Broadened ligand responsiveness of androgen receptor mutants obtained by random amino acid substitution of H874 and mutation hot spot T877 in prostate cancer

Authors

  • Karine Steketee,

    1. Department of Pathology, Josephine Nefkens Institute, Erasmus University, Rotterdam, The Netherlands
    Search for more papers by this author
    • The first two authors contributed equally to this article.

  • Leon Timmerman,

    1. Department of Pathology, Josephine Nefkens Institute, Erasmus University, Rotterdam, The Netherlands
    Search for more papers by this author
    • The first two authors contributed equally to this article.

  • Angelique C.J. Ziel-van der Made,

    1. Department of Pathology, Josephine Nefkens Institute, Erasmus University, Rotterdam, The Netherlands
    Search for more papers by this author
  • Paul Doesburg,

    1. Department of Pathology, Josephine Nefkens Institute, Erasmus University, Rotterdam, The Netherlands
    Search for more papers by this author
  • Albert O. Brinkmann,

    1. Department of Endocrinology & Reproduction, Erasmus University, Rotterdam, The Netherlands
    Search for more papers by this author
  • Jan Trapman

    Corresponding author
    1. Department of Pathology, Josephine Nefkens Institute, Erasmus University, Rotterdam, The Netherlands
    • Department of Pathology, Josephine Nefkens Institute, Erasmus University, P.O. Box 1738, 3000 DR Rotterdam, The Netherlands
    Search for more papers by this author
    • Fax: +31-10-4089487


Abstract

In a subset of endocrine therapy-resistant prostate cancers, amino acid substitutions H874Y, T877A and T877S, which broaden ligand specificity of the ligand binding domain (LBD) of the androgen receptor (AR), have been detected. To increase our knowledge of the role of amino acid substitutions at these specific positions in prostate cancer, codons 874 and 877 were subjected to random mutagenesis. AR mutants were screened in a yeast readout system for responsiveness to 5α-dihydrotestosterone, progesterone and dehydroepiandrosterone. At position 874, only the histidine to tyrosine substitution could broaden AR ligand specificity. At position 877, 4 ligand specificity broadening substitutions were found: T877A, T877S, T877C and T877G. The latter 2 were not found in prostate cancer. The AR mutants were tested in mammalian (Hep3B) cells for responsiveness to 13 different ligands. All mutants displayed their own ligand specificity spectrum. Importantly, AR(H874Y) and AR(T877A) could be activated by cortisol. According to the 3-dimensional structure of the AR LBD, T877 interacts directly with the 17β-hydroxyl group of androgens. All amino acid substitutions identified at position 877 had smaller side chains than the threonine in the wild-type receptor, indicating that increased space in the ligand binding pocket is important in broadened ligand specificity. Because H874 does not interact directly with the ligand, its substitution by a tyrosine is expected to change the ligand binding pocket conformation indirectly. For T877C and T877G substitutions, 2-point mutations are required, and for H874Y, T877A and T877S substitutions, only a 1-point mutation is sufficient. This most likely explains that the latter 3 have been found in prostate cancer. © 2002 Wiley-Liss, Inc.

Androgens (testosterone [T] and 5α-dihydrotestosterone [DHT]) are essential for development and maintenance of the male phenotype. They mediate their function by activation of the androgen receptor (AR), which is a member of the nuclear receptor family of transcription factors. The AR also plays a pivotal role in prostate tumor growth. Because growth of most prostate cancers depends on continuous androgenic stimulation, therapy of metastatic disease is generally based on androgen withdrawal or blockade of AR function by antiandrogens. However, after an initial regression, essentially all tumors continue to grow.

Like other nuclear receptors, the AR displays a modular structure: a carboxy-terminal ligand binding domain (LBD), a central DNA binding domain (DBD), and an amino-terminal transactivation domain (TAD). Upon ligand binding, the AR regulates transcription by binding to specific androgen response elements in regulatory regions of target genes. Together with coactivators, general transcription factors and RNA polymerase II, a stable transcription initiation complex is formed (for reviews, see refs. 1–3). The size of the AR can be variable, due to variation in the length of polyglutamine and polyglycine stretches in the TAD. The amino acid numbering in our article corresponds to an AR with a length of 919 amino acids, which is employed by The Androgen Receptor Gene Mutations Database (http://www.mcgill.ca/androgendb).

