Stromal-epithelial interactions are responsible for prostate tumor progression through an androgen-related mechanism

Abstract

While several hypotheses have been put forward to explain how prostate tumors become resistant to androgen deprivation therapy, the mechanism by which prostate tumors have increased androgen concentrations as compared to the serum has been poorly explored. Using a stromal/epithelial cell co-culture model, Mizokami et al. have demonstrated how prostate-, bone- and prostate tumor-derived stromal cells participate with tumor-derived epithelial cells (i.e., LNCaP cells) to produce active androgens from a readily available substrate during androgen deprivation therapy, dehydroepiandrosterone (DHEA). Although these experiments are conducted in vitro, they provide a basis for the possibility of intratumoral DHEA-mediated androgen synthesis mechanisms that may underlie androgen receptor reactivation during androgen deprivation in many prostate tumors. Moreover, Mizokami et al. have shown that dutasteride, previously considered an SRD5A inhibitor, also inhibits the interplay between stromal and epithelial cells in the synthesis of testosterone. Herein, we summarize this study and comment on therapeutic implications.

After the diagnosis of prostate cancer, the prostate is most often resected and androgen deprivation therapy (ADT) is given to manage disease recurrence—typically in the form of bone metastases. ADT can be performed via chemical castration, using agents such as leuprolide and goserelin, or via surgical castration.¹ As the name implies, ADT mediates its effect by reducing serum androgen concentration by approximately 23C23c-90%, thereby inhibiting intratumoral androgen receptor (AR) signaling by depriving the tumor of testosterone (T). However, the intratumoral concentration of androgens is typically −20% of the original value prior to therapy, and the tumor: serum androgen concentration ratio may increase with disease progression during ADT.¹ The median progression free survival of ADT is estimated to be approximately 15–20 mo,¹ and once the therapy fails, the disease deteriorates into castration resistant prostate cancer (CRPC).² Unfortunately, despite therapeutic advances, the vast majority of prostate metastases will eventually become resistant to ADT.²

Prostate tumors typically become castration resistant by effectively utilizing minute concentrations of androgens present during ADT, or by relying on various other signaling pathways in order to survive and proliferate.² The former mechanism is most often mediated by overexpression of the AR, so that far more receptors are present in the cell.³ Similar responses occur when the AR assumes a hypersensitive state where binding of a single androgen can elicit a response many times greater than normal.⁴ This occurs when cells develop the ability to enhance the transcriptional response of AR through the overexpression of transcriptional coactivator proteins,⁵ or by the overexpression of various kinases and pathways that activate AR signaling.² Finally, cells can enhance AR activity via mutations that are permissive to AR signaling in the presence of weak ligands such as dehydroepiandrosterone (DHEA) as well as estradiol and androstenediol.⁶ ADT may actually be responsible for the aforementioned changes because only those cells that develop the ability to survive in low androgen concentration survive and proliferate.²

Though it is known that in most cases ADT only partially reduces the original tumor androgen concentration, despite drastic decreases in serum androgen concentration;¹ the mechanism by which many prostate tumors are able to supplant their androgen source has been poorly explored. Mizokami et al. have clarified how prostate and bone stromal cells are able to convert DHEA into testosterone thereby increasing the local concentration of useful androgens within primary tumors and metastases respectively.⁶ Moreover, this study provides evidence that dutasteride may inhibit the conversion of DHEA into testosterone.

Mizokami et al. utilized the androgen-responsive LNCaP tumor cell line with a promiscuous AR that is activated by DHEA (albeit weakly), testosterone, DHT,² and other ligands. In response to androgen, LNCaPs proliferate and secrete prostate specific antigen (PSA).⁶ In order to measure AR activity, a luciferase expression plasmid was linked to the PSA promoter and transfected into LNCaP cells. The LNCaP transfects were then grown in monoculture, or in co-culture with prostate stromal cells (PrSC) and luciferase activity was measured in the presence and absence of DHEA. The transcriptional activity and proliferation of LNCaP cells were each dependent on the dose of DHEA administered and on the number of stromal cells in coculture. Furthermore, by inhibiting the action of AR via bicalutamide (AR antagonist) and siRNA knockdown, Mizokami et al. demonstrated that the effect of DHEA and stromal cells on LNCaP proliferation was dependent on the AR in LNCaP cells and not the AR in PrSC. The above results suggest that PrSCs alone, or in participation with LNCaP cells, convert DHEA into active androgens resulting in prostate tumor proliferation via the prostate tumor AR.

