Dnmt3a is an epigenetic mediator of adipose insulin resistance

  1. Dongjoo You
  2. Emma Nilsson
  3. Danielle E Tenen
  4. Anna Lyubetskaya
  5. James C Lo
  6. Rencong Jiang
  7. Jasmine Deng
  8. Brian A Dawes
  9. Allan Vaag
  10. Charlotte Ling
  11. Evan D Rosen  Is a corresponding author
  12. Sona Kang  Is a corresponding author
  1. University of California, Berkeley, United States
  2. Lund University Diabetes Centre, Scania University Hospital, Sweden
  3. Beth Israel Deaconess Medical Center, United States
  4. Broad Institute of Harvard and MIT, United States
  5. Weill Cornell Medical College, United States
  6. Rigshospitalet, University of Copenhagen, Denmark
  7. AstraZeneca, Innovative Medicines, Sweden
8 figures and 5 additional files

Figures

Figure 1 with 1 supplement
Increased adipose expression of Dnmts in obesity.

(A) Dnmt mRNA expression in adipose from chow- and high fat-fed (HFD) C57Bl/6 mice (A); n = 6 mice, p<0.05, Student’s t-test, mean ±s.e.m.) and (B) in adipose tissue from ob/+and ob/ob mice in the …

https://doi.org/10.7554/eLife.30766.002
Figure 1—figure supplement 1
Expression of Dnmts in various metabolic tissues.

(A) Dnmt mRNA expression in inguinal WAT (iWAT, n = 6), BAT (n = 6), and mesenteric WAT (mesWAT, n = 5) from chow- and high fat-fed C57Bl/6 mice (p<0.05, Student’s t-test, mean ±s.e.m.). (B, C) Dnmt

https://doi.org/10.7554/eLife.30766.003
Figure 1—figure supplement 1—source data 1

Figure 1—figure supplement 1—source data.

https://doi.org/10.7554/eLife.30766.004
Figure 2 with 2 supplements
Dnmt3a is sufficient and necessary to mediate cell autonomous IR in vitro.

(A) Two independent hairpins against ​Dnmt3a (versus scrambled control shRNA; shScr) were delivered to mature 3T3-L1 adipocytes (Day 8) via lentiviral transduction. Cells were then treated with ​Dex …

https://doi.org/10.7554/eLife.30766.006
Figure 2—figure supplement 1
Knock-down efficiency of Dnmt hairpins.

(A-C) Two independent hairpins against ​Dnmt1, 3a, and 3b (versus scrambled shRNA; shScr) were delivered to mature 3T3-L1 adipocytes (Day 8) via lentiviral transduction. Shown is mRNA (n = 3, p<0.05,…

https://doi.org/10.7554/eLife.30766.007
Figure 2—figure supplement 1—source data 1

Figure 2—figure supplement 1—source data.

https://doi.org/10.7554/eLife.30766.008
Figure 2—figure supplement 2
Dnmt3a is necessary for Dex- and TNF-mediated IR.

(A–E) Two independent hairpins against ​Dnmt1, Dnmt3a, and Dnmt3b (versus scrambled shRNA; shScr) were delivered to mature 3T3-L1 adipocytes (Day 8) via lentiviral transduction, and cells were then …

https://doi.org/10.7554/eLife.30766.009
Figure 2—figure supplement 2—source data 1

Figure 2—figure supplement 2—source data.

https://doi.org/10.7554/eLife.30766.010
Figure 3 with 1 supplement
DNA methyltransferase activity is critical for Dnmt3a to mediate IR.

Insulin-stimulated glucose uptake assay was conducted using 3T3-L1 adipocytes treated with Dnmt inhibitors (A, 5-Aza; B, RG-108) and Dex or TNF for 4 days. Shown is the % of insulin-stimulated …

https://doi.org/10.7554/eLife.30766.012
Figure 3—figure supplement 1
Dnmt inhibitors confer protection from Dex- and TNF-mediated IR.

(A–F). Insulin-stimulated glucose uptake assay was conducted on 3T3-L1 adipocytes co-treated with Dex or TNF and the Dnmt inhibitors (A) RG-108, (B) 5-Azacytidine (5-Aza), (C, D) …

https://doi.org/10.7554/eLife.30766.013
Figure 3—figure supplement 1—source data 1

Figure 3—figure supplement 1—source data.

https://doi.org/10.7554/eLife.30766.014
Figure 4 with 5 supplements
Adipose-specific Dnmt3a KO mice display improved whole body insulin sensitivity and glucose tolerance on HFD.

(A) Body weight of Dnmt3aAdiKO and control mice on HFD (n = 7, 9, p<0.05, Student’s t-test, mean ±s.e.m.). (B) Body composition after 10 weeks of HFD (n = 7, 9). (C) Effect of long-term fasting (16 …

https://doi.org/10.7554/eLife.30766.016
Figure 4—figure supplement 1
Adipose-specific deletion of Dnmt3a.

