember that swift killing of a host is actually a disadvantage to the parasite, which needs time to propagate and transform to the sexual stages that can escape to the mosquito vector. This is particularly important for P. falciparum, which requires up to 14 days to develop the transmissible sexual stages, termed gametocytes. P. falciparum is the only human malaria parasite which requires this extended maturation period [164]. Therefore, it can be argued that it is beneficial to both host and parasite to limit parasite burden. Although the concept of apoptosis or regulated cell death in a single-celled organism may sound counterintuitive, malaria parasites in the bloodstream constitute a population which collectively aims for advancement to the Anopheles vector for onward transmission. Engelbrecht and Coetzer [119] reported that growing cultures to high parasitaemias induces markers of regulated cell death including DNA fragmentation, loss of mitochondrial polarisation, morphological changes, and reduced growth, in addition to PS exposure. Parasite death may play an important role in density regulation and allow an expanding parasite population to regulate its burden on the host [165]. In such a scenario, PS exposure in a subpopulation of iRBCs may be a strategy to support host survival, and ultimately parasite transmission. However, more direct evidence is required to support this hypothesis.

Membrane asymmetry in other parasite life cycle stages

The asexual blood stage of Plasmodium represents only one phase of a complicated lifecycle, involving multiple hosts and regions of the body (Fig. 4). It is difficult to collect data on membrane asymmetry in P. falciparum during other stages of the lifecycle, such as hepatic infection in human patients. However, studies in the murine malaria model have suggested that P. berghei parasites can manipulate apoptosis in dying host hepatocytes, which includes reducing PS exposure, prior to blood infection [166, 167]. This would benefit the parasites by allowing them to progress to the bloodstream for the next life cycle stage while limiting immune detection. This process possibly involves parasite-mediated sequestration of Ca²⁺ to limit scramblase activation [166]. So far, it is unknown if this process extends to human hepatic infection.

Fig. 4.

Simplified lifecycle of Plasmodium, showing stages where PS exposure has been investigated at the host or parasite membrane. Since not all life-cycle stages of the human malaria parasite P. falciparum are easily accessible for study, information on the murine malaria parasite P. berghei is also included, serving as an in vivo model for human malaria. PS phosphatidylserine, RBC red blood cell

PS exposure could also occur within the mosquito host, such as during the sporogony phase of the lifecycle in the midgut. During this stage, studies have demonstrated markers of caspase-dependent programmed cell death in P. falciparum and P. berghei, which in the latter case included PS exposure in the plasma membrane of ookinete-stage parasites [168–170]. The underlying cause for PS exposure has not been definitively demonstrated and could be involved in interactions with mosquito defence mechanisms.

Membrane asymmetry at the parasite plasma membrane

To our knowledge, no studies to date have successfully investigated lipid asymmetry at the parasite plasma membrane (PPM). Isolating parasites from the surrounding host membranes is notoriously difficult. Saponin is commonly used to lyse host membranes, but leaves host fragments attached to the parasite; Annexin V can then bind to both inner and outer layers, overshadowing any signal from the PPM and hindering appropriate data interpretation [171]. Instead, indication for the dynamic regulation of PPM lipid distribution comes from genomic data indicating the presence of parasite flippase, floppase and scramblase genes [172].

The P. falciparum genome encodes several putative flippases, though no studies to date have reported localisation or functional analyses of these proteins [173]. In the murine parasite P. berghei, two of these proteins, ATP2 (encoded by PBANKA_143480) and ATP8 (encoded by PBANKA_143830), localise at or near the PPM [174]. Targeted experimental genetics approaches were unable to remove these genes, supporting the notion of essential functions for blood infection [174]. Support for critical roles of flippase proteins in P. falciparum asexual blood-stage originates from an untargeted mutagenesis screen in P. falciparum, which indicated that the orthologues (PF3D7_1219600 and PF3D7_1223400, respectively) are likely to be essential, or that their inactivation causes severe growth defects [175]. The localisation of these regulatory candidate flippases in the PPM could indicate that the membrane is asymmetrically maintained, but more direct evidence is clearly required. The parasite genome also includes other putative flippase proteins, such as ATP7 (PF3D7_0319000, or PBANKA_080630 in P. berghei), although these appear to be dispensable, and localise to structures inside the murine parasite [174–176], suggesting they have other roles within the cell.

