Glucocorticoid and TGFβ Crosstalk in the Regulation of PAI-1 Gene Expression

T. D. Gelehrter, G. Li, S. Wang

Departments of Human Genetics and Internal Medicine, University of Michigan Medical School, Ann Arbor MI 48109-0618, U.S.A.

Transforming growth factor β (TGFβ) is a major regulator of extracellular matrix (ECM) deposition and a potent inducer of type-1 plasminogen activator inhibitor (PAI-1) gene expression. Glucocorticoids inhibit TGFβ-induced expression of ECM proteins including fibronectin, collagen, TIMPS and PAI-1, and thus have an important opposing role in wound healing and fibrosis. We have reported previously that liganded glucocorticoid receptor (GR) represses TGFβ transactivation of PAI-1 in Hep3B human hepatoma cells, and that it interacts functionally and physically with the C-terminal activation domain of Smad3, a major mediator of TGFβ signaling (Song et al. Proc Natl Acad Sci USA 96 : 11 776, 1999).

To identify the domain(s) in the glucocorticoid receptor involved in repression of TGFβ signaling, we have examined the ability of various GR truncation, deletion, and substitution mutants to repress TGFβ transactivation in Hep3B human hepatoma cells, which lack functional endogenous GR. Partial deletions in the ligand binding domain (LBD), including the τ2 and τC (AF2) regions, greatly reduced or eliminated GR repression, while deletion of the N-terminal AF1 (τ1) domain, and substitution mutations in the DNA binding domain (DBD) had little or no effect. Liganded androgen receptor repressed TGF-β transactivation as strongly as GR, whereas mineralocorticoid receptor (MR) did not. Studies with rat GR-MR chimeras confirmed that the GR C-terminal domains, but not the N-terminal domains, were required for repression. RU486, a strong antagonist of transactivation by GR, only partially reversed repression by wild-type GR, and was itself a partial agonist for repression.

Co-immunoprecipitation experiments in Hep3B cells demonstrated that all GR mutants that repressed TGFβ transactivation interacted physically, in vivo, with endogenous Smad3. However, most GR mutants that did not repress also interacted with Smad3, suggesting that physical interaction between GR and Smad3 is necessary but not sufficient for repression. GST pull-down assays confirmed these results and demonstrated that several regions of the LBD were sufficient for GR-Smad3 physical interaction in vitro. Interestingly, the physical interaction required activation of Smad3 by TGFβ, but not dexamethasone binding to GR. In contrast, glucocorticoid binding to GR was necessary for transrepression. However, substitution of alanine for a conserved serine in the τ2 region of the LBD blocked transrepression even though it did not interfere with dexamethasone binding; thus, the τ2 domain may play an important role in transrepression independent of its role in glucocorticoid binding.

We conclude that the LBD of GR is required for GR-mediated transrepression of TGFβ transactivation. Unlike the GR transrepression of AP-1 or NFκB, in which the DBD of GR is essential, the GR DBD is dispensable in GR transrepression of TGFβ signaling. Ligand binding to GR is necessary but not sufficient for transrepression of TGFβ activation of PAI-1 gene expression. These results may have important implications for the development of ”dissociated steroids” for the therapy of chronic inflammatory and fibrotic diseases.

This work was supported by a Biomedical Science grant from the Arthritis Foundation.