Divergent Roles of IRAK4-Mediated Innate Immune Responses in Two Experimental Models of Severe Malaria

Abstract

Severe malaria represents a clinical spectrum of disease. We propose that innate immune inflammatory responses to malaria play key roles in the pathogenesis and clinical outcomes of distinct severe malaria syndromes. To investigate this hypothesis, mice deficient in IRAK4, central to Toll-like receptor (TLR)-mediated signaling, were studied in two experimental models of malaria: Plasmodium berghei (PbA) and Plasmodium chabaudi (PccAS). Irak4^–/–mice had decreased pro-inflammatory cytokine production during infection in both models. However, animals were relatively protected from PbA-associated symptoms compared with wild-type mice, whereas Irak4^–/– animals were more susceptible to PccAS-associated disease. These results show that IRAK4-mediated innate immune inflammatory responses play critical roles in divergent clinical outcomes in murine malaria models. As such, integrated approaches, using more than one model, are required to fully understand the parasite/host interactions that characterize severe malaria, and more importantly, to fully assess the effect of adjunctive therapies targeting innate immune responses to malaria.

The clinical presentation of a number of bacterial (e.g., leprosy¹) and parasitic diseases (e.g., leishmaniasis²), manifest across a broad clinical spectrum defined by the host immune response to infection. The poles of this spectrum are defined by either an excessive or a deficient inflammatory response to the infecting pathogen. Appropriate management of affected individuals varies according to the patient's position along this clinical and immunological continuum.

Of the estimated one million deaths from malaria each year, most are attributable to severe malaria caused by Plasmodium falciparum infection.³ Severe malaria has been increasingly recognized to be a multi-system disorder with a broad spectrum of clinical manifestations that include overlapping, evolving, or complex presentations of disease.⁴ Severe malaria primarily occurs in non-immune individuals where host defense is dependent, at least partly, on innate immune responses to infection.⁵

We propose that malaria may also present along a “clinical-immunological” spectrum with differential host innate immune responses to infection, resulting in divergent clinical outcomes.^6,7 Individuals with an early deficient inflammatory response that fails to clear infection, define one pole of this spectrum. Whereas “hyper-responsive” hosts generate excess inflammation and cellular activation, which can ultimately contribute to the cerebral complications associated with cerebral malaria (CM),⁸ define the opposite pole.⁶ Representing the mid-range of immunological response are those with regulated responses sufficient to control parasite replication without invoking immunopathological tissue injury. In actual human infection, individuals may have complex responses and resultant mixed clinical syndromes. However, we propose a conceptual framework by which the clinical spectrum may be extended to murine models. Plasmodium berghei ANKA (PbA), a model of cerebral malaria where inflammation exacerbates disease^9,10 and Plasmodium chabaudi chabaudi AS (PccAS), a model of malaria infection where pro-inflammatory responses are required for parasite clearance and survival^11,12 represent the framework's divergent poles, which correspond to differing host inflammatory responses to infection.

The PbA infection of susceptible mice (e.g., C57BL/6) is characterized by excessive early inflammation associated with high levels of pro-inflammatory cytokines, such as tumor necrosis factor (TNF)/lymphotoxin-α and interferon-γ (IFN-γ),^13,14 contributing to host immunopathology and death within 6–10 days of infection caused by cerebral complications.¹⁵ The PccAS infection is commonly used as a murine malaria model of blood stage parasite replication without cerebral complications.⁵ Susceptible mice lack strong pro-inflammatory responses, resulting in higher parasite loads and ultimately death.¹⁶ Resistant mice show symptoms of disease, usually peaking 10–12 days post-infection, but survive through a controlled inflammatory response.¹⁷ As such, strong inflammatory immune responses to malaria are linked with mortality in the PbA model, but necessary for survival in the PccAS model.

Based on the hypothesis that the innate immune-mediated inflammation plays a key role in defining pathogenesis and determining clinical outcome in severe malaria, we investigated mice deficient in inflammatory signaling caused by genetic deletion of the interleukin-1 receptor-associated kinase 4 (IRAK4) in both the PbA and PccAS murine models of malaria. IRAK4 integrates signals from all three Toll-like receptors (TLRs) so far implicated in the pathogenesis of malaria¹⁸ and from IL-1 and IL-18 cytokine signaling.¹⁹ Unlike TLRs, IL-1 and IL-18 signaling do not appear to be necessary for the development of CM.^20–22 TLRs, a family of pattern recognition receptors, are essential components of the innate immune response to a variety of microbial pathogens. After ligand binding, TLRs interact with adaptor proteins to activate transcription factors, induce inflammatory responses, and activate adaptive immunity.²³ Mutations in the TLR adaptor protein IRAK4 have been shown to cause a primary immunodeficiency syndrome characterized by increased susceptibility to pyogenic bacterial infections in humans.²⁴

The role of TLRs in the host inflammatory response to malaria has been intensely studied. Several pro-inflammatory Plasmodium components are recognized by TLRs in vitro. The malaria toxin glycosylphosphatidylinositol,²⁵ PbA peroxiredoxin,²⁶ and malarial DNA complexed to parasite hemozoin^27,28 have been shown to signal through TLR2, 4, and 9, respectively. However, the degree of stimulation and the contribution of each receptor to the overall host inflammatory response and outcome of infection remain controversial.^20,21,29 Furthermore, human studies, have implicated polymorphisms in TLR4, TLR9, and MAL (an adaptor protein in MyD88-depedent signaling), but not TLR2, with disease severity.^30–32 Precisely how these polymorphisms affect the clinical manifestations of malaria infection remains unclear.

