Estrogen receptor activation remodels TEAD1 gene expression to alleviate hepatic steatosis

Figures

• 

• 

• 

  Figure 1. Male mice are severely affected by the high-fat diet. (A) Schematic representation of the mouse experimentation. Five-week-old female (f) and male (m) C57BL/6 mice (n = 4) received either a control (CD, 10% fat) or high-fat diet (HFD, 60% fat) for 13 weeks. HFDm subgroups were injected with estrogen receptor α (ERα, E2 or PPT) or ERβ (E2, DPN or DIP) agonists every other day from weeks 15 to 18. Isolated livers were histologically and molecularly assessed. (B) Liver cross-sections of female (left) and male (right) mice on different diets and ER-agonist treatments were stained with hematoxylin and eosin. Cross-sections of all four mouse replicates are shown in Fig. EV1D. Scale bar: 50 μm. (C) A factorial map of the principal components (PC) analysis separates global gene expression levels. The percentage of PC variance is shown (parentheses). Color-coded small circles illustrate individual mice on different diets and treatments. Color-coded large ellipses group mice by sex, diet, and treatment. (D) Horizontal bars present the number (highlighted) of downregulated (blue) and upregulated (red) genes for sex (purple), diet (yellow) or treatment (blue) comparisons (n = 4, except PPT:…

• 

  Figure EV1. Male but not female mice exhibit HFD-induced liver steatosis. (A–C) Bar charts display physiological parameters assessed upon sacrifice in female (f) and male (m) mice (n = 4 mice per condition, +SD), which were previously examined (Hases et al, 2020). (A) Total weight, (B) liver weight and (C) blood glucose after 2 h fasting were measured. Color gradient indicates female and male mice on different diets (CD or HFD) and male mice upon ERα/β-agonist treatments (DPN, DIP, E2 and PPT). P values highlight non-significant, and asterisks indicate significant differences (*P < 0.05, **<0.01, ***P < 0.005, one-way ANOVA followed by Tukey’s post-hoc test). (D) Hepatic cross-sections of four female and male mice on different diets (CD or HFD) and HFD males treated with an ER-agonist (DPN, DIP, E2 or PPT) were stained with hematoxylin and eosin. Images shown in Fig. 1B are highlighted in black boxes. Scale bar: 100 μm. (E) Two-way Venn diagram (left) shows the intersection and number (n) of genes deregulated exclusively in HFD male (top) or HFD female (bottom) mice compared to CD or deregulated in both sexes (middle). Bubble plots (right) show the top eight gene ontology terms (biological…

• 

  Figure 2. ERα/β-agonist treatment largely reverts HFD-induced transcriptome alterations in males. (A) Line charts depict four clusters (gray-scaled) of gene expression trends (z-score) for unified deregulated genes (DEGs, n = 1477) in mice on different diets and ER-agonist treatments (color-coded, n = 4, except PPT: n = 3). Cluster centroid (dashed black line) represents all deregulated genes (gray lines). Number of genes per cluster is shown (parentheses). (B) Three-way Venn diagram (left) presents intersections of unified DEGs (clustered in (A)). Genes are categorized (right) into non-reverted (rose), reverted (denim), ERβ-specific (DPN-DIP, ocher) and ERα-specific (E2-PPT, violet) gene sets. Horizontal bar chart displays the proportional occurrences of gene sets in the four clusters (gray scale as in (A)). Number (n) indicates gene set size. (C) PCA factorial map separates the semantic space of enriched gene ontology (GO) terms in reverted and ERβ-specific gene sets (circles). Enriched GO terms are collapsed at the parent term level and separated based on similarity (x–y axis). Circle size corresponds to number of enriched GO terms. A hypergeometric test with Benjamini–Hochberg…

