Cannabinoid Receptor 2 Modulates Susceptibility to Experimental Cerebral Malaria through a CCL17-dependent Mechanism *

Cerebral malaria is a severe and often fatal complication of Plasmodium falciparum infection. It is characterized by parasite sequestration, a breakdown of the blood-brain barrier, and a strong inflammation in the brain. We investigated the role of the cannabinoid receptor 2 (CB2), an important modulator of neuroinflammatory responses, in experimental cerebral malaria (ECM). Strikingly, mice with a deletion of the CB2-encoding gene (Cnr2−/−) inoculated with Plasmodium berghei ANKA erythrocytes exhibited enhanced survival and a diminished blood-brain barrier disruption. Therapeutic application of a specific CB2 antagonist also conferred increased ECM resistance in wild type mice. Hematopoietic derived immune cells were responsible for the enhanced protection in bone marrow (BM) chimeric Cnr2−/− mice. Mixed BM chimeras further revealed that CB2-expressing cells contributed to ECM development. A heterogeneous CD11b+ cell population, containing macrophages and neutrophils, expanded in the Cnr2−/− spleen after infection and expressed macrophage mannose receptors, arginase-1 activity, and IL-10. Also in the Cnr2−/− brain, CD11b+ cells that expressed selected anti-inflammatory markers accumulated, and expression of inflammatory mediators IFN-γ and TNF-α was reduced. Finally, the M2 macrophage chemokine CCL17 was identified as an essential factor for enhanced survival in the absence of CB2, because CCL17 × Cnr2 double-deficient mice were fully susceptible to ECM. Thus, targeting CB2 may be promising for the development of alternative treatment regimes of ECM.

CD11b ؉ cells that expressed selected anti-inflammatory markers accumulated, and expression of inflammatory mediators IFN-␥ and TNF-␣ was reduced. Finally, the M2 macrophage chemokine CCL17 was identified as an essential factor for enhanced survival in the absence of CB2, because CCL17 ؋ Cnr2 double-deficient mice were fully susceptible to ECM. Thus, targeting CB2 may be promising for the development of alternative treatment regimes of ECM.
Cerebral malaria (CM) 6 is a life-threatening complication of Plasmodium falciparum infection and one of the leading causes of mortality worldwide (1)(2)(3). CM pathology is associated with a sequestration of parasitized red blood cells within the CNS microvasculature, which triggers an excessive inflammatory CNS response and an increase in blood-brain barrier permeability (1,2). CM can be modeled in mice through inoculation with Plasmodium berghei ANKA (PbA)-infected red blood cells (1)(2)(3)(4)(5)(6)(7)(8). Those mice develop severe neurological symptoms of ECM within 5-8 days after infection that lead to coma and death by days 7-9 (4). Disease pathogenesis is associated with a widespread activation of microglial cells, the production of inflammatory cytokines, and endothelial cell damage (1,6). It was suggested that macrophages support pathogenic T cell responses in the CNS leading to the death of mice with experimental CM (ECM). Conversely, macrophages are also capable of mediating anti-parasite responses by clearance of parasites or production of IL-10, a cytokine shown to limit disease progression (9 -11).
The endocannabinoid system plays a key role in immune modulation of the CNS by signaling via CB2 (12)(13)(14)(15)(16)(17)(18)(19). It may therefore hold a therapeutic potential for the treatment of infectious CNS disorders. CB2 signaling affects various macrophage functions, such as antigen uptake, antigen presentation, and chemokine/cytokine production. For example, macrophages isolated from Cnr2 Ϫ/Ϫ mice presented reduced T cell stimulatory potential in vitro compared with WT controls, indicating CB2-mediated alterations in antigen processing (20,21). In addition, it was demonstrated that activation of CB2 negatively regulated IL-12p40 and NO production but enhanced IL-10 release in LPS-activated macrophages (22,23).
In this study, we investigated the role of CB2 in ECM by using Cnr2 Ϫ/Ϫ mice, mixed bone marrow (BM) chimeras, and a pharmacological approach. We could demonstrate that CB2 modulates susceptibility to ECM associated with anti-inflammatory macrophage effector responses. Our findings and the fact that the therapeutic application of the CB2 antagonist SR144528 conferred enhanced CM resistance in C57BL/6 mice provide the basis for the development of novel therapeutic strategies targeting CB2 for the treatment of ECM.
