Rosiglitazone Modulates the Innate Immune Response to Plasmodium falciparum Infection and Improves Outcome in Experimental Cerebral Malaria

Abstract

For severe malarial syndromes such as cerebral malaria, adverse clinical outcomes are often mediated by the immune system rather than caused by the parasite directly. However, few therapeutic agents have been developed to modulate the host’s immunopathological responses to infection. Here, we report that the peroxisome proliferator-activated receptor γ (PPARγ) agonist rosiglitazone modulated the host response to malaria by enhancing phagocytic clearance of malaria-parasitized erythrocytes and by decreasing inflammatory responses to infection via inhibition of Plasmodium falciparum glycosylphosphatidylinositol-induced activation of the mitogen-activated protein kinase (MAPK) and nuclear factor–κB (NF-κB) signaling pathways. We found that, in the Plasmodium berghei strain ANKA experimental model of cerebral malaria, rosiglitazone modified the inflammatory response to malarial infection and improved the survival rate even when treatment was initiated as late as day 5 after infection. Furthermore, rosiglitazone reduced the parasitemia in a CD36-dependent manner in the Plasmodium chabaudi chabaudi hyperparasitemia model. These data suggest that PPARγ agonists represent a novel class of host immunomodulatory drugs that may be useful for treatment of severe malaria syndromes

Plasmodium falciparum malaria is a major determinant of the childhood mortality rate, causing an estimated 1–3 million deaths annually [1, 2]. Cerebral malaria is among the deadliest complications of P. falciparum infection and affects an estimated 785,000 African children each year [1, 2]. There is no specific therapy for cerebral malaria, and despite the use of rapidly active antimalarial therapy such as parenteral quinine or artesunate, mortality rates remain high [3–5]. This may be attributable to the observation that interventions for malaria are antiparasitic, even though poor outcomes associated with cerebral malaria appear to be mediated more by the host’s immunopathological responses to infection than by the parasite per se [3–6]

Key events in the pathogenesis of severe and cerebral malaria include the sequestration of parasitized erythrocytes within the cerebral microvasculature; dysregulated inflammatory responses to infection, contributing to immune-mediated tissue injury, endothelial activation, and upregulation of sequestration receptors; and high parasite burdens that further enhance parasite sequestration and immunopathological responses [7]. Defining the mechanisms underlying the host’s response to malaria may identify novel targets for immunomodulation and interventions to improve the outcome of persons with severe malaria syndromes

Malaria-associated fatalities occur predominately among individuals who are not immune to Plasmodium infection. Survival of infected humans and in murine models of malaria appears to be critically linked to the host’s ability to contain replication of blood-stage parasites during the acute phase of infection [6]. Because malaria-specific immune responses are largely absent during acute Plasmodium infection, innate mechanisms appear to be essential in controlling parasite replication and in decreasing the risk of progression to severe and fatal disease. The innate immune response therefore represents an attractive target for therapeutic intervention [6]

Mononuclear phagocytes represent an essential first line of innate defense against malaria [6, 8–12]. Macrophage pattern-recognition receptors, including Toll-like receptors (TLRs) and scavenger receptors such as CD36, are important components in the regulation of immune responses [13, 14]. Pattern-recognition receptors sense a wide range of microbial molecules and activate proinflammatory responses to infection. Parasites and parasite products, such as P. falciparum glycosylphosphatidylinositol (pfGPI) and hemozoin plus parasite DNA, induce the release of proinflammatory cytokines via interaction with pattern-recognition receptors, including TLR2, TLR9, and CD36 [11, 15–22]. Macrophages in general [6, 9, 10] and macrophage CD36 in particular have been shown to mediate the clearance of parasitized erythrocytes and, in experimental models of malaria, to help control replication of blood-stage parasites during acute infection and to enhance survival of the host [6, 11, 19–21]

