Androgen Receptor Splice Variants Are Not Substrates of Nonsense-Mediated Decay

Abstract

Background: Androgen receptor (AR) splice variants have been clinically associated with progressive cancer, castration-resistance, and resistance to AR antagonists and androgen synthesis inhibitors. AR variants can be generated by genomic alterations and alternative splicing, and their expression is androgen-regulated. There has been a suggestion that AR variants bearing premature termination codons and coding for truncated proteins should be regulated by the nonsense-mediated decay (NMD) mRNA surveillance pathway, suggesting that either the NMD pathway is dysfunctional in variant-expressing cell lines or that variants are somehow able to evade degradation by NMD.

Methods: We first used siRNA knockdown of the NMD regulator, UPF1, in an NMD reporter assay to determine if this surveillance pathway is functioning normally in AR variant-expressing cell lines. We then used UPF1 knockdown to determine if expression of the AR variants ARV3 and ARV7 is affected by inhibition of NMD. Next, we analyzed androgen regulation of UPF1 and used transcript expression analysis to determine if there is any association between UPF1 expression, resistance, and ARV3 or ARV7 expression.

Results: We found that the NMD pathway functions normally in the AR variant-expressing cell line 22Rv1 and that inhibition of NMD does not increase expression of ARV3 or ARV7. Furthermore, we found that expression of UPF1 is not androgen-regulated. We also found that UFP1 expression levels do not differentiate castration-sensitive from resistant cell line and that UPF1 expression does not correlate with expression of ARV3 or ARV7 in cells in which these variants are highly expressed.

Conclusion: This study eliminates a possible mechanism of regulation of certain AR variants. Future research into the regulation of AR variants should focus on other mechanisms to better understand the origin of these variants and to possibly inhibit their expression for the resensitization of resistant cancers. Prostate 77:829-837, 2017. © 2017 Wiley Periodicals, Inc.

Keywords: UPF1; androgen receptor; nonsense-mediated decay; prostate cancer; splice variants.

Figure 1. Nonsense-mediated decay assay in 22Rv1 cells

A) ZsG green fluorescent protein is inserted in-frame into a TCRβ exon such that when the minigene is transcribed and spliced, the ZsG stop codon is located upstream of an EJC, is recognized as a PTC, and is subject to degradation by NMD. (Cartoon adapted from Paillusson et al. [19].) B) siRNA inhibition of UPF1 in 22Rv1 cells causes a significant (p = 0.03) increase in reporter expression. Shown are the mean +/- SEM of three replicate experiments.

Figure 2. Nonsense-mediated decay regulation of ARV3 and ARV7

A) Structures of AR, ARV3, and ARV7. Stop codons are indicated by black triangles. B) Sequence structure of ARV3. Cryptic exon 4 is highlighted in gray. The sequences before and after cryptic exon 4 are of exons 2 and 3, respectively. The ARV3 stop codon is indicated in bold text. qPCR primers, spanning the exon 2-cryptic exon 4 junction and within cryptic exon 3, are underlined. There are 124 nucleotides between the terminal nucleotide of the stop codon and the putative EJC at the junction of cryptic exon 4 and exon 3. Because the stop codon occurs >50 nucleotides upstream of an EJC, ARV3 would be expected to be a substrate of NMD. C) siRNA knockdown of UPF1 did not alter expression of ARV3 or ARV7 in 22Rv1 (p = 0.1 and 1, respectively) or VCaP (p = 0.6 and 1.0, respectively) but did decrease expression of ARV3 and ARV7 in LN95 (p = 0.005 and 0.02, respectively). Shown are the mean +/- SEM of three replicate experiments.

Figure 3. Androgen regulation of AR, ARV7, and UPF1 expression

AR and ARV7 expression was significantly decreased in all cell lines by R1881 (AR: p = 0.02, 0.01, and 0.004; ARV7: p = 0.001, 0.0002, and 0.03 for 22Rv1, LN95, and VCaP, respectively). UPF1 expression in LN95 and VCaP was unaffected by androgen (p = 0.2 and 0.3, respectively), but expression of UPF1 was significantly decreased by androgen in 22Rv1 (p = 0.0003) cells. Shown are the mean +/- SEM of three replicate experiments.

Figure 4. Expression of UPF1 in prostate cancer cell lines

AR expression is statistically increased in resistant cell line LAPC4-cr relative to the sensitive parent cell line, LAPC4 (p = 0.01). There was no statistical difference in AR expression between CWR22 and LNCaP and its resistant derivatives, 22Rv1 and LN95, respectively (p = 0.08 and 0.4, respectively). Both ARV3 and ARV7 expression are increased between CWR22 and 22Rv1 (p < 0.0001 and p = 0.0004, respectively) and between LNCaP and LN95 (p = 0.002 and 0.0003, respectively). UPF1 expression is unchanged between sensitive and resistance cell lines (p = 0.3 CWR22/22Rv1, p = 0.9 LNCaP/LN95, and p = 0.5 LAPC4/LAPC4-cr).

Figure 5. Linear Regression Analysis of ARV3 and ARV7 v. UPF1

Correlation between expression levels of ARV3 or ARV7 vs. UPF1 was determined by linear regression analysis. Solid lines indicate linear regression, and dashed lines indicate confidence interval bands. Significant deviations from slopes of zero (indicated by p-values < 0.0.5) were found in CWR22 (ARV7 v. UPF1, p = 0.04), LAPC4 (ARV3 v. UPF1, p = 0.03), and LAPC4-cr (ARV3 v. UPF1, p =0.03 and ARV7 v. UPF1, p = 0.006).

Figure 6. Translation reinitiation of ARV3

Putative nonsense transcripts and NMD substrates can evade NMD by a variety of mechanisms, one of which is reinitiation of translation downstream of the PTC. ARV3 harbors an in-frame translation start codon (outlined in black) downstream of the PTC (in bold text) located in cryptic exon 4, which could explain why inhibition of NMD does not effect expression.