We found that PbA-infected WT mice developed severe ECM starting at day 5 and succumbed to ECM 24 -48 h later (Fig. 1, D and E). In contrast, infected Cnr2 Ϫ/Ϫ mice exhibited a diminished clinical score and an enhanced survival (p Ͻ 0.0001). Approximately 50% of the infected Cnr2 Ϫ/Ϫ mice were protected from development of ECM and finally succumbed to hyperparasitemia and severe anemia after Ͼ20 days. This variability in survival points toward threshold effects influencing cerebral symptoms and survival of Cnr2 Ϫ/Ϫ mice. Parasitemia in Cnr2 Ϫ/Ϫ mice was similar to controls, indicating that a reduced parasite burden in the blood did not account for enhanced survival (Fig. 1F). However, attenuated ECM in Cnr2 Ϫ/Ϫ mice was accompanied by a reduction of parasitespecific 18S RNA levels in the brain (Fig. 1G). The reduced 18S RNA in the presence of a comparable parasitemia in Cnr2 Ϫ/Ϫ versus WT mice indicates a reduction in parasite sequestration in the brain (25). We therefore investigated the integrity of the blood-brain-barrier (BBB) by assessing the diffusion of Evans Blue into the brain (1,2). In contrast to WT mice, infected Cnr2 Ϫ/Ϫ brains displayed a reduced Evans Blue staining, indicative of an attenuated BBB disruption (Fig. 1H). Thus, the severity of PbA infection and incidence of ECM were significantly reduced in Cnr2 Ϫ/Ϫ mice, which could account for the higher survival rate of these mice.
We next asked whether severity of ECM could also be modulated by pharmacological blockade of CB2 (26). PbAinfected C57BL/6 mice were treated daily with the CB2 antagonist SR144528. Most strikingly, these mice also showed an enhanced survival rate (p ϭ 0.0116; Fig. 1I). These data demonstrate that the reduced ECM incidence in Cnr2 Ϫ/Ϫ mice is due to a disrupted CB2 signaling rather than being a consequence of compensatory developmental changes. They also indicate a therapeutic potential of CB2 antagonists/inverse agonists.
Cnr2 Ϫ/Ϫ Myeloid-derived Cells Mediate Enhanced ECM Resistance in BM Chimeras-Recruitment of immune cells to the brain is an important step in ECM pathology (7,8). The absolute number of mononuclear brain-infiltrating cells was markedly reduced in Cnr2 Ϫ/Ϫ mice at day 6 after PbA infection ( Fig. 2A). Within this population, we observed increased percentages of CD11b ϩ cells; Ͼ31% of cells that immigrated into the Cnr2 Ϫ/Ϫ brain expressed CD11b, whereas Ͻ14% of CD11b ϩ cells were found in WT brains (Fig. 2B). Equivalent percentages of CD4 ϩ T cells were found in the WT and Cnr2 Ϫ/Ϫ brains (data not shown), whereas percentages of CD8 ϩ T cells were reduced in the brains of Cnr2 Ϫ/Ϫ mice (Fig.  2C). We also found a reduced number of CD4 ϩ CD25 ϩ FoxP3 ϩ T cells in the Cnr2 Ϫ/Ϫ brain, indicating that a T reg -mediated mechanism does not contribute to disease protection in the absence of CB2 signaling (WT: 0.02 ϫ 10 4 Ϯ 0.01; Cnr2 Ϫ/Ϫ : 0.01 ϫ 10 4 Ϯ 0.005) (27).
100% mortality (Fig. 2D). These data point toward a contribution of CB2-expressing myeloid cells to the enhanced susceptibility to ECM. To analyze a possible role of CB2 in modulating CD8 T cell responses in this model, we investigated Th1 responses and cytotoxicity in Cnr2 Ϫ/Ϫ and WT T cells (Fig. 2E).
CB2 Does Not Affect Antigen Presentation Capacities of BM DCs but Polarization of BM Macrophages in Vitro-To characterize CD11b ϩ myeloid cells in Cnr2 Ϫ/Ϫ mice in more detail, we first analyzed dendritic cells and macrophages generated from the BM of Cnr2 Ϫ/Ϫ and WT mice. GM-CSF-generated Cnr2 Ϫ/Ϫ BM DCs induced equivalent OT-II T cell responses when compared with WT BM DCs, as represented by proliferation, IFN-␥ release, and GrzB production (Fig. 3, A and B). These data indicate that CB2 does not affect basic T cell priming properties of BM DCs. M-CSF-generated CD11b ϩ BM macrophages from Cnr2 Ϫ/Ϫ mice expressed enhanced Ym1 and arginase-1 (Arg-1) mRNA levels under IL-4-stimulatory conditions (Fig. 3C). They also exhibited enhanced expression of mannose macrophage receptors (MMRs) and secreted reduced levels of the pro-inflammatory cytokines TNF and IL-6 ( Fig. 3, D and E), pointing toward anti-inflammatory properties of these cells (32).
Enhanced Numbers of Macrophages and Neutrophils in the Spleens of Cnr2 Ϫ/Ϫ Mice after PbA Infection-In a next step, we investigated myeloid cell responses in vivo in infected Cnr2 Ϫ/Ϫ and WT mice. In the blood, numbers of CD11b ϩ CD11c Ϫ cells and Ly6G hi neutrophils increased equivalently in Cnr2 Ϫ/Ϫ and WT mice (Fig. 3F). In addition, comparable numbers of CD11c hi MHC II ϩ DCs were found at all time points after infection with equal proportions of conventional DCs (CD11b ϩ CD11c ϩ , CD4 ϩ CD11c ϩ DCs, and CD8 ϩ CD11c ϩ DCs) as well as B220 ϩ plasmacytoid DCs (Fig. 3G).
