The AR (androgen receptor) is a hormone-dependent transcription factor that translates circulating androgen hormone levels into a physiological cellular response by directly regulating the expression of its target genes. It is the key molecule in e.g. the development and maintenance of the male sexual characteristics, spermatocyte production and prostate gland development and growth. It is also a major factor in the onset and maintenance of prostate cancer and a first target for pharmaceutical action against the further proliferation of prostate cancer cells. The AR is a member of the steroid hormone receptors, a group of steroid-inducible transcription factors sharing an identical consensus DNA-binding motif. The problem of how specificity in gene activation is achieved among the different members of this nuclear receptor subfamily is still unclear. In this report, we describe our investigations on how the AR can specifically activate its target genes, while the other steroid hormone receptors do not, despite having the same consensus monomeric DNA-binding motif. In this respect, we describe how the AR interacts with a newly identified class of steroid-response elements to which only the AR and not, for example, the glucocorticoid receptor can bind.

Introduction

The AR (androgen receptor) is a member of the steroid hormone receptor subfamily of the nuclear receptors and therefore shares with its co-members a conserved three-domain structure consisting of an NTD (N-terminal domain), a central DBD (DNA-binding domain) and a C-terminal LBD (ligand-binding domain) [1]. The NTD of the AR harbours the main activation function and is therefore responsible for a large part of the transcriptional activation potential of the receptor [24]. The DBD is the region of the protein that interacts with DNA. The LBD is the site of interaction of the ligand (the androgen hormone), binding of which will trigger a cascade of events that will eventually lead to a migration of the receptor to the cellular nucleus and the activation of the receptor's target genes. We will pay particular attention to the structure–function relations within the AR-DBD and its role in androgen-related pathologies.

Schematic representation of the AR-DBD (A) and a three-dimensional model of the AR-DBD bound to a direct repeat ARE (B)

Figure 1
Schematic representation of the AR-DBD (A) and a three-dimensional model of the AR-DBD bound to a direct repeat ARE (B)

(A) Schematic representation of the human AR-DBD. The regions containing the three α-helices are boxed, the CGSCKVF motif in the first helix is the P-box region responsible for the base-specific contacts with the DNA major groove. The three AR amino acids described by Schoenmakers et al. [19] to be crucial for AR-selective interaction with direct repeat elements are double underlined. The two amino acids proposed from the AR-DBD crystal by Shaffer et al. [20] to be essential for a better interaction by the AR-DBD compared with the GR-DBD are encircled. The dotted box delineates the part of the AR-DBD that was unstructured in the crystal. Closed arrowheads indicate AR-DBD amino acid mutations that are found in AIS patients, and open arrowheads depict AR-DBD mutations that are found in prostate cancer cells (according to the AR mutations database [5]). Arrows indicate the amino acids, the mutation of which was recently proposed to play a role in infertility due to oligozoospermia [6]. (B) Three-dimensional model of the AR-DBD bound to a perfect direct repeat of the 5′-TGTTCT-3′ DNA motif, according to Shaffer et al. [20]. Peptide backbone and DNA phosphate backbone are given. Represented as ‘ball and stick’ are Thr602 in the dimerization interface and Gly627 in the CTE that are important for direct repeat recognition and that are part of the crystal structure. Reprinted from Molecular Genetics and Metabolism, vol. 78, G. Verrijdt, A. Haelens and F. Claessens, Selective DNA recognition by the androgen receptor as a mechanism for hormone-specific regulation of gene expression, pp. 175–185, © 2003, with permission from Elsevier.

Figure 1
Schematic representation of the AR-DBD (A) and a three-dimensional model of the AR-DBD bound to a direct repeat ARE (B)

