SCD2-mediated monounsaturated fatty acid metabolism regulates cGAS-STING-dependent type I IFN in CD4+ T cells

Intimate interactions exist between the host lipid metabolism and viral responses. However, how acquired immune systems adapt lipid metabolism to meet demands and whether or not the metabolic rewiring confers a selective advantage in host immunity remains unclear. We found that viral infection attenuated the expression of genes related to lipid metabolism in CD4 + T cells, which in turn increased the anti-viral gene expression. The inhibition of the fatty acid synthesis pathway substantially increased the basal expression of anti-viral genes via the spontaneous production of type I interferon. Using a combination of CRISPR/Cas9-mediated genome editing technology and a global lipidomics analysis, we found that the decreased monounsaturated fatty acid by genetic deletion of Scd2 was crucial for the induction of an anti-viral response through the activation of cGAS-STING pathway. These ndings demonstrate the novel relationship between fatty acid biosynthesis and type I IFN responses that enhances the anti-viral response. (H1N1) was used. BALB/c were anesthetized by i.p. injection with midazolam, domitor, and butorphanol tartrate. In the administrative transfer experiment, 5 x 10 6 of control Th1 or sgScd2 Th1 cells suspended in 30 μl were administered intranasally the day before infection. Infected mice were weighed daily, with 75% weight loss as the criterion for euthanasia. After three kinds of mixed anesthesia, antisedan was also administered intraperitoneally in all experiments. More than three independent experiments were performed and showed similar results (a-j and l-p = 3, k = 4). Three technical replicates were performed with quantitative RT-PCR (e-g and l, m). ACC1-/-, sgScd2, and sgFads2 Th1 cells compared to control Th1 cells. Normalized values were shown here. Two independent samples were conducted for lipidomics analysis. (n=2 per each group biologically independent sample). g, Pie chart showed that the ratio of fatty acid species incorporated into lipid detected by lipidomics analysis. h, Species of lipid contents with or without oleic acid were shown. i, A heat map depicts the relative contents of TGs. j, A scatter plot depicts lipid contents of control and sgScd2 Th1 cells. The dashed lines indicate 50% changed line for the difference in lipid contents. The X-axis shows the mean value of control and sgScd2 Th1 cells, and the Y-axis shows the percentage of sgScd2 Th1 cells relative to control. k, qRT-PCR analyses of the relative expression of ISGs in control, sgScd2, and sgGpam Th1 cells. Relative expression (normalized to Hprt) with SD is shown. l, qRT-PCR analyses of the relative expression of ISGs in control, sgScd2, and sgDgat1 Th1 cells. Relative expression (normalized to Hprt) with SD is shown. Two biological replicates were performed for lipidomics analysis (d-j). More than three independent experiments were performed and showed similar results (a, b, c, k, l). Three technical replicates were performed with quantitative RT-PCR (k, l).


Introduction
Evidences concerning the changes in the lipid metabolic pathway in host cells during virus infection has accumulated 1,2 . These changes in the lipid metabolism have both harmful and bene cial effects on host cells. Human cytomegalovirus alters the host cell lipid metabolism to produce long-chain fatty acids for virus replication 3 . Hansen et al. reported that the DNA virus herpes simplex type 2 induces nitro-fatty acid production, resulting in the inhibition of host cell-derived interferon (IFN) generation 4 . In contrast, it was reported that polyunsaturated fatty acid (PUFA)-derived lipid mediator protectin D1 attenuates in uenza virus replication via RNA export machinery as a bene cial role of lipids in host defense 5 . Interestingly, in macrophages, the abrogation of the cholesterol synthesis pathway triggers the spontaneous production of type I IFN (IFN-I), resulting in the upregulation of anti-viral activity 6 . IFN-I elicits resistance to invading viral pathogens by inducing a diverse range of genes related to the antiviral response, called interferon-stimulated genes (ISGs). Due to the ubiquitous expression of the IFN-I receptor (IFNAR), most cells can receive IFN-I signaling. In addition to ISGs upregulation, IFN-I has been shown to in uence the ability of CD4 + T cells to construct follicular helper T cells and help B cells produce antibodies 7 . Although IFN-I is mainly produced by innate immune cells, such as plasmacytoid dendritic cells (pDCs) and macrophages, CD4 + T cells can also produce IFN-I in response to viral infection 8 .
It was recently reported that IFN-I induces changes in the core lipid metabolism that are critical for the immune function against anti-viral responses 9 . IFN-I signaling increases the fatty acid oxidization (FAO) and oxidative phosphorylation (OXPHOS) required for the full activation of pDCs 9 . In macrophages, IFNβ treatment reduces the total amount of cellular cholesterol and increased the level of the anti-viral lipid 25hydroxycholesterol (25HC) 10 . 25HC restricts mammalian retrovirus infection by inhibiting viral DNA replication, viral entry and growth. Thus, the lipid metabolism and anti-viral response are mutually connected in innate immune cells. However, little is known about the speci c lipid metabolic regulators that control the adaptive immune cell function against virus infection.
The intracellular metabolism reportedly contributes to the regulation of proliferation, activation and memory generation of T cells 11,12 . In naïve CD8 + T cells, the metabolic switch from OXPHOS to glycolysis is required for effector differentiation upon antigen invasion 13 . In the context of acquired immunity, cell intrinsic lysosomal lipolysis is necessary for the generation and maintenance of memory CD8 + T cells 14 .
CD4 + T cells also require lipid metabolism for their proliferation, activation and memory generation. mTOR-PPARγ axis-mediated fatty acid metabolic reprogramming is required for the early activation of CD4 + T cells 15 .
Acetyl-CoA carboxylase 1 (ACC1), a rate-limiting enzyme of fatty acid biosynthesis, contributes to Th17 cell differentiation in mouse and human obesity 16 . It has also been reported that each effector Th cell subset uses different chain length of fatty acids for their differentiation 17 . The middle-and long-chain fatty acids support Th1 and Th17 cell differentiation and short-chain fatty acids support regulatory T cell differentiation. We also previously reported that ACC1 characterizes the memory potential of individual CD4 + T cells 18 . ACC1 lo effector Th1 cells are more likely to become memory cells compared to ACC1 hi effector Th1 cells. Interestingly, type I IFN-related genes were upregulated in ACC1 lo effector Th1 cells. However, while mounting evidence indicates the importance of lipid metabolism for the T cell differentiation and function, the role of lipid metabolism in anti-viral responses remains unclear in CD4 + T cells.
In the present study, we investigated the relationship between the fatty acid synthesis pathway and antiviral responses in CD4 + T cells. We found that in uenza virus infection attenuated the expression of genes associated with lipid metabolism in CD4 + T cells. The genetic deletion or pharmacological inhibition of ACC1 resulted in a dramatic increase in ISGs expression in both mouse and human Th1 cells. Deeper analyses revealed that the ISGs expression was augmented following IFN-I production in an autocrine (and/or paracrine) fashion through IFNAR signaling. Furthermore, a mechanistic study indicated that limiting the levels of triacylglycerol containing MUFA was crucial for the regulation of the T cell anti-viral responses induced by the cGAS-STING signaling axis. Our data therefore indicate that changes in the MUFA metabolism in T cells during virus infection trigger cGAS-STING-mediated IFN-Irelated immune responses

