IFNAR1 Controls Progression to Cerebral Malaria in Children and CD8⁺ T Cell Brain Pathology in Plasmodium berghei–Infected Mice

Cerebral malaria (CM) is a severe, and often fatal, complicated form of Plasmodium falciparum malaria, affecting mostly children in endemic regions (1–3). Although adhesion and sequestration of infected RBCs (iRBCs) in brain microvasculature leads to blood vessel occlusion and is central to CM pathogenesis (4), immunological mechanisms operating during the development of CM remain unclear.

Accessibility to human tissues limits research into understanding the progression and control of CM development in humans (5). Therefore, use of the mouse model, termed “experimental CM” (ECM), has enabled researchers to uncover cellular and molecular components occurring in CM. Recent articles addressed the relevance of the mouse model of ECM and emphasized the need for identifying consistent factors between the mouse model and human studies (6–9). Because such factors are seldom confirmed, confounding lists of players in CM development have emerged. However, one factor remaining implicit are CD8⁺ T cells (10, 11), although the exact mechanisms that drive their pathogenesis in CM remain unclear. Additionally, the role of cytokines in human CM development was brought to light in recent genetic studies (reviewed in Ref. 12). Initial studies by Hill and coworkers (13, 14) began exploring the necessary role of type I IFN (type I IFN) in CM progression and found that genetic variants in IFNAR1 led to protection from CM development. Because IFNAR1 is the key signaling receptor mediating type I IFN activity (15) and is expressed on virtually every cell surface (16, 17), the impact that type I IFN has on the immune response is vast. In particular, type I IFN was shown to be involved in the priming of CD8⁺ T cells (18, 19) and in the generation of a robust immune response to infection (20, 21). Recent studies addressed the actions of type I IFN, through the IFNAR1 receptor, in response to hemozoin derivatives (22), as well as in controlling parasite burden upon Plasmodium infection (23). However, the exact caliber of IFNAR1 action in CM development remains unexplored.

We performed a genetic study to analyze IFNAR1 involvement in the progression of CM versus uncomplicated malaria (UM). We selected children, the primary risk group of severe malaria, from an area of Angola where malaria transmission is perennial (24). We report in this article that variants of IFNAR1 strongly associate with CM protection, which drove the use of Ifnar1^−/− mice to interrogate the action of IFNAR1 in CM pathogenesis. We describe ECM protection in Ifnar1^−/− mice and, through design of a novel cell-transfer protocol, we expose the necessity of IFNAR1 expression in CD8⁺ T cells during the development of ECM.

A total of 272 children, living in Luanda and ranging from 6 mo to 13 y of age, were enrolled in the current study. Ethical permission for this study was granted by the Ethical Committee of the Hospital Pediátrico David Bernardino in Luanda, appointed by the Angolan Ministry of Health. Written, informed consent was obtained from the parents or guardians of each child. Subjects were selected from patients at Hospital Pediátrico David Bernardino. The details of the data set were described previously (25). Briefly, samples were collected from children aged 6–156 mo, between February 2005 and May 2007, and included 130 patients with CM and 142 patients with UM. Malaria was diagnosed on the basis of a positive asexual parasitemia detected on a Giemsa-stained thick smear (26). CM was defined according to the World Health Organization criteria: a coma score < 3 on the Blantyre Scale for children <60 mo or a coma score < 7 on the Glasgow Scale for children ≥60 mo. Meningitis and encephalitis were ruled out by cerebrospinal fluid analysis after lumbar puncture. Exclusion criteria were different known etiologies of encephalopathy and hypoglycemia (glycemia < 40 mg/dl). Patients with consciousness disturbances or with other diseases were also excluded from this group. The UM group represents patients with malaria diagnosis and febrile illness without any clinical finding suggestive of other causes of infection and with no manifestations of severe malaria.

Genomic DNA was extracted from whole blood using the Chemagen Magnetic Bead technology. DNA preparations were quantified using PicoGreen reagents, according to the supplier’s instructions. Twenty-seven single nucleotide polymorphisms (SNPs) covering the IFNAR1 region were initially genotyped using the Sequenom’s iPlex assay (San Diego, CA) and the Sequenom MassArray K2 platform by the Genomics Unit of the Instituto Gulbenkian de Ciência.

Extensive quality control was performed using eight HapMap (http://hapmap.ncbi.nlm.nih.gov/) controls of diverse ethnicity, Hardy–Weinberg equilibrium with p > 0.01, and a minimum 90% call rate for each SNP. Genotype determinations were performed blinded to affection status. Only SNPs with minor allele frequency > 1% in both groups were analyzed for association. Nine SNPs in IFNAR1 (rs224359, rs2252930, rs12626750, rs2834197, rs2226300, rs2254315, rs1012943, 2843996, and 2254305) did not meet the quality control criteria and were excluded. Samples with <75% call rate and duplicates were excluded from analysis; a total of 33 samples was excluded. The final data set used in the analysis consisted of 239 children: 110 CM patients (mean age, 54.6 mo) and 129 UM patients (mean age, 49.1 mo).

The χ2 tests for Hardy–Weinberg equilibrium in both groups, allelic and haplotypic association of SNPs with progression to CM, and linkage disequilibrium plots were performed using Haploview 4.2 (27). A diagram with representation of the allelic genetic-association results was produced by the snp.plotter R package (28). Association analyses were performed with logistic regression and adjusted for potential confounding effects of age using the SNPassoc v.1.4-9 package (29) implemented in R freeware (http://cran.r-project.org/). Results were considered suggestive below the conventional probability level of 0.05. Bonferroni corrections for multiple tests were carried out to exclude type I errors (the significance level for 18 tests is set at p value < 2.78 E−3).

Male, 8–12-wk-old C57BL/6, B6.Ly5.1, Ifnar1^−/−, and BALB/c mice were bred and used as provided by the Instituto Gulbenkian de Ciência in-house animal facility. Ifnar1^−/− mice (Ifnar1^-tm-Agt) (30) were originally a gift to the institute from Michel Aquet, Institute of Molecular Biology, University of Zurich, Switzerland. Ifnar1^−/− mice were backcrossed onto the C57BL/6 background (98.7–99.36% C57BL/6 background outside of the congenic region, as confirmed by the animal facility at Instituto Gulbenkian de Ciência). All procedures were in accordance with national regulations on animal experimentation and welfare, as approved by the Instituto Gulbenkian de Ciência Ethics Committee.

All infections used 1 × 10⁶ Plasmodium berghei ANKA-GFP (PbA)-parasitized iRBCs (31) given via i.p injection. Frozen iRBC stocks were expanded in C57BL/6 mice prior to infection. Parasitemia was determined as the percentage of GFP⁺ RBCs, using flow cytometry analysis (FACScan Cell Analyzer; Becton Dickinson) and noninfected RBCs as negative control. ECM development was monitored from day 5 postinfection (PI), including head deviations, paralysis, ataxia, and convulsions (32). ECM-susceptible mice developed symptoms at days 6–7 PI and died within 4–5 h. Mice resistant to ECM died around days 22–25 from anemia and hyperparasitemia without displaying neurologic symptoms.

Upon display of ECM-induced coma in C57BL/6 mice, the experimental groups of mice were injected with 100 μl 2% Evan’s blue dye (Sigma), sacrificed 1 h postinjection, and perfused intracardially with 15 ml ice-cold PBS. Brains were dissected, weighed, and immersed in 2 ml formamide (Merck), covered at 37°C for 48 h, to allow extraction of Evan’s blue dye; images were taken after the 48 h period. Noninfected mice from each group were used as controls. Absorbance of dye was measured at 620 and 740 nm using spectrophotometry (Thermo Spectronic Helios Delta). Using a standard curve, data were calculated and expressed as μg Evans blue dye/mg brain tissue.

Infected C57BL/6 and Ifnar1^−/− mice were sacrificed upon observation of clinical signs of ECM in C57BL/6 mice, ∼day 6 PI. Noninfected C57BL/6 and Ifnar1^−/− mice were used as controls and sacrificed at the same time point. For H&E staining, brains were carefully removed, fixed in 10% neutral-buffered formalin, embedded, sectioned (4 μm), mounted with Entellan (Merck, Darmstadt, Germany), and stained with H&E, following standard procedures. CM and brain pathology were assessed by routine histopathology in coronal brain sections of the same anatomical location. Bright-field images were captured using a Leica DMD108 microscope (Leica Microsystems); 10× (numerical aperture: 0.40) and 40× (numerical aperture: 0.95) objectives were used. Adobe Photoshop software (Adobe) was used to compose images and adjust the contrast.

