Rocaglates As Dual-Targeting Agents for Experimental Cerebral Malaria

Rocaglates as dual-targeting agents for experimental cerebral malaria

David Langlaisa,b,1, Regina Cencica,b, Neda Moradina,b, James M. Kennedya,b, Kodjo Ayic, Lauren E. Brownd, Ian Crandalle, Michael J. Tarrya, Martin Schmeinga, Kevin C. Kainc, John A. Porco Jr.d, Jerry Pelletiera,b, and Philippe Grosa,b,1

aDepartment of Biochemistry, McGill University, Montreal, QC, H3G 0B1 Canada; bMcGill University Research Center on Complex Traits, McGill University, Montreal, QC, H3G 0B1 Canada; cDepartment of Medicine, Toronto General Hospital-University Health Network, University of Toronto, Toronto, ON, M5G 2C4 Canada; dDepartment of Chemistry, Center for Molecular Discovery, Boston University, Boston, MA 02215; and eLeslie Dan Faculty of Pharmacy, University of Toronto, Toronto, ON, M5S 3M2 Canada

Edited by David A. Fidock, Columbia University, New York, NY, and accepted by Editorial Board Member Carl F. Nathan January 19, 2018 (received for review July 24, 2017) Cerebral malaria (CM) is a severe and rapidly progressing complica- rent WHO-approved standard of care for uncomplicated malaria. tion of infection by Plasmodium parasites that is associated with high ACTs include Art (artesunate, dihydro-Art, artemether, and rates of mortality and morbidity. Treatment options are currently arteether) in combination with other drugs with pharmacokinetic few, and intervention with artemisinin (Art) has limited efficacy, a profiles showing longer circulation half-lives, such as mefloquine, problem that is compounded by the emergence of resistance to Art in lumefantrine, amodiaquine, piperaquine, and sulfadoxine/pyri- Plasmodium parasites. Rocaglates are a class of natural products methamine (15, 16). In contrast, treatment options for CM are derived from plants of the Aglaia genus that have been shown to few and are effective only when infection is detected early before interfere with eukaryotic initiation factor 4A (eIF4A), ultimately block- irreversible neurological symptoms (5–7). Administration of i.v. ing initiation of protein synthesis. Here, we show that the rocaglate artesunate is the WHO-approved first-line therapy, but efficacy CR-1-31B perturbs association of Plasmodium falciparum eIF4A is limited, with 18–30% of CM cases still being fatal in large (PfeIF4A) with RNA. CR-1-31B shows potent prophylactic and thera- randomized clinical trials in children and adults (17–19). Because peutic antiplasmodial activity in vivo in mouse models of infection of the lack of equipotent drug alternatives, the recent emergence with Plasmodium berghei (CM) and Plasmodium chabaudi (blood- and spread of Art resistance in different areas of Southeast Asia is a stage malaria), and can also block replication of different clinical iso- major threat to global health (8, 12). Although the mechanisms of lates of P. falciparum in human erythrocytes infected ex vivo, includ- Art resistance in Plasmodium are being characterized (20–24), it is ing drug-resistant P. falciparum isolates. In vivo, a single dosing of unclear that novel treatment options based on this knowledge will CR-1-31B in P. berghei-infected animals is sufficient to provide pro- be immediately forthcoming. Thus, there is a great urgency to detection against lethality. CR-1-31B is shown to dampen expression of velop new therapeutic approaches to overcome Art resistance in the the early proinflammatory response in myeloid cells in vitro and damp- malarial parasite. Likewise, increasing effectiveness of Art-based ens the inflammatory response in vivo in P. berghei-infected mice. The therapies may improve the clinical outcome of CM, the most se- dual activity of CR-1-31B as an antiplasmodial and as an inhibitor of the vere and most difficult-to-treat complication of malaria. inflammatory response in myeloid cells should prove extremely valu- Recent studies have pointed to initiation of protein synthesis as able for therapeutic intervention in human cases of CM. a novel pharmacological target in P. falciparum for the development of effective antimalarial drugs (25–27). Considering the Plasmodium | cerebral malaria | severe malaria | protein translation | dual target Significance alaria is a severe threat to global health, with an estimated Severe complications of malaria, such as anemia and cerebral 200–300 million cases annually and 445,000 deaths in 2016; M malaria (CM) (neuroinflammation), are responsible for ∼25% of it accounts for ∼25% of pediatric deaths in endemic areas of infant mortality in certain regions of Africa. Notwithstanding Africa (1–3). Malaria is also a significant threat to immunologically current effective antiplasmodial treatment, drug resistance is on naive travelers visiting or working in these areas (4). It is caused by the increase in the Plasmodium parasites, and therapy to treat infection with members of the Plasmodium parasite family, with CM is nonexistent. Here, we show that rocaglates derived from a clinical symptoms occurring at the blood stage, when Plasmodium natural product originally identified from plants of the Aglaia merozoites rapidly replicate and lyse erythrocytes (RBCs), leading to species, effectively block blood-stage parasite replication in anemia and high fever (2, 3). Furthermore, parasitized RBCs several mouse models and in infected human RBCs. Importantly, (pRBCs) can adhere to the brain microvasculature, causing local neurological inflammation is mitigated and survival is promoted inflammatory and vascular tissue damage, which, in turn, leads to in experimental CM, highlighting the strong potential of this cerebralmalaria(CM),aseveresyndromethatcanrapidlyevolveto class of compounds to treat complicated malaria. fatal encephalomyopathy and is mainly caused by Plasmodium falciparum (5, 6). Despite antimalarial treatment, CM has a high fatality Author contributions: D.L., J.P., and P.G. designed research; D.L., R.C., N.M., J.M.K., K.A., rate (18–40%) and up to one-third of survivors are afflicted with L.E.B., I.C., M.J.T., M.S., K.C.K., and J.A.P. performed research; D.L., M.S., and J.A.P. con- tributed new reagents/analytic tools; D.L., R.C., N.M., J.M.K., K.A., K.C.K., J.P., and P.G. neurocognitive deficits (7). The malaria problem is further compli- analyzed data; and D.L., K.C.K., J.A.P., J.P., and P.G. wrote the paper. – cated by widespread parasite resistance to antimalarial drugs (8 12), Conflict of interest statement: A provisional patent application based on the technology resistance of the Anopheles vector to insecticides (13), and the ab- described in this work has been filed. sence of an effective antimalaria vaccine despite enormous research This article is a PNAS Direct Submission. D.A.F. is a guest editor invited by the Editorial Board. efforts (14). Published under the PNAS license. Treatment of clinical malaria (uncomplicated disease) relies 1To whom correspondence may be addressed. Email: [email protected] or philippe. entirely on antimalarial drugs, with artemisinin (Art) being the [email protected]. most effective drug currently available (8). To avoid emergence of This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. Art resistance, Art monotherapy is discouraged and has been 1073/pnas.1713000115/-/DCSupplemental. replaced by Art combination therapy (ACT). ACTs are the cur- Published online February 20, 2018.