One of the causes of transition from androgen-dependent to apparent androgen-independent prostate tumor growth is modification of AR functioning. In a proportion of endocrine therapy-resistant tumors, AR gene amplification has been detected.4, 5, 6 This can lead to AR overexpression. Another mechanism, which directly affects AR function, can be activation of the AR by aberrant crosstalk with other signal transduction pathways.7, 8, 9, 10, 11 A third mechanism is modification of AR properties by missense mutations. In a subgroup of endocrine therapy-resistant prostate cancers, amino acid substitutions in the AR LBD have been found, which result in a broadened ligand response spectrum. The most common substitution, T877A, was first described in the LNCaP prostate cancer cell line.12 Subsequently, it was repeatedly found in prostate cancer tissue specimens of patients with advanced disease.13, 14, 15, 16, 17, 18 The T877A substitution renders the AR responsive to natural low-affinity ligands and antiandrogens; T877S and H874Y substitutions, which have also been found in prostate cancer, induce similar properties to the AR.12, 17, 19–23

During the past few years, the 3-dimensional structures of many nuclear receptor LBDs have been elucidated.24, 25, 26, 27, 28, 29, 30, 31, 32, 33 Crystallographic data have revealed a 3-layer structure composed of 10–12 α-helices. Ligand binding induces a specific conformational change in the helical LBD structure, which makes it accessible to coactivators.34, 35 Antagonists induce a different LBD conformation than agonists, indicating the importance of the LBD conformation for activation or inhibition of nuclear receptor function.27, 31, 34

Knowledge of the LBD structure is invaluable for explanation of the molecular and biologic effects of specific amino acid substitutions in the AR in prostate cancer.

Homology modeling predicted a 3-dimensional structure of the AR LBD that is similar to that of other nuclear receptors.36, 37, 38 The crystal structures of the DHT and R1881 complexed wild-type AR LBDs and the DHT complexed T877A mutant AR LBD have recently been elucidated; most of the earlier assumptions were confirmed.39, 40

We investigated the biologic effects of amino acid substitutions at positions 874 and 877, which are both in helix 11 of the AR LBD.36, 37, 39, 41 We also addressed the question of whether, in addition to H874Y, T877A and T877S, other as yet unidentified amino acid substitutions at these positions could give rise to similar functional alterations. Previously, we have shown that both wild-type AR and AR(T877A) retained their ligand specificity in yeast.42 Therefore, AR expression libraries with random mutations at codons 874 or 877 were screened for ligand specificity in a yeast readout system. AR mutants with an altered ligand specificity, as identified in the yeast system, were analyzed in mammalian cells for their responsiveness to a large series of sex steroids, antiandrogens and adrenal steroids, including glucocorticoids.

Abbreviations:

AR, androgen receptor; ASD, androstenedione; β-gal, β-galactoside; CPA, cyproterone acetate; DBD, DNA binding domain; DEX, dexamethasone; DHEA, dehydroepiandrosterone; DHT, 5α-dihydrotestosterone; E2, estradiol; LBD, ligand binding domain; OH-Fl, hydroxy-flutamide; Pg, progesterone; RLU, relative light unit; T, testosterone; TAA, triamcinolone acetonide; TAD, transactivation domain.

MATERIAL AND METHODS

Hormones

DHT, ASD, Pg, E2, DHEA, DEX, cortisol, aldosterone and TAA were purchased from Steraloids (Wilton, NH), and R1881 (methyltrienolone) was from NEN (Boston, MA). CPA was a gift from Schering (Berlin, Germany), OH-Fl from Schering USA (Bloomfield, NJ) and bicalutamide (Casodex) from Zeneca Pharmaceuticals (Macclesfield, UK).

