Differential regulation of metabolic pathways by androgen receptor (AR) and its constitutively active splice variant, AR-V7, in prostate cancer cells

Abstract

Metastatic prostate cancer (PCa) is primarily an androgen-dependent disease, which is treated with androgen deprivation therapy (ADT). Tumors usually develop resistance (castration-resistant PCa [CRPC]), but remain androgen receptor (AR) dependent. Numerous mechanisms for AR-dependent resistance have been identified including expression of constitutively active AR splice variants lacking the hormone-binding domain. Recent clinical studies show that expression of the best-characterized AR variant, AR-V7, correlates with resistance to ADT and poor outcome. Whether AR-V7 is simply a constitutively active substitute for AR or has novel gene targets that cause unique downstream changes is unresolved. Several studies have shown that AR activation alters cell metabolism. Using LNCaP cells with inducible expression of AR-V7 as a model system, we found that AR-V7 stimulated growth, migration, and glycolysis measured by ECAR (extracellular acidification rate) similar to AR. However, further analyses using metabolomics and metabolic flux assays revealed several differences. Whereas AR increased citrate levels, AR-V7 reduced citrate mirroring metabolic shifts observed in CRPC patients. Flux analyses indicate that the low citrate is a result of enhanced utilization rather than a failure to synthesize citrate. Moreover, flux assays suggested that compared to AR, AR-V7 exhibits increased dependence on glutaminolysis and reductive carboxylation to produce some of the TCA (tricarboxylic acid cycle) metabolites. These findings suggest that these unique actions represent potential therapeutic targets.

Keywords: LNCaP; androgen receptor; metabolism; prostate cancer; splice variant.

Figure 1. Characterization of AR isoforms

A. Schematic of full-length androgen receptor (AR) composed of distinct functional domains: amino-terminal transactivation domain (encoded by exon 1), DNA-binding domain (DBD encoded by exon 2 and 3), a hinge region (H encoded by exon 4), and a ligand-binding domain (LBD encoded by exons 4-8). Note that the domains are not proportional to actual size in this diagram. The naturally-occurring V7 splice variant is truncated at exon 3 (amino acids 1-627) followed by 16 unique amino acids. B. Inducible LNCaP-AR-V7 cells were changed to stripped serum and treated with vehicle (EtOH), 10 nM R1881, or 20 ng/mL Doxycycline (Dox) for 24 hrs and protein detected by western blot. C. LNCaP and LNCaP-AR-V7 cells were changed to stripped serum and treated with vehicle (EtOH), 10 nM R1881, or Doxycycline for 24 hrs and harvested for RNA. AR target gene (FKBP5 and PSA) mRNAs were measured by q-PCR and normalized to 18S mRNA. D. LNCaP-AR-V7 cells were treated with vehicle (EtOH), 1 nM R1881 or 20 ng/ml Dox in stripped serum for the time periods indicated. Cells were counted using a Coulter Counter. E. Migration chambers were used to examine migratory ability of the cells. LNCaP-AR-V7 cells were treated with vehicle (EtOH), 1 nM R1881 or 20 ng/mL Dox in serum-free medium (top chamber) and movement into the full-serum medium (bottom chamber) was measured after 48 hours. **p < 0.01 compared to respective vehicle, n = 3.

Figure 2. Analysis of AR isoform regulation of ECAR and OCR

A.-C. LNCaP-AR-V7 cells were changed to 10% stripped serum and treated with vehicle (EtOH), 10 nM R1881, or 20 ng/mL Dox for 24 hours prior to Seahorse assay. A. Extracellular acidification rate (ECAR), B. area under the curve calculations (AUC) for ECAR, and C. oxygen consumption rate (OCR) were measured using the Seahorse Bioscience -The XF^e Analyzer. Values were normalized to cell number. Cells were treated with different mitochondrial inhibitors (i.e. A. Oligomycin, B. FCCP, and C. Rotenone) over the course of the flux experiment. Oligomycin is a Complex V inhibitor that blocks ATP synthase. FCCP is a mitochondrial oxidative phosphorylation inhibitor. Rotenone is a Complex I inhibitor that blocks electron flow. Values were normalized to cell number. n = 3.*p < 0.05 compared to respective vehicle, n = 3.

Figure 3. Full-length AR and AR-V7 have unique metabolic profiles in LNCaP cells

A. A simplified diagram of metabolism highlighting the basic pathways of glycolysis, the TCA cycle, and glutaminolysis. B. Using liquid chromatography-mass spectrometry (LC-MS), we examined the effect of AR or AR-V7 activation on the levels of a series of metabolites. In this study, the vehicle group is parental LNCaP cells treated with EtOH and 20 ng/mL Dox. The R1881 group is parental LNCaP cells treated with 10 nM R1881 and 20 ng/mL Dox. The AR-V7 group is inducible LNCaP-AR-V7 cells treated with 20 ng/mL Dox. All treatments were for 48 hours. The metabolic profile of Vehicle versus R1881 (i.e. activated AR) versus AR-V7 (20 ng/ml Dox treated cells) is depicted. G6P/F6P = glucose-6-phosphate / fructose-6-phosphate, 3PG/2PG = 3-phosphoglycerate / 2-phosphoglcerate, AMP = adenosine monophosphate, FBP/GBP = fructose-bisphosphate / glucose-bisphosphate, and DPA = docosapentaenoic acid.

Figure 4. AR and AR-V7 differentially regulate metabolites in LNCaP cells

AR isoform-specific regulation of specific metabolites involved in A. glycolysis, B. the tricarboxylic acid (i.e. citric acid) cycle (TCA), and C. glutaminolysis are depicted from the following treatments: vehicle (EtOH), 10 nM R1881, and 20 ng/mL Dox (i.e. AR-V7 expressing cells). *p < 0.05, **p < 0.01, and ***p < 0.001 compared to respective vehicle, n = 4. 3PG/2PG = 3-phosphoglycerate / 2-phosphoglcerate, AKG = α-ketoglutarate, and OAA = oxaloacetate.

Figure 5. Castration resistant tumors have reduced levels of citrate

The relative citrate levels in the samples from Figure 4B were compared with a data set [17] measuring citrate in 16 benign prostate tissue samples, 12 androgen-dependent prostate cancer tumor samples, and 14 metastatic, castration-resistant prostate cancer tumor samples. A. The steady-state levels of citrate from the LNCaP and LNCaP-AR-V7 cells treated with vehicle, 10 nM R1881, and 20 ng/mL Dox (i.e. AR-V7 expressing cells) are graphed as boxplots from our metabolomics data set (Figure 4B). B. Citrