Direct glucocorticoid receptor-Stat5 interaction in hepatocytes controls body size and maturation-related gene expression

Abstract

The glucocorticoid receptor regulates transcription through DNA binding as well as through cross-talk with other transcription factors. In hepatocytes, the glucocorticoid receptor is critical for normal postnatal growth. Using hepatocyte-specific and domain-selective mutations in the mouse we show that Stat5 in hepatocytes is essential for normal postnatal growth and that it mediates the growth-promoting effect of the glucocorticoid receptor through a direct interaction involving the N-terminal tetramerization domain of Stat5b. This interaction mediates a selective and unexpectedly extensive part of the transcriptional actions of these molecules since it controls the expression of gene sets involved in growth and sexual maturation.

Figure 1.

Growth deficiency in Stat5^AlfpCre, GR^AlfpCre, and Stat5/GR^AlfpCre mice. Growth curves from control mice (black dashed line), mice lacking Stat5 in the hepatocytes (red line), mice lacking GR in hepatocytes (green line), and mice lacking both Stat5 and GR in hepatocytes (blue line). All mutants show a very similar growth defect starting around 3 wk. Error bars indicate standard error of the mean.

Figure 2.

Substantial similarities in expression profiles from Stat5^AlfpCre, GR^AlfpCre, and Stat5/GR^AlfpCre mice. (A) Venn diagrams showing overlap of genes significantly changed in livers from mice with hepatocyte-specific deletion of Stat5 or GR. As criterion for calling a gene significantly changed we used p < 0.001, or the combined criterion of p < 0.0025 and a log ratio >1 or less than −1 (fold change >2 or <0.5). Note the substantial overlap of genes down-regulated by the two mutations. (B) Comparison between Stat5^AlfpCre and GR^AlfpCre mutants of the expression changes for genes significantly changed in both mutants. All of these genes are changed in the same direction in the two mutants and the magnitude of change correlates tightly. (C) Comparison of expression changes in Stat5 mutant and Stat5 + GR mutant livers for the same genes as in B. There is no additional effect of deleting GR in the absence of Stat5 for the expression levels of these genes.

Figure 3.

GR–Stat5 interaction is dependent on the Stat5b N terminus. (A, top) Liver extracts were prepared from untreated or GH-injected mice, followed by quantitative α-Stat5b immunoprecipitation (I.P.) and subsequent immunoblotting with anti-P-Y-Stat5 antiserum. (Bottom) Blots were stripped and reprobed with anti-Stat5b-specific antiserum. Stat5b and Stat5b^ΔN proteins are fully tyrosine-phosphorylated upon GH stimulation. The amounts of total Stat5b were reduced to ∼40% in Stat5^ΔN mice. (B) In vivo binding of Stat5 in proximal enhancers of ALS, Socs2, and IGF-1. The fold enrichment for each DNA fragment upon immunoprecipitation by anti-Stat5 followed by quantitative PCR is illustrated as histograms (mean ± SEM). Upon GH stimulation, wild-type Stat5b is highly bound to chromatin. Stat5b^ΔN displays ∼50%–70% of the DNA-binding capacity of full-length Stat5b. (C) The N terminus of Stat5b is essential for the GR–Stat5b interplay. Reciprocal immunoprecipitation of liver extracts using anti-Stat5b (left) or anti-GR (right) antisera. (Left) Stat5b^ΔN as well as full-length Stat5b proteins are efficiently immunoprecipitated with anti-Stat5b antisera (bottom), but only wild-type Stat5b interacts with the GR as shown via co-IP experiments (top). (Right) GR proteins are comparably pulled down (bottom), but only full-length Stat5b is coimmunoprecipitated (top). Four similar experiments were performed and one representative analysis is shown.

Figure 4.

Deletion of the Stat5 N terminus selectively reduces the expression of the Stat5-responsive genes, which are also dependent on GR. (A,B) Magnitude (A) and significance (B) of expression changes of GR-independent (top panel) and GR-dependent (bottom panel) Stat5-responsive genes in livers from mice with a deletion of the N terminus of Stat5. The Stat5-responsive genes that are not dependent on GR show very little dependence on the N terminus while many of the Stat5-responsive genes also dependent on GR show severely reduced expression levels. (C) Comparison within the group of GR-dependent Stat5 targets. The genes that are only partially GR dependent (white and gray dots; left panel) are also only partially dependent on the N terminus of Stat5 (right panel).