Infected C57BL/6 and Ifnar1^−/− mice were sacrificed at days 5–7 PI. Ifnar1^−/− mice were also sacrificed at days 9 and 11 PI; C57BL/6 mice were not collected at these time points due to death from ECM. Noninfected C57BL/6 and Ifnar1^−/− mice were used as controls and represent day 0. Brains were dissected, fast frozen in liquid nitrogen, and stored at −80°C. Brains were then homogenized, and a final brain weight of 20 mg was used for total RNA extraction (RNeasy Mini Kit; QIAGEN). RNA quality was determined using Nanodrop, and a final elution of 50 μl was extracted. One microgram of total RNA was converted to cDNA (Transcriptor High Fidelity cDNA Synthesis Kit; Roche) using random hexamer primers, and a final volume of 10 μl the cDNA reaction was diluted 1:3 in RNase-free water to be used for mRNA quantification. Tnf-α and Il-10 mRNA was quantified using TaqMan Gene Expression Assays from ABI (Mm00443258_m1 and Mm99999062_m1, respectively). P. berghei ANKA rRNA was quantified using specific primer sequences: forward 5′-CCG ATA ACG AAC GAG ATC TTA ACC T-3′, reverse 5′-CGT CAA AAC CAA TCT CCC AAT AAA GG-3′ and probe 5′-ACT CGC CGC TAA TTA G-3′ (FAM/MGB), all with TaqMan Universal PCR Master Mix. T lymphocyte (Cd3e) expression was quantified using a specific primer sequence, forward 5′-TCT CGG AAG TCG AGG ACA GT-3′ and reverse 5′-ATC AGC AAG CCC AGA GTG AT-3′ (33), using Power SYBR Green PCR Master Mix (Applied Biosystems). Gene expression–quantification reactions were performed on an ABI Prism 7900HT system, according to the manufacturer’s instructions; a total volume of 10 μl PCR reactions was amplified for 45 cycles. Relative quantification of specific mRNA was normalized to the mouse housekeeping gene ACTB (Mouse ACTB Endogenous Controls; ABI), used in multiplex PCR with the target genes and calculated using the ΔΔCt method.

For infected blood donor mice, blood was collected by mandibular vein puncture into an Eppendorf tube containing 10 μl heparin (Heparina Leo; LEO Pharmaceutical Products) from infected mouse at day 6 PI, upon display of ECM symptoms and when parasitemia was ∼30%. Parasitemia was measured to calculate injection of 4 × 10⁶ iRBCs into cell donor mice. Collected blood was then placed on ice in a 50-ml Falcon tube and irradiated at 20,000 rad (Gammacell 2000; Mølsgaard Medical). Irradiated iRBCs were then resuspended in PBS and injected i.p. into cell donor mice. At day 6 postirradiated iRBC injection, cell donor mice were sacrificed, and purified spleen cell populations were injected into recipient mice; recipient mice were left noninfected or were infected with 1 × 10⁶ PbA i.p. 1 h after transfer. Parasitemia and ECM development were monitored from day 5 PI. For control noninfected blood donor mice, mice were bled, and an equivalent volume of noninfected RBCs was irradiated and diluted in PBS in the same manner as above.

Single-cell suspensions were prepared from spleens of cell donor mice. For total spleen cell transfers, cells were washed twice in PBS, counted in a hemocytometer, resuspended in PBS, and injected at 5 × 10⁶/100 μl into recipient mice. For spleen CD8⁺ cell transfers, cells were washed twice in PBS, 2% FCS and incubated with anti-mouse CD8–Alexa Fluor 647 Ab (clone YTS169; in-house production) for 30 min on ice. For CD8⁺Ly5.1/2⁺ cell transfers, cells were incubated with anti-mouse CD8–Alexa Fluor 647 Ab (clone YTS169; in-house production), anti-mouse CD45.1-FITC Ab (clone 820; BD Pharmingen), or anti-mouse CD45.2-PE Ab (clone 104.2; in-house production) for 30 min on ice. Stained cells were washed in PBS, 2% FCS and sorted (FACSAria Multicolor Cell Sorter; Becton Dickinson) by combining lymphocyte morphological gating and relevant Ab labels. For CD8⁺ transfers, cells were collected as purified CD8⁺ cells and CD8⁻ (depleted) groups of cells. For CD8⁺Ly5.1/2 transfers, cells were gated for CD8⁺ cells, within which CD45.1⁺ or CD45.2⁺ populations were collected. Purified cell populations were washed in PBS and counted. Cells were injected i.v. at 3 × 10⁶/100 μl into recipient mice. Respective groups of recipient mice were then infected with 1 × 10⁶ iRBCs/100 μl PbA 1 h after cell transfer. Mice were monitored for parasitemia and ECM development from day 5 PI.

