INTRODUCTION: MALARIA DISEASE

Malaria is an infectious disease that causes enormous global medical and economic burdens. Of the four species of malaria parasites known to infect humans, the bulk of severe disease and complications is caused by the species Plasmodium falciparum. More than 80% of the approximately 2 million estimated cases of malaria-related mortality worldwide occur in Africa, mostly affecting infants and pre-school children aged 6 months to 5 years. Malarial fatalities are associated with a spectrum of discrete and overlapping disease syndromes of complex aetiologies. Humans affected and dying of malaria may variously suffer systemic, single- or multi-organ involvement, including acute respiratory distress, coagulopathy, shock, metabolic acidosis, hypoglycaemia, renal failure, pulmonary oedema and cerebral involvement including seizures and coma (White & Ho, 1992). Basic mechanisms controlling these processes are thought to be the site-specific localization of parasites (by cytoadherence to vascular endothelial markers such as the adhesin ICAM-1 (Berendt et al. 1989), and both local and systemic inflammatory responses arising from the action of cytokines produced predominantly by the innate immune system (Stevenson & Riley, 2004). The ability of the intraerythrocytic parasites to cytoadhere to endothelium and sequester in the microvasculature of vital organs (Miller et al. 2002) is thought to promote the accumulation of high local concentrations of bioactive parasite products (GPI-anchored antigens and glycolipids, metabolite toxins) that can trigger a cascade of pro-inflammatory immune responses in the brain, lungs, spleen, liver, kidney and placenta. This cytokine-dependent inflammatory cascade is believed to contribute to the spectrum of acute clinical symptoms that manifest during severe malaria bouts. In order to better understand the molecular mechanisms underlying severe malarial pathogenesis, our group has focused its attention on the interaction between P. falciparum and the immune system of the vertebrate host in the context of malaria infection.

MOLECULAR BASIS OF SEVERE MALARIA – CYTOKINE CONTRIBUTIONS

When examining the molecular basis of malarial pathogenesis it is clear that many factors from both the parasite and the host immune system contribute to the development of severe disease and there is an intricate relationship between these factors (reviewed by Chen, Schlichtherle & Wahlgren 2000; Stevenson & Riley, 2004). Both human and murine malaria infections elicit cells of the innate immune compartment to produce high levels of pro-inflammatory cytokines such as interferon-gamma (IFN-γ), interleukin-1 (IL-1), IL-6, IL-12 and tumor necrosis factor alpha (TNF-α) (Grau et al. 1987, 1989b; Lucas et al. 1997; Lyke et al. 2004). These pro-inflammatory cytokines appear to be essential for mediating acquired immunity to the invasive sporozoite during the early stages of infection (Schofield et al. 1987), but their role in mediating host resistance to blood stage infection is less clear.

The over-production of inflammatory cytokines can also be harmful to the host. Serum IFN-γ levels are elevated in patients with acute P. falciparum malaria (Ringwald et al. 1991), and death resulting from cerebral malaria (CM) was found to be associated with heterozygosity for an IFN-γ receptor-1 polymorphism in Gambian children (Koch et al. 2002). In the murine model both IFN-γ and IFN-γ receptor knock-out mice are resistant to Plasmodium berghei-mediated CM (Amani et al. 2000), and injection of neutralizing monoclonal antibody (mAb) against IFN-γ was shown to protect against CM in C57BL/6 susceptible mice (Grau et al. 1989a). These data suggest a role for IFN-γ as a key regulator of downstream immune system processes in the pathogenesis of severe malaria. IFN-γ has multiple activities in both the innate and acquired compartments of the immune system. Importantly, IFN-γ induces macrophages to become highly activated and, when exposed to live parasites, to generate high levels of free radical nitrogen intermediates (RNIs) via an inducible nitric oxide synthase (iNOS). RNIs can cause irreversible damage to neuronal tissues and have been implicated in the pathogenesis of CM (Al Yaman et al. 1996). In addition, when stimulated with parasite products in vitro, IFN-γ-primed macrophages can produce high levels of other pro-inflammatory cytokines associated with severe malaria, including IL-1, IL-6, IL-10, IL-12 and TNF-α (Grau et al. 1987, 1989b; Lucas et al. 1997; Lyke et al. 2004). Severe malaria is also associated with polymorphisms in the TNF-α promoter region, which may be linked to increased TNF-α expression (Mcguire et al. 1994), and high circulating levels of TNF-α are a prognostic indicator of fatality (Grau et al. 1989b). In concert, IFN-γ and TNF-α cytokines are capable of synergistically up-regulating intracellular adhesion molecules such as ICAM-1, VCAM-1, CD36 and thrombospondin on various cell types, including vascular endothelium and leukocytes. This augments sequestration of parasitized erythrocytes to endothelium and intensifies the deposition of malaria toxins in the deep vasculature of organs such as the brain. Malaria fatalities are thus strongly associated with an exacerbated systemic or organ-specific inflammatory cascade, but the molecular nature of the parasite toxin(s) that trigger the overproduction of pro-inflammatory cytokines has for a long time remained unclear.