• 

  Figure EV2. ERα/β-agonist treatments control molecular responses. (A, B) Four-way Venn diagrams show intersections and numbers (n) of gene sets that are either (A) upregulated (red arrow) or (B) downregulated (blue arrow) by ER-agonist treatments compared to HFD in male mice (n = 4 mice per condition). (C) Horizontal bar plots (top) display the proportional frequency of up- (red) and downregulated (blue) genes in male mice on HFD compared to HFD male mice treated with an ERα/β-agonist (DPN, DIP, E2 or PPT). Horizontal bar plots (bottom) show the respective occurrences of the deregulated genes upon ER-agonist treatment in a diet comparison of male CD versus HFD. Genes significantly higher in CD male (red), higher in HFD male (blue) or unaltered (gray) are shown. Alluvial line width connecting upper and lower side indicate how many HFD-deregulated genes are recovered upon treatment (from red to red or blue to blue). Genes indicated in light gray are unchanged by HFD. Genes changed in both comparisons but not recovered upon treatment (from red to blue or blue to red) were categorized as non-reverted. (D) Bubble plot displays significantly enriched gene ontology terms (biological processes) per…

• 

  Figure 3. ERα/β-agonist treatment reverts HFD-induced transcriptional changes by affecting central cellular pathways and liver cell types. (A) Network connects major Reactome pathway clusters (numbered nodes). Colored bars inside each node present pathway cluster enrichment for CDm and ER-agonist-treated HFDm compared to HFDm (normalized enrichment score, NES). Edges connect positively-correlated (green) or negatively-correlated (black) nodes based on NES profiles. (B) UMAP space-projected enrichment plots highlight liver cell types with enhanced signal for the gene sets (defined in Fig. 2B). Hepatocyte nuclear fraction (snRNA-seq) is labeled. (C) Spatial transcriptomics maps show liver zonation patters of the gene sets (defined in Fig. 2B). (D) Bubble plots display activity scores of altered pathway clusters (A) across all liver cell types in control (left), HFD (middle) male mice and their differences (right). Color-code and circle diameter for enrichment score: low (yellow, narrow), high (black, wide). Arrows and enrichment change indicate higher abundance in control (red) and HFD (blue) mice.

• 

  Figure EV3. ER-responsive molecular gene signatures show cell type-specificity and are maintained in primates. (A) Single-cell map projected in UMAP space displays reference annotation of liver cell types (numbered according to Fig. EV3B). (B) Bar plots show the relative abundance in percent of distinct liver cell populations in control (left) and MASLD (middle) mice as well as their difference (right). Arrows indicate higher abundance in control (red) and MASLD (blue) mice. (C) Enrichment maps present liver cell type specificities of murine gene sets (defined in Fig. 2B) in human (top) and macaque (bottom) livers. Four relevant mouse- and primate-maintained cell types are labeled. (D) Enrichment maps show signal distribution of enriched pathways across liver cell types in mouse. (E) Spatial transcriptomics maps display liver zonation patters of the selected pathways. (C–E) Enrichment score: low: yellow, high: black.

• 

  Figure 4. ERα/β-agonist treatment recovers HFD-induced changes at enhancers and promoters. (A) Metaplots show H3K27ac read aggregation in promoters (left) and enhancers (right), centered at the peak summits. Number (n) indicates significant H3K27ac signal gains or losses in livers of HFDm compared to CDm. The average signal is depicted (n = 3 mice per condition). (B) Genome browser view (mm10) illustrates genomic region around the Tead1 gene locus. Black boxes represent exons and UTRs. Arrows indicate gene transcription directionality. The scale bar shows genomic region length in kilobases (kb). Black arcs display promoter-capture Hi-C (CHi-C) 3D connections. Genomic regions are enriched for CTCF (black peaks) with CTCF motif orientations determined with FIMO (plus or minus symbols), ERα (black peaks), ERα input (black peaks), significant ERα peaks (black insets), H3K27ac (color-coded peaks) in CDm, HFDm and ER-agonist-treated HFDm, H3K4me3, and H3K4me1 (horizontal gray bar; dark: high, light: low). One replicate per condition is shown. The y axis of each track specifies normalized read density. Genomic location of enhancers (numbered from 1 to 4) paired with the Tead1 gene locus are…