Single-cell suspensions were prepared from spleens of infected C57BL/6 and Ifnar1^−/− mice at day 5 PI; noninfected mice from both groups were used as control. Cells used for total cell counts and activation profile were prepared as single-cell suspensions: from 5 ml total cells, 200 μl/sample was plated in a 96-well round-bottom plate, and cells were centrifuged; pelleted; resuspended in Fc Block for 20 min on ice; washed in PBS, 2% FCS; and resuspended in either anti-mouse CD8–Alexa Fluor 647 Ab (clone YTS169; in-house Ab) or CD4–Alexa Fluor 647 (clone GK 1.5; in-house Ab) with CD44-FITC (clone IM7; BD Pharmingen 553133) and CD62L-PE (clone Mel 14; eBioscience 12-0621-85) for 20 min on ice. Cells were then washed, resuspended in 100 μl PBS, 2% FCS plus 20 μl propidium iodide (1 μg/ml; P-4170; Sigma) and 10 μl 5 × 10⁵ beads (Beckman Coulter nominal 10 μm Latex Beads). For intracellular staining, cells were washed and resuspended in RBC lysis buffer; washed and resuspended in 100 μl RPMI 1640 medium supplemented with 10% FCS, 10 U/ml Pen-Strep, 1% sodium pyruvate, 1% HEPES 1 M, 1% l-glutamine, and 0.01% 2-ME; centrifuged; and resuspended in 200 μl RPMI 1640 supplemented medium containing 10 μg/ml brefeldin A for fixation of intracellular cytokine production and 100 ng/ml of PMA and 500 ng/ml ionomycin used for cell stimulation. Cells were stimulated at 37°C, 5% CO₂ for 4 h. Cells were then processed for extracellular staining; cells were centrifuged, pelleted, resuspended in Fc Block for 20 min on ice, washed in PBS, 2% FCS, and resuspended in anti-mouse CD8–Alexa Fluor 647 Ab (clone YTS169; in-house Ab) for 20 min on ice. Cells were then washed and resuspended in 2% paraformaldehyde for fixation for 30 min at room temperature (RT), washed twice in PBS, resuspended in 200 μl PBS, 2% FCS, and stored at 4°C overnight. The following day, cells were pelleted and resuspended in permeabilization buffer (10% saponin in PBS, 2% FCS) for 10 min at RT in preparation for intracellular staining. Cells were then washed, pelleted, and incubated with anti-mouse granzyme B (GrB)–FITC (clone NGZB; eBioscience) and anti-mouse IFN-γ–PE (clone XMG1.2; BD Pharmingen) for 30 min at RT. Cells were then washed twice in permeabilization buffer and then twice in PBS, 2% FCS and resuspended in 150 μl PBS, 2% FCS plus 5 × 10⁵ beads (as before). All cell analyses were performed using a FACSCalibur cell analyzer and FlowJo software (version 7). Single-staining preparations were used for control compensation of double-staining analysis.

Infected C57BL/6 and Ifnar1^−/− mice were sacrificed at day 6 PI and perfused pericardially (as described above), and brains were placed in 10 ml HBSS supplemented with Collagenase VIII (0.2 mg/ml; Sigma). Brains were then smashed and incubated for 30 min at 37°C. Each sample was passed through a 100-μm strainer and centrifuged for 10 min at 1500 rpm. Supernatant was discarded, and samples were resuspended in 10 ml PBS and centrifuged. Samples were resuspended in 10 ml 30% Percoll gradient (Percoll; GE Healthcare, BioSciences) and centrifuged for 20 min at 2500 rpm at RT, without brake. The supernatant was aspirated carefully, and the pellet was resuspended in 50 ml PBS and centrifuged. The supernatant was discarded, and each sample pellet was resuspended to a final volume of 200 μl. Cells were centrifuged and resuspended in 100 μl RBC lysis buffer, incubated for 5 min at RT, washed twice in 200 μl PBS, 2% FCS, resuspended in supplemented RPMI 1640 medium, and prepared for extra- and intracellular staining and analysis, as described above.

The nonparametric Mann–Whitney U test was used for analysis when comparing data from two groups of mice. One-way ANOVA was used to compare three or more groups of data. Survival analysis was assessed by the log-rank test. All tests were performed using Prism GraphPad 4. The p values ≤0.05 were considered statistically significant.

The IFNAR1 gene has been associated with severe forms of malaria (13, 14). In this study, we used a cohort of Angolan children to identify IFNAR1 genetic variants that specifically modify the risk for progressing from uncomplicated forms of malaria to CM syndrome. We compared genotype frequencies of SNPs in infected children that developed CM (n = 110) versus children that presented with clinically UM (n = 129). After age correction, 8 of 18 IFNAR1 SNPs showed a suggestive association (p < 0.05) with progression to CM (Fig. 1B), possibly under a dominant mode of action (Supplemental Table II). The two markers having the strongest association with progression to CM, rs2856968 (p = 1.33E−6) and rs2253923 (p = 5.77E−5), remained significantly associated after conservative Bonferroni correction for multiple testing (p < 2.78E−3) (Fig. 1B). Because the two SNPs are in strong linkage disequilibrium (r² = +0.77), this result suggests a single genetic effect in CM protection (Fig. 1D).

Haplotypic-association analysis of eight SNPs in the 5′ region of IFNAR1 (encompassing upstream regulatory region intron 1 and intron 2) identified one haplotype (hap) that shows robust association with CM protection (P_hap = 6.89E−7; OR_hap = 0.34). Conversely, the most frequent haplotype for the same eight SNPs shows an increased risk for CM progression (P_hap = 8.0E−4; OR_hap = 1.87). Together, this genetic evidence strongly suggests that specific minor frequency alleles of the IFNAR1 gene provide protection against CM. This suggests a decisive role for IFNAR1 during the inflammatory response to malaria infection and, specifically, at dictating progression to CM.

We used the mouse model of ECM to ascertain whether IFNAR1 is involved in inflammation evoked by malaria infection (6). In line with a recent report (22), we observed that PbA infection in Ifnar1^−/− mice led to strong resistance to ECM (75% ECM resistance, Fig. 2A) compared with susceptible C57BL/6 control mice. Ifnar1^−/− mice that resist ECM development die between days 25 and 30 from anemia and hyperparasitemia and showed parasitemia comparable to C57BL/6 mice up to day 6 PI (Fig. 2B). This indicated that ECM resistance in Ifnar1^−/− mice was not related to decreased peripheral parasite burden.

Disruption of the blood–brain barrier (BBB) is a hallmark of ECM pathology (34). We analyzed infected and noninfected C57BL/6, Ifnar1^−/−, and ECM-resistant BALB/c mice for barrier compromise using Evans blue. Like BALB/c mice, Ifnar1^−/− mice displayed no BBB compromise (Fig. 3A, 3B), as opposed to C57BL/6 mice, which showed abundant dye leakage in the brain. Histopathological analysis of infected C57BL/6 brain sections exhibited prominent mononuclear cell accumulations within brain microvessels (Fig. 3D), evidence of disruption of vessel walls and endothelial cell destruction (Fig. 3E), and parenchymal hemorrhagic lesions containing iRBCs (Fig. 3F). In contrast, ECM-resistant Ifnar1^−/− mice displayed mild intravascular accumulation of mononuclear cells (Fig. 3H), intact endothelial cells (Fig. 3I), and absence of hemorrhagic lesions with iRBCs seen within the lumen of blood vessels only (Fig. 3J). These results show that, despite abnormal mononuclear accumulation within brain microvessels and the presence of intraluminal iRBCs, infected Ifnar1^−/− mice do not exhibit pathology typical of ECM.

To further investigate the mechanisms underlying ECM protection in Ifnar1^−/− mice, we evaluated cytokine production and parasite burden within the brain tissue of infected C57BL/6 and Ifnar1^−/− mice. Comparison of P. berghei rRNA quantification showed no significant differences during the early time course of infection, strongly suggesting that ECM protection in Ifnar1^−/− mice was not attributable to reduced brain parasite burden (Fig. 4A). We also observed that kinetics of Cd3 mRNA expression in the brain leading up to ECM manifestations were slightly lower in Ifnar1^−/− mice but not significantly different from C57BL/6 mice; this suggests that T cell migration to the brain during infection was not impaired in Ifnar1^−/− mice (Fig. 4B). Nevertheless, we found that, in the brain of Ifnar1^−/− mice, the expression of inflammatory markers (i.e., Tnf and Il10) had lower induction in the period preceding ECM onset and showed delayed kinetics compared with infected C57BL/6 mice. Overall, these results indicate that, despite the accumulation of iRBCs and the presence of T cells, Ifnar1^−/− mice show a diminished exacerbation in their inflammatory responses, deterring the development of brain pathology within the time frame of ECM development.

We questioned in which cell type IFNAR1 was fundamental to cause ECM development in mice. To this end, we developed a novel cell-transfer protocol to determine the ability of individual cell types to restore ECM susceptibility in Ifnar1^−/− mice (Fig. 5A). Donor mice were exposed to irradiated iRBCs, allowing relevant blood-stage parasite Ags to prime in vivo immune system cells that were then used in transfer experiments. The irradiated parasite is unable to reinvade RBCs; therefore, (exposed) donor mice do not develop parasitemia or ECM. We analyzed the T cell profiles of C57BL/6 mice that received iRBCs, as well as mice that received irradiated iRBCs (Fig. 5B, 5C). We found that the numbers of CD8⁺ T cells in mice that received irradiated iRBCs were not as high as those found in infected C57BL/6 mice but were similar in number to activated CD8⁺ T cells (CD44^hi and CD62L^low) (Fig. 5B). We noted that CD4⁺ T cell numbers and activation profile were not significantly different between mice that received irradiated iRBCs or iRBCs (Fig. 5C). Together, this suggests that irradiated iRBCs induce activation of both CD8⁺ and CD4⁺ T cells. A group of C57BL/6 cell donor mice was injected with irradiated iRBCs and then infected with 1 × 10⁶ nonirradiated PbA iRBCs 7 d later (as per normal infection); all mice developed ECM and died by day 7 PI (data not shown). We concluded that our protocol does not confer protective immunity but rather is a protocol of immune cell presensitization that occurs in the absence of parasite expansion or disease development. We anticipate that this method will be useful in investigating pathogenic mechanisms of malaria blood-stage infection in future mouse models.

