Edited by Michael B. A. Oldstone, The Scripps Research Institute, La Jolla, CA, and approved June 4, 2018 (received for review January 30, 2018)
Cerebral malaria (CM) is a neurological complication of malaria infection that, despite antimalarial drug treatment, results in fatality or neurodisability in approximately 25% of cases. Thus, there is an urgent clinical need to develop therapies that can improve the efficacy of antimalarial drugs to prevent or reverse cerebral pathology. Here, we show in an experimental mouse model of CM (ECM) that IL33 administration can improve survival and reduce pathology in the brain over antimalarial drugs alone. Mechanistically, we demonstrate that IL33 enhances recovery from ECM by inhibiting NLRP3 inflammasome-induced inflammatory responses within the brain. These results suggest that IL33 and NLRP3 inflammasome inhibitors may be effective adjunctive therapies for CM.
Cerebral malaria (CM) is a serious neurological complication caused by Plasmodium falciparum infection. Currently, the only treatment for CM is the provision of antimalarial drugs; however, such treatment by itself often fails to prevent death or development of neurological sequelae. To identify potential improved treatments for CM, we performed a nonbiased whole-brain transcriptomic time-course analysis of antimalarial drug chemotherapy of murine experimental CM (ECM). Bioinformatics analyses revealed IL33 as a critical regulator of neuroinflammation and cerebral pathology that is down-regulated in the brain during fatal ECM and in the acute period following treatment of ECM. Consistent with this, administration of IL33 alongside antimalarial drugs significantly improved the treatment success of established ECM. Mechanistically, IL33 treatment reduced inflammasome activation and IL1β production in microglia and intracerebral monocytes in the acute recovery period following treatment of ECM. Moreover, treatment with the NLRP3-inflammasome inhibitor MCC950 alongside antimalarial drugs phenocopied the protective effect of IL33 therapy in improving the recovery from established ECM. We further showed that IL1β release from macrophages was stimulated by hemozoin and antimalarial drugs and that this was inhibited by MCC950. Our results therefore demonstrate that manipulation of the IL33–NLRP3 axis may be an effective therapy to suppress neuroinflammation and improve the efficacy of antimalarial drug treatment of CM.
Cerebral malaria (CM) is a severe manifestation of Plasmodium falciparum infection, which affects 2–3 million people each year, mainly young children in Africa (1). The only treatment for CM is antimalarial drugs, typically in the form of parenteral artesunate or quinine compounds. Such treatment fails to prevent mortality in a quarter of CM patients, leading to the death of ∼300,000 people each year (1⇓–3). Moreover, up to 26% of individuals develop residual neurological deficits following antimalarial drug treatment and recovery from CM (4, 5). Thus, CM remains a leading cause of mortality and neurodisability in tropical regions (1⇓⇓⇓–5). Consequently, there is a critical clinical need for development of more effective therapies for CM that will enhance the protective effects of antimalarial drugs.
The cerebral processes contributing to the pathophysiology of CM and those that undermine recovery from the syndrome after antimalarial drug treatment are poorly understood (1, 6⇓–8). However, there is a growing consensus that targeting the host proinflammatory immune response to infection may be an effective strategy to enhance the antimalarial drug treatment success of CM (7, 8). Indeed, serological and/or cerebral spinal fluid concentrations of proinflammatory cytokines and chemokines, including TNFα, IL6, IL1β, IFN-γ, and CXCL10, frequently correlate with the development of CM and, in some cases, the severity of CM (7, 8). Proinflammatory processes may disrupt CM recovery by activating the brain endothelium, causing permeability of the blood–brain barrier, activation of astrocytes and microglia, disruption of neuronal signaling, and recruitment of circulating leukocytes (1, 7⇓–9). All of these events have been observed in brains of individuals with fatal CM (1, 6⇓⇓–9). In particular, it is believed that cerebrovascular dysfunction is a critical pathological process in CM development and fatal outcome (1, 7, 9). Therefore, intracerebral inflammatory responses at time of treatment may prevent re-establishment of brain homeostasis, leading to the failure of antimalarial drug treatment.
