Original Chemical Series of Pyrimidine Biosynthesis Inhibitors That Boost the Antiviral Interferon Response

ABSTRACT De novo pyrimidine biosynthesis is a key metabolic pathway involved in multiple biosynthetic processes. Here, we identified an original series of 3-(1H-indol-3-yl)-2,3-dihydro-4H-furo[3,2-c]chromen-4-one derivatives as a new class of pyrimidine biosynthesis inhibitors formed by two edge-fused polycyclic moieties. We show that identified compounds exhibit broad-spectrum antiviral activity and immunostimulatory properties, in line with recent reports linking de novo pyrimidine biosynthesis with innate defense mechanisms against viruses. Most importantly, we establish that pyrimidine deprivation can amplify the production of both type I and type III interferons by cells stimulated with retinoic acid-inducible gene 1 (RIG-I) ligands. Altogether, our results further expand the current panel of pyrimidine biosynthesis inhibitors and illustrate how the production of antiviral interferons is tightly coupled to this metabolic pathway. Functional and structural similarities between this new chemical series and dicoumarol, which was reported before to inhibit pyrimidine biosynthesis at the dihydroorotate dehydrogenase (DHODH) step, are discussed.

enzymatic step is catalyzed by the dihydroorotate dehydrogenase (DHODH), which is bound to the inner membrane of mitochondria, where it converts dihydroorotate (DHO) to orotate (3). Finally, the multifunctional UMP synthase uses orotate to produce UMP, a common precursor of all other pyrimidine nucleosides.
It has been recently shown that compounds inhibiting the de novo pyrimidine biosynthesis pathway exhibit potent broad-spectrum antiviral activity (4)(5)(6)(7)(8)(9)(10)(11). Indeed, several screening campaigns for antiviral molecules led to the identification of either CAD or DHODH inhibitors. Such molecules were found to efficiently block the in vitro replication of many viruses, including both DNA and RNA viruses. In the presence of pyrimidine biosynthesis inhibitors, cellular pools of pyrimidines collapse, and the lack of pyrimidine is usually considered to be directly responsible for the inhibition of viral growth. However, it was also reported that inhibiting de novo pyrimidine biosynthesis stimulates the innate immune response, in particular the transcription of some interferon-stimulated genes (ISGs) independently of interferons (IFNs) and the canonical JAK-STAT pathway (8,(12)(13)(14)(15)(16)(17)(18). In addition, the antiviral activity of pyrimidine biosynthesis inhibitors was found to be strictly dependent on cellular gene transcription and nuclear export machinery and required interferon regulatory factor 1 (IRF1), a key transcription factor driving the expression of antiviral genes, including ISGs (8). More recently, it was shown that pyrimidine biosynthesis inhibitors could increase the expression of retinoic acid-inducible gene 1 (RIG-I), a cytoplasmic sensor inducing the expression of innate immunity genes and IFNs in response to RNA virus infections (16). Altogether, these different reports support a key role of the innate immune response in the antiviral activity of compounds inhibiting the de novo pyrimidine biosynthesis pathway. However, the mechanisms linking the intracellular pool of pyrimidines to the innate immune response remain to be characterized.
Here, we describe a novel series of 3-(1H-indol-3-yl)-2,3-dihydro-4H-furo [3,2-c]chromen-4-one derivatives inhibiting de novo pyrimidine biosynthesis. The lead molecule from this series, called DD363, was isolated from a screening campaign that was previously described and aimed at identifying stimulators of antiviral genes (8). The phenotypic assay we used was based on human HEK-293T cells transiently transfected with a luciferase reporter gene controlled by five interferon-stimulated response elements (ISRE). This regulatory element is present in promoter sequences of ISGs, where it binds transcription factors activated in type I interferon-stimulated or virus-infected cells, such as STAT1/STAT2/IRF9 (ISGF3) or IRFs. It was therefore expected that any compound inducing the ISRE-luciferase construct would also stimulate the expression of endogenous ISGs and exhibit some potent broad-spectrum antiviral activity. This phenotypic assay was used to screen a total of 41,353 chemical compounds for their capacity to stimulate ISRE-luciferase expression. Two compounds from the chemical library of Institut Curie were finally selected for further studies, including DD264, which has already been described (8), and DD363, which is novel in terms of structure and activity. Most interestingly, a functional study of this chemical series led us to show for the first time that in cells transfected with RIG-I ligands mimicking a viral infection, the production of type I interferon (IFN-I) and IFN-III is strongly boosted when de novo pyrimidine biosynthesis is blocked. This new observation unravels a mechanism by which cells modulate their communication with neighboring cells as a function of their metabolic status.