Construction of Androgen Receptor CDNA Libraries with Random Mutation of Codons 874 and 877

The yeast AR cDNA expression vector pG1ARII42 was used to generate pG1ARIIΔ(863–919) as a cloning vector for construction of the control AR expression vector pG1ARIII, and the AR expression libraries pG1ARIII(874X) and pG1ARIII(877X). All deletions and mutations were generated essentially as described.43 First, a PCR fragment was synthesized utilizing pG1ARII as a template, with the forward primer 5′-CACTGAGGAGACAACCCAGAAGCT-3′ and the reverse primer 5′-AAGACGTCGACTACGCGGCGCGCAATAGGCTGCACGG-3′. A SalI restriction site in the reverse primer is boldfaced and underlined; a BssHII site in this primer is underlined. The amplified fragment was TthIII- and SalI-digested and exchanged with the corresponding AR fragment in pG1ARII, resulting in pG1ARIIΔ(863–919).

To generate the pG1ARIII(874X) library, PCR mutagenesis was carried out on the pG1ARII template, utilizing the forward primer 874X: 5′-ATTGCGCGCGAGCTGNNNCAGTTCACTTTTGACCTG-3′ (BssHII boldfaced and underlined; codon 874 underlined) combined with the reverse primer RP 5′-AAGACGTCGACCGGATCCGCTTCACTGGGTGTGG-3′ (SalI boldfaced and underlined; BamHI underlined; stop codon boldfaced). The amplified fragment was BssHII/SalI-digested and inserted in the corresponding sites in pG1ARIIΔ(863–919). The pG1ARIII(877X) AR cDNA library was generated by the same procedure, utilizing the forward primer 877X 5′-ATTGCGCGCGAGCTGCATCAGTTCNNNTTTGACCTGCTAATC-3′ and the RP reverse primer. Similarly, PG1ARIII was generated, utilizing forward primer 5′-ATTGCGCGCGAGCTGCATCAGTTCAC-3′ and the reverse primer RP, resulting in an AR cDNA expression vector with an internal BssHII site and a BamHI site in the slightly shorter 3′-UTR, compared with pG1ARII. The internal BssHII site does not result in an altered AR amino acid composition. Random codon representation at codons 874 and 877 in the pG1ARIII(874X) and pG1ARIII(877X) libraries was verified by sequencing of 14 clones of each library. In both libraries the sequenced clones were unique.

Yeast LacZ-reporter Plasmids

The androgen-inducible yeast integration vector pGRE3LacZi was constructed by insertion of a 100 bp HindIII/EcoRI fragment of pARE3tkCAT,44 containing a triple arranged repeat of the −174/−152 prostate-specific antigen (PSA) promoter region, in the corresponding sites of pLacZi (Clontech, Palo Alto, CA). The androgen-inducible yeast LacZ-reporter plasmid pUCΔSS-26X, containing a triple-arranged 26 bp GRE oligonucleotide, was provided by Dr. Didier Picard (Department Cell Biology, University of Geneva, Geneva, Switzerland).45

Construction of Mammalian Androgen Receptor Expression Plasmids

Mammalian AR expression plasmids pSVARIII, pSVARIII(H874Y), pSVARIII(T877A), pSVARIII(T877C), pSVARIII(T877G) and pSVARIII(T877S) were constructed by exchanging the TthIII/BamHI fragments of pG1ARIII(mutant) constructs for the corresponding fragment of pSVAR0.46

Yeast Strains, Growth and Transformation

Yeast strain YM4271(GRE3LacZ) was utilized for AR cDNA library screening. YM4271(GRE3LacZ) was derived from YM4271 (Clontech) by integration of NcoI linearized pGRE3LacZi into its nonfunctional ura locus. Yeast strain BJ2168, a gift from Dr. Picard, was used for quantitative measurement of AR activity.42 Yeast cells were grown in the appropriate selective media (0.67% w/v yeast nitrogen base without amino acids, 2% glucose, pH 5.8) supplemented with the required amino acids. Yeast transformation was carried out according to the lithium acetate method.47

Yeast Screening of Androgen Receptor Mutants

Approximately 400 clones of YM4271(GRE3LacZ) transformed with pG1ARIII(874X) or pG1ARIII(877X) were grown on a master plate with the appropriate selective medium. After replica plating on Hybond-N filters (Amersham, Buckinghamshire, UK), colonies were grown for 16 hr on the same medium supplemented with different hormones: DHT (10−8 M), Pg (10−7 M) and DHEA (10−6 M) or in the absence of hormone. Yeast colonies were made permeable by freezing the filters in liquid nitrogen. Next, LacZ expression was visualized by incubation on Whatmann paper soaked in Z-buffer (60 mM Na2HPO4, 40 mM NaH2PO4, 10 mM KCl, 1 mM MgSO4, pH 7.0) supplemented with 0.27% β-mercaptoethanol and 0.1% X-GAL.