In this study, to identify immune candidates for therapy of CM, we optimized a preclinical model of Plasmodium berghei (Pb) ANKA-induced murine experimental cerebral malaria (ECM) (10) where antimalarial drug treatment of established ECM leads to suboptimal recovery, associated with significant mortality and development of severe cerebral pathology. Using this infection–drug cure model of ECM, we have performed a nonbiased whole-brain RNA-seq time-course analysis during antimalarial drug chemotherapy. We subsequently identified IL33 as a key regulator of cerebral inflammatory pathways during fatal ECM and in the acute period after antimalarial drug treatment. Injection of IL33 along with antimalarial drugs significantly improved the recovery of mice with established ECM, potentially through reduction of NLRP3-dependent inflammasome activation. Consistent with this, direct inhibition of the NLRP3 inflammasome using the specific inhibitor MCC950 phenocopied the protective capacity of IL33 in improving recovery from ECM. Overall, these data indicate that pharmacological strategies targeting the IL33–NLRP3 axis could potentially be beneficial for the treatment of CM.
To study the recovery from established malaria-induced cerebral pathology, we adapted the conventional Pb ANKA ECM model (10) to recapitulate the clinical settings associated with the treatment of CM. C57BL/6 mice infected with Pb ANKA were treated daily with the antimalarial drugs artesunate [the front line drug for treatment of severe malaria (2)] and chloroquine (as a representative quinine compound), both at 30 mg/kg, or vehicle alone. Treatment began at the onset of neurological dysfunction, as defined by a rapid murine coma and behavior scale (RMCBS) score of ≤15 (11), on day 6 post infection (d6) (SI Appendix, Fig. S1).
Peripheral parasitemia developed exponentially before rapidly reducing upon antimalarial drug treatment (Fig. 1A). Despite their potent parasiticidal activity, administration of antimalarial drugs [artesunate and chloroquine (AC)] failed to prevent mortality in ∼25% of mice (Fig. 1B). Interestingly, in the cases where antimalarial drug treatment was unsuccessful, drug-treated mice succumbed more rapidly to ECM than vehicle-treated controls (80% compared with 20% of deaths on day 6, respectively) (Fig. 1B). Antimalarial drug treatment also failed to prevent significant deterioration in neurological function in the critical 6- to 12-h period post treatment (d6.5), with drug-treated mice exhibiting comparable levels of neurological dysfunction to those of vehicle-treated mice (Fig. 1C). Drug-treated mice still exhibited substantial neurological impairment at 24 h post administration (d7), although this was ameliorated compared with the level of neurological dysfunction observed in untreated mice with fatal ECM (Fig. 1C). We then compared the neuropathology between mice that survived following treatment with antimalarial drugs (d7: 16–24 h post treatment), with mice that were not drug-treated and were therefore in the agonal stages of the disease (d7: 16–24 h post vehicle treatment). Consistent with observations of residual neurological deficits in drug-treated mice (Fig. 1C), mice that survived following treatment with antimalarial drugs (d7) exhibited a reduction in, but not complete abrogation, of various neuropathological features associated with CM (1, 6⇓⇓⇓–10) including cerebrovascular parasitized red blood cell (pRBC) accumulation (Fig. 1D), hemostasis (Fig. 1E), vascular leakage (Fig. 1F), hemorrhage (Fig. 1G) axonal injury (Fig. 1H), and myelin damage (Fig. 1I). None of the neuropathological features were observed in naive mice (as we have previously shown in ref. 10). Collectively, these data demonstrate that administration of antimalarial drugs to mice with established ECM resulted in a similar mortality rate as antimalarial drug treatment of CM (2, 3) and did not fully prevent or reverse associated neuropathology.