MALARIAL GPIS AS CANDIDATE TOXINS AND PATHOGENICITY FACTORS

Several years ago, in vitro and in vivo studies suggested that glycosylphosphatidylinositol (GPI)-anchored molecules of P. falciparum function as the dominant malarial toxin in the context of infection. This hypothesis was based on the finding that purified GPI-anchored glycoproteins and glycolipids of P. falciparum and P. berghei are very potent activators of innate immune cells, capable of inducing pro-inflammatory cytokines TNF-α and IL-1 from macrophages (Schofield & Hackett, 1993). This activity has subsequently been confirmed by others (Naik et al. 2000; Vijaykumar, Naik & Gowda, 2001) (Table 1). The original study also showed that when administered to mice, GPI alone is sufficient to cause symptoms similar to acute malaria infection such as transient fever and hypoglycemia and death of recipients due to TNF-α-mediated sepsis (Schofield et al. 1993). Initial data showed that the endotoxin activity of major P. falciparum surface antigens released during schizont rupture (i.e. MSP-1, MSP-2) is restricted to the GPI moiety, since protein denaturation and exhaustive digestion with pronase does not abolish or reduce the potency of these molecules for the induction of cytokines in vitro (Schofield & Hackett, 1993). In contrast, enzymatic cleavage with phospholipase A₂ or chemical treatment of the GPI moiety with mild alkali (for selective removal of fatty acid esterified at the glycerol sn-1 and sn-2 postitions) or nitrous acid (for selective breakage of the glucosamine-myo-inositol bond) reduces most of the pro-inflammatory activity (Schofield & Hackett, 1993) – indicating a requirement for intact GPI-associated glycolipid structure. Additional studies found that purified P. falciparum GPI glycolipids were also able to induce, either directly or in synergy with IFN-γ, a range of other activities in various host tissues and cell types, including insulin-mimetic signalling in adipocytes (Schofield et al. 1994, 1996), the induction of iNOS in macrophages and vascular endothelium (Tachado et al. 1996), and the up-regulation of cell adhesion molecules on the surface of host leukocytes and endothelial cells (Schofield et al. 1996). The bioactivity of purified GPI glycolipid was comparable to that obtained with live parasites in vitro, in the range of 0·1–10 parasite equivalents per macrophage (Schofield & Hackett, 1993; Schofield et al. 1996). Importantly, monoclonal or polyclonal antibodies that recognize malarial GPI completely block the induction of TNF-α or iNOS output from macrophages and ICAM-1 expression on endothelial cells by crude extracts of P. falciparum (Schofield et al. 1994, 1996, 2002). Thus, Plasmodium GPIs appear both necessary and sufficient for the induction of pro-inflammatory host responses by malarial parasites in vitro.

These findings were then extended to GPIs of other parasitic protozoa (Table 1). For example, the GPI anchor of Trypanosoma brucei VSG and highly purified GPIs of Trypanosoma cruzi trypomastigote origin are potent macrophage activators ( Tachado & Schofield, 1994; Magez et al. 1998; Almeida & Gazzinelli, 2001), and this may account for many features of pathogenesis of trypanosomiasis and Chagas disease. More recently, it was shown that highly purified GPIs of Toxoplasma gondii tachyzoites, the infective form of the parasite which causes severe encephalitis, are bioactive factors that participate in the production of inflammatory cytokines from macrophages during toxoplasmal pathogenesis (Debierre-Grockiego et al. 2003). In contrast, GIPLs and lipophosphoglycans derived from infective promastigote forms of Leishmania spp. suppress several functions of the host immune system – thereby allowing this parasite to evade the innate immunological control of infection (Camargo et al. 1997b; Tachado et al. 1997). Taken together, these data support the view that GPIs of the parasitic protozoa are both immunostimulatory and immunoregulatory components in the context of protozoal infections and the important differences in bioactivity that exist depend upon GPI fine structure.

GPI ACTS AS A CRITICAL TOXIN IN THE CONTEXT OF SEVERE DISEASE SYNDROMES

The data reviewed so far indicate that malaria GPI is sufficient to act as a toxin but do not establish that it plays this role in the context of authentic disease processes. We therefore sought to test this proposition in a credible pre-clinical model of disease. P. berghei ANKA murine malaria has salient features in common with several aspects of the human severe and cerebral malaria syndromes and is thus an accepted model for certain important aspects of the human disease (De Souza & Riley, 2002). It manifests a cytokine-dependent encephalopathy associated with up-regulation of adhesins on the cerebral microvascular endothelium and attendant neurological complications (Grau et al. 1987, 1989a; 1991; Jennings et al. 1997). The more generalized syndromes of pulmonary oedema, lactic acidosis, coagulopathies, shock and renal failure are also observed associated with fatalities (Chang et al. 2001), indicating that the P. berghei ANKA infection models many aspects of clinically severe malaria beyond narrow definitions of cerebral symptomology. Unlike most, but not all, human cerebral malaria cases, however, there is a macrophage infiltrate and compromised blood-brain barrier in the terminal or agonal stages of the murine syndrome. Nonetheless, in the proximal or developmental stages the murine disease is accepted to reflect more accurately the cytokine-dependent inflammatory cascade leading to cerebral and systemic involvement in humans. As recently reviewed (De Souza &Riley, 2002; Miller et al. 2002), P. berghei ANKA thus appears the best available small animal model of clinically severe malaria, suitable for investigations into proximal events controlling the systemic inflammatory cascade, and particularly the regulation of cytokine-dependent events and pathology.

