Increased transcription and high translation efficiency lead to accumulation of androgen receptor splice variant after androgen deprivation therapy

Abstract

Upregulation of androgen receptor splice variants (AR-Vs), especially AR-V7, is associated with castration resistance of prostate cancer. At the RNA level, AR-V7 upregulation is generally coupled with increased full-length AR (AR-FL); consequently, AR-V7 and AR-Vs collectively constitute a minority of the AR population. However, Western blotting showed that the relative abundance of AR-V proteins is much higher in many castration-resistant prostate cancers (CRPCs). To address the mechanism underlying this discrepancy, we analyzed RNA-seq data from ~350 CRPC samples and found a positive correlation between all canonical and alternative AR splicing. This indicates that increased alternative splicing is not at the expense of canonical splicing. Instead, androgen deprivation releases AR-FL from repressing the transcription of the AR gene to induce coordinated increase of AR-FL and AR-V mRNAs. At the protein level, however, androgen deprivation induces AR-FL, but not AR-V, degradation. Moreover, AR-V7 is translated much faster than AR-FL. Thus, androgen-deprivation-induced AR-gene transcription and AR-FL protein decay, together with efficient AR-V7 translation, explain the discrepancy between the relative AR-V mRNA and protein abundances in many CRPCs, highlighting the inevitability of AR-V induction after endocrine therapy.

Keywords: AR negative autoregulation; AR-V mRNA stability; AR-V protein stability; AR-V translation; Castration resistance; Castration-resistant prostate cancer.

Figure 1.. Canonical and alternative AR splicing are tightly correlated in CRPC samples.

A. Schematic diagram of AR gene structure with canonical exons, cryptic exons (CE), and the relative locations of alternative splicing donors and acceptors. Drawings are not to scale. B. Spearman correlation analysis showing tight positive correlations between canonical (black) and alternative (blue) AR splicing in 51 CRPC samples in the SU2C polyA-selected RNA-seq dataset. The red-blue gradient bar indicates the scale of correlation coefficient scores. C. RNA-seq data from LNCaP95 and 22Rv1 cells and hormone sensitive (HS) and castration-resistant (CR) LuCaP 23.1, 35, and 96 PDXs showing no statistically significant impact of androgen deprivation (AD) or hormone sensitivity on AR-V/AR-FL mRNA ratios. DHT: 10 nM.

Figure 2.. AR-FL is required for increase in AR-V levels after depriving AR-FL-expressing cells of androgen.

RT-qPCR (A & C) and Western blot analyses (B & D) showing that increase in AR-V7 mRNA and protein levels after androgen deprivation was attenuated by AR-FL protein depletion (A & B) or AR-FL knockdown (C & D). A & B, LNCaP 95 cells cultured in androgen-deprived condition were pre-treated with or without 50 nM A1175, an AR-FL PROTAC, for 6 h prior to exposure to vehicle or 1 nM R1881 for 16 (A) or 48 h (B). Data are expressed as relative to the levels in cells treated with R1881 alone. C & D, LNCaP 95 cells infected in triplicate with lentiviruses containing control (shCtrl) or AR-FL shRNA (shAR-FL) were treated with vehicle or 1 nM R1881 for 24 (C) or 48 h (D). Data are expressed as relative to the levels in shCtrl-lentivirus-infected cells treated with R1881. Graphs in Panels B & D: Western blot quantitation. Long expo: long exposure. V7 Ab: AR-V7 antibody.

Figure 3.. AR-V7 does not regulate the transcription of the AR gene.

RT-qPCR (A) and Western blot analyses (B) showing that cumate-inducible expression of AR-V7 in LNCaP cells, at a low level (3 days of cumate treatment), did not affect the levels of AR pre-mRNA, AR-FL mRNA, or AR-FL protein. Data are expressed as relative to the levels in cumate-inducible AR-V7-expressing cells not treated with cumate. Graphs in Panel B: quantitation of AR-FL and -V7 protein levels in AR-V7-expressing cells from the Western blots. Data from the mock control cells are shown in Supplementary Fig. S5. Cells were cultured in androgen-deprived condition. Long expo: long exposure.

Figure 4.. ARBS2 is critically involved in increase in AR-FL and AR-V mRNA levels after androgen deprivation.

A & C. Top: strategy for deleting (A) or mutating (C) ARBS2 by CRISPR/Cas9. Arrows indicate forward (F) and reverse (R) PCR primers for detecting ARBS2 deletion or mutation. Bottom: RT-qPCR analyses showing that ARBS2 deletion (A) and mutation (C) attenuated androgen deprivation (AD)-associated increase in AR-FL and AR-V7 mRNA levels in LNCaP95 cells. R1881: 1 nM. Data are expressed as relative to the levels in mock control cells treated with R1881. B. AR ChIP-seq read coverage in ARBS2 region from LNCaP95 cells is displayed on IGV. The ARBS2, sgRNA10, sgRNA2, AR-binding peak, and androgen-response element (ARE) loci are indicated below the coverage. DHT: 10 nM.

Figure 5.. Androgen deprivation does not affect AR-FL or AR-V mRNA stability.

RT-qPCR was performed on LNCaP95 (A), VCaP (B), and 22Rv1 (C) cells treated with 10 μM actinomycin D for the indicated time. Data are expressed as relative to the levels in the respective 0-time-point group and presented as semi-log regression line plots. R1881: 1 nM. DHT: 10 nM. P > 0.05 between the AD and the R1881 groups for all transcripts, between the AR-FL and AR-V7 transcripts, between the AR-FL and AR-V9 transcripts, and between the AR-V7 and AR-V9 transcripts in all cell models. P < 0.05 between the AR-V7 and AR-V1 transcripts, the AR-V7 and AR-V6 transcripts, the AR-FL and AR-V1 transcripts, and the AR-FL and AR-V6 transcripts in androgen-deprived condition.

Figure 6.. Androgen deprivation (AD) reduces AR-FL protein stability without affecting AR-V protein stability.

Because of differences in sensitivity to cycloheximide (CHX) cytotoxicity, LNCaP95 (A), R1-AD1, and R1-D567 cells (C) were treated with 10 μg/ml cycloheximide. M12 cells stably infected with lentiviruses containing AR-FL or -V7 (B) were treated with 100 μg/ml cycloheximide. R1881: 1 nM. Graphs: semi-log regression line plots of Western blot signals relative to the signal in the respective 0-time-point group. There is no statistically significant difference between the AD and the R1881 groups for AR-V decay or between AR-FL and AR-V7 or AR^v567es in androgen-deprived condition. Long expo: long exposure.

Figure 7.. AR-V7 is translated at a much faster rate than AR-FL.

A, ³⁵S-methionine pulse-chase analysis of AR-FL and AR-V7 protein decay and translation rates. LNCaP95 cells cultured in androgen-deprived (AD) condition or with 1 nM R1881 were pulse labeled with ³⁵S-methionine for 30 min and chased for the indicated time. AR proteins were then immunoprecipitated (IP) and subjected to phosphorimaging and Western blot analysis of radio-labeled and steady-state AR proteins in the immunoprecipitates, respectively. Quantitation of protein decay rates confirming androgen deprivation facilitating AR-FL degradation without affecting AR-V7 protein stability. Quantitation of translation rates showing more efficient translation of AR-V7 than AR-FL in both androgen-deprived and - replenished conditions. Translation rates are calculated as the pulse-labeled amount of a protein normalized to its mRNA level and the number of methionine in that protein and are presented as relative to the respective AR-FL rate. Short/long expo: short/long exposure. B, Non-radioactive nascent protein synthesis analysis confirming more efficient translation of AR-V7 than AR-FL. LNCaP95 cells were pulse labeled with L-AHA for 2 h. Labeled proteins were then conjugated with biotin using a click reaction, purified using streptavidin beads, and subjected to Western blotting with a pan-AR antibody. The amount of AR-FL or AR-V7 protein, after subtracting for non-specific binding to streptavidin beads, was normalized to its mRNA level to derive the translation rates, which are presented as relative to the respective AR-FL rate.

Figure 8.. AR-FL and AR-V7 mRNAs are distributed similarly in individual polysome fractions.

LNCaP95 (A) and VCaP (B) cells cultured in androgen-deprived condition or with 1 nM R1881 or 10 nM DHT were subjected to polysome gradient profiling. Top panels: representative absorbance profiles at 254 nm to locate the monosome and polysome regions. Middle panels: agarose gel analysis of RNA isolated from each fraction to show the integrity of the RNA. Bottom panels: distributions of SNHG5 (cytoplasmic non-coding RNA control), AR-FL, and AR-V7 mRNAs in each fraction as measured by RT-qPCR. There is no statistically significant difference between the AR-FL and the AR-V7 groups.