Joanne Edwards, Endocrine Cancer Group, Section of Surgical and Translational Research, Division of Cancer and Molecular Pathology, University Department of Surgery, Queen Elizabeth Building, Glasgow Royal Infirmary, Glasgow, Scotland G31 2ER, UK. e-mail: firstname.lastname@example.org
In 2001, prostate cancer was responsible for ≈ 10 000 deaths in the UK, making it the second most common cause of male cancer-related deaths (http://www.prostate-research.org.uk). Treatment for advanced or metastatic prostate cancer has relied on androgen-deprivation therapy (ADT) for the past 50 years . At present, few treatment options offer effective relief for patients who develop resistance to ADT. The lack of novel and effective therapies to treat this disease reflects a poor understanding of the mechanisms underlying the development of both the primary disease and more particularly those events which drive resistance. In Part 1 of this review we described how prostate cancer growth is stimulated in response to androgens, and consequently how ADT acts to combat this. We then explained how modifications to the androgen receptor (AR) via mutations, amplification and phosphorylation may affect the development of androgen escape. However, in recent years it has become increasingly apparent that androgen escape may also involve mechanisms that do not directly modify the AR. It is these mechanisms which will be discussed in this part of the review, including AR cofactors and how they might influence AR transactivation, and how signal-transduction pathways can act independently of the AR to influence prostate cancer cell growth and survival.
ACTIVATION OF THE AR
Androgen-dependent transcription, as described in Part 1, may be significantly enhanced by interactions between the AR and ‘coactivators’, proteins that generally do not themselves bind DNA, but are recruited to gene-promoter regions through protein–protein interactions with AR, usually in a ligand-dependent manner. A comprehensive list of currently known proteins that interact with the AR is available in the appendix of a recent review by Lee and Chan . Coactivators function to facilitate the assembly of transcription factors into a stable pre-initiation complex. In addition, some coactivators, including steroid receptor coactivator-1 (SRC-1), cAMP response element-binding protein (CBP) and p300, can also remodel chromatin by acetylating histones and recruiting the p300/CBP-associated factor which harbours intrinsic histone acetyltransferase activity. When the ligand-bound AR dimer binds to androgen-receptor elements, coactivators and p300/CBP-associated factor are recruited. This loosens the nucleosomal structure of the gene, by targeted histone acetylation, and initiates the stable assembly of the pre-initiation complex via their bridging function. The end result is an enhanced rate of transcription initiation by RNA polymerase II.
HORMONE-RESISTANT PROSTATE CANCER
As discussed in Part 1, the function of ADT is to prevent the activation of AR-mediated gene transcription. Recently it was shown that androgen escape may not only arise by modification of the AR, but may also involve the action of AR coactivators or pathways independent of the AR. In this part we summarize the current understanding of the molecular mechanisms underlying androgen escape, with particular emphasis on AR coactivators and AR bypass pathways.
These by definition are proteins that, through binding directly or in a multiprotein complex to the AR, increase or inhibit the transcriptional activity of the AR. It is most likely that AR coactivators contribute to the development of androgen-insensitive prostate cancer by increasing AR transcriptional activity in the presence of low ligand concentrations or by altering the ligand specificity of the AR, allowing antiandrogens and oestrogens to act as agonists .
Coactivators known to alter ligand specify of the AR include AR-associated proteins CBP, β catenin, ARA55 and ARA70 [5–7]. These coactivators can change the action of antagonists to agonists or allow other steroids to activate the AR, and may thus be important in the development of clinical androgen escape.
CBP is overexpressed in hormone-refractory prostate cancer and allows hydroxyflutamide to function as an agonist in vitro. The AR coactivators ARA55 and β catenin/S33F, alter AR ligand specificity and enhance AR transactivation in response to oestradiol . Phosphorylation of ARA55 by proline-rich tyrosine kinase-2 (PYK2) decreases AR transcriptional activity, as phosphorylated ARA55 cannot interact with the AR. Both PYK2 and ARA55 are expressed in normal prostate epithelium, but as prostrate cancer progresses the expression of PYK2 is reduced, resulting in decreased ARA55 phosphorylation and increased AR/ARA55 interaction . This ultimately results in an increase in AR-mediated transcription, and increased PSA expression.
