Targeting Glutamine Metabolism Rescues Mice from Late-Stage Cerebral Malaria

Targeting glutamine metabolism rescues mice from late-stage cerebral malaria

Emile B. Gordona,1,2, Geoffrey T. Harta,1, Tuan M. Trana,1,3, Michael Waisberga,4, Munir Akkayaa, Ann S. Kima, Sara E. Hamiltonb, Mirna Penaa, Takele Yazewa, Chen-Feng Qia, Chen-Fang Leec,d, Ying-Chun Loc, Louis H. Millere,5, Jonathan D. Powellc,5, and Susan K. Piercea,5

aLaboratory of Immunogenetics, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Rockville, MD 20852; bDepartment of Laboratory Medicine and Pathology, Center for Immunology, University of Minnesota Medical School, Minneapolis, MN 55454; cSidney Kimmel Comprehensive Cancer Research Center, Department of Oncology, Johns Hopkins University School of Medicine, Baltimore, MD 21287; dChang-Gung Transplantation Institute, Department of Liver and Transplantation Surgery, Chang-Gung Memorial Hospital, Taoyuan 333, Taiwan; and eLaboratory of Malaria and Vector Research, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Rockville, MD 20852

Contributed by Louis H. Miller, August 26, 2015 (sent for review June 19, 2015; reviewed by James R. Mitchell) + The most deadly complication of Plasmodium falciparum infection intravasculature leukocytes, including CD8 T cells, sequestered in the + + is cerebral malaria (CM) with a case fatality rate of 15–25% in brain vessels (13, 14). In ECM monocytes and both CD4 and CD8 T African children despite effective antimalarial chemotherapy. cells have been shown to accumulate in the brain by both flow There are no adjunctive treatments for CM, so there is an urgent cytometry and by intravital imaging (15). Current evidence indicates + need to identify new targets for therapy. Here we show that the that CD8 T cells are the major mediators of death in ECM (16) and + glutamine analog 6-diazo-5-oxo-L-norleucine (DON) rescues mice that antigen-specific CD8 T cells engage parasite antigens cross-pre- from CM when administered late in the infection a time at which sented on MHC class I molecules on brain endothelium, resulting in mice already are suffering blood–brain barrier dysfunction, brain endothelial cell dysfunction by a perforin-dependent mechanism (17). swelling, and hemorrhaging accompanied by accumulation of par- A critical role for metabolic reprogramming in regulating immune + asite-specific CD8 effector T cells and infected red blood cells in responses is becoming increasingly appreciated. Upon activation, T cells the brain. Remarkably, within hours of DON treatment mice showed undergo metabolic reprogramming to meet the increased energetic and blood–brain barrier integrity, reduced brain swelling, decreased biosynthetic demands of growth and effector T-cell functions (18–20). function of activated effector CD8+ T cells in the brain, and levels Reprogramming involves a shift to aerobic glycolysis and increased of brain metabolites that resembled those in uninfected mice. glutaminolysis. Activated T cells import large quantities of Gln and These results suggest DON as a strong candidate for an effective increase their expression of glutaminase (21–23). Because the pathology adjunctive therapy for CM in African children. leading to death in CM is believed to be in part immune mediated, we hypothesized that blocking T-cell metabolism might effectively mitigate + cerebral malaria | adjunctive therapy | CD8 T cells | glutamine the pathology leading to death in HCM. To this end, in the present metabolism | DON

