Molecular mechanisms of gene activation by nuclear receptors and their associated coregulators

Abstract

Nuclear receptors (NRs) are ligand-dependent transcription factors. The DNA-bound receptors either positively or negatively regulate transcription of target genes. The receptors regulate transcription through interaction with transcription coregulators and recruitment of components of the basal transcription machinery. Most nuclear receptors have at least two transactivation regions, the activation function 1 (AF-1) which resides in the N-terminus, and the activation function 2(AF-2), which is localised in the C-terminus. This thesis deals with molecular aspects of these activation functions. The AF-1 region and the AF-1 core region (tau 1 core) of the glucocorticoid receptor are disordered in solution. The activation function of the tau 1 core domain was investigated with regard to transactivation potential in vivo and binding capacity to target proteins in vitro.

The results obtained are consistent with models in which activation domains are built up of short activation modules, each of which can interact with coregulator(s) to give efficient activation of transcription. Our results also suggest that an increased number of activation modules may not result in correspondingly higher affinity for target factors, but instead the concentration of binding sites is increased which could give rise to a higher association rate. Thus, the association rate may be the important parameter for efficient activation rather than affinity. Investigation of the N-terminal regions (containing AF-1) of the two estrogen receptor subtypes, ERalpha and ERbeta, showed that both the N-terminal region of ERalpha (ERalpha-N) and ERbeta (ERbeta-N) are unstructured in solution. We also characterised the interaction between the TATA Binding Protein (TBP) and the N-terminal ERs.

Our results show that the intrinsically unstructured ERalpha-N interacts with TBP, and suggest that conformational changes are induced in ERalpha-N upon TBP interaction. ERbeta-N did not interact with T13P. This difference in TBP binding could imply differential recruitment of coregulatory proteins by ERalpha-N and ERbeta-N and could be one explanation for the lower transcriptional activation potential of ERbeta compared to that of ERalpha usually seen in in vivo experiments. The ligand-binding domain/AF-2 domain of NRs undergoes a conformational change upon binding of ligand, which facilitates the association of coactivator proteins.

A large number of coactivators have been shown to bind to the AF-2 domains of many nuclear receptors. Two such coactivators are the TRAP220 and the TIF2, which both contain conserved LXXLL NR interaction motifs (L, leucine; X, any amino acid), so called NR-boxes. Although they utilize the same motifs for binding to NRs, they have been shown to be components of distinct coactivator pathways. The TRAP220 protein is part of a large multi-subunit coactivator complex, the TRAP/Mediator complex. The TIF2 protein, on the other hand, is believed to form a complex with other coactivators such as CBP and P/CAF, which are involved in chromatin accessibility. We have shown that TRAP220 and TIF2 can compete with each other for binding to both DNA bound and non-DNA bound thyroid hormone receptor in a ligand dependent manner.

Our findings support a model where competition between NR-binding proteins/complexes constitutes a regulatory step in NR mediated transactivation. We have also demonstrated that TRAP220 displays an association preference for ERbeta over ERalpha. Such differences with regard to coactivator recruitment indicate that the relative importance of individual coregulators in estrogen signaling could depend on the dominant ER subtype expressed in a cell. TIF2 did not show such preference and could readily interact with both receptors. In an additional study we investigated the interactions between ERalpha and the different LXXLL motifs of TIF2. In crystallographic studies of agonist-bound ERalpha ligand-binding domain in complex with an NR-box 2 or NR-box 3 peptide we revealed distinct binding modes for the two peptides. The Box 2 peptide adopted the 'classical' binding mode that has been observed for other NR / coactivator peptide complexes while Box 3 displayed a novel mode of interaction. In contrast, quantitative protein-protein interaction studies using wild type and mutated NR-box 3 peptides favoured the 'classical binding' mode also in the case of NR-box 3. The conflicting crystallographic and solution binding data highlights the potential pitfalls encountered when using short peptide motifs to mimic interactions between large macromolecules in structural studies.