A Virus Hosted in Malaria-Infected Blood Protects against T Cell-Mediated Inflammatory Diseases by Impairing DC Function in a Type I IFN-Dependent Manner.

Coinfections shape immunity and influence the development of inflammatory diseases, resulting in detrimental or beneficial outcome. Coinfections with concurrent Plasmodium species can alter malaria clinical evolution, and malaria infection itself can modulate autoimmune reactions. Yet, the underlying mechanisms remain ill defined. Here, we demonstrate that the protective effects of some rodent malaria strains on T cell-mediated inflammatory pathologies are due to an RNA virus cohosted in malaria-parasitized blood. We show that live and extracts of blood parasitized by Plasmodium berghei K173 or Plasmodium yoelii 17X YM, protect against P. berghei ANKA-induced experimental cerebral malaria (ECM) and myelin oligodendrocyte glycoprotein (MOG)/complete Freund's adjuvant (CFA)-induced experimental autoimmune encephalomyelitis (EAE), and that protection is associated with a strong type I interferon (IFN-I) signature. We detected the presence of the RNA virus lactate dehydrogenase-elevating virus (LDV) in the protective Plasmodium stabilates and we established that LDV infection alone was necessary and sufficient to recapitulate the protective effects on ECM and EAE. In ECM, protection resulted from an IFN-I-mediated reduction in the abundance of splenic conventional dendritic cell and impairment of their ability to produce interleukin (IL)-12p70, leading to a decrease in pathogenic CD4+ Th1 responses. In EAE, LDV infection induced IFN-I-mediated abrogation of IL-23, thereby preventing the differentiation of granulocyte-macrophage colony-stimulating factor (GM-CSF)-producing encephalitogenic CD4+ T cells. Our work identifies a virus cohosted in several Plasmodium stabilates across the community and deciphers its major consequences on the host immune system. More generally, our data emphasize the importance of considering contemporaneous infections for the understanding of malaria-associated and autoimmune diseases.IMPORTANCE Any infection modifies the host immune status, potentially ameliorating or aggravating the pathophysiology of a simultaneous inflammatory condition. In the course of investigating how malaria infection modulates the severity of contemporaneous inflammatory diseases, we identified a nonpathogenic mouse virus in stabilates of two widely used rodent parasite lines: Plasmodium berghei K173 and Plasmodium yoelii 17X YM. We established that the protective effects of these Plasmodium lines on cerebral malaria and multiple sclerosis are exclusively due to this virus. The virus induces a massive type I interferon (IFN-I) response and causes quantitative and qualitative defects in the ability of dendritic cells to promote pathogenic T cell responses. Beyond revealing a possible confounding factor in rodent malaria models, our work uncovers some bases by which a seemingly innocuous viral (co)infection profoundly changes the immunopathophysiology of inflammatory diseases.

B y extension of the "hygiene hypothesis" (1), the host immune system is "conditioned" by lifelong environmental encounters with micro-and macroorganisms (also known as the multibiome [2]), ranging from commensals to pathogens. In turn, this interplay is expected to influence, positively or negatively, the development and severity of inflammatory disorders, including allergy and autoimmune diseases (3).
With respect to multiple sclerosis (MS), an autoimmune demyelinating disease of the central nervous system (CNS), neurotropic virus infections were reported to confer an added risk of developing the disease in adolescents and young adults (4,5). Although the mechanisms are still debated, it has been hypothesized that virus infections may lead to breakdown of self-tolerance in genetically predisposed individuals, in part by antigenic molecular mimicry and/or by bystander activation of autoreactive T cells via upregulation of major histocompatibility complex (MHC) and costimulatory molecules on dendritic cells (DC) (6). In experimental autoimmune encephalomyelitis (EAE), a mouse model of MS, virus-specific brain-resident T cells generated in early life following CNS virus infection are indeed able to promote the activity of autoreactive T cells, resulting in brain autoimmune attacks (7). In contrast to viruses, helminthic parasites are known to play a protective role in MS, most likely because (i) they promote tissue repair and (ii) they induce regulatory T and B cells as well as anti-inflammatory cytokines (e.g., interleukin 10 [IL-10] and transforming growth factor ␤ [TGF-␤]), which suppress pathogenic autoreactive granulocyte-macrophage colony-stimulating factor (GM-CSF) ϩ CD4 ϩ T cell responses (reviewed in references 8 to 10).