Antimalarial drug treatment promotes suboptimal recovery from ECM. Mice were infected with Pb ANKA GFP and treated with artesunate and chloroquine (AC) or vehicle (Veh) at the onset of ECM. (A) Peripheral parasitemia, (B) survival curves, and (C) RMCBS scores of mice after infection (d0) and drug treatment (gray box). (D–I) Brains were examined 16–24 h after treatment (d7) for (D) GFP+ parasites (green), costained with lectin (red) and DAPI (blue); (E) erythrocyte-congested vessels indicative of hemostasis (H&E); (F) extravascular IgG indicative of vasogenic edema (DAB counterstained with hematoxylin); (G) hemorrhage (H&E); (H) β-APP accumulation (green) indicative of axonal injury, costained with erythrocytes (red) and DAPI (blue); (I) myelin damage (H&E). Data are presented as means ± SEM (A–C) n = 12–97 from 2 to 10 infections and (D–I) n = 6 from two infections. (Scale bars: 25 μm.) #P < 0.05 d0 versus d7 in AC-treated. (A and C) Specified comparisons for parasitemia and RMCBS were made by Mann–Whitney U tests. (B) Comparison made by log rank test (D–I) Comparisons made by Mann–Whitney U or t tests as detailed in SI Appendix, SI Methods. *P < 0.05, **P < 0.01, ***P < 0.001, (all vs. Veh), #P < 0.05 AC group day 7 versus day 0.
As therapeutic strategies targeting only the parasite failed to prevent substantial mortality or morbidity, we utilized a nonbiased systems approach to identify potential targets for additional therapy. We compared the cerebral (whole-brain) transcriptomes of mice by RNA-seq before infection (d0), at the onset of ECM (d6), in late-stage (agonal) ECM without drug treatment (d7), and at various time points after drug treatment (d7+AC, d10, d14, d30, and d60). Principal component analysis (PCA) demonstrated that antimalarial drug administration led to a rapid change in the brain transcriptome (d7+AC and d10) compared with that in mice with early onset ECM (d6) and agonal ECM (d7), the latter two of which exhibited largely overlapping PCA transcriptome signatures (Fig. 2A). The brain transcriptome returned to homeostasis quickly post resolution of ECM, with d14, d30, and d60 samples clustering with d0 (Fig. 2A). Antimalarial drug administration did not reverse the majority of the gene changes (< or >1.5-fold change and q value < 0.05, compared with d0) that were established in the brain at onset of ECM and that were also observed in fatal ECM (Fig. 2B). Drug treatment did, however, lead to segregated expression of many genes compared with agonal ECM (Fig. 2B). Very few genes were differentially expressed in brains at d14, d30, or d60 compared with d0 (Fig. 2C).
Expression profiling and pathway analyses indicate that IL33 is a potentially important gene negatively regulating pathogenesis in the late stages of ECM. Whole-brain transcriptomic analyses were performed before infection (d0), at the onset of ECM (d6), in agonal ECM (d7), and after antimalarial treatment (d7+AC, d10, d14, d30, d60). (A) Principal component analysis of whole-brain transcriptomes. (B and C) Venn diagrams defining overlap of DEGs (< or > 1.5-fold change and q value < 0.05). (D) K-means and hierarchical clustering of DEGs (normalized to d0). (E) Bipartite cytoscape network defining enriched (GO slim) biological processes within the 13 filtered upstream regulator combined protein–protein interaction networks (DEGs within protein–protein interaction network identified within d7 and/or d7+AC groups, compared with d0). (F) IL33 gene expression in the brain compared by one-way ANOVA. (G) IL33 protein in brain homogenates measured by BioLegend LEGENDPlex, compared by t test. Data are presented as mean ± SEM **P < 0.01, ***P < 0.001, ****P < 0.0001 all versus d0 or naive.
We then sought to understand in more detail the transcriptional responses that undermined the effectiveness of antimalarial drug treatment of established ECM. A total of 4,825 differentially expressed genes (DEGs) were identified when all time points were compared, separately, to d0. DEGs were clustered by k-means into eight clusters and ranked by hierarchical clustering (Fig. 2D, with the gene expression pattern in each cluster visually represented in SI Appendix, Fig. S2A). We then performed gene ontology analysis to assess the biological processes significantly enriched within each cluster (SI Appendix, Fig. S2B). In general, antimalarial drug treatment did not acutely modify the expression of the majority of the biological processes involved in inflammation and immunological activation (clusters I, VI, and VII) established at the onset of ECM at d6 (Fig. 2D). Instead, antimalarial drug treatment altered the expression of genes involved in nervous system development, metabolism, and axogenesis (cluster VIII), transcription, apoptosis, and cell adhesion (clusters II and IV), and DNA repair and regulation of lymphocyte activation (cluster V) (SI Appendix, Fig. S2B). Together, these data show that antimalarial drugs failed to rapidly alter the intracerebral expression of large numbers of genes defining the inflammatory signature of the brain during and post ECM. Instead, in the surviving mice, antimalarial drug administration appeared to significantly modulate expression of genes involved in brain function.