Prostate tumor, bone, bone marrow and lung stromal cells were also examined, and their effect on LNCaP proliferation was compared to that of PrSCs (i.e., normal prostate stroma). Human marrow and lung stromal cells enhanced the PSA promoter activity of LNCaP cells, although PrSCs were more effective. Bone derived stromal cells (BDSCs) induced an approximate 2-fold induction of PSA promoter activity compared to PrSCs and were similar to prostate cancer stromal cells (PCaSCs; ~3-fold induction over PrSCs). When PCaSCs were evaluated, two PCaSC cell lines were each able to cause nearly a two-fold increase in LNCaP proliferation relative to PrSC. These data suggest that bone-derived stromal cells from healthy bone and PCaSCs are supportive of AR-related signaling and prostate cancer proliferation; thus, prostate tumor cells may metastasize to bone since bone stromal cells are particularly suited to support AR signaling during ADT. Unfortunately, Mizokami et al. did not evaluate stromal cells derived from a bone metastasis, and further experiments will be required to determine if BDSCs from prostate metastases are more or less effective than normal BDSCs.

To elucidate a potential mechanism behind the effect of tumor-derived stromal cells on LNCaP proliferation, Mizokami et al. evaluated the concentrations of T and DHT, as well as their precursors, adione (androstenedione) and adiol (androstenediol) in the culture medium over a period of 24 h. At the end of the experiment, both T and DHT concentrations were approximately ten-fold greater in the PCaSC/LNCaP co-culture than in the LNCaP monoculture, and approximately five-fold greater than in the PrSC/LNCaP coculture. Adione concentrations within the PCaSC/LNCaP co-culture were similar to the concentrations observed in PrSC coculture, but adiol concentrations were −4–5 fold higher in PCaSC/LNCaP coculture than in LNCaP monoculture or in PrSC/LNCaP coculture. Thus, it appears that prostate tumor stroma may develop alterations in the biosynthetic route responsible for making testosterone from DHEA such that adiol formation is favored over adione, and PCaSC/LNCaP co-culture produces significantly higher concentrations of active androgens (i.e., testosterone and DHT) than do LNCaP cells alone, or in co-culture with PrSCs.

As Mizokami et al. determined that PCaSCs appeared to be responsible for increases in local production of T, DHT and adione, they then sought to determine if the expression of genes encoding steroid metabolism enzymes were responsible for the differences in hormone concentrations between co-cultures with PrSCs and PCaSCs. PCaSCs express high levels of type 5 17β-HSD (aldo-keto reductase, AKR1C3) that converts DHEA into adiol and adione into T; however, its expression was not related to DHEA-induced PSA promoter activity in PrSC or PCaSC monoculture. Similarly, when the activity of AKR1C3 was evaluated using naproxen, a specific inhibitor of this enzyme, and AKR1C3-specific shRNA knockdown, there was no difference in PSA-promoter activity in PCaSC/LNCaP coculture. Thus, it appears that although AKR1C3 is expressed differentially in prostate cancer stroma versus normal stroma, it is not itself responsible for increases in the production of active androgens observed in co-culture experiments.