ARA70 may also modify AR ligand specificity in the development of hormone-refractory cancer. Yeh et al. first reported ARA70 as an AR-specific coactivator in 1996, and ARA70 is overexpressed in prostate cancer and hormone-refractory CWR22 xeonografts . ARA70 interacts primarily with the AR ligand-binding domain (LBD), and enables the antiandrogens hydroxyflutamide and bicalutamide to function as AR ligands, increasing transcriptional activity . Elevated ARA70 expression in hormone-refractory prostate cancer promoting AR activation by antiandrogens may contribute towards the failure of maximum androgen blockade even in the presence of wild-type AR. In addition to the action that ARA70 has on antiandrogens, in vitro experiments show that increased ARA70 expression allows low concentrations of adrenal androgens (similar to those found in serum during maximum androgen blockade) or oestradiol to activate AR .
However, the function of ARA70 as a specific AR coactivator is disputed by two groups who reported that ARA70 binds to other nuclear receptors and that up-regulation of AR activity by ARA70 does not add to the enhancement of activity caused by other coactivators [11,12]. In addition Alen et al. reported that in vitro mutations in the LBD of the AR that impaired the interaction with ARA70 and AR only moderately decreased AR transcriptional activity. However, the weight of evidence supports the role of ARA70, interacting with both wild-type and mutated AR in the development of hormone-refractory disease .
Coactivators that influence the development of androgen escape by activating the AR in the absence of ligand (or at low ligand concentrations) include SRC-1, p300, Tat interactive protein 60 kDa (Tip60), SRC-3 and c-Jun. C-Jun functions as an AR coactivator by binding to the N-terminal binding domain at amino acids 503–555 . This region contains many phosphorylation consensus sites and is critical for ligand-independent transactivation of the AR , but the role of phosphorylation in promoting or inhibiting AR/C-Jun interaction remains unclear. Binding of c-Jun to the N-terminal binding domain promotes AR homodimerization (via an AR N-C domain interaction) allowing AR to bind to DNA in a sequence-specific manner and act as a transcription factor even in the absence of ligand . However, in vivo, either in normal physiological or androgen-depleted states, it is likely that this interaction serves to potentiate the action of AR in the presence of low concentrations of ligand. It is thought that c-Jun can act in conjunction with the coactivator TIF-2 (SRC-2), which is also overexpressed in hormone-refractory tumours, to potentiate AR transactivation. The effect of c-Jun and TIF-2 binding on AR transactivation is additive .
TIF-2 is a member of the steroid receptor cofactor family (SRC1, TIF2 [SRC2] and AIB1 [SRC3/RAC3]). This family is commonly overexpressed in hormone-refractory prostate cancer and is known to potentiate AR transcriptional activity in the presence of androgens . The formation of AR homodimers may be mediated by SRC-1, which targets both the N-terminal domain and the LBD . It has also been reported that the mitogen-activated protein kinase (MAPK) may increase AR activity by phosphorylating SRC-1, independent of AR phosphorylation (Fig. 1) [13,18]. This offers an alternative route for MAPK signal transduction to influence the development of androgen escape . SRC-1 is increased in a large proportion of recurrent prostate tumours, and in LNCaP cells it enhances ligand-independent activation of the AR by binding to the N-terminal binding domain. Physical interaction between the N-terminal binding domain of AR and SRC-1 is critical for androgen-independent AR signalling in LNCaP cells . Although such physical AR/SRC-1 interaction does not require phosphorylation of SRC-1 by MAPK, it is only when SRC-1 is phosphorylated by MAPK that the AR is activated in the absence of androgens . The interaction of phosphorylated SRC-1 with the AR results in activation of the AR to the same magnitude as that obtained by dihydrotestosterone (DHT) . In the physiological situation it may be a combination of MAPK phosphorylating the AR to sensitize it to DHT, allowing it to enter the nucleus, and MAPK phosphorylating SRC-1 to increase transcriptional activity. SRC3 expression correlates with decreased disease-free survival and facilitates RNA polymerase II recruitment to a distant enhancer element of the PSA gene, resulting in increased PSA levels in response to very low level adrenal androgens [17,20].