To define the key genes controlling the cerebral transcriptional landscape during agonal ECM and following antimalarial drug treatment, we identified the upstream regulators (URs) within each cluster (SI Appendix, Fig. S2C). A transcription factor (TF) enrichment analysis revealed that most of the URs were controlled by a genetic regulatory network involving several TFs. Based on this information, we filtered this list to identify 13 URs the expression of which was not regulated by TFs, as we hypothesized that these genes were strong candidates for independently and rapidly controlling the transcriptome of the brain during and following treatment of ECM. Importantly, these 13 genes were predicted to control multiple inflammation and immune-related processes in the brain during agonal ECM (d7) and immediately following antimalarial drug treatment (d7+AC) (Fig. 2E).
Of the 13 identified independently controlled master URs, IL33, which was present in cluster VIII of the heat map (Fig. 2D), was of particular interest due its protective role in other inflammatory neuropathologies, including Alzheimer’s disease, stroke, and spinal cord injury (12⇓⇓–15). IL33 gene expression was significantly down-regulated in the brain during agonal ECM and in the acute phase post antimalarial drug treatment, before returning to levels observed in naive mice from day 10 (Fig. 2F). IL33 protein levels were similarly reduced in the brain following antimalarial drug treatment of ECM (d7+AC) compared with levels in naive mice (Fig. 2G). These data identified IL33 as a potential immunotherapy to dampen inflammation, re-establish homeostasis in the brain, and improve the success of antimalarial drug treatment of established ECM.
To investigate whether administration of IL33 could reduce the mortality and/or neuropathology associated with antimalarial chemotherapy of established ECM, we administered antimalarial drugs alone or together with IL33 to Pb ANKA-infected mice at the onset of neurological dysfunction (d6). IL33 was administered as a single dose [0.02 mg/kg, human equivalent dose (HED) 0.0016 mg/kg] alongside antimalarial drugs (both at 30 mg/kg) on the first day of treatment. IL33 administration did not alter peripheral parasitemia (Fig. 3A); however, IL33 treatment significantly improved survival over antimalarial drugs alone (100% with IL33 vs. 71% without) (Fig. 3B). Furthermore, IL33 significantly improved RMCBS scores of mice, compared with mice treated with antimalarial drugs alone, at both 6–12 (d6.5) and 16–24 (d7) h after treatment (Fig. 3C). We then assessed the effects of IL33 on the neuropathology that we had previously observed in mice that had survived following antimalarial drug treatment (Fig. 1). We compared the neuropathology of mice treated with antimalarial drugs alone (d7) with that of mice treated with combined IL33 and antimalarial drugs (d7). IL33 administration significantly reduced a number of indices of cerebral pathology, including cerebrovascular pRBC accumulation (Fig. 3D), hemostasis (Fig. 3E), vascular leakage (Fig. 3F), hemorrhage (Fig. 3G), and axonal injury (Fig. 3H). Myelin damage was unaltered (Fig. 3I). When IL33 treatment was administered without antimalarial chemotherapy (on d6), all mice succumbed to ECM on d7, demonstrating that IL33 alone was not able to promote recovery from established ECM (SI Appendix, Fig. S3). These results demonstrate that IL33 significantly improved the effectiveness of antimalarial drug treatment of established ECM.
IL33 improves efficacy of antimalarial drug treatment of ECM. Mice were infected with Pb ANKA (n = 12–28 from two to six infections) and treated with antimalarial drugs, either alone (AC) or together with IL33 (AC+IL33), at the onset of ECM. (A) Peripheral parasitemia, (B) survival curves, and (C) RMCBS scores of mice after infection (d0) and drug treatment (gray box). (D–I) Brains were examined at 16–24 h after treatment (d7) for (D) GFP+ parasites (green), costained with lectin (red) and DAPI (blue); (E) erythrocyte-congested vessels indicative of hemostasis (H&E); (F) extravascular IgG indicative of vasogenic odema (DAB counterstained with hematoxylin); (G) hemorrhage (H&E); (H) β-APP accumulation (green) indicative of axonal injury, costained with erythrocytes (red) and DAPI (blue); (I) myelin damage (H&E). Data are presented as means ± SEM (A–C) n = 12–28 from two to six infections. (D–I) n = 6 from two infections. (Scale bars: 25 μm.) (A and C) Separate comparisons were made between groups at d6.5 and d7 by Mann–Whitney U test. (B) Comparison made by log-rank test. (D–I) Comparisons made by Mann–Whitney U or t test as detailed in Methods. *P < 0.05, **P < 0.01, ***P < 0.001.
We next examined the mechanism(s) through which IL33 improved the recovery from ECM. Analyzing IL33’s protein–protein interaction network revealed that a large number of IL33-regulated genes significantly up-regulated in the brains of mice following antimalarial drug treatment were directly or indirectly related to the NLRP3 inflammasome pathway (Fig. 4A and SI Appendix, Fig. S4). It has previously been shown that the malarial parasite product hemozoin (Hz) can directly activate the NLRP3 inflammasome to promote IL1β production (16, 17). Consistent with this, we found that Hz induced release of mature IL1β from bone marrow-derived macrophages (BMDMs) (SI Appendix, Fig. S5A). We also found that artesunate and chloroquine, as well as pyrimethamine, another commonly used antimalarial drug (18), induced IL1β release from BMDMs (SI Appendix, Fig. S5 B–D). The release of IL1β from BMDMs induced by Hz and antimalarial drug stimulation, individually and in combination, was completely inhibited by MCC950, a selective inhibitor of the NLRP3 inflammasome (19) (Fig. 4B and SI Appendix, Fig. S5 A–D). These data, therefore, implied that antimalarial drugs and malaria-parasite products may directly induce damaging inflammasome-induced neuroinflammation, possibly undermining recovery from ECM. In agreement, apoptosis-associated speck-like protein containing a caspase-recruitment domain (CARD) specks, indicative of inflammasome activation, were observed extensively within the brains of infected mice 16–24 h after antimalarial drug treatment of ECM (Fig. 4C). ASC specks were visualized adjacent to, and within, microglial cells, intravascular monocytes, and endothelial cells (Fig. 4C). Critically, IL33 treatment significantly reduced the number of ASC specks in the brain, compared with mice treated only with antimalarial drugs (Fig. 4D).
IL33 suppresses NLRP3 and IL-1β responses that undermine antimalarial drug treatment of ECM. (A) Cytoscape network defining DEGs in the IL33 protein–protein interaction network in brains 16–24 h post drug treatment (d7+AC) compared with d0. (B) BMDMs were treated with antimalarial drugs and hemozoin (AC+Hz) with or without the NLRP3 inhibitor MCC950, with IL1 release measured by ELISA (n = 4) and mature IL1β in the supernatant confirmed by Western blot. (C and D) Pb ANKA-infected ASC-citrine reporter mice were treated at ECM onset with AC alone (AC) or together with IL33 (AC+IL33). (C) Cortical gray matter of AC-treated mice showing ASC specks associated with Iba1+ microglia, intravascular CD68+ monocytes, or lectin+ endothelial cells. (D) ASC specks per field of view (20 fields total from n = 2 for each group). (E–F) Pb ANKA-infected C57BL/6 mice were treated at ECM onset with AC alone (AC), or together with IL33 (AC+IL33), and brains examined by flow cytometry (n = 8 from two infections). (E) Total numbers of microglial cells and intracerebral monocytes. (F) Production of IL1β by microglia and monocytes. (G–I) Pb ANKA-infected C57BL/6 mice (n = 12 from two infections) were treated at ECM onset with AC alone (AC), together with IL33 (AC+IL33) or MCC950 (AC+MCC950). (G) Peripheral parasitemia, (H) survival curves, and (I) RMCBS scores. Data are presented as means ± SEM. (B) Comparisons made by ANOVA. (E and F) Comparisons made by Mann–Whitney U tests. (G and I) Separate comparisons were made between groups at d6.5 and d7 by Kruskal–Wallis test, with Dunn’s correction for multiple comparisons. (H) Comparisons made by log-rank test. (B) *P < 0.05, **P < 0.01, versus AC+Hz. (D–I) *P < 0.05, **P < 0.01, ***P < 0.001, all versus AC. #P < 0.05 MCC950 vs. AC.
As our results indicated that IL33 administration reduced the numbers of monocytes and microglia expressing inflammasomes, we examined whether IL33 treatment modified the polarization or activation of the cells. Both microglia and recruited monocytes/macrophages expressed the IL33 receptor ST2 following drug treatment of established ECM (SI Appendix, Fig. S6 A–D). IL33 administration reduced the numbers of monocytes, but not microglia, at 16–24 h post treatment, and significantly reduced IL1β production in both cell types (Fig. 4 E and F). This effect of IL33 was not mediated through alteration in M1 (based on TNFα and CD40) or M2 (based on CD36, PDL1, and Relmα expression) polarization in either monocytes or microglia (SI Appendix, Fig. S6 E and F). Collectively, these data indicate that IL33 therapy selectively inhibited the NLRP3 inflammasome–IL1β axis in microglia and monocytes during the acute recovery period following treatment of ECM.
CD8+ T cells have been shown to play an important role in the development of ECM (20). Although intracerebral CD8+ T cells also expressed the ST2 receptor following antimalarial drug treatment of ECM (SI Appendix, Fig. S7A), IL33 administration did not significantly alter CD8+ T cell accumulation in the brain (SI Appendix, Fig. S7B). IL33 also had no effect on intracerebral CD8+ T cell effector functions, as defined by intracellular levels of Granzyme B and cell-surface expression of the degranulation marker CD107a (SI Appendix, Fig. S7 C and D).
We then assessed whether administration of a selective NLRP3 inhibitor alongside antimalarial drugs could improve ECM recovery. MCC950 was administered as a single dose (50 mg/kg, HED 4.0541 mg/kg) alongside antimalarial drugs (both at 30 mg/kg) on the first day of treatment (d6). MCC950 did not significantly alter peripheral parasitemia (Fig. 4G). However, comparable to IL33, MCC950 cotreatment along with antimalarial drugs significantly improved survival from established ECM (Fig. 4H). Furthermore, MCC950 administration also significantly improved the RMCBS scores of mice 6–12 h (d6.5) after treatment, compared with mice treated with antimalarial drugs alone (Fig. 4I). Consistent with our findings regarding IL33 monotherapy, MCC950 administration alone (on d6) did not promote improved recovery from ECM (SI Appendix, Fig. S8). Thus, NLRP3 inhibitor treatment also significantly improved the efficacy of antimalarial drug treatment of ECM comparable to the effects of IL33 treatment.
In this study we have shown that antimalarial drugs are unable to prevent mortality in a quarter of mice with established ECM, analogous to the failure rates for CM treatment (2, 3). Furthermore, even when antimalarial drug treatment was successful and animals survived, they were left with significant levels of residual neuropathology. This is consistent with the long-lasting neurological sequelae commonly found in drug-cured CM patients (4, 5). Therefore, our experimental model effectively recapitulates both the primary and secondary clinical challenges associated with the antimalarial drug treatment of CM. Using this model, we assessed the effectiveness of adjunctive therapies in improving existing antimalarial drug therapy. We have discovered that adjunctive IL33 or NLRP3 inhibitor therapy dramatically improved the survival and enhanced the recovery of mice that underwent antimalarial drug treatment.
