The long non-coding RNA PCGEM1 is regulated by androgen receptor activity in vivo

We performed surgical castration (androgen ablation) of mice bearing the LTL-331 PCa xenograft. Tissue samples were collected just prior to castration and 3 weeks after castration, and were analyzed for gene expression. We observed a significant >500-fold decrease in PCGEM1 expression post-castration relative to the pre-castration level (Figure 1A). Notably, this down-regulation was persistent (>180-fold) even twelve weeks following castration (Additional file 2: Figure S1). In agreement with this data, qPCR analysis on RNA collected before and 12 weeks after castration from a second in vivo model, LTL-313B, confirmed the down-regulation of PCGEM1 (>70-fold; Figure 1B). In both models, reduced AR activity in response to castration was confirmed by the comparable down-regulation of PCAT18 (Figure 1A,B), an AR-regulated lncRNA [13], and reduction in serum prostate-specific antigen (PSA) levels (at least >45-fold; Figure 1C). These findings led us to investigate the response of PCGEM1 to AR stimulation in vivo. This was achieved by using intact mouse hosts supplemented with or without pure testosterone at two distinct dosages (1.0 mg and 5.0 mg/mouse) for PCa xenotransplantation of the LTL-331 tumor line. Corresponding to the dosages, serum testosterone levels of 3.96 ± 0.679 ng/ml and 20.6 ± 14.1 ng/ml were achieved in the testosterone supplemented animals versus 1.95 ± 0.106 ng/ml in the intact hosts. Notably, the hosts supplemented with pure testosterone showed a dose-dependent up-regulation of PCGEM1, from 5-fold to greater than 22-fold, relative to the hosts without the augmented testosterone (Figure 1D). This coincided with a comparable PCAT18 up-regulation and serum PSA jump of >2-fold at both dosages.

Notably, previous in vitro studies investigating AR-regulation of PCGEM1 produced conflicting results. Srikantan et al. [9] concluded PCGEM1 is AR-regulated, while Prensner et al. [17] found no up-regulation of PCGEM1 upon AR stimulation. The reason for this discrepancy could lie in the varied experimental conditions employed; in particular, the use of different agonists (viz. dihydrotestosterone (DHT) and synthetic R1881) for AR activation. Although the patterns of DHT-mediated and R1881-mediated gene expression changes are largely parallel, they do not completely overlap [24]. Consequently, we repeated this experiment using DHT, a physiological AR ligand, coupled with the sensitive TaqMan qPCR for gene quantification. Two AR+ PCa cell line models that expressed the highest levels of PCGEM1, LNCaP and VCaP, were chosen for the in vitro studies (Additional file 2: Figure S2). DHT treatment (at 6 h and 12 h) induced a modest (1.8 – 2.2-fold) up-regulation of PCGEM1 in LNCaP cells (Figure 2A). However, the activation of PCGEM1 did not continue to escalate in a manner similar to that of canonical AR-regulated genes such as PSA and PCAT18 over extended time-points (12 h and 24 h). Similar observations were made in VCaP cells upon treatment with DHT (10nM; Figure 2B). Most remarkably, the weak AR-activation of PCGEM1 in vitro is in accordance with the recent observation that genes actively regulated by the AR in patient tumor tissue show a strong in vivo response to castration but a distinctly lesser response to androgen stimulation in vitro [22]. Notably, also contradicting the original findings [9], no significant increase in PCGEM1 expression was observed in vitro (at 6 h, 12 h or 24 h) following stimulation of LNCaP cells with R1881 – a synthetic AR ligand (at 10nM; Figure 2C). Similar results were obtained on treatment with a super-physiological concentration of DHT (100nM; Figure 2D). Taken together, our results categorically demonstrate that substantial regulation of PCGEM1 expression by androgen occurs exclusively in vivo.

It was previously suggested that a direct AR-PCGEM1 interaction regulates AR’s transcriptional activity [16]. In line with this mechanistic model, PCGEM1’s suspected function as a transcriptional co-regulator would imply it to be predominantly contained in the nucleus, in particular when the AR is transcriptionally active. However, sub-cellular localization of PCGEM1 had never been investigated. Addressing this issue, we performed cellular fractionation of PCa patient-derived xenograft cells to isolate the nuclear and cytoplasmic RNA fractions, and quantified gene expression.