E2366–E2375 | PNAS | vol. 115 | no. 10 www.pnas.org/cgi/doi/10.1073/pnas.1713000115 Downloaded by guest on October 2, 2021 functional and structural conservation of components of the expressed as recombinant proteins and purified from Escherichia PNAS PLUS translation initiation machinery, we reasoned that known mam- coli (Fig. 1C), and their ability to interact with [32P]-labeled RNA malian translation initiation inhibitors may display activity against was quantified in the presence and absence of two rocaglates: their Plasmodium counterparts, including antimalarial activity in silvestrol and the hydroxamate CR-1-31B (Fig. 1 D and E). Both vivo. One target of this pathway that has been extensively scruti- eIF4A proteins showed similar behavior, including low-level nized for small-molecule discovery is eukaryotic initiation factor RNA binding under control conditions, which was significantly 4F (eIF4F), a heterotrimeric complex required for ribosome re- stimulated by silvestrol and CR-1-31B (approximately six- to 10- cruitment to capped (m7GpppX; X is any nucleotide) mRNAs fold increase). Clamping of eIF4A onto RNA, especially when in (28). eIF4F is composed of eIF4E, the cap-binding subunit; the context of 5′ leaders, has the potential to block ribosome eIF4A, a DEAD-box RNA helicase required to remodel the scanning and could be one contributor to the inhibition of mRNA 5′ leader region; and eIF4G, responsible for recruiting translation imparted by these compounds. We directly tested the the small ribosomal subunit (and associated factors) to the effect of CR-1-31B on parasite protein synthesis by metabolic mRNA (28). Rocaglates, the most characterized and potent labeling using [35S]-methionine incorporation in PbA pRBCs small molecules found to target eIF4F (29), are a class of natural (Fig. 1 F and G). In these experiments, cycloheximide (CHX) products derived from plants of the Aglaia genus (30) that in- was used as a positive control, while the antimalarial drug terfere with eIF4A activity in two ways. Rocaglates cause eIF4A chloroquine (CQ), which does not target translation, was used as to clamp onto RNA, which, if it occurs in the mRNA 5′ leader a negative control (Fig. 1 F and G). The effect of CR-1-31B on region, may act as a steric barrier to the initiation process, and protein synthesis was monitored directedly by SDS/PAGE of they deplete the eIF4F complex of its eIF4A helicase subunit P. berghei-labeled cell extracts and quantified by scintillation (31–34). The net result is reduced initiation complex formation. counting. CR-1-31B could inhibit P. berghei protein synthesis Since eIF4F discriminates between mRNAs during the initiation potently with an IC50 in the low nanomolar range, proving to be a process, rocaglates, unlike general inhibitors of elongation, exert better inhibitor than CHX (Fig. 1 F and G). In addition, we have a selective inhibitory effect on a subset of translating mRNAs measured the level of P. falciparum lactate dehydrogenase (31–34). They display tolerable safety profiles in vivo and exhibit (pLDH; the terminal enzyme of their glycolytic pathway) pro- anticancer properties in tumor xenograft models (31, 35). In duced in in vitro cultures exposed to increasing concentrations of addition, certain rocaglates can interfere with nuclear factor of CR-1-31B or CHX. The level of pLDH is potently reduced by κ activated T cells (NFAT) and nuclear factor-kappa B (NF- B) CR-1-31B with an IC50 of ∼3 nM, compared with an ∼300 nM signaling pathways in T cells in vitro (36, 37), suggesting po- IC50 for CHX in both 3D7 and ItG P. falciparum isolates (Fig. tential antiinflammatory properties. S3). Altogether, these initial observations strongly suggest that In the present study, we have investigated the possible anti- rocaglates can efficiently interfere with PfeIF4A activity and plasmodial activity of these compounds in vivo in mouse models of associated protein synthesis in a manner similar to their dem- blood-stage malaria (Plasmodium chabaudi) and experimental ce- onstrated action on mammalian eIF4A. rebral malaria [ECM; Plasmodium berghei ANKA (PbA)], as well as against multiple drug-resistant P. falciparum lines in vitro. We Antiplasmodial Activity of Rocaglates in a Mouse Model of CM. The show that rocaglates are potent antimalarial drugs, demonstrating potential antiplasmodial activity of rocaglates was assessed in

both direct antiplasmodial and antiinflammatory activities in vivo. vivo in a murine model of ECM induced by infection with PbA. MICROBIOLOGY Single dosing is sufficient to protect mice against ECM induced by In this model, mice infected with PbA develop blood-stage para- PbA.Ourdatasuggeststhatthiseffect is a consequence of roca- sitemia detectable within 2 d. PbA-parasitized erythrocytes become glates suppressing both the activity of Plasmodium falciparum trappedinthebrainmicrovasculature, leading to a pathological eIF4A (PfeIF4A) and the excessive host inflammatory response. inflammatory response in situ and appearance of drastic cerebral symptoms (paralysis, seizure, and coma), with a clinical end point Results reached by day 5–6 postinfection. Cerebral disease typically de- Rocaglates Interact with PfeIF4A. We investigated translation ini- velops with levels of blood parasitemia <10%. Using a prophylactic tiation as a potential new target for antimalarial drug discovery. treatment regimen where mice received daily dosing for 10 d We noted a very strong sequence conservation (70% identity, (starting at day −2 and ending at day +8), we observed that 83% total similarity) between mammalian eIF4A1 and PfeIF4A treatment with the structurally related rocaglates CR-1-31B that extended beyond the conserved signature motifs that define (0.2 mg/kg) and SDS-1-021 (0.05 mg/kg) (Fig. 2A)almost DEAD-box helicase family members (Fig. S1). Of note, human completely protected mice from ECM, with 80–90% of and mouse parasites have near-complete sequence identity treated mice surviving beyond day 15, while untreated mice conservation (Fig. S1). succumbed by day 6. Blood parasitemia monitored at days We asked whether this strong structural similarity translated to 3 and 6 postinfection showed a significant reduction of Plas- functional conservation, including interaction with small-molecule modium-infected RBCs (pRBCs) in both the CR-1-31B– and modulators of mammalian eIF4A. Members of the rocaglates SDS-1-021–treated groups (Fig. 2B). These data indicate that family, characterized by a common cyclopenta[b]benzofuran skel- prophylactic treatment of mice with rocaglates can prevent eton (Fig. 1A), have been shown to be among the most potent and significantly reduce mortality in the ECM model. inhibitors of eIF4A, and structure–activity relationships have We next tested the capacity of rocaglates to act therapeutically identified highly potent translation inhibitors from within this family and cure an established PbA infection. For this, treatment of (33, 35). CR-1-31B and (−)-SDS-1-021 act by increasing clamping PbA-infected mice with CR-1-31B (0.2 mg/kg) was performed of eIF4A onto RNA with a concomitant decrease in eIF4F- daily starting at various times postinfection (day 1, day 3, and day associated eIF4A, ultimately leading to impaired 43S ribosome 4) and continuing to day 8 with disease outcome monitored. CR- loading onto mRNA (31–35) (Fig. 1B). For both molecules, only 1-31B could significantly improve survival from PbA, even when the (−) enantiomers are active; the molecules used in the current treatment was initiated later in the course of infection up to day study were prepared as single, bioactive enantiomers using our +4 (Fig. 2C). Similar results were obtained using SDS-1-021 (Fig. previously published methods (38, 39) [Fig. S2; >95% pure com- S4). With both SDS-1-021 and CR-1-31B, the protective effect of pounds, as confirmed by ultraperformance liquid chromatography rocaglates was correlated with length of treatment and total dose (UPLC)/evaporative light scattering detection (ELSD) analysis]. received (Fig. 2C and Fig. S4) and was associated with a signif- We tested whether PfeIF4A could also be targeted by roca- icant reduction in blood parasitemia (measured at days 3, 6, and glates. Both murine eIF4A1 (meIF4A1) and PfeIF4A were 8 postinfection; Fig. 2D). Interestingly, in CR-1-31B–treated

Langlais et al. PNAS | vol. 115 | no. 10 | E2367 Downloaded by guest on October 2, 2021 A B

(-)-SDS-1-021 CR-1-31B

C D 20000 ** E 10000 ** kDa 245 180 16000 135 7500 ** * 100 12000 75 5000 63 8000 48 P-labelled RNA P-labelled

P-labelled RNA P-labelled 2500 (Retained cpms) 32

4000 (Retained cpms) 32 35 0 0 25 meIF4A1 -++--- meIF4A1 -++--- 20 PfeIF4A ----++ PfeIF4A ----++ DMSO ++-++- DMSO ++-++- Silvestrol (20µM) --+--+ CR-1-31B (20µM) --+--+

F PbA-infected uninfected G nM CHX nM CQ nM CR-1-31B nM CR-1-31B CHX 4 2 3 4 3 4 4 3 2 2 3 2 10 10 10 1 10 0 10 10 0 10 1 10 1 10 0 10 1 10 10 0 10 10 10 10 CR-1-31B kDa 1.0 180 - 135 - 100 - 75 -

63 - S]-Met incorporation 0.5 35 48 -

35 - 25 - 0.0 20 - [ Relative 110102 103 104 [nM]

Fig. 1. Rocaglates stabilize the interaction between PfeIF4A and mRNA. (A) Chemical structure of rocaglates used in this study: silvestrol, CR-1-31B, and (−)-SDS-1-021. (B) Mechanism by which rocaglates inhibit translation initiation (stabilization of the eIF4A/RNA complex leads to depletion of the eIF4A helicase from the eIF4F complex and eIF4A clamped onto mRNA, which may interfere with ribosome scanning); components of relevant proteins of the translation initiation machinery are identified. AUG, translation initiation codon. (C) Analysis of murine eIF4A1 (meIF4A1) and PfeIF4A recombinant proteins by gel electrophoresis and Coomassie blue staining. (D and E) Effect of silvestrol and CR-1-31B on binding and retention of purified meIF4A1 and PfeIF4A recombinant proteins to radiolabeled RNA assessed by filter-binding assay; significance was determined using an unpaired two-tailed Student’s t test for the indicated pairs (*P < 0.05; **P < 0.01). (F and G) Incorporation of [35S]-methionine to assess protein synthesis in PbA-infected blood and to test the effect of increasing doses of CR-1-31B, CHX, or CQ. (F) One sample per treatment condition was resolved on SDS/PAGE, and radiolabeled proteins were revealed by autoradiography; noninfected blood was included as a labeling control. (G) Experiment was performed twice in triplicate, and pooled data are presented as mean ± SD of the [35S]-methionine incorporation as counts relative to vehicle-treated parasites.

