SETDB is ubiquitously expressed in mouse tissues and

SETDB1 is ubiquitously expressed in mouse tissues, and more than 900,000 ERVs are dispersed through the mouse genome. Nevertheless, the repertoire of ERVs under the control of SETDB1 was limited and highly cell-type specific. Only 17,349 ERVs were associated with H3K9me3 domains in Th2 cells. The vast majority of these ERVs were not associated with H3K9me3 domains in white adipose cells. Moreover, the biological functions of genes associated with the ERVs marked by H3K9me3 in Th2 and white adipose l-ascorbic acid sale were fully different: they were associated with genes involved in immune processes in differentiated lymphocytes, whereas they had no direct link with immunity in adipocytes. The enrichment of H3K9me3 at a specific set of ERVs might be explained by the fact that SETDB1 is recruited to the chromatin by Krüppel-associated box zinc-finger proteins (KZFPs) that use the scaffold protein TRIM28 as a molecular intermediate. The mouse genome encodes hundreds of KZFPs, whose expression depends on the cell type and its physiological state (Imbeault et al., 2017). These transcriptional regulators have different DNA binding sites, and ERVs are one of their main genomic targets. Indeed, KZFPs have co-evolved with transposable elements and use evolutionarily conserved regions located mainly within their regulatory sequences to control gene expression (Chuong et al., 2017, Imbeault et al., 2017). The cell-type-specific and SETDB1-dependent H3K9me3 deposition that we observed at ERVs in Th2 cells was therefore probably orchestrated by a specific set of KZFPs that await identification. Of the three lysine methyltransferases from the SUV39H family, only SETDB1 was necessary for silencing ERVs in differentiated lymphocytes. This finding is consistent with those obtained in neural progenitor cells and immortalized mouse embryonic fibroblasts, in which SUV39H1 deficiency does not severely affect ERV silencing (Bulut-Karslioglu et al., 2014). Although they apparently argue against a direct collaboration of SUV39H1 and SETDB1 in H3K9 trimethylation at ERVs, our data do not fully exclude cooperation between these two enzymes in regulating Th2 cell commitment. In fact, the H3K9me3-dependent epigenetic regulation of CD4+ T cell differentiation involves both SUV39H1 and SETDB1. In differentiated Th2 cells, SUV39H1 controls H3K9me3 deposition at the Ifng promoter (Allan et al., 2012), and our data demonstrate that SETDB1 regulates the entire Th1 gene network through repression of ERVs overlapping or flanking Th1-specific enhancers. To guarantee Th2 cell stability in a changing environment, two non-redundant epigenetic silencing pathways therefore converge to lock the Th1 transcriptional program at different genomic locations. In conclusion, our data support that SETDB1 controls CD4+ T cell identity by repressing ERVs that flank or overlap Th1-specific enhancers. This enzyme is thus a potential target for drugs that might be useful, for example, for promoting Th1 cell differentiation in various infectious diseases or preventing harmful Th2 responses in allergic disorders.
STAR★Methods
Acknowledgments We greatly acknowledge T. Jenuwein (Max Planck Institute, Freiburg, Germany) for providing the Suv39h1-deficient mice. We also thank F.-E. L’Faqihi-Olive, V. Duplan-Eche, and A.-L. Iscache for technical assistance at the flow-cytometry facility of INSERM U1043, the personnel of the US006 ANEXPLO/CREFRE animal facility for expert animal care, and D. Rozet for administrative assistance. We are grateful to the GeT and Bioinformatics platforms from Genotoul (Toulouse, Région Occitanie, France) and to the Transcriptomic & Genomic Platform Marseille Luminy (Marseille, France) for providing sequencing, computing, and storage resources. We acknowledge M. Lebeurrier for bioinformatic analysis at the Genomic and Transcriptomic platform from the Center for Physiopathology of Toulouse Purpan (Toulouse, Région Occitanie, France). We are also grateful to members of the T-Cell-Mediated Immune Tolerance laboratory for advice and discussions. We would like to thank D. Dunia for his careful reading of the manuscript. This work was supported by Agence Nationale de la Recherche JCJC “EpiTreg” (ANR-14-CE14-0021-01) to O.P.J., Région Occitanie NVEQ 2014 to O.P.J., and Fondation pour la Recherche Médicale (FRM) AJE201212 to O.P.J. B.B. was funded by a fellowship from the Association pour la Recherche sur le Cancer (ARC, DOC20160604329). A.M. was supported by fellowships from the Région Midi-Pyrénées and FRM (FDT20170437348).