RESULTS
DD363 is an ISRE-luciferase-inducing antiviral compound. DD363 was first selected in a chemical screen because, compared to other tested molecules, it significantly stimulated the expression of the ISRE-luciferase reporter plasmid transiently transfected in HEK-293T cells (6.2-fold induction at 56 M in the initial screen). DD363 is 3-(1H-indol-3-yl)-2,3-dihydro-4H-furo [3,2-c]chromen-4-one, which corresponds to a novel chemical series never reported before (Fig. 1A). It is different from DD264 that we previously identified in the same screen (8), although parts of the molecule exhibit similarities (Fig. 1A). To confirm DD363 activity, this compound was tested in a dose-response experiment on STING-37 cells, a reporter cell line derived from human HEK-293 cells by stable transfection of the ISRE-luciferase reporter gene. As shown in Fig. 1B, a 3-fold increase in luciferase expression was observed upon DD363 treatment. This is slightly inferior to the results obtained with DD264 but in a similar range. DD363 and DD264 are, however, weak inducers of the ISRE-luciferase reporter gene compared to recombinant beta interferon (IFN-␤) (Fig. 1B). Nevertheless, we decided to further characterize this molecule and evaluate its antiviral activity in secondary assays.
At first, HEK-293T cells were infected with a recombinant strain of measles virus (MV) expressing luciferase as a reporter of viral replication (MV-Luc) and then treated with increasing concentrations of DD363 for 24 h. As shown in Fig. 2A, DD363 efficiently inhibited MV growth, as assessed by decreased levels of luciferase activity in cultures. The half-maximal inhibitory concentration (IC 50 ) of DD363 (mean Ϯ standard deviation [SD]) was estimated to be 11 Ϯ 3 M in this assay (Table 1). The inhibition of MV-Luc was persistent at 48 h postinfection but was partially lost at 72 h and limited to the highest concentration of DD363, suggesting that the virus progressively overcame the antiviral effects (see Fig. S1A in the supplemental material). We also showed that DD363 at 30 M reduced by 1 log the production of infectious MV particles in cell cultures, as determined by virus titration at 48 h postinfection (Fig. 2B). Besides, this compound showed limited cytotoxicity at these concentrations ( Fig. 2A), as determined by quantification of the number of living cells in DD363 versus dimethyl sulfoxide (DMSO)treated wells after 24 h of culture. Under these conditions, the 50% cytotoxic concentration (CC 50 ) of the molecule (mean Ϯ SD) was estimated to 87 Ϯ 7 M. However, kinetic studies showed that DD363 essentially inhibited cellular proliferation rather than killed cells at concentrations below 60 M. Indeed, the number of viable cells increased less rapidly in DD363-treated cultures but did not collapse below the initial number of cells seeded at T ϭ 0 h (Fig. S1B). Furthermore, the fraction of dead cells in cultures did not increase from 0 to 72 h of culture, and it always remained below 15% in the presence or absence of DD363, as assessed by flow cytometry analysis (LIVE/ DEAD fixable green dead cell stain kit; data not shown). Then, DD363 antiviral activity was tested on two other RNA viruses: chikungunya virus (CHIKV) and human coronavirus 229E (HCoV-229E). HEK-293T cells were infected with CHIKV and treated with DD363 or DMSO alone for 24 h, and cells were immunostained for the expression CHIKV glycoprotein E2. As shown in Fig. 2C, DD363 efficiently suppressed CHIKV replication in these cells. Inhibition by DD363 was also confirmed by the infection of HEK-293T cells with a recombinant strain of CHIKV expressing Renilla luciferase as a reporter of viral growth ( Fig. 2D; mean Ϯ SD IC 50 ϭ 3.6 Ϯ 0.6 M) (19). Similarly, DD363 was found to inhibit HCoV-229E growth in human MRC5 cells (Fig. 2E). We also showed that DD363 reduced the cytopathic effects of HCoV-229E, as assessed by quantifying the number of viable cells in infected MRC5 cultures (Fig. 2F). Altogether, this demonstrated that DD363 is active against different negative-strand (MV) or positive-strand (CHIKV and HCoV-229E) RNA viruses, while it also inhibits cellular proliferation.