(A) Schematic representation of the human AR-DBD. The regions containing the three α-helices are boxed, the CGSCKVF motif in the first helix is the P-box region responsible for the base-specific contacts with the DNA major groove. The three AR amino acids described by Schoenmakers et al. [19] to be crucial for AR-selective interaction with direct repeat elements are double underlined. The two amino acids proposed from the AR-DBD crystal by Shaffer et al. [20] to be essential for a better interaction by the AR-DBD compared with the GR-DBD are encircled. The dotted box delineates the part of the AR-DBD that was unstructured in the crystal. Closed arrowheads indicate AR-DBD amino acid mutations that are found in AIS patients, and open arrowheads depict AR-DBD mutations that are found in prostate cancer cells (according to the AR mutations database [5]). Arrows indicate the amino acids, the mutation of which was recently proposed to play a role in infertility due to oligozoospermia [6]. (B) Three-dimensional model of the AR-DBD bound to a perfect direct repeat of the 5′-TGTTCT-3′ DNA motif, according to Shaffer et al. [20]. Peptide backbone and DNA phosphate backbone are given. Represented as ‘ball and stick’ are Thr602 in the dimerization interface and Gly627 in the CTE that are important for direct repeat recognition and that are part of the crystal structure. Reprinted from Molecular Genetics and Metabolism, vol. 78, G. Verrijdt, A. Haelens and F. Claessens, Selective DNA recognition by the androgen receptor as a mechanism for hormone-specific regulation of gene expression, pp. 175–185, © 2003, with permission from Elsevier.

DBD of the AR: involvement in AIS (androgen insensitivity syndrome), prostate cancer and fertility problems

Most naturally occurring mutations in the AR-DBD are correlated with the phenotype of partial to complete androgen insensitivity (AIS) and, less frequently, with prostate cancer {AR mutations database (http://www.androgendb.mcgill.ca/) [5]}.

Frame-shift mutations, mutations that introduce stop codons and mutations of Zn-co-ordinating cysteine residues, obviously destroy the three-dimensional structure of the DBD and result in an inactive receptor and complete AIS. Of all the other mutations, 83% cause mild, partial or complete AIS, while 17% of the mutations are linked with prostate cancer. All of the latter are somatic mutations. Recently, it was also speculated that mutations in the AR-DBD can be a causal factor in fertility problems correlated with severe oligozoospermia [6].

A schematic structure of the AR-DBD with an indication of these natural mutations and how these could affect receptor function is depicted in Figure 1(A).

General principles of DNA binding by steroid hormone receptors

The DBD of the nuclear receptors is a region of approx. 80 amino acids, organized in two zinc-co-ordinating modules and containing three helices (Figure 1A). In each zinc-finger, a Zn2+ ion is co-ordinated by four cysteine residues. The first helix extends from the C-terminal part of the first zinc-finger into the region between the two zinc-fingers and contains the amino acids responsible for the sequence-specific recognition of the DNA. The second helix lies in the tip of the second zinc-finger and borders the dimerization interface. The third helix stretches from the C-terminal end of the second zinc-finger into the CTE (C-terminal extension) of the DBD, and is important for the overall three-dimensional structure of the DBD. The amino acids in the first helix, responsible for the sequence-specific DNA binding (CGSCKVF), are 100% conserved between all but one of the steroid hormone receptors [1].

This degree of conservation of the DBDs allows the delineation of a subset of receptors [the AR, the progesterone receptor, the GR (glucocorticoid receptor) and the mineralocorticoid receptor], which all recognize the same DNA motif. All these receptors interact with the consensus 5′-TGTTCT-3′ hexamer, which is arranged as a three-nucleotide-spaced palindromic repeat [7].

Given the large diversity in physiological responses elicited by the different steroid hormones, it remains largely unclear exactly how the different receptors, despite sharing identical DNA-binding motifs, distinguish between the response elements of their respective target genes.

Differential receptor and cofactor expression, cell-type-specific chromatin organization or hormone metabolism etc. may account for many, but probably not all physiological situations where a gene is specifically regulated by a single steroid hormone [8].

DNA binding by the AR: direct versus indirect repeats

Our laboratory and others have previously demonstrated the functional importance of AREs (androgen-response elements) that subtly differ from the consensus SRE (steroid-response element) and that were found in genes that are specifically stimulated by androgens and not any of the other steroid hormones (reviewed in [9]).

These genes include the rat PB (probasin) gene [1012], the human secretory component gene [13], the mouse sex-limited protein gene [1416], Rhox5 [17] and SARG (specifically androgen-regulated gene) [18]. All these genes are driven by enhancers and/or promoters that contain AREs that share the unique characteristic of being ordered as direct, rather than indirect, repeats of the core 5′-TGTTCT-3′ monomer-interaction site [9]. The introduction of a single point mutation essentially converting these motifs into a partial inverted repeat invariably gives rise to a strong increase in its glucocorticoid responsiveness, while the amplitude of the androgen response remains virtually unchanged [13,15,16]. These direct repeat elements can therefore be considered androgen-specific SREs.

