Nuclear receptor CTDs serve a dual function, possessing both the LBD and AF-2 coactivator binding surface. The model for activation of the AR and other steroid receptors is that ligand binding induces an AF-2 conformation that is permissive for docking with NR boxes (LxxLL and related motifs) of coactivator molecules, which are required for transcriptional activation. For the AR, this model is complicated by the fact that AR AF-2 can also bind an FxxLF motif in the AR NTD, and thus mediate an interaction between the N- and C-termini (9
). In in vitro
assays, the affinity of AF-2 for AR-derived FxxLF peptides is higher than for coactivator-derived LxxLL peptides (6
). As a result, the importance and role of AR AF-2 in transcriptional activation of target genes has been a topic of debate. Indeed, AR AF-2 functions as a very weak ligand-dependent transcriptional activation domain in isolation (9
). However, the isolated AR NTD is stronger in reporter gene assays than the isolated AR CTD (27
). Our previous studies have demonstrated that the AR NTD may play a major role in mediating PCa therapy resistance (10
). The data we present in this study supports this concept, and has shown for the first time that naturally-occurring AR isoforms, consisting of the NTD and DBD, can support the expression of many endogenous AR target genes as well as the growth of androgen-refractory PCa cells in the absence of androgens.
Various mechanisms of resistance to androgen ablation therapy have been proposed and demonstrated in various models of PCa progression. For example, mutations in the AR LBD broaden the ligand specificity of the AR, thus permitting activation by alternative steroids or even antiandrogens (2
). AR overexpression, which may result from AR gene amplification and/or enhanced AR transcription, has been shown to sensitize the AR to castrate levels of androgens (24
). Indeed, tissue androgens in androgen refractory PCa may persist at levels sufficient to elicit AR activation, and may arise from aberrant intra-tumor androgen production (31
). All of these mechanisms are ligand-dependent in nature, and have called into question the importance of ligand-independent AR activation for the development and progression of androgen-refractory PCa. However, in this study, our data demonstrates that the androgen-refractory phenotype of 22Rv1 cells is due to novel AR isoforms that function in a completely ligand-independent fashion. Moreover, these isoforms would be completely resistant to all current androgen ablation regimens, which are targeted to the AR CTD. Our observation that other commonly-used PCa cell lines possess the machinery required to synthesize the ligand-independent AREx1/2/2b
isoform provides a novel and simple avenue for these PCa cells to achieve androgen-independence. Indeed, in two xenograft-based models of PCa progression, we observed increased expression of AR Exon 1/2/2b mRNA in AI vs. AD tumors. Moreover, we observed a similar increased expression of a 75kDa AR species in these AI vs. AD tumors. Our data strongly suggests that this protein species is AREx1/2/2b
; however, a definite conclusion will have to await the development of antibodies specific for this AR isoform. Importantly, in these models, AREx1/2/2b
overexpression was accompanied by overexpression of full-length AR, which has been previously described (24
). Since mRNAs encoding both full-length AR and AREx1/2/2b
are spliced from the same pre-mRNA, enhanced transcription of the AR gene would result in enhanced overall expression of both isoforms. It will therefore be important to assess the relative contributions of AR and AREx1/2/2b
overexpression to PCa progression during androgen ablation.
Two previous studies have investigated the nature of the small AR isoform(s) expressed in 22Rv1 cells, and have concluded that these species arise from calpain-mediated cleavage of AREx3dup
). Our data oppose this conclusion, and demonstrates that the AR NTD/DBD isoforms expressed in 22Rv1 cells arise from splicing of a novel AR
Exon 2b after either AR
Exon 2 or Exon 3. AR
Exon 2b spliced after Exon 2 or Exon 3 gives rise to protein species possessing the entire AR NTD fused to the first zinc finger of the AR DBD (termed AREx1/2/2b
), or the entire AR NTD fused to the complete AR DBD (termed AREx1/2/3/2b
), respectively. Structural studies with AR DBD demonstrate that the first zinc finger, encoded by Exon 2, harbors the recognition helix that directly engages with one hexameric half-site in an androgen response element (ARE) (5
). The second zinc finger, encoded by Exon 3, mediates dimerization with an AR molecule engaged with the neighboring ARE half-site (5
). Our finding that AREx1/2/2b
is able to constitutively activate AR-responsive promoters demonstrates that the first zinc finger is sufficient for the AR to engage with AREs, which agrees with this structural data. However, since the second Exon 3-encoded zinc finger is responsible for AR dimerization and subsequent stabilization of ARE-bound ARs, the repertoire of genes that are activated by AREx1/2/2b
may be diminished compared to the repertoire of genes that are activated by full-length, liganded AR. Dissecting the differential patterns of gene activation by AREx1/2/2b
and full-length, liganded AR will be important for understanding the relative abilities of these isoforms to mediate PCa growth.
In summary, our data indicates that AR truncation, resulting from aberrant splicing of a novel AR Exon 2b, uncouples androgen signaling from AR activity in 22Rv1 cells. AR mRNA and protein containing Exon 2b were also identified in commonly used PCa cell lines. Our data further shows that AR mRNA species containing Exon 2b were enriched in xenograft-based models of PCa progression. Therefore, inclusion of Exon 2b in AR mRNAs represents an effective mechanism by which constitutively active AR proteins can be produced that are completely resistant to current CTD-directed PCa therapies.