• 

  Figure EV4. ERα/β-agonist treatments alter metabolic gene regulation. (A) Histograms show frequency of genome-wide (top) or differentially acetylated (DAc, bottom) promoters (teal) and enhancers (dark blue) (y axis) with respect of distance to the closest annotated transcription start site (TSS, log10 scale, x axis). Median distance from nearest annotated TSS to promoters (teal) and enhancers (dark blue) is highlighted. (B) Stacked bar plot indicates percentage of DAc or genome-wide (all or downscaled to the number of DAc) promoter and enhancer elements binned according to the distance to the closest annotated TSS in bp. The numbers (n) of DAc and all promoter and enhancer regions are specified in parenthesis. (C) Box plots illustrate number of normalized (1× genome coverage) reads in peaks (log2) for the same regions and diet comparisons as in Fig. 4A (n = 3 mice per condition). Each box represents interquartile range (IQR) divided by the median (horizontal line), and whiskers span a maximum of 1.5×IQR. Outliers (circles) are shown. (D) Workflow displays identification of estrogen-sensitive enhancer-gene pairs (ES-E-Gs). First, for each enhancer, which is DAc in CDm versus HFDm (n = 2181),…

• 

  Figure 5. ER-sensitive genes are associated with MASLD progression and reveal TEAD1 as a clinical target. (A) Heatmap displays changes in expression levels for the 45 orthologous ES-E-G genes in MASLD patients (panels 1–3, n = 216) and mice (panels 4 and 5, n = 4). Color gradient indicates z-score-normalized gene expression counts (blue: low, red: high). Four k-means clusters group genes by expression in healthy (CTRL), MASL and MASH patients (panel 1) as well as patients with different NAS (early (E): NAS0-1, moderate (M): NAS2-6, advanced (A): NAS7-8, panel 2) and fibrosis stages (E: F0-1, M: F2, A: F3-4, panel 3). Expression levels of the 45 genes in HFDm (panel 4) and HFDf (panel 5) mice are shown. Color codes distinguish downregulated and upregulated genes in HFDm versus CDm (purple: low, green: high) and HFDf versus CDf (brown: low, gray: unchanged, orange: high). Gene names follow human nomenclature. (B) Box plot shows CPM-normalized TEAD1 gene expression in the MASLD patient cohort depicted in (A). Each box indicates the interquartile range (IQR), median (horizontal line) and 1.5×IQR (whiskers). (C) Bar chart displays TPM-normalized Tead1 gene expression in male mice (n = 4 per…

• 

  Figure EV5. TEAD inhibition changes key cellular processes in the liver. (A) Bar plot displays mean of basal, minimum, and maximum oxygen consumption rate in HepG2 cells with control (siNT, dark gray) and siRNA-mediated TEAD1-KD (siTEAD1, light gray) (n = 63 per condition, +SD). Dots indicate single wells across two biological replicates. Significant P values are indicated (two-sided t test). (B) Bar plot shows the adjusted P values (log10) of the top 12 KEGG pathways enriched for genes deregulated upon TEADap treatment. Red vertical line indicates adjusted P value threshold (P < 0.05). Color gradient indicates gene number per pathway (low: gray, high: black). One-sided Fisher’s exact test with Benjamini–Hochberg correction was used for the overrepresentation analysis. (C) Heatmap displays gene expression changes of the top 12 KEGG pathways altered upon TEADap treatment. Left panel indicates normalized gene expression (z-score) in free fatty acid-fed controls, TEADsf and TEADap treatments (low: blue, high: red). Right panel shows log2FC for ER-agonist treatments compared to untreated male mice on HFD (low: green, high: brown). Gene names follow human nomenclature. Ortholog absence in mouse…