Beside extracellular helminths, intracellular parasites such as Plasmodium, the malaria-causing parasite, which affects almost one-half of the world's population, can also elicit immune modulatory pathways, which could affect autoimmune reactions. Following an asymptomatic liver stage, the parasites reach the bloodstream and develop within red blood cells (RBC). Blood-stage malaria can be associated with mild to severe clinical symptoms, including anemia, acute respiratory distress syndrome, acidosis, renal failure, or cerebral malaria (CM), which is characterized by the sequestration of leukocytes and parasitized RBC (pRBC) in brain microcapillaries, leading to hypoxia and vascular damage. Interestingly, most of the reported modulatory effects of Plasmodium, which can affect DC and B and T cells, occur during blood stage. Examples of this modulation on DC include the increased apoptosis of blood-circulating DC (11,12), an atypical/partial DC maturation profile (13), and a crippled ability to present antigens to CD4 ϩ and CD8 ϩ T cells (11,14,15). The impact of Plasmodium on DC may be direct, such as exposure to parasite effectors or by-products such as the heme crystal hemozoin (16), or indirect, such as the systemic activation by pattern recognition receptors such as Toll-like receptors (TLRs), which imprint a "refractory" state on DC (17), or by type I interferon (IFN-I), which impairs their Th1-promoting property (18). With regard to T cells, blood-stage malaria may cause T cell exhaustion, which can be restored by checkpoint inhibitor therapy (19). CD4 ϩ T follicular helper (Tfh) cells normally play a critical role in parasite control during blood stage, as they enhance the activation of germinal center B cell responses and enable long-lasting more-efficient humoral immunity (20,21). Yet during severe malaria, a strong Th1-polarizing environment promotes the development of dysfunctional T-bet ϩ "Th1-like" CD4 ϩ Tfh cells (22,23), which exhibit poor help activity on B cell responses and lead to B cell apoptosis or differentiation into short-lived plasma cells and atypical memory B cells (24).
While such immune modulatory processes are thought to partially underlie the poor naturally acquired immunity to malaria observed in areas of endemicity, they may also have a beneficial impact on the course of autoimmune disorders. More than half a century ago, the incidence of two autoimmune diseases, rheumatoid arthritis and systemic lupus erythematosus, was found to be up to 6 times less frequent in Nigerians than in Europeans, and it was proposed that parasitic infections, in particular, malaria, were responsible for alleviating the development of autoimmunity (25). In accordance, experimental infection with Plasmodium berghei suppressed the spontaneous development of renal disease in a mouse lupus model (26). Intriguingly, the prevalence and incidence of MS has increased following malaria eradication in Sardinia (27), and work using rodent-adapted Plasmodium strains has revealed an overall protective effect of malaria infection on EAE. Infection with Plasmodium chabaudi chabaudi AS (PccAS) reduced EAE severity, possibly due to the induction of regulatory CD4 ϩ T cells (Treg) and the production of IL-10 and TGF-␤ (28). Moreover, transfer of DC incubated with extracts of P. berghei NK65 pRBC ameliorated EAE (29); although paradoxically, when induced in mice cured from that same parasite, EAE was aggravated (30). Currently, little is known about the molecular and cellular mechanisms by which Plasmodium infection influences CNS autoimmunity.
In addition, beside autoimmune contexts, the clinical evolution of malaria itself is influenced by coinfection with another Plasmodium species. In humans, the risk of developing symptomatic malaria seems to be lower in mixed P. falciparum/P. malariae or P. falciparum/P. vivax infections (31,32). In mice, the development of experimental cerebral malaria (ECM), a deadly vascular pathology during which Th1 CD4 ϩ T cells promote the sequestration of pathogenic CD8 ϩ T cells in the brain vasculature (33), is inhibited by coinfection with Plasmodium yoelii yoelii 17X clone 1.1 (34) and by P. berghei K173 (35). In the former case, replication of the ECM-causing P. berghei ANKA parasites was found to be hampered by mixed infection, but in the latter case, there was no effect on parasite growth. Rather, protection was associated with an early production of IFN-␥ and IL-10 cytokines at 24 h postinfection. Yet, the exact protective mechanisms remain ill defined.
Here, in order to elucidate the bases of Plasmodium-mediated protection against T cell-mediated inflammatory diseases, we have investigated rodent malaria strains that block ECM and EAE. We show that live blood, as well as blood extracts, parasitized by P. berghei K173 or P. yoelii 17X YM confers full protection against P. berghei ANKAinduced ECM and myelin oligodendrocyte glycoprotein (MOG)/complete Freund's adjuvant (CFA)-induced EAE and that this is associated with a strong IFN-I signature. We report the identification of an RNA virus called lactate dehydrogenase-elevating virus (LDV) in the protective Plasmodium stabilates, and we find that LDV infection alone recapitulates all the protective effects. In brief, LDV infection triggers a massive IFN-I response, which leads to a decrease in the number and functional capacity of conventional dendritic cells (cDC), thereby preventing the provision of signal 3 cytokines that are responsible for the pathogenic polarization of CD4 ϩ T cells in ECM and EAE.
(This article was submitted to an online preprint archive [36].)