On the basis of these observations, we hypothesized that pharmacological modulation of innate immunity through pathways involving CD36 and related pattern-recognition receptors might increase parasite clearance, modify deleterious host inflammatory responses to infection, and improve outcome. CD36 transcription is regulated by the nuclear receptor heterodimer peroxisome proliferator-activated receptor γ–retinoic X receptor (PPARγ-RXR) [23]. PPARγ-RXR is activated when either partner is ligand bound, and it regulates the transcription of a variety of genes, including those encoding pattern-recognition receptors [23, 24]. PPARγ agonists have also been shown to modulate inflammatory responses, including decreased secretion of proinflammatory cytokines via inhibition of the activity of transcription factors, such as activator protein–1 (AP-1) and nuclear factor–κB (NF-κB) [25–28]. We postulated that US Food and Drug Administration (FDA)–approved PPARγ agonists could, via their potential to improve parasite clearance and regulate inflammatory responses to infection, serve as immunomodulatory agents for the treatment of malaria. Here, we demonstrate in vitro and in 2 complementary models in vivo that rosiglitazone, a PPARγ agonist [29], enhanced phagocytic clearance of parasites, regulated inflammatory responses to infection, and, in an experimental model of fatal cerebral malaria, improved survival

Methods

ParasitesCultures of the P. falciparum laboratory clone ItG, maintained and synchronized as described elsewhere [19, 20], were treated with a mycoplasma removal agent (ICN). All cultures tested negative for mycoplasma by polymerase chain reaction analysis before use. Culture supernatants were collected, aliquoted, and frozen for subsequent use. Plasmodium berghei strain ANKA (PbA; Malaria Resource Center) and Plasmodium chabaudi chabaudi strain AS (PccAS; provided by Dr. M. Stevenson [McGill University, Montreal, Canada]) were maintained by regular passage in naive mice

Phagocytosis assaysPhagocytosis assays were performed as described elsewhere [19–21] (see also the supplemental methods in the Appendix, which is not available in the print edition of the JournalAppendix, which appears only in the electronic edition of the Journal). A total of 1×10⁶ peripheral blood mononuclear cells (PBMCs) or 2×10⁵ murine macrophages were plated and treated for 48 h with rosiglitazone or with dimethyl sulfoxide (DMSO) as a control. Fragment crystallizable (Fc) regions (concentration, 20 μg/mL) were used to block Fc receptors. Various monoclonal antibodies (concentration, 5 μg/mL) were used when appropriate. Synchronized parasites were layered on top for a final ratio of 10 parasitized erythrocytes to 1 monocyte. Hypotonic lysis was used to remove nonphagocytosed parasitized erythrocytes. Phagocytosis was quantified microscopically by counting the total number of parasitized erythrocytes observed in 500 monocytes/macrophages

Detection of CD36 surface expression by flow cytometryMacrophages were treated for 48 h with rosiglitazone or with DMSO as a control, and surface expression of CD36 was detected by staining the macrophages with anti-CD36 monoclonal antibodies. Isotype matched antibody controls were also evaluated. Cells were fixed in 1% paraformaldehyde/phosphate-buffered saline and analyzed using the Epics Elite flow cytometer (Beckman-Coulter). Data were analyzed using FlowJo software (TreeStar)

Isolation and purification of GPI fromP. falciparumProtein-free pfGPI was isolated and purified by high-performance liquid chromatography (HPLC) as described elsewhere [16, 17] (see also the supplemental methods in the appendix)

Tumor necrosis factor (TNF) assaysHuman PBMCs (concentration, 5×10⁵ cells/well) or murine thioglycollate-elicited peritoneal macrophages (concentration, 2×10⁵ cells/well) were treated with rosiglitazone or with DMSO as a control, with or without the addition of HPLC-purified pfGPI (concentration, 200 nmol/L per mL). After incubation for 24 h at 37°C, the supernatants were collected and assayed for TNF, using an enzyme-linked immunosorbant assay

Signal transduction assaysThioglycollate-elicited murine peritoneal macrophages were pretreated with rosiglitazone for 24 h and stimulated with pfGPI for various periods. Cell lysates were collected, separated by 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis, and transferred to polyvinylidene fluoride membranes. Blots were probed with antibodies recognizing total and phosphospecific extracellular signal–regulated kinase 1/2 (ERK1/2), c-Jun N-terminal kinase (JNK), p38, and IκB kinase α (IκBα), incubated with the appropriate horseradish peroxidase–conjugated secondary antibody, and developed using an electrochemiluminescence-based system [17]

Murine survival and serum cytokine studiesAll experiments involving animals were reviewed and approved by the University of Toronto animal use committee and were performed according to the university’s animal ethics guidelines. Male C57BL/6 mice (Charles River Laboratories) aged 8–12 weeks were used in most experiments. Cd36^−/− C57BL/6 mice were bred in the animal facility at the University of Toronto. Mice were kept under pathogen-free conditions with a 12-h light cycle. One week before infection or 1, 3, or 5 days after infection, mice were fed ad libitum a regular powder diet or powder chow containing 50 mg/kg of rosiglitazone in powder format. For each mouse, infection was initiated by intraperitoneal injection of 1×10⁶ freshly isolated erythrocytes parasitized with PbA or PccAS. Mice in the rosiglitazone arm continued to receive the drug for the remainder of the experiment. The course of infection was monitored daily for 18 or 21 days via Giemsa-stained thin-blood smears to determine parasitemia level

In some experiments, mice were sacrificed on day 3, 5, or 7, and peripheral blood was collected by cardiac puncture. Serum was stored at −80°C and was later analyzed for levels of TNF and transforming growth factor β (TGF-β) by an enzyme-linked immunosorbant assay (R&D Systems)

Statistical analysisAll in vitro experiments were performed in duplicate or triplicate and repeated at least 3 times. Data are mean values (± standard deviations), unless otherwise noted. Statistical significance, defined as a P value of <.05, was determined by analysis of variance with a post hoc Tukey-Kramer test or a Mann-Whitney&amp;rank sum test, depending on whether the data were normally distributed. Normality was assessed using the Kolmogorov-Smirnov test. Survival studies for PbA infections were done using 5 mice per group and were repeated 3 times. Survival studies for PccAS were done using 6–10 mice per group and were repeated once. Statistical significance was determined by a log-rank test. Parasitemia time courses were analyzed using 2-way analysis of variance with a Bonferroni post-hoc test. Statistical analyses were performed using GraphPad Prism software

Results

Rosiglitazone Up-regulates CD36, Increases Phagocytosis of P. falciparum–Parasitized Erythrocytes, and InhibitspfGPI-Induced TNF Secretion

Failure to control the replication of blood-stage parasites during acute Plasmodium infection and dysregulated inflammatory responses to malaria are associated with a poor clinical outcome. We hypothesized that therapeutic interventions to modify these components of malaria pathogenesis would be of clinical benefit. Macrophage-mediated uptake of parasitized erythrocytes, primarily via engagement of scavenger receptors such as CD36, has been shown to contribute to the control of acute infection [6, 9–11, 19, 20]. To determine whether rosiglitazone could up-regulate CD36 surface expression and parasitized-erythrocyte phagocytosis, human and thioglycollate-elicited murine peritoneal macrophages were treated with increasing concentrations of rosiglitazone. Rosiglitazone induced a dose-dependent increase in surface expression of CD36 (figure 1A1B1D and 1E) that was associated with a dose-dependent increase in phagocytosis of nonopsonized parasitized erythrocytes (figure 1C and 1F). Phagocytosis was significantly inhibited by monoclonal antibody blockade of CD36, suggesting that it is dependent on CD36

Figure 1

Rosiglitazone up-regulates surface expression of CD36 and CD36-mediated clearance of Plasmodium falciparum–parasitized erythrocytes. Human and murine macrophages treated with rosiglitazone for 48 h were assessed for CD36 surface expression (by flow cytometry) and for their ability to phagocytose nonopsonized P. falciparum–parasitized erythrocytes. A and D CD36 fluorescence intensities (FIs) for human (A) and murine (D) macrophages. Data for macrophages treated with 100 μmol/L of rosiglitazone are shown in black data for vehicle control–treated macrophages are shown in dark gray and data for macrophages stained with isotype antibody control are shown in light gray. B and E CD36 geometric mean FIs (GMFIs) for human (B) and murine (E) macrophages. Here and elsewhere, white bars denote the control dimethyl sulfoxide (DMSO [vehicle control]), and black bars denote various rosiglitazone concentrations. C and F Phagocytosis of nonopsonized parasitized erythrocytes by human (C) and murine (F) macrophages. Hatched bars uptake by macrophages pretreated with 5μg/mL of monoclonal anti-CD36 antibody. Data are from a representative experiment performed in triplicate. All experiments were repeated at least 3 times. For C *P=.02, P<.003, and P<.001 for comparisons of DMSO with rosiglitazone concentrations of 10 μmol/L, 50 μmol/L, and 100 μmol/L, respectively. For F *P=.008 and P<.004 for comparisons of DMSO with rosiglitazone concentrations of 50 μmol/L and 100 μmol/L, respectively. All statistical comparisons were performed by analysis of variance and a post hoc 2-tailed Tukey-Kramer test

To determine whether rosiglitazone could modulate malaria-induced inflammatory responses, human PBMCs and murine thioglycollate-elicited peritoneal macrophages were incubated with increasing concentrations of rosiglitazone and HPLC-purified pfGPI [11, 16, 17]. Rosiglitazone inhibited pfGPI-induced TNF secretion by human cells (figure 2A) and murine cells (figure 2B) in a dose-dependent manner

Figure 2

Rosiglitazone inhibits Plasmodium falciparum glycosylphosphatidylinositol (pfGPI)–induced signaling and tumor necrosis factor (TNF) production. A and B Human peripheral blood mononuclear cells (A) and murine macrophages (B) were treated with rosiglitazone (concentration, 10–100 μmol/L; black bars) or dimethyl sulfoxide (DMSO [vehicle control]; white bars) plus pfGPI (200 nmol/L per mL). TNF levels were assessed by enzyme-linked immunosorbant assay 24 h after treatment. Experiments were repeated at least 3 times with similar results. Data shown are from a representative experiment. For A *P<.001 and P<.001 for comparisons of DMSO with rosiglitazone concentrations of 50 μmol/L and 100 μmol/L, respectively (n = 4). For B *P<.008, P<.001, and P<.001 for comparisons of DMSO with rosiglitazone concentrations of 10 μmol/L, 50 μmol/L, and 100 μmol/L, respectively (n = 6). All statistical comparisons were performed by analysis of variance and a post hoc 2-tailed Tukey-Kramer test. C and D Murine macrophages were pretreated with rosiglitazone (Rosi) or DMSO for 24 h and incubated with pfGPI (200 nmol/L per mL) for 30 min (C) or 0–60 min (D). Cell lysates were analyzed for phosphospecific extracellular signal–regulated kinase 1/2 (ERK1/2), p38, c-Jun N-terminal kinase (JNK), and IκB kinase α (IκBα) degradation. Total p38 and ERK1/2 levels were used to assess equal loading. Experiments were repeated 3 times. +, present; −, absent

pfGPI promotes cytokine induction via TLR2-dependent activation of the mitogen-activated protein kinase (MAPK) and NF-κB signaling pathways [11, 16, 17]. To investigate the mechanisms involved in rosiglitazone-mediated inhibition of the pfGPI-stimulated inflammatory response, we examined the effects of rosiglitazone on pfGPI-induced signaling pathways. pfGPI induced phosphorylation of JNK, ERK1/2, and p38, and it promoted the degradation of IκBα, a precursor step to NF-κB activation (figure 2) [16, 17]. Treatment with rosiglitazone resulted in a dose-dependent inhibition of the phosphorylation of JNK, ERK1/2, and p38 (figure 2C). In addition, rosiglitazone inhibited pfGPI-induced IκBα degradation (figure 2D). These observations indicate that rosiglitazone reduced pfGPI-induced signaling and proinflammatory cytokine production in vitro

Rosiglitazone Modulates the Innate Immune Response to Malaria and Improves Infection Outcome In Vivo