Our analysis of the brain transcriptome following antimalarial drug treatment provides insights into why antimalarial drugs fail to promote optimal recovery from ECM. Specifically, our data highlight that the neuroinflammatory response associated with agonal ECM is not rapidly down-regulated by antimalarial drugs alone. Importantly, many of the inflammatory pathways that continue to be up-regulated in the brains of mice following antimalarial drug treatment of ECM (e.g., response to IFN gamma, cytotoxic T cell and macrophage activation, and blood coagulation) likely converge to affect the activation of brain endothelial cells (1, 6⇓⇓–9). Concordantly, significant vasculopathy was still evident in mice 24 h after antimalarial drug treatment of established ECM. Thus, our data are consistent with the notion that suboptimal recovery from CM is associated with excessive levels of neuroinflammation and continued disruption to the neurovascular unit (1, 6⇓⇓–9).
Analysis of the upstream regulators controlling the brain transcriptional response during ECM identified 13 genes that could potentially be targeted by additional therapies. We prioritized IL33 because exogenous administration of IL33 has been shown to resolve inflammation and promote repair in other neuropathologies, including Alzheimer’s disease, stroke, and spinal cord injury (12, 13, 15). Moreover, we have previously shown that IL33 administration (without concurrent antimalarial treatment) can attenuate ECM development when given at early stages of infection (21). We hypothesized that the observed reduction in cerebral IL33 during ECM allowed cerebral inflammation to become dysregulated and undermined the success of antimalarial drug treatment. Consistent with this, adjunctive administration of IL33 significantly improved survival and reduced neurological dysfunction in drug-treated mice, compared with antimalarial drugs alone. Importantly, in addition to reducing parasite levels in the brain of surviving mice (examined 16–24 h post treatment), IL33 therapy protected against ECM-induced cerebrovascular damage, as shown by reduced levels of vascular occlusion, edema, and hemorrhage.
Our gene expression analysis from antimalarial drug-treated animals suggests that there is an interaction between the decrease in IL33 gene expression and the increase in expression of genes in the NLRP3 inflammasome pathway (Fig. 4A). While the NLRP3 inflammasome is reportedly not a contributor to the development of ECM (22), its activation could account for the mortality observed after drug treatment of CM and ECM. Indeed, high levels of IL1β have been observed in the brains of individuals with fatal CM (23, 24). Moreover, the NLRP3–IL1β axis is a key driver of acute cerebrovascular dysfunction (25) and progressive neuroinflammation in a number of brain pathologies (26). We observed that administration of IL33 reduced ASC speck formation and IL1β production in the brain compared with mice given antimalarial drugs alone. Furthermore, the selective NLRP3-inflammasome inhibitor MCC950 also significantly improved recovery of mice following antimalarial drug therapy (as with IL33, MCC950 treatment by itself without antimalarial drugs was not protective). Together, our results therefore suggest that IL33 improves antimalarial drug treatment of ECM by altering the brain transcriptome, resulting in suppression of NLRP3-dependent inflammation. This model of protection is in agreement with reports suggesting that administration of IL33 suppresses the expression of NLRP3-inflammasome components in an Alzheimer’s disease model (12) and in a model of intracerebral hemorrhage (27). However, our results are in contrast to a recent report that oligodendrocyte-derived IL33 acts to promote production of IL1β from microglia, subsequently causing cognitive deficits and ECM development (28). Where we examined the NLRP3-supresssing effects of IL33 in vivo, Reverchon et al. (28) defined the IL33-IL1β cycle in an in vitro-mixed glial culture derived from naive mice. IL33 treatment may exert fundamentally different direct and/or indirect activities in vivo within an established inflammatory brain environment than in in vitro-mixed glial cultures in the absence of any other inflammatory or pathogenic signal(s) (29, 30).
NLRP3 inflammasome activation in the acute recovery period following treatment with antimalarial drugs could be caused by the drugs themselves, malaria parasite products, or damage-associated signaling molecules. Consistent with previous studies (16, 17), hemozoin, which we postulate phagocytic cells will be exposed to in significant amounts following antimalarial drug treatment and death of high numbers of parasites, induced NLRP3-dependent release of mature IL1β from BMDMs. A variety of antimalarial drugs (chloroquine, artesunate, and pyrimethamine) also induced predominantly NLRP3-dependent mature IL1β release from BMDMs. Thus, we speculate that antimalarial drug treatment of CM may directly and indirectly provoke inflammasome activation in intracerebral mononuclear phagocytes, impairing the effectiveness of antiparasitic chemotherapy to resolve malaria-induced cerebral pathology. In support of this, we consistently observed accelerated neurological dysfunction and mortality within the subset of mice that succumbed to ECM following antimalarial drug treatment, compared with vehicle-treated controls. Collectively, our data therefore suggest that fatality and neurological sequelae in antimalarial drug treatment of CM may occur, at least partially, as a result of related iatrogenic effects, which can be prevented through IL33 or NLRP3 inhibitor administration.