However, in the splenic red pulp of infected Cnr2 Ϫ/Ϫ mice, CD11b expression was markedly increased at day 6 after infection when compared with WT controls (Fig. 4A). Flow cytometry further verified enhanced percentages of CD11b ϩ cells in the spleens of Cnr2 Ϫ/Ϫ mice at this time point (Fig. 4B). Because CD11b is expressed on several myeloid cell types comprising macrophages, neutrophils, and DCs, we further characterized CD11b-expressing cells in Cnr2 Ϫ/Ϫ mice (33)(34)(35). Higher percentages of CD11b ϩ cells that were negative for CD11c accumulated in the spleens of infected Cnr2 Ϫ/Ϫ mice, indicating that these cells do not exclusively represent CD11bexpressing DC subsets (Fig. 4C). Rather, numbers of CD11c hi DCs decreased equally in the spleens of both mouse strains after infection (Fig. 4D). Furthermore, we identified no differences in the percentages of conventional and plasmacytoid DCs in WT versus Cnr2 Ϫ/Ϫ mice at day 6 after infection (Fig. 4E). However, counts of CD11b ϩ Ly6G hi neutrophils markedly increased at this time point of infection in Cnr2 Ϫ/Ϫ mice but not in WT mice (Fig. 4F). We further investigated whether increased numbers of CD11b ϩ cells in the Cnr2 Ϫ/Ϫ spleen also comprise macrophages by utilizing previously published gating strategies (36) (Fig. 4G). Red pulp macrophages (CD11b lo F4/80 ϩ CD169 lo MHC II lo ), marginal zone macrophages (F4/80 Ϫ CD11b Ϫ MARCO ϩ ; MMs) and a subset of CD11b ϩ F4/80 hi CD169 lo MHC-II lo cells accumulated in the spleens of infected Cnr2 Ϫ/Ϫ mice, with the exception of metallophilic marginal zone macrophages (F4/80 Ϫ CD11b Ϫ CD169 ϩ ; MMMs) (Fig. 4G). These data suggest that accumulated CD11b ϩ cells in Cnr2 Ϫ/Ϫ versus WT spleens contain macrophages as well as neutrophils.
CD11b ϩ Cells in the Spleens of Infected Cnr2 Ϫ/Ϫ Mice Exhibit an Anti-inflammatory Phenotype and Function-We next investigated the phenotypic and functional profiles of splenic CD11b ϩ cells in PbA-infected Cnr2 Ϫ/Ϫ and WT mice. At day 6 after infection, CD11b ϩ cells in the Cnr2 Ϫ/Ϫ spleen exhibited enhanced expression of MMRs and Arg-1 activity (Fig. 5, A and  B). Furthermore, isolated Cnr2 Ϫ/Ϫ CD11b ϩ cells secreted higher amounts of the anti-inflammatory cytokine IL-10 at this time point, suggesting that these cells exerted protective functions in Cnr2 Ϫ/Ϫ mice after PbA infection (Fig. 5C).
To clarify the identity of IL-10-producing CD11b ϩ cells in Cnr2 Ϫ/Ϫ mice, we measured cytokine release in DCs, neutrophils, and macrophages isolated from the spleens of PbA-infected mice. We found no difference in the percentages of TNFand IL-10-producing CD11c hi MHC-II ϩ DCs in the spleens of day 6 infected Cnr2 Ϫ/Ϫ and WT mice in vitro and an equivalent increase of TNF ϩ DCs after LPS restimulation of these cells (Fig. 5D). However, Ly6G ϩ cells bona fide neutrophils purified from the Cnr2 Ϫ/Ϫ spleen at day 3 after infection exhibited reduced TNF release and markedly enhanced amounts of IL-10 at a later time point as compared with WT controls (Fig. 5E). In accordance, higher percentages of IL-10-producing CD11b ϩ Ly6G hi neutrophils from the spleens of day 6 infected Cnr2 Ϫ/Ϫ mice were detected by flow cytometry after iRBC restimulation when compared with WT control cells, whereas no differences were observed between the genotypes with regard to frequencies of TNF-producing cells (Fig. 5F). Proportions of TNF-producing MMMs were lower in the spleens of Cnr2 Ϫ/Ϫ when compared with WT mice, whereas TNF-producing cells among the other macrophage subsets were equivalent in both mouse strains (Fig. 5G). All macrophage subsets in the spleens of day 6 infected Cnr2 Ϫ/Ϫ mice exhibited higher percentages of IL-10producing cells when compared with WT mice after iRBC stimulation, with the exception of CD11b ϩ F4/80 ϩ cells (Fig. 5G). Taken together, these findings indicate that in the spleens of PbA-infected Cnr2 Ϫ/Ϫ mice, higher frequencies as well as absolute numbers of IL-10-producing neutrophils and macrophages were present as compared with WT mice.