Structure-activity analysis of DD363 analogs. To characterize structural features that account for DD363 activity, we first decided to separate the two enantiomers from this racemic mixture, as they could differ in their antiviral potencies. Thus, enantiomers of DD363 were separated by chiral chromatography and tested on HEK-293T cells infected by MV-Luc. Interestingly, only one of the two enantiomers (ϩ) exhibited antiviral potency (Fig. 3). Altogether, this demonstrated that there are spatial requirements with respect to the connection between the indole moiety and the 2,3-dihydro-4H-furo[3,2-c]chromen-4-one.
To further identify chemical features accounting for DD363 potency and select for more potent molecules, a set of 17 structural analogs was synthesized (see supplemental material) (20,21). The antiviral potencies of these compounds were tested on HEK-293T cells infected by MV-Luc, and the corresponding IC 50 s are reported in Table  1 and Fig. 4A. Interestingly, the addition of a halogen group (DD776, DD771, and JP52), a methyl (DD778) group, or an ethyl group (JP49) at position R1 increased the antiviral potency (2-to 9-fold higher than for DD363; Table 1). As a control, DD778 was shown to inhibit not only MV ( Fig. 4B and Table 1) but also CHIKV ( Fig. 4C; IC 50 ϭ 1.4 Ϯ 0.2 M) replication, with an increased potency compared to DD363 (Fig. 2D). Then, highly potent analogs with a halogen group (DD771) or a methyl group (DD778) at position R1 were further modified at position R2, R3, R4, or R5 (Table 1). The addition of a methyl group at position R2 showed no major effect on the antiviral potency (DD775). In contrast, all compounds modified on the indole moiety were much less potent (JP10, JP25, DD783, and DD784), suggesting that this part of the molecule is essential (Table  1). We have also tested analogs of DD363 and DD771 without a nitro group on the furan ring ( Fig. 4A; JP42 and JP46, respectively). Although DD363 and JP42 showed comparable antiviral potencies, JP46 was much less potent than DD771 (9-fold decrease). Besides, opening the furan ring (JP39) led to some loss of the antiviral potency, supporting the role played by this part of the molecule (Fig. 4A). Other structural modifications were tested on 2,3-dihydro-4H-furo[3,2-c]chromen-4-one, showing no benefit (DD787) or abrogating the antiviral potency (DD785 and DD786; Fig. 4A). Furthermore, replacement of the indole moiety group by a 3-chlorophenyl group (DD311; previously described in reference 20), as in DD264 (8), greatly decreased the antiviral activity by comparison with DD363 (Fig. 4A). Altogether, this established structure-activity relationships for this chemical series, and analogs that are up to 9-fold more potent than the initial hit DD363 were identified. DD778 is an inhibitor of pyrimidine biosynthesis. As mentioned above, DD363 was selected from a screening campaign together with DD264, a compound subsequently characterized as an inhibitor of de novo pyrimidine biosynthesis (8). Although DD363 and its active analogs are chemically unrelated to DD264, they have similar antiviral properties and capacities to induce the ISRE-luciferase reporter gene. This suggested a similar mode of action. We thus determined whether DD778, one potent analog of DD363 showing a better selectivity index, could inhibit de novo pyrimidine biosynthesis like DD264.