Our data shows that PCGEM1 is evenly distributed between the nucleus and the cytoplasm of LTL-331 cells harvested from intact mouse hosts or intact hosts supplemented with testosterone (Figure 3A,B). Expected cytoplasmic localization of GAPDH and Actin (>85% for both) and nuclear localization of small nucleolar RNA 55 (snoRNA55) and MALAT1 (>80% for both) authenticate our findings. A similar sub-cellular distribution of PCGEM1 was observed in DHT treated and untreated LNCaP cells as well (Additional file 2: Figure S3A, B). While these results do not conclusively rule out the possibility that PCGEM1 may be interacting with the AR, they demonstrate for the first time that PCGEM1 is not selectively localized in the nucleus; or is shuttled to it upon transcriptional activation of the AR. These findings also highlight the gap in our understanding of lncRNAs with proposed cell compartment-specific functions. The unresolved question is whether lncRNAs with nuclear roles are contained in the nucleus throughout their lifetime, or do they also shuttle to the cytoplasm where they can adopt additional roles? For instance, H19, one of the most well characterized lncRNAs, is associated with both nuclear and cytoplasmic functions i.e. chromatin modulation [25] and micro-RNA generation [26,27], respectively. Collectively, our results warrant further verification of the alleged function of PCGEM1 as a transcriptional co-regulator of the AR, in addition to investigating its potential role in the cytoplasm. For detailed information about all the experiments refer to “Additional file 3: Supplementary methods and materials.”

Our results to this point indicate PCGEM1 to be an in vivo androgen regulated gene, with uniform distribution in PCa cell nucleus and cytoplasm. One remaining question, however, is the stage of PCa in which PCGEM1 might be biologically most relevant. To this end we investigated a patient database consisting of microarray and clinical data on 131 primary PCa, 19 metastasized PCa and 29 normal prostatic tissue samples [28]. Here we observed that PCGEM1 was significantly up-regulated only in primary PCa but not in metastasized PCa relative to normal prostate tissue. In fact, PCGEM1 expression was significantly down-regulated in metastasized tumors relative to primary tumors (Additional file 2: Figure S4). Adding to a recent study discrediting the involvement of PCGEM1 in progression of localized PCa to CRPC [17], our analysis further raises the question of the biological relevance of PCGEM1 in survival and proliferation of metastasized PCa cells themselves. In view of this, PCGEM1 is most likely implicated only in the early stages of the disease warranting further experimental validation. To explore this further, we performed Pearson’s Correlation analysis to identify genes that are positively associated with PCGEM1 expression, hereafter referred to as PCGEM1-associated gene expression signature (PES) (Additional file 1: Table S2), and uploaded the list into the Oncomine database. As previously demonstrated for PCGEM1, the PES was also significantly up-regulated in PCa relative to other neoplasms and in PCa versus normal prostate tissue (Table 1). Corroborating the relevance of PCGEM1 in early stages of PCa, PES was consistently repressed in metastatic and high Gleason PCa relative to primary and low Gleason neoplasms, respectively (Table 1). Notably, this trend was observed in several independent patient cohorts totaling more than 1100 samples. Accordingly, the PES was significantly repressed in patients exhibiting poor clinical outcomes.

A review of literature also reveals that PCGEM1 is up-regulated in normal prostate epithelial cells of men with a family history of PCa [15] and certain single-nucleotide polymorphisms in the PCGEM1 gene are associated with a higher risk of PCa development in Chinese men [29]. Together, these findings support the testable hypothesis that PCGEM1 is more important in the early stages, possibly initiation, of primary PCa. Interestingly, the PCGEM1 signature was also under-expressed in hormone-refractory PCa relative to hormone-naïve PCa (Table 1); and the pathway analysis revealed PES to be most significantly down-regulated (p = 5.47E-14, odds ratio = 121.4) in patients receiving androgen-deprivation therapy (Additional file 1: Table S3) lending further support to in vivo AR-regulation of PCGEM1.

Our work brings in light an even more pertinent concern with exploring AR transcriptional activity using cell line models. Recently, cell line models in accordance with their metastatic origin were demonstrated to exhibit the AR regulatory program that is active in metastatic advanced PCa as opposed to primary PCa [22]. The AR regulatory activity in cultured cell line models (including LNCaP) had a greater overlap to CRPC (31% overlap) than untreated primary PCa (merely 3% overlap) (see [22]). Besides the absent cell microenvironment, this describes further limitations of cell line models for investigating AR-regulation, in particular for genes with a clinical expression profile akin to that of PCGEM1. Simultaneously, it underscores the importance of using clinically relevant models that best represent the original malignancy. With mounting evidence for lncRNAs implicated in carcinogenesis and cancer progression, we regard lncRNAs as essential in decrypting cancer cell biology. Our study corroborates the irrelevance of PRNCR1 in PCa, and confirms PCGEM1 as an in vivo AR-regulated transcript as well as rationalizes its oncogenic involvement in early stages of the disease.