Total splenocytes were transferred from C57BL/6 mice that had received irradiated iRBCs (cell donors) to Ifnar1^−/− mice. Upon infection, recipient Ifnar1^−/− mice showed 75% development of ECM, 6–12 d PI, within the ECM time frame of susceptible C57BL/6 mice (Fig. 5D). This demonstrates that C57BL/6 splenocytes presensitized to irradiated iRBCs are effective at restoring ECM susceptibility in Ifnar1^−/− mice. In a control experiment, total splenocytes were transferred from C57BL/6 donor mice that had received noninfected irradiated RBCs to Ifnar1^−/− mice. When infected 1 h posttransfer, recipient mice developed parasitemia like Ifnar1^−/− mice that received infection only (no cell transfer). Neither group of mice developed ECM, with all mice dying around day 25 PI from anemia and hyperparasitemia (Supplemental Fig. 1). This shows that C57BL/6 splenocytes exposed to noninfected irradiated RBCs are unable to induce ECM susceptibility in Ifnar1^−/− mice, indicating that RBC irradiation per se has no impact on ECM development in recipient mice. We then determined whether CD8⁺ spleen cells induced ECM in Ifnar1^−/− mice. Sorted CD8⁺ or CD8⁻ depleted cells, isolated from donor C57BL/6 mice that received irradiated iRBCs, were transferred into Ifnar1^−/− mice (Fig. 5E). Recipient mice were then infected 1 h after cell transfer. Parasitemia did not differ between the two groups of mice compared with Ifnar1^−/− mice that received infection only (and no cell transfer). In contrast, we found ECM development only in the group of Ifnar1^−/− mice that received purified CD8⁺ cells. This result indicates that transfer of presensitized C57BL/6 CD8⁺ cells is sufficient to restore susceptibility to ECM in Ifnar1^−/− mice. Upon transfer of presensitized Ifnar1^−/− CD8⁺ T cells to Ifnar1^−/− mice, we found that recipient mice did not develop ECM; this led us to conclude that only CD8⁺ IFNAR1⁺ cells from exposed C57BL/6 mice were able to cause ECM (Fig. 5F). By assessing BBB integrity, we confirmed the ability of presensitized CD8⁺ cells isolated from C57BL/6 mice to induce bone fide ECM. We observed BBB compromise only in Ifnar1^−/− mice that received C57BL/6 CD8⁺ cells and not in CD8^- cell–depleted recipient mice (Fig. 5G, 5H). Together, these results show that IFNAR1 is critical for CD8⁺ cell–mediated abrogation of ECM resistance in Ifnar1^−/− mice.

We modified our cell-transfer protocol to ascertain whether ECM induction was dependent on IFNAR1 expression in CD8⁺ cells (Fig. 6A). CD8⁺ cells were sorted from noninfected Ifnar1^−/−.Ly5.2 mice and transferred to previously nonlethally irradiated C57BL/6.Ly5.1 mice, allowing for the discrimination of endogenous and transferred CD8⁺ populations. Ly5.1/5.2 chimeric mice then received irradiated iRBCs from a B6.Ly5.1 mouse. Seven days later, Ly5.1 CD8⁺ and Ly5.2 CD8⁺ cell populations were sorted (Fig. 6B) and transferred to Ifnar1^−/− recipient mice. No differences in parasitemia progression were observed in either group compared with Ifnar1^−/− mice that received no cell transfer and infection only (Fig. 6C). We found that only recipient Ifnar1^−/− mice that had received Ly5.1 IFNAR1⁺ CD8⁺ cells developed ECM, showing 100% disease incidence, whereas recipients of Ly5.2 IFNAR1⁻ CD8⁺ cells maintained ECM resistance comparable to Ifnar1^−/− mice that received no cell transfer and infection only. This strongly suggests that deficiency in IFNAR1 signaling in CD8⁺ cells impairs their ability to induce ECM in Ifnar1^−/− mice during PbA infection.

