Fatty acid metabolism controls Th9 cell differentiation in vitro
First, we confirmed that fatty acid metabolism is involved in the differentiation of murine Th9 cells by examining the expression of four genes (Ldlr, Lrp8, Scarb1, and Vldlr) that encode enzymes involved in fatty acid uptake by means of quantitative real-time PCR (qRT-PCR) analysis. The expression of all four genes was significantly upregulated in Th0 (Th cells generated under neutral conditions) and Th9 cells compared with that in naïve CD4+ T cells, indicating that fatty acid metabolism is indeed involved in the differentiation of murine Th9 cells (Supplementary Fig. 1a).
Next, we confirmed that murine Th cells acquire free fatty acids from the external environment by examining the uptake of fluorescently labelled palmitate (BODIPY FL C16; Invitrogen). Consistent with the results of the qRT-PCR analysis, Th0 and Th9 cells acquired significantly higher levels of palmitate from the external environment than did naïve CD4+ T cells (Fig. 1a), indicating that the fatty acid uptake program is activated during the differentiation of Th0 and Th9 cells.
We then examined IL-9 production by murine Th9 cells and found that the proportion of IL-9 producing cells was significantly higher under fatty acid-free conditions than under control conditions (Fig. 1b). Consistent with this finding, the expression of Il9 mRNA was also significantly increased in murine Th9 cells cultured under fatty acid–free conditions compared with control (Fig. 1c). Similar results were also found in human Th9 cells (Fig. 1d, e). We also examined the mRNA expression of several transcription factors known to be involved in the differentiation of mouse and human Th9 cells and found that the expression of Batf3/BATF3 was significantly increased in both types of Th9 cells cultured under fatty acid–free conditions (Supplementary Fig. 1b, c).
We next examined the involvement of de novo fatty acid biosynthesis in the differentiation of murine Th9 cells. qRT-PCR analysis revealed significant upregulation of several mRNAs encoding enzymes involved in fatty acid biosynthesis, including Acaca, Scd1, Scd2, Acsl3, Elovl1, Elovl5, and Fads2, in Th9 cells compared to naïve CD4+ T cells (Supplementary Fig. 1d). We also observed increased levels of ACC1 and SCD2 in murine Th0 and Th9 cells (Fig. 1f), as well as a large, significant increase in lipid droplet in these cells as compared with that in naïve CD4+ T cells (Fig. 1g).
To confirm that fatty acid biosynthesis plays a role in the differentiation and function of Th9 cells, we treated mouse and human Th9 cells with 5-(tetradecyloxy)-2-furoic acid (TOFA), a pharmacological inhibitor of ACC1. In both the mouse and human Th9 cells, IL-9 production and expression of Il9 mRNA were significantly increased in the TOFA-treated group compared with the untreated control (mouse, Fig. 1h–j; human, Fig. 1k, l). Furthermore, TOFA treatment significantly increased the mRNA expression of several transcription factors associated with the differentiation of Th9 cells: mouse Batf3 and Irf4 and human BATF3, IRF4, and SPI1 (Supplementary Fig. 1e, f).
Together, this series of results indicates that the fatty acid uptake program and de novo fatty acid biosynthesis are both deeply involved in the differentiation and function of mouse and human Th9 cells.
Extrinsic supplementation with fatty acids restores IL-9 production in AcacaΔT Th9 cells
To directly assess the role of ACC1-mediated fatty acid biosynthesis in Th9 cell differentiation, we used mice in which the biotin carboxyl carrier protein domain in the Acaca gene in CD4+ T cells was conditionally deleted by the expression of Cre-recombinase driven by the Cd4 promoter (hereafter referred to as AcacaΔT)26,30. We observed a significant, 2 to 3-fold increase in the proportion of IL-9–producing cells and a significant increase in the expression of Il9 in murine AcacaΔT Th9 cells compared with those detected in murine Acacafl/fl Th9 cells (Fig. 2a–c). Batf3 expression was also significantly increased in murine AcacaΔT Th9 cells compared with that in Acacafl/fl Th9 cells (Supplementary Fig. 2a).