To extend our analysis to the in vivo setting, we examined the efficacy of rosiglitazone in modulating innate responses in murine models of malaria. Because no single experimental murine model of malaria exists that encompasses all aspects of clinical disease due to P. falciparum infection in humans, we investigated the efficacy of rosiglitazone in 2 complementary murine models. To examine the ability of rosiglitazone to modulate the immunopathological response in the host, we used the PbA model of experimental cerebral malaria [8]. To examine the drug’s ability to enhance parasite clearance and reduce parasite burden, we used the PccAS model of hyperparasitemia [6, 8]

Rosiglitazone modifies inflammatory responses and improves survival in an experimental model of cerebral malariaSimilar to P. falciparum in humans, PbA-susceptible mice (e.g., strain C57BL/6) develop symptoms of cerebral malaria during infection, including a cytokine-associated encephalopathy, neurological symptoms, and acidosis culminating in a fatal outcome [8]. PbA-associated cerebral malaria is characterized by an unbalanced cytokine response to infection, with elevated levels of proinflammatory cytokines and inadequate induction of regulatory cytokines [6, 8]. We used this experimental model to investigate whether rosiglitazone favorably modulates host inflammatory responses to infection and confers protection against cerebral malaria

Control and rosiglitazone-treated C57BL/6 mice were infected with 1×10⁶ PbA-parasitized erythrocytes and assessed once daily for parasitemia and serum TNF and TGF-β levels and twice daily for morbidity and survival (figure 3). All control mice developed neurological signs characteristic of cerebral malaria, including limb paralysis, movement disorder, ataxia, convulsions, and coma, and succumbed to infection 6–10 days after infection onset. In contrast, rosiglitazone-treated mice had a significantly greater survival rate (43% vs. 0%; P=.007) (figure 3A). The rosiglitazone-treated mice that succumbed to infection also had neurological signs. The parasitemia level did not differ between control and rosiglitazone-treated mice (figure 3B), consistent with previous studies in which dysregulated cytokine responses, rather than parasitemia per se, were shown to be responsible for mortality during the acute phase of PbA infection [8, 30]

Figure 3

Rosiglitazone improves survival and modulates inflammation in a murine model of cerebral malaria. C57BL/6 mice that received chow with or without rosiglitazone (concentration, 50 mg per kg of chow) were infected with 1×10⁶Plasmodium berghei strain ANKA–parasitized erythrocytes via intraperitoneal injection. A Survival, assessed twice daily (15 mice per group; P<.008, by the log-rank test). B Parasitemia level, assessed daily. C and D Serum tumor necrosis factor (TNF; C) and transforming growth factor β (TNF:TGF-β ratio; D) levels were assessed on days 3, 5, and 7 of infection (5 mice per group). *P<.003, by a 2-tailed Mann-Whitney&amp;rank sum test

During acute infection, early proinflammatory responses, particularly release of IFN-γ and TNF, are required to facilitate parasite clearance, but they must be balanced, in part by anti-inflammatory/immunoregulatory cytokines such as TGF-β and IL-10, to limit the host-mediated immunopathological response [30–35]. To assess the effect of rosiglitazone on PbA-induced inflammatory responses, we examined serum levels of TNF and TGF-β on days 3, 5, and 7 of infection. Circulating levels of TNF in rosiglitazone-treated mice were significantly lower than those in control mice (figure 3C). Moreover, the ratio of TNF to TGF-β for rosiglitazone-treated mice was significantly lower than that for control mice on all days assessed (figure 3D). Taken together, these data demonstrate that rosiglitazone treatment altered host inflammatory responses to PbA infection and enhanced survival in an in vivo experimental model of cerebral malaria