In the serum of WT and Cnr2 Ϫ/Ϫ mice, no differences were observed in TNF and CCL2 levels during infection (Fig. 5H). However, reduced levels of IFN-␥, IL-1␣, and IL-12 at day 3 as well as higher levels of IL-10 at day 6 hint toward reduced Th1 responses and an anti-inflammatory cytokine milieu in infected Cnr2 Ϫ/Ϫ mice. These data indicate that CB2 also has an impact on systemic cytokine levels after PbA infection.

Discussion
In this study, we demonstrate a role of CB2 in the pathophysiology of experimental CM. Cnr2 Ϫ/Ϫ mice exhibit enhanced survival, a reduced parasite load in the brain, and a diminished BBB disruption after PbA infection in comparison with WT mice. Importantly, we demonstrate that therapeutic application of a CB2 antagonist conferred increased CM resistance in C57BL/6 mice. These results indicate a therapeutic potential of targeting CB2 signaling in ECM and may also facilitate the design of human intervention trials for the treatment of parasite-induced encephalitis.
It was suggested that CB2 could act as a chemokine receptor and is therefore functionally involved in immune cell trafficking (41)(42)(43)(44). Here we showed that CB2 controls the recruitment of peripheral immune cells into the brain during ECM. The degree of mononuclear cells infiltrating the brain was markedly reduced in PbA-infected Cnr2 Ϫ/Ϫ mice. The most striking finding was the nature of the immune cells invading the brain in these animals. Brain infiltrates in infected Cnr2 Ϫ/Ϫ mice showed reduced numbers of lymphoid CD8 ϩ T cells. In contrast, we found that CD11b ϩ cells markedly expanded in the spleen, and higher proportions of these cells were determined in brain infiltrates of infected Cnr2 Ϫ/Ϫ mice.
Our findings suggest that increased numbers of CD11b ϩ cells in Cnr2 Ϫ/Ϫ versus WT spleens represent myeloid macrophages as well as neutrophils. Among those, the specific subset of CD11b ϩ F4/80 hi CD169 lo MHC-II lo cells increased in the spleens of infected Cnr2 Ϫ/Ϫ animals. A corresponding population of F4/80 ϩ CD11b ϩ cells has recently been described as a macrophage subset that accumulated after IL-33 treatment in the spleens of PbA-infected C57BL/6 mice and mediated resistance to CM due to anti-inflammatory properties (45). However, CD11b has also been described on subsets of adaptive immune cells among those CD8 ϩ T cells (46). Although our findings exclude a differential increase of the major adaptive immune cell subsets in infected Cnr2 Ϫ/Ϫ mice, we cannot fully exclude the possibility that minor populations of non-myeloid CD11b-expressing cells also accumulate in the Cnr2 Ϫ/Ϫ spleen.
We identified CB2-expressing myeloid cells as a critical cell type contributing to enhanced susceptibility to ECM in mixed BM chimeras. It is well established that myeloid-derived macrophages express pattern recognition receptors and phagocytose blood pathogens for subsequent transfer to local DCs for cross-presentation to specific T cells (47). Alternatively activated macrophages are specialized in down-regulating immune responses by a specific anti-inflammatory response pattern (32). In vitro, we found a polarization of Cnr2 Ϫ/Ϫ BM macrophages toward M2. In the spleens of infected Cnr2 Ϫ/Ϫ mice, expanded CD11b ϩ cells expressed higher levels of selected M2 markers, such as CD206 (macrophage mannose receptor), arginase-1 activity, and IL-10, an anti-inflammatory cytokine known to confer protection against ECM (48,49). Macrophage subsets and neutrophils expanded in the spleens of Cnr2 Ϫ/Ϫ mice after infection and further exhibited an enhanced capacity to produce IL-10. Also, in the brains of infected Cnr2 Ϫ/Ϫ mice, immigrating CD11b ϩ cells expressed higher CD206 and reduced MHC-II levels, associated with enhanced arginase-1 production. Furthermore, expression of M1-associated cytokines was reduced in the Cnr2 Ϫ/Ϫ brain when compared with WT. In contrast, we found a concomitant and equivalent expression of the M1-associated cytokine TNF in the majority of splenic macrophages in both genotypes. Thus, rather then representing canonical M2 macrophages, expanded CD11b ϩ cells in infected Cnr2 Ϫ/Ϫ mice share only selected factors with anti-inflammatory myeloid cells after PbA infection. In fact, multiple reports defined CD11b ϩ cells as a highly heterogeneous mixture of cells that also include a subset of myeloid-derived suppressor cells (MDSCs) (50). MDSCs have been shown to stain positive for Gr-1, an antibody