We next performed a lipidomics analysis to clarify which fatty acids were directly involved in the differentiation of murine Th9 cells. A global cellular lipidomics analysis identified 240 lipid species, and the overall lipid profile differed between the cells lacking ACC1 depending on whether they were AcacaΔT Th9 cells or cells treated with TOFA (Fig. 2d, e). Of the 240 identified lipids, the expression of 46 was selectively reduced in AcacaΔT Th9 cells, and the expression of 26 was selectively reduced in TOFA-treated Th9 cells (Fig. 2d, e). We also observed that oleic acid (18:1), palmitoleic acid (16:1), and palmitic acid (16:0) were the three major components of the cellular lipids that were selectively reduced in AcacaΔT Th9 cells (Fig. 2f).
Previously, we reported that extrinsic supplementation with fatty acids restored the function of AcacaΔT Th17 cells, and improved the proliferation, survival, and metabolic reprogramming of TOFA-treated activated CD4+ T cells17,26,31. We therefore next sought to determine whether supplementation of fatty acids would cancel the differentiation induced in Th9 cells cultured under fatty acid–limited conditions. As expected, in AcacaΔT murine Th9 cells, supplementation with oleic acid, the most common component of the monounsaturated fatty acids with reduced expression in AcacaΔT murine Th9 cells, or palmitic acid, the most common component of the saturated fatty acids with reduced expression in AcacaΔT murine Th9 cells, partially suppressed the increase of the proportion of IL-9-producing cells and the increase of Il9 expression, whereas supplementation with docosahexaenoic acid, a typical poly unsaturated fatty acid, did not (Fig. 2g, h and Supplementary Fig. 2b). Similar results were found for TOFA-treated human Th9 cells (Fig. 2i, j). Thus, supplementation with a saturated or monounsaturated fatty acid prevented the augmentation of IL-9 production induced by the pharmacological inhibition or genetic deletion of ACC1 in mouse and human Th9 cells.
ACC1 controls a permissive chromatin landscape at the Il9 gene locus
Since Th17-cell skewing conditions (IL-6 plus TGF-β) are somewhat similar to those for Th9 (IL-4 plus TGF-β), we differentiated naïve CD4+ T cells into Th17 cells in the presence of TOFA in vitro and observed that Th17 cells significantly produced much more IL-9 when Acaca was pharmacologically inhibited (Supplementary Fig. 3a).
To better understand the effects of fatty acid metabolism on the Th9 cell program, we performed an RNA-seq analysis of Th9 and Th17 cells differentiated under control, TOFA-treatment, or AcacaΔT conditions. Principal component analysis revealed that TOFA-treated or AcacaΔT Th9 or Th17 cells formed clusters that were distinct from those of control Th9 cells (Fig. 3a and Supplementary Fig. 3b). A heatmap obtained by unsupervised hierarchical clustering showed many upregulated and downregulated genes in the cells cultured under the TOFA-treatment or AcacaΔT conditions compared with control cells, suggesting extensive Th9 reprogramming under these conditions (Fig. 3b). We found 1056/865 differentially expressed genes (DEGs) among control, TOFA, and AcacaΔT Th9 cells, including 495/449 genes were upregulated, and 479/341 genes were downregulated (Fig. 3b). A similar tendency was observed for TOFA-treated and AcacaΔT Th17 cells (Supplementary Fig. 3c). MA plots showed that Il9 was one of the most highly upregulated genes in TOFA-treated Th9 and Th17 cells (Fig. 3c and Supplementary Fig. 3d).
Murine Th9 cell differentiation is known to be controlled by several Th9-associated transcription factors, including Spi1, Irf4, Batf, Batf3, Stat5, Stat6, Hif1α, and Foxo132,33,34. In TOFA-treated cells in comparison with the control, an increased expression of two of these factors, Irf4 and Batf3, was detected in RNA-seq analysis, as in the earlier qRT-PCR analysis (Supplementary Fig. 1e), but the expression of the other factors remained largely unchanged (Supplementary Fig. 3e). In response to these results, we examined whether CRISPR-mediated deletion of Batf3 in TOFA-treated murine Th9 cells would prevent the increase in IL-9 production; however, no changes in IL-9 production were detected (Supplementary Fig. 3f).