AR expression plasmids were isolated from LacZ-positive yeast clones as described.48 Plasmids were sequenced to identify specific mutations at codons 874 and 877 and to confirm proper PCR amplification of the inserted fragments in the pG1ARIII vectors.

Quantitative Analysis of Androgen Receptor Mutants in Yeast

A liquid β-galactosidase assay was performed to quantify the activity of selected AR mutants, utilizing yeast strain BJ2168 containing the pUCΔSS-26X LacZ-reporter plasmid. Overnight cultures of yeast transformants grown in selective medium were diluted to an OD600 of 0.3 in the same medium supplemented with ligand (DHT, Pg or DHEA) or without hormone and grown until an OD600 of approximately 1.0 was reached. Next, β-galactosidase activity was determined as described previously.42

Mammalian Cell Culture, Transfection and Luciferase Assay

Hep3B (human liver) cells were maintained in α-minimal essential medium (α-MEM) supplemented with 5% FCS and antibiotics. Cells were seeded at a density of 5 × 104 cells/well (1.9 cm2) and grown for 24 hr. Four hours prior to transfection, the medium was replaced by 250 μl α-MEM, supplemented with 5% charcoal-stripped FCS, antibiotics and 1 of the following hormones: DHT, R1881, ASD, Pg, E2, OH-Fl, CPA, bicalutamide, DHEA, DEX, cortisol, aldosterone or TAA. For transfection, 25 μl α-MEM containing 1 μl Fugene 6 (Boehringer Mannheim, Germany), 0.5 μg AR expression plasmid (pSVARIII constructs) and 1 μg MMTV-LUC reporter plasmid were added per well. After 24 hr of incubation, cells were lysed, and luciferase activity was assayed as described previously.49 In the absence of ligand, wild-type and mutant ARs displayed comparable background activities.

RESULTS

Detection of Androgen Receptor H874 and T877 Mutants in A Yeast Screening System

Using random mutagenesis, 2 AR cDNA libraries were generated in a yeast expression vector, 1 with mutations in codon 874, AR(874X), and 1 with mutations in codon 877, AR(877X). Approximately 400 independent yeast colonies from each library were screened for activation of an AR-inducible LacZ reporter by DHT, Pg and DHEA (see Material and Methods). Pg was tested because H874Y and T877 mutant ARs exhibited an increased response to this hormone.12, 19–23, 50 DHEA was chosen because the H874Y and T877A mutant ARs were known to be responsive to this adrenal androgen.21 In each library, approximately 200 of the 400 colonies were β-galactosidase-positive upon incubation with DHT, indicating that the other half of the colonies contained inactivating AR mutations.

Screening of the AR(874X) library resulted in 4 yeast colonies, which were positive after DHT, Pg and DHEA incubation. Colonies positive with 2 tested hormones or with Pg or DHEA alone were not found. Sequencing revealed that the DHT+/Pg+/DHEA+ colonies contained at codon 874 the sequences TAC or TAT, which both encode a tyrosine residue. From this finding it was concluded that a tyrosine residue at position 874 is unique in the generation of an AR, which can be activated not only by DHT, but also by Pg and by DHEA.

Screening of the AR(877X) library resulted in 63 colonies, which were DHT+/Pg+/DHEA+; 14 colonies were DHT+/Pg+/DHEA−. Sequencing of the DHT+/Pg+/DHEA+ colonies revealed 3 different amino acid substitutions at 877: T877A, T877G and T877S. The DHT+/Pg+/DHEA− colonies contained a cysteine residue at 877. Interestingly, as described above, H874Y, T877A and T877S have been found in prostate cancer, whereas T877C and T877G substitutions have not been detected in these tumors.

Hormone-induced Transcriptional Activity of the Androgen Receptor H874Y Mutant in Yeast and in Mammalian Cells

Activation of AR(H874Y) by DHT, Pg and DHEA was quantified in a yeast liquid β-galactosidase assay. The results are summarized in Figure 1. AR(H874Y) exhibited a decrease in AR activation by DHT, compared with wild-type AR (Fig. 1a). At the highest hormone concentrations, Pg and DHEA responses of AR(H874Y) were clearly stronger than that of wild-type AR (Fig. 1b,c).