animals, parasitemia measured at day 8 was lower than at day 6, growth (measured at day 5 postinfection). These data suggest indicating that continued treatment with CR-1-31B caused a that therapeutic intervention with rocaglates can prevent and reduction of parasitemia from the peak measured at day 6 (Fig. significantly reduce mortality in the ECM model. 2D). Finally, the therapeutic effect of CR-1-31B was found to be dose-dependent: Animals dosed daily (starting at day 3 and Antiplasmodial Activity of Rocaglates Against Blood-Stage Parasites. continuing to day 8) with CR-1-31B at either 0.1 or 0.2 mg/kg We conducted additional experiments to further validate the were protected against lethal PbA-induced CM, while animals effect of the rocaglate CR-1-31B on blood-stage replication of receiving a lower daily dose of 0.02 mg/kg were as susceptible as the malarial parasite. First, we monitored the course of blood untreated controls (Fig. 2E). Parasitemia levels were also af- parasitemia in a mouse mutant that does not develop CM fol- fected in a dose-dependent manner (Fig. 2F and Fig. S5), with lowing PbA infection due to an inactivating mutation in the the lower 0.02-mg/kg dose having no effect, while 0.1- and Themis gene, which causes impairment of T lymphocyte-driven 0.2-mg/kg dosing caused a dose-dependent decrease in parasite neuroinflammation (40). Control untreated Themis mice infected

E2368 | www.pnas.org/cgi/doi/10.1073/pnas.1713000115 Langlais et al. Downloaded by guest on October 2, 2021 A B PNAS PLUS 100 *** 20 *** 15 Ctl (n=7) Veh (n=11) 50 10 CR-1-31B (n=11)

Survival (%) (-)-SDS-1-021 (n=11) 5 **

% parasitemia % parasitemia (± SD) *** 0 -2 0 5 10 15 3 6 Days post-infection Days post-infection C D 25 100 ***

*** 20 Veh (n=12) 15 * CR-1-31B d+1 (n=7) *** 50 CR-1-31B d+3 (n=14) *** * *** 10 CR-1-31B d+4 (n=7) Survival (%) Survival *** ** Parasitemia (%) 5

0 0 0 5 10 15 3 6 3 6 8 3 6 8 Days post-infection Veh d+1 d+3 Days post-infection

E *** F 100 15 ** (%)

10 Veh (n=7) 50 CR-1-31B 0.02mg/kg (n=7) *

parasitemia 5 Survival (%) CR-1-31B 0.1mg/kg (n=7) *** CR-1-31B 0.2mg/kg (n=7) Day 5 Day 0 0 0 5 10 15 Days post-infection CR-1-31B

[mg/kg] MICROBIOLOGY

Fig. 2. Rocaglates protect mice from CM. Mice infected with PbA were left either untreated or treated with different daily i.p. regimens (triangles) of CR-1- 31B or(−)-SDS-1-021, and the extent of parasite replication in blood (percentage of infected erythrocytes measured at the indicated days postinfection), appearance of neurological symptoms (CM phase, shaded area), and overall survival (percentage of surviving mice, total number of mice tested indicated) were monitored. (A and B) Effect of prophylactic treatment regimen with CR-1-31B and (−)-SDS-1-021 on PbA infection parameters (survival and blood-stage parasitemia). (C and D) Effect of different therapeutic treatment regimens (initiated at different days postinfection and given i.p.) with CR-1-31B on PbA infection parameters. (E and F) Effect of different doses of CR-1-31B (0.02, 0.1, and 0.2 mg/kg) in a therapeutic regimen initiated at day 3 on PbA infection parameters. Survival data were compared using the Mantel–Cox test, and significance was established relative to the vehicle (Veh)-treated mice; parasitemia values were compared by ANOVA, followed by a Bonferroni posttest (indicated level of significance: *P < 0.05; **P < 0.01; ***P < 0.001). Ctrl, control.

with PbA do not develop cerebral symptoms and survive past day 6 Antiplasmodial Activity of Rocaglates Against Human P. falciparum (compared with B6 controls) (Fig. 3A), but they do sustain pro- Parasites. In dose–response studies with human erythrocytes in- gressive blood-stage parasite replication, leading to very high levels fected in vitro, we also monitored CR-1-31B activity against the of blood parasitemia by day 15 (>40%). Treatment of Themis mice human P. falciparum malarial parasite (Fig. 4 A and B), using with CR-1-31B (0.2 mg/kg) starting at either day 3 or day 5 post- previously described experimental conditions (42). In these ex- infection and continuing to day 11 caused a significant reduction in periments, we used a number of different P. falciparum isolates, blood parasitemia compared with untreated mice (Fig. 3B). Sec- including four freshly isolated from patients seen at the Uni- ond, we investigated the effect of CR-1-31B on replication of P. versity of Toronto Health Network (UHN1–4), standard labo- chabaudi AS (PcA). PcA is a blood-stage parasite that does not ratory lines (3D7, ItG, CS2, and FCR3), and isolates from induceCM(41),withinfectedmicerathersuccumbingtohyper- Thailand that display low levels of CQ and mefloquine resistance parasitemia and associated severe anemia by day 10 (Fig. 3 C and (MS-97 lines). Infected blood cultures were exposed to various D). Treatment of PcA-infected mice with CR-1-31B (0.2 mg/kg, concentrations of CR-1-31B or CQ for 48 h, and the effect of standard 4-d test) fully protected animals against lethality from drug exposure on parasite growth inhibition was monitored and PcA infection. CR-1-31B–induced protection was associated with quantified (IC50) using a fluorescence-based assay (SYBR-Green a highly significant reduction in blood parasitemia measured daily I) (Fig. 4 A and B). CR-1-31B showed strong activity against all between days 3 and 6 postinfection (Fig. 3D), an effect that was P. falciparum lines tested, with IC50 values in the low nanomolar comparable to protection afforded by treatment with high-dose range, varying between 0.9 and 2.8 nM, whereas CQ had an IC50 artesunate (1.5 mg/kg) (Fig. 3D). These results strongly support average of ∼100 nM (Fig. 4C). Together, these results demon- a direct antiplasmodial effect for CR-1-31B in vivo against mu- strate that CR-1-31B acts to block replication of human malarial rine Plasmodium parasites that induce blood-stage infection or parasites, arguing for a possible therapeutic value of CR-1-31B cerebral disease. in clinical management of malaria.

Langlais et al. PNAS | vol. 115 | no. 10 | E2369 Downloaded by guest on October 2, 2021 A B *** 15 100 * B6 Veh (n=7) 10 Themis Veh (n=7) 50 * Themis d+3 (n=7) 5

Themis d+5 (n=7) Parasitemia (%) Survival (%) Survival ***

0 0 0 5 10 15 357357357 Days post-infection Veh d+3 d+5 Days post-infection C D *** 80 100

60 Veh (n=5)

50 CR-1-31B 0.2mg/kg (n=5) 40 Art 1.5mg/kg (n=5) Survival (%) Survival Parasitemia (%) 20 *** *** *** 0 0 051015 3 4 5 6 Days post-infection Days post-infection

Fig. 3. Rocaglate CR-1-31B blocks blood-stage replication of different Plasmodium malarial parasites. (A and B) Immunodeficient ThemisI23N mutant mice (Themis) do not develop CM symptoms following PbA infection (as shown in A), and were used to assess the effect of different therapeutic i.p. regimens of CR-1-31B (initiated at day 3 and day 5; green and red triangles, respectively) on blood-stage parasite replication. (C and D) Effect of CR-1-31B (0.2 mg/kg i.p.) on parameters of infection with the blood-stage parasite PcA were determined and compared with that of Art (1.5 mg/kg i.p.) using a standard 4-d test (57) and PcA-suceptible A/J mice. Statistical analyses were done as described in the legend for Fig. 2 and compared drug-treated groups with groups receiving vehicle (Veh) only (*P < 0.05; ***P < 0.001).