Result
In uenza virus infection reprograms lipid metabolism in lung CD4 + T cells.
To elucidate the role of IFN in cellular metabolism of T cells, we rst performed a RNA-sequence analysis following treatment of Th1 cells with the anti-viral protein IFNβ. Gene ontology and pathway analyses using the NIAID DAVID and KEGG databases 19 showed a signi cant enrichment of several functional categories, including the fatty acid biosynthetic process (Fig. 1a). A gene set enrichment analysis (GSEA) also showed the decreased expression of gene sets including the fatty acid biosynthetic process (Fig. 1b). Importantly, IFNβ treatment resulted in the decreased expression of Acaca, which encodes ACC1, a rate-limiting enzyme for fatty acid biosynthesis (Fig. 1c). Quantitative real-time polymerase chain reaction (qRT-PCR) also showed the decreased expression of genes related to lipid metabolism (Fig. 1d).
These data suggest that IFN-I signaling suppresses the lipid metabolism in T cells during virus infection.
We next examined the expression of genes associated with lipid metabolism in lung CD4 + T cells following in uenza virus infection (X31 strain). The expression of CD317, which is a surface protein known to restrict virus infection 8 , was increased on lung CD4 + T cells and comparable to that on other adaptive immune cells, such as CD8 + T cells and B cells (Fig. 1e, and Supplementary Fig. 1a, 1b). Splenic CD4 + T cells, CD8 + T cells and B cells showed very slight changes in the CD317 expression ( Supplementary Fig. 1c). Consistent with these data, the enhanced expression of representative ISGs, such as Ddx60, I t1, Mx1 and Oas3, was detected in lung CD4 + T cells obtained from infected mice ( Supplementary Fig. 1d). In sharp contrast, a substantially reduced expression of genes involved in the lipid metabolism was detected in CD4 + T cells from X31-infected mice (Fig. 1f). Interestingly, slight changes were observed in the expression of genes related to the lipid metabolism in CD8 + T cells ( Supplementary Fig. 1e).
To explore whether or not a similar phenomenon was observed in human CD4 + T cells, we treated primary human PBMC-derived CD4 + T cells with IFNβ. Consistent with our ndings in the mouse experiment, IFNβ treatment decreased the expression of genes related to lipid metabolism (Fig. 1g).
The fatty acid biosynthesis pathway controls the ISGs expression in CD4 + T cells.
The results in Fig. 1 prompted the hypothesis that reduced levels of fatty acid biosynthesis might be required for the induction of basal level of ISGs in CD4 + T cells. We used mice in which the biotin carboxyl carrier protein domain in the Acaca gene had been conditionally deleted in CD4 + T cells driven by the Cd4 promoter (herein referred to as ACC1 −/− ) 20 . To investigate the role of ACC1 in T cell development, we analyzed thymus and spleen derived from ACC1 −/− mice. ACC1 −/− mice showed normal proportion and numbers of CD4 + and CD8 + T cells in the thymus ( Supplementary Fig. 2a, upper panel), whereas the proportion and numbers of CD4 + and CD8 + T cells in the spleen was slightly reduced in ACC1 −/− mice compared to ACC1 +/+ mice ( Supplementary Fig. 2a, lower panel)). However, we did not nd signi cant changes in the proportion of memory phenotype CD4 + T cells and naïve CD4 + T cells between ACC1 +/+ and ACC1 −/− mice ( Supplementary Fig. 2b). To examine the effect of Acaca deletion on the expression of ISGs, we analyzed the global gene expression pro les of Th1 cells in ACC1 −/− mice. A total of 395 genes showed a greater than 2-fold change, including 185 up-regulated and 210 down-regulated genes in ACC1 −/− Th1 cells. Importantly, the genes concerning anti-viral responses were signi cantly enriched in