P. berghei K173 infection does not cause ECM and elicits lower
Th1 responses than P. berghei ANKA. While the P. berghei ANKA isolate is widely used to induce ECM in C57BL/6 mice, the outcomes of P. berghei K173 infection vary. Some studies report its potential to induce ECM (37)(38)(39), while others suggest that it does not cause ECM (40) or that it even protects from this pathology (35). We first evaluated the immune responses and clinical outcomes following infections with our "in-house" P. berghei ANKA and P. berghei K173 stabilates. We infected mice intravenously (i.v.) with P. berghei ANKA-or P. berghei K173-infected pRBC and monitored the development of ECM and circulating parasitemia. Both parasites replicated in vivo (Fig. 1A), but only P. berghei ANKA led to ECM, characterized by brain edema and early death within 7 to 8 days of infection (Fig. 1B). At day 6 postinfection, the absence of ECM development upon P. berghei K173 infection was consistent with a lower number of brainsequestered CD4 ϩ and CD8 ϩ T cells than with P. berghei ANKA (Fig. 1C). Moreover, the percentage of brain CD4 ϩ (Fig. 1D) and CD8 ϩ T cells (Fig. 1E) producing IFN-␥, an essential cytokine in ECM pathogenesis (33,41,42), was significantly decreased upon P. berghei K173 infection compared to that with P. berghei ANKA. In the spleen, the percentage of CD4 ϩ T cells producing IFN-␥ upon pRBC restimulation was reduced in P. berghei K173-infected mice (Fig. 1F), but no difference was observed in the percentage of IFN-␥ ϩ CD8 ϩ T cells (Fig. 1G). These data suggest that the inability of P. berghei K173 to elicit ECM is related to an impaired differentiation of CD4 ϩ Th1 T cells in the spleen and a reduced sequestration of pathogenic CD8 ϩ T cells in brain microcapillaries. P. berghei K173 infection triggers an early type I IFN response. To investigate the molecular bases behind the lower Th1 responses in P. berghei K173-infected mice, we performed a kinetic analysis of several cytokines in the serum after infection. In accordance with the above findings, bioactive IL-12p70, a hallmark cytokine of Th1 responses, was induced exclusively by P. berghei ANKA but not by P. berghei K173 at day 3 postinfection (p.i.) ( Fig. 2A). Instead, P. berghei K173 triggered an early production of IFN-␣, CCL4, CCL5, tumor necrosis factor alpha (TNF-␣), IL-6, and IL-12p40 at day 2 ( Fig. 2A). In the spleen, P. berghei K173 infection induced a global activation of lymphocytes and DC, as exemplified by the upregulation of CD69 on T cells (Fig. 2B), B cells, and NK cells (not shown), and of CD86 on DC (Fig. 2C). These changes were suggestive of a systemic type I IFN (IFN-I) response (43). To confirm this hypothesis, we assessed whether they were dependent on IFN-I signaling. We analyzed the cytokine induction and T cell/DC activation in P. berghei K173-infected wild-type versus Ifnar1 FIG 1 Reduced Th1 responses and absence of ECM pathology following P. berghei K173 infection. C57BL/6 mice infected i.v. by injection of 10 6 P. berghei ANKA (Pb ANKA) or P. berghei K173 (Pb K173) pRBC. Blood circulating parasitemia (A) and ECM development (B) monitored following infection. Brain edema was visualized by Evans blue coloration. (C) Total numbers of CD4 ϩ or CD8 ϩ T cells collected from brain at day 6 after infection. Cells collected from brain (D and E) and spleen (F and G) at day 6 after infection were restimulated in vitro with MutuDC preloaded or not with pRBC. IFN-␥ production by CD4 ϩ T cells (D and F) or CD8 ϩ T cells (E and G) detected by intracellular staining. Percentages on the representative dot plots show the median percentages of IFN-␥ ϩ cells of total CD4 ϩ or CD8 ϩ T cells Ϯ IQRs. Bar graphs show the medians Ϯ IQRs of absolute numbers. Data are representative of 2 independent experiments. N ϭ 5 mice per group. knockout (KO) mice. Both the induction of IFN-␣, IFN-␤, TNF-␣, and IL-6 genes and the systemic activation of splenic T cells and DC were abrogated in Ifnar1 KO mice ( Fig. 2D to F). These data indicate that P. berghei K173-but not P. berghei ANKA-parasitized blood triggers a systemic IFN-I response, which correlates with defective Th1 differentiation and with ECM protection.