Chemical synthesis of the non-toxic glycan moiety of mature Plasmodium GPI has recently provided a means to test the hypothesis that malaria GPI is causally involved in rodent malaria pathogenesis (Schofield et al. 2002). To prepare an immunogen, the synthetic GPI glycan was conjugated to the carrier protein keyhole limpet haemocyanin (KLH-glycan). Antibodies from mice immunized with KLH-glycan reacted only with malaria-infected rather than uninfected red blood cells, as demonstrated by immunofluorescence and Western blot analyses (Schofield et al. 2002). These results suggested that GPI of parasite origin was sufficiently different from host endogenous GPI molecules (which differ by virtue of amino-sugar or phosphoethanolamine modifications to the core glycan) to be recognized as foreign (non-self). These antibodies specifically neutralized the production of TNF-α from macrophages induced by P. falciparum extracts, confirming their anti-toxic properties in vitro. When challenged with P. berghei ANKA, naïve and sham-immunized mice all died within 5–8 days post infection (p.i.) with severe cerebral syndrome, whereas 75% of those mice immunized with the KLH-glycan were protected (Fig. 1a). There were no significant differences in parasite counts evident at this early stage of infection, but ultimately all mice succumbed at ~15 days p.i. to severe haemolytic anaemia (Fig. 1b) (Schofield et al. 2002) – indicating that parasite growth was unaffected by the antibodies. In studying the effectiveness of KLH-glycan vaccination, the immunized mice were found to be protected against occlusion of brain vasculature (Fig. 1c), pulmonary oedema (Fig. 1d) and blood acidosis (Fig. 1e) (Schofield et al. 2002). These findings not only demonstrated the efficacy of this prototype anti-toxic vaccine, but also showed that Plasmodium GPI is central to the pathogenesis of systemic disease endpoints and the development of a lethal cerebral syndrome in this model. Thus, GPI of P. falciparum satisfies the criteria for a predicted malaria toxin and may thereby contribute to life-threatening disease in humans, albeit a requirement in human malarial pathogenesis and fatality remains to be proven (Boutlis et al. 2002; Suguitan et al. 2004).

Fig. 1. Immunization against synthetic malaria GPI glycan substantially protects against murine cerebral malaria, pulmonary oedema and peripheral blood acidosis. (a), (b) Kaplan-Meier survival plots and parasitemia levels of KLH-glycan-immunized (filled circles) and sham-immunized (open squares) C57BL/6 mice challenged with P. berghei ANKA. (c) Haeomotoxylin and eosin-stained sections of brain tissue showing blood vessels from KLH-glycan immunized (left and centre panels) and sham-immunized (right panel) mice killed 6 days p.i. Pulmonary oedema and blood acidosis indices given as a proportion of the lung wet:dry weight ratio (d) or serum pH (e) form KLH-glycan-immunized and, sham-immunized and naïve infected mice at day 6 p.i. compared to age/sex-matched uninfected mice. Taken from Schofield et al. (1993). Permission pending.

PLASMODIUM FALCIPARUM GPI DISTRIBUTION, METABOLISM AND STRUCTURE

GPI molecules represent a ubiquitous class of glycolipids that function as anchors for transmembrane proteins in eukaryotic cells. The parasitic protozoa are unusual in that they express very high levels of GPI-anchored proteins, GPI-anchored glycoconjugates and a range of unconjugated ‘free’ GPIs and GPI-like structures (GIPLs). Compared with mammalian cells, which typically express in the order of 10⁵ GPI copies per cell, various parasitic protozoa express over 10⁷ GPI copies per cell and most of these GPI-anchored molecules are displayed at the cell surface (Ferguson et al. 1994). In P. falciparum, which shows uniquely low levels of N- and O-linked glycosylation, GPI anchors constitute over 95% of the post-translational carbohydrate modification of parasite proteins (Davidson & Gowda, 2001; Gowda, Gupta & Davidson, 1997). Recent genome-wide sequence analyses predict that approximately 0·5% of the P. falciparum proteome may be post-translationally glycosylated with a GPI anchor, which is typical for a lower eukaryote (Eisenhaber, Bork & Eissnhaber, 2001). From microarray transcription data about 2/3 of these proteins are thought to be expressed in asexual blood stages (P. Gilson, personal communication). Among the GPI-anchored proteins in Plasmodium, the most important are the circumsporozoite protein (CS-2) which coats the insect stage sporozoites, and the merozoite surface proteins (e.g. MSP-1, MSP-2, MSP-4, MSP-5, MSP-10) expressed at the extracellular blood stage of the parasite. All of these proteins are of particular interest because of their central role in host/parasite interactions and in the response to host immunity, and several are under consideration as vaccine candidates. In addition to GPI-anchored proteins, Plasmodium species also contain an abundant pool of structurally identical GPI molecules which are not conjugated to proteins or other structures but exist free in the plasma membrane (free GPIs). In fact, quantification of the relative amount of GPI molecules purified from late schizonts shows a 4-fold molar excess of free versus serine-linked GPIs (resulting from exhaustive pronase digestion of GPI-anchored proteins) by gas chromatography mass spectrometry (GC/MS), suggesting that free GPIs are the predominant GPI-anchored molecules present at the time of schizont rupture (K. Evans, personal communication). These free GPIs appear to be metabolic end-products and functionally important components in their own right.