ACC1 −/− Th1 cells without IFN-I treatment (Fig. 2a, and Supplementary Fig. 2c). A GSEA con rmed statistically signi cant enrichment of IFN-I-inducible genes in ACC1 −/− cells (Fig. 2b). Similarly, the dramatic up-regulation of ISGs was detected when we treated Th1 cells with TOFA, an allosteric inhibitor of ACC1 (Supplementary Fig. 2d and 2e). An ontology analysis also showed that the pharmacological inhibition of ACC1 resulted in the enrichment of the pathway related to the anti-viral response ( Supplementary Fig. 2f). These results indicate that the cell-autonomous basal expression of anti-viral genes was induced by both genetic deletion and pharmacological inhibition of ACC1 in CD4 + T cells.
We next compared the gene expression pro les of ACC1 −/− and IFN-I treated Th1 cells, and surprisingly, 99 up-regulated genes (53.5% and 72.3% of up-regulated genes in ACC1 −/− and IFN-I-treated Th1 cells, respectively) were shared between the two groups (Fig. 2c). Quantitative RT-PCR analysis con rmed that the basal expression of ISGs was increased in ACC1 −/− or TOFA-treated cells as a comparable level to IFN-I-treated cells ( Fig. 2d and Supplementary Fig. 2g). We also found the upregulation of ISGs in ACC1 −/− or TOFA-treated CD4 + T cells cultured under Th2 polarization conditions ( Supplementary Fig. 2h). Similarly, the expression of ISGs was enhanced in TOFA-treated Th0, Th9, Th17 or regulatory T cells as comparable levels to those of cells treated with IFN-I. (Supplementary Fig. 2i-l). In addition, we also observed that the ISGs expression was upregulated in CD8 + T cells by the treatment of TOFA ( Supplementary Fig. 2m). IRF7, interferon regulatory factor 7, is known as an essential transcription factor controlling the expression of ISGs for anti-viral immune responses 21 . Protein expression of IRF7 was signi cantly upregulated by the inhibition of fatty acid biosynthesis ( Fig. 2e and 2f). The basal expression level of ISGs was also increased by the inhibition of the fatty acid biosynthesis pathway in human Th1 cells (Fig. 2g). These data indicate inhibition of fatty acid synthesis increases the basal expression of ISGs in CD4 + and CD8 + T cells.
Spontaneous production of IFNa by limiting the fatty acid biosynthesis pathway triggers a type I IFN response in CD4 + T cells.
To investigate the mechanism underlying the induction of ISGs by the inhibition of de novo fatty acid biosynthesis in CD4 + T cells, we rst quanti ed the amount of IFNa in the Th1 cell culture supernatant without TCR re-stimulation. Signi cantly higher concentrations of IFNa were detected in the supernatants cultured from ACC1 −/− and TOFA-treated Th1 cells (Fig. 3a). Importantly, the induction of ISGs in ACC1 −/− Th1 cells was clearly prevented by IFNAR blockade, suggesting that the secreted IFNa might have induced the ISGs expression (Fig. 3b). We also found that the protein level of IRF7 was signi cantly decreased by the inhibition of IFNAR signaling in TOFA-treated Th1 cells ( Fig. 3c and 3d).
We additionally investigated the in uence of IFNa produced by ACC1 −/− Th1 cells on the induction of ISGs using co-culture systems ( Supplementary Fig. 3a). Co-culture of wild-type (WT) and ACC1 −/− Th1 cells led to the upregulation of ISGs and IRF7 protein of co-cultured WT Th1 cells, with levels almost the same as those in ACC1 −/− Th1 cells (Fig. 3e, Supplementary Fig. 3b, and 3c). These results reinforced that IFNa produced by ACC1 −/− Th1 cells plays a role in autocrine and paracrine signaling through IFNAR to induce the expression of ISGs. Upon the binding of IFN-I on IFNAR, the JAK/STAT pathway is activated, inducing the construction of ISGF3, composed of STAT1, STAT2 and IRF9 for the production of ISGs 22 .