Plasmodium-parasitized blood protects against ECM and EAE independently from live parasites. We next investigated if upon P. berghei K173/P. berghei ANKA coinfection, P. berghei K173 had a "dominant" protective effect on ECM. Coinfected mice were indeed protected from ECM (see Fig. S1A in the supplemental material) and, consistent with an impaired development of Th1 responses, blood CD4 ϩ T cells displayed lower production of IFN-␥ (Fig. S1B). To assess if live parasites were required for protection, we sonicated the P. berghei K173 stabilate, which precluded the parasites from replicating in vivo (not shown). Killed P. berghei K173 extracts still conferred full protection against ECM (Fig. 3A), and protection correlated with an early systemic activation of splenic T cells and DC ( Fig. 3B and C). To examine if the protective potential of P. berghei K173 extracts could extend to other T cell-mediated inflammatory diseases, we evaluated their impact on experimental autoimmune encephalomyelitis (EAE), a murine model of multiple sclerosis. EAE was induced by immunization with an MHC II peptide from the myelin oligodendrocyte self-antigen (MOG 35-55 ) combined with CFA. Control mice injected with P. berghei ANKA blood extracts 1 day prior to MOG 35-55 /CFA immunization developed a classical EAE disease characterized by a progressive ascendant paralysis, but mice that received extracts of P. berghei K173 blood showed no sign of disease ( Fig. 3D and E). This indicates that inhibition of ECM and EAE by P. berghei K173-parasitized blood operates independently from live parasites.
It was previously reported that coinfection with certain clones of P. yoelii parasites can inhibit ECM (34) and that the P. yoelii 17X YM strain triggers an IFN-I response (44,45). Thus, we tested whether P. yoelii 17X YM, obtained here from the MR4/BEI distributor, would phenocopy P. berghei K173. Although P. yoelii 17X YM does not elicit ECM, it causes hyperparasitemia and is known to be lethal around day 7. To mitigate this confounding factor, we sonicated the P. yoelii 17X YM stabilate before coadministration with P. berghei ANKA. Akin to P. berghei K173 extracts, P. yoelii 17X YM extracts fully prevented the development of ECM (Fig. 3F) and induced T cell and DC activation at day 2 postinoculation (Fig. 3G and H). These data establish that at least two commonly used rodent malaria strains (P. berghei K173 and P. yoelii 17X YM) induce a systemic immune activation and protect from ECM. P. berghei K173 and P. yoelii 17X YM stabilates contain LDV virus, which alone, disrupts Th1 responses and protects from ECM. Intriguingly, we noticed that the protective activity of P. berghei K173 blood was contained in the plasma (see Fig. S2A) and was disrupted by UV treatment (Fig. S2B), two observations suggestive of the presence of a viral element. To verify this hypothesis, we subjected P. berghei ANKA and P. berghei K173 stabilates to comprehensive PCR rodent infectious agent (PRIA) testing. We found the presence of a virus, the lactate dehydrogenase-elevating virus (LDV), in P. berghei K173 (titer of 3,600 particles per 100 l plasma) and P. yoelii 17X YM (titer of 14,600 particles per 100 l plasma) stabilates, but not in P. berghei ANKA stabilates.
LDV is an enveloped, single-stranded positive RNA virus of the family Arteriviridae, order Nidovirales, known to cause persistent asymptomatic infection in mice, with long-lasting circulating viremia (46). This virus has been reported to modulate host responses, thereby either alleviating (47,48) or exacerbating (49) immune-mediated diseases. To assess the direct implication of LDV in ECM protection, we decontaminated P. berghei K173 stabilates by fluorescence-activated cell sorting (FACS) of erythrocytes. We verified by PRIA that these new P. berghei K173 pRBC stocks were LDV free. We also prepared parasite-free LDV-containing plasma by collecting the plasma from mice injected with P. berghei K173 sonicated blood extracts. P. berghei ANKA-infected mice coinjected with LDV-containing plasma were protected from ECM, while those coinfected with LDV-free P. berghei K173 were no longer protected (Fig. 4A). Strikingly, the LDV-free P. berghei K173 stabilate was now causing ECM in the majority of mice (Fig. 4A), and these effects were not due to faster parasite growth (Fig. 4B). Since we had observed that ECM protection was linked to a reduced Th1 response, we analyzed the activation and production of Th1 cytokines by parasite-specific CD4 ϩ T cells at day 6 postinfection. While there was no major difference in the percentages of activated (CD11a ϩ CD49d ϩ ) splenic CD4 ϩ T cells between the groups (Fig. 4C), activated CD4 ϩ T cells were unable to produce IFN-␥ and TNF-␣ upon restimulation with the I-A brestricted ETRAMP 272-288 antigenic peptide (Fig. 4D). This inhibition was also observed in the total CD4 ϩ T cell population upon anti-CD3 stimulation (Fig. 4E). The results demonstrate that LDV is necessary and sufficient for ECM protection, most likely by abrogating the Th1 effector activity of CD4 ϩ T cells.
