In T cells whereas SETDB is implicated in OX

In T cells, whereas SETDB1 is implicated in OX40-dependent repression of the Il17a locus in Th17 cells (Xiao et al., 2016), SUV39H1 controls Th2 cell stability by depositing H3K9me3 at the Ifng promoter (Allan et al., 2012). However, the deregulation of the Ifng locus observed in Suv39h1−/− cells cannot by itself explain a loss of Th2 cell integrity. Other critical Th1-cell-lineage-specific loci might therefore be controlled by H3K9me3-dependent repressive mechanisms. In addition, whereas a clear H3K9me3 signal is detected at the gene encoding T-bet in Th2 cells, SUV39H1 has no effect on the deposition of the repressive mark at this locus (Allan et al., 2012). Together with the fact that H3K9me3 disappearance at euchromatin and facultative heterochromatin is limited in SUV39H1-deficient cells (Peters et al., 2002), these observations suggest that other H3K9me3-dependent epigenetic pathways critically control Th2 cell stability. Here, we examined the effects of SETDB1-dependent H3K9me3 deposition on CD4+ T cell differentiation. We found that SETDB1 restricts Th1 cell priming and ensures Th2 cell integrity. Unlike their wild-type counterparts, Setdb1−/− Th2 cells readily expressed Th1-associated genes when exposed to the Th1-instructing cytokine interleukin-12 (IL-12). SETDB1 repressed Th1-related loci by depositing H3K9me3 at a subset of ERVs that flank and repress Th1 enhancers or behave themselves as cis-regulatory elements of Th1 genes. Our findings reveal a repertoire of ERVs that have been co-opted to behave as Th1-specific cis-regulatory modules and outline a model wherein H3K9me3 deposition by SETDB1 locks the Th1 gene network and ensures Th cell lineage integrity by repressing these repeat elements.
Results
Discussion Up to 10% of the mouse genome is composed of ERVs, which have long been considered to be junk DNA sequences. Recently, however, regulatory functions over gene expression have been assigned to transposable elements. In mouse CD4+ T cells, we documented that a set of ERVs enriched in Fmoc-Arg(Pbf)-OH for pro-Th1 transcription factors overlapped or flanked the enhancers of genes from the Th1 cell transcriptomic signature. We further showed that the accessibility of these repeat elements was regulated at the epigenetic level by SETDB1. Indeed, in Setdb1−/− cells, the lack of deposition of H3K9me3 at this subset of ERVs correlated with their activation and with the increased expression of their closest genes. At the cellular level, this deregulation of gene expression translated into increased Th2 cell plasticity and enhanced Th1 cell priming. Together, these data suggest that SETDB1 controls Th2 cell integrity by repressing a restricted and cell-type-specific repertoire of ERVs. Although the use of Setdb1−/− cells allowed us to establish cause and effect links between SETDB1 depletion, H3K9me3 disappearance, and ERV de-repression, we did not strictly demonstrate that the ERVs marked by H3K9me3 in Th2 cells acted as Th1 gene enhancers. The most direct way to prove that SETDB1 controlled the Th1 gene expression program through the regulation of ERVs that behave as Th1 gene enhancers would have been to selectively inactivate the transposable elements marked by H3K9me3 in Th2 cells by using CRISPR/Cas9. Unfortunately, because of the large number of ERVs and the absence of a consensus sequence to target, we have not been able to perform this experiment. ERVs potentially control Th1 gene expression through two non-mutually exclusive mechanisms: they behave as cis-regulatory elements or they regulate chromatin accessibility at nearby enhancers. Our evidence that the binding motifs for critical Th1-associated transcription factors were enriched in H3K9me3+ ERV sequences suggests that the ERVs directly act as cis-regulatory elements. The existence of such a subset of regulatory ERVs, which might have shaped the Th1 transcriptional network over time, is supported by a recent study showing that ERVs containing binding sites for IFN-induced transcription factors are necessary for AIM2 inflammasome activation (Chuong et al., 2016). However, when we fractionated the repertoire of H3K9me3+ ERVs associated with Th1 enhancers in Th2 cells, we observed that most of them only flanked STAT1 or STAT4 binding sites. This result suggests that ERVs regulate the Th1 gene network mainly by modulating the activity of the Th1 enhancers located in their vicinity. Interestingly, the H3K9me3+ ERVs that flanked Th1 enhancers accumulated at a distance of 3–5 kb from the STAT peaks. This distribution of the retrotransposons overlaps the distribution of the H3K9me3 signal observed on the flanking regions of enhancers whose activity is regulated by this histone mark in dendritic cells (DCs) and fibroblasts (Zhu et al., 2012). Although the authors of that study do not implicate SETDB1 in H3K9me3 deposition and do not identify ERVs as the targeted genomic elements, they correlate the accumulation of H3K9me3 at this location with the repression of adjacent enhancer activity. This observation reinforces (and extends to other cell types) our model supporting that retrotransposons are the genetic elements that are targeted by the H3K9me3-dependent silencing machinery to regulate enhancer activity in a cell-type-specific manner. The underlying molecular mechanism probably relies on local heterochromatin spreading from ERVs to nearby regulatory elements, as suggested by our ChIP-seq data and by studies from the literature (Rebollo et al., 2011).