Limiting fatty acid biosynthesis protects against viral infection.
Our data suggested that the inhibition of the fatty acid biosynthesis pathway induced spontaneous IFNa production by CD4 + T cells, resulting in the ISGs upregulation in surrounding cells. These data prompted us to investigate whether or not the inhibition of fatty acid synthesis induces anti-viral activity.
We collected supernatants from control, ACC1 −/− or TOFA-treated Th1 cells and supplemented them to MLE-15 cells, a murine lung epithelial cell line ( Supplementary Fig. 3d). The expression of ISGs and IRF7 protein was strongly induced in MLE-15 cells supplemented with culture supernatant from ACC1 −/− or TOFA-treated Th1 cells ( Fig. 3i and Supplementary Fig. 3e). Consistent with the augmentation of the ISGs expression, the supplementation of MLE-15 cells with the culture supernatant from TOFA-treated Th1 cells conferred resistance against in uenza virus (Fig. 3j). A set of data indicate that spontaneous IFNa production induced by the inhibition of de novo fatty acid biosynthesis in CD4 + T cells can activate basal level of ISGs expression and induce anti-viral responses in the surrounding cells.
The MUFA metabolism plays a crucial role in the regulation of the type I IFN response in CD4 + T cells.
Since ACC1 is a rate-limiting enzyme of fatty acid biosynthesis, the inhibition of ACC1 can globally alter the cellular lipid pro les 20 . To clarify which type of fatty acids play an essential role in the regulation of ISGs induction, we focused on stearyl-CoA desaturase (SCD) and fatty acid desaturase (FADS), which control the production of MUFA and polyunsaturated fatty acid (PUFA), respectively (Supplementary Fig. 4a). When we pharmacologically inhibited SCD enzyme, the basal expression level of ISGs was substantially induced in Th1 cells. Treatment of FADS inhibitor modestly increased basal expression level of ISGs compared to control group ( Fig. 4a and Supplementary Fig. 4b).
To further evaluate the role of SCD and FADS in ISGs regulation in CD4 + T cells, we performed CRISPR/Cas9-mediated genome editing of Scd2 and Fads2, which are preferentially expressed in Th1 cells ( Supplementary Fig. 4c-f). The genetic deletion of Scd2 (sgScd2) in Th1 cells increased the basal ISGs expression almost to the same level seen in ACC1 −/− cells (Fig. 4b). Although the genetic deletion of Fads2 (sgFads2) resulted in the higher level of ISGs expression as compared to control Th1 cells, those expression levels were much lower than sgScd2 Th1 cells. Consistently, the supplementation of MLE-15 cells with the culture supernatant from sgScd2 Th1 cell, which contained high levels of IFNa, signi cantly enhanced anti-viral activity against in uenza as compared to control or sgFads2 groups ( Fig. 4c and 4d). We therefore suspected that either the accumulation of saturated fatty acid (SFA) or the de ciency of MUFA caused the induction of a type I IFN response following the deletion of Scd2. The upregulation of ISGs expression was clearly canceled in sgScd2 Th1 cells by the addition of oleic acid (OA), which is a kind of MUFA (Fig. 4e). However, extrinsic SFA supplementation was not able to decrease the basal ISGs expression in sgScd2 Th1 cells (Fig. 4e). The supplementation of SFA into WT Th1 cell culture did not substantially increase the basal expression of ISGs ( Supplementary Fig. 4g).
Reduced cholesterol biosynthesis reportedly enhances the anti-viral responses in macrophages 6 .
However, pharmacological inhibition of the sterol synthesis pathway showed only slight changes in the ISGs expression in Th1 cells ( Supplementary Fig. 4h). In consistent with this data, the addition of cholesterol failed to suppress the basal expression of ISGs (Fig. 4f). Taken together, these ndings indicate that a reduction in MUFA metabolites can induce the expression of ISGs and encourage anti-viral responses in Th1 cells.
Genetic deletion or pharmacological inhibition of SCD2 confer resistance to lethal in uenza virus infection in vivo.
To investigate whether the inhibition of fatty acid biosynthesis exhibits anti-viral responses in vivo, mice were infected with A/Puerto Rico/8/34 (PR8), as causes more severe infection than X31 in mice. Mice infected with 5LD50 of PR8 virus were monitored for 12 days to analyze weight change and survival. Although the mice injected with control Th1 cells failed to improve the weight loss and survival rate, 3 of 10 mice in the group injected with sgScd2 Th1 cells survived and there was a moderate rate of weight loss on day 5 (No-Transfer: 83.3%, Control Th1 cells: 83.1%, sgScd2 Th1 cells: 90.5%) ( Fig. 4g and 4h). Taken together, these data indicate that genetic deletion of SCD2 augmented anti-viral responses and protected mice from in uenza virus infection in vivo settings.
Rewiring the MUFA metabolism triggers cGAS-STING activation in CD4 + T cells.
Next, we sought to gain more mechanistic insight into the role of fatty acid metabolism in the induction of basal ISGs expression. To this end, we analyzed the involvement of the MAVS and STING pathways for the production of IFN-I in Th1 cells. MAVS and STING work as upstream enzymes activating TANKbinding kinase 1 (TBK1) for the production of IFN-I 23 . Higher levels of phospho-TBK1 were detected in TOFA treated and sgScd2 Th1 cells ( Fig. 5a and Supplementary Fig. 5a). Although slight phosphorylation of TBK1 was observed in sgFads2 Th1 cells compared to control Th1 cells, those levels were much lower than TOFA treated and sgScd2 Th1 cells ( Fig. 5a and 5b). We next targeted Mavs and Tmem173, which encode MAVS and STING respectively, to clarify the role of these molecules in the regulation of IFN-I production in sgScd2 Th1 cells. Gene targeting of STING clearly inhibited the induction of phospho-TBK1 in sgScd2 Th1 cells, but not Mavs (Fig. 5c, Supplementary Fig. 5b, and 5c). Likewise, the phosphorylation of STING was induced in sgScd2 Th1 cells dependent on cGAS (Fig. 5d, Supplementary Fig. 5d, and 5e).
Consistent with the level of pTBK1, gene targeting of Mavs did not affect the basal expression levels of ISGs in sgScd2 Th1 cells (Fig. 5e). In contrast, double knockout of Scd2 with Mb21d1 or Tmem173 failed to increase basal ISGs expression (Fig. 5f). Similar trend was observed when we treated sgTmem173 or sgMb21d1 Th1 cells with SCD inhibitor (Supplementary Fig. 5f). It was reported that palmitoylation of STING at the Golgi is required for STING activation 24 . Thus, we next examined the effect of palmitoylation of STING on its activation using 2-bromopalmitate (2-BP), an inhibitor of STING palmitoylation and found that 2-BP suppressed STING phosphorylation and the induction of basal ISGs expression (Fig. 5g, 5h, and Supplementary Fig. 5g). Furthermore, supplementation of OA clearly suppressed ISGs induction and STING phosphorylation ( Fig. 5i and Supplementary Fig. 5h). These results indicated that the modulation of MUFA metabolism triggers STING phosphorylation, resulting in the ISGs induction.