Our experiments show that the survival of Irak4^–/– animals was enhanced in PbA-infected mice, but worsened in PccAS-infected animals; altering early host inflammatory responses had opposite results at either pole of the proposed conceptual framework. The Irak4^–/– mice showed marked differences to wild-type (WT) mice during infection in both experimental malaria models. During PccAS infection, mortality was significantly increased in Irak4^–/– mice compared with WT controls (Fisher's exact test, P < 0.0001, N = 54, Figure 1A). Parasitemia, the control of which has been linked to host immune responses,^33–35 was increased in Irak4^–/– mice (P < 0.0001, N = 75, Figure 1B), whereas inflammatory cytokine levels were markedly decreased (Figure 1C). Elevated levels of IFN-γ^12,36 and TNF^11,17 are necessary for resolving PccAS infection. The peaks of plasma IFN-γ and TNF (347.7 pg/mL and 94.5 pg/mL, respectively, N = 40, Figure 1C) observed in WT mice on Day 10 of PccAS infection, coincided with peak parasitemia and were reduced to near-background levels in Irak4^–/– mice (IFN-γ: 48.4 pg/mL, P < 0.0001; and TNF: 35 pg/mL, P = 0.01; N = 45, Figure 1C). Levels on measured on Day 1 or post-peak parasitemia (d15 and d18) showed no such difference. Interestingly, levels of the anti-inflammatory cytokine IL-10, which has been implicated as a regulator of inflammation in the PccAS model³⁷and shown to impede parasite clearance during Plasmodium yoelii infection in mice,³⁸ were similar in Irak4^–/– and WT animals over the course of infection (Figure 1C). Collectively, these results show that IRAK4 plays a key role in regulating pro-inflammatory responses required for parasite clearance and survival during PccAS infection. It was recently demonstrated that macrophages from Irak4^–/– mice are impaired in their ability to phagocytose malaria-infected red blood cells.³⁹ Because this mechanism has been shown to significantly contribute to parasite clearance in the PccAS model,⁴⁰ impaired parasite clearance may contribute to the increased mortality observed in the Irak4^–/– animals infected with PccAS.

Figure 1.

Divergent outcomes of PbA and PccAS infections in Irak4^–/– animals compared with wild-type (WT) controls. (A) Unlike in WT controls (8–12 week-old C57BL/6 mice), survival was significantly decreased in Irak4^–/– animals infected with 1*10⁶ PccAS parasites by intraperitoneal injection (Fisher's exact test, P < 0.0001, N = 54). (B) Parasitemia, measured every 2 days by thin blood smear, was significantly increased during PccAS infection (4.9% mean increase over time course (mixed linear model [SPSS 16.0] of the response variable as a function of time and genotype [WT, Irak4^–/–] as fixed effects, with animal number and experiment identifiers as random effects, P < 0.0001, N = 75). (C) Median levels of plasma IL-10 (top panel), IFN-γ (mid panel), and TNF (bottom panel) collected on Days 5 and 10 and measured by Cytometric Bead Array (CBA) (BD Biosciences, Canada). Day 10 levels of TNF and IFN-γ were decreased in Irak4^–/– animals compared with WT mice (Mann-Whitney test, P = 0.01, and P < 0.0001, respectively, N = 40). (D) Survival was significantly increased (Fisher's exact test, P < 0.0001, N = 66) in Irak4^–/– compared with WT animals after infection with 1*10⁶ PbA parasites by intraperitoneal injection (E) Cerebral mononuclear cell infiltration during PbA infection was not apparent in Irak4^–/– animals. Representative images from histopathological examinations of brains from WT and Irak4^–/– mice, isolated on Day 6 post-infection. Brains were formalin fixed, prepared into 5-μM sections and stained with hematoxylin and eosin. Cerebral malaria (CM)-susceptible mice (WT) displayed characteristic histological evidence of CM. (F) Parasitemia, measured every 2 days by thin blood smear, was similar between both groups during PbA infection (P = 0.1, N = 66). (G) Decreases in TNF levels (33.1 pg/mL vs. 19.5 pg/mL, Mann Whitney test, P = 0.01, N = 34 [bottom panel], data represented as median and individual values) were observed at Day 5 of PbA infection as were decreases in IFN-γ levels (253.4 pg/mL vs. 44.8 pg/mL, Mann-Whitney test, P = 0.03, N = 34 [top panel], data represented as median and individual values). Statistical details for all panels can be found in Supplementary Table 1. This figure appears in color at www.ajtmh.org.

