Divergent androgen regulation of unfolded protein response pathways drives prostate cancer

Abstract

The unfolded protein response (UPR) is a homeostatic mechanism to maintain endoplasmic reticulum (ER) function. The UPR is activated by various physiological conditions as well as in disease states, such as cancer. As androgens regulate secretion and development of the normal prostate and drive prostate cancer (PCa) growth, they may affect UPR pathways. Here, we show that the canonical UPR pathways are directly and divergently regulated by androgens in PCa cells, through the androgen receptor (AR), which is critical for PCa survival. AR bound to gene regulatory sites and activated the IRE1α branch, but simultaneously inhibited PERK signaling. Inhibition of the IRE1α arm profoundly reduced PCa cell growth in vitro as well as tumor formation in preclinical models of PCa in vivo. Consistently, AR and UPR gene expression were correlated in human PCa, and spliced XBP-1 expression was significantly upregulated in cancer compared with normal prostate. These data establish a genetic switch orchestrated by AR that divergently regulates the UPR pathways and suggest that targeting IRE1α signaling may have therapeutic utility in PCa.

Keywords: ER stress; UPR; androgen receptor; androgens; prostate cancer.

Figure 1

Correlation of AR and UPR gene expression in prostate cancer cohorts

1. Possible correlation between AR- and UPR-associated gene expression was assessed in the global gene expression data available in the TCGA Prostate Adenocarcinoma cohort (n = 190) (http://www.cbioportal.org/public-portal/index.do). Tumors were stratified according to AR status into three groups, that is AR_low (n = 60), AR_medium (n = 70), and AR_high (n = 60). The levels of UPR gene expression in the three groups were compared using Pearson's correlation analysis by the R software and presented as a heatmap. There were significant differences between the three groups (Supplementary Table S2).

2. The expression profiles of some prominent UPR genes from the data in (A), including ERN1 (IRE1), EDEM1, ATF6, and DNAJC3 (P58IPK), are presented. P-values of the different genes are given.

Figure 2

Androgens divergently regulate the UPR arms LNCaP cells were cultured and treated with R1881 for the indicated times.

1. A, B mRNA expression levels in LNCaP cells for the indicated genes were investigated using quantitative PCR (qPCR). Controls were treated with vehicle for 84 h and set to 100. Data represent the mean of three independent experiments in triplicate, and bars represent SE. P-values ranged between 1.66 × 10⁻⁵ and 0.025, and IRE1α expression in R1881 48 h was *P = 0.013 with respect to vehicle-treated cells using unpaired Student's t-test.

2. C–E IRE1α, XBP-1S, and XBP-1U mRNA in CWR22 xenografts grown in nude mice and collected at the indicated times after castration. The value at t = 0 was set to 1. Columns represent the mean of at least three independent tumors for each time point, and bars represent SE. P-values ranged between 5.7  ×  10⁻¹⁰ and 0.0002. IRE1α expression at 48 h post-castration was 3.17 × 10⁻⁶ with respect to t = 0 using unpaired Student's t-test.

3. F, G Protein expression in LNCaP cells upon treatment with R1881 for the indicated times by Western blot analysis. Data presented are representative of three independent experiments.

Source data are available online for this figure.

Figure 3

AR knockdown differentially influences transcription of the different UPR members After starvation, LNCaP cells were transfected with control (CTRL) siRNA or AR siRNA. Cells were then treated with R1881 for the indicated times. Controls were treated with vehicle for 48 h.

1. A Expression level of AR mRNA upon siRNA treatment for 48 h assessed by qPCR in LNCaP cells. Expression in cells transfected with CTRL siRNA was set to 1. Bars represent SE with *P = 0.001 indicating significant difference between AR siRNA- and control siRNA-transfected cells using paired Student's t-test.

2. B–E Same as in (A), but mRNA expression levels of the indicated UPR genes were determined by qPCR at indicated time points after R1881 stimulation. Expression in cells transfected with CTRL siRNA was set to 100. Bars represent SE. P-values are shown indicating significant difference between AR siRNA- and control siRNA-transfected cells using unpaired Student's t-test.

3. F Expression of the indicated proteins under conditions indicated on the top label was determined by Western blot analysis. Representative blots for three independent experiments are shown.

Source data are available online for this figure.

Figure 4

AR directly binds in the vicinity of different UPR genes

1. A, B LNCaP (A) or VCaP cells (B) were cultured and treated with vehicle (C) for 48 h or R1881 for 24 and 48 h. The cells were then fixed, and ChIP assay was performed as described in Materials and Methods using AR antibody. The data shown are representative of one experiment in duplicate. Error bars represent SE. *P < 0.01 for LNCaP and *P < 0.04 for VCaP indicate significant difference between C (control) and R1881 using unpaired Student's t-test.