To determine at what point during the course of infection rosiglitazone might still confer enhanced survival, we examined C57BL/6 mice that were infected with PbA and that commenced treatment 1 week before infection or 1, 3, or 5 days after infection (figure 4). As in the previous experiment, all control mice died of infection (all died <9 days after infection), and the survival rate among mice that received rosiglitazone 1 week prior to infection was significantly greater than that among control mice (60% vs. 0%; P<.002). Initiation of rosiglitazone treatment on day 1 or 3 after infection resulted in survival rates similar to that for mice commencing treatment 1 week before infection (P<.389 for the comparison between mice treated 1 week before infection and mice treated on day 1 after infection, and P=.731 for the comparison between mice treated 1 week before infection and mice treated on day 3 after infection). Likewise, survival rates for these 2 treatment groups were greater than that for control mice (0% for the control group, compared with 40% for mice treated on day 1 [P=.014] and 50% for mice treated on day 3 [P<.003]). Survival among mice that started rosiglitazone therapy as late as day 5 after infection was significantly greater than that among control mice (10% vs. 0%; P=.016) but significantly less than that among pretreated mice (10% vs. 60%; P=.041). These data indicate that therapeutic administration of rosiglitazone as late as day 5 after infection enhanced the rate of survival in an experimental model of cerebral malaria

Figure 4

Therapeutic administration of rosiglitazone as late as 5 days after infection enhances survival. C57BL/6 mice received chow plus rosiglitazone (concentration, 50 mg per kg of chow; gray lines) 1 week before infection (A) or 1 day (B) 3 days (C) or 5 days (D) after infection. Control mice (black lines) received no additions to their chow. Mice were infected with 1×10⁶Plasmodium berghei strain ANKA–parasitized erythrocytes via intraperitoneal injection. Survival was assessed daily. P<.002 (A) P<.015 (B) P<.003 (C) and P=.016 (D) (n=10 mice per group). All statistical comparisons were performed by the log-rank test

Rosiglitazone improves parasite clearance in a CD36-dependent manner, in a hyperparasitemia model of malariaThe acute phase of PccAS infection is characterized by hyperparasitemia that results in death if parasite replication is not adequately contained [6, 8]. Macrophages have been shown to be key effector cells in the control of hyperparasitemia [9, 10]. Early control of parasite replication is dependent on innate mechanisms involving scavenger receptors, including CD36, and is largely independent of opsonins, including parasite-specific IgG and complement components [9–11, 36–38]

We used the PccAS model of malaria to examine the efficacy of rosiglitazone in facilitating the clearance of parasitized erythrocytes and reducing the parasite burden in vivo. Control and rosiglitazone-treated C57BL/6 mice were infected with 1×10⁶ PccAS-parasitized erythrocytes (figure 5, which appears only in the electronic edition of the Journal). Parasitemia, morbidity, and mortality were assessed daily, starting on day 5 of infection. C57BL/6 mice are innately resistant to PccAS infection [6, 8], and accordingly, survival rates did not differ significantly between rosiglitazone-treated mice and control mice (90% and 80%, respectively). However, rosiglitazone had a significant effect on parasitemia (P<.001, by 2-way analysis of variance). Compared with control mice, the parasitemia level (defined in terms of the mean percentage of parasitized cells [± standard error of the mean]) was lower in the rosiglitazone-treated group at all times after day 8 of infection and was significantly lower on day 11 of infection (41.9%±3.84% vs. 31.0%±4.61%; P=.03, by the Bonferroni post-hoc test) and on day 12 of infection (41.3%±2.48% vs. 26.5%±3.61%; P<.001, by the Bonferroni post-hoc test)

Figure 5

Rosiglitazone reduces parasitemia in a murine model of hyperparasitemia in the presence but not the absence of CD36. C57BL/6 mice (A and B) or Cd36^−/− C57BL/6 mice (C and D) that received chow with or without rosiglitazone (concentration, 50 mg/kg of chow) were infected with 1×10⁶Plasmodium chabaudi chabaudi strain AS–parasitized erythrocytes via intraperitoneal injection. A and C Survival was assessed daily (n=10 C57BL/6 mice per group, and n=6 Cd36^−/− C57BL/6 mice per group). Analysis by means of the log-rank test revealed no significant differences between control mice and rosiglitazone-treated mice. B and D Parasitemia level was assessed daily beginning on day 5 after infection (n=10 C57BL/6 mice per group, and n=6 Cd36^−/− C57BL/6 mice per group). For wild-type mice (B) the parasitemia level among the rosiglitazone-treated animals was significantly less than that among the control mice on days 11 and 12 (*P=.03 for day 11 and P<.001 for day 12, by 2-way analysis of variance with a Bonferroni post-test). However, for Cd36^−/− mice (D) the parasitemia level did not vary significantly between rosiglitazone-treated mice and control mice