To address this question, HEK-293T cells were treated with DD778 or DMSO alone, and nucleoside levels in cellular extracts were determined by high-performance liquid chromatography (HPLC) and spectrophotometry analysis. As shown in Fig. 5A, uridine (U) and cytidine (C) levels collapsed in DD778-treated cells, whereas purine levels (A and G) were unaffected. This demonstrated that DD778 interferes with pyrimidine metabolism and homeostasis. Of its multiple functions, uridine is also required to synthesize UDP-sugar complexes, which are essential to protein glycosylation. We thus determined whether DD778 could also alter intracellular levels of UDP-galactose (UDP-Gal) and UDP N-acetylgalactosamine (UDP-GalNac). UDP-Gal collapsed in DD778treated cells, as predicted. Surprisingly, UDP-GalNac was maintained at high levels, suggesting that specific mechanisms help secure intracellular concentrations of this metabolite (Fig. 5B).
We then determined if the lack of pyrimidine nucleosides in DD778-treated cells accounts for the antiviral potency of this compound. Thus, HEK-293T cells were infected with MV-Luc and then cultured with DD778 in a medium supplemented with either uridine or precursors of pyrimidine nucleoside biosynthesis, including dihydroorotate (DHO) or orotate (Fig. 5C). Uridine and orotate were found to restore viral replication in DD778-treated cells, as assessed by luciferase expression, whereas DHO showed no effect. This established that low levels of pyrimidine nucleosides account for the inhibition of viral growth in DD778-treated cells. Furthermore, the fact that orotate but not DHO could revert the antiviral effects of DD778 suggested that DHODH, the fourth enzyme of pyrimidine biosynthesis that converts DHO to orotate, is the target of DD778 (Fig. 5C). This is in line with previous reports, including ours, showing that DHODH is the target of numerous broad-spectrum antiviral compounds (4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18).

DD778 amplifies interferon expression in cells stimulated with RIG-I ligands.
We investigated potential interactions between DD778 and the innate antiviral response, in line with previous reports showing that inhibitors of de novo pyrimidine biosynthesis stimulate the expression of ISGs. First, it was confirmed that DD778 induced the ISRE-luciferase gene in STING-37 cells, like DD363, whereas the chemical analog DD786 that did not inhibit MV replication (Table 1)

New Pyrimidine Synthesis Inhibitors as IFN Activators
Antimicrobial Agents and Chemotherapy S2A). We then determined if, as expected, DD778 could also amplify the cellular response to synthetic ligands of RIG-I that mimic viral nucleic acids from RNA viruses (22). STING-37 reporter cells were transfected with low doses of short synthetic 5=triphosphate RNA molecules (ssRNA) to prime the innate immune response without reaching saturation and were cultured with DD778 or DMSO alone for 24 h. As shown in Fig. 6A, DD778 amplified ISRE-luciferase expression in response to ssRNA transfection. In contrast, DD786 had no effect on cellular response toward ssRNA (Fig. S2B). We also tested the effects of DD778 on ISRE-luciferase induction by a virus. Thus, STING-37 cells were infected with MV at different multiplicities of infection (MOIs) and incubated for 9 h to let the virus trigger the innate immune response, and DD778 or DMSO alone was added for an additional incubation period of 16 h. As shown in Fig. 6B, DD778 efficiently boosted the cellular response to MV infection, as assessed by increased expression levels of the ISRE-luciferase reporter gene. Finally, we determined if DD778 could also boost the expression of IFNs in response to ssRNA, something that was not investigated before. To address this question, supernatants from cell cultures transfected with ssRNA and treated with DD778 or DMSO alone for 48 h were collected, diluted twice, and applied to fresh STING-37 reporter cells to measure IFN biological activity with the ISRE-luciferase reporter gene. As shown in Fig. 6C, DD778 strongly amplified the IFN activity in culture supernatants of ssRNA-transfected cells. It should be noticed that the supernatants of cells treated with DD778 alone did not show any IFN activity, in line with previous reports using other pyrimidine biosynthesis inhibitors (8,16). To further document the consequences of DD778 treatment on the cellular response to ssRNA, mRNA levels for RIG-I, MDA5, oligoadenylate synthase-like protein (OASL), IFN-␣4, IFN-␤, IFN-1, and IFN-2/3 were determined by reverse transcriptionquantitative PCR (RT-qPCR). The results showed that in response to ssRNA transfection, DD778 amplified the induction of RIG-I, MDA5, OASL, IFN-␤, IFN-1, and IFN-2/3 (Fig.  7A). Furthermore, the effects of DD778 on the expression of these genes could be reversed by the addition of uridine in culture medium. This demonstrated that pyrimidine depletion is responsible for the amplified response to ssRNA, in agreement with our previous report on DD264 (8). To confirm these results at the protein level, we determined MDA5 and RIG-I expression by Western blotting. As shown in Fig. 7B, DD778 promoted the expression of MDA5 and RIG-I in cells transfected with ssRNA. In parallel, we measured in culture supernatants the concentrations of IFN-␤, IFN-1, IFN-2/3, and interferon-inducible protein of 10 kDa (IP-10), which is a well-characterized IFN-inducible chemokine (Fig. 7C). We found that DD778 also increased the expression of IFN-␤, IFN-1, IFN-2/3, and IP-10 when cells were stimulated with ssRNA. Altogether, our results showed that DD778 amplifies the expression of ISGs in cells stimulated with ssRNA but also produced higher levels of type I and III IFNs. This illustrates an original mechanism by which cells can communicate with neighbors on their metabolic status.

DD778 modulates the innate immune response in MV-infected cells.
Finally, we determined if DD778 also amplified the expression of IFN-␤, IFN-1, IFN-2/3, IP-10, MDA5, and RIG-I in the context of a viral infection. STING-37 cells were infected with MV (MOI ϭ 2) and incubated for 9 h to let the virus trigger the innate immune response, and DD778 or DMSO alone was added for an additional incubation period of 39 h. IFN-␤, IFN-1, IFN-2/3, and IP-10 were quantified in culture supernatants (Fig. 8A), whereas MDA5 and RIG-I expression levels were determined by Western blotting on cellular pellets (Fig. 8B). As shown in Fig. 8A, MV alone induced the different cytokines, and DD778 efficiently increased the expression of IFN-1, IFN-2/3, and IP-10. The expression of RIG-I in MV-infected cells was also boosted in the presence of DD778. Surprisingly, IFN-␤ expression in MV-infected cells was not statistically enhanced by DD778 treatment (Fig. 8A), in contrast to results obtained with ssRNA (Fig. 7C). The effects on MDA5 expression were also limited, even though we observed a significant induction of MDA5 by MV plus DD778, but not with MV infection alone. One explanation is that MV encodes virulence factors that interfere with the innate immune response, in particular the V protein that binds directly to MDA5 to prevent its activation and counteract IFN-␤ expression (23)(24)(25). Nevertheless, the results obtained for RIG-I, IFN-1, IFN-2/3, and IP-10 clearly demonstrate that DD778 can boost the innate immune response in virus-infected cells, in particular the expression of type III IFNs, which were produced at high levels in this system.