Mutational and deletion analyses of both the response elements and the AR-DBD fragments suggest that one amino acid residue (Thr602) in the dimerization interface in the second zinc-finger and two amino acids (Gly627 and Leu634) in the CTE of the AR-DBD play a crucial role in AR-DBD binding to specific AREs (Figure 1A; [19]).

The recently solved molecular structure of a dimer of the AR-DBD bound to a perfect direct repeat response element (depicted schematically in Figure 1B) clearly shows a classical head-to-head conformation of the two monomers as also seen on an inverted repeat [20]. A closer look at the dimerization interface, and a comparison with that of the GR-DBD bound to DNA [21], supported the hypothesis that two amino acids (Thr602 and Ser597) in this interface are responsible for a stronger binding of the AR-DBD compared with the GR-DBD to motifs diverging from the consensus SRE (Figure 2).

Schematic representations of the dimerization interfaces of the AR- and GR-DBDs

Figure 2
Schematic representations of the dimerization interfaces of the AR- and GR-DBDs

Schematic depiction of the dimerization interfaces (view from the top) of the AR-DBD (left) and the GR-DBD (right) as described in the respective protein structures [20,21]. Hydrogen bonds are depicted as dotted lines. The ‘glycine hole’ in the GR-DBD dimerization interface is indicated by a broken circle. The numbering is according to the human receptors.

Figure 2
Schematic representations of the dimerization interfaces of the AR- and GR-DBDs

Schematic depiction of the dimerization interfaces (view from the top) of the AR-DBD (left) and the GR-DBD (right) as described in the respective protein structures [20,21]. Hydrogen bonds are depicted as dotted lines. The ‘glycine hole’ in the GR-DBD dimerization interface is indicated by a broken circle. The numbering is according to the human receptors.

Briefly, the hydroxy side chain of Ser597 in the AR-DBD can form a hydrogen bond with the Ser597 of the dimer partner. In the GR-DBD, the glycine at this position leaves what is denoted as ‘the glycine hole’ (Figure 2). In a similar way, the hydroxy group in the side chain of the threonine residue at position 602 of the AR-DBD forms a hydrogen bond with the backbone oxygen of Ala596 of the dimer partner. The GR amino acid at this position is an isoleucine and is therefore excluded from forming a hydrogen bond with the backbone oxygen.

DNA binding and transactivation by wild-type and dimerization interface mutants of AR and GR

Figure 3
DNA binding and transactivation by wild-type and dimerization interface mutants of AR and GR

(A) Binding curves of recombinant wild-type and mutated AR-DBD (left) and GR-DBD (right) to the non-specific rTAT-GRE (top) and the specific PB ARE2 (bottom). The percentage of maximum binding (y-axis) is plotted as a function of the concentration (in nM) of the DBD (x-axis). Note the significant shift to the right of the GR-DBD curves compared with those of the AR-DBD for the PB-ARE2 element, illustrating the AR specificity of this motif. ––●––, wild-type; – –○– –, S-G mutant; −−·▼−−·, T-I mutant; ····▽····, double mutant. (B) Results of transient transfection assays in HeLa cells using a 2×rTAT-GRE- (left panel) and a 2×PB-ARE2- (right panel) containing luciferase reporter construct. Indicated are the induction factors (the ratios of luciferase values of induced over those of non-induced samples) mediated by the wild-type and mutated AR (closed bars) and GR (open bars) upon stimulation with their respective hormones (1 nM R1881 and 10 nM dexamethasone). Results are means±S.E.M. for at least four experiments performed in triplicate.