To determine whether the rosiglitazone-induced reduction in parasite burden observed in the PccAS model was dependent on CD36, we infected control and rosiglitazone-treated Cd36^−/− mice (figure 5, which appears only in the electronic edition of the Journal). Although these mice are on the resistant C57BL/6 genetic background, their inability to produce CD36 has previously been shown to render them more susceptible to PccAS infection [11]. The mortality rate (50% for the control group and the rosiglitazone group) and parasitemia level did not differ significantly between control mice and rosiglitazone-treated mice. In summary, rosiglitazone treatment in the absence of CD36 did not decrease the parasite burden or improve the survival rate among PccAS-infected mice, suggesting that CD36 must be present for rosiglitazone to have a beneficial effect

Discussion

Host immunopathological responses are important contributors to severe and fatal outcomes in a number of life-threatening infectious disease syndromes, including cerebral malaria [3, 39]. Despite this, few therapeutic agents have been developed to modulate deleterious host immune responses to infection. Here, we show that rosiglitazone, an FDA-approved PPARγ agonist, enhanced macrophage phagocytosis of parasitized erythrocytes in vitro and reduced parasitemia level in vivo, modified innate inflammatory responses to malaria in vitro and in vivo, and conferred improved survival in an experimental model of cerebral malaria. In experiments mimicking clinical conditions, rosiglitazone enhanced survival even when initiated up to 5 days after infection, at a time when symptoms of cerebral malaria are manifesting. This suggests that rosiglitazone may be useful for the treatment of established Plasmodium infections

For humans with Plasmodium infection and in murine models of malaria, survival is linked to the ability of the host to generate a regulated inflammatory response and control parasite replication during the acute stage of infection [6]. Excessive or dysregulated inflammatory responses to infection have been consistently implicated in malaria-associated immunopathological processes [6, 31, 34, 35]. In this study, we demonstrate that rosiglitazone inhibited pfGPI-induced signaling and TNF secretion in vitro and modulated innate inflammatory responses, particularly the balance between pro- and immunoregulatory cytokine levels, in mice infected with PbA. pfGPI is thought to be a major proinflammatory mediator, and CD36 cooperates with TLR2 in recognizing and initiating responses to it. The cerebral malaria syndrome observed in PbA infection occurs even in the absence of CD36 [40], and therefore, the beneficial actions of rosiglitazone observed in this model may be independent of its effects on CD36 expression. Instead, rosiglitazone may exert anti-inflammatory activity via direct transcription regulation, by inhibition of the activity of transcription factors such as AP-1 and NF-κB [25–28]

Studies in humans and mice also support an important role for mononuclear phagocytes in the clearance of blood-stage parasites and the early control of the parasite burden during acute infection [6, 9–11, 19–21]. Here, we demonstrate that rosiglitazone up-regulated macrophage CD36 expression and CD36-mediated uptake of P. falciparum–parasitized erythrocytes in vitro. Using the in vivo model of PccAS-induced hyperparasitemia, we demonstrate that rosiglitazone treatment resulted in a CD36-dependent decrease in parasite burden. In mice with CD36 sufficiency, rosiglitazone significantly reduced parasitemia level. Notably, rosiglitazone failed to decrease the parasitemia level when it was administered to CD36-deficient mice, suggesting that the effects of rosiglitazone on parasitemia are CD36-dependent and may be attributable to an increase in CD36-mediated clearance. In the PbA model of experimental cerebral malaria, rosiglitazone did not affect the level of parasitemia. However, unlike P. falciparum and P. chaubaudi it is unclear whether P. berghei–parasitized erythrocytes are cleared by CD36-mediated phagocytosis [11, 40]