To investigate CD8⁺ T cell impairments in Ifnar1^−/− mice during PbA infection, we analyzed the activation and expansion of CD8⁺ T cells in the spleen at day 5 PI, prior to ECM onset. We found a significant difference between noninfected C57BL/6 and Ifnar1^−/− mice that was attenuated upon infection. This indicates that infection-driven CD8⁺ T cell expansion was not impaired in Ifnar1^−/− mice (Fig. 7A). Similarly, splenic CD8⁺ T cells from Ifnar1^−/− mice do not show impairments in GrB or IFN-γ expression upon infection (Fig. 7B, 7C). We also note that CD4⁺ T cell expansion and activation, as measured by IFN-γ, are not hindered in Ifnar1^−/− mice upon infection compared with C57BL/6 mice (Supplemental Fig. 2).

Likewise, we analyzed profiles of CD8⁺ T cells from the brain of infected C57BL/6 and Ifnar1^−/− mice at day 6 PI, upon display of end-stage ECM symptoms in control C57BL/6 mice. We found that the total number of CD8⁺ T cells sequestered in the brain of Ifnar1^−/− mice was significantly reduced compared with C57BL/6 mice (Fig. 7D). Nevertheless, the vast majority of brain-sequestered CD8⁺ T cells in both Ifnar1^−/− and C57BL/6 mice were activated and showed a cytotoxic profile (Fig. 7E, 7F). This indicates no faltering in sequestered Ifnar1^−/− CD8⁺ T cells to exhibit a cytotoxic profile, as supported by mean fluorescence intensity analysis of GrB and IFN-γ in CD8⁺ T cells upon infection compared with C57BL/6 mice (Supplemental Fig. 3).

We confirmed these results by analyzing sequestered CD8⁺ T cells on day 6 PI from Ifnar1^−/− mice that received either CD8⁺ T cells (and were restored with ECM susceptibility) or CD8⁻ depleted cells. As expected, higher numbers of CD8⁺ T cells, as well as GrB⁺ and IFN-γ⁺ CD8⁺ T cells, were seen in the brain of mice that received CD8⁺ T cell transfer (Fig. 7G–I). However, these results were not significantly different, suggesting that induction of ECM in Ifnar1^−/− mice is not strictly dependent on accumulation of effector IFNAR1⁺ CD8⁺ T cells. Together, these results suggest that ECM resistance in Ifnar1^−/− mice operates through impaired IFNAR1 signaling in CD8⁺ T cells.

This work strengthens the role of type I IFNs/IFNAR1 signaling in human cerebral malaria and uncovers the actions of IFNAR1 in ECM development. We found that IFNAR1 genetic variants showed protection against progression to CM in a population of Angolan children. In Gambian children, two SNPs, which were identified in the IFNAR1 gene in intron 3 and exon 4, were associated with severe malaria and CM (13). However, the study of Gambian, Kenyan, and Vietnamese populations (14) highlighted positive associations in the 5′ region of the IFNAR1 locus. This was also found as a region of strong association in our study, suggesting that the 5′ region of the IFNAR1 gene known to control transcriptional activity (35) represents a malaria-susceptibility factor throughout diverse populations. Furthermore, the design established in our study proved to be advantageous, allowing us to evaluate specifically the association of IFNAR1 with disease progression from UM to CM. IFNAR1 has also been associated with AIDS progression (15) and clinical presentation of hepatitis B virus infection (36), which allows the inference that IFNAR1 variants can affect type I IFN responses and govern the course and progression of infectious diseases.

These results motivated the use of Ifnar1^−/− mice to address the mechanisms underlying the pathogenesis of CM in the ECM mouse model. We found that protection from ECM in Ifnar1^−/− mice was not controlled via a reduction in parasite burden. Conflicting results have been found; one study (22) showed no decrease in parasitemia when using PbA infection in Ifnar1^−/− mice, whereas another study (23) reported a decrease in parasite burden upon PbA-GFP + luciferase infection. It remains to be determined whether different strains of PbA parasite dictate parasitemia burden in Ifnar1^−/− mice infection. In our experimental system (PbA-GFP), type I IFN appears to be dominant during other stages of ECM pathogenesis. Despite rapid and steep development of parasitemia throughout infection and parasite accumulation within the brain tissue at levels comparable to susceptible C57BL/6 mice, this was not enough to provoke a sufficient inflammatory response for BBB compromise and mononuclear cell accumulation within brain microvessels. It is worth mentioning that recombinant human IFN-α evoked a striking reduction in parasite burden and protection from ECM in C57BL/6 mice, suggesting that the type I IFN/IFNAR1 system has a role in parasite development (37).