Epigenetic chromatin modifications can selectively control the expression of genes that function in the immune system35. We therefore explored whether ACC1-mediated fatty acid biosynthesis regulates the chromatin status of the Th9-associated gene loci in Th9 cells. We performed chromatin immunoprecipitation sequencing (ChIP-seq) analysis of H3K9 acetylation to evaluate the changes of the epigenetic landscape in detail. Examination of the profile revealed more than 10,000 unique peaks in control and TOFA-treated Th9 and Th17 cells, and the H3K9 acetylation profiles around the transcription start sites of the genes showed marked differences (Fig. 3d, e and Supplementary Fig. 3g, h). We also used ATAC-seq (assay for transposase-accessible chromatin using sequencing) to examine the permissive chromatin landscape in control and TOFA-treated Th9 and Th17 cells. The chromatin in the Il9 promoter region was in a permissive state in TOFA-treated Th9 and Th17 cells (Fig. 3f). Genome-wide H3K9 acetylation and ATAC-seq signal profiles revealed striking increases in the levels and extent of epigenetic marks at the Il9 gene locus in TOFA-treated Th9 and Th17 cells (Fig. 3f and Supplementary Fig. 3i). Furthermore, TOFA-treated Th9 cells showed exceptionally high levels of H3K9 acetylation, which spanned a large genomic segment. To evaluate the importance of histone acetylation for IL-9 production, we examined whether administration of the potent histone acetyltransferase inhibitor curcumin would suppress IL-9 production in TOFA-treated Th9 cells, and found that curcumin did significantly suppress it (Fig. 3g). In addition, treatment with trichostatin A, a representative histone deacetylase inhibitor, significantly increased IL-9 production compared with control, but the increase was not as much as that induced by TOFA treatment (Fig. 3h).
Together, this series of results indicates that the inhibition of ACC1-mediated de novo fatty acid biosynthesis confers a permissive chromatin landscape at the Il9 gene locus that is critical for robust Th9 cell differentiation.
ACC1-mediated fatty acid biosynthesis controls the TGF-β–Smad2/3 pathway and limits Th9 cell differentiation
To understand more about the mechanisms underlying ACC1-mediated fatty acid biosynthesis, we used a combination of genome-wide RNA-seq and ChIP-seq profiling. A Venn diagram of the results of RNA-seq and ChIP-seq analyses showed that the expression of 76 and 109 genes, including Il9, were increased commonly in TOFA-treated murine Th9 and Th17 cells, respectively, as compared with that in control cells (Fig. 4a, b, Supplementary Fig. 4a, b, and Supplementary Table 2). Of the 76 genes with changed expression in Th9 cells, 42 genes were shared with those identified in Th17 cells, nearly half of which were related to the TGF-β signaling pathway and included Il9, Gadd45a, Il1rn, and Tmcc336–39 (Fig. 4b, Supplementary Fig. 4b, and Supplementary Table 2).
To examine how ACC1-mediated fatty acid biosynthesis controls murine Th9 cell differentiation, we investigated the differentiation of TOFA-treated murine Th9 cells cultured at different concentrations of IL-4 and TGF-β. TOFA-treated Th9 cells were able to produce IL-9 even in the absence of IL-4 (Supplementary Fig. 4c). In addition, IL-4 concentration had only a mild effect on IL-9 production, especially in TOFA-treated Th9 cells (Supplementary Fig. 4d). Unlike the limited effect of IL-4 on Th9 cell differentiation, TGF-β significantly increased IL-9 production in a dose-dependent manner, especially in TOFA-treated Th9 cells (Fig. 4c and Supplementary Fig. 4e). Th17 and regulatory T cells, which are both reliant upon TGF-β for their differentiation, exhibited a significant increase of IL-9 production consequent to the treatment with TOFA, whereas no such effect was detected in Th1 cells (Fig. 4d and Supplementary Fig. 4f).