Despite the diversity of GPI-anchored molecules found in parasitic protozoa, fungi and mammalian cells, all eukaryotic GPI membrane anchors have a common core structure, consisting of a conserved trimannosylglucosaminyl glycan (Man₃-GlN: Manα1,2-Manα1,6-Manα1,4-GlcN; Fig. 2, bold letters) linked to the inositol residue of a phosphatidylinositol (Man₃-GlN-PI). The glycan core of GPI anchors carries an ethanolamine phosphate (EtN-P) group on the third mannose (Etn-P-Man₃-GlN-PI), making it a substrate for GPI transamidase. This enzyme cleaves the peptide bond at the GPI-anchor attachment site near the carboxy-terminal end and creates an amide linkage between the ethanolamine of GPI and the newly generated carboxyl group of the cleaved protein precursor. GPI molecules are assembled in the endoplasmic reticulum (ER) by the sequential addition of N-acetylglucosamine, core mannose and EtN-P residues to PI through the coordinated action of a series of glycsyl- and ethanolaminephosphate-transferases. Critical evaluation of sequence similarity data for malaria homologues of proteins essential for GPI synthesis indicates that the machinery for GPI assembly has been reduced to minimal requirements in Plasmodium (Delorenzi et al. 2002). Although the essential features of this pathway are conserved in all eukaryotes, considerable differences occur in the extent and nature of side chain additions to the glycan backbone as well as the nature of the lipid anchor (for excellent reviews on GPI structure and biosynthesis in parasitic protozoa see (McConville & Ferguson, 1993; McConville & Menon, 2000). As a result, the core glycan may be decorated with accessory carbohydrate (galactose, N-acetylglucosamine, sialic acid) side branches or EtN-P groups in a species- and tissue-specific manner (Fig. 2a-f). Other variations in the GPI structure are found in the lipid backbone (glycerol versus ceramide) and the number (mono-, di-), linkage (acyl versus alkyl), as well as the length and degree of saturation of the lipid chains (i. e. C14[ratio ]0; C16[ratio ]0, C18[ratio ]0, C18[ratio ]1, C18[ratio ]2) in various GPIs (Fig. 2a–f).

Fig. 2. Schematic structure of parasitic GPI-anchors: (a) Leishmania mexicana GIPLs (free iM2, iM3, [EP]iM3 and iM4 GPI forms); Trypanosoma cruzi epimastigote (b) and trypomastigote (c) mucin-derived GPI anchors; (d) Toxoplasma gondii-derived GPI anchor (e) Trypanosoma brucei VSG-derived GPI anchor (f) immature (Man3-GPI) and mature MSP-1-derived (Man4-GPI) GPI anchors of Plasmodium falciparum. Abbreviations: EtN, ethanolamine; P, phosphate; Man, mannose; GlcN, unacetylated glucosamine; Gal, Galactose; GalNAc, acetylated galactosamine, Glu, glucose; m-Ins, myo-Inositol; O, ether and ester bonded Oxygen molecules; Cn:n, number of saturated/ unsaturated carbon molecules in the lipid chain.

The structure of P. falciparum GPIs was first characterized by Schwarz and colleagues (Gerold, Diekmann-Schuppert & Schwartz, 1994), and is one of the simplest described to date (Fig. 2f). Analysis of radiolabeled GPIs suggested that the parasite synthesizes two mature GPI species, which were shown to correspond to a trimannosyl GPI (Fig 2f, Man₃-GPI) and a tetramannosyl GPI (Fig 2f, Man₄-GPI). The Man₃-GPI consists of the evolutionarily conserved core glycan linked to a diacylglycerol (EtN-P-Man₃-GlN-PI-DAG), whereas the Man₄-GPI carries a terminal fourth mannose residue in α1,2 linkage to the third mannose residue (EtN-P-Man₄-GlN-PI-DAG). A report that the terminal fourth mannose residue carries an additional charged substituent (Gowda et al. 1997) remains unconfirmed. Detailed structural analysis by GC-MS has essentially confirmed the monosaccharide composition of the glycan moiety, but found considerable structural heterogeneity of native P. falciparum Man₄-GPI due to variation in fatty acid substituents of the lipid moiety (Naik et al. 2000). GC-MS analysis of HPLC-purified parasite GPIs treated with hydrofluoric acid or phospholipase A2 (for selective removal of fatty acids at the sn-1 and/or sn-2 position of the PI moiety, respectively) detected the presence of at least five distinct GPI diacylglycerol structures in the total free Man₄-GPI pool (Naik et al. 2000). These contained predominantly saturated stearic (C18[ratio ]0) acid and unsaturated oleic (C18[ratio ]1) acid esterified at the sn-1 and sn-2 positions, respectively, and had mainly palmitic (C16[ratio ]0) acid on C2 of the inositol ring of the core glycan (Naik et al. 2000). The differences in fatty acid composition compared with earlier reports may reflect the preferential incorporation of various fatty acids present in different growth media into early GPI precursors (Mitamura et al. 2000; Krishnegowda & Gowda, 2003). Importantly, parasite proteins were found to be anchored almost exclusively by the Man₄-GPIs (Naik et al. 2000; Gerold et al. 1994), indicating a high degree of selectivity towards the presence of the fourth, terminal mannose residue in the GPI anchor moiety. By contrast, all five Man₄-GPI structures were found to be used for protein anchoring (Naik et al. 2000), irrespective of their lipid composition. Overall, Plasmodium GPIs therefore appear to represent an evolutionarily conserved glycolipid structure with a high level of variation in the lipid moiety (the precursors to GPI synthesis) and a structural requirement for the mature Man₄-GPI species (the substrate of GPI transamidase) with regard to protein anchoring.