Adipose tissue deletion of Dnmt3a was confirmed in 8 wk-old Dnmt3aAdiKO (KO) mice by qPCR (n = 3, p<0.05, Student’s t-test, mean ±s.e.m.) (A) and western blot (B).

https://doi.org/10.7554/eLife.30766.017
Figure 4—figure supplement 1—source data 1

Figure 4—figure supplement 1—source data.

https://doi.org/10.7554/eLife.30766.018
Figure 4—figure supplement 2
Metabolic features of Dnmt3aAdiKO mice on chow.

(A) Body weight of Dnmt3aAdiKO mice on chow. Glucose (B) and insulin tolerance test (C) on 11 and 12 wk old KO and WT mice on chow. (D) Insulin levels of fed and fasted chow-fed KO and WT mice as …

https://doi.org/10.7554/eLife.30766.019
Figure 4—figure supplement 2—source data 1

Figure 4—figure supplement 2—source data.

https://doi.org/10.7554/eLife.30766.020
Figure 4—figure supplement 3
Metabolic features of female Dnmt3aAdiKO on HFD.

(A) Body weight was measured at 8-wk-old mice prior to HFD regimen and at 22-wk-old mice on HFD. Glucose (B) and insulin tolerance test (C) on 19 and 20 wk old KO and WT mice on HFD. (n = 6, 6, …

https://doi.org/10.7554/eLife.30766.021
Figure 4—figure supplement 3—source data 1

Figure 4—figure supplement 3—source data.

https://doi.org/10.7554/eLife.30766.022
Figure 4—figure supplement 4
Gene expression profile of thermogenic genes in Dnmt3aAdiKO mice on HFD.

The expression of major thermogenic genes was measured by qPCR in iWAT (A) and BAT (B) from WT and KO mice on HFD (iWAT: n = 6,6; BAT n = 5,5 Student’s t-test, mean ±s.e.m.).

https://doi.org/10.7554/eLife.30766.023
Figure 4—figure supplement 4—source data 1

Figure 4—figure supplement 4—source data.

https://doi.org/10.7554/eLife.30766.024
Figure 4—figure supplement 5
Gene expression analysis of proinflammatory genes and histology examination of eWAT from Dnmt3af/f and Dnmt3aAdiKO mice on HFD.

The expression of major proinflammatory genes was measured by qPCR in iWAT (A) and BAT (B) from WT and KO mice on HFD (iWAT: n = 6, 6; BAT n = 5,5 Student’s t-test, mean ±s.e.m.). (C) Representative …

https://doi.org/10.7554/eLife.30766.025
Figure 4—figure supplement 5—source data 1

Figure 4—figure supplement 5—source data.

https://doi.org/10.7554/eLife.30766.026
Figure 5 with 1 supplement
Fgf21 is a key target gene of Dnmt3a.

Venn diagram showing the number of positive (A) and negative (B) Dnmt3a target genes through comparative analysis of RNA-seq profiles from Dnmt3a knockdown and overexpressing adipocytes. (C) Heat …

https://doi.org/10.7554/eLife.30766.028
Figure 5—figure supplement 1
Serum Fgf21 levels in Dnmt3aAdiKO mice.

(A) Serum Fgf21 levels measured by ELISA in WT and KO mice on 12-wk-old chow (n = 6) or after 14 weeks of HFD (n = 7, p<0.05, Student’s t-test, mean ±s.e.m.).

https://doi.org/10.7554/eLife.30766.029
Figure 5—figure supplement 1—source data 1

Figure 5—figure supplement 1—source data.

https://doi.org/10.7554/eLife.30766.030
Dnmt3a-mediated methylation of the Fgf21 gene promoter.

(A) Schematic showing the tested regions for MeDIP-qPCR at the Fgf21 locus. Target regions were chosen using MethFinder (http://www.urogene.org/methprimer) on regions proximal (5 kb) and distal (50 …

https://doi.org/10.7554/eLife.30766.032
FGF21 DNA methylation in adipose tissue from subjects with T2D compared with controls.

(A) Four CpG sites annotated to FGF21 with significant difference in adipose tissue DNA methylation between subjects with type 2 diabetes and controls. (p<0.01, paired Wilcoxon statistics, mean ±s.d.…

https://doi.org/10.7554/eLife.30766.034
Author response image 1
(A) Histograms of PPARγ-ChIP and histone modification fragments near the mouse Dnmt3a locus (Mikkelsen et al., 2010).

Depicted with arrow are PPARγ peaks that co-localize with H3K4me1 and/or H3K27ac enhancer peaks. (B) The effect of Rosi (1uM) on the expression of Dnmt3a in the presence/absence of cycloheximide …

Additional files

Supplementary file 1

Genes that are up-regulated (UP) and down-regulated (DOWN) Dnmt3a vs.

GFP and shDnmt3a/shScr.

https://doi.org/10.7554/eLife.30766.035
Supplementary file 2

DNA methylation of four CpG sites annotated to Fgf21 showing differential DNA methylation in adipose tissue of a case control study for type 2 diabetes.

https://doi.org/10.7554/eLife.30766.036
Supplementary file 3

Oligonucleotide sequences used in this manuscript.

https://doi.org/10.7554/eLife.30766.037
Supplementary file 4

Clinical characteristics of study subjects included in the discordant twin cohort and the case control cohort.

https://doi.org/10.7554/eLife.30766.038
Transparent reporting form
https://doi.org/10.7554/eLife.30766.039

Download links