The TGF-β–Smad2/3 signaling pathway regulates several transcription factors that participate in the differentiation of Th9 cells, including PU.1 and Batf37. We therefore next assessed the involvement of Smad2/3 in TOFA-treated murine Th9 cells. Phosphorylation of Smad2/3 in these cells was significantly higher than in control Th9 cells (Fig. 4e). Consistent with this observation, immunofluorescence microscopy revealed that nuclear localization of Smad2/3 was enhanced by treatment with TOFA, and that the effect was cancelled by oleic acid supplementation in TOFA-treated murine Th9 cells (Fig. 4f). Furthermore, CRISPR/Cas9-mediated double knockout of Smad2/3 in TOFA-treated murine Th9 cells resulted in a significant reduction of IL-9 production (Fig. 4g, h). We also found that double knockout of Smad2/3 cancelled TOFA-mediated H3K9 acetylation at the Il9 gene locus (Fig. 4i).
The TAK1–ID3 pathway, which is a SMAD-independent TGF-β signaling pathway, also contributes to the differentiation of Th9 cells10,11. Thus, we investigated whether this pathway could be involved in the Th9 cell differentiation induced by inhibition of ACC1. Consistent with the previous reports, TAK1 inhibitor (5Z-7-oxozeaenol; MCE) significantly decreased IL-9 production in control Th9 cells but had minimal effect on IL-9 production in TOFA-treated cells (Supplementary Fig. 4g).
ACC1-mediated fatty acid biosynthesis inhibits Th9 cell differentiation via retinoic acid–retinoic acid receptor-alpha signaling
To gain more insight into the effects of inhibition of fatty acid biosynthesis in Th9 cells, we evaluated the DEGs in TOFA-treated or AcacaΔT Th9 and Th17 cells. A gene set enrichment analysis revealed statistically significant enrichment of retinoic acid regulatory genes in the TOFA-treated and AcacaΔT Th9 cells (Fig. 5a and Supplementary Fig. 5a). In particular, the expression of a group of genes whose expression is suppressed in the presence of retinoic acid (RA) in Th9 cells was found to be upregulated by TOFA treatment (Fig. 5b and Supplementary Fig. 5b)33.
Because RA is known to act through its receptor, RARα, to inhibit the Th9 transcriptional program33, we hypothesized that RA–RARα signaling was involved in ACC1-mediated suppression of Th9 cell differentiation. Indeed, RA administration significantly decreased IL-9 production in TOFA-treated Th9 cells (Fig. 5c). Similarly, treatment of TOFA-treated Th9 cells with BMS753, a selective agonist of RARα33, significantly suppressed Th9 cell differentiation (Fig. 5d). Furthermore, Il9 expression was also significantly reduced by administration of RA or BMS753 in TOFA-treated Th9 cells (Fig. 5e, f). Consistent with these results, immunofluorescence microscopy showed that nuclear localization of Smad2/3 in TOFA-treated Th9 cells was prevented by treatment with RA or BMS753 (Fig. 5g). Conversely, BMS195614, a selective RARα antagonist, has been shown to enhance IL-9 production33. Therefore, to further evaluate the effect of RA–RARα signaling on fatty acid metabolism, BMS195614 was added to TOFA-treated Th9 cell cultures supplemented with oleic acid. Oleic acid abrogated IL-9 production by TOFA-treated Th9 cells, but this was not reversed by the BMS195614 treatment (Fig. 5h, i).
Together, these data indicate that ACC1-mediated fatty acid biosynthesis inhibits IL-9 production via RA–RARα signaling, at least in part, by reducing the nuclear localization of Smad2/3 during Th9 cell differentiation.