Figure 3
DNA binding and transactivation by wild-type and dimerization interface mutants of AR and GR

(A) Binding curves of recombinant wild-type and mutated AR-DBD (left) and GR-DBD (right) to the non-specific rTAT-GRE (top) and the specific PB ARE2 (bottom). The percentage of maximum binding (y-axis) is plotted as a function of the concentration (in nM) of the DBD (x-axis). Note the significant shift to the right of the GR-DBD curves compared with those of the AR-DBD for the PB-ARE2 element, illustrating the AR specificity of this motif. ––●––, wild-type; – –○– –, S-G mutant; −−·▼−−·, T-I mutant; ····▽····, double mutant. (B) Results of transient transfection assays in HeLa cells using a 2×rTAT-GRE- (left panel) and a 2×PB-ARE2- (right panel) containing luciferase reporter construct. Indicated are the induction factors (the ratios of luciferase values of induced over those of non-induced samples) mediated by the wild-type and mutated AR (closed bars) and GR (open bars) upon stimulation with their respective hormones (1 nM R1881 and 10 nM dexamethasone). Results are means±S.E.M. for at least four experiments performed in triplicate.

Mutations involving Ser597 have been described in PAIS (partial AIS) patients [2224]. In one of these patients, the serine residue is mutated to the corresponding GR amino acid glycine [22]. This patient, however, had an additional R617P mutation and his phenotype was similar to that of another patient with a single R617P mutation [25]. A T602P mutation has been described in a patient with a PAIS phenotype [26].

Taken together, this seemed a plausible mechanism to explain the apparent higher affinity of the AR-DBD for diverging SRE motifs, compared with the GR-DBD. Unfortunately, the part of the CTE containing Leu634, previously shown to also be involved in direct repeat binding [19], is not included in the crystal structure.

Amino acid exchanges between the AR and GR in their DBD dimerization interfaces do not influence the specificity of interaction

We have now introduced point mutations in the AR- and the GR-DBD, switching the two amino acids in the dimerization interface between the two receptors. These amino acids were mutated both separately and in combination. Mutations were introduced in the context of the full-length receptors in a CMV (cytomegalovirus)-driven mammalian expression vector for transient transfection experiments. The same mutations were also inserted in GST (glutathione S-transferase)-fusion vectors expressing the AR- or GR-DBD for use in in vitro bandshift assays.

Transient transfection experiments were performed in HeLa cells. The luciferase reporter constructs used in these experiments are driven by the thymidine kinase minimal promoter and contain two copies of either the non-specific rTAT (rat tyrosine aminotransferase)–GRE (glucocorticoid-response element) [27] or the AR-specific PB-ARE2 [10,11]. Bandshift assays were performed using radiolabelled 15-mers containing either the PB-ARE2 (as described in [1012]) or the rTAT-GRE (as described in [28]) binding motifs as probes, and using the recombinant DBD fragments of either wild-type AR or GR or the single or double mutants.

The results of these experiments (Figure 3) clearly demonstrate that none of the mutations changed the DNA binding affinities or activities of either receptor. This shows that AR- versus GR-specificity of DNA binding cannot solely be attributed to the presence of two receptor-specific amino acids in the DBD's dimerization surface, but probably also involves differences between the DBDs and/or CTEs of the two receptors.

Allosteric effects of the DNA-response element determine steroid specificity

As we had demonstrated previously [19], residues in the CTE are likely candidates to act in concert with one or both amino acids in the second zinc-finger to mediate AR selectivity of DNA interaction. This would be in agreement with our hypothesis that an AR-DBD interacting with a selective, directly repeated ARE can adopt a non-conventional head-to-tail conformation, whereby amino acids in the CTE of one dimer partner would contact amino acids in the dimerization interface of the other. This mode of interaction would be uncommon for a steroid hormone receptor, but is widely used by many other nuclear receptors that are able to interact with partial direct repeats of their monomeric binding element [29]. We speculate that the AR is able to adopt both types of conformation, depending on the type of DNA to which it is bound. This could be one example of allosteric influence on receptor conformation or signal transduction, depending on the type of DNA motif to which the receptor is bound, as was described for the GR by Lefstin et al. [30], and later reviewed by Geserick et al. [31] and Lefstin and Yamamoto [32]. Such allosteric effects are documented for selective versus classical AREs, e.g. in the report on the DNA sequence dependence of the effects of mutations in the AR-NTD affecting NTD–LBD interaction [33], or altering sumoylation of the AR-NTD [34].

Nuclear Receptors: Structure, Mechanisms and Therapeutic Targets: A Focus Topic at BioScience 2006, held at SECC Glasgow, U.K., 23–27 July 2006. Edited by C. Bevan (Imperial College London, U.K.), D. Black (Organon, U.K.) and I. McEwan (Aberdeen, U.K.).