Next, to investigate how STING signaling was activated, we addressed whether cGAMP is involved in MUFA-mediated STING activation in CD4 + T cells. Although the amount of cGAMP in control Th1 cells was below the detection limit, substantial amount of cGAMP was detected in sgScd2 Th1 cells (Fig. 5j).
Taken together, we considered the activation of STING signaling was induced by cGAS-cGAMP pathway. To support this hypothesis, we observed signi cantly higher amounts of cytosolic DNA in sgScd2 Th1 cells compared to control Th1 cells (Fig. 5k). Furthermore, the number of sequences containing Hprt and Gapdh in the genomic DNA was signi cantly increased in sgScd2 Th1 cells while the number of sequences containing Nadh and cytochromeb encoded in mitochondrial DNA was unchanged (Fig. 5l). In addition, due to the results in similar levels of mass and membrane potential of mitochondria between control and sgScd2 Th1 cells, we considered cGAMP was originated from genomic DNA ( Supplementary  Fig. 5i).
Since IFN-I treatment reduced gene expression related to lipid metabolism (Fig. 1a), we next explored whether extrinsic supplementation of fatty acids affected the expression of ISGs. Supplementation of OA moderately suppressed the ISGs upregulation induced by IFN-I treatment (Fig. 5m). Furthermore, phosphorylation of STING and TBK1 was also suppressed by supplementation of OA, but not palmitic acid or stearic acid ( Fig. 5n and 5o, Supplementary Fig. 5j). These data indicate that the IFN-I-mediated reduction in fatty acid biosynthesis contributed at least in part, to a positive-feedback mechanism of ISGs induction in CD4 + T cells. Finally, we found that the anti-viral responses induced by sgScd2 Th1 cells were signi cantly suppressed by gene targeting of cGAS or STING (Fig. 5p). Given these data, we concluded that rewiring the MUFA metabolism in CD4 + T cells triggers the activation of the cGAS-STING pathway, thereby enhancing anti-viral responses.
OA-containing TG is essential for the activation of type I IFN response in CD4 + T cells.
Finally, we investigated which types of lipids are important for the regulation of basal ISGs expression when fatty acid synthesis is inhibited. Lipid droplets (LDs) are cellular organelles for the storage of neutral lipids. Although the role of LDs in adipocytes has been well studied, accumulating evidence suggests the existence of LDs in immune cells 14,25,26 . As previously described, LDs were detected in primary Th1 cells ( Fig. 6a and Supplementary Fig. 6a). Intriguingly, both immuno uorescence microscopy and FACS analyses revealed that LDs of sgScd2 Th1 cells were smaller and lower compared to control Th1 cells (Fig. 6b, 6c, Supplementary Fig. 6a, and 6b). We therefore performed a lipidomics analysis to clarify which type of lipid was responsible for inducing basal ISGs expression in sgScd2 Th1 cells. A global cellular lipidomics analysis identi ed 616 lipid species including 177, 49, 10, 47, 216, 34, 52, 28, 3 species of glycerolipid, ganglioside, acylcarnitine, phospholipid, glycerophospholipid, lysophospholipid, glycolipid, cholesterol, and free fatty acid ( Supplementary Fig. 6c). Lipid groups such as glycerophospho lipid (38.3%), glycerol lipid (36.42%) and phospholipid (10.22%) were abundant in control Th1 cells (Fig. 6d). We found that the genetic deletion of Scd2, Fads2, or Acaca signi cantly altered the overall composition of cellular lipids (Fig. 6e). Among them, the ratio of glycerophospho lipid and glycerolipid was markedly altered (Supplementary Fig. 6d). Especially, the ratio of TG, which was the most abundant lipid in control Th1 cells, was reduced by genetic deletion of Scd2, Fads2, or Acaca (Fig. 6f). In contrast, some lipid groups including ganglioside, PE, LPC, LPE and Sph were increased ACC1 −/− cells ( Supplementary Fig. 6e left and right). To identify the lipid species responsible for inducing ISGs, we focused on the lipid group, which of contents changed similarly in ACC1 −/− or sgScd2 Th1 cells. We found the species of decreased lipid were TG, TG [e], BMP, PC [e], and PC [sn2 + O] (Fig. 6f and Supplementary Fig. 6e). We found that OA (16.8%), palmitic acid (12.1%) and stearic acid (8.1%) were major components of cellular lipids (Fig. 6g). Due to the abundance of OA in Th1 cells and the importance of OA for basal expression of ISGs (Fig. 4e), we classi ed the individual lipid species based on the OA content (Fig. 6h) (Fig. 6h, and Supplementary Fig. 6f left). Likewise, we also observed lower occupancy of OA in lysophospholipid, and glycolipids ( Supplementary Fig. 6f right). A heatmap showed a marked reduction of TGs in ACC1 −/− Th1 cells and moderate reduction of TGs in sgScd2 Th1 cells compared to control Th1 cells or sgFads2 Th1 cells (Fig. 6i). Furthermore, deeper analysis showed that the signal value of TG containing oleic acid was much higher than TG not containing oleic acid, and 85% of the TGs that were reduced in sdScd2 Th1 cells contained oleic acid (Fig. 6j, Supplementary Fig. 6g, and 6h). Taken together, these data indicate that the amount of TG containing OA was strongly affected by the inhibition of MUFA biosynthesis via the deletion of Scd2. We also addressed whether or not a reduction in TG containing MUFA contributes to the induction of ISGs in Th1 cells. When we targeted glycerol-3phosphate acyltransferase (GPAM), which codes the initial enzyme to catalyze the phospholipid, a modest induction of ISGs was observed (Fig. 6k). In addition, gene targeting of Dgat1, which codes diacylglycerol O-acyltransferase1 (DGAT1), upregulated the ISGs expression to levels comparable to those detected in sgScd2 Th1 cells (Fig. 6l). Although we tested the importance of lipolysis for ISGs induction, treatment of lipolysis inducing hormone isoproterenol failed to increase ISGs expression in Th1 cells (Supplementary Fig. 6i). Taken together, these results indicate that inhibition of MUFA synthesis limited the generation of TG containing OA, contributing the enhancement of basal expression of ISGs in Th1 cells.