Pharmacologic inhibition of ACC1 enhances Th9 cell differentiation and facilitates IL-9–dependent anti-tumor activity in vivo
Having established a function for ACC1-dependent fatty acid biosynthesis in Th9 cell differentiation in vitro, we next determined the roles of ACC1 in Th9 cell–mediated anti-tumor activity in vivo. Because previous studies have shown that Th9 cells have superior anti-tumor properties compared to those of Th1 or Th17 cells upon adoptive cell transfer40–42, we first used an experimental mouse model of melanoma in which B6 mice were injected with B16-OVA (ovalbumin (OVA)-transfected B16F10) melanoma cells and treated with adoptive transfer of OVA-specific OT-II transgenic Th9 cells five days after tumor inoculation (Supplementary Fig. 6a). Consistent with our in vitro experiments, mice that received TOFA-treated cells, which had increased IL-9 expression, showed significant tumor regression as compared with mice that received untreated Th9 cells (Fig. 6a, b and Supplementary Fig. 6b). Analysis of the tumor-infiltrated cells revealed that administration of the TOFA-Th9 cells resulted in significantly increased infiltration of CD45+ cells and CD8+ T cells as compared with administration of untreated Th9 cells (Fig. 6c, d). We also observed that the numbers of tumor-infiltrating granzyme B- or interferon gamma (IFNγ)–producing CD8+ T cells were significantly increased by administration of TOFA-Th9 cells compared with administration of untreated Th9 cells (Supplementary Fig. 6c). Furthermore, the number of tumor-infiltrating CD4+ T cells was also significantly increased by administration of TOFA-Th9 cells compared with administration of untreated Th9 cells (Supplementary Fig. 6d). The tumor-infiltrating CD4+ T cells in the TOFA-Th9 group contained more transferred cells than did those in the control Th9 group (Supplementary Fig. 6e). The significantly stronger anti-tumor activity of TOFA-treated Th9 cells was also observed in a MC38-OVA colon adenocarcinoma tumor model (Fig. 6e, f and Supplementary Fig. 6f, g). Also, in agreement with the results obtained with the B16-OVA model, tumor-infiltrating CD45+, granzyme B- or IFNγ-producing CD8+ T, and CD4+ T cells were significantly increased in the TOFA-treated Th9 group compared with administration of untreated Th9 cells (Fig. 6g, h and Supplementary Fig. 6h, i).
Checkpoint blockade therapy activates anti-tumor immunity by targeting proteins that inhibit T cell proliferation and function43. We therefore examined combination therapy with TOFA-treated Th9 cell transfer and checkpoint antibody blockade. We transplanted B16-OVA cells into mice that received Th9 cells with or without TOFA treatment and then treated these mice with isotype or PD-1–blocking antibodies (Supplementary Fig. 6j). Combination therapy with TOFA-treated Th9 cells and PD-1 blockade resulted in significant tumor regression as compared to transfer of TOFA-treated Th9 cells alone (Fig. 6i, j and Supplementary Fig. 6k, l). Surprisingly, this combination therapy fully eradicated the tumor in 1 of 5 mice.
Th9 cells are known to elicit strong host anti-tumor CD8+ cytotoxic T lymphocyte (CTL) responses by promoting Ccl20/Ccr6-dependent recruitment of dendritic cells to the tumor tissues via IL-9 production40. To further investigate the link between Th9 cells and ACC1-mediated fatty acid biosynthesis in the cancer setting, we next treated tumor-inoculated mice with IL-9–neutralizing antibodies (αIL-9). One day before Th9 cell transfer, B6 mice were given one dose of cyclophosphamide to induce temporary lymphopenia conditions, which is frequently induced as part of clinical adoptive cell therapy protocols to promote homeostatic proliferation of transferred T cells (Supplementary Fig. 6m)41. Mice also received adjuvant OVA peptide–pulsed dendritic cell vaccination on the day of transfer, which is used to boost the anti-tumor response during adoptive cell therapy 41. TOFA-treated Th9 cells mediated a tumor regression that resulted in long-term survival (Fig. 6k). Anti-IL-9 antibody treatment significantly abrogated the beneficial effect of TOFA in Th9 cells, indicating that the anti-tumor activity is likely driven mainly by enhanced IL-9 production (Fig. 6k and Supplementary Fig. 6n, o).
Finally, we performed publicly available dataset analysis of RNA-seq data derived from 73 patients with melanoma who received anti-PD1 antibodies such as nivolumab or pembrolizumab (Fig. 6l)44,45. We sought to determine whether Th9-related genes especially up-regulated genes by ACC1 inhibition including IL9, BATF3, TNFSF8, SPI1, and IRF4 was associated with a 10-year overall survival (OS) or progression free survival (PFS). The PFS and OS were significantly longer for anti-PD1 antibody-treated patients with high expression of IL9/BATF3/TNFSF8/SPI1/IRF4 Th9 phenotype as compared with those with low expression (Fig. 6l).
Taken together, these data indicate that targeting ACC1-mediated augmentation of IL-9 production by Th9 cells could be a new tool for the clinical study of combination therapy with PD-1 blocking antibody.