LDV induces Ifnar1-dependent depletion and functional impairment of splenic cDC. To study how LDV impedes the Th1 CD4 ϩ T cell response, we analyzed splenic conventional DC (cDC), which are primarily responsible for Th1 priming and polarization in this context (50)(51)(52). Two days following P. berghei ANKA/LDV coinfection, we observed a strong reduction in the proportion and numbers of total CD11c ϩ MHC II ϩ cDC, which was not observed in Ifnar1 KO mice ( Fig. 5A to C). In proportion, the decrease was more pronounced for the CD8␣ ϩ cDC1 subset, but the CD172a ϩ cDC2 subset was also affected (Fig. 5D to F). In addition, we analyzed the capacity of the remaining cDC to make the pro-Th1 cytokine IL-12p70, which is composed of the IL-12p35 and IL-12p40 subunits. Upon day 2 of infection, P. berghei ANKA induced the upregulation of the IL-12p35 and IL-12p40 genes, which was blocked by LDV coinfection but rescued in Ifnar1 KO mice ( Fig. 5G and H). Logically, this resulted in the lack of bioactive IL-12p70 in the serum at day 3 postinfection, an effect that was reversed in Ifnar1 KO mice (Fig. 5I). Taken together, our data establish that IFN-I signaling induced by LDV infection causes a quantitative reduction in cDC and a functional impairment in their ability to polarize Th1 CD4 ϩ T cell responses.
Type I IFNs can affect multiple cell types, including CD4 ϩ T cells. Therefore, to determine to which extent the dysfunctional cDC compartment was responsible for LDV-mediated ECM protection, we tested if the transfer of functional cDC could restore disease in P. berghei ANKA/LDV coinfected mice. cDC were isolated from P. berghei ANKA-infected mice and transferred into LDV/P. berghei ANKA coinfected mice on days 3, 4, and 5 (Fig. 6A). This procedure restored ECM in only 3 of 10 transferred mice. We reasoned that even if fully functional at the time of injection, the transferred cDC may immediately be negatively conditioned by IFN-I in the LDV-infected environment.
To overcome this effect, we isolated "IFN-I-insensitive" cDC from Ifnar1 KO P. berghei ANKA-infected mice and transferred them into LDV/P. berghei ANKA coinfected wildtype (WT) mice. This procedure restored ECM in 8 of 10 mice (Fig. 6B), indicating that crippling of cDC by IFN-I signaling is the predominant mechanism that underlies the LDV-mediated protection against ECM.
LDV prevents EAE by disrupting the polarization of pathogenic autoreactive CD4 ؉ T cell responses. Finally, we investigated to which extent the protective mechanisms of P. berghei K173 extracts against EAE were analogous to those described in ECM. We first confirmed that LDV infection alone was protective against EAE. Mice injected with phosphate-buffered saline (PBS) or with sonicated extracts of LDV-free P. berghei K173 prior to MOG 35-55 /CFA immunization developed EAE, while mice that received LDV-containing plasma were fully protected ( Fig. 7A and B). Upon day 2 of MOG 35-55 /CFA immunization, the IL-12p40, p35, and p19 subunit genes were induced in cDC isolated from draining lymph nodes (dLN), but this upregulation was blocked by LDV infection in an interferon ␣/␤ receptor (IFNAR)-dependent manner (Fig. 7C to E). These findings show that IFN-I signaling blocks the production by cDC of IL-23 (composed of IL-12p40 and p19 subunits), a key cytokine in the polarization of  stimulation was lower in LDV-infected mice ( Fig. S3B and D). This decrease was confirmed by dosing the 3 cytokines in the supernatants of MOG 35-55 -stimulated spleen cells (Fig. S3E). In parallel to the impaired CD4 ϩ T cell functional polarization in the spleen and dLN, we observed a major drop in the percentages and numbers of CD4 ϩ T cells producing IFN-␥, IL-17, and GM-CSF in the brain (Fig. 8A) and spinal cord (Fig. 8B), in line with the absence of disease. In conclusion, our data show that LDV-mediated protection against EAE is mediated by an IFN-I signaling-dependent blockade of IL-23-dependent polarization of autoreactive GM-CSF-producing encephalitogenic CD4 ϩ T cells.

DISCUSSION
The host immune status is shaped by continuous exposure to commensal as well as potentially pathogenic microorganisms. Depending on whether they exacerbate or alleviate immunopathology, the surrounding microbes may be harmful or beneficial. In this study, we initially sought to address why malaria infection influences the clinical outcome of a concurrent Plasmodium infection and autoimmune reactions. To investigate these mechanisms, we took advantage of rodent Plasmodium lines that prevent the development of ECM and EAE, two inflammatory diseases mediated by T cells. Unexpectedly, we found that the beneficial effects of the protective Plasmodium berghei K173 strain are entirely due to LDV, a nonpathogenic mouse virus, and that this virus is cohosted in stabilates of other parasite strains (e.g. P. yoelii 17X YM) commonly used by the community. We demonstrated that protection against ECM results from an IFN-I signaling-dependent depletion and functional impairment of splenic DC, which become incapable of producing IL-12p70 and polarizing pathogenic Th1 CD4 ϩ T cell responses. Analogous to this mechanism, we found that the protection against EAE is linked to an IFN-I-dependent reduction in the production of encephalitogenic IL-23 and GM-CSF cytokines.