To further examine testosterone biosynthesis, Mizokami et al. used the SRD5A inhibitor dutasteride to determine if dutasteride could inhibit stromal cell AR- and growth activation of LNCaP cells. Dutasteride was effective at inhibiting PSA promoter activity in LNCaP cells and DHT production at a concentration of 5 μM. Most interestingly, dutasteride also inhibited biosynthesis of testosterone from DHEA at —65—80% following a 24 h incubation. Adione concentration was significantly reduced in accordance with the decrease in testosterone concentration, although adiol was not changed. These data suggest that dutasteride inhibits the formation of adione and testosterone via an unknown mechanism in addition to inhibiting DHT synthesis via SRD5A1/2. Mizokami et al. then determined that the presence of two cell types (i.e., stroma and prostate cancer epithelium) acted synergistically to produce more testosterone and DHT than either cell in monoculture. Further experiments determined that, in co-culture, the total adione may be primarily due to 3β-HSD activity in LNCaP cells and that adiol synthesis is enhanced over monoculture experiments. Dutasteride inhibited adione synthesis in LNCaP monoculture, but it did not inhibit adione synthesis in PCaSC monoculture. Therefore, it appears that the reduction in testosterone concentration observed in co-culture is due to the inhibition of 3β-HSD activity in LNCaP cells that disrupts further synergistic androgen synthesis when both PCaSC and LNCaP cells are grown together. These results suggest that, at least for some prostate tumors, stromal cells require androgen precursor synthesis from prostate cancer cells that are then used in the formation of testosterone that drives tumor progression through an AR mechanism. They also suggest that dutasteride may inhibit the interplay between both cell types by inhibiting 3β-HSD activity.

As was previously mentioned, there have been many theories to explain the resurgence of the cancer in the form of CRPC following ADT. The results of Mizokami et al. provide some of the first evidence for another mechanism where stromal cells (both in the primary tumor and in bone) drive the formation of androgens that result in increases in local production of androgen via DHEA. While ADT targets the formation of androgens by the testes, it does not inhibit the formation of adrenal DHEA, which is known to drive tumor progression although the precise mechanism has not been well established.¹ ADT itself only remains efficacious for a period of roughly 15–20 mo, at which time the cancer returns in an androgen resistant form (i.e., CRPC). Mizokami et al. suggest that >50% of prostate cancers relapse following ADT are due to intratumoral production of active androgens through DHEA; this assertion is justified by the success of abiraterone, a CYP17 inhibitor that blocks DHEA formation, in CRPC.^7,8 Moreover, the use of ketoconazole, an inhibitor of adrenal DHEA synthesis, has been effective in treating CRPC when applied certain settings.

Several stuaies have shown that dutasteride improves phenotypic markers of disease progression in men when administered as a neoadjuvant prior to radical prostatectomy.^9–12 Prostate cancer risk may also be decreased in healthy men treated with SRD5A inhibitors.¹³ However, since patients treated with dutasteride in these contexts do not receive ADT, and thus have a ready supply of T from the testes, it is likely that dutasteride mediates these responses by reducing intraprostatic DHT synthesis (via SRD5A inhibition) and not by inhibition of testosterone in the stroma. Others have treated men with prostate cancer with a combination ADT and dutasteride with a low response rate.¹⁴ Therefore, while work by Mizokami et al. provides some of the first in vitro evidence that dutasteride may inhibit other enzymes involved in testosterone synthesis, it is unlikely that dutasteride treatment would be an effective adjuvant during ADT.

While the data presented by Mizokami et al. is not necessarily novel, as previous investigators had noticed that certain stromal cells enhance androgen synthesis and PSA formation,¹⁵ the data presented in this article shed significant light on how the interplay between the two cell types is responsible for the formation of intratumoral androgens. It should be noted that there is a clear distinction between prostate stromal cells (PrSCI) versus bone derived stromal cells, (BDSCs) and prostate cancer stromal cells (PCaSC)—PcaSCs and BDSCs elicit a far greater response from epithelial tumor cells (LNCaP), and this appears to be due to a greater release of steroidogenic substrates by stromal and epithelial cells (such as adiol, adione, etc.).⁶ It is also very interesting that these androgens are formed from a readily available precursor during ADT (DHEA), and that dutasteride may inhibit synergistic androgen formation between the two cell types. Therefore, Mizokami et al. may have begun to characterize a mechanism that may contribute to androgen receptor reactivation during ADT wherein stromal cells are responsible for producing active androgens from low activity precursors (i.e., DHEA).² This mechanism would be expected to apply to primary tumors and metastases alike and may be inhibited by a known therapeutic, dutasteride.