detecting both Ly6C and Ly6G. A major subset of CD11b ϩ cells in infected Cnr2 Ϫ/Ϫ mice also expressed Ly6C hi and Ly6G hi surface molecules, thus reflecting the phenotype of certain MDSC subsets phenotypically identical to neutrophils or inflammatory monocytes (50). MDSCs exhibit potent immunosuppressive activity on T cell functions ex vivo via mechanisms that depend on nitric oxide, arginase, IFN-␥, and cell-cell contact and have functionally FIGURE 5. Enhanced expression of MMR, arginase-1 activity, and IL-10 in CD11b ؉ cells and higher IL-10 production by macrophages and neutrophils in the spleens of infected Cnr2 ؊/؊ mice. A, percentages of MMR ϩ CD11b ϩ cells in the spleens of day 6 infected WT and Cnr2 Ϫ/Ϫ mice as determined by flow cytometry (mean Ϯ S.E. (error bars), n ϭ 5-6 mice/group). B, measurement of functional arginase-1 activity in spleen-isolated CD11b ϩ cells of control or day 6 infected mice. One representative experiment of two independent experiments is shown in each section (mean Ϯ S.E., n ϭ 5-10 mice/group). C, IL-10 secretion in spleen-isolated CD11b ϩ cells from WT and Cnr2 Ϫ/Ϫ mice at day 3 and day 6 after PbA infection after rechallenge with LPS for 15 h (mean Ϯ S.E., n ϭ 5-6 mice/group). D, diagrams show mean percentages of cytokine-secreting CD11c hi MHC II ϩ DCs from the spleens of WT and Cnr2 Ϫ/Ϫ mice 15 h after stimulation with iRBCs or unstimulated as determined by flow cytometry (mean Ϯ S.E., n ϭ 4 mice/group). E, TNF and IL-10 production by MACS-purified Ly6G ϩ neutrophils isolated from the spleens of WT and Cnr2 Ϫ/Ϫ mice at day 3 or 6 after PbA infection as determined after rechallenge with LPS for 15 h by cytokine bead assays (mean Ϯ S.E., n ϭ 3-5 mice/group). F, diagrams show percentages of cytokine-secreting Ly6G hi neutrophils from the spleens of day 6 infected Cnr2 Ϫ/Ϫ and WT mice 15 h after stimulation with iRBCs or unstimulated as determined by flow cytometry (mean Ϯ S.E., n ϭ 4 mice/group). G, diagrams show mean percentages of cytokine-secreting macrophage subsets from day 6 infected WT and Cnr2 Ϫ/Ϫ mice 15 h after stimulation with iRBCs or unstimulated as determined by flow cytometry (mean Ϯ S.E., n ϭ 4 mice/group). H, cytokine concentrations in the blood from WT and Cnr2 Ϫ/Ϫ mice at days 0, 3, and 6 after PbA infection as determined by cytokine bead assays (mean Ϯ S.E., n ϭ 3-5 mice/group). *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001. been involved in parasitic infections (51,52). Previous studies demonstrated that CB1 as well as CB2 are involved in the induction of MDSC in mice following THC treatment (53). We found that CD11b ϩ cells in infected Cnr2 Ϫ/Ϫ mice shared markers with MDSC, such as low levels of MHC class II and enhanced arginase-1 activity. Future studies will precisely define the ori-gin and function of expanded CD11b ϩ cells in infected Cnr2 Ϫ/Ϫ mice.
Parasite-specific cytotoxic T cells in the spleen induced by antigen-presenting cells are critical mediators of ECM-associated pathology (8, 56 -63). We demonstrate that anti-parasitic T cell responses were reduced in the brains and spleens of infected Cnr2 Ϫ/Ϫ mice. We also identified a reduced vascular leakage in the brains of infected Cnr2 Ϫ/Ϫ mice. It is possible that the maintenance of the BBB in Cnr2 Ϫ/Ϫ mice is related to the observed reduction in brain-immigrating CD8 ϩ T lymphocytes, which have been implicated as cellular mediators of BBB breakdown via perforin-dependent mechanisms (61,62). However, production of cytolytic factors was only reduced in whole spleen cultures but not in isolated splenic CD8 T lymphocytes of infected Cnr2 Ϫ/Ϫ mice. This suggests that suppressive effects in the Cnr2 Ϫ/Ϫ spleens affected anti-parasitic T cell responses required for ECM induction but did not operate in a T cell intrinsic way. Generation of mixed BM chimeras further demonstrated that the expression of CB2 on myeloid cells was associated with mortality after PbA infection. Corroborating our in vivo data, we found comparable GrzB and IFN-␥ production by Cnr2 Ϫ/Ϫ and WT CD8 ϩ T cells upon stimulation by peptideloaded CB2-expressing RAG-2 Ϫ/Ϫ DCs in vitro, indicating that CB2 modulates myeloid responses but does not directly affect T cell functions under these conditions. Nevertheless, we cannot fully exclude the possibility that CD8 ϩ T cells contributed to enhanced protection in PbA-infected Cnr2 Ϫ/Ϫ mice by using alternative effector functions, such as perforin.
The M2 macrophage chemokine CCL17 is an essential factor that mediates enhanced protection against ECM in the absence of CB2 as demonstrated by the full susceptibility of CCL17 ϫ Cnr2 double-deficient mice to ECM. We demonstrated before that the CCL17/CCR4 axis affects the function of CCR4 ϩ myeloid cells, but not T lymphocytes, recruited into the brain during CNS autoimmunity (64). It is thus possible that CCL17 favors brain recruitment or function of myeloid cells rather than CD8 ϩ T cells in ECM with beneficial outcome in disease in Cnr2 Ϫ/Ϫ mice. Here we propose a model whereby CB2 is critical for the development of potent anti-parasitic immune responses early during PbA infection in vivo, thereby causing disease pathology. Pharmacological intervention with a CB2 antagonist induced enhanced disease protection, pointing toward its potential for the treatment of fatal malaria in humans. Clinical studies will help to move toward the development of applications using antagonists/inverse agonists acting on CB2 receptors in fatal malaria.
Endocannabinoid Measurements-Brain endocannabinoid and arachidonic acid levels were measured by GC/MS as reported previously (65). In short, immediately frozen brains were cut along their longitudinal axis and weighed. Chloroform was added (1 ml/100 mg), and the half-brains were homogenized with a Polytron homogenizer (15,000 rpm for 1 min) and sonicated twice for 10 s. The homogenate was added to 10 ml of ice-cold chloroform containing the internal standards (328 pmol of AA-d 8 , 529 pmol of 2-AG-d 5 ). Folch extraction was performed by adding 5 ml of methanol and 2.5 ml of PBS. The mixture was vigorously vortexed and sonicated for 5 min at 4°C, followed by centrifugation for 5 min at 800 ϫ g. The organic phase was recovered into a glass vial and dried under N 2 . The dried organic phase was reconstituted into 1 ml of EtOH and vortexed, and 9 ml of water were added. Solid phase extraction was performed with Sep-Pak cartridges (Waters). Columns were conditioned with 3 ml of methanol (99.8%) and 3 ml of 10% ethanol, and then the samples were applied and washed with 3 ml of 10% ethanol. The arachidonic acid and 2-AG were eluted with 3 ml of ethyl acetate/acetonitrile (8:2). Eluates were dried under N 2 and reconstituted in 50 l of acetonitrile. 20 l of N,N-diisopropylethylamine (Sigma) and 20 l of pentafluorobenzyl bromide (Sigma) (1 g in 3 ml of acetonitrile) were added and incubated for 25 min at 45°C to derivatize arachidonic acid. Excess reagent was evaporated, the sample was reconstituted in 25 l of dimethylisopropylsilylimidazole (TCI Europe), and 2-AG was derivatized with it for 1 h at room temperature. Samples were stored at Ϫ20°C until GC/MS analysis. Samples were injected in splitless mode into an Agilent 6890N GC (HP-5MS column, 30 m). The oven temperature program was as follows: initial temperature 150°C for 1 min, followed by an 8°C/min increase up to 280°C, which was held for 10 min. Helium was used as a carrier gas at a flow rate of 1.5 ml/min. Coupled to the GC was an Agilent 5975C MSD system. The following ions were used for selected ion monitoring analysis: 2-AG/1-AG, m/z 535; 2-AG-d 5 /1-AG-d 5 , m/z 540; AA, m/z 386, 484; AA-d 8 , m/z 392. Because of a significant amount of spontaneous acyl group migration (isomerization) of 2-AG during extraction, the peak areas of 1(3)-AG and 2-AG were combined for quantitative analysis.
Infection, Treatment, and Vascular Leakage-PbA or OVAexpressing P. berghei (PbTg) parasitized red blood cells (RBCs) (kindly provided by Rachel Lundie and Andrew Waters) were used (59). 8 -12-week-old mice were infected intravenously with 5 ϫ 10 4 PbA-RBCs obtained from donor mice infected intraperitoneally with 1 ϫ 10 7 PbA-RBCs. Parasitemia was assessed daily after day 4 in Giemsa-stained tail blood smears. The course of disease was monitored twice daily using the following score according to Amante et al. (27): 0, without symptoms; 1, ruffled fur; 2, hunching; 3, wobbly gait; 4, limb paralysis; 5, convulsions; 6, coma. Each sign was given a score of 1. Animals with severe ECM (cumulative score of 5) and severe anemia were sacrificed according to ethics guidelines. For pharmacological treatment, C57BL/6 mice were daily intraperitoneally injected with 25 g of CB2 antagonist SR144528 (RTI International) in DMSO and PBS. Control groups were injected with DMSO and PBS. The dose of SR144528 was chosen in accordance with earlier studies (66 -68). For assessment of vascular leakage, mice were intravenously injected with 200 l of 2% Evans Blue (Sigma-Aldrich) at day 6 post-infection, and 1 h later, brains were analyzed for dye leakage. After photodocumentation, brains were weighed and incubated in 2 ml of formamide for 48 h at 37°C. 100 l of solution were analyzed in triplicate. Absorbance was measured at 620 nm in an ELISA reader. Concentration of Evans Blue was calculated with a standard curve starting at 200 g/ml and is expressed as g of Evans Blue/g of brain tissue.
