The KRAB zinc finger protein RSL1 modulates sex-biased gene expression in liver and adipose tissue to maintain metabolic homeostasis

Abstract

Krüppel-associated box zinc finger proteins (KRAB-ZFPs) are a huge family of vertebrate-specific repressors that modify gene expression in an epigenetic manner. Despite a well-defined repression mechanism, few biological roles or gene targets of KRAB-ZFP are known. Regulator of sex-limitation 1 (RSL1) is a mouse KRAB-ZFP that enforces male-predominant expression in the liver, affecting body mass and pubertal timing. Here we show that female but not male Rsl1(-/-) mice gain more weight than wild-type mice on a high-fat diet (HFD) and that key liver and white adipose tissue (WAT) metabolic genes are altered in both Rsl1(-/-) sexes in response to dietary stress. Expression profiling of Rsl1-sensitive genes in liver and WAT indicates that RSL1 accentuates sex-biased gene expression in liver but greatly diminishes it in WAT. RSL1 expression solely in liver is sufficient to limit diet-induced weight gain and suppress lipogenic genes in WAT, indicating that RSL1 balances metabolism via liver-to-adipose-tissue communication. RSL1's effects on adult physiology exemplify a significant modulatory capacity of KRAB-ZFPs, in the absence of which there is widespread metabolic dysregulation. This ability to buffer against gene expression noise, coupled with extensive individual genetic variation, highlights the enormous potential of KRAB-Zfp genes as candidate risk factors for complex diseases.

FIG 1

RSL1 influences response to HFD. (A) Mean body weights and standard errors of the means (SEM) for mice fed standard lab chow from weaning to 10 weeks of age (for WT males, n = 27; for Rsl1^−/− males, n = 18; for L-Rsl1-tg males, n = 12; for WT females, n = 28; for Rsl1^−/− females, n = 21; and for L-Rsl1-tg females, n = 13). The t test for each group versus the WT determined differences within each sex. *, P < 0.001. (B) Percent weight gain throughout 60 days of HFD. Data are plotted as mean percentages ± SEM over time. WT males (n = 8), Rsl1^−/− males (n = 4), L-Rsl1-tg males (n = 5), WT females (n = 6), Rsl1^−/− females (n = 6), and L-Rsl1-tg females (n = 10) were studied. t tests evaluated differences from WT mice. †, P < 0.10; *, P < 0.05. (C) Mean percent weight gain ± SEM after 60 days of HFD. *, P < 0.05.

FIG 2

Heat production in Rsl1^−/− females is transiently greater than that in WT mice at the beginning of their active phase, and both sexes have enhanced insulin sensitivity. (A) Gas exchange values from the CLAMS metabolic chambers (n = 4 mice per sex and genotype) were used to calculate energy expenditure as heat, based on the following equation: heat (kcal/h) = [3.82 + (1.23 × RER)] × VO₂. Data are plotted as means ± SEM. Means were considered significantly different (*) if the difference in WT versus Rsl1^−/− mice was ≥20% and the t test P value was <0.05. (B) For glucose tolerance tests, mice were fasted overnight and injected intraperitoneally with d-glucose (2 g/kg body weight). Blood glucose was monitored 0, 30, 60, 90, 120, and 150 min after injection. Data are plotted as means ± SEM over time. WT males (n = 7), Rsl1^−/− males (n = 6), WT females (n = 4), and Rsl1^−/− females (n = 6) were studied. t tests evaluated differences from the WT group. *, P < 0.05. AUC were calculated for each mouse and plotted as means ± SEM. t tests evaluated differences from the WT group. *, P < 0.05.

FIG 3

Sex-specific effects of RSL1 on hepatic response to dietary stress. Hepatic mRNA abundance was measured by qRT-PCR (pools for ≥4 mice). (A) Mice were fed a chow diet ad lib (fed) or fasted for 16 h (fast). (B and C) Mice were fed a chow diet (cd) or high-fat diet (hfd) for 8 weeks. The data are plotted as the means and standard deviations (SD) for duplicate reactions relative to WT males. In panel C, note that the graphs for males and females differ substantially in scale. t tests evaluated differences from the WT group. *, P < 0.05.

FIG 4

RSL1 influences response to dietary stress in white adipose tissue. WAT mRNA abundance was measured by qRT-PCR (pools for ≥4 mice). Mice were fed a chow diet (cd), fasted for 16 h (fast), or given 8 weeks of high-fat diet (hfd). The data are plotted as the means and SD for duplicate reactions relative to WT males. Note that the results for Elovl6 and Elovl3 differ substantially in scale depending on sex. t tests evaluated differences from the WT group. *, P < 0.05.