DISCUSSION
Components of the de novo pyrimidine biosynthesis pathway, and DHODH in particular, are becoming drug targets of prime interest in the treatment of multiple pathologies, including viral and microbial infections, cancer, and autoimmune diseases (3). A most recent example is a breakthrough study showing the therapeutic potential of DHODH inhibitors in the treatment of acute myeloid leukemia (26). So far, teriflunomide and its prodrug leflunomide, which are used as immunosuppressors in the treatment of multiple sclerosis and rheumatoid arthritis, are the only two DHODH inhibitors on the market (Fig. 9). However, it has been shown that the immunosuppressive properties of these drugs and their capacity to inhibit lymphocyte proliferation in vivo rely on interactions with secondary targets, and in particular tyrosine kinases (27). Therefore, the identification of highly active and specific inhibitors of de novo pyrimidine biosynthesis is a field of intense competition. In the last decade, many lead compounds that are mostly targeting DHODH have been characterized in detail and are currently at a different stage of development. However, these molecules can be . Total RNAs were extracted from cellular pellets collected at T ϭ 24 h, and the expression levels of specified genes were determined by RT-qPCR. Data represent means Ϯ SD of the results from 3 independent experiments. Bold figures correspond to statistically significant differences comparing DD778 to DMSO-treated samples or DD778 ϩ uridine to DD778 alone (P Ͻ 0.05, calculated by two-way ANOVA with Bonferroni's post hoc test). (B) Same as above, but MDA5 and RIG-I expression levels were determined by Western blotting on total protein extracts from STING-37 cells at T ϭ 48 h. Upper images correspond to representative experiments. Graphs below correspond to the quantification of Western blotting results, where values were normalized to actin (means Ϯ SD of the results from 3 independent experiments; cells were transfected with 120 ng of ssRNA). *, P Ͻ 0.05; and **, P Ͻ 0.01, as calculated by one-way ANOVA with Bonferroni's post hoc test. (C) Same experiment as above, but total concentrations of indicated cytokines in culture supernatants at T ϭ 48 h were determined using the LEGENDplex analysis kit (cells were transfected with 120 ng of ssRNA). Data represent means Ϯ SD of the results from 4 independent experiments. *, P Ͻ 0.05; and **, P Ͻ 0.01, as calculated by two-tailed standard t test. clustered in a relatively limited number of chemical families, and there is still a great need for original structures that could help in the development of molecules suitable for the clinic (3).
In this study, we describe a new class of pyrimidine biosynthesis inhibitors based on 3-(1H-indol-3-yl)-2,3-dihydro-4H-furo[3,2-c]chromen-4-one. Complementation experiments in cell-based assays strongly suggest that DD778 and related analogs target DHODH, but this will require confirmation by crystallography or enzymatic studies. We previously determined the antiviral potencies of brequinar and teriflunomide in HEK-293T cells infected by MV-Luc, and IC 50 s were estimated at 40 nM and 5 M, respectively (Fig. 9). Therefore, the antiviral potency of DD778 in this assay is comparable to that of teriflunomide but is much lower than that of brequinar or other DHODH inhibitors, such as the series of 2-(3-alkoxy-1H-pyrazol-1-yl)pyrimidines that we recently described (IC 50 ϭ 0.7 nM for 21q) (9). Accordingly, brequinar induced the ISRE-luciferase reporter gene in STING-37 cells and inhibited de novo pyrimidine biosynthesis at submicromolar concentrations ( Fig. S3A and B). Thus, DD778 and other tested analogs are clearly not among the most active DHDOH inhibitors, but their interest lies in their chemical structure. Indeed, these compounds are devoid of the carboxylic group characterizing DHODH inhibitors from the brequinar family, but they are also missing the amide group that is central to teriflunomide and its derivatives (Fig. 9). Furthermore, the novel chemical series described herein is based on two edge-fused polycyclic moieties linked together, something relatively unique among pyrimidine biosynthesis inhibitors described in the literature (3). To our knowledge, the only known DHODH inhibitor formed from two edge-fused polycyclic moieties is dicoumarol (IC 50 ϭ 6.6 M when tested on the recombinant enzyme) (28), but the description of its activity was limited, and no effect on innate immunity or viral growth has been reported before ( Fig.  9) (28,29). We thus tested dicoumarol in our different assays and found that it induced the ISRE-luciferase reporter gene in STING-37 cells (Fig. S3C), amplified the cellular response to ssRNA transfection (Fig. S3D), and efficiently inhibited MV replication (IC 50 ϭ 12 Ϯ 0.3 M; Fig. S3E). Most importantly, and despite the fact that dicoumarol has multiple cellular targets and biological activities, the antiviral effect was fully reversed when supplementing the culture medium with uridine ( Fig. S3F), thus establishing a functional link with the inhibition of pyrimidine biosynthesis. Quite interestingly, the 4-hydroxycoumarin group of dicoumarol is showing obvious similarities with the dihydrofurochromenone of DD778. Molecular docking experiments on DHODH structures also suggest that 4-hydroxycoumarin can be regarded as a substitute to the quinoline-4-carboxylic acid of brequinar to ensure critical interactions with Arg136 of DHODH (28). It is thus tempting to speculate that the 2,3-dihydro-4H-furo[3,2-c]chromen-4-one moiety of DD778 and other active analogs play the same role. Altogether, our findings expand our current knowledge of chemical possibilities to target the pyrimidine biosynthesis pathway and should help in the development of drugs amenable to the clinic.