Significance MICROBIOLOGY he World Health Organization estimates that there are nearly T200 million clinical cases of Plasmodium falciparum malaria Cerebral malaria (CM) is a deadly complication of Plasmodium annually (1). For most individuals living in endemic areas, malaria falciparum infection in African children despite effective anti- is uncomplicated and resolves with time. However, in about 1% of malarial treatment. Once signs of neurologic disease have cases, almost exclusively among young children, malaria becomes commenced, there is no adjunctive treatment for CM, and severe and life threatening, resulting in 525,000 deaths each year overall mortality remains high. Thus, a treatment that arrests in Africa alone. One of the most deadly complications of P. fal- disease and promotes healing in the late stages is urgently ciparum infection in humans is cerebral malaria (HCM) charac- needed. Here we report, in an animal model of CM, that the terized by the onset of severe neurological signs such as altered glutamine analog 6-diazo-5-oxo-L-norleucine (DON) is an ef- consciousness, seizures, and coma (2). Autopsy and MRI analyses fective therapy even when treatment is initiated after infected of brains of children with HCM indicate sequestration of infected animals show neurological signs of disease. Within hours of – red blood cells (iRBCs), microhemorrhaging, breakdown of the DON treatment blood brain barrier integrity was restored, and blood–brain barrier (BBB) (3), and a fatal increase in intracranial brain swelling was reduced. These results suggest DON as a pressure resulting from edema (4, 5). At present, despite effective strong candidate for an effective adjunctive therapy for CM in antimalarial drug treatment, mortality for children presenting with African children. HCM remains high, at 15–25%. HCM takes a second toll on Author contributions: E.B.G., G.T.H., T.M.T., M.W., L.H.M., J.D.P., and S.K.P. designed re- African children, leaving survivors at risk for debilitating neuro- search; E.B.G., G.T.H., T.M.T., M.W., M.A., A.S.K., S.E.H., M.P., T.Y., C.-F.Q., C.-F.L., and Y.-C.L. logical defects (6). Thus, there is an urgent need for the devel- performed research; E.B.G., G.T.H., T.M.T., M.W., M.A., A.S.K., C.-F.Q., C.-F.L., Y.-C.L., and J.D.P. opment of effective adjunctive therapies that can be used in analyzed data; and E.B.G., G.T.H., T.M.T., L.H.M., J.D.P., and S.K.P. wrote the paper. conjunction with antimalarials to treat children with HCM. Reviewers included: J.R.M., Harvard School of Public Health. Experimental cerebral malaria (ECM) in mice is a widely used The authors declare no conflict of interest. model of HCM and provides a valuable tool for elucidating the 1E.B.G., G.T.H., and T.M.T. contributed equally to this work. mechanisms involved in CM pathogenesis and identifying cellular 2Present address: Perelman School of Medicine, University of Pennsylvania, Philadelphia, and molecular targets for adjunctive therapy (7). In ECM, 6–7dafter PA 19104. infection with Plasmodium berghei ANKA (PbA), mice of susceptible 3Present address: Division of Infectious Diseases, Indiana University School of Medicine, strains, such as C57BL/6, develop ataxia, paralysis, seizures, and coma Indianapolis, IN 46202. and ultimately die (8). ECM displays key features of HCM, including 4Present address: Department of Pathology, University of Virginia School of Medicine, BBB breakdown, focal hemorrhaging, and brain swelling (9–11). Charlottesville, VA 22903. ’ ECM s pathology also requires sequestration of iRBCs in the brain 5To whom correspondence may be addressed. Email: [email protected], Jpowell@jhmi. vasculature (12), a hallmark of HCM (3). Histological analysis of the edu, or [email protected]. brains of children who died of HCM showed leukocytes, primarily This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. monocytes with phagocytized hemozoin and platelets but also 1073/pnas.1516544112/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1516544112 PNAS | October 20, 2015 | vol. 112 | no. 42 | 13075–13080 Downloaded by guest on October 1, 2021 A C signs) to 10 (moribund) using previously described criteria (25). 15 No Rx Nearly all the untreated mice that were infected with PbA developed 100 No Rx DON Rx d5a DON Rx d5a neurological signs by day 5p p.i. that in most cases were severe 10 DON Rx d5p DON Rx d5p DON Rx d6a > DON Rx d6a (clinical score 6) by day 6a p.i. (Fig. 1B). Treatment with DON 50 5 beginning on day 5a p.i. prevented the development of neurological

% Survival % symptoms in all PbA-infected mice (Fig. 1B). Treatment of mice with % Parasitemia 0 0 456789 DON beginning on day 5p p.i., a point at which most mice had 0510 Day p.i. Day/Time p.i. clinical scores of 2, not only prevented the worsening of clinical signs B D but promoted the rapid resolution of symptoms (Fig. 1B). Re- 10 4 No Rx n.s. markably, treatment of mice on day 6a p.i., when many mice already DON Rx d5a 3 DON Rx d5p had developed clinical scores of 5, blocked the progression of the 5 DON Rx d6a 2 disease and rapidly resolved the symptoms in half of the mice (Fig. 1B). Thus, DON was able to arrest disease and promote healing even (d5p to d6a) to (d5p

Pb A 18s F.C. 1

Clinical Score Clinical when the mice already were displaying signs of neurologic damage. 0 0 5678910 PbA infected: + + A scatter plot showing the clinical scores of individual mice on the DON Rx day p.i.: - d5p Day/Time p.i. RNA day p.i.: d6a d6a day of treatment and the resulting outcome of treatment is given in Fig. S2. For mice treated with DON on d5p p.i., the clinical scores on Fig. 1. DON treatment reduced the mortality associated with ECM. C57BL/6 the day of DON treatment (which ranged from 0 to 6) did not dis- mice were infected with PbA on day 0 and were injected i.p. with saline tinguish the mice that survived from those that died, even though the = (NoRx) (n 49) or with DON (1.3 mg/kg) beginning on day 5 p.i. at 7:00 AM average clinical score was significantly higher for mice that died (DON Rx d5a) (n = 28), on day 5 p.i. at 11:00 PM (DON Rx d5p) (n = 28), or on day 6 p.i. at 7:00 AM (DON Rx d6a) (n = 28). DON treatment was continued every day or every other day as shown in Fig. S1.(A) Kaplan–Meier survival plots. (B) Clinical scores from 0 (no symptoms) to 10 (moribund) of mice in A. (C) Peripheral blood parasitemia for mice in A. Data for DON Rx d5a were A combined from three independent experiments, data for DON Rx d5p were combined from four independent experiments, and data for DON Rx d6a were combined from two independent experiments. Data in B and C are shown as mean and SEM. (D) Fold changes in PbA 18s RNA in brains of DON- PbA infected: + + + + - DON Rx day: - - d5p d5p - treated and untreated PbA-infected mice on day 6a p.i. compared with PbA Assay day: d5p d6a d6a d7a 18s RNA in brains of PbA-infected untreated mice on day 5p p.i. Each dot *** B *** C represents a mouse with the mean and SD given. The results shown are *** *** 1.25 *** 81.0 ** combined from three independent experiments, each having three or four ** mice per group. A Mann–Whitney test showed no significant difference. 1.00 80.5 80.0 0.75 79.5 0.50