Figure 5

IRE1α and XBP-1 are proliferative factors in PCa cells both in vitro and in vivo

1. A Knockdown of IRE1α or XBP-1 leads to a decrease in cell survival. LNCaP cells were transfected with siRNA targeting either IRE1α or XBP-1 (5 nM) and starved in 2% CT-FCS medium for 3 days before cell viability was measured using the CCK-8 assay. The graph is representative of one experiment in triplicate and was repeated three times with similar results. Error bars represent SD with *P = 6.6 × 10⁻⁵ and 4.5 × 10⁻⁵ for comparison between Ctrl and siRNA against IRE1α and XBP-1, respectively, using paired Student's t-test.

2. B XBP-1 rescues the growth defect of siIRE1α-transfected LNCaP cells. LNCaP cells were transfected with 5 nM of indicated siRNA using Lipofectamine RNAiMax reagent. One day after siRNA transfection, the cells were transfected with either vector control (Empty) or Flag-XBP-1S (XBP-1S). Three days after transfection, cells were harvested for Western analysis or cultured for three more days before being applied to cell proliferation assay using the CCK-8 reagent. The data are representative of two experiments in triplicate. Error bars represent SE. *P = 0.02, **P = 8.54 × 10⁻⁷ using paired Student's t-test.

3. C, D IRE1α and XBP-1 knockdown inhibits clonogenic capacity of LNCaP cells. Control LN-Scr (Scr), LN-shIRE1 (shIRE1), or LN-shXBP1 (shXBP-1) cells were cultured for 3 weeks. The colonies formed were stained with crystal violet and photographed. The extent of IRE1α and XBP-1 knockdown was determined by Western blot analysis. The area covered by colonies was quantified using the Gene Tools software (SynGene). The data are representative of three experiments in triplicate. Error bars represent SEM. *P = 4.38 × 10⁻⁷ and **P = 3.39 × 10⁻²⁰ using paired Student's t-test.

4. E Growth analysis of xenografted LNCaP tumors in nude mice. LNCaP cells expressing shRNA against IRE1α (LN-shIRE1), XBP1 (LN-shXBP-1), or control shRNA (LN-Scr) were subcutaneously implanted into both flanks of male nude mice (6 mice per group). Tumor size was measured at the indicated time points. Representative pictures of the tumors at harvest are shown. Error bars indicate SEM. *P = 0.03 for shIRE1 at week 7, P = 0.02 for shXBP-1 at week 7, **P = 0.01 for both shIRE1 and shXBP-1 at week 8 using unpaired Student's t-test.

5. F PCNA immunostaining in tumors from animals bearing LN-shIRE1, LN-shXBP-1, or LN-Scr tumors. Scale bars: 100 μm.

Figure 6

A small molecule IRE1α inhibitor interferes with prostate cancer cell growth in vivo LNCaP xenografts were grown in nude mice until palpable. Mice were then intraperitoneally injected with 0.5 mg/kg toyocamycin or saline (Vehicle) (tumor numbers: n = 15, or n = 10, respectively) twice weekly.

1. Tumor sizes were measured weekly with calipers. Error bars indicate SEM. *P = 0.04; **P = 0.0004; ***P = 0.002 using unpaired Student's t-test.

2. Same procedure was repeated on VCaP xenografts, with similar findings as in LNCaP xenografts (tumor numbers: n = 9, n = 11). Error bars indicate SEM. *P = 0.02 and **P = 0.0006 using unpaired Student's t-test.

3. XBP-1S mRNA levels in LNCaP xenografts from animals treated with toyocamycin or vehicle. Tumors were harvested, RNA-extracted, and qPCR performed. Error bars indicate SEM. *P = 0.04 for LNCaP and **P = 0.02 for VCaP using unpaired Student's t-test.

4. PCNA staining in tumors from animals treated with either toyocamycin or saline. Scale bar: 100 μm.

Figure 7

XBP-1S expression in clinical prostate cancer specimens XBP-1S expression was determined by IHC in two different cohorts of human prostatectomy samples.

1. Representative pictures of benign and tumor samples.

2. Tissue microarrays (Wang et al, 2010) containing 25 benign and 283 tumor samples were stained with a XBP-1S specific antiserum and scored by a pathologist. The P-value indicates the difference between XBP-1S staining (strong and moderate) in normal vs cancer cells using Mann–Whitney test.

3. XBP-1S expression was determined by IHC of a neoadjuvant hormone therapy (NHT) tissue microarray containing samples from hormone naïve (untreated) (n = 25), patients that received NHT for 1–6 months (n = 33), and patients that received NHT for 7–12 months (n = 50), as indicated. Representative images are shown. Scale bars: 100 μm.

4. Quantitative presentation of the data from (C). *P = 0.006 in unpaired Student's t-test.

5. A model for AR regulation of UPR in PCa cells: Liganded AR activates the IRE1α pathway and coordinately inhibits the PERK arm of the UPR. In addition, AR inhibits the proapoptotic JNK pathway that may be activated by IRE1α or other pathways. The end result of these AR effects is PCa cell proliferation and survival. An arrow with a solid line indicates direct promotion. An arrow with dashed line indicates indirect/unexplored interactions.