Kris Schauwaers

Loss of androgen receptor binding to selective androgen response elements causes a reproductive phenotype in a knockin mouse model

Androgens influence transcription of their target genes through the activation of the androgen receptor (AR) that subsequently interacts with specific DNA motifs in these genes. These DNA motifs, called androgen response elements (AREs), can be classified in two classes: the classical AREs, which are also recognized by the other steroid hormone receptors; and the AR-selective AREs, which display selectivity for the AR. For in vitro interaction with the selective AREs, the androgen receptor DNA-binding domain is dependent on specific residues in its second zinc-finger. To evaluate the physiological relevance of these selective elements, we generated a germ-line knockin mouse model, termed SPARKI (SPecificity-affecting AR KnockIn), in which the second zinc-finger of the AR was replaced with that of the glucocorticoid receptor, resulting in a chimeric protein that retains its ability to bind classical AREs but is unable to bind selective AREs. The reproductive organs of SPARKI males are smaller compared with wild-type animals, and they are also subfertile. Intriguingly, however, they do not display any anabolic phenotype. The expression of two testis-specific, androgen-responsive genes is differentially affected by the SPARKI mutation, which is correlated with the involvement of different types of response elements in their androgen responsiveness. In this report, we present the first in vivo evidence of the existence of two functionally different types of AREs and demonstrate that AR-regulated gene expression can be targeted based on this distinction.

Functional Interplay between Two Response Elements with Distinct Binding Characteristics Dictates Androgen Specificity of the Mouse Sex-limited Protein Enhancer

Many of the aspects involved in steroid-specific transcriptional regulation are still unsolved to date. We describe here the detailed characterization of the mouse sex-limited protein enhancer as a paradigm for androgen-specific control of gene expression. By deletion analysis, we delineate the minimal enhancer region displaying androgen sensitivity and specificity. We also show that each of the three hormone response elements (HRE), which constitute this minimal enhancer region, is essential but not sufficient for its functionality. When investigated as isolated elements, HRE1 is inactive and HRE3 is a potent androgen response element as well as GRE. Only the non-canonical HRE2 (5-TGGTCAgccAGTTCT-3′) is capable of conferring an androgen-specific transcriptional response to a heterologous promoter. This finding is correlated with the fact that HRE2 is recognized in binding assaysin vitro by the DNA-binding domain (DBD) of the androgen but not the glucocorticoid receptor, while HRE3 is recognized by both DBDs. Differential binding of the androgen receptor to HRE2 in the context of the enhancer was analyzed in more detail in footprinting assays in vitro. In transient transfection experiments using chimeric receptors, the inability of the glucocorticoid receptor to transactivate via the slp-ARU as well as the isolatedslp-HRE2 was rescued by the replacement of its DNA-binding domain with that of the androgen receptor. Our data suggest that the functional interplay between the weak, but highly androgen-specific HRE2 and the adjacent strong, but non-selective HRE3 is the major determinant in the generation of androgen specificity of transcriptional response via the sex-limited protein enhancer.

Molecular biology of the androgen responses

The androgen receptor is a ligand-inducible transcription factor with very specific target genes. This definition implies the activation by the cognate ligand through the ligand-binding domain, the recognition of the target genes by means of the DNA-binding domain and the transcriptional activation through different activation functions. When the first androgen-responsive genes were cloned, we identified receptor-binding sites by means of a DNAcellulose competition assay with partially purified androgen receptor from rat prostate (Claessens et al., 1990). Once the receptor cDNA was cloned, the separate DNAbinding domain was expressed and shown to have similar, if not identical DNA recognition properties as the full size receptor. The binding sites were proven functional in transient transfection experiments with reporter genes cloned downstream of these sites (Claessens et al., 1993). The motifs which are recognized by the receptor are called androgen response elements (ARE), and a consensus of the first identified AREs pointed out that it is very similar to the glucocorticoid/progesterone response element (GRE/PRE) consensus 5¢-GGTACAnnnTGTTCT-3¢. Not surprisingly, these AREs also act as GRE/PRE in transient transfections. The probasin promoter region also contains two AR-binding sites, but in contrast to what was observed for the earlier AREs, these are not recognized by the glucocorticoid receptor. Later on, several other selective AREs were characterized in the slp and sc enhancers (Verrijdt et al., 2000). A comparison of the DNA-binding domains of the androgen and glucocorticoid receptors revealed specific residues which are involved in the recognition of these selective AREs, but not in the recognition of the classical AREs. These residues are not situated within the first zinc-coordinated module or zinc finger, but rather in the second one, as well as in a carboxy-terminal extension of the DNA-binding motif (Schoenmakers et al., 2000). This hinted to us that the recognition of the selective AREs occurs through an alternative dimerization of the DNA-binding domain that would be specific for the androgen receptor. Indeed, when the direct repeat nature of the selective AREs was changed into inverted repeat nature, the selectivity of the AREs and of the enhancers, of which they form part, was lost (Verrijdt et al., 2000). The silico screening of human genome has led to the definition of several additional selective AREs. In collaboration with the group of Daniel Gewirth, we were able to solve a crystal structure of the DNA-binding domain of the androgen receptor complexed to a perfect direct repeat of the 5¢TGTTCT-3¢ hexamer (Shaffer et al., 2004). This revealed that the domain is folded into two zinc-coordinated modules very similar to what has been described for other nuclear receptors. The two monomers are organized in a head-to-head configuration. Specific for the androgen receptor is the increased strength of the dimerisation interface due to an enlarged contact surface as well as to three additional hydrogen bonds. A functional analysis of the carboxyterminal extension of the DBD, which is part of the hinge region, revealed that it has more functions besides contributing to selective DNA binding. It overlaps with part of a nuclear localization signal and it is involved in the control of transactivation. Indeed, opposite to what is expected, deletions within this region result in a superactive androgen receptor, even when DNA binding in band shifts becomes difficult to demonstrate. The transcription activation by the androgen receptor is complex in the sense that different domains are contributing to it. For all steroid receptors, two activation functions have been described: the activation function 1 (AF1) in the amino-terminal domain and activation function 2 (AF2) in the ligand-binding domain. The androgen receptor is an exception since the AF2 is weak and in most experiments difficult to demonstrate. A possible explanation for this was found in a strong interaction between the ligand-binding domain and the amino-terminal domain of the androgen receptor. This occurs through a motif at the amino-terminal end of the receptor that interacts with AF2, described as a hydrophobic cleft on the surface of the ligand-binding domain. This interaction seems to prevent recruitment of the known p160 co-activators to