study we focus on targeting Gln metabolism for an adjunctive therapy Blue Evans 79.0 for CM using the Gln analog 6-diazo-5-oxo-L-norleucine (DON). DON 0.25 78.5

broadly inhibits Gln metabolism, in part by blocking Gln transport and 0.00 Brain water content78.0 (%) PbA infected: + + + + -- PbA infected: + + + -- inhibiting all three isoforms of glutaminase as well as other Gln-using DON Rx day: -- -- d5p d5p -- DON Rx day: -- -- d5p -- Assay day: d5p d6a d6a d7a enzymes such as the amidotransferases and glutamine synthetase (24). Weight day:d5p d6a d6a Consequently, DON has been shown to be a potent inhibitor of T-cell D proliferation (22). Here we show that DON treatment rescues PbA-infected mice from ECM at late stages in the disease, at a time when the animals X10 X10 X10 show clinical signs of neurological damage and physical loss of BBB PbA infected: -- + + DON Rx day: -- -- d5p integrity, brain swelling, and hemorrhaging. The ability of DON to Assay day: d6a d6a arrest disease is concomitant with a decrease in the effector function ** *** + 30 of parasite-specific CD8 T cells in the brains of treated mice. E However, the striking ability of DON treatment to arrest pathology 20 and promote survival so late in the disease suggests a fundamental

and potentially direct role for Gln metabolism in promoting neu- 10 ropathology. Overall, these results suggest DON as a candidate for Total hemorrhages an adjunctive therapy for HCM in African children. 0 DON Rx day: -- -- d5p d5p Results Assay day: d5p d6a d6a d15a DON Treatment Promotes Survival Even in Late-Stage ECM. To de- Fig. 2. DON treatment promoted BBB function and reduced brain swelling termine if blocking Gln metabolism would inhibit death caused by im- but did not acutely resolve brain hemorrhages in PbA-infected mice. All mice munopathology in a mouse model of ECM, C57BL/6 mice were were infected with PbA and treated with saline or DON (1.3 mg/kg) on day infected with PbA on day 0 and were injected i.p. with DON (1.3 mg/kg) 5p p.i., and the brains were removed and analyzed on the days and times or saline, beginning at 7:00 AM on the morning of day 5 p.i. (day 5a p.i.), indicated. (A) Representative images of the brains of mice injected with EB. at 11:00 PM on day 5 p.i. (day 5p p.i.), or at 7:00 AM on day 6 p.i. (B) EB levels in the brains were quantified and expressed relative to the EB (day 6a p.i.). DON treatments were repeated either every day or levels in the brains of PbA-infected, untreated mice on d6a p.i. Each symbol every other day (Fig. S1). The majority of untreated PbA-infected represents one mouse. The data are combined from three independent experiments and are shown as mean and SD. (C) Brain water content expressed mice died during day 6 p.i., and all died by day 7 p.i. (Fig. 1A). In as the weight of each brain after desiccation divided by the weight before contrast, all the mice treated with DON beginning on day 5a p.i. desiccation × 100 is given. Data are combined from two independent ex- and 80% of the mice treated on day 5p p.i. survived as followed until periments and are shown as mean and SD. (D) Representative images of brain day 12 p.i. Remarkably, nearly 50% of mice treated as late as day 6a sections; hemorrhages are indicated by white arrows. (E) Quantification of p.i. survived. brain hemorrhages. Each symbol represents one mouse. Data are combined Mice were evaluated for the development of neurological signs from three independent experiments. Mann–Whitney tests were used for associated with ECM and were given clinical scores between 0 (no comparison of groups (**P < 0.005, ***P < 0.0005).

13076 | www.pnas.org/cgi/doi/10.1073/pnas.1516544112 Gordon et al. Downloaded by guest on October 1, 2021 following treatment. For mice treated with DON on day 6a p.i., DON Treatment Promotes Recovery of BBB Integrity and Reduction in clinical scores below 8 did not distinguish the mice that survived Brain Swelling but Does Not Have an Immediate Effect on Brain from those that did not. However, all mice with clinical scores of 9 Hemorrhaging. Compromise of the BBB, brain swelling, and hemor- or 10 died following DON treatment, indicating that there is a rhaging are major components of the neuropathology observed in tipping point clinically at whichmicenolongercanberescuedby ECM. Thus, we assessed the effect of DON treatment on the integrity DON treatment. of the BBB by measuring the leakage of the dye Evans blue (EB) into In both DON-treated and untreated PbA-infected mice para- the brains (26, 27). Mice were injected retroorbitally with EB and sitemia increased similarly with time, reaching peak levels of ∼8– were killed 3 h later; then their brains removed, and EB was quan- 12% between d5 and d6 p.i. (Fig. 1C). Nearly all untreated mice died tified. Both visual inspection (Fig. 2A) and EB quantification (Fig. 2B) by day 7 p.i., with peripheral parasitemia of 10–12%. In the DON- showed significant leakage of EB into the brains of PbA-infected, treated mice that survived beyond day 6 p.i., the peripheral para- untreated mice on day 5p p.i., and the leakage increased further on sitemia decreased beginning on day 7 p.i., reaching low levels by day day 6a p.i. (Fig. 2 A and B). Remarkably, on day 6a p.i. EB leakage 8 p.i. However, during the critical period of day 6 p.i., during which was significantly less in the brains of PbA-infected DON-treated mice untreated mice die and DON treated mice survive, the parasite loads than in the brains of untreated mice, and the EB leakage in the brains were indistinguishable in the brains of treated and untreated mice of PbA-infected DON-treated mice decreased significantly further by (Fig. 1D), suggesting that inhibition of parasite growth by DON is day 7a p.i. (Fig. 2 A and B). not the primary mechanism promoting survival. Consistent with the We also determined the effect of DON treatment on the water ability of DON to inhibit parasite replication, continued DON content of the brain as a measure of cerebral edema by weighing the treatment suppressed parasitemia to under 5% as followed out to brains before and after desiccation. The brains of PbA-infected mice day 17 p.i., and when DON treatment was stopped, parasitemia in- had significantly greater water content than the brains of uninfected creased to a mean of 40% by day 21 p.i. (Fig. S3). mice on both day 5p p.i. and day 6a p.i. (Fig. 2C). Treating mice with