In contrast, Irak4^–/– mice survived significantly longer (P < 0.0001, N = 66, Figure 1D) than WT mice following infection with PbA, with little evidence of cerebral mononuclear cell infiltration (Figure 1F, left and right panels). During PbA infection, increased production of TNF⁹ and IFN-γ¹⁰ is associated with appearance of neurological signs of CM. On Day 5 of PbA infection, levels of pro-inflammatory cytokines in plasma were markedly reduced in Irak4−/− versus WT mice (IFN-γ: 253.4 pg/mL versus 44.8 pg/mL, P = 0.03, N = 34 and TNF: 33.1 pg/mL versus 19.5 pg/mL, P = 0.01, N = 34, Figure 1G). These data implicate IRAK4 in mediating early inflammatory responses during PbA infection, which are deleterious to the host. The prolonged survival observed in Irak4^–/– mice, despite equivalent parasitemia, compared with CM-susceptible WT controls (Figure 1E), highlights the critical role of the inflammatory response, rather than parasite burden, in mediating immunopathology. During PbA infection in susceptible C57BL/6 animals, the expansion of regulatory T cells is limited.²⁰ Cerebral malaria development can be inhibited through regulatory T cell expansion,⁴¹ although this remains controversial,⁴² and may be limited in the Irak4^−/− mice as a result of increases in regulatory T cells.

Our studies show a non-redundant role for the TLR-adaptor molecule IRAK4 as an essential regulator of inflammatory response and disease outcome in two experimental models of severe malaria (Figure 1). As such, our data may in part reconcile the currently discordant observations regarding the role of TLRs in malaria,^21,43,44 by confirming the need to disrupt a common adaptor to impart a functional difference. Loss of a common TLR-adaptor such as IRAK4 would overcome any presumed functional redundancy that might exist between the individual TLRs previously implicated in malaria pathogenesis (i.e., TLR2, 4, and 9).

Suppressed inflammation conferred by IRAK4 deficiency improved survival in the “hyper-responsive” model, but conferred a worse clinical outcome in the “hypo-responsive” model (Figure 2). Interestingly, work by other groups supports this hypothesis. Griffith and others²⁰ reported that MyD88 deficiency on a C57BL/6 background conferred protection to experimental CM, whereas it conferred susceptibility on a Balb/c background. As with our results, modulating TLR signaling led to divergent outcomes, emphasizing the need to consider the spectrum of host response to infection and the use of integrated models of disease pathogenesis. Importantly, in human infection, complex responses to infection will result in mixed clinical syndromes, rendering prediction of disease outcome even more challenging.

Figure 2:

Summary of findings. Infection with PccAS or PbA malaria parasites in C57BL/6 mice leads to divergent outcomes, which can be reversed by disrupting inflammatory responses. The C57BL/6 mice are resistant to PccAS infection (left panel); the mice show symptoms of malaria but control parasite burden and survive. However, in Irak4^–/– animals, inflammatory responses to malaria are decreased while parasite loads are increased, a process which may be caused by impaired macrophage function, resulting in reduced survival. When infected with PbA (right panel), C57BL/6 mice generate an excessive inflammatory response, leading to immunopathology and ultimately death of the host. Irak4^–/– animals, by contrast, have ablated malaria inflammatory responses resulting in decreased cerebral pathology, possibly through the expansion of regulatory T cells, and improved survival. Lower levels of inflammatory cytokines generated during malaria infection in Irak4^–/– animals reverses the outcomes of both PbA and PccAS infections. Animals lacking IRAK4 become susceptible to PccAS, but resistant to PbA, supporting our hypothesis that PbA and PccAS reflect a clinical spectrum of disease defined by host response to infection. This figure appears in color at www.ajtmh.org.

Treatment with potent anti-malarials such as artesunate has decreased mortality in severe malaria, however the case-fatality rate remains high.^45,46 Modulating the host immune response during malaria infection^47–52 has been suggested as potential adjunctive therapy in conjunction with anti-malarials to further improve outcome in severe malaria.⁵³ However, our data highlight possible limitations of this approach and the potential to exacerbate disease manifestations at other points along the clinical spectrum, because inhibiting TLR-associated responses may prove beneficial in CM, but may potentially aggravate other clinical manifestations of malaria. Before initiating clinical trials of putative disease immunomodulators, an integrated approach that examines models representing the clinically relevant spectrum of disease may be informative in mitigating harmful effects that might result from unanticipated manipulation of protective host responses to infection, especially in light of the ever-increasing resistance to all anti-malarials.⁵⁴ Future studies of adjunctive therapies for malaria may need to consider individualized assessment of patient's clinical-immunological response to infection before instituting immunomodulatory therapies for severe malaria.

Note: A supplemental table appears at www.ajtmh.org.