AR coactivator Tip60 expression and nuclear localization increase in response to androgen withdrawal in both CWR22 prostate xenografts and LNCaP prostate cancer cells . In hormone-refractory tumours the exclusive nuclear location of Tip60 may mediate increased AR sensitivity to low concentrations of androgens, as Tip60 is linked with transcription of the PSA gene in hormone-refractory cell lines and is thought to influence transcription of other AR genes by inducing changes to the acetylation status of AR .
Expression of the AR coactivator p300 correlates with a high Gleason score and is associated with prostate cancer progression . P300 is associated with proliferation of prostate cancer cells both in vitro and in vivo, and is thought to be involved with the cell cycle. In prostate cancer cell line models, interleukin-6-stimulated growth in the absence of androgens requires p300, and early apoptosis is not detected after silencing p300. Therefore p300 may be important in the development of hormone-refractory prostate cancer .
The balance of coactivators to co-repressors was also shown to influence the development of androgen escape, especially in the presence of AR antagonists. Cell-line studies showed that adding bicalutamide to cell culture medium results in a slight reduction in the interaction of SRC-1 with AR. However, more significantly, when bicalutamide is added to the cell culture medium, there is a large increase in the interaction between the AR and the co-repressor SMRT. However, if SRC-1 is over-expressed in this cell line, this interacts with the receptor in preference to the co-repressor. Therefore, as with breast cancer resistance, androgen independence may be a combination of the association of coactivators with AR and recruitment of co-repressors. This balance should be investigated in the clinical situation in more detail . It was suggested that hormone-independent transcriptional activity of the AR may be mediated solely through interactions with coactivators such as the SRC family and p300 .
The third mode by which AR cofactors can influence the development of androgen escape is not as well established and involves binding of cofactors to the AR, resulting in AR translocation to the nucleus. An example of this is STAT3, which is a member of the JAK/STAT3 pathway. In vitro studies show that interleukin-6 activation of the JAK/STAT3 pathway is accompanied by transition from androgen-sensitive to -insensitive prostate cancer cell growth  (Fig. 2). Levels of activated STAT3 are significantly higher in the hormone-refractory prostate cancer cell lines (DU145 and PC3) than in hormone-sensitive cell lines (LNCaP cells) . In LNCaP cells the activated dimer of STAT3 binds ligand-free AR before entering the nucleus, therefore facilitating the translocation of AR to the nucleus in the absence of androgens . The AR/STAT-3 complex can activate androgen-regulated gene transcription, and PSA expression is elevated even in the absence of androgens [27,28]. This mechanism is supported by data showing that interleukin-6 can activate the AR in a ligand-independent manner . However, the oncogenic role of STAT3 in prostate cancer is not clearly established and STAT3 has also been correlated with interleukin-6-induced growth arrest in cell lines, including LNCaP cells [30,31].
In summary, there is now strong in vitro evidence that implicates AR cofactors in the development of androgen escape via three routes: altering ligand specificity, activation in the presence of low levels of androgens and translocation to the nucleus. However, these have not all been confirmed in vivo. The clinical evidence available to support the role of cofactors in altering the specificity of AR to surrogate ligands (e.g. flutamide) may explain why PSA levels decrease in some patients after antiandrogen withdrawal. As in breast cancer, this may indicate a degree of antisteroid therapy dependence in these tumours. These patients may respond well to removal of antiandrogen therapy. In the light of strong in vitro evidence for the involvement of AR cofactors in the promotion of steroid resistance, studies seeking to identify the in vivo significance of these findings are urgently required. Only when such evidence is available can we determine the potential of these pathways as therapeutic targets in hormone-insensitive prostate cancers.