ABC 6 5 5 10 n.s. 10 10 n.s. n.s. 5 10 4 104 10 104 3

3 NK cell #

DC cell # 10 10 103 Macrophage/ Neutrophil cell # 2 2 102 10 10 PbA Infected: + + - PbA Infected: + + - PbA Infected: + + - DON Rx day: -- d5p -- DON Rx day: -- d5p -- DON Rx day: -- d5p --

6 n.s. Tetramer DE105 10 FNo Enrichment Enrichment No Enrichment n.s. 5 104 10

3 4 10 10 MICROBIOLOGY CD44 CD44 CD44 3 102 CD8+ T cell # 10 CD4+ T cell #

2 101 10 GAP50 Tet GAP50 Tet GAP50 Tet PbA Infected: + + - PbA Infected: + + - PbA Infected: - + + DON Rx day: -- d5p -- DON Rx day: -- d5p -- DON Rx day: - -- d5p

6 GH10 **

5 10 n.s. CD44 CD44 CD44 4 6.45 32.2 17.7 10

CD107 CD107 CD107 # CD8+, GAP50Tet+ # CD8+, 3 10 PbA Infected: - + + PbA Infected: + + + + - DON Rx day: - -- d5p DON Rx day: -- d5p -- d5p - Tissue:Spleen Spleen Brain Brain Spleen + LN n.s. IJ70 80 60 60 50 *** *** 40 * 40 30 CD107a/b+

20 20

10 % CD8+, CD107a/b+ % CD8+, GAP50 Tet+, 0 0 PbA Infected: + + + + PbA Infected: + + + + DON Rx day: -- d5p -- d5p DON Rx day: -- d5p -- d5p Tissue: Spleen Spleen Brain Brain Tissue:Spleen Spleen Brain Brain

Fig. 3. DON treatment reduced CD8+ T-cell degranulation but not the accumulation of immune cells in the brains of PbA-infected mice. Uninfected mice and mice infected with PbA were treated with DON (1.3 mg/kg) or saline on day 5p p.i. (A–E) On day 6a p.i. mice were perfused with cold PBS, the brains and/or spleens were removed, and single-cell suspensions were prepared. Cells were analyzed by flow cytometry using the gating strategy shown in Fig. S4. Shown + + are the number of cells per brain for macrophage/dendritic cells (DC) (A), neutrophils (B), NK cells (C), CD 4 T cells (D), and CD8 T cells (E). Data were + combined from three independent experiments. (F) Representative flow cytometry plots of GAP50 MHC class I Db tetramer-binding CD8 T cells in unenriched spleen/lymph node cell populations (Left and Right) and in GAP50-tetramer–binding cells enriched by GAP50-tetramer-bound magnetic beads (Center). + (G) The number of GAP50-tetramer–binding CD8 T cells in spleen and brains. Data were combined from three independent experiments. (H) Representative flow cytometry plots of CD8+ T cells in the spleens of mice that received fluorescently labeled CD107-specific Abs i.v. 1 h before being killed to allow labeling of CD107-expressing cells in vivo. (I and J)The percent of total CD8+ T cells that expressed CD107 (I) and the percent of GAP50-tetramer–binding CD8+ T cells that expressed CD107 (J) are shown. Data were combined from three independent experiments. Mann–Whitney tests were used for statistical analysis. ns, not significant; *P < 0.05, **P < 0.005, ***P <0.0005.