Most importantly, we characterized in detail the interactions between this new series of pyrimidine biosynthesis inhibitors, which were identified in an ISRE-luciferase screen, and the innate antiviral response. It has been previously shown that pyrimidine biosynthesis inhibitors are weak stimulators of ISGs when applied alone (8,(13)(14)(15)(16)(17)(18), but these drugs boost the expression of ISGs in response to either extracellular or intracellular stimuli, such as IFN-␤, RIG-I ligands, or viruses (8,12,13,15,18). Besides, pyrimidine biosynthesis inhibitors do not induce any type of IFNs when applied alone, and virus growth inhibition by these drugs seems to occur independent of IFN-␣/␤ synthesis (8,15,16), despite some recent report showing no antiviral effects in Vero cells which are deleted for the IFN-␣/␤ gene cluster (18). Here, we showed that IFN-␤, IFN-1, and IFN-2/3 syntheses induced by the transfection of 5=-triphosphate RNA

New Pyrimidine Synthesis Inhibitors as IFN Activators
Antimicrobial Agents and Chemotherapy molecules are significantly amplified in DD778-treated cells. In MV-infected cells, DD778 also increased the expression of IFN-1 and IFN-2/3, which were produced at high levels in this system. It should be noted that cells were infected 9 h before the addition of DD778, thus allowing the accumulation of viral RNAs necessary to trigger the innate immune response before the antiviral state was settled. This was not always performed in previous reports and might account for the contradictory results obtained by others (6,16). These new observations do not challenge the idea that the antiviral state induced by pyrimidine deprivation is essentially IFN independent. Indeed, it has been previously reported that chemical inhibitors of JAK1, which is signaling downstream of type I, II, and III IFN receptors, did not reverse the antiviral state induced by pyrimidine biosynthesis inhibitors (15). Our results, rather, expand the impact of pyrimidine deprivation on the innate antiviral response by documenting the consequences on cytokine secretion and cell-to-cell communication. Indeed, we showed that when cells are deprived of pyrimidines and suffer such a metabolic stress, the alarm signal they send to neighboring cells in response to RIG-I ligands is enhanced. Quite interestingly, this links cellular metabolism to cytokine exchanges and cell-to-cell communication in the context of innate immunity. However, the way in which pyrimidine biosynthesis inhibitors influence the expression of innate immunity genes remains a pending question. Although this is clearly a consequence of pyrimidine deprivation, since it is abolished when culture medium is supplemented with uridine, the mechanisms involved are still poorly understood. Most recently, it has been shown in Vero cells that RIG-I is among the ISGs induced by pyrimidine biosynthesis inhibitors, which could account for some enhanced sensing of 5=-triphosphate RNA (16). However, we did not confirm this observation in HEK-293 cells (Fig. 7A and B and 8B), suggesting that RIG-I induction by pyrimidine biosynthesis inhibitors alone is cell type dependent. In the future, a major challenge will be to identify the factors that sense intracellular levels of pyrimidines and the signaling pathways triggered to induce an antiviral state, to promote the expression of ISGs, and to enhance the synthesis of IFNs.