Malaria-Associated Virus Protects against ECM and EAE
® Based on the facts that (i) in our experiments, CD8 ϩ T cell accumulation in the CNS was reduced (see Fig. 1C and E) despite a stronger induction of parasite-specific CD8 ϩ T cell responses in the spleen (see Fig. 1G) and (ii) Th1 CD4 ϩ T cells are known to enhance CD8 ϩ T cell sequestration in the brain through the induction of CXCL9/10 (33), we hypothesize that the defective Th1 responses underlie ECM protection. However, because IL-12p70 may also heavily influence CD8 ϩ T cells, we do not rule out the possibility that dysfunctional cDC also directly cripple the pathogenic CD8 T cell response.
This work reiterates the importance of screening products that are injected into animal models for the presence of unsuspected contaminants. LDV was initially discovered as a companion virus of transplantable tumor cell lines more than half a century ago (54). Our data establish that the virus is carried over in commonly used rodent Plasmodium stocks. Although LDV should not be an issue for Plasmodium strains that are regularly passaged through mosquitoes, the results of some studies on innate immune sensing and/or IFN-I responses performed with blood-passaged Plasmodium strains may need to be reinterpreted if these strains are LDV positive. So far, the timing of the contamination of certain Plasmodium stocks by LDV and the causes remain unclear. As LDV can be, though very poorly, transmitted to cage mates (46,55), one may speculate that at times when the confinement of cages in animal facilities was less stringent, some Plasmodium-infected mice may have been exposed to LDVcontaminated bedding. Another possibility is that at some point, mice used to expand Plasmodium stocks have been injected with a biological product containing LDV. LDV is indeed a known contaminant of mouse-passaged ascitic fluids (56). In any case, we are confident that the contamination of strains used in this study occurred prior to their manipulation in our facility for various reasons: (i) for P. yoelii 17X YM, quantitative PCR (qPCR) detection of LDV was performed immediately on one-half of the obtained cryovials, before any mouse passage, (ii) all cages were kept isolated from each other on ventilated racks, (iii) for stock preparation, mice infected with different parasite strains were never mixed in the same cage, (iv) our "clean" Plasmodium stocks have remained LDV free in the course of this study, (v) the data on P. berghei K173 are fully consistent with earlier observations of ECM inhibition by this strain (35).
LDV is a mouse-specific virus that causes limited pathology but persists lifelong with detectable circulating viremia (46). Only one study from 1978 examined the outcome of LDV/malaria coinfection. It was observed that LDV exacerbates P. yoelii virulence by increasing parasitemia, leaving open the possibility that the phenotype of the lethal and nonlethal P. yoelii 17X substrains may be in part determined by the presence of LDV (57). Why, in this case, LDV aggravated the pathology remains unclear, but one could speculate that with malaria strains that do not cause cerebral malaria, the suppression of Th1 responses is deleterious for parasite control. Concerning autoimmunity, the beneficial role of LDV has been reported in various contexts such as lupus (58), diabetes (59), and EAE (47). In lupus, LDV-mediated protection was correlated with a decrease in IFN-␥ production in the serum (60), but no mechanism was provided to explain the suppression of diabetes and EAE.
Beyond the importance of our findings for the community working on malaria models, our work has fundamental relevance to understand how viruses shape inflammatory diseases. We indeed reveal some of the molecular bases that underlie the protective effects of LDV in two settings involving T cell-mediated immunopathology. In the case of ECM, protection results from the functional paralysis and the strong depletion of splenic cDC, both of which are caused by IFN-I signaling. It was previously shown that LDV induces a rapid systemic IFN-I production by pDC in a TLR7-dependent manner (61). IFN-I signaling has been reported to play a dual role on the IL-12 pathway. On the one hand, IFN-I drives DC activation and maturation (62) and production of IL-12 (63), which may augment Th1 CD4 responses. On the other hand, IFN-I can selectively suppress IL-12p70 (64). It is likely that IFN-I dosage determines the outcome of IFN-I signaling on Th1 responses (65). While low levels of IFN-I signaling positively regulate Th1 responses, higher doses of IFN-I, as observed here upon massive viral replication in the first 2 days, rather impede the development of Th1. In addition to the functional alterations of cDC, we reported a major drop in the number of splenic cDC that is dependent on IFN-I signaling. It will be interesting to analyze whether the reduction in cDC is due to apoptosis induced by BH3-only proteins, such as the proapoptotic factor Bim (66).