Gordon et al. PNAS | October 20, 2015 | vol. 112 | no. 42 | 13077 Downloaded by guest on October 1, 2021 DON on day 5p p.i. significantly reduced the water content of the that in the spleens of uninfected mice the frequency of GAP50-tet- + brains measured on day 6a p.i. Thus, DON not only arrested the ramer–binding CD8 T cells was less than 0.1% (Fig. 3F, Left). En- disease process; it also promoted resolution even when treatment richment for GAP50-tetramer–binding cells in the spleen and lymph was initiated at a time when significant BBB dysfunction and brain nodes of uninfected mice by GAP50-tetramer–bound magnetic bead + swelling already were manifest. purification showed ∼2,700 GAP50-tetramer–binding CD8 Tcells, − To quantify brain hemorrhages, brain sections were stained with the majority of which were resting CD44 cells (Fig. 3F, Center). In H&E. PbA-infected, untreated mice developed petechial hemor- contrast, in PbA-infected mice the frequency of GAP50-tetramer– + rhages throughout the brain by day 5p p.i. that increased in number binding T cells in the spleen was ∼3%,andthesewereCD44 effector by day 6a p.i. (Fig. 2 D and E). Treatment of mice with DON be- T cells (Fig. 3FRight). DON treatment of PbA-infected mice resulted + ginning on day 5p p.i. had no significant effect on the number of in a small decrease in the number of GAP50-tetramer–binding CD8 hemorrhages that developed. However, by day 15 p.i. the hemor- T cells in the spleens but had little effect on the number of GAP50- rhages were no longer evident (Fig. 2 D and E). Thus, DON-treated tetramer–binding T cells in the brains (Fig. 3G). mice survived local hemorrhaging in the brain when the overall in- Having observed that DON did not inhibit expansion of parasite- + + tegrity of the BBB was restored and brain swelling was reduced. specific CD8 T cells, we determined the effect of DON on CD8 T-cell effector function. The development of ECM has been shown + DON Treatment Results in a Reduction in the Number of CD8+ Effector to be dependent on degranulation and perforin release of CD8 + T Cells That Degranulate in the Brains of Infected Mice. CD8 T cells T cells that accumulate in the brains of PbA-infected mice (16). + have been shown to play a major role in promoting death in ECM (16), Degranulation of CD8 T cells results in the expression of CD107 on and Gln metabolism is critical for the differentiation, proliferation, and their plasma membranes. To assess the ability of DON to inhibit + function of effector T cells. Thus, we determined the effect of DON CD8 T-cell degranulation in the brains of infected mice, fluo- treatment on T cells in the brains of infected mice. DON was admin- rescently labeled CD107-specific mAbs were administered i.v. to istered to PbA-infectedmiceonday5pp.i.,andonday6ap.i.mice mice 1 h before tissues were removed for analysis to allow in vivo + were terminally anesthetized and transcardially perfused with cold PBS. labeling of CD107 cells. Shown are representative flow cytometry + + The brains were collected, and single-cell suspensions were prepared plots showing the percent of CD44 , CD8 T cells that express and analyzed by flow cytometry. First we examined immune cell in- CD107 in the spleens of uninfected mice, PbA-infected mice, and filtration of both the treated and untreated mice. Compared with un- PbA-infected mice treated with DON (Fig. 3H). DON treatment of infected mice, the brains of infected mice showed large increases in all PbA-infected mice resulted in significant decreases in the percent of + + + + immune cell types analyzed, including CD8 and CD4 T cells, neu- CD8 T cells that were CD107 in both the spleens and brains (Fig. + trophils, macrophages, and natural killer (NK) cells. Interestingly, de- 3I). The percent of GAP50-tetramer–binding CD8 T cells that were + spite its ability to reverse disease, DON treatment had no effect on the CD107 also decreased significantly in the brains of PbA-infected numbers of immune cells in the brains of infected mice (Fig. 3 A–E). mice (Fig. 3J). Thus, although the late-stage treatment of DON does + Next, we examined the effect of DON on the expansion and not block the expansion of the PbA-specific CD8 T cells, it appears + function of parasite-specific CD8 T cells. The accumulation of PbA- to block their effector function as measured by degranulation. + specific CD8 T cells in the brains and spleens was quantified using a peptide-MHC class I tetramer composed of the PbA glideosome- The Metabolic Profiles of Brains from DON-Treated PbA-Infected Mice associated protein 50 (GAP50) peptide, SQLLNAKYL, bound to Resemble Those of Uninfected Mice. The ability of DON to arrest MHC class I Db (referred to as a GAP50-tetramer) that identifies ECM rapidly at such a late stage of disease and to promote recovery + ∼5% of splenic CD8 TcellsinPbA-infected mice (28). We verified led us to wonder whether, in addition to inhibiting immune-mediated

A BC PC 2 Uninfected, untreated PC 2 PC 2 Uninfected, DON treated Infected, DON treated, low clinical scores (<6) Infected, DON treated, high clinical scores (≥6) Infected, untreated

PC 3 PC 1 PC 3 PC 1 PC 3 PC 1

Brain Liver Serum

Infected, untreated Infected, DON treated, low clinical scores (<6) vs. Uninfected, untreated 144 81 8 vs. Infected, untreated

Fig. 4. DON significantly alters metabolism in the brain during ECM. (A–C) Principle components analysis for each of three tissue types collected from mice at day 6a p.i. for the five experimental groups defined by infection, treatment, and clinical outcome (Table S1). Principal components were determined from all detectable metabolites for the brain (438 metabolites) (A), liver (544 metabolites) (B), and serum (563 metabolites) (C). Each sphere represents one tissue sample from one mouse. (D) Venn diagram showing the number of differentially abundant brain metabolites for the two main comparisons: infected, untreated mice vs. uninfected, untreated mice and infected, DON-treated mice with low clinical scores vs. infected, untreated mice. Differential abundance thresholds were an absolute fold-change in abundance of ≥1.2 and a false-discovery rate of <5% (Welch’s t test).

