Epigenetic regulation is thought to determine the individual response to environmental changes. So far, the underlying biological regulation is not fully understood. We hypothesized that enhancer activity might be able to (1) precede future transcriptional response and (2) to memorize environmental changes. 

To model such an adaption, we treated C57BL6 mice with either High Fat Diet (HFD) for 10 weeks or Low Fat Diet (LFD) as a control. Additionally, we included a group of reversed feeding (HFD for 5 weeks and LFD for 5 weeks). Based on bodyweight, the last group completely recovered from HFD-induced phenotype (LFD (n=15): 27.76g ± 0.37g, HFD (n=15): 32.93g ± 0.59g, Reverse (n=15): 28.01g ± 0.67g; mean ± SEM). Using ChIPseq (chromatin immunoprecipitation with massively parallel DNA sequencing) from isolated cardiomyocytes, we found 330 H3K27ac-positive enhancers after HFD (FDR < 0.1). Only one large (49kb) superenhancer (SE), containing one coding region (Inhibitor of DNA-Binding 1, Id1) and two H3K4me1-positive enhancers (‘ME1’ and ‘ME2’) were persistently activated despite reverse feeding. To further investigate the biological relevance of this epigenetic regulation, we generated three cardiomyocyte-specific knockout mouse lines by using CRISPR/Cas9 (Id1-cKO, ME1cKO and ME2-cKO). 

Id1-cKO and ME-cKOs only showed minimal changes of left ventricular ejection fraction (LVEF) under HFD conditions compared to WT controls (cre+/loxP+ cre-). On a transcriptional level, HFD leads to a significantly upregulation of a specific gene program in the heart from WT animals (n=6, DESeq2 FDR<0.05) but not in ID1-cKO. These genes (e.g., Aldh1l2 fc WT 1.47 vs. fc KO 0.87, Pdk1 fc WT 1.21 vs. fc KO 0.78, Acot3 fc WT 2.31 vs. fc KO 1.58, Cpt1b fc WT 1.28 vs. fc KO 1.02 and Gpam fc WT 1.52 vs. fc KO 0.75) are known to regulate fatty acid metabolism (FAM). Strikingly, single deletion of one H3K4me1-positive enhancer impaired the transcriptional response to the same extent. This Id1- independent regulation is indicating an enhancer-dependent mechanism. To identify the upstream regulator of the SE, we performed a pulldown of nuclear proteins, bound to biotin-labeled ME2 in vitro. From 514 uniquely bound epigenetic modifiers (e.g., EHMT1, EHMT2), we found 21 transcription factors. We performed a complementary transcription factor analysis of the promoter regions of the enhancer-dependent FAM-genes. By overlapping mass spectrometry data with bioinformatic in silico analysis, we identified the retinoid acid receptor (ROR) binding motif (p=1e-13) and the related receptor (RORa) as the regulatory element of the SE. Regulatory capabilities were validated by promoter Luciferase-assays. Functionally, the PDK1/4-phosphorylation target ‘pyruvate dehydrogenase’ was hyperphosphorylated at Serin 293 in WT animals, but not in ME2-cKO (PDH pS293/PDH: FC WT HFD/LFD: 6.46; FC ME2 HFD/LFD: 1.84, p=0.002, Mann-Whitney Test), indicating a missing metabolic flexibility upon SE deletion.

In human myocardial samples from obese donors (n= 6, mean BMI: 36.6 kg/m² ± 1.7 SEM), Id1, ME1 and ME2 were hyperacetylated at H3K27 compared to samples from normal weighted donors (n=4, mean BMI: 21.2 kg/m² ± 1.1 SEM).

In conclusion, we discovered an evolutionary conserved mechanism of obesity-induced RORa-dependent epigenetic imprinting regulating adaptive FAM and glycolysis in the heart.