In the case of EAE, LDV infection prior to MOG/CFA immunization blunted the production of IL-12 and IL-23 by cDC in the dLN, resulting in the absence of IFN-␥, IL-17, and GM-CSF production by MOG-specific CD4 ϩ T cells, both in the periphery and in the CNS. It has now become clear that IL-12p35 (67), IFN-␥, and IL-17 (68) are dispensable for the development of MOG-induced EAE, while the key pathogenic cytokines are GM-CSF produced by CD4 ϩ Th cells and IL-23, which may be produced by multiple sources (68,69). Also, in contrast to ECM pathology in which cDC are required for the priming and polarization of Th1 CD4 ϩ T cells (50)(51)(52), cDC are not necessary for the priming of encephalitogenic CD4 ϩ Th cells in MOG protein-induced EAE (70), and mice lacking cDC even display an aggravated disease (71). Based on this, we propose that LDV-mediated protection against EAE is due to the global effect of IFN-I signaling, which inhibits IL-23 production by several different cell types of the dLN, including cDC, and that this prevents the differentiation of GM-CSF-secreting CD4 ϩ T cells. In human DC, IFN-␤ has been shown to inhibit IL-23p19 production by DC, possibly through a STAT3-dependent induction of the negative regulator SOCS3 (72). The exact transcriptional and/or epigenetic reprogramming that cripples the IL-12 and IL-23 synthesis pathways in response to IFN-I remains to be fully unraveled in this context.
Type I IFNs are powerful signals that can modify the functionality of many cell types, including CD4 ϩ T cells, which play a major role in the two diseases. For example, in EAE, IFN-I signaling in T cells can directly impair Th17 differentiation (73). During blood-stage malaria, CD4 ϩ T cell-intrinsic IFN-I signaling induces T-bet and Blimp-1 expression, thereby promoting IL-10 ϩ Tr1 responses in mice (74) as well as in humans (75). Therefore, it was possible that CD4 ϩ T cells impacted by IFN-I may contribute to protection. Because ECM pathology in LDV-infected animals could be restored by transferring activated "IFN-I-insensitive" (Ifnar1 KO) DC from P. berghei ANKA-infected mice, we infer that the role, if any, of non-DC in the protection phenotype should be modest.
There have been other examples of viruses modifying the clinical course of a parasitic disease. In experimental malaria, coinfection with the chikungunya arbovirus also prevented ECM but through a distinct mechanism involving the IFN-␥-mediated retention of CXCR3-expressing pathogenic CD8 ϩ T cells in the spleen (76). Another striking example is the discovery of a double-stranded RNA virus in Leishmania parasites (LRV), which is responsible for the exacerbation of the severity of mucocutaneous leishmaniasis via a TLR3/IFN-I-dependent proinflammatory response (77,78). A notable difference is that LRV is an endogenous viral element of the parasite itself, while LDV is an exogenous mouse-specific virus that is cohosted in Plasmodium-parasitized blood. Nevertheless, it was observed that the aggravation and metastasis of Leishmania lesions were not exclusively associated with LRV and that they could also be enhanced by coinfection with exogenous viruses (79). Interestingly, a new RNA virus called Matryoshka RNA virus 1 (MaRNAV-1) was recently identified in P. vivax-infected blood samples (80). Like LRV, which infects the Leishmania parasite, MaRNAV-1 is thought to be a virus of the Plasmodium parasite. Future investigations should reveal whether this virus modulates the pathogenicity of P. vivax, as P. vivax malaria is associated with lower severity than P. falciparum malaria (81) and was suggested to be protective in cases of coinfection (32).
In conclusion, our data emphasize the notion that viral infections, even with viruses that are seemingly innocuous, can have dramatic consequences on a concurrent infectious or autoimmune disease. LDV is a mouse-specific virus, but the IFN-I-mediated modulatory mechanisms highlighted here may be fully relevant in humans. As constant exposure to an immense variety of commensal and noncommensal microbes profoundly shapes our health, it appears essential to actively and systematically charac-terize environmental microbial agents. From a fundamental standpoint, one could foresee the discovery of new languages in the host-pathogen cross talk. From a medical perspective, this research could suggest novel therapeutic immune modulatory molecules. Parasites and experimental infections. Plasmodium berghei ANKA and Plasmodium berghei Kyberg 173 parasites were gifts from S. Picot (University of Lyon, France) and I. Landau (National Museum of Natural History, Paris, France), respectively. The identity of these parasites was confirmed by wholegenome sequencing. Plasmodium yoelii subs p. yoelii, strain YM (P. yoelii 17X YM, MRA-755) was obtained through BEI Resources, NIAID, NIH, in 2017, contributed by David Walliker. All Plasmodium strains were propagated in C57BL6/J mice. To prepare pRBC stocks, mice were bled with heparin, and the proportion of pRBC was evaluated by blood smear. The pRBC concentration was adjusted to 10 7 pRBC/ml in Alsever's solution with 10% glycerin. One-milliliter aliquots were stored at Ϫ80°C. All infections were conducted by intravenous inoculation of 10 6 pRBC. For Evans blue staining, mice were injected i.v. with a solution of 1% Evans blue dissolved in 0.9% NaCl, and brain coloration was examined 1 h after dissection. To prepare the sonicated blood extracts, total blood was sonicated using an ultrasonic liquid processor (Branson, Sonifier) under the following parameters: 40 mA, pulse 2, 1 min. Blood was sonicated twice with a 1-min rest. For UV treatment, plasma was exposed for 40 min.