13078 | www.pnas.org/cgi/doi/10.1073/pnas.1516544112 Gordon et al. Downloaded by guest on October 1, 2021 pathology, DON also might affect pathology induced by altered brain treatment. These observations suggest that DON’sprotectiveeffectis + metabolism. To investigate this possibility, we profiled metabolites in mediated not by preventing the proliferation and generation of CD8 + the brain, liver, and serum from five groups of mice that differed in effector T cells but rather by blocking CD8 T-cell effector function. + their infection, treatment, and/or clinical status (Table S1). Principal The ability of DON to block CD8 T-cell function may account, in component analysis of all detected metabolites revealed distinct part, for the remarkably fast kinetics by which DON mediates its clustering of samples according to infection status in all tissues (Fig. 4 protective effect. The ability of DON to promote survival at such a late A–C). Notably, we observed that infected, DON-treated mice with low clinical scores (<6) clustered with uninfected mice for metabolites in stage of the disease when animals are suffering from BBB dysfunction the brain (Fig. 4A). In contrast, when we examine the effect of DON and brain swelling distinguishes our findings from all other attempts to on systemic metabolism, as determined by serum metabolites, or the treat and reverse ECM. effect of DON on liver metabolism, this correlation did not hold (Fig. Although our studies were motivated by the ability of DON to 4 B and C), suggesting that DON treatment in ECM results in specific inhibit immune function, we also investigated the effect of DON effects on brain metabolites. Because DON blocks the initial step in cellular glutaminolysis by

inhibiting glutaminase (24) (Fig. S5A), we first examined the metabolites Uninfected, untreated (n = 5) of glutaminolysis and observed that PbA infection alone affects these Uninfected, DON treated (n = 5) metabolites differently in the brain than in the liver and serum. However, Infected, DON treated, low clinical scores (<6) (n = 10) we did not observe a clear pattern predicted for treatment with a glu- Infected, DON treated, high clinical scores (≥6) (n = 6) taminase inhibitor (Fig. S5B). Of the 144 brain metabolites that changed Infected, untreated (n = 8) significantly in infected versus uninfected mice, 81 overlapped with the 89 significantly affected metabolites identified by comparing infected, DON-treated mice with low clinical scores and infected, untreated mice (Fig. 4D). Pathways significantly affected by PbA infection involved cit- Pathway rulline metabolism, the urea cycle, or nitric oxide biosynthesis (aspartate, citrulline, ornithine, and urea) and were similar to pathways affected by DON treatment of PbA-infected mice (Fig. S6). Strikingly, all 81 over- Fatty Acid Metabolism lapping metabolites were metabolites that were reversed by DON treatment (Table S2). Unsupervised hierarchal clustering analysis on Fatty Acid, Monohydroxy these 81 brain metabolites accurately grouped the samples by infection, Glutamate Metabolism treatment, and clinical status (Fig. 5), with 9 of 10 PbA-infected, DON- treated mice with low clinical scores (<6) clustering together and dem- Histidine Metabolism onstrating remarkable metabolic similarities to uninfected mice. DON appears to reverse brain metabolism associated with disease or to pre- Leucine, Isoleucine vent the metabolic perturbation, presumably acting through its immu- and Valine Metabolism nomodulatory effects or acting directly on brain metabolism. Although these data do not identify the specific metabolites promoting disease, they do identify a set of metabolites which are specifically associated with Lysine Metabolism ECM and are selectively blocked by treatment with DON. MICROBIOLOGY Discussion HCM is a deadly complication of P. falciparum malaria despite treatment with effective antimalarial drugs. That is, even in when the rep- Lysolipid lication of P. falciparum is effectively suppressed, there is no effective adjunctive treatment for HCM once signs of neurologic disease have commenced, and overall mortality remains high. Therefore a treatment Medium Chain Fatty Acid that can arrest disease and promote healing in the late stages is urgently Methionine, Cysteine, SAM needed. Here we report that the Gln analog DON is an effective and Taurine Metabolism therapy for ECM even when treatment is first initiated after infected Phenylalanine and animals show neurological signs of disease. This clinical response was Tyrosine Metabolism accompanied by the ability of DON to inhibit pathology as measured by Phospholipid Metabolism decreases in BBB dysfunction, brain swelling, and degranulation of + parasite-specific CD8 T cells that accumulated in the brain. Further- Primary Bile AcidMetabolism more, DON is able to reverse or prevent metabolic changes associated Purine Metabolism with the disease state. Previously, DON was shown to have antiparasitic activity, albeit weak, Pyrimidine Metabolism both in vitro and in vivo (29, 30). Although the antiparasitic activity of DON may contribute to its ability to arrest ECM and promote healing, Tryptophan Metabolism Urea cycle; Arginine and we do not believe plays a critical role. Indeed, DON had little effect on Proline Metabolism the parasite load in the brains of PbA-infected mice during the critical Samples period when the BBB was restored and brain swelling decreased in DON-treated mice. + On the other hand, CD8 effector T cells that release perforin have −40-2 2 been shown to play a critical role in promoting the pathogenesis that log(normalized abundance) + constitutes ECM (16). CD8 T-cell activation results in a dramatic shift Fig. 5. Recovery from ECM with DON treatment is associated with normal from oxidative metabolism to aerobic glycolysis and glutaminolysis – brain metabolites. Clustering heat map showing the normalized abundance necessary for T-cell expansion and effector function (18 20). Both of the 81 metabolites that were shared between the two main comparisons T-cell cytokine production and proliferation are blocked by restricting in Fig. 4D. Each row is a metabolite, and each column represents brain tissue the availability of extracellular Gln (21) or by treating T cells with DON from a single mouse. Unsupervised hierarchical clustering of samples was (22). DON treatment of PbA-infected mice did not prevent or decrease performed using Ward’s method for linkage analysis and Pearson’s dissimi- + the accumulation of CD8 T cells in the brain. Rather, the number larity as distance measure. Metabolites were grouped according to pathway + of CD8 T cells that degranulated was decreased following DON annotations (labels at right).