Figure 1.

Transcriptional activity of wild-type AR and AR(H874Y) in yeast. Yeast cells were cotransfected with the wild-type AR or AR(H874Y) yeast expression vector and the androgen-inducible LacZ reporter pUCΔSS-26X. (a) DHT activation. (b) Pg activation. (c) DHEA activation. Values (±SEM) are the mean of 3 independent experiments each carried out in duplicate.

For direct comparison with the yeast data, Hep3B mammalian cells were cotransfected with the AR(H874Y) mutant or wild-type AR expression plasmid and an MMTV-luciferase reporter plasmid. Transfected cells were incubated in the absence of hormone or in the presence of serial dilutions of a large set of different ligands: the androgens DHT and R1881; the steroidal antiandrogen CPA, the nonsteroidal antiandrogens OH-Fl and bicalutamide; and the steroids Pg and E2. R1881 activated wild-type AR and AR(H874Y) equally; bicalutamide was inactive on both wild-type AR and AR(H874Y) (data not shown). The activities of wild-type AR and AR(H874Y) induced by DHT, Pg, E2, OH-Fl and CPA are shown in Figure 2a and b, respectively. As expected, wild-type AR activation was DHT-specific. Only at high concentrations was some agonistic activity of Pg, E2 and CPA observed; OH-Fl was unable to activate the wild-type AR at all concentrations tested. As in yeast, DHT was found to be a less potent activator of AR(H874Y) than of wild-type AR, which was not due to a lower expression level (data not shown). Both E2 and Pg induced AR(H874Y) activity to almost the same extent as DHT. At high concentrations, OH-Fl exhibited some agonistic activity; agonistic activity of CPA on AR(H874Y) was as low as on wild-type AR.

Figure 2.

Transcriptional activity of wild-type AR and AR(H874Y) in mammalian cells. Hep3B cells were cotransfected with the wild-type AR or AR(H874Y) mammalian expression vector and a MMTV-LUC reporter and activated by different hormones. (a) Wild-type AR. (b) AR(H874Y). Ligands: (♦) DHT, (▪) Pg, (×) E2, (•) CPA, (▴) OH-Fl. Values (±SEM) represent the mean of 3 independent experiments each carried out in duplicate.

Hormone-induced Transcriptional Activity of the Androgen Receptor T877 Mutants in Yeast and Mammalian Cells

Activation of the AR 877 mutants by DHT, Pg and DHEA was analyzed in the quantitative yeast assay as described above for AR(H874Y). The results are summarized in Figure 3. DHT activation of all mutants, AR(T877A), AR(T877S), AR(T877G) and AR(T877C), was comparable to that of wild-type AR (Fig. 3a. AR(T877A) and AR(T877S) displayed the most prominent altered ligand specificity at the 2 Pg and DHEA concentrations tested. Both exhibited increased activation by Pg and DHEA compared with wild-type AR (Fig. 3b,c). AR(T877C) and AR(T877G) were activated by Pg (Fig. 3b), but only AR(T877G) was DHEA-inducible (Fig. 3c), in agreement with the qualitative yeast screening.

Figure 3.

Transcriptional activity of wild-type AR and AR(T877) mutants in yeast. Yeast cells were cotransfected with wild-type AR yeast expression vector, or AR(T877A), or AR(T877C), or AR(T877G) or AR(T877S) mutant yeast expression vectors and the androgen-inducible LacZ-reporter pUCΔSS-26X. (a) DHT activation. (b) Pg activation. (c) DHEA activation. Values (±SEM) represent the mean of 3 independent experiments each carried out in duplicate.