STRUCTURE: ACTIVITY

The studies described in the previous sections demonstrated that Plasmodium-derived GPIs display potent pro-inflammatory activities that contribute to parasite-associated pathogenesis. Understanding the molecular details of the GPI structure-activity relationship and cell signalling mechanisms involved in these processes may therefore offer valuable insights that could lead to the development of novel therapeutics for malaria. Initial work on murine macrophages and human endothelial cells demonstrated that GPIs of P. falciparum and T. brucei origin can impart at least two discrete signals through structurally distinct elements. The core GPI glycan moiety was found to act as an agonist for the activation of protein tyrosine kinase (PTK), whereas the GPI diacylglycerol moiety was shown to function as a second messenger substrate required for protein kinase C (PKC) activation (Tachado et al. 1997). Of the seven or so PKC members detectable in macrophages, activation by GPI diacylglycerols was restricted to the ε isoform. Activation was associated with a translocation from the cytoplasm to detergent-insoluble membranes. Transient tyrosine phosphorylation of src-family non-receptor tyrosine kinases (e.g. hck, lyn), which peaks within minutes after the addition of low concentrations (10 nM – 1 μM) of GPI to cells, is the earliest measurable event in GPI-mediated signal transduction (Tachado et al. 1997). Although lyso-GPI and deacylated inositolglycan generated by specific enzymatic degradations (i.e. PLA₂, PLD) were shown to be sufficient to activate PTKs, they were found not to promote the full range of signalling events required for the activation of macrophages. In addition, inhibition of signal transduction by specific PKC and NF-κB antagonists could block the production of cytokines in response to P.falciparum-derived GPIs (Tachado et al. 1997). Taken together, these data suggested that GPI-mediated PTK and PKC signals synergise to activate transcription factors from the NF-κB family for the induction of de novo expression of pro-inflammatory loci (e.g. TNF-α, IL-1, IL-6, iNOS, ICAM-1).

One of the important questions raised by these studies relates to the fine structure requirements with respect to the bioactivity of P. falciparum GPIs on host cells. Using structurally defined Plasmodium GPI precursors which differ in the degree of glycosylation together with specific enzymatic and chemical treatments for the selective removal of diacylglycerol or PI-diacylglycerol structures, it was possible to map the minimum structural requirement for PTK activation to the evolutionarily conserved Man₃-GlcN-PI moiety (Tachado et al. 1997). Consistent with this result, the biological activity of total parasite extract, including TNF-α output, can be inhibited by high concentrations of GPI partial structures such as α-methyl mannoside, glucosamine, and phosphatidylinositol. Vijaykumar et al. essentially confirmed and extended these findings by studying the inhibition profile of P. falciparum GPI bioactivity (Vijaykumar et al. 2001). In this study, authors used chemically defined GPI fragments generated by enzymatic treatment of HPLC-purified Man₄-GPI with α-mannosidase (for selective removal of the terminal fourth mannose residue on the core glycan) or chemical treatment with hydrofluoric acid or nitrous acid (for selective cleavage of phosphate or amino bonds of the core glycan, respectively). Consistent with results by Tachado et al. (1997), none of the GPI partial structures alone were able to induce TNF-α production by macrophages (Vijaykumar et al. 2001). Prior incubation of macrophages with inactive Man₄-containing carbohydrate fragments (but not the PI or DAG fragments) released by HF or nitrous acid was found to effectively inhibit the activity of intact Man₄-GPI (Vijaykumar et al. 2001). In contrast to intact Man₄-GPI, the authors reported that P. falciparum Man₃-GPI failed to activate macrophages, nor did Man₃-containing carbohydrate fragments inhibit the ability of Man₄-GPI to induce TNF-α production (Vijaykumar et al. 2001). Surprisingly, the same group subsequently published a paper reporting 80% of TNF-α output from macrophages in response to Man₃-GPI compared to Man₄-GPI (Krishnegowda et al. 2005). The data are therefore inconsistent and at odds with the different inhibition profile of soluble Man₄-containing and Man₃-containing carbohydrate fragments on P. falciparum GPI activity (Vijaykumar et al. 2001). Although discrepancies with regard to the bioactivity of intact Man₃-GPI remain to be resolved, current observations indicate that the fourth, terminal α-mannose residue linked to the core glycan moiety may be critical for the activity of P. falciparum GPIs.

The precise structural requirement of the lipid moiety for the bioactivity of P. falciparum GPIs still remains to be determined. Structure-activity data from different protozoan parasites indicates that ceramide or saturated lipid chains in the monoalkyl or 1-alkyl-2-acyl-glycerol GPIs moieties of Leishmania spp (Fig. 2a) or T. cruzi epimastigote forms (eGPI, Fig. 2b) induce relatively low or no activity in macrophages (Camargo et al. 1997b), compared to the bioactive T. gondii (Fig. 2d), T. brucei (Fig. 2e) and P. falciparum (Fig. 2f) diacylglycerol-containing GPI structures. Moreover, the bioactive GPIs of T. cruzi trypomastigote forms (tGPI, Fig. 2c) and P. falciparum GPI anchors (Fig. 2f) contain mainly unsaturated fatty acid chains at the sn-2 position in the 1-alkyl-2-acyl-glycerol and 1,2-diacylglycerol moieties, respectively (reviewed by (Almeida & Gazzinelli, 2001; Ropert et al. 2002). Whether the presence of the unsaturated fatty acid chain at the sn-2 position of P. falciparum GPI is actually required for its bioactivity is however unclear. Tachado et al. (1997) found that the potency of phospholipase A₂-generated sn-2 lyso GPIs of P. falciparum is drastically reduced, whereas Vijaykumar et al. (2001) and Krishnegowda et al. (2005) reported that these molecules retained nearly all activity. The latter findings are at odds with the observation that chemical removal of fatty acids at the sn-2 position is sufficient for a drastic (~100fold) reduction in the ability of highly purified T. brucei and T. cruzi trypomastigote GPIs to activate macrophages (Camargo et al. 1997b; Tachado et al. 1997, Almeida et al. 2000). The reason for this discrepancy is not clear, but is likely to be due to the difficulty in obtaining adequate amounts of pure, structurally-defined native GPIs. The preparation of native malarial GPI is a labour-intensive process that requires the large-scale growth of P. falciparum in human red blood cell cultures. All growth media used must be endotoxin-free and cultures need to be rigorously tested for mycoplasma contamination before being processed, to eliminate the likelihood of potential contamination by bacterial lipopolysaccharides and mycoplasma-derived lipopeptides. The preparation of malarial GPI itself involves a series of optimized glycolipid extraction and fractionation steps using organic solvents and at least one round of RP-HPLC followed by analytical HP-TLC and GC-MS analyses. The yield of pure native GPI resulting from this procedure is typically low – in the range of 10 μg per 1 litre of culture – and this has proven to be probably one of the most limiting factors in the laboratory setting. The chemical synthesis of partial Plasmodium GPI structures (Schofield et al. 2002) as well as entire protozoan GPI-anchors (Debierre-Grockiego et al. 2003) may soon help to resolve some of the questions regarding the detailed structural requirements for malarial GPI bioactivity.

SIGNALLING PATHWAYS TRIGGERED BY PLASMODIUM GPI – THE TWO-SIGNAL MODEL

The molecular mechanisms of protozoan glycolipid-host cell interaction are poorly understood. In particular, how activation signals mediated by protozoan GPIs are transmitted across the plasma membrane to the interior of host immune cells remains unclear. Tachado et al. (1997) proposed a working hypothesis to explain how bioactive GPIs of T. brucei and P. falciparum can activate host macrophages. The initial interaction is predicted to involve the binding of parasite-derived glycolipids to a putative glycan-specific transmembrane receptor present in the surface of host macrophages. This agrees with the observation that stimulation of macrophages with parasite GPI glycan fragments alone is sufficient to cause rapid transient activation of src-family non-receptor PTKs in host macrophages. Although PTK activation does not provide the full range of signalling events leading to downstream inflammatory gene expression, capture by a glycan-specific receptor is predicted to catalyze the transfer of GPI from the extracellular space to the host membrane. Within the plasma membrane, GPI anchored molecules are known to diffuse laterally and partition into sphingolipid- and cholesterol-rich liquid-ordered microdomains which function both as signalling platforms and endosomal entry and sorting sites. Membrane-permeant diacylglycerol, formed via the hydrolysis of parasite-derived GPI molecules by PLD present in the serum or on the surface of mammalian macrophages (Krishnegowda et al. 2005), could cross cellular membranes to activate an intracellular PKC. Both PTK and PKC signalling pathways were shown to collaborate to activate transcription factors of the NF-κB/rel family and thereby promote expression of pro-inflammatory cytokines. Although certain aspects of this model remain unclear, it has provided a useful hypothetical framework that agrees with the observed structural requirement for both carbohydrate and fatty acid moieties of intact GPI-anchors of P. falciparum, T. brucei, T. cruzi and T. gondii origin for their full pro-inflammatory activity on macrophages (Almeida et al. 2000; Sileghem et al. 2001; Debierre-Grockiego et al. 2003). The major unresolved issues concerning this two-signal model for GPI-specific signal transduction are the identities of the glycan-specific transmembrane receptor and the molecular mechanism by which GPI-derived lipids contribute to the signalling pathways triggered by parasite-derived GPIs.

RECOGNITION OF PARASITE GPIS BY NON-TLR PATTERN RECOGNITION RECEPTORS

In addition to Toll-like receptors, cells of the immune system are equipped with many other glycan-specific surface proteins that function as pattern recognition and uptake receptors on macrophages and DCs. Many of these are members of the Ca²⁺-dependent C-type lectin family and recognize their ligands through the structurally related Ca²⁺-dependent carbohydrate-recognition domains (CRDs) (Cambi & Figdor, 2003). Pathogen recognition by lectin-like receptors (LLRs), such as the soluble serum mannose-binding protein (MBP) (Gadjeva, Takahashi & Tniel, 2004), the macrophage cell-surface mannose receptor (MMR) (Ezekowitz et al. 1990) or the dendritic cell-specific ICAM3-grabbing non-integrin (DC-SIGN) receptor (Cambi et al. 2003), is mediated by binding of terminal mannose residues characteristic of viral, bacterial, fungal and parasite cell surfaces. Ligand binding most likely depends on subtle differences in the arrangements of carbohydrate residues and their branching – creating unique sets of carbohydrate recognition profiles.

Current observations support the idea of carbohydrate-specific interactions in the recognition of protozoan glycolipids by macrophages. It has been suggested, for example, that the potent cytokine producing activity of T. cruzi mucin GPIs purified from trypomastigote forms may be due to the precise number and/or structure of carbohydrate residues, since mild treatment with periodate largely destroys this activity (Camargo et al. 1997a, b). Compared to the low potency of T. cruzi mucin GPIs purified from epimastigote forms the high potency of trypomastigote GPIs is probably due to up to four extra α-galactosyl residues linked to the first mannose residue of the glycan core (Almeida & Gazzinelli, 2001). In agreement with this hypothesis, the presence of the galactose side-chain additions to the core glycan of highly purified GPI anchors of T. brucei origin were also shown to be essential for optimal TNF-α output from macrophages (Magez et al. 1998). It is possible that this carbohydrate pattern is recognized by TLR2, as evidenced by the potent induction of NFκB-dependent gene expression in transfected cell lines by trypomastigote compared to epimastigote GPIs (Campos et al. 2001). Recent work by Vijaykumar et al. (2001) suggested that cell signalling by P. falciparum GPI molecules, which lack the galactosyl side chain modifications of T. cruzi and T. brucei GPIs, may represent a novel mechanism involving the initial recognition of the distal fourth α-mannosyl residue. This proposal is based on the carbohydrate inhibition profile of P. falciparum Man₄-GPI activity on macrophages in vitro, as discussed above.