Gordon et al. PNAS | October 20, 2015 | vol. 112 | no. 42 | 13079 Downloaded by guest on October 1, 2021 treatment on infected brain metabolism by interrogating the brain ACUC approved the Animal Study Proposal identification number LIG-1E, for known metabolites. To this end, we observed more than 81 which adheres to the regulations of the Animal Welfare Regulations and metabolic changes in the brains of infected mice compared with Public Health Service Policy on Humane Care and Use of Laboratory Animals. uninfected mice that were either reversed or blocked by DON. Detailed materials and methods can be found in SI Materials and Methods. Unequivocally, such findings do not reveal the mechanism by which C57BL/6 mice were infected with PbA and monitored for peripheral blood DON mediates its effects. The complexity of the fluctuations in Gln parasitemia (31) and ECM clinical scores (25). DON (1.3 mg/kg) was admin- and Glu levels in the brain during infection and upon treatment most istered i.p. Parasite loads were quantified by quantitative PCR in brain tissue likely reflects DON’s inhibition not only of glutaminase activity but taken from mice that were anesthetized and transcardially perfused (32). also of Gln transport and other Gln-using enzymes (24). Rather, the Terminal assays also were carried out to quantify brain pathology. BBB in- changes we identified provide a metabolic profile, a biomarker for tegrity was assessed by quantifying EB in brains of mice that were injected diseased brains in ECM. Notably, although these 81 metabolites with EB intraorbitally 3 h before the mice were anesthetized and trans- correlated with DON treatment and disease state in the brain, such cardially perfused and the brains were removed (33). Brain swelling was correlations were not observed in the serum or the livers of the measured by weighing brains before and after desiccation. Brain hemor- DON-treated mice. rhages were quantified by histological examination of H&E-stained brain In summary, we demonstrate, for the first time to our knowledge, sections. To characterize leukocytes that accumulated in brains, single-cell the ability of a pharmacologic intervention to arrest disease and pro- suspensions were prepared of brains from anesthetized and transcardially mote survival in the late stages of ECM. Because of the high mortality perfused mice. Cells were analyzed by flow cytometry using appropriately + in HCM once neurologic symptoms are manifested, our findings have labeled antibodies specific for cell-surface markers and a CD8 T-cell–specific relevant and immediate clinical implications. Furthermore, our studies GAP50-tetramer (28). Single-cell suspensions of spleens were analyzed sim- reveal a potentially selective metabolic signature for ECM and the ilarly. To quantify degranulation of T cells in vivo, mice were given phyco- subsequent reversal of diseases. It will be of interest to determine if this erythrin (PE)-conjugated antibodies specific for CD107a and CD107b i.v. 1 h signature is applicable to other neurologic disorders and whether DON before mice were killed and brains and spleens were removed for analysis by might be capable of arresting and/or reversing a broad spectrum of flow cytometry (34). Metabolic profiling of brain, liver, and serum was neuroinflammatory diseases. performed at Metabolon (35). Materials and Methods ACKNOWLEDGMENTS. This study was supported by the Intramural Research All experiments were approved by the National Institute of Allergy and In- Program of the NIH, National Institute of Allergy and Infectious Diseases and fectious Diseases Animal Care and Use Committee (NIAID ACUC). The NIAID by NIH Grant R01AI077610 (to J.D.P.).