Effect of CR-1-31B on Host Inflammatory Response in Vivo. It is highly remained largely unstained, a feature that is indicative of an intact desirable for new antimalarials to be effective once clinical BBB and is similar to what is observed in the ECM-resistant symptoms have started and, ideally, to be efficacious at a single mouse mutant, Irf8R294C (43) (Fig. 5 D and E). By flow cytom- dosing. Therefore, we tested whether treatment with CR-1-31B etry analysis, we observed reduced cellular infiltration of leuko- + late during infection could blunt the development of disease and cytes (Fig. S6A), of both CD11b myeloid cells (Fig. 6 A and B) + cure infected animals. We observed that treatment of PbA- and TCRb lymphoid cells (Fig. 6 D and E and Fig. S6 B and C), infected mice with CR-1-31B starting at day 5 (when the first in the brain of CR-1-31B–treated mice compared with control neurological symptoms appear) and continuing until day 8 caused PbA-infected untreated mice. Reduced infiltration of inflamma- significant protection against ECM in these mice (Fig. 5A). Fur- tory cells in the brain coincided with reduced RNA expression of thermore, we observed that even a single dosing of CR-1-31B at genes encoding for the myeloid (Cebpd and C4b) and for the either 0.2 or 0.5 mg/kg administered on day 5 had beneficial ef- lymphoid cell lineages (Gzmb, Cd3g,andIfng) (Fig. 6 C and F and fects on disease and, at the higher dose, caused a significant in- Fig. S6D). Hence, ECM resistance in mice treated with CR-1-31B crease in overall survival of infected mice (Fig. 5B). This correlates with reduced infiltration of leukocytes in the brain, re- intervention is similar to what is needed in human cases of CM, duced expression of proinflammatory genes in situ, and preser- where high mortality is associated with rapidly evolving and irre- vation of BBB integrity. versible neurological symptoms. Finally, we note that the observed therapeutic effect of a single dose of CR-1-31B given at day Effect of CR-1-31B on Host Inflammatory Response in Vitro. Finally, 5 occurs in the absence of a strong reduction in parasite load and we directly tested the effect of different doses of CR-1-31B on associated blood parasitemia measured at day 5 or 6 (Fig. 5C). activation of isolated primary myeloid cells in response to IFN-γ, Excessive inflammatory response in situ is a major pathological which was monitored both by induction of the master proin- feature of ECM. This response is mediated, in part, by infiltration flammatory transcription factor IRF1 and by production of early of activated leukocytes in the brain, with concomitant production inflammatory markers (GBP2 and CXCL10) (Fig. 7). The protein of proinflammatory cytokines. Cells of the myeloid lineage play a level of STAT1, another proinflammatory transcription factor critical role in this pathological response, and mouse mutants downstream of IFN signaling that forms transcription complexes harboring reduced numbers or impaired activity of these cells with members of the IRF and ETS families to activate transcrip- confer resistance in the ECM model (43, 44). The observation that tion in response to proinflammatory stimuli (45), was not affected very late treatment with CR-1-31B can protect PbA-infected mice by CR-1-31B exposure (Fig. 7A). On the other hand, exposure of against CM without affecting blood parasitemia (Fig. 5C)ledus bone marrow-derived macrophages (BMDMs) to CR-1-31B sup- to test the possibility that CR-1-31B may also have an impact on pressed production of IRF1 (Fig. 7), a major transcriptional acti- the host inflammatory response in situ in the brain. Neuro- vator mediating myeloid cell activation in response to type I and inflammation in ECM is associated with breakdown of the blood– type II interferons, as well as other viral and bacterial stimuli. This brain barrier (BBB). An Evans Blue dye extravasation assay was was associated with reduced expression of protein markers of ac- used to assess BBB integrity in PbA-infected mice (Fig. 5 D and tivation in these cells (GBP2) and production of the critical early E). As opposed to the brains of PbA-infected B6 controls that inflammatory chemokine CXCL10 (Fig. 7). The effect of CR-1- cannot exclude the blue dye, brains from CR-1-31B-treated mice 31B on inflammatory protein production was dose-dependent and

E2370 | www.pnas.org/cgi/doi/10.1073/pnas.1713000115 Langlais et al. Downloaded by guest on October 2, 2021 A stable to impair ribosome scanning if they occur in the mRNA 5′ PNAS PLUS leader regions and if they deplete eIF4A from the eIF4F com- 125 UHN1 plex, resulting in decreased initiation events (31–34). Impor- CS2 tantly, systemic suppression of eIF4F is well-tolerated at the 100 UHN2 organismal level, increasing the therapeutic potential of roca- FCR3 glates (31, 46, 47). Recent studies have suggested that translation 75 UHN3 MS97.10 is a potential target for antimalarial drug discovery (26). Indeed, 50 UHN4 2,6-disubstituted quinoline-4-carboxamide analogs show strong MS97.78 activity against different life forms of the human P. falciparum 25 MS97.08 parasite, targeting translation elongation factor 2 (PfEF2) (26). ITG Considering the high degree of structural and functional con- Relative Fluorescence (%) Fluorescence Relative 3D7 0 servation among proteins involved in the translation machinery, -4 -3 -2 -1 0 we investigated the possibility that rocaglates may (i) interact Log[CR-1-31B] (µM) with PfeIF4A and (ii) display antimalarial activity in in vivo and ex vivo models of infection. B We obtained strong experimental evidence indicating that the 125 UHN1 rocaglate CR-1-31B has direct antimalarial activity. First, we CS2 demonstrate that CR-1-31B can stabilize complexes formed in 100 UHN2 vitro between recombinant PfeIF4A and RNA, and this in a FCR3 manner similar to mammalian eIF4A1 tested under the same 75 UHN3 conditions. Second, Plasmodium protein synthesis is inhibited by MS97.10 low nanomolar concentrations of CR-1-31B in a 1-h radioactive 50 UHN4 MS97.78 metabolic labeling assay and in a 24-h pLDH assay. Third, CR-1- 31B can block replication of P. falciparum in human erythrocytes 25 MS97.08 ITG infected ex vivo, with IC50 values in the low nanomolar range. Relative Fluorescence (%) Fluorescence Relative 3D7 0 Importantly, CR-1-31B shows similar IC50 values for P. falcipa- -4 -3 -2 -1 0 rum clinical isolates from different regions of the world that display diverse degrees of resistance to established antimalarial Log[CQ] (µM) drugs. Fourth, when tested in prophylactic and therapeutic C protocols, CR-1-31B protected mice against lethal CM caused by 300 CR-1-31B infection with P. berghei. In all cases, the antimalarial effect of CQ CR-1-31B is dose- and exposure-dependent and is associated 200 with a significant decrease in blood parasitemia in treated mice. Fifth, CR-1-31B is also active against P. chabaudi, a strict blood- stage Plasmodium parasite that replicates rapidly in blood but

100 does not cause CM. Finally, it is important to note that a single MICROBIOLOGY dose of CR-1-31B administered to P. berghei-infected mice at day IC50 (nM) 5, a time point at which control untreated infected mice display 5 early neurological symptoms (lethargy), is sufficient to significantly improve survival and BBB integrity. Both single-dose cure and novel modes of action with no cross-resistance to current 0 drugs are two key selection criteria in the global search for novel antimalarial drugs (48, 49). Taken together, these studies establish that CR-1-31B has direct antiplasmodial activity by acting P. falciparum isolates on PfeIF4A, but we cannot formally exclude that CR-1-31B may also affect other targets. Fig. 4. Activity of CR-1-31B against different human P. falciparum isolates. The rocaglate CR-1-31B was previously shown to have good (A and B) Laboratory and field P. falciparum isolates were cultured in vitro in pharmacokinetic properties (35) that include high solubility and human erythrocytes for 48 h in the presence of increasing doses – – membrane permeability at physiological pH. Incubation of CR- (0.1 100 nM) of CR-1-31B or CQ (1 500 nM). Cultures were stained with 1-31B in human plasma showed that it was 82–84% bound to SYBR-Green to quantify P. falciparum genomic DNA, and the results (representative of two biological replicates) are shown as relative fluorescence plasma proteins and that it was highly stable (100% remaining compared with the values measured at the lowest tested drug concentra- after 3 h). Experiments with murine and human liver microsomes tion. (C) Calculated IC values for CR-1-31B and CQ for all P. falciparum demonstrated that CR-1-31B has a significant hepatic stability 50 > isolates tested are shown as a histogram (average IC50 ± SD from two in- ( 95% after 1 h). In mice receiving repeated daily injections at dependent experiments). 0.2 mg/kg, no observable toxic effects were detected and CR-1- 31B did not display toxicity toward cells of hematopoietic origin (35). Altogether, these preliminary CR-1-31B pharmacological was not due to transcriptional inhibition. These results suggest that and safety data suggest possible use in humans for CM. rocaglates can also dampen host inflammatory function in myeloid Our studies extend those of Bhattacharya et al. (50) in doc- cells, with a beneficial impact on neuroinflammation that drives umenting an antiinflammatory response by rocaglates. Impor- pathogenesis in the ECM model. tantly, we demonstrate that CR-1-31B, and related rocaglates, can be used to interdict this process in the brain, highlighting its po- Discussion tential use as an antimalarial drug for intervention in CM in hu- Rocaglates (flavaglines) are a class of natural products that we mans. An antiinflammatory effect of CR-1-31B to dampen have previously characterized as novel anticancer drugs (29). neuroinflammation was initially suggested by results from single- These compounds act as potent inhibitors of translation initia- dosing experiments, where dramatic improvements in survival of tion by targeting eIF4A in two ways. They cause clamping of free PbA-infected CR-1-31B–treated mice (i) occurred in the absence eIF4A onto RNA, resulting in complexes that may be sufficiently of a substantial reduction in blood parasitemia (Fig. 5), (ii)were