The parasite genomes also encode 11 ABC family members with transmembrane domains, some of which could function as floppases. Several have been localised to the PPM, including PfMRP1/ABCC1 (PF3D7_0112200), PfMRP2/ABCC2 (PF3D7_1229100), PfMDR2/ABCB2 (PF3D7_1447900), and PfMDR5/ABCB5 (PF3D7_1339900) [177–180]. To our knowledge, no published studies have investigated the potential floppase activity of any of the parasite ABC transporters, although their roles in drug resistance have been extensively examined, indicating that unequivocal assignment to floppase function might be challenging [181].

A candidate scramblase protein, now termed PfPLSCR (PF3D7_1022700), has recently been described and characterised [182]. When expressed in proteoliposomes, this protein scrambled lipids, particularly PE, in a Ca²⁺-dependent manner. The protein appears to localise within the parasite, rather than at the PPM, perhaps with the exception of merozoites and some gametocyte stages. No defects to asexual parasite growth or invasion were evident when the gene was removed. Further evidence is required to determine if this protein has any effect on asymmetry in the PPM or internal parasite membranes.

Conclusions

Phospholipid asymmetry in the RBC membrane is established, maintained, and eventually collapsed by the coordination of flippase, floppase, and scramblase enzymes. This intricate regulation protects the RBC during its lifespan, and eventually marks the cell for degradation by the appearance of PS in the outer membrane layer during RBC senescence. Infection with the malaria parasite P. falciparum imposes a huge obstruction and induces changes to membrane asymmetry in RBCs, which the intracellular parasite must limit to reach full maturity. Although there has been considerable debate, and a number of methodological issues need resolution, the broad consensus in the published literature is that the iRBC population has a higher degree of PS exposure than the uRBC population, indicating that membrane asymmetry is disrupted in parasitized RBCs. A better understanding of the underlying molecular mechanisms that trigger the regulation and collapse of asymmetry in these cells represents an exciting field for further study. Of particular interest is to understand the consequences of altered asymmetry in malaria patients, including interactions with the host immune system. This includes immune avoidance by cytoadherence, recognition of exposed PS by phagocytes, and interaction of antibody- and complement-mediated recognition. Further studies in this field could add to our understanding of host-parasite interactions, particularly the role of the innate immune system in malaria infection. The most pertinent questions for further research are highlighted in Box 1.

Box 1 Outstanding questions for the field.

How are flippase, floppases, and scramblase regulated in RBCs to ensure dynamic fine-tuning of membrane asymmetry in response to changing environments?

What is the impact of experimental artefacts on PS exposure in P. falciparum iRBCs?

What are the mechanisms underlying scramblase activation and disrupted membrane asymmetry in iRBCs? How do different mechanisms interact?

How much do human RBC flippase, floppase, and Ca²⁺ efflux pumps contribute to regulating membrane asymmetry of iRBCs? To what extent is the activity of these proteins affected by the parasite?

To what extent is PS exposed in iRBCs of human malaria patients?

Does PS increase or modulate cytoadherence of iRBCs in malaria patients?

Does PS exposure in iRBCs affect phagocyte recognition, antibody interaction, and complement-mediated lysis during malaria infection?

Could PS exposure play a beneficial role in limiting parasite burden to prolong host survival until transmission?

Will it be possible to develop adjunct therapies and interfere with the regulation of membrane asymmetry to improve patient outcomes?

Supplementary information

Below is the link to the electronic supplementary material.

Funding

Not applicable for this study.

Declarations

Conflict of interest

Not applicable for this study.

Ethics approval

Not applicable for this study.

Consent to participate

Not applicable for this study.

Consent for publication

Not applicable for this study.