1. World Health Organization (2014) World Malaria Report 2014 (WHO, Geneva). 20. Pollizzi KN, Powell JD (2014) Integrating canonical and metabolic signalling pro- 2. Molyneux ME, Taylor TE, Wirima JJ, Borgstein A (1989) Clinical features and prog- grammes in the regulation of T cell responses. Nat Rev Immunol 14(7):435–446. nostic indicators in paediatric cerebral malaria: A study of 131 comatose Malawian 21. Carr EL, et al. (2010) Glutamine uptake and metabolism are coordinately regulated by children. Q J Med 71(265):441–459. ERK/MAPK during T lymphocyte activation. J Immunol 185(2):1037–1044. 3. Taylor TE, et al. (2004) Differentiating the pathologies of cerebral malaria by post- 22. Wang R, et al. (2011) The transcription factor Myc controls metabolic reprogramming mortem parasite counts. Nat Med 10(2):143–145. upon T lymphocyte activation. Immunity 35(6):871–882. 4. Potchen MJ, et al. (2012) Acute brain MRI findings in 120 Malawian children with 23. Nakaya M, et al. (2014) Inflammatory T cell responses rely on amino acid transporter cerebral malaria: New insights into an ancient disease. AJNR Am J Neuroradiol 33(9): ASCT2 facilitation of glutamine uptake and mTORC1 kinase activation. Immunity – 1740–1746. 40(5):692 705. 5. Seydel KB, et al. (2015) Brain swelling and death in children with cerebral malaria. N 24. Thangavelu K, Chong QY, Low BC, Sivaraman J (2014) Structural basis for the active Engl J Med 372(12):1126–1137. site inhibition mechanism of human kidney-type glutaminase (KGA). Sci Rep 4:3827. 6. Shikani HJ, et al. (2012) Cerebral malaria: We have come a long way. Am J Pathol 25. Waisberg M, Vickers BK, Yager SB, Lin CK, Pierce SK (2012) Testing in mice the hy- 181(5):1484–1492. pothesis that melanin is protective in malaria infections. PLoS One 7(1):e29493. 7. Grau GE, Craig AG (2012) Cerebral malaria pathogenesis: Revisiting parasite and host 26. Yen LF, Wei VC, Kuo EY, Lai TW (2013) Distinct patterns of cerebral extravasation by contributions. Future Microbiol 7(2):291–302. Evans blue and sodium fluorescein in rats. PLoS One 8(7):e68595. 8. Engwerda C, Belnoue E, Grüner AC, Rénia L (2005) Experimental models of cerebral 27. Nag S (2003) Blood-brain barrier permeability using tracers and immunohistochem- istry. Methods Mol Med 89:133–144. malaria. Curr Top Microbiol Immunol 297:103–143. 28. Howland SW, et al. (2013) Brain microvessel cross-presentation is a hallmark of ex- 9. Nacer A, et al. (2012) Neuroimmunological blood brain barrier opening in experi- perimental cerebral malaria. EMBO Mol Med 5(7):916–931. mental cerebral malaria. PLoS Pathog 8(10):e1002982. 29. Plaimas K, et al. (2013) Computational and experimental analysis identified 6-diazo-5- 10. Promeneur D, Lunde LK, Amiry-Moghaddam M, Agre P (2013) Protective role of brain oxonorleucine as a potential agent for treating infection by Plasmodium falciparum. water channel AQP4 in murine cerebral malaria. Proc Natl Acad Sci USA 110(3): Infect Genet Evol 20:389–395. 1035–1040. 30. Queen SA, Jagt DL, Reyes P (1990) In vitro susceptibilities of Plasmodium falciparum to 11. Penet MF, et al. (2005) Imaging experimental cerebral malaria in vivo: Significant role compounds which inhibit nucleotide metabolism. Antimicrob Agents Chemother of ischemic brain edema. J Neurosci 25(32):7352–7358. 34(7):1393–1398. 12. Baptista FG, et al. (2010) Accumulation of Plasmodium berghei-infected red blood 31. Malleret B, et al. (2011) A rapid and robust tri-color flow cytometry assay for moni- cells in the brain is crucial for the development of cerebral malaria in mice. Infect toring malaria parasite development. Sci Rep 1(118):14. – Immun 78(9):4033 4039. 32. Gordon EB, et al. (2015) Inhibiting the Mammalian target of rapamycin blocks the 13. Grau GE, et al. (2003) Platelet accumulation in brain microvessels in fatal pediatric development of experimental cerebral malaria. MBio 6(3):00725–00715. – cerebral malaria. J Infect Dis 187(3):461 466. 33. Kim JV, Kang SS, Dustin ML, McGavern DB (2009) Myelomonocytic cell recruitment 14. Dorovini-Zis K, et al. (2011) The neuropathology of fatal cerebral malaria in malawian causes fatal CNS vascular injury during acute viral meningitis. Nature 457(7226): – children. Am J Pathol 178(5):2146 2158. 191–195. 15. Pai S, et al. (2014) Real-time imaging reveals the dynamics of leukocyte behaviour 34. Yuzefpolskiy Y, Baumann FM, Kalia V, Sarkar S (2015) Early CD8 T-cell memory pre- during experimental cerebral malaria pathogenesis. PLoS Pathog 10(7):e1004236. cursors and terminal effectors exhibit equipotent in vivo degranulation. Cell Mol + 16. Howland SW, Claser C, Poh CM, Gun SY, Renia L (2015) Pathogenic CD8 T cells in Immunol 12(4):400–408. experimental cerebral malaria. Semin Immunopathol 37(3):221–223. 35. Shin SY, et al.; Multiple Tissue Human Expression Resource (MuTHER) Consortium 17. Nitcheu J, et al. (2003) Perforin-dependent brain-infiltrating cytotoxic CD8+ T lym- (2014) An atlas of genetic influences on human blood metabolites. Nat Genet 46(6): phocytes mediate experimental cerebral malaria pathogenesis. J Immunol 170(4): 543–550. 2221–2228. 36. Haluszczak C, et al. (2009) The antigen-specific CD8+ T cell repertoire in unimmunized 18. MacIver NJ, Michalek RD, Rathmell JC (2013) Metabolic regulation of T lymphocytes. mice includes memory phenotype cells bearing markers of homeostatic expansion. Annu Rev Immunol 31:259–283. J Exp Med 206(2):435–448. 19. O’Sullivan D, Pearce EL (2015) Targeting T cell metabolism for therapy. Trends Immunol 37. Storey JD, Tibshirani R (2003) Statistical significance for genomewide studies. Proc 36(2):71–80. Natl Acad Sci USA 100(16):9440–9445.

13080 | www.pnas.org/cgi/doi/10.1073/pnas.1516544112 Gordon et al. Downloaded by guest on October 1, 2021