Langlais et al. PNAS | vol. 115 | no. 10 | E2371 Downloaded by guest on October 2, 2021 A B Veh (n=7) Veh (n=14) 100 100 CR-1-31B d+5 (n=7) d+5 0.2mg/kg (n=7)

d+5 0.5mg/kg (n=9) *** 50 50 *** Survival (%) Survival (%) Survival **

0 0 0 5 10 15 0 5 10 15 Days post-infection Days post-infection C D E ** 16 100 ns NI * * 12 80 Veh 8 * 60

CR-1-31B 40

Parasitemia (%) 4

Evan's blue dye blue dye Evan's 20 (µg/g of brain tissue) brain of (µg/g 0 IRF8R294C 5 65656 0 Veh d+5 d+5 0.2mg/kg 0.5mg/kg

Fig. 5. CR-1-31B–mediated protection against neuroinflammation in late-stage CM. (A) Effect of a therapeutic regimen with CR-1-31B (0.2 mg/kg; four daily i.p. injections) initiated at day 5 post-PbA infection on survival (significance determined using the Mantel–Cox log-rank test: ***P < 0.001). Veh, vehicle. (B and C) Effect of late-stage, single high doses of CR-1-31B (0.2 mg/kg and 0.5 mg/kg) given i.p. at day 5 on infection parameters in PbA-infected mice. The effects on survival (B) and on blood-stage parasitemia (C) are shown. Parasitemia levels were compared using ANOVA with a Bonferroni posttest. (D) Integrity of the BBB in normal mice (NI) and PbA-infected mice treated with Veh or CR-1-31B (single 0.5-mg/kg dose, i.p., day 5) was assessed by Evan’sBlueex- travasation assay. The CM-resistant Irf8R294C mutant mice (43) were included as positive controls. (E) Evan’s Blue dye accumulation in the brain was quantified and expressed as micrograms per gram of brain tissue. Data are presented as mean ± SD, and groups were compared using an unpaired Student’s t test (*P < 0.05; **P < 0.01; ***P < 0.001).

associated with decreased infiltration of myeloid and lymphoid tempting to speculate that inhibition of eIF4A by CR-1-31B could inflammatory cells in the brain and maintenance of the BBB in- likewise dampen early proinflammatory responses such as type I IFN, tegrity (Figs. 5 and 6), and (iii) led to reduced cerebral expression thereby benefiting the host under situations that arise such as CM, of molecular markers of proinflammatory cells and inflammatory where pathogenesis is driven by acute inflammation. Indeed, we and mediators (i.e., Cebpd, Cd3g, Ifng, C4b, Gzmb) at the peak of others have previously reported rapid activation of the IFN response neuroinflammation (Fig. 6 and Fig. S6D). Finally, we showed that in situ in the brain of PbA-infected mice, and have shown that in- exposure of myeloid cells to CR-1-31B causes a significant activation of Irf7 and other genes that are part of this response (IFN- dampening of the inflammatory response of these cells following stimulated genes: Usp15, Trim25, Stat1, Irf1, Irf3, Irf8,andSocs1/Socs3) exposure to IFN-γ, as demonstrated by reduced protein expression causes resistance to CM (43, 55, 56). These findings are consistent of the master transcriptional regulator IRF1 and the inflammatory with the proposal that inhibition of translation initiation in general, chemokine CXCL10 by these cells (Fig. 7). These findings and of eIF4A in particular, dampens the inflammatory response. establish a rapid and beneficial antiinflammatory activity of Our studies also tease out an important functional property of CR-1-31B during PbA-induced experimental CM and associated rocaglates in Plasmodium-infected mice: the ability to act as dual- neuroinflammation. targeting agents displaying strong antimicrobial activity against this A possible explanation of these unexpected findings is that CR- parasite while also dampening the acute inflammatory response in 1-31B–mediated inhibition of mammalian eIF4A/eIF4F prefer- infected brains of the host. The end result is a potent perturbation entially inhibits translation of mRNAs encoding proinflammatory of parasite replication and pathogenesis at low nanomolar con- effectors. There is a growing body of literature suggesting that centrations. These properties, documented in preclinical models early activation of proinflammatory pathways involves regulation of blood-stage and cerebral disease, make CR-1-31B a high-profile at the level of translation, including proteins that are themselves candidate for intervention in CM in humans. The efficacy of CR- key transcriptional activators of the inflammatory response (51– 1-31B and other rocaglates against human malaria will require 53). For example, Colina et al. (54) showed that translational future testing in clinical studies either alone or in combination control is critical for expression of early defenses, including type I with Art, which is the current standard of care for CM. IFN. They observed increased resistance to a number of viral in- − − fections in 4E-BP1/2 / mice, two redundant translational re- Materials and Methods pressors of eIF4F, and that this was associated with increased Mice and Murine Plasmodium Parasites. Eight-week-old female C57BL/6 mice production of type I IFN. They further demonstrated that 4E-BP1/ were purchased from the Charles River Laboratories. Mouse mutant strains carrying loss of function at either Irf8 (Irf8R294C; Irf8m/m) (43) or Themis 2 inactivation caused increased translation of the type I IFN (ThemisI23N) (40) are on a C57BL/6 background and are resistant to CM as transcriptional activator IRF7 from existing RNA pools in plas- previously described. All mice were housed at McGill University Animal Care macytoid dendritic cells. This establishes that the translational Center according to the guidelines of the Canadian Council on Animal Care. repressors 4E-BP1/2 act as negative regulators of type I IFN and All protocols were approved by the Animal Care Committee of McGill Uni- that the translation of IRF7 is regulated by eIF4F (54). It is versity (Protocol 5287), and include procedures to minimize distress and

E2372 | www.pnas.org/cgi/doi/10.1073/pnas.1713000115 Langlais et al. Downloaded by guest on October 2, 2021 PNAS PLUS ns A 750 B NI Veh CR-1-31-B C Cebpd 0.52 * 16 14 cells

+ 500 12 10 8 *** 250 6 4 SSC Total CD11b Relative expr. to NI to expr. Relative 2 0 0

CD11b

ns D E F Gzmb 1250 * * NI Veh CR-1-31-B 2500 1000 2000 cells

+ 750 1500

500 1000 * Total TCRb Relative expr. to NI to expr. Relative 250 SSC 500

0 0

TCRb

Fig. 6. CR-1-31B–mediated protection against neuroinflammation is associated with reduced infiltration of immune cells. PbA-infected mice were perfused, and the effect of CR-1-31B on the total number of infiltrating cells, both CD11b+ myeloid cells (A) and TCRb+ lymphoid cells (D), was determined by FACS analysis. Scatter plots for primary FACS data are shown for CD11b+ myeloid cells (B) and lymphoid cells (E) obtained from noninfected (NI) and PbA-infected mice treated i.p. with vehicle (Veh) or CR-1-31B. P values were calculated by ANOVA, followed by a Bonferroni posttest. (C and F) Relative mRNA expression (expr.) for representative myeloid (Cebpd) and lymphoid (Gzmb) cell markers in the brain of CR-1-31B–treated and control PbA-infected mice, as determined by RT-qPCR. Data were normalized to Hprt, used as an internal control, and are expressed relative to noninfected controls (P values were calculated using an unpaired Student’s t test: *P < 0.05; ***P < 0.001).

improve welfare. An isolate of PcA was provided by Mary M. Stevenson, were dissolved in DMSO solution before being used to prepare serial twofold MICROBIOLOGY McGill University Health Center Research Institute, and was maintained by dilutions across a 96-well plate. Parasite culture material was then added to weekly passage in A/J mice for a maximum of four consecutive passages, as give a final hematocrit of 2% and a parasitemia of 2%, and the plate was then

previously described (41). An isolate of PbA was initially obtained from the incubated for 24 h at 37 °C in an atmosphere of 95% N2,3%CO2,and2%O2.