FIG 5

Genomewide identification of RSL1-sensitive genes in WAT. (A) Venn diagram of genes differentially expressed ≥2-fold in WT and Rsl1^−/− abdominal WAT. Males had nearly three times as many differentially expressed genes as females; however, less than 15% of these genes were shared between the sexes. Data shown are the numbers of genes that were either elevated (upward-facing arrows) or reduced (downward-facing arrows) in Rsl1^−/− compared to WT mice. (B) Top 10 genes with the greatest differential of up- and downregulated genes in female (left) and male (right) WAT. (C) Expression by qRT-PCR of genes selected from the microarray in WT and Rsl1^−/− WAT (n ≥ 4). Mup genes appeared to be decreased by microarray assay, where probes detected 10 Mup family members, but gene-specific primers for Mup1 and Mup3 revealed a male bias and RSL1 dependence in liver (21, 22) and a female bias unaffected by RSL in WAT. The data are plotted as means and SEM. t tests evaluated differences from the WT group. *, P < 0.05.

FIG 6

Comparison of RSL1-sensitive genes in liver and WAT. (A) Heat maps highlighting genes that are sexually dimorphic (>1.5-fold difference between males and females) in liver and WAT. The intensity of color reflects the magnitude of the sex bias (red shows male-predominant genes, and blue shows female-predominant genes). The arrangement of genes in Rsl1^−/− mice is identical to that in WT mice. (B and C) Venn diagrams of sexually dimorphic genes (>1.5-fold difference between males and females) in liver and WAT and their distributions among WT and Rsl1^−/− mice. Circles are not to scale but are approximated to convey relative relationships among gene sets. (D) Sex-biased genes in liver and WAT were grouped into categories based on the magnitude of their sex bias (1.5- to 2-fold, 2- to 3-fold, and >3-fold) and plotted as percentages of the total. In WT liver, 71.9% of the sex-biased genes were in the 1.5- to 2-fold range (58.4% in Rsl1^−/− liver), 16.2% were in the 2- to 3-fold range (21.8% in Rsl1^−/− liver), and 12.0% were in the >3-fold range (19.9% in Rsl1^−/− liver). In WAT, the differences between WT and Rsl1^−/− mice were much smaller (in the 1.5- to 2-fold range, 53.5% for WT mice and 55.9% for Rsl1^−/− mice; in the 2- to 3-fold range, 24.6% for WT mice and 23.2% for Rsl1^−/− mice; and in the >3-fold range, 21.9% for WT mice and 20.8% for Rsl1^−/− mice). (E) Venn diagrams of RSL1-responsive genes in liver compared to RSL1-responsive genes in WAT. A differential expression (WT versus Rsl1^−/−) threshold of 2-fold was used, with significance for P values of <0.05.

FIG 7

Liver-autonomous RSL1 action and impact on WAT lipogenic genes. (A) G6pc expression in primary hepatocytes isolated from WT and Rsl1^−/− livers and cultured for 18 h in starvation medium (“fasted”) or starvation medium supplemented with glucose plus insulin. qRT-PCR results from duplicate experiments were normalized to that for WT males with glucose plus insulin, which was arbitrarily set to 1. Means and SD were plotted. (B) Genes involved in lipogenesis exhibit attenuated expression in WAT due to Rsl1 expression in liver. WAT mRNA abundance was measured by qRT-PCR (n ≥ 4 mice per genotype). To ease comparisons for transcripts that vary in abundance, results were normalized and the means and SEM plotted relative to the level in WT females, which was arbitrarily set to 1. t tests evaluated differences from the WT group. *, P < 0.05.

FIG 8

RSL1-sensitive pathways in liver that may contribute to metabolic phenotypes. CD, chow diet; HFD, high-fat diet. “Lean” and “obese” refer to the state of Rsl1^−/− mice relative to normal wild-type mice. RSL1 limits gluconeogenesis and MUP1 secretion, which in turn may derepress lipogenesis, leading to a normal phenotype. In the absence of RSL1, upregulation of these pathways promotes a lean phenotype. With HFD, the RSL1-deficient female liver fails to repress ELOVL3 and fatty acid transporters (e.g., MUP1 and LCN2), which triggers a cascade in systemic signaling, resulting in a failure to suppress WAT lipogenesis and in excessive weight gain.