Discussion
In the present study, we identi ed a molecular mechanism to regulates anti-viral genes via the rewiring of the MUFA metabolism in CD4 + T cells. We initially showed that the pharmacological inhibition or genetic deletion of ACC1 led to spontaneous IFN-I secretion, which induced the expression of ISGs in CD4 + T cells in both an autocrine and paracrine manner. Since the IFNa secreted by these cells can stimulate both themselves and surrounding cells, CD4 + T cells with a reduced activity of fatty acid biosynthesis can also exhibit an increased antiviral response to the surrounding cells. Indeed, genetic deletion or pharmacological inhibition of SCD2 increased anti-viral activity both an in vivo and in vitro experimental settings. Through a combination of CRISPR/Cas9-mediated genome editing technology and a comprehensive lipidomic analysis, we found that the decreased production of MUFA incorporated into TG was crucial for the induction of an anti-viral response in CD4 + T cells. We also found that the changes in the lipid metabolic mode led to the induction of STING and TBK1 phosphorylation via the activation of the cGAS-STING signal pathway. Thus, our ndings provide a novel evidence that the MUFA reduction and IFN signaling pathway in CD4 + T cells are part of a positive-feedback mechanism.
The interrelationship between sterol biosynthesis and IFN-I signaling has been reported in macrophages 6 .
The inhibition of the cholesterol biosynthesis pathway activates the IFN-I response in bone-marrowderived macrophages (BMDMs). Furthermore, it has been also reported that the IFN response in BMDMs reduces the gene expression of cholesterol 25-hydroxylase, causing the suppression of Il1 family-induced in ammation 27 . It is noteworthy that the cholesterol biosynthesis pathway is important for the induction of anti-viral responses in macrophages, whereas the fatty acid biosynthesis pathway is critical for type I IFN responses in CD4 + T cells. Bensinger et al. mentioned that altering the balance between cholesterol biosynthesis and scavenging, rather than reducing the endogenous lipid pool size, is essential for achieving anti-viral responses in macrophages 6 . In contrast, the inhibition of cholesterol biosynthesis did not affect the type I IFN response in CD4 + T cells. Our data further indicated that selective decrease in FA biosynthesis, especially MUFA biosynthesis, activates the type I IFN response in CD4 + T cells and primes these cells for anti-viral immunity. Thus, although dynamic changes in cellular lipid metabolismincluding fatty acid and cholesterol metabolism-are essential for anti-viral immunity, the mode of acquisition and its contribution to the regulation of the type I IFN responses appear to differ among immune cell types. Further detailed studies focused on the difference in the degree of lipid metabolic dependency on the induction of IFN-I responses in acquired and innate immune cells will help expand our understanding of the crosstalk between lipid metabolism and host immune systems.
One striking point revealed in this study is that the limitation of MUFA-containing TG triggers the basal expression of ISGs in CD4 + T cells. When MUFA biosynthesis was inhibited by pharmacological inhibition or genetic deletion of Scd2 in CD4 + T cells, cells spontaneously produced IFN-I, which strongly induced the expression of ISGs in surrounding cells. This was clearly reversed by the extrinsic supplementation of OA. While how MUFA is incorporated into TG or DG is unclear, gene targeting of Dgat1 or Gpam partially augmented the expression of ISGs, implying that these enzymes further metabolize MUFA into TG to Several researchers have also reported on the relationship between lipid metabolism and anti-viral responses. In pDCs, metabolic changes in FAO and OXPHOS induced by IFN-I were required for fully activation 9 . In terms of HCV infections, LDs in host cells were redistributed for the construction of a complex virus assembly. 25-hydroxychoresterol (25HC) reduces the syncytia formation of HIV, which resulted in an impairment of HIV infection. Furthermore, 25HC inhibited HIV in a humanized mouse model, and Ch25h -/mice was more susceptible to MHV68 lytic infection than wild-type mice 28 . HIV infects with CD4 + T cells for a long latency period. During HIV invasion, IFN-I regulates HIV infection in the initial phase 8 . In addition, HIV-1 acquires its envelop from the host cell at sites enriched in lipid raft 29 .
Thus, IFN-I response through lipid metabolism in CD4 + T cells may be important against HIV infection. Further detailed studies focused on the role of lipid metabolism and IFN-I response in T cell during the other kinds of virus infection may lead to the discovery of novel therapeutic targets for the treatment of various viral infection.
In addition to the role of type I IFN in direct regulation of the ISGs expression and anti-viral activity, it has been also reported that type I IFN limited cytotoxic activity of NK cells against antiviral CD8 + T cells during LCMV infection 30,31 . Thus, IFN-mediated protection of CD8 + T cells from NK cell cytotoxicity resulted in the proper generation of antiviral CD8 + T cells. Since both of CD4 + and CD8 + T cells dramatically induced ISGs expression via the modulation of fatty acid metabolism, this phenomenon could confer protection from NK cell-mediated cytotoxicity.
We found that cGAS-STING-mediated TBK1 activation, but not MAVS activation, is involved in the regulation of anti-viral genes caused by a limitation of MUFA in CD4 + T cells. We considered the possibility that a reduction in MUFA might increase the availability of palmitic acid and subsequently augment the palmitoylation of STING 24 . However, supplementation of palmitic acid did not affect the inductions of ISGs in CD4 + T cells. We therefore believe it unlikely that changes in MUFA metabolism are signi cantly altering the palmitic acid availability in CD4 + T cells. cGAMP generated from cytosolic DNA is known to activate the cGAS-STING pathway, which triggers IFN-I production. The accumulation of cytosolic DNA reportedly enhances the IFN response in mice and humans 32,33 . Furthermore, we observed higher amount of cGAMP and cytosolic DNA originated from genome in sgScd2 Th1 cells, indicating that generation of cGAMP was critical for STING activation and the induction of ISGs. A recent study has demonstrated that TG synthesis is coupled with the nuclear membrane remodeling 34 . Thus, our results together with the previous study suggest that organelle interaction between lipid droplets and nuclear membrane control cGAMP-mediated STING activation in CD4 + T cells.
In conclusion, our results demonstrate a novel anti-viral mechanism involving the fatty acid biosynthesis pathway in CD4 + T cells. The decreased production of MUFA caused the activation of the cGAS-STING pathway, resulting in spontaneous IFNa production from CD4 + T cells. In particular, TG containing OA regulated the activation of the cGAS-STING pathway. Since inhibition of MUFA increased basal ISGs expression in tested CD4 + T cell subsets including Th0, Th1, Th2, Th17, Treg, and Th9 cells, these data suggested that SCD2 mediated IFN production occurs in all CD4 + T cell subsets. Thus, our ndings shed light on novel mechanistic insights into the process of lipid metabolism that is essential for cGAS-STING activation, IFN-I response and antiviral activity in CD4 + T cells. Our results underscore the need to understand the role of lipids in the enhancement of basal ISGs expression and the lipid environment that favors host cell defenses.