Malaria Reference and Research Reagent Resource Center (MR4); PbA was The IC50 values of CR-1-31B and CHX were determined using a four-variable passaged in C57BL/6J (B6) mice for a maximum of three consecutive passages nonlinear regression analysis using GraphPad Prism v7.0 and represent the (57). In passage mice, blood parasitemia was monitored by calculating the mean ± SEM (n = 4). percentage of pRBCs on thin blood smears stained with Diff-Quick (Dade Behring), as described (41). Infected blood was diluted in pyrogen-free saline Protein Sequence Alignment. Human (NP_001407.1) and murine (NP_659207.1) for infection of experimental animals. eIF4A1 protein sequences were retrieved from the National Center for Bio- technology Information Refseq database. The murine eIF4A1 protein sequence Plasmodium Human Parasites. Several different P. falciparum isolates were was used as input in the PlasmoDB BLASTp tool to identify similar proteins in used to test the antimalarial activity of CR-1-31B. These included four freshly Plasmodium parasites. Putative PfeIF4A proteins were identified in all human isolated clones from patients seen at The University of Toronto Health Network and murine parasites, having 70% sequence identity and 83–84% sequence of the University of Toronto (UHN1–4). We also tested standard laboratory lines similarity. Plasmodium protein sequences were downloaded from PlasmoDB (62) 3D7, ItG, CS2, and FCR3, as well as recent isolates from Thailand showing a low for P. falciparum 3D7 (PF3D7_1468700), Plasmodium vivax Sal-1 (PVX_117030), level of CQ and mefloquine resistance (MS-97 isolates). Plasmodium knowlesi strain H (PKNH_1212500), PbA (PBANKA_1331900), and All P. falciparum clones were mycoplasma-free and were maintained in P. chabaudi AS (PCHAS_1336500). Multiple sequence alignment was performed continuous culture with fresh O+ erythrocytes from healthy blood donors as using Clustal Omega (63) and was visualized with Jalview v.2 (64) (Fig. S1). previously described (58). Ring-form cultures were synchronized with 5% sorbitol treatment twice over two intraerythrocytic cycles (59). To assess the Infections. Mice were infected i.v. (0.2 mL) with 107 P. chabaudi pRBCs or 106 antimalarial activity of the drugs tested, we used a SYBR-Green method to PbA into the tail vein and monitored under biosafety level 2 containment quantify parasite genomic DNA (60). Briefly, a 96-well plate with a series of twofold dilutions was produced containing R-10G medium plus drug and R- conditions. At predetermined times postinfection, blood was sampled, and > 10G medium alone (last well). The parasite culture, 1% hematocrit and 2% blood parasitemia was determined (five fields of 100 erythrocytes were parasitemia, was added to each well, and the plates were incubated at 37 °C counted per mouse, and results are expressed as the percentage of parasitized erythrocytes) following staining of thin blood smears with Diff-Quick. and gassed (93% N2, 5% CO2, and 2% O2) for 48 h. Relative fluorescence was assessed using a Biotek Synergy-4 Multi-Mode Microplate Reader (Biotek P. berghei-infected mice were similarly monitored for blood parasitemia, but Instruments, Inc.) and a Fluostar Optima plate reader (BMG Labtech). The also for the appearance of neurological symptoms (shivering, tremors, ruf-

IC50 values of CR-31 or CQ for individual isolates were determined using a fled fur, seizures, and paralysis), and were euthanized when reaching clinical four-variable nonlinear regression analysis of the relative fluorescence of or experimental end points. Survival curves were compared using the individual wells following the addition of SYBR-Green I using GraphPad Mantel–Cox log-rank test. Prism v7.0 software. In vitro P. falciparum protein inhibition was performed by using a Plas- Plasmodium Metabolic Labeling. Irf8m/m mice were infected i.p. with PbA modium-specific LDH assay (61) after a 24-h incubation period and de- pRBCs and monitored for parasitemia. When the blood parasitemia reached termining the absorbance at 595 nm using a plate reader. Briefly, compounds 10–15%, blood and was recovered by cardiac puncture in heparinized tubes.

Langlais et al. PNAS | vol. 115 | no. 10 | E2373 Downloaded by guest on October 2, 2021 For drug treatments, CR-1-31B and (−)-SDS-1-021 were solubilized in 100% CR-1-31B [nM] 0 0 20 50 100 100 A DMSO and freshly diluted in sterile PBS to the required concentration with IFNg - ++++ - 10% DMSO. Vehicle-treated mice received PBS containing 10% DMSO only.

3hrs stimulation The antimalarial activity of CR-1-31B and (−)-SDS-1-021 was tested according IRF1 to experimental protocols we have previously described (57).

STAT1 Evans Blue Dye Extravasation. At day 6 postinfection with PbA, 24 h after receiving a single dose (0.5 mg/kg) of CR-1-31B, groups of mice were injected ACTIN i.v. with 0.2 mL of 1% Evans Blue dye (E2129; Sigma) prepared in sterile PBS. The dye was allowed to circulate for 1 h, and mice were then exsanguinated by cardiac puncture and perfused with PBS containing 2 mM EDTA. Brains CXCL10 6hrs stimulation were excised, weighed, imaged, and incubated with 1 mL of dimethyl formamide for 48 h to extract Evan’s Blue dye from the tissues. The optical GBP2 density of 50 μL of dimethyl formamide that received brain Evan’s Blue dye was measured at 620 nm, and measurements were converted into micro- STAT1 grams of dye extravasated per gram of tissue by interpolation of a standard curve from serial dilutions of Evan’s Blue dye in dimethyl formamide. ACTIN Leukocyte Infiltration in the Brain. Six days postinfection with PbA, groups of mice that received single-dose treatment on day 5 were exsanguinated and B IRF1 CXCL10 perfused as above. Brains were harvested; Dounce-dissociated; and ho- 30000 9000 mogenized in RPMI media containing 0.5 mg/mL collagenase D (Gibco Life Technologies), 0.01 mg/mL DNase I (Roche), and 2 mM EDTA. Infiltrating cells

20000 6000 were separated on a 34% Percoll solution. Cells were counted and stained for flow cytometry analyses with the following: Zombie Aqua Dye-V500 (BioLegend), anti–CD4-peridinin chlorophyll protein complex (PerCP)-Cyanine5.5 (clone RM4- 10000 3000 5; eBioscience), anti–CD8α-eFluor450 (clone 53-6.7; eBioscience), anti–TCR- β–phycoerythrin (PE) (clone H57-597; eBioscience), anti–B220-FITC (RA3-6B2;

Relative protein level (AU) level protein Relative eBioscience), CD69-BV605 (H1.2F3; BD Bioscience), CD44-PE-Cy7 (clone IM7; 0 0 eBioscience), anti–CD11b-eFluor450 (clone M1/70; eBioscience), anti–CD11c- IFNg -++++- -++++- PerCP-Cyanine5.5 (clone N418; eBioscience), and anti–Ly6G-FITC (clone 1A8; CR-1-31B [nM] 0 0 20 50 100 100 0 0 20 50 100 100 BioLegend). Flow cytometry was performed on a BD LSRFortessa (BD Bioscience) Fig. 7. CR-1-31B blocks IFN-γ (IFNg)–induced production of inflammatory and analyzed using FlowJo software v.10. A fraction of total brain cells, before proteins in myeloid cells. (A) Whole-cell extracts from wild-type BMDMs Percoll separation, were lysed, and total RNA was prepared using RNeasy treated or not treated with IFN-γ in combination with increasing doses of CR- columns (Qiagen). cDNA was synthesized with Moloney murine leukemia 1-31 for 3 or 6 h, were used to monitor expression of a series of diagnostic virus reverse transcriptase (Thermo Fisher Scientific). Myeloid and lymphoid proinflammatory proteins (IRF1, CXCL10, STAT1, and GBP2) by immuno- markers, as well as IFN-stimulated gene expression levels, were assessed by blotting, using actin as an internal control. (B) Quantification of IRF1 and qPCR using SsoAdvanced SYBR-Green Supermix (BioRad), and gene ex- CXCL10 protein expression from A [relative arbitrary units (AU), normalized pression was normalized to Hprt; oligonucleotide sequences were described to actin]. previously (55). Fold changes in gene expression were expressed relative to noninfected mice.