Abbreviations

     
  • AIS

    androgen insensitivity syndrome

  •  
  • AR

    androgen receptor

  •  
  • ARE

    androgen-response element

  •  
  • CTE

    C-terminal extension

  •  
  • DBD

    DNA-binding domain

  •  
  • GR

    glucocorticoid receptor

  •  
  • GRE

    glucocorticoid-response element

  •  
  • LBD

    ligand-binding domain

  •  
  • NTD

    N-terminal domain

  •  
  • PAIS

    partial AIS

  •  
  • PB

    probasin

  •  
  • SRE

    steroid-response element

  •  
  • rTAT

    rat tyrosine aminotransferase

We thank Hilde de Bruyn, Rita Bollen and Kathleen Bosmans for excellent technical assistance. G.V., A.H. and L.C. are holders of a postdoctoral fellowship of the FWO-Vlaanderen (Research Foundation Flanders), and U.M. is the holder of a grant from the Association of International Cancer Research, U.K.

References

References
1
Tsai
M.-J.
O'Malley
B.W.
Annu. Rev. Biochem.
1994
, vol. 
63
 (pg. 
451
-
486
)
2
Bin
H.
Bowen
N.T.
Minges
J.T.
Wilson
E.M.
J. Biol. Chem.
2001
, vol. 
276
 (pg. 
42293
-
42301
)
3
Alen
P.
Claessens
F.
Verhoeven
G.
Rombauts
W.
Peeters
B.
Mol. Cell. Biol.
1999
, vol. 
19
 (pg. 
6085
-
6097
)
4
Callewaert
L.
Van Tilborgh
N.
Claessens
F.
Cancer Res.
2006
, vol. 
66
 (pg. 
543
-
553
)
5
Gottlieb
B.
Beitel
L.K.
Wu
J.H.
Trifiro
M.
Hum. Mutat.
2004
, vol. 
23
 (pg. 
527
-
533
)
6
Forresta
C.
Garolla
A.
Bartoloni
L.
Bettella
A.
Ferlin
A.
J. Clin. Endocrinol. Metab.
2005
, vol. 
90
 (pg. 
152
-
156
)
7
Mangelsdorf
D.J.
Thummel
C.
Beato
M.
Herrlich
P.
Schutz
G.
Umesono
K.
Blumberg
B.
Kastner
P.
Mark
M.
Chambon
P.
Evans
R.M.
Cell
1995
, vol. 
83
 (pg. 
835
-
839
)
8
Funder
J.W.
Science
1993
, vol. 
259
 (pg. 
1132
-
1133
)
9
Verrijdt
G.
Haelens
A.
Claessens
F.
Mol. Genet. Metab.
2003
, vol. 
78
 (pg. 
175
-
185
)
10
Rennie
P.S.
Bruchovsky
N.
Leco
K.J.
Sheppard
P.C.
McQueen
S.A.
Cheng
H.
Snoek
R.
Hamel
A.
Bock
M.E.
McDonald
B.S.
Mol. Endocrinol.
1993
, vol. 
7
 (pg. 
23
-
36
)
11
Claessens
F.
Alen
P.
Devos
A.
Peeters
B.
Verhoeven
G.
Rombauts
W.
J. Biol. Chem.
1996
, vol. 
271
 (pg. 
19013
-
19016
)
12
Schoenmakers
E.
Alen
P.
Verrijdt
G.
Peeters
B.
Verhoeven
G.
Rombauts
W.
Claessens
F.
Biochem. J.
1999
, vol. 
341
 (pg. 
515
-
521
)
13
Verrijdt
G.
Schoenmakers
E.
Alen
P.
Haelens
A.
Peeters
B.
Rombauts
W.
Claessens
F.
Mol. Endocrinol.
1999
, vol. 
13
 (pg. 
1558
-
1570
)
14
Adler
A.J.
Scheller
A.
Hoffman
Y.
Robins
D.M.
Mol. Endocrinol.
1991
, vol. 
5
 (pg. 
1587