All animal work was approved following local ethical review by the University of Manchester Animal Procedures and Ethics Committees and was performed in accordance with the UK Home Office Animals (Scientific Procedures) Act 1986 (approved Home Office Project Licenses 70/7293 and P8829D3B4). Female and male C57BL/6 mice (8–10 wk old) were purchased from Charles River. ASC-citrine reporter mice (31) mice were bred at the University of Manchester. All mice were maintained in individually ventilated cages. Cryopreserved Pb ANKA GFP parasites (32) or Pb ANKA parasites (33) were thawed and passaged once through C57BL/6 mice before being used to infect experimental animals. Animals were infected via i.v. injection of 1 × 104 pRBCs. Peripheral parasite burdens of infected mice were followed from day 3 by microscopic examination of Giemsa-stained thin blood smears. The development of, and subsequent recovery from, ECM was assessed using the RMCBS (12). Mice exhibiting early signs of ECM (score ≤15 on the RMCBS, always on d6) received up to six daily i.p. injections of 30 mg/kg artesunate (Sigma) and 30 mg/kg chloroquine (Sigma) in PBS or, alternatively, PBS alone. In some experiments mice received single doses of 0.02 mg/kg (HED 0.0016 mg/kg, calculations based on ref. 34) recombinant IL33 (BioLegend) or 50 mg/kg (HED 4.0451 mg/kg) MCC950 (Sigma) on day 6 via i.p. injection, concomitant with antimalarial drug administration. Detailed information describing protocols for microscopy of brain pathology, RNA purification from whole brain and paired-end RNA-seq analysis, and flow cytometry of intracerebral leukocytes is provided in SI Appendix, Supplementary Methods.
BMDMs, generated as described in SI Appendix, Supplementary Methods, were seeded at 100,000 cells per well in 96-well plates and then left to adhere overnight before priming with 1 µg/mL lipopolysaccharide (0127:B8; Sigma) for 4 h. Following priming, media was replaced with fresh DMEM containing 10% FBS for Hz (Invivogen) or serum-free for antimalarial drug treatments. MCC950 (CP-456773; Sigma) or vehicle control were preincubated for 15 min before inflammasome activation. For Hz assays, cells were treated with Hz or PBS for 24 h. Malaria drugs or appropriate vehicles were incubated for 5 h. In the case of coincubation of Hz and drugs, cells were treated for 24 h. Supernatants were removed and analyzed for IL1β content by ELISA (DuoSet; R&D Systems). IL-1β cleavage within activated BMDMs was performed by Western blot as described in SI Appendix, Supplementary Methods.
The study was supported by the Medical Research Council (MRC) Grants MR/L008564/1 and MR/R010099/1 and by MRC Career Development Award G0900487 (to K.N.C.). Contributions from S.M.A., M.J.H., and D.B. were supported by MRC Grants MC_PC_16033 and MR/N003586/1.
↵1M.J.H., M.G.A., and J.B. contributed equally to this work.
↵2D.B. and K.N.C. contributed equally to this work.
Author contributions: P.S., S.M.C., S.M.A., A.C., F.Y.L, D.B., and K.N.C. designed research; P.S., M.J.H., J.B., T.S., R.D., and E.W. performed research; T.-C.T., D.T.G., F.Y.L., and D.B. contributed new reagents/analytic tools; P.S., M.J.H., M.G.A., J.B., L.Z., S.M.B., E.W., and K.N.C. analyzed data; and P.S., M.J.H., D.B., and K.N.C. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
Data deposition: The sequence reported in this paper has been deposited in the ArrayExpress database (accession no. E-MTAB-6474).
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1801737115/-/DCSupplemental.
Published under the PNAS license.