Mice
The animals used in this study were backcrossed to C57BL/6 mice ten times. Acaca / mice were crossed with CD4-cre mice (Jackson Laboratory) and were maintained on a C57BL/6 background.
BALB/c and C57BL/6 mice were purchased from CLEA Japan. Ly5.1 mice were from Sankyo Laboratories. All mice were used at 6-8 weeks old and were maintained under speci c-pathogen-free conditions. Animal procedures were approved by Animal Ethics Committee of the National Institute of Infectious Diseases, Japan. Animal care was conducted in accordance with the guidelines of Chiba University, KAZUSA DNA research institute, and the Institutional Animal Care and Use Committee of the National Institute of Infectious Diseases, Japan.

Cell Preparation
Splenic naïve CD4 + T cells were obtained by the negative selection using Mojo Sort Mouse CD4 T Cell Isolation Kit (Biolegend) and positive selection using CD62L MicroBeads, mouse (Miltenyi Biotec).

Medium transfer experiments
Naïve CD4 + T cell were cultured described as Cell Preparation with or without TOFA. On day5, supernatant of culture medium was harvested. After 48 hours incubation of MLE-15 cells with culture supernatant, MLE-15 cells were harvested for FACS analysis and RNA isolation.

Infection
In uenza virus was diluted to the appropriate concentration in PBS supplemented with 3% BSA. MLE-15 cells prepared described as Medium transfer experiments were infected with in uenza virus for 3 day at 34°C. Then, MLE-15 cells were xed with 4% PFA and stained with amido black solution. TCID50 was calculated as described previously 35 .
Cas9 mediated-genome-editing The short guide RNA was designed using the online tool provided by CHOPCHOP (http://chopchop.cbu.uib.no) 36 . Freshly isolated naïve CD4 T cells were activated with plate bound anti-CD3 and CD28 antibodies. Cas9 proteins were prepared immediately before experiments by incubating 1μg Cas9 with 0.3 μg sgRNA in transfection buffer at room temperature for 10 min. 24 hours after T cell activation, these cells were electroporated with a Neon transfection kit and device (Thermo).

In vivo Experiment
The in uenza A virus X31 (H3N2) were grown in a 10-d-old embryonated hen eggs and were puri ed through a 10-50% sucrose gradient as described 35 . BALB/c were anesthetized by i.p. injection with sodium pentobarbital and inoculated intra-nasally with 5 lethal dose50 (LD50) of in uenza virus X31 in a volume of 20 μL. mRNA expression of lipid enzymes or cell surface pro les of CD317 on T cells were analyzed on day 6 after X31 infection. To investigate anti-viral activity, in uenza A virus A/Puerto Rico/8/34 (H1N1) was used. BALB/c were anesthetized by i.p. injection with midazolam, domitor, and butorphanol tartrate. In the administrative transfer experiment, 5 x 10 6 of control Th1 or sgScd2 Th1 cells suspended in 30 μl were administered intranasally the day before infection. Infected mice were weighed daily, with 75% weight loss as the criterion for euthanasia. After three kinds of mixed anesthesia, antisedan was also administered intraperitoneally in all experiments.
Isolation of lymphocytes from lung tissues.
Mice were euthanized by anesthesia and their lungs were perfused with 5 mL cold PBS solution through the right ventricle. Lungs were transferred to a tube containing ice-cold digestion buffer [RPMI 1640 supplemented with collagenase type III (200 U/mL; Worthington) and DNase I (200 μg/mL; Sigma-Aldrich)]. Individual lungs were dissociated in 3 mL digestion buffer using a GentleMACS tissue dissociator (Miltenyi Biotec). This was followed by incubation for 30 min at 37 °C with frequent agitation and a nal GentleMACS dissociation. Lung mononuclear cells were separated by centrifugation on Percoll (GE Healthcare).