A good example of a recognition molecule able to differentiate between defined mannose-terminating patterns is the hepatocyte-derived soluble mannan-binding protein. MBP is an abundant C-type lectin found in various mammalian sera including human sera, and plays an important role in innate immunity. It is able to specifically recognize terminal mannose-, N-acetyoglucosamine and fucose-rich carbohydrate patterns on the surfaces of pathogenic microorganisms and initiates complement deposition on relevant (non-self) surfaces (reviewed by Gadjeva et al. 2004). In addition to carbohydrate patterns present on bacteria, viruses and fungi, it has been observed that MBP is also able to recognize major surface glycoconjugates and glycoproteins of Leishmania and P. falciparum parasites. For example, using thin layer chromatogram-overlay assays it was shown that MBP binds to the mannose-terminating di-, tri- and tetrasaccharide fragments derived from purified LPG and GIPLs that are highly expressed on promastigote and amastigote stages of most Leishmania species (Green et al. 1994). As demonstrated by immunofluorescence microcopy of live L. major and L. mexicana promastigotes (Green et al. 1994), FACS analyses show that human MBL may also function as an opsonin for schizont-infected erythrocytes (Klabunde et al. 2002). Attempts to identify specific parasite-derived surface ligands found that MBP binds to at least two glycoproteins that incorporated ³H-glucosamine and were immunogenic in humans (Klabunde et al. 2002; Garred et al. 2003). In patients with homozygous mutations in the MBL gene complicated malaria was associated with significantly higher parasite burdens compared to MBL-competent counterparts (Garred et al. 2003), but whether MBP deficiency is a significant risk factor for severe malaria is still a matter of debate (Bellamy et al. 1998).

Whether tissue macrophages or dendritic cells express a specific lectin-like receptor, such as cell-surface mannose receptor or DC-SIGN capable of recognizing the terminal mannose residue on the core glycan of Plasmodium GPIs remains to be determined. Evidence is emerging that LLRs not only play a role as phagocytic antigen uptake receptors, but may allow signalling upon fusion of late endosomes/lysosomes via TLRs expressed on endosomal membranes (i.e. TLR-3, TLR-7, TLR-9). Whereas TLR signalling through MyD88, IRAK and MAPK resulting in the activation of NF-κB is quite well understood, signalling through lectin-like receptors via alternative pathways is still relatively obscure. Nevertheless, in view of the different dose-response and TLR/MyD88 signalling profiles of T. cruzi trypomastigote and P. falciparum GPIs, it appears that cellular immune responses to GPIs of various parasitic protozoa may involve multiple lipid- and carbohydrate-specific pattern recognition and signal transduction mechanisms.

SIGNALLING BY GPI-ANCHORED PROTEINS AND GLYOLIPIDS THROUGH LIPID-DEPENDENT MICRODOMAINS

A striking property of mammalian GPI-anchored proteins as well as glycolipids is that their ligation on the cell surface by natural ligands or suitable antibodies regulates numerous signal transduction and activation responses (reviewed by (Horejsi et al. 1998; Kasahara & Sanai, 2000). Such signalling capacity is surprising, considering that these molecules have no transmembrane or intracellular moieties and thus no direct contact with the cell interior. The signalling capacity of both endogenously expressed and exogenously added GPI-anchored proteins appears in many cases to involve their association with cytoplasmic signalling molecules in detergent-resistant membrane domains (DRMs). DRMs are stabilized against disruption by cold detergents through liquid-ordered packing of component cholesterol, (glycol)sphingolipids and glycerolipids. An important factor in this lipid-dependent organisation appears to be the length and fluidity (saturation) of the fatty acid residues attached with the GPI-anchored molecules (Schroeder et al. 1998; Benting et al. 1999). Low buoyant density DRMs, often called lipid rafts, can be isolated by density gradient ultracentrifugation. Such membrane fractions are characteristically enriched in GPI-anchored proteins, dually acylated cytoplasmic signalling or adaptor proteins (e.g. Src-family kinases, trimeric G proteins, Ras, LAT) and different raft-organizing proteins (e.g. caveolins, stomatins, flotillins) (Brown & London, 2000). In vivo, liquid-ordered membrane domains are thought to be quite small and/or extremely dynamic and consequently below the level of resolution of the light microscope (Jacobson & Dietrich, 1999). Thus, they are optimally investigated with biophysical techniques that can detect short-range interactions. Recent studies employing chemical crosslinking and fluorescence resonance energy transfer (FRET) microscopy techniques estimated that a fraction (20–40%) of GPI-anchored proteins form extremely tight clusters of nanometer size (~5–50 nm), each containing a few (4–15) different molecules on the surface of living cells (Friedrichson & Kurzchalia, 1998; Varma & Mayor, 1998; Sharma et al. 2004). Following cross-linking with antibodies or natural ligands these rafts can coalesce into visible and stable functional domains. Lipid-dependent membrane organization has been proposed to be responsible for the endocytosis and sorting of raft-associated proteins, and for providing platforms for the dynamic assembly of GPI-anchored proteins, transmembrane proteins and cytoplasmic signalling molecules in a variety of biological contexts (e.g. T cell receptor, B cell receptor and FcεRI signalling) (Simons & Toomre, 2000; Anderson & Jacobson, 2002; Mayor & Riezman, 2004).