-
1596
)
15
Verrijdt
G.
Schoenmakers
E.
Haelens
A.
Peeters
B.
Verhoeven
G.
Rombauts
W.
Claessens
F.
J. Biol. Chem.
2000
, vol. 
275
 (pg. 
12298
-
12305
)
16
Verrijdt
G.
Schauwaers
K.
Haelens
A.
Rombauts
W.
Claessens
F.
J. Biol. Chem.
2002
, vol. 
277
 (pg. 
35191
-
35201
)
17
Barbulescu
K.
Geserick
C.
Schuttke
I.
Schleuning
W.D.
Haendler
B.
Mol. Endocrinol.
2001
, vol. 
15
 (pg. 
1803
-
1816
)
18
Steketee
K.
Ziel-van der Made
A.C.J.
van der Korput
H.A.G.M.
Houtsmuller
A.B.
Trapman
J.
J. Mol. Endocrinol.
2004
, vol. 
33
 (pg. 
477
-
491
)
19
Schoenmakers
E.
Verrijdt
G.
Peeters
B.
Verhoeven
G.
Rombauts
W.
Claessens
F.
J. Biol. Chem.
2000
, vol. 
275
 (pg. 
12290
-
12297
)
20
Shaffer
P.L.
Jivan
A.
Dollins
D.E.
Claessens
F.
Gewirth
D.T.
Proc. Natl. Acad. Sci. U.S.A.
2004
, vol. 
101
 (pg. 
4758
-
4763
)
21
Luisi
B.F.
Xu
W.X.
Otwinowski
Z.
Freedman
L.P.
Yamamoto
K.R.
Sigler
P.B.
Nature
1991
, vol. 
352
 (pg. 
497
-
505
)
22
Zoppi
S.
Marcelli
M.
Deslypere
J.-P.
Griffin
J.E.
Wilson
J.D.
McPhaul
M.J.
Mol. Endocrinol.
1992
, vol. 
6
 (pg. 
409
-
415
)
23
Giwercman
Y.L.
Ivarsson
S.
Richthoff
J.
Lundin
K.B.
Gewircman
A.
Horm. Res.
2003
, vol. 
61
 (pg. 
58
-
62
)
24
Nordenskjöld
A.
Friedman
E.
Tapper-Persson
M.
Söderhäll
C.
Leviav
A.
Svensson
J.
Anvret
M.
Urol. Res.
1999
, vol. 
27
 (pg. 
49
-
55
)
25
Marcelli
M.
Zoppi
S.
Grino
P.B.
Griffin
J.E.
Wilson
J.D.
McPhaul
M.J.
J. Clin. Invest.
1990
, vol. 
87
 (pg. 
1123
-
1126
)
26
Melo
K.F.S.
Mendonca
B.B.
Billerbeck
A.E.C.
Costa
E.M.F.
Inacio
M.
Silva
F.A.Q.
Leal
A.M.O.
Latronico
A.C.
Arnhold
I.J.P.
J. Clin. Endocrinol. Metab.
2003
, vol. 
88
 (pg. 
3241
-
3250
)
27
Jantzen
H.M.
Strahle
U.
Gloss
B.
Stewart
F.
Schmid
W.
Boshart
M.
Miksicek
R.
Schutz
G.
Cell
1987
, vol. 
49
 (pg. 
29
-
38
)
28
Verrijdt
G.
Haelens
A.
Schoenmakers
E.
Rombauts
W.
Claessens
F.
Biochem. J.
2002
, vol. 
361
 (pg. 
97
-
103
)
29
Shaffer
P.L.
Gewirth
D.T.
EMBO J.
2002
, vol. 
21
 (pg. 
2242
-
2252
)
30
Lefstin
J.A.
Thomas
J.R.
Yamamoto
K.R.
Genes Dev.
1994
, vol. 
8
 (pg. 
2842
-
2856
)
31
Geserick
C.
Meyer
H.A.
Haendler
B.
Mol. Cell. Endocrinol.
2005
, vol. 
236
 (pg. 
1
-
7
)
32
Lefstin
J.A.
Yamamoto
K.
Nature
1998
, vol. 
392
 (pg. 
885
-
888
)
33
Callewaert
L.
Verrijdt
G.
Christiaens
V.
Haelens
A.
Claessens
F.
J. Biol. Chem.
2003
, vol. 
278
 (pg. 
8212
-
8218
)
34
Callewaert
L.
Christiaens
V.
Haelens
A.
Verrijdt
G.
Verhoeven
G.
Claessens
F.
Biochem. Biophys. Res. Commun.
2003
, vol. 
306
 (pg. 
46
-
52
)