Blood was washed three times with 1 vol of cold methionine-free RPMI 1640 Preparation of BMDMs. BMDMs were differentiated from bone marrow iso- (ThermoFisher) and resuspended to 5% hematocrit in RPMI 1640 methionine- lated from the femur and tibia of 8- to 16-wk-old C57BL/6 mice, as described free medium containing L-glutamine (2 mM), 10% heat-inactivated charcoal- previously (45). Briefly, bone marrow cells were cultured in Dulbecco’s filtered FBS, 0.15 mM hypoxanthine, and 25 mM Hepes. Ninety-six–well plates modified Eagle’s medium (DMEM; Wisent) containing 10% heat-inactivated containing 5 μL of vehicle or an increasing concentration (from 1 nM to 10 μM) FBS, antibiotics, and 20% L cell-conditioned medium (LCCM). The cells were of CHX, CR-1-31B, or CQ were supplemented with 90 μLofPbA-infectedblood supplemented with 10% LCCM 4 d later, and at day 6, cells were harvested (or noninfected blood) and incubated at 37 °C for 30 min. Cells were then by gentle washing of the monolayer with PBS-citrate. Cells were plated in labeled by addition of [35S]-Met (150–225 μCi/mL; PerkinElmer), followed by tissue culture-grade six-well dishes in DMEM containing 10% FBS and 20% incubation for 30 min at 37 °C. At the end of the labeling procedure, plates LCCM, and were used the next day. were immediately put on ice and cold PBS was added. Plates were centrifuged for 10 min at 500 × g at 4 °C, supernatant was removed, and erythrocyte Molecular Biology. A codon-optimized variant of PfeIF4A (XM_001348793.1) pellets were lysed by addition of 40 μL of lysis buffer (1× PBS, 0.15% saponin, for increased bacterial expression was synthesized by Genscript and cloned and 0.02% sodium deoxycholate), followed by incubation for 15 min 4 °C with into pET15b utilizing NdeI and BamHI restriction sites. The resulting plasmid agitation. Twenty microliters of lysate was used for TCA precipitation, and was transfected in BL21(DE3) codon+ E. coli cells, and the overexpressed radioactivity was determined by scintillation counting. Also, 5 μLofonerep- protein was purified using nickel affinity (HiTrap IMAC FF; GE Healthcare), licate was analyzed on 10% SDS/PAGE; the gels were treated with En3Hance anion exchange (HiTrap Q HP; GE Healthcare), and gel filtration (Superdex (PerkinElmer), dried, and exposed to Kodak BioMax XAR Film for 36 h at 20 °C. 200 10/300 GL; GE Healthcare) chromatography. Fractions containing PfeIF4A were pooled and concentrated using Amicon Ultra centrifugal filter Drugs and Drug Treatments. CQ hydrochloride (Sigma) was prepared in PBS, units. The construct pET15b/eIF4AI for expression of murine eIF4AI has been while Art (artesunate; a gift from Dafra Pharmaceuticals, Turnout, Belgium) described previously (66). was prepared fresh in 5% sodium bicarbonate. Asymmetrical synthesis of CR- The ability of different compounds to affect binding and retention of 1-31B and (−)-SDS-1-021 using biomimetic kinetic resolution of chiral, race- eIF4A to RNA templates was determined using purified recombinant proteins mic aglain ketone precursors was performed as previously published (38, 65) and [32P]-labeled RNA templates as previously described (31). To monitor the followed by amide formation (39). Purity of the rocaglate derivatives effect of CR-1-31B on the proinflammatory response of primary macro- CMLD010513 [(−)-CR-1-31B] and CMLD011604 [(−)-SDS-1-021] was assessed phages, whole-cell protein extracts were prepared and analyzed by SDS/ using UPLC-MS-UV-ELSD analysis (Fig. S2). A Waters Acquity UPLC instrument PAGE. Immunoblotting was performed using primary antibodies directed (with a binary solvent manager, single quadrupole (SQ) mass spectrometer, against IRF1 (sc-640), STAT1 (sc-591), and GBP2 (sc-10588) (all from Santa and Waters 2996 photodiode array detector) and an ELSD instrument were Cruz Biotechnology); CXCL10 (R&D Systems); and B-Actin (EMD Millipore), used with an Acquity UPLC BEH C18 1.7-μm column. For the analysis, a sol- and it was revealed using IR680 or IR800-conjugated secondary antibodies vent gradient of 10–90% acetonitrile in water over 2 min, followed by 1 min (Li-Cor). Protein band intensities were measured using Image Studio Lite of 90% acetonitrile, was employed. software (Li-Cor) and normalized against B-Actin protein level.

E2374 | www.pnas.org/cgi/doi/10.1073/pnas.1713000115 Langlais et al. Downloaded by guest on October 2, 2021 ACKNOWLEDGMENTS. We thank Susan Gauthier, Geneviève Perreault, and Québec, the CIHR Neuroinflammation training program. J.P. is supported by PNAS PLUS Patrick Sénéchal for technical assistance. This work was supported by a research CIHR Research Grant FDN-148366. M.S. is supported by a CIHR Foundation grant. grant (to P.G.) from the Canadian Institutes of Health Research (CIHR) (Founda- J.A.P. is supported by NIH Grant R35 GM118173. Work at the Boston University tion Grant). J.P. and P.G. are supported by a James McGill Professorship salary Center for Molecular Discovery is supported by Grant R24 GM111625. K.C.K. was award. D.L. is supported by fellowships from the Fonds de recherche santé supported by a CIHR Foundation Grant and the Canada Research Chair program.

1. World Health Organization (2017) World Malaria Report 2017 (WHO, Geneva). 35. Rodrigo CM, Cencic R, Roche SP, Pelletier J, Porco JA (2012) Synthesis of rocaglamide 2. White NJ, et al. (2014) Malaria. Lancet 383:723–735. hydroxamates and related compounds as eukaryotic translation inhibitors: Synthetic 3. Cowman AF, Healer J, Marapana D, Marsh K (2016) Malaria: Biology and disease. Cell and biological studies. J Med Chem 55:558–562. 167:610–624. 36. Proksch P, et al. (2005) Rocaglamide derivatives are immunosuppressive phytochem- 4. Massad E, Behrens RH, Burattini MN, Coutinho FA (2009) Modeling the risk of malaria icals that target NF-AT activity in T cells. J Immunol 174:7075–7084. for travelers to areas with stable malaria transmission. Malar J 8:296. 37. Baumann B, et al. (2002) Rocaglamide derivatives are potent inhibitors of NF-kappa B 5. Mishra SK, Newton CRJC (2009) Diagnosis and management of the neurological activation in T-cells. J Biol Chem 277:44791–44800. complications of falciparum malaria. Nat Rev Neurol 5:189–198. 38. Stone SD, Lajkiewicz NJ, Whitesell L, Hilmy A, Porco JA, Jr (2015) Biomimetic kinetic 6. Hora R, Kapoor P, Thind KK, Mishra PC (2016) Cerebral malaria–Clinical manifesta- resolution: Highly enantio- and diastereoselective transfer hydrogenation of aglain tions and pathogenesis. Metab Brain Dis 31:225–237. ketones to access flavagline natural products. J Am Chem Soc 137:525–530. 7. Dorovini-Zis K, et al. (2011) The neuropathology of fatal cerebral malaria in malawian 39. Gerard B, Cencic R, Pelletier J, Porco JA, Jr (2007) Enantioselective synthesis of the children. Am J Pathol 178:2146–2158. complex rocaglate (-)-silvestrol. Angew Chem Int Ed Engl 46:7831–7834. 8. White NJ (2008) Qinghaosu (artemisinin): The price of success. Science 320:330–334. 40. Torre S, et al. (2015) THEMIS is required for pathogenesis of cerebral malaria and 9. Dondorp AM, et al. (2010) Artemisinin resistance: Current status and scenarios for protection against pulmonary tuberculosis. Infect Immun 83:759–768. containment. Nat Rev Microbiol 8:272–280. 41. Min-Oo G, Fortin A, Poulin J-F, Gros P (2010) Cysteamine, the molecule used to treat 10. Ashley EA, et al.; Tracking Resistance to Artemisinin Collaboration (TRAC) (2014) cystinosis, potentiates the antimalarial efficacy of artemisinin. Antimicrob Agents Spread of artemisinin resistance in Plasmodium falciparum malaria. N Engl J Med 371: Chemother 54:3262–3270. 411–423. 42. Min-Oo G, et al. (2010) Cysteamine, the natural metabolite of pantetheinase, shows 11. Amaratunga C, et al. (2016) Dihydroartemisinin-piperaquine resistance in Plasmo- specific activity against Plasmodium. Exp Parasitol 125:315–324. dium falciparum malaria in Cambodia: A multisite prospective cohort study. Lancet 43. Berghout J, et al. (2013) Irf8-regulated genomic responses drive pathological in- Infect Dis 16:357–365. flammation during cerebral malaria. PLoS Pathog 9:e1003491. γ 12. Krishna S, Kremsner PG (2013) Antidogmatic approaches to artemisinin resistance: 44. Amani V, et al. (2000) Involvement of IFN- receptor-medicated signaling in pathol- Reappraisal as treatment failure with artemisinin combination therapy. Trends ogy and anti-malarial immunity induced by Plasmodium berghei infection. Eur J – Parasitol 29:313–317. Immunol 30:1646 1655. 13. Ranson H, et al. (2011) Pyrethroid resistance in African anopheline mosquitoes: What 45. Langlais D, Barreiro LB, Gros P (2016) The macrophage IRF8/IRF1 regulome is required are the implications for malaria control? Trends Parasitol 27:91–98. for protection against infections and is associated with chronic inflammation. J Exp – 14. Long CA, Zavala F (2016) Malaria vaccines and human immune responses. Curr Opin Med 213:585 603. Microbiol 32:96–102. 46. Graff JR, et al. (2007) Therapeutic suppression of translation initiation factor eIF4E – 15. Sinclair D, Zani B, Donegan S, Olliaro P, Garner P (2009) Artemisinin-based combi- expression reduces tumor growth without toxicity. J Clin Invest 117:2638 2648. 47. Lin CJ, et al. (2012) Targeting synthetic lethal interactions between Myc and the eIF4F nation therapy for treating uncomplicated malaria. Cochrane Database Syst Rev, complex impedes tumorigenesis. Cell Rep 1:325–333. CD007483. 48. Wells TNC, Gutteridge WE (2011) Malaria: New medicines for its control and eradi- 16. Eastman RT, Fidock DA (2009) Artemisinin-based combination therapies: A vital tool cation. Neglected Diseases and Drug Discovery, eds Palmer MJ, Wells TNC (Royal in efforts to eliminate malaria. Nat Rev Microbiol 7:864–874. Society of Chemistry, Cambridge, UK), pp 1–32. 17. Kyu HH, Fernández E (2009) Artemisinin derivatives versus quinine for cerebral ma- 49. malERA Consultative Group on Drugs (2011) A research agenda for malaria eradica- laria in African children: A systematic review. Bull World Health Organ 87:896–904. tion: Drugs. PLoS Med 8:e1000402. 18. Dondorp AM, et al.; AQUAMAT group (2010) Artesunate versus quinine in the 50. Bhattacharya B, et al. (2016) Fine-tuning of macrophage activation using synthetic