Of particular interest in the present context of Plasmodium GPI signalling is accumulating evidence indicating that the innate recognition of different pathogens may involve the interaction of CD14 with various transmembrane receptors within supramolecular activation clusters (i.e. TLR-2, TLR-4, TLR-6, Mac-1, CXCR4, GDF5 and others) (Henneke et al. 2001; Pfeiffer et al. 2001; Triantafilou et al. 2001a, b). By analogy to the immunological synapse, it has been proposed that the entire bacterial recognition system may be based around the formation of a signalling complex of receptors at the site of CD14-LPS ligation within membrane microdomains (Triantafilou & Triantafilou, 2002). Consistent with this theory, FRET analyses have shown that ligation of CD14 by LPS and ceramide provokes ligand-specific clustering and translocation of TLR4 into lipid rafts (Pfeiffer et al. 2001; Triantafilou et al. 2002). A recent study demonstrates that the atypical PKC isoform, PKC-ζ, is critical in the regulation of LPS-induced TLR4 lipid raft mobilization within macrophages (Cuschieri, Umanskiy & Solomkin, 2004). The cellular mechanisms involved in this process remain incompletely understood, but recruitment of PKCζ to plasma membrane lipid raft microdomains during insulin stimulation of adipocytes was shown to be mediated by TC10, a member of the Rho family of small GTP-binding proteins that is constitutively localized to plasma membrane lipid raft microdomains (Kanzaki et al. 2004).

It is possible that similar processes are required for signalling by biologically active parasite GPIs. We recently generated fluorescently-labelled semi-synthetic P. falciparum GPIs to study the specio-temporal distribution of exogenously added parasite glycolipids in the host cell membrane by confocal microscopy. When added to live mammalian tissue culture cells in vitro, these molecules were initially distributed homogeneously, but after several hours concentrated at cell-to-cell contacts and acquired a punctuate distribution on the cell surface (unpublished observation). These preliminary data demonstrate that a proportion of Plasmodium glycolipids incorporate into host plasma membrane, but whether the observed cell surface puncta represent sites involved in uptake or signalling by Plasmodium glycolipids is currently unknown. FRET analyses using fluorescently-tagged semi- or fully synthetic Plasmodium GPI molecules may allow us to further investigate the possible association of P. falciparum GPI with CD14, TLRs and other signalling molecules within plasma membrane microdomains in live cells.

CONCLUSIONS AND PERSPECTIVES

While our understanding of malarial GPI bioactivity has improved significantly over the past decade, several key issues remain to be resolved. For example, while the concept that the GPI anchors of parasitic protozoa constitute immunostimulatory and regulatory agents in the context of protozoan infections has been widely recognized, it is still difficult to pinpoint the precise molecular features of parasite GPIs that underlie the differences in their bioactivities. This is especially true in the case of malarial GPI toxin, with independent studies coming to different conclusions with regard to the structural requirement for the presence of the terminal fourth mannose residue on the core glycan and/or the unsaturated fatty acid chain at the sn-2 position of P. falciparum GPI for full activation of pro-inflammatory responses of macrophages. One of the main factors contributing to inconsistencies in the present data, and at the same time limiting the pace of our research, is the difficulty in obtaining adequate amounts of pure, structurally defined P. falciparum GPIs. The chemical synthesis of milligram quantities of partial GPI structures as well as the entire Plasmodium GPI-anchor will make this key reagent more accessible and facilitate the creation of novel probes that will help to resolve some of the questions regarding fine structural requirements for malarial GPI bioactivity, in the near future.

A significant development regarding the potential mechanism of GPI bioactivity has been the demonstration that GPI anchors of various parasitic protozoa can mediate cellular immune responses via the Toll-like receptor family. The original observation that T. cruzi trypomastigote GPIs can specifically activate TLR2/MyD88-dependent signalling pathways to produce inflammatory responses in host macrophages has recently been confirmed and extended using P. falciparum GPI. The use of macrophages from mice deficient in individual or multiple TLRs (e.g. TLR-2, TLR-4, TLR-2 & 4, TLR-9 etc.), essential adaptor molecules (i.e. MyD88), and downstream signalling molecules (e.g. ERK1/2, JNK-1 or JNK-2) has allowed investigators to dissect the relative contributions of specific TLR(s) and downstream signalling pathways activated by P. falciparum GPI and other malarial ligands. Recent studies indicate that malarial GPIs and haemozoins are preferentially recognized by TLR2/TLR1 and TLR9, respectively. Both rely on the MyD88-dependent activation of specific signalling pathways that mediate the production of proinflammatory cytokines and nitric oxide. However, the contribution of bioactive malaria products to severe malarial pathology and the role of TLRs or other pattern recognition receptors in innate immunity is only just beginning to be characterized. Importantly, the role of TLR/MyD88 pathways in mediating pro-inflammatory cytokine production and clearance of blood-stage parasites during the acute phase of malaria infections in mice infected with various Plasmodium species, as well as the role of GPI as a critical toxin in the context of human malaria infections remains to be firmly established. Given the complexity of the host/pathogen interface, signalling of malarial GPIs via TLR/MyD88-dependent pathways most likely represents only one example of how multiple pattern recognition receptors may function in regulating innate and adaptive immune responses in the context of protozoal infections. Further analysis of the interaction of P. falciparum parasites and the host immune system may shed new insights into the molecular mechanisms underlying severe malarial pathogenesis with potentially important implications for the development of novel therapeutics for malaria treatment.