Isolation of Mouse Parasites-Preparation of iRBCs was performed based on earlier published protocols (36). In brief, blood from PbA-infected mice was diluted with PBS (pH 7.4) and centrifuged on Lymphoprep cushions at room temperature for 15 min. After removal of the buffy coat, cell pellets were resuspended in PBS before usage.
Generation of BM Chimeric Mice-Recipient mice were irradiated with 9.5 Grays and reconstituted with 0.8 -1.2 ϫ 10 7 BM cells (or 0.4 -0.6 ϫ 10 7 cells mixed with 0.4 -0.6 ϫ 10 7 BM cells from the indicated donors in mixed chimeras) via tail vein injection ϳ7-8 h after irradiation. Mice were treated with antibiotics (trimethoprim and sulfamethoxazole) for 9 days after reconstitution. The level of chimerism was determined by assessing the percentage of CD45.2 donor cells present in the blood of CD45.1 recipients by flow cytometry 7-8 weeks after BM cell transfer. Mice exhibiting Ͼ95% CD45.2 ϩ donor cells in peripheral blood cells were PbA-infected as described above.
Cell Isolation-Brain tissue samples were collected from perfused animals, homogenized by digestion with collagenase/dispase (Roche Applied Science) and DNase I (Roche Applied Science, Mannheim, Germany) at 37°C for 45 min, and gently pressed through a sieve to obtain single cell suspensions. For enrichment of brain mononuclear cells, 30 and 70% Percoll gradients (GE Healthcare UK Ltd., Buckinghamshire, England) were performed and centrifuged at 922 ϫ g for 25 min without brake at room temperature. Cells within the interphase were collected. Splenocytes were isolated by digestion with collagenase type VIII and DNase I (Sigma-Aldrich). CD11b ϩ cells from spleens were purified in two steps. First, CD11c ϩ cells were purified from spleens via bead-coupled antibodies (CD11c MicroBeads, Miltenyi Biotec, Bergisch Gladbach, Germany), and the flow-through was subsequently purified for CD11b ϩ cells (CD11b MicroBeads, Miltenyi Biotec). To isolate splenic neutrophils, the Anti-Ly-6G MicroBead Kit (Miltenyi Biotec) was used. Purified cell populations were cultured in 1 ϫ 10 6 cells/ml in DMEM supplemented with 10% (v/v) heat-inactivated FCS, 1% MEM, 1% penicillin-streptomycin, 0.1% 2-mer-captoethanol (all from Gibco) and stimulated with 100 ng/ml Escherichia coli LPS serotype 0127:B8 (Sigma-Aldrich) for 16 h. For isolation of lymphoid cells, spleens were gently pressed through a sieve to obtain single lymphocyte suspensions. Lymphocyte isolation from the blood of mice was performed by pipetting freshly drawn blood into ice-cold PBS containing 5 mM EDTA (150 l of blood, 2 ml of PBS-EDTA). Samples were inverted immediately to prevent clotting. Red blood cells were digested using ACK lysing buffer (Gibco) as described by the manufacturer. Erythrocytes from the blood were isolated via Pancoll gradient isolation (Pan-Biotech) with centrifugation at 1400 rpm/room temperature without brake for 20 min.
Coculture and Proliferation Assays-1 ϫ 10 5 BM DCs were incubated for 3 h at 37°C, 5% CO 2 in a 100-l total volume of BM DC medium with or without 1 M I-Ab OVA(323-339) peptide. Splenocytes were incubated with CFSE (1 M in PBS) for 20 min at 37°C in the dark. 1 ϫ 10 5 CFSE-labeled CD4 ϩ OT-II cells were cocultured with peptide-pulsed or control BM DCs for 3 days. Proliferation was determined on day 3 via flow cytometry.
Generation and in Vitro Stimulation of BM Macrophages and Dendritic Cells-BM cells for generation of BM DCs were isolated from the hind legs of mice, and 5 ϫ 10 5 cells/ml were cultured in DMEM supplemented with 10% (v/v) heat-inactivated FCS, L-glutamine, MEM, penicillin/streptomycin, 2-mercaptoethanol (all from Gibco), and 10% GM-CSF, and medium was changed at day 3 of culture. BM cells for generation of macrophages were cultured in 1.5 ϫ 10 5 cells/ml in RPMI supplemented with 10% (v/v) heat-inactivated FCS, L-glutamine, MEM, penicillin/streptomycin, 2-mercaptoethanol (all from Gibco), and 15% conditioned medium of L292 cells. On day 3, adherent cells were recultured in complete medium and harvested on day 6. BM-derived macrophages were stimulated in vitro for the indicated times with 100 ng/ml E. coli LPS serotype 0127:B8 (Sigma-Aldrich). For the generation of M2 macrophages, M-CSF-differentiated BM macrophages from WT and Cnr2 Ϫ/Ϫ mice were stimulated with IL-4 at a concentration of 100 units/ml for 48 h (eBiosciences; biological activity 8 ϫ 10 5 units/mg). After stimulation, cells were used for flow cytometry or frozen in TRIzol for further quantitative real-time PCR analyses. BM macrophage cells or the indicated splenic cells were stimulated on day 5 or 6, respectively, for 15 h with 5 ϫ 10 4 iRBCs/well.