Ligand specificity studies of the 4 AR 877 mutants were extended to mammalian Hep3B cells, using the same set of ligands as used for the wild-type AR and AR(H874Y) studies shown in Figure 2. R1881 activation, which was identical for wild-type AR and the 4 877 mutants, is not shown. Bicalutamide did not activate any of the 877 mutants (data not shown). The T877A and T877S substitutions introduced the most dramatic alterations in ligand specificity (Fig. 4; for wild-type AR, see Fig. 2a). Both AR(T877A) and AR(T877S) exhibited a strong activation by Pg, E2 and the antiandrogen CPA (Fig. 4a,d). AR(T877A) was more responsive to OH-Fl than AR(T877S) (Fig. 4a,d). AR(T877C) and AR(T877G) ligand specificity was less altered. Although Pg activation could clearly be established, agonistic activity of other ligands was limited (Fig. 4b,c). Differences in ligand responses between the AR mutants were not due to different protein levels (data not shown).

Figure 4.

Transcriptional activity of AR(T877) mutants in mammalian cells. Hep3B cells were cotransfected with (a) AR(T877A), (b) AR(T877C), (c) AR(T877G) and (d) AR(T877S) mammalian expression vector and an MMTV-LUC reporter. Ligands: (♦) DHT, (▪) Pg, (×) E2, (•) CPA, (▴) OH-Fl. Values (±SEM) represent the mean of 3 independent experiments each carried out in duplicate.

Transcriptional Activation of Androgen Receptor H874 and T877 Mutants by Adrenal Steroids and Synthetic Glucocorticoids

The AR 874 and AR 877 mutants were also assayed in Hep3B cells for their activation by the adrenal steroids DHEA and ASD (androgens), cortisol (glucocorticoid) and aldosterone (mineralocorticoid), as well as the synthetic glucocorticoids DEX and TAA. Activation of wild-type AR and all mutant ARs by ASD was identical; TAA was unable to activate wild-type and mutant ARs (data not shown). For the other ligands a remarkable variation in activation patterns of the different mutants was observed. Figure 5a displays the activities of wild-type AR and all AR mutants induced by high concentrations DHEA, cortisol, DEX and aldosterone (10−6 M). In Figure 5b–e, the ligand concentration dependent activation of selected mutants is shown. AR(H874Y), AR(T877A) and AR(T877S) were clearly responsive to DHEA (Fig. 5a,b). Also, a concentration-dependent activation of AR(H874Y) and AR(T877A) by aldosterone, cortisol and DEX was observed (Fig. 5a,c–e). In contrast, AR(T877S) and AR(T877G) could hardly be activated by these ligands (Fig. 5a). AR(T877C) did not respond to any of the ligands (Fig. 5a).

Figure 5.

DHEA- and corticoid-induced transcriptional activity of wild-type AR and AR mutants. Hep3B cells were cotransfected with the wild-type AR, AR(H874Y) or AR(T877) mutant mammalian expression vector and an MMTV-LUC reporter. (a) Incubation in the absence of hormone, or in the presence of DHEA, cortisol, DEX or ALD (all at 10−6 M). (b) DHEA activation. (c) ALD activation. (d) Cortisol activation. (e) DEX activation. (♦)WT, wild-type AR; (▴)Y, AR(T874Y); (▪)A, AR(T877A); and (•)S, AR(T877S). Values (±SEM) represent the mean of 3 independent experiments each carried out in duplicate.

DISCUSSION

Mutations in the AR have been described in several diseases. In androgen insensitivity, which is an inherited defect of male development, over 100 amino acid substitutions in the AR LBD have been documented (http://www.mcgill.ca/androgendb).1, 51 These mutations completely or partially inactivate AR function. In Kennedy's disease or spinal and bulbar muscular atrophy (SBMA), an expanded (CAG)n repeat results in a longer glutamine stretch in the AR TAD.52 In prostate cancer, AR mutants are rare in primary and locally progressive tumors but are more frequent in metastatic disease, after endocrine therapy.16–18, 22, 53–57 The relevance of most AR mutants in progressive prostate cancer remains to be established. The mutants investigated in more detail are functionally different from AR mutants in androgen insensitivity and tend to cluster in different regions of the LBD.57

The most frequently described AR mutations in prostate cancer are substitutions of H874 and T877, which are both in helix 11 of the AR LBD.36, 37, 39, 41 AR(T877A), which has originally been detected in the LNCaP cell line, seems to be the preferred amino acid substitution in endocrine therapy-resistant prostate cancer.12, 14–17 Like the less common H874Y and T877S amino acid substitutions, T877A broadens AR ligand specificity in such a manner that not only androgens but also other sex steroids and antiandrogens can activate the AR.12, 19–23 The recently identified T877A&L701H AR double mutant exhibited an even broader ligand specificity than the T877A single mutant,58 adding cortisol to the spectrum of strong activators. There is increasing evidence that the AR LBD mutants with less specific ligand responsiveness are of clinical relevance in a subset of endocrine therapy resistant prostate cancers.16, 17