ELISA for the Measurement of Cytokine or cGAMP Concentration
Cells were cultured as described in section of Cell Preparation. After collecting the samples, centrifugation was performed. The supernatant was used as the sample for IFNα and the cells were used as the sample for cGAMP. The concentration of IFNα or cGAMP was assessed by Mouse IFN Alpha All Subtype ELISA Kit (PBL Assay Science) or by 2'3' -cGAMP ELISA kit (Cayman) as according to the manufacture's protocol.
Quantitative Real-Time PCR Total RNA was isolated with the TRIzol reagent (Invitrogen). cDNA was synthesized with an oligo (dT) primer and Superscript II RT (Invitrogen). Quantitative RT-PCR was performed with the Applied Biosystems StepOnePlus TM Real-Time PCR Systems as described previously. Primers and Roche Universal probes were purchased from Sigma and Roche, respectively. Gene expression was normalized with the Hprt mRNA signal or the 18S ribosomal RNA signal.

Viral genome detection
A lobe of the lung was frozen in liquid nitrogen and then crushed to make an RNA sample. Puri ed RNA was prepared as described in section of Quantitative Real-Time PCR. One-step RT-PCR was conducted using One Step TB Green PrimeScript RT-PCR Kit (Takara) according to the manufacture's protocol. The sequence of primer is as follows. Differentially expressed genes (fold changes >= 2 in ACC1 KO and TOFA -treated Th1 cells or fold changes <= 1.5 in IFNβ-treated Th1 cells) were performed to analyze for GO biological processes using DAVID.

Lipidomics Analysis
Unbiased metabolic pro ling was performed by KAZUSA DNA Research Institute using control, ACC1 -/-, sgScd2, or sgFads2 Th1 cells. At the end of culturing, 10 million cells were spun down and the pellets were washed with PBS (-). After treatment of mixture of chilled methanol (1:13, v/w) and methyl tert-butyl ether, samples were mixture by vigorous vortex and sonication. The required water of water (1:2.5, v/w) was added into sample before vigorous vortex. After the centrifugation, upper layer was collected for LC/MS.

Immuno uorescence microscopy
The cells were xed with 4.0% paraformaldehyde for 10 min, before permeabilized with 0.1% Triton X-100 in PBS for 10 min on ice. After then, sample was blocked with 3% BSA in PBS for 15 min and stained with indicated reagents for 30 min in the dark.

Isolation of cytosolic DNA
Cytosol fraction was prepared with Mitochondria/Cytosol Fractionation Kit (Abcam) according to the manufactured protocol. Brie y, after the treatment of RNaseA, cytosolic DNA was puri ed from cytosolic fraction with QIAquick Gel Extraction Kit (QIAGEN) and quanti ed using Qubit dsDNA HS Assay kit , n=4 biologically independent sample). f, qRT-PCR analyses of the relative expression of fatty acid synthesis genes in lung CD4+ T cell derived from control or X31 infected mice. Relative expression (normalized to S) with SD is shown. Each dot showed averaged expression of genes in single sample (Control, n=3; Infection (+), n=4 biologically independent sample). g, qRT-PCR analyses of the relative expression of genes relevant to lipid metabolism in human Th1 cells with or without IFNβ. Relative expression (normalized to 18S) with SD is shown. Two biological replicates were performed for RNA-sequencing analysis (a-c). More than two independent experiments were performed and showed similar results (e, f =2; d, g=3). Three technical replicates were performed with quantitative RT-PCR (d, f, g).   The MUFA metabolism plays a crucial role in the regulation of the type I IFN response and anti-viral response in CD4+ T cells. a, qRT-PCR analyses of the relative expression of ISGs in Th1 cells treated with 10 μM TOFA, 1 μM Mk-8245, or 30 μM Sc-26196. Relative expression (normalized to Hprt) with SD is shown. b, qRT-PCR analyses of the relative expression of ISGs in control, sgScd2 and sgFads2 Th1 cells.
Relative expression (normalized to Hprt) with SD is shown. c, MLE-15 cells were cultured with culture supernatants of control, sgScd2 and sgFads2 Th1 cells and infected with x31. Virus titers were determined 72 hours p.i. (n=3 per each group biologically independent sample). d, The amount of IFN in the cell supernatant was measured by ELISA. Data are means ± SD. e, f, qRT-PCR analyses of relative expression of ISGs in control, sgScd2 Th1 cells supplemented with 30 μM palmitic acid, 10 μM stearic acid, 30 μM oleic acid (e) or 10 μg/ml cholesterol (f) for 72 hours. g, h, Survival rate (g) or weight change (h) of PR8 infected mice was assessed 1 day after administrative transfer of control or sgScd2 Th1 cells. Relative expression (normalized to Hprt) with SD is shown. More than two independent experiments were performed and showed similar results (a-f = 3). Three technical replicates were performed with quantitative RT-PCR (a, b, e, f). Data is a summary of two independent experiments (g-h). Rewiring the MUFA metabolism triggers cGAS-STING activation in CD4+ T cells. a, b Western blot analysis of phospho-TBK1 (pTBK1) and total TBK1 from control, TOFA-treated, sgScd2, and sgFads2 Th1 cells (a). Band intensity was determined by image j and summary of three independent experiments was shown (b). c, Western blot analysis of pTBK1 and total TBK1 from control, sgScd2, sgScd2/sgTmem173 DKO and sgScd2/sgMavs DKO of Th1 cells. d, Western blot analysis of phospho-STING (pSTING) and total STING from control, sgScd2, sgScd2/sgMb21d1 DKO and sgScd2/sgTmem173 DKO of Th1 cells. e, qRT-PCR analyses of the relative expression of ISGs in control, sgScd2, and sgScd2/Mavs DKO of Th1