The contribution of CD36 to Plasmodium pathogenesis or, conversely, to protection during Plasmodium infection remains undetermined. CD36 was initially identified as a sequestration receptor for parasitized erythrocytes, leading to the assumption that it contributes to the pathogenesis of cerebral malaria; however, several lines of evidence would appear to challenge this premise [11, 9–21, 40–43]. Because CD36 expression in the brain is minimal or absent, direct cytoadherence of parasitized erythrocytes to endothelial CD36 is unlikely to account for cerebral sequestration, although it has been proposed that adhesion may be mediated via bridging through platelet CD36 [44, 45]. CD36 is highly expressed in microvascular endothelium of skin and adipose tissue and may direct parasites to these nonvital sites and away from cerebral microvasculature [20, 21]. This hypothesis is supported by studies demonstrating that significantly greater frequency of parasitized-erythrocyte binding to CD36 occurs in cases of nonsevere disease, by reports showing that protection against cerebral malaria observed in individuals with Southeast Asian ovalocytosis is associated with increased adhesion of parasitized ovalocytic erythrocytes to CD36 [41, 42], and by population data linking CD36 deficiency with an increased susceptibility to cerebral malaria [43]. In contrast, recent reports examining other erythrocyte disorders associated with protection from severe disease, including sickle cell and thalassemia, have reported decreased CD36-mediated adhesion due to altered PfEMP-1 expression on parasitized erythrocytes [46]. However, these were in vitro studies performed in the absence of serum, and direct in vivo evidence from human infections is at present lacking. Although the precise contribution of CD36 to protection or pathogenesis will require additional investigation, our in vitro and in vivo data support a beneficial role for PPARγ agonists in acute malaria infection that is at least partly mediated by the modulation of innate immune responses, including those involving macrophage CD36

Rosiglitazone’s ability to modify the sequestration of parasitized erythrocytes at the endothelium is of potential concern. Although PPARγ agonists have been shown to up-regulate both CD36 and ICAM-1 expression, these agonists had minimal effects on the expression of sequestration receptors on endothelial cells and did not up-regulate endothelial cell adhesion of parasitized erythrocytes [47]

There is an urgent need for adjunctive therapeutic interventions that improve the outcome of cerebral malaria. However, given that the current costs of new drug discovery exceed $750 million per new chemical entity, development of new therapeutic agents for diseases with a primary burden in the developing world faces sizable economic barriers [48]. High-throughput screening of FDA-approved drugs for novel indications represents one potential approach to expedite drug discovery and overcome financial obstacles. Another strategy, exploited in this study, is to identify disease critical pathways and make use of available drugs predicted to act on these pathways through known interactions with transcriptional response elements or other cellular targets

The data presented in this article suggest that PPARγ agonists such as rosiglitazone may have clinical usefulness as adjuncts to standard therapy for falciparum malaria in humans. However, recent systematic reviews have highlighted discordance between treatment outcomes in animal models and those in humans via clinical trials [49, 50]. Therefore, we have extended our animal studies and recently completed a randomized, double-blind, placebo-controlled treatment trial of P. falciparum malaria acquired on the Thai-Myanmar border to investigate the efficacy of rosiglitazone as adjunctive therapy in falciparum malaria. In this randomized clinical trial, rosiglitazone treatment significantly improved parasite clearance, lowered the parasite burden, and decreased the levels of inflammatory biomarkers associated with adverse clinical outcomes

In summary, these data provide direct in vitro and in vivo evidence that rosiglitazone can modulate host response and improve outcome during experimental cerebral malaria, thereby suggesting a potential role for this class of immunomodulators in the management of severe and cerebral malaria

Potential conflicts of interest: The University Health Network holds intellectual property pertaining to the use of PPARγ-RXR agonists for the treatment of inflammatory states

Presented in part: 55th Meeting of the American Society of Tropical Medicine and Hygiene, Atlanta, Georgia, 12–16 November 2006 (abstract 61A)

Financial support: Canadian Institutes of Health Research (CIHR) Team Grant in Malaria (to K.C.K.); operating grant MT-13721 (to K.C.K.); Genome Canada, through the Ontario Genomics Institute (to K.C.K.); McLaughlin-Rotman Centre and McLaughlin Centre for Molecular Medicine (to K.C.K. and W.C.L.); CIHR Canada Research Chair (to K.C.K. and W.C.L.)