In the present study 2 types of experiments were carried out. First, ARs randomly mutated at positions 874 or 877 were screened for broadened ligand specificity in a yeast readout system. Second, AR mutants with broadened ligand responsiveness were assayed in mammalian cells for activation by 13 different ligands, including sex steroids, adrenal steroids and antiandrogens. From our findings, several important conclusions can be drawn:

  • 1In the AR cDNA library randomly mutated at codon 874, AR(H874Y) was the only mutant able to broaden AR ligand specificity. In the 877 AR cDNA library, an alanine, serine, glycine or cysteine residue at position 877 broadened AR ligand response. As pointed out above, AR(H874Y), AR(T877A) and AR(T877S) are well known from prostate cancer; AR(T877G) and AR(T877C) have never been described in prostate cancer. The random mutagenesis system used in our study allowed the screening of all triplets possible for codons 874 and 877. As expected, among the identified AR mutants, 1-, 2- and 3-base deviations from the wild-type codon were detected. In nature, the chance of more than 1 point mutation within a codon is extremely low. Indeed, all amino acid substitutions found at AR codons 874 and 877 in prostate cancer are due to single-point mutations: H874Y: CAT>TAT,17, 21, 22 T877A: ACT>GCT,12–17, 22 T877S: ACT>TCT.17, 22 For T877C or T877G substitutions, at least 2 bases need to be mutated. This can explain their absence in prostate cancer. In conclusion, the H874Y substitution is not only unique at this position in prostate cancer but is also the only possibility at this position to broaden AR ligand specificity. The T877A and T877S substitutions in prostate cancer are not unique, in that they are not the only substitutions that can broaden AR ligand specificity at this position, but they seem to be sequence-driven selections of 4 possible amino acid substitutions.
  • 2The results of the extensive series of transactivation experiments with wild-type and mutated ARs in mammalian cells are summarized in Table 1.
Table I. Ligand Responsiveness of Ar Mutants to 13 Different Hormones as Tested on An MMTV-LUC Reporter in Transiently Transfected Mammalian (HEP3B) Cells1
 Wild-typeAR (H874Y)AR (T877A)AR (T877C)AR (T877G)AR (T877S)
  • 1

    Degree of ligand responsiveness: −, no activity; +, low activity; ++, moderate activity; +++, high activity, comparable to wild-type AR responsiveness to DHT.Importantly, each mutant displayed its own characteristic spectrum of ligand responsiveness. Differences in ligand affinities, as well as differences in ligand-induced conformational changes, may account for this variation. Most remarkable are the similarities between activation of AR(H874Y) and AR(T877A) by the various ligands. Although they are completely different, both exhibit identical responses to the glucocorticoids cortisol and DEX, as well as the mineralocorticoid aldosterone. Zhao et al.58 described activation of the AR double mutant T877A&L701H by cortisol, but they did not find cortisol responsiveness of the single mutant AR(T877A). The apparent discrepancy with our data might be due to a less sensitive assay or a different cell line used for transfection experiments. Our findings warrant a further investigation of the role of glucocorticoids in prostate cancer patients carrying a mutated AR.

Androgens      
 DHT++++++++++++++++++
 R1881++++++++++++++++++
Sex steroids      
 Pg++++++++++++++
 E2+++++++++
Antiandrogens      
 CPA++++++++++
 OH-F1+++++
 Bicalutamide
Adrenal androgens      
 DHEA++±+
 ASD++++++++++++
Glucocorticoids      
 Cortisol++
 DEX++
 TAA
Mineralocorticoid      
 Aldosterone++

Based on the crystal structures of closely related steroid hormone receptor LBDs, homology models of the AR LBD have been constructed.36, 37, 38 These models indicated that T877 is part of the ligand binding pocket and interacts directly with the ligand; H874 does not participate directly in ligand binding. Recently, the predictions of AR LBD folding and ligand interaction were modified and extended by elucidation of the crystal structures of the wild-type and T877A mutant AR LBD complexed with androgens.39, 40 For the wild-type AR LBD, 18 amino acids were found to contact the ligand directly. Importantly, T877 in helix 11 of the LBD, together with N705 in helix 3, forms hydrogen bonds to the 17β-hydroxyl group of R1881, which is also present in DHT (Fig. 6). As predicted, H874, which is also in helix 11, projects away from the ligand binding pocket.

Figure 6.

Chemical structures of steroids used in our study.

The crystal structure of the T877A mutant AR LBD complexed with DHT revealed an increased space in the ligand binding pocket.40 The amino acid residues serine, glycine and cysteine at position 877 have in common that, like alanine, they are all smaller than the threonine residue at this position in the wild-type AR.59 Thus, substitution of T877 by S, G or C will also increase the space of the ligand binding pocket. This larger space will facilitate appropriate entrance by ligands with more bulky side chains at C17 like Pg, cortisol, DEX and aldosterone (Fig. 6). This may allow a conformational change in the LBD, which is favorable for the AR transactivation function.

A larger binding pocket may also explain appropriate folding of the LBD induced by the antiandrogen CPA, resulting in agonistic activity. The synthetic glucocorticoid TAA might be too big for proper entrance into an enlarged AR ligand binding pocket (Fig. 6). Not only the size of the C17 side chain, but also slight differences in overall conformation of a steroid, as determined by the A- to D-ring moities, might contribute to positioning in the ligand binding pocket. Particularly in the case of E2 and DHEA, which both have a small C17 side chain (Fig. 6), but not excluding other ligands, the larger binding pocket might be needed for appropriate binding of these different conformations. Elucidation of the crystal structures of the various AR mutants complexed with different ligands is needed to prove and extend these hypotheses.

Many other mutations have been described in the AR LBD in prostate cancer, but most of the mutants have not been characterized.22, 55, 56, 60, 61 In contrast to L701 and T877, none of the mutated amino acid residues can be predicted to contact the ligand directly.39, 40 This is also true for mutant V715M, which clearly displays broadened ligand response.23, 61, 62 Thus, as for H874Y, for the V715M mutant a different mechanism of activation can be predicted.

One of the most important questions that remain to be addressed is the identification of the physiologic ligand of the AR mutants in endocrine therapy-resistant prostate cancer. Most AR LBD mutants with a broadened ligand response seem to be induced or selected during antiandrogen therapy.17 Our findings suggest that in androgen-depleted patients after antiandrogen withdrawal, Pg or cortisol might be the physiologic ligands for activation of a mutated AR. The concentration of circulating Pg in men (0.3–0.9 nM)63 seems sufficient for such a function, because of the strong response especially of AR(T877A) and AR(T877S) to this ligand (Fig. 4). Although cortisol is a less potent activator of the mutants (Fig. 5), its high concentration in the circulation (70–550 nM)63 warrants further investigation of its role in patients carrying an AR(H874Y) or AR(T877A) mutation. Activation of these mutants by cortisol would be in line with activation of the AR(L701H) single and AR(L701H/T877A) double mutant by cortisol.58 Circulating E2, DHEA and aldosterone concentrations (73–184 pM, 6–28 nM and 83–832 pM, respectively),63 seem too low to account for such a function. However, inactive DHEA sulfate, with a serum level of 1–9 μM,63 can be converted into DHEA in the prostate, and the resulting local DHEA concentrations may be sufficient for activation of the various AR mutants (Fig. 5, Table 1).64

To obtain relevant data on noncognate ligands as activators of mutant ARs, extended studies with prostate cell lines containing mutated ARs are needed. In a previous study with DEX-incubated LNCaP cells, applied hormone concentrations were too low to be able to observe an effect on the endogenous AR(T877A).65 Monitoring of the response of LNCaP (T877A) and CWR22 (H874Y)12, 21, 66 cells to physiologic concentrations of different ligands, including glucocorticoids and DHEA, will give important supportive information about the role of mutated ARs in prostate cancer.

Acknowledgements

We thank D. Picard for supplying plasmid pUCΔSS-26X, P. Farla for help in preparation of the manuscript and E.J. Dubbink for fruitful discussions.

Ancillary