AR BYPASS PATHWAYS
As discussed in Part 1, the MAPK (Fig. 1), phosphatidylinositol 3-kinase (PI3K) (Fig. 3) and protein kinase C (PKC) (Fig. 4) cascades may be involved in the development of androgen escape via activation of the AR. However, these pathways may also be involved in the development of androgen escape by increasing cell proliferation and decreasing apoptosis, completely independently of the AR. The Ras/Raf/MAPK cascade may influence cell-cycle regulation and/or increase cell proliferation via AP-1, c-MYC and NF-κB transcription factors [32–34] (Fig. 1). Members on the MAPK cascade are amplified in hormone-refractory prostate cancer  and cell-line studies show that androgen escape may be induced by transfection with Ras, resulting in increased expression and activation of MAPK . Weber et al. showed that after castration, prostate cancer recurrence in mice correlated with up-regulation of phosphorylated and hence activated MAPK. It was also reported [36,37] that in human tissue an increase in Raf expression in the transition from hormone-sensitive to hormone-refractory prostate cancer is associated with time to relapse, and expression of activated MAPK increases with Gleason score, tumour grade and androgen resistance. Hence it is evident that the MAPK cascade is associated with the development of hormone-refractory cancer, but the downstream events remain to be clarified. We showed recently that those patients who express high levels of phosphorylated c-Jun survive for significantly less time than those who express low levels of phosphorylated c-June . These data support a role for AP-1 activation, possibly via the MAPK cascade in the development of hormone-refractory prostate cancer. AP-1 is involved in the control of cell growth and differentiation, and is composed of the nuclear proteins c-Jun and c-Fos, encoded by c-jun and c-fos proto-oncogenes. AP-1 can either be a c-Jun/c-Jun homodimer or a c-Jun/c-Fos heterodimer, the latter being the most stable . Formation of either dimer requires c-Jun phosphorylation at serine residues 63 and 73 by c-Jun N-terminus kinase . AP-1 induces transcriptional activation by binding to the TPA responsive element (TRE) . TREs are recognized by both AP-1 c-Jun homodimers, and c-Jun/c-Fos heterodimers . AP-1 is thought to influence the development of androgen escape by competing with the AR to alter expression of androgen-regulated genes (Fig. 1) . AR and AP-1 are capable of binding to each other; this protein/protein interaction prevents either from being able to bind to DNA, and hence results in a decrease in gene transcription . However, evidence also suggests that AP-1 can increase the expression of androgen-regulated genes by binding to a TRE domain within the promoter region . Therefore, the effect of AP-1 on androgen-regulated gene expression could depend on the ratio of AR to AP-1 and the ability of free AP-1 or AR to bind to specific promoter regions within the androgen-regulated gene . Such competition could influence the ability of AP-1 to increase androgen-regulated genes in the absence of androgens, and hence might influence the development of androgen escape . This is especially important in androgen-regulated genes which contain multiple TREs in the promoter region, e.g. PSA and prostate-specific membrane antigen . In a situation where the ratio of AP-1 to AR is high (e.g. in the absence of androgens), there would be less AR available to initiate transcription by binding to the ARE . However, there would be excess AP-1 available for binding to an alternative TRE, resulting in an increase in androgen-regulated gene expression. Therefore it is conceivable that such a situation could influence the development of androgen escape, i.e. increase androgen-regulated gene expression in the absence of androgens [39,42]. In vitro work showed that in PC3 cells (prostate cancer cells which have progressed to androgen independence), the intracellular concentration of c-Jun and c-Fos is seven times greater than in LNCaP cells (androgen-sensitive prostate cancer cells) . This suggests that AP-1 influences androgen escape in the PC-3 cell line . Our work in vivo substantiates that found in cell-line studies, that AP-1 is involved in the development of hormone-refractory prostate cancer .
Similarly, AKT may influence the development of androgen escape independent of AR phosphorylation. AKT has roles in the control of cell apoptosis and proliferation in prostate cancer cell lines, and may inhibit apoptosis by suppressing the pro-apoptotic functions of BAD, via Ser136 phosphorylation and caspase 9 (Fig. 3) . AKT may also signal for G1 cell-cycle progression by mTOR and p70S6K which act via p21CIP/WAF1, and hence CDK4 and cyclin D1 (Fig. 3) . In addition, AKT may inactivate the Forkhead family of transcription factors to decrease protein expression of p27KIP1, a cell-cycle regulator . Therefore there are many routes by which AKT may influence the development of hormone-refractory prostate cancer and all of these mechanisms function in prostate cancer cell lines [44,45].
PKC also influences the development of androgen escape in cell-line studies. We recently showed that those patients who have an increase in PKC expression with the development of hormone-refractory prostate cancer survive for significantly less time than those whose PKC expression remains unchanged or falls . This does not, in contrast to data from in vitro studies, appear to be mediated via AP-1 activation [40,41] (Fig. 4). PKC is widely expressed in tissue, and abnormal levels have been found in many transformed cell lines and tumours . The PKC family consists of at least 12 isoforms that reportedly have different and occasionally opposing roles in cell growth and differentiation . The diversity of PKC isoforms was highlighted in a recent review by Mackay and Twelves ; they reported that PKC α, δ and ɛ may activate the Raf1/MAPK pathway via Raf phosphorylation, PKC θ may activate the Rac1/JNK pathway via Rac1, and PKC α, β1 and γ may specifically inactivate GSK-3β by phosphorylation, leading to activation of the c-Jun transcription factor . There is therefore significant potential for PKC to interact with many of the pathways described above. Data on the in vivo expression of specific PKC isoforms in hormone-resistant prostate cancer is currently lacking, and therefore it is difficult to speculate at present which of the above mechanisms may be responsible for the association between PKC expression and hormone escape.
In conclusion, there is strong evidence that in vitro signal-transduction mechanisms may promote androgen escape independent of the AR, by regulating apoptosis and cell proliferation. However, as discussed below, these pathways interact significantly with AR coactivators and co-repressors, as well as directly modifying the AR itself. The pluripotency of these signalling pathways, linked to a growing body of evidence that they provide effective therapeutic targets, suggests that future therapies for hormone-resistant prostate cancer may be directed against specific targets within these pathways.
It is likely that prostate cancer cells achieve the transition from androgen-sensitivity to androgen-independence by different multistep routes, including adapting the AR pathway via the MAPK, PI3K, JAK/STAT pathways or through bypassing the AR by inhibiting apoptosis or increasing cellular proliferation. To develop future therapies it is crucial that the molecular alterations underlying the development of androgen escape are fully understood.
As discussed in Part 1, it is now apparent that the control of AR function involves interaction of the receptor with many co-activators and co-repressors, and that these interactions, and the function of the AR itself, are modified significantly by post-translational modification (generally via phosphorylation). The signalling pathways which mediate these modifications can also promote tumour growth by bypassing the AR completely. Given the complexity of these pathways, it is likely that prostate cancer cells achieve the transition to androgen-independent growth by different multistep routes. These include cofactors, adapting the AR pathway via the MAPK, PI3K and PKC pathways, or by bypassing the AR by inhibiting apoptosis or increasing cellular proliferation. Further, many mechanisms may be active within one cell population. However, despite this complexity, significant progress has been made and it may now be possible to predict the most fruitful avenues for future therapeutic studies.
The dominant pathways involved in the development of androgen escape both via AR modifications (as described in Part 1) and independent of the AR are the MAPK and PI3K pathways. These may act by altering AR sensitivity to androgens and altering expression of genes responsible for promoting tumour growth and inhibiting apoptosis (Figs 1 and 3). Already there seems to be sufficient evidence to begin early clinical trials of drugs that inhibit these pathways, with agents such as farnesyl transferase inhibitors or AKT/mTOR-based inhibitors.
Current evidence suggests that in the future the most effective way of treating prostate cancer would involve profiling individual tumours to identify appropriate therapies. It appears that it is only by matching therapies to individual tumours that patients can be offered significant improvements on the current approach to treating prostate cancer. By using this approach we think that we will begin to solve the problem of androgen escape.
CONFLICT OF INTEREST
None declared. Source of funding: Prostate Cancer Research Foundation.