Our data also point to type I IFN/IFNAR1 controlling disease tolerance to malaria. Tolerance is a defense mechanism that does not limit pathogen load (38), and this is in agreement with recent studies that illustrate Ifnar1^−/− mice providing a survival advantage against Plasmodium infection (39). As a side note, the type I IFN/IFNAR1–signaling system can also be regulated by heme oxygenase-1, a stress-response enzyme associated with CM pathogenesis in children, as we reported previously (25).

The development of a novel transfer protocol allowed us to focus on cellular mechanisms involved in ECM protection conferred by IFNAR1. These experiments uncovered that IFNAR1 is needed unequivocally in primed splenic CD8⁺ T cells to abrogate ECM resistance in Ifnar1^−/− mice. The cytotoxic profile of CD8⁺ T cells in ECM induction has been shown using cell transfers; perforin (40) and GrB (41) were identified as essential elements to produce cytotoxicity and effector CD8⁺ T cell responses. Because we saw no hindrance in the cytotoxic profile of splenic Ifnar1^−/− CD8⁺ T cells upon infection (Fig. 7A–C), this suggests that priming in the spleen is not impaired. However, because we observed lower retention of activated CD8⁺ T cells in the brain of Ifnar1^−/− ECM-resistant mice, type I IFN could influence ECM via its action in the brain. Type I IFN was shown to be critical in maintaining CD8⁺ T cells at the site of their effector function (42); therefore, this hints that unresponsive Ifnar1^−/− CD8⁺ T cells may have an impaired ability to accumulate within the brain tissue. We question whether enabling the accumulation of IFNAR1⁻ CD8⁺ T cells in the brain could generate enough power to trigger an ECM response. We noted that Ifnar1^−/− mice that received primed CD8⁺ splenocytes accumulated relatively low numbers of activated CD8⁺ T cells within the brain tissue, yet they developed ECM (Fig. 7G–I). This supports the hypothesis that induction of ECM in transferred mice was not due to CD8⁺ T cell accumulation but rather was caused by their expression of IFNAR1, allowing CD8⁺ T cells to become reactivated by type I IFN within the brain and unleashing their cytotoxic effector capacity.

Evidence also showed that CD8⁺ T cells require constant Ag presentation to remain at their effector site (43–45). Type I IFN was shown to directly stimulate CD8⁺ T cell responses during Ag cross-presentation (46). This suggests that resistance of Ifnar1^−/− CD8⁺ T cells to inducing ECM could be due to their inability to respond to type I IFN signals during Ag presentation within the brain tissue. Moreover, the exact way in which CD8⁺ T cells are presented in the brain is still unknown. We hypothesize that IFNAR1 affects CD8⁺ T cell stimulation at different stages of Plasmodium infection. We support the spleen as a primary site of parasite Ag presentation and propose that a secondary site of stimulation and presentation to CD8⁺ T cells occurs within the brain tissue. In fact, transfer of Ifnar1^−/− CD8⁻ splenocytes caused a slight increase in susceptibility to ECM, possibly reflecting the effect of other immune cells, predominantly activated CD4⁺ T cells, in promoting ECM development (47) by an IFNAR1-independent mechanism.

Ifnar1^−/− mice also have proved fundamental in uncovering the therapeutic use of type I IFN in cancer (48), as well as in cases of experimental autoimmune encephalomyelitis (17). However, in support of our results, Ref. 45 reviews how type I IFN therapy has led to the induction of autoimmune diseases. Our results induce us to think that blocking IFNAR1 would be therapeutic for CM protection; however, this would have to done with extreme precision because of the vast array of cells expressing IFNAR1, and, more importantly, this would not directly mimic an IFNAR1 gene polymorphism.

Our work collectively highlights the cohesive role of IFNAR1 during CM development, which is that it confers human CM protection and it is implicated in CD8⁺ T cell pathology in mouse ECM. These observations allow us to speculate that IFNAR1 may operate in human CM through the stimulation of CD8⁺ T cells. Together, these results identify responses of type I IFN through IFNAR1 as vital components in CM pathogenesis.

We thank Miguel Soares (Instituto Gulbenkian de Ciência) for critical revision of the manuscript and use of reagents and Ivo Marguti for technical support and discussion. We thank Dr. Tânia Carvalho, resident pathologist at Instituto Gulbenkian de Ciência, for analysis of pathology samples and the Cell Imaging Unit for expertise in cell sorting. We are most grateful to the children who participated in the study and thank the staff of the Hospital Pediátrico David Bernardino. We are grateful to Drs. Filomena Silva, Emingarda Castelo Branco, and Rosa Silva from the Instituto de Saúde Pública de Angola in Luanda for help with sample processing and Dr. Rui Pinto (Clínica Sagrada Esperança, Luanda, Angola) for sample collection materials.

The authors have no financial conflicts of interest.