MATERIALS AND METHODS
Compound library, screening procedure, and DD363 analogs. The initial screen that allowed the identification of DD363 was described in detail elsewhere (8). Briefly, HEK-293T cells were transfected with the pISRE-luciferase reporter plasmid (reference no. 219089; Stratagene) and dispensed at 2 ϫ 10 4 cells/well in white 96-well plates already containing 1 l of compounds from the Institut Curie library at 2 mg/ml in DMSO. Since the final culture volumes were 100 l/well, compounds were screened at 20 g/ml, corresponding to 57 M for DD363. Compound synthesis is described in in the supplemental material and Table 1.
Cell lines, culture medium, and luciferase assays. Cells were cultured at 37°C and 5% CO 2 in Dulbecco's modified Eagle's medium (DMEM; Gibco-Invitrogen) containing 10% fetal calf serum (FCS), penicillin, and streptomycin. Human MRC5 and HEK-293T cells were from the ATCC. The reporter cell line STING-37, corresponding to HEK-293 cells stably transfected with the ISRE-luciferase reporter gene, was previously described (8). Luciferase induction in STING-37 cells was determined using the Bright-Glo (Promega) or Britelite plus reagents (PerkinElmer), according to the manufacturer's recommendations. Bioluminescence was measured for 0.1 s with a luminometer (EnSpire; PerkinElmer). Cellular viability was determined by quantification of ATP in culture wells using the CellTiter-Glo assay (Promega).
Viruses, infections, and immunostaining. Experiments with measles virus (MV) were performed with the vaccine strain Schwarz from the ATCC. The recombinant MV strain expressing firefly luciferase (MV-Luc) from an additional transcription unit was derived from the vaccine strain Schwarz and was previously described (30). Virus stocks were produced on Vero cells and titrated by 50% tissue culture infective dose (TCID 50 ) on Vero cells. Chikungunya virus (CHIKV) infections were performed with wild-type strain 05115 from La Réunion Island or the recombinant strain CHIKV/Ren expressing Renilla luciferase (kindly provided by Philippe Desprès [19]). HCoV-229E was originally obtained from the ATCC (VR-740), and virus stocks were produced on MRC5 cells, as previously described (31). Firefly and Renilla luciferase expression from MV-Luc and CHIKV/Ren recombinant viruses was determined using the Bright-Glo (Promega) and Renilla-Glo (Promega) reagents, respectively.
For immunofluorescence detection of viral proteins, HEK-293T cells were infected with CHIKV at an MOI of 0.1 and cultured on microscopy slides (Ibidi, Munich, Germany). At 24 h postinfection, cells were fixed for 15 min with 4% paraformaldehyde (PFA) and permeabilized for 5 min at 4°C in phosphatebuffered saline (PBS) containing 0.1% Triton X-100. Cells were incubated for 1 h in PBS supplemented with 5% goat serum and incubated again in the same solution containing the Cy3-conjugated antibody against the CHIKV E2 protein (clone 3E4; kindly provided by Philippe Desprès) (32). For HCoV-229E, MRC5 cells were infected at an MOI of 0.1 and cultured on microscopy slides (Ibidi) for 48 h. Cells were fixed and permeabilized as described above and stained with an anti-spike glycoprotein monoclonal antibody Aldrich (clone AC-15, catalog no. A5441). Secondary anti-mouse and anti-rabbit horseradish peroxidase (HRP)-conjugated antibodies were from Dako (catalog no. P0447) and Sigma-Aldrich (catalog no. A0545), respectively. Protein detection was performed using the SuperSignal West Pico chemiluminescent substrate (Thermo Fisher Scientific) and quantified with the Adobe Photoshop software.
Cytokine quantification in culture supernatants. Cytokine levels in culture supernatants of HEK-293 cells stimulated with ssRNA or MV with or without DMSO or DD778 at 30 M were determined using the LEGENDplex analysis kit, according to the manufacturer's recommendations (human anti-virus response panel with V-bottom plate, catalog no. 740390).

SUPPLEMENTAL MATERIAL
Supplemental material for this article may be found at https://doi.org/10.1128/AAC .00383-17.