treatment of severe falciparum malaria in African children (AQUAMAT): An open- MICROBIOLOGY rocaglate derivatives. Sci Rep 6:24409. label, randomised trial. Lancet 376:1647–1657. 51. Anderson P (2010) Post-transcriptional regulons coordinate the initiation and reso- 19. Dondorp A, Nosten F, Stepniewska K, Day N, White N; South East Asian Quinine lution of inflammation. Nat Rev Immunol 10:24–35. Artesunate Malaria Trial (SEAQUAMAT) group (2005) Artesunate versus quinine for 52. Hilliard A, et al. (2006) Translational regulation of autoimmune inflammation and treatment of severe falciparum malaria: A randomised trial. Lancet 366:717–725. lymphoma genesis by programmed cell death 4. J Immunol 177:8095–8102. 20. Cheeseman IH, et al. (2012) A major genome region underlying artemisinin resistance 53. Schott J, et al. (2014) Translational regulation of specific mRNAs controls feedback in malaria. Science 336:79–82. inhibition and survival during macrophage activation. PLoS Genet 10:e1004368. 21. Miotto O, et al. (2015) Genetic architecture of artemisinin-resistant Plasmodium fal- 54. Colina R, et al. (2008) Translational control of the innate immune response through ciparum. Nat Genet 47:226–234. IRF-7. Nature 452:323–328. 22. Ariey F, et al. (2014) A molecular marker of artemisinin-resistant Plasmodium falci- 55. Torre S, et al. (2017) USP15 regulates type I interferon response and is required for parum malaria. Nature 505:50–55. pathogenesis of neuroinflammation. Nat Immunol 18:54–63. 23. Straimer J, et al. (2015) K13-propeller mutations confer artemisinin resistance in 56. Edwards CL, et al. (2015) Spatiotemporal requirements for IRF7 in mediating type I – Plasmodium falciparum clinical isolates. Science 347:428 431. IFN-dependent susceptibility to blood-stage Plasmodium infection. Eur J Immunol 45: 24. Mok S, et al. (2015) Population transcriptomics of human malaria parasites reveals the 130–141. – mechanism of artemisinin resistance. Science 347:431 435. 57. Moradin N, et al. (2016) Cysteamine broadly improves the anti-plasmodial activity of 25. Ahyong V, et al. (2016) Identification of Plasmodium falciparum specific translation artemisinins against murine blood stage and cerebral malaria. Malar J 15:260. inhibitors from the MMV malaria box using a high throughput in vitro translation 58. Trager W, Jensen JB (1976) Human malaria parasites in continuous culture. Science screen. Malar J 15:173. 193:673–675. 26. Baragaña B, et al. (2015) A novel multiple-stage antimalarial agent that inhibits 59. Ayi K, Turrini F, Piga A, Arese P (2004) Enhanced phagocytosis of ring-parasitized – protein synthesis. Nature 522:315 320. mutant erythrocytes: A common mechanism that may explain protection against 27. Garg A, et al. (2015) Targeting protein translation, RNA splicing, and degradation by falciparum malaria in sickle trait and beta-thalassemia trait. Blood 104:3364–3371. morpholino-based conjugates in Plasmodium falciparum. Proc Natl Acad Sci USA 112: 60. Smilkstein M, Sriwilaijaroen N, Kelly JX, Wilairat P, Riscoe M (2004) Simple and in- 11935–11940. expensive fluorescence-based technique for high-throughput antimalarial drug 28. Hershey JWB, Sonenberg N, Mathews MB (2012) Principles of translational control: An screening. Antimicrob Agents Chemother 48:1803–1806. overview. Cold Spring Harb Perspect Biol 4:a011528. 61. Makler MT, et al. (1993) Parasite lactate dehydrogenase as an assay for Plasmodium 29. Chu J, Pelletier J (2015) Targeting the eIF4A RNA helicase as an anti-neoplastic ap- falciparum drug sensitivity. Am J Trop Med Hyg 48:739–741. proach. Biochim Biophys Acta 1849:781–791. 62. Aurrecoechea C, et al. (2009) PlasmoDB: A functional genomic database for malaria 30. Pan L, Woodard JL, Lucas DM, Fuchs JR, Kinghorn AD (2014) Rocaglamide, silvestrol parasites. Nucleic Acids Res 37:D539–D543. and structurally related bioactive compounds from Aglaia species. Nat Prod Rep 31: 63. Sievers F, et al. (2011) Fast, scalable generation of high-quality protein multiple se- 924–939. quence alignments using clustal omega. Mol Syst Biol 7:539. 31. Cencic R, et al. (2009) Antitumor activity and mechanism of action of the 64. Waterhouse AM, Procter JB, Martin DMA, Clamp M, Barton GJ (2009) Jalview version cyclopenta[b]benzofuran, silvestrol. PLoS One 4:e5223. 2–A multiple sequence alignment editor and analysis workbench. Bioinformatics 25: 32. Iwasaki S, Floor SN, Ingolia NT (2016) Rocaglates convert DEAD-box protein eIF4A into 1189–1191. a sequence-selective translational repressor. Nature 534:558–561. 65. Gerard B, Sangji S, O’Leary DJ, Porco JA, Jr (2006) Enantioselective photocycloaddition 33. Bordeleau M-E, et al. (2008) Therapeutic suppression of translation initiation modu- mediated by chiral Brønsted acids: Asymmetric synthesis of the rocaglamides. JAm lates chemosensitivity in a mouse lymphoma model. J Clin Invest 118:2651–2660. Chem Soc 128:7754–7755. 34. Chu J, et al. (2016) CRISPR-mediated drug-target validation reveals selective phar- 66. Lindqvist L, et al. (2008) Selective pharmacological targeting of a DEAD box RNA macological inhibition of the RNA helicase, eIF4A. Cell Rep 15:2340–2347. helicase. PLoS One 3:e1583.

Langlais et al. PNAS | vol. 115 | no. 10 | E2375 Downloaded by guest on October 2, 2021