Measurement of Cytokines-Cytokines in culture supernatants were determined by a specific sandwich enzyme-linked immunosorbent assay. TNF, IL-6, IFN-␥, IL-10, and GrzB (ELISA Duo-Set, R&D Systems) were used according to the manufacturer's instructions. Plates were read at 450 nm using a SpectraMax 340 microtiter reader (Molecular Devices, Sunnyvale, CA). To determine cytokine levels in the serum or supernatants of CD11b ϩ and Ly6G ϩ splenocytes, the LEGENDplex TM multi-analyte flow assay kits (Biolegend) were used according to the manufacturer's recommendations. Serum samples were diluted 2-fold with assay buffer, whereas cell culture supernatants were assayed without dilution. Data analysis was performed using the LEGENDplex TM data analysis software (Biolegend).
Serum Preparation-To collect serum samples, blood was drawn and transferred into clotting activator containing 1.3-ml microtubes (Sarstedt, Nuembrecht, Germany). Coagulation occurred for 30 min at room temperature.
Measurement of Arginase Activity-Splenocytes were prepared as described above. The homogenate was purified for CD11b ϩ cells via bead-coupled antibodies (CD11b Micro-Beads, human and mouse, Miltenyi Biotec) using the autoMACS Pro Separator (Miltenyi Biotec). The measurement of arginase from supernatant of lysed cells was performed according to the manufacturer's protocol (Arginase Assay Kit, Abnova, Jhongli, Taiwan).
Immunofluorescence-For histological analysis, brain tissues were fixed in 4% buffered formalin and embedded in paraffin. Tissue was then sliced (4 -6 m) and stained with hematoxylin. Spleens were fixed in 4% paraformaldehyde at 4°C for 30 -120 min, washed twice in PBS, saturated in 20% sucrose for 8 h, embedded in tissue-freezing medium (Leica), and snap-frozen in 2-methylbutane (Merck) prechilled with liquid nitrogen. Samples were then sliced into 10 -12-m slices, fixed in acetone (Merck), and stained with biotinylated monoclonal rat anti-mouse Abs for CD11b, CD11c, B220, TCR, CD4, and CD8 (all purchased from eBiosciences). Bound antibody was detected using Streptavidin Alexa 647 (eBiosciences), and samples were mounted with Fluoromount-G (Southern Biotechnology Associates, Inc.) for imaging with confocal microscopy. Adobe Photoshop was used for final image processing.
Total RNA Preparation and Taqman Analysis-Mouse tissue or cells were rapidly dissected, snap-frozen in isopentane, and stored at Ϫ80°C. Total RNA was prepared according to the TRIzol method (Invitrogen). Up to 5 g of RNA and 0.5 g of oligo(dT) 20 primer (Invitrogen) were heated at 70°C for 4 min, chilled on ice, and then reverse transcribed at 42°C for 50 min. A total volume of 20 l included 4 l of first strand buffer (Invitrogen), 2 l of 0.1 M DTT, 1 l of 10 mM dNTPs, 0.5 l of RNase OUT (Invitrogen), and 200 units of Superscript II reverse transcriptase (Invitrogen). Quantitative RT-PCR of cDNA samples was performed using an ABI 7900 sequence detector (PerkinElmer Life Sciences) and Universal PCR Master Mix (PerkinElmer Life Sciences). After incubation of the samples at 50°C for 2 min and 95°C for 10 min, 40 cycles of 95°C for 15 s and 60°C for 1 min were applied. Taqman primer and probe sets were ordered from Applied Biosystems as follows: Cnr1 Mm01212171_s1, Cnr2 Mm00438286_m1, diacylglycerol lipase ␣ Mm00813830_m1, diacylglycerol lipase ␤ Mm00523381_m1, fatty acid amide hydrolase Mm00515684_ m1, monoacylglycerol lipase Mm00449274_m1, ARG Mm00475988_m1. Results are expressed as -fold change relative to the control samples normalized to GAPDH. RT-PCRs for Ifng, Tnf, and inos were performed as described before (69). RT-PCR was performed in accordance for YM1 (ym-1 primer, gcccttattgagaggagcttta (forward) and tacagcagacaagacatcc (reverse); plasmodial 18S RNA, ctaacatggctttgacgggta (forward) and tgtcactaccctcttattt (reverse)).
Statistical Analysis-Analysis of variance and Bonferroni post hoc tests or two-tailed Student's t tests were used to determine differences between groups unless stated otherwise. Mann-Whitney U tests were used as indicated. Data are given as mean Ϯ S.E. The Mantel-Cox log-rank test was used to compare differences in survival and clinical score between groups. All tests were performed with GraphPad Prism version 5 (GraphPad Software, San Diego, CA). p Յ 0.05 was considered significant, and asterisks indicate significance: *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001.