LDV detection and dosage and decontamination of P. berghei K173 blood. For the initial detection of LDV, P. berghei ANKA and P. berghei K173 stabilates were subjected to a mouse/rat comprehensive clear panel for PCR infectious agent testing (PRIA; Charles River). Further detection and quantitation of LDV particles were conducted with a simple PCR LDV test (Charles River) on mouse plasma.
To free P. berghei K173 blood from the virus, C57BL6/J mice were injected with 10 6 P. berghei K173 pRBC; 10 days after injection, intracardiac blood was collected with heparin and washed with PBS. RBC and platelets were stained with anti-Ter119 (allophycocyanin [APC], REA847, 1/50; Miltenyi Biotec) and CD41 (PE, MWReg30, 1/300; BD Biosciences), and then RBC were separated from platelets by magnetic sorting using anti-PE microbeads (Miltenyi Biotec). The Ter119 ϩ CD41 Ϫ fraction was stained again with the same anti-Ter119 and CD41 antibodies, and RBC were FACS sorted with an Aria cell sorter (BD Biosciences). New B6 mice were injected with 10 6 sorted pRBC. Five days later, a new stock of pRBC was prepared as indicated above. Confirmation of the absence of LDV was performed by PCR (Charles River).
Induction of EAE and clinical score. The MOG  (MEVGWYRSPFSRVVHLYRNGK) peptide was purchased from Polypeptide Laboratories (San Diego, CA, USA). Mice were immunized subcutaneously with 50 g of MOG  peptide emulsified in complete Freund's adjuvant (CFA; BD Difco, Franklin Lakes, NJ, USA) containing 500 g of killed Mycobacterium tuberculosis (strain H37a; Difco). Mice were injected intravenously with 200 ng of pertussis toxin (List Biological Laboratories, Campbell, CA, USA) at days 0 and 2 postimmunization. Clinical scores were recorded daily by an experimenter blinded to the experimental groups. Scores were assigned as follows: 0, no sign of disease; 1, loss of tone in the tail; 2, hind limb paresis; 3, hind limb paralysis; 4, tetraplegia; 5, moribund.
Real-time quantitative PCR for gene expression. Total mRNA was extracted from tissue and cells by standard TRIzol-chloroform precipitation. For reverse transcription, the iScript cDNA synthesis kit (Bio-Rad) was used. Quantitative PCR (qPCR) reactions were prepared with LightCycler 480 DNA SYBR green master reaction mix (Roche Diagnostics). Primers were used at 0.2 M, and sequences are available upon request. qPCRs were run in duplicates on a LightCycler 480 System (Roche diagnostics), and cDNA abundance was normalized to the reference gene Hprt. Results are expressed as the threshold cycle difference (ΔC T ) (Hprt minus gene of interest).
Sorting and transfer of splenic dendritic cells. Spleens were injected with 1 ml of PBS supplemented with Liberase TL (0.125 mg/ml; Roche) and DNase I (100 g/ml; Sigma-Aldrich), incubated for 30 min at 37°C, and mashed through a 100-m cell strainer. RBC were lysed and cells washed in magnetically activated cell sorting (MACS) buffer. For DC sorting, cells were incubated 10 min with CD11c microbeads (Miltenyi Biotec), and CD11c ϩ cells were magnetically sorted according to the manufacturer's instructions. Postsorting flow cytometry analysis showed that less than 0.5% of transferred CD11c ϩ cells were CD8 ϩ CD3 ϩ T cells. After sorting, 10 6 CD11c ϩ cells in 100 l PBS were transferred to recipient mice by i.v. injection.
Statistical analyses. Statistical evaluation of differences between the experimental groups was done by using two-way analysis of variance followed by a Bonferroni's posttest for clinical monitoring and parasitemia, Kaplan-Meier test for monitoring of ECM-free mice, Mann-Whitney U test for group versus group comparison, and Kruskal-Wallis test for multiple-group comparisons. All tests were performed with GraphPad Prism 5.00 (GraphPad Software Inc., San Diego, CA, USA). All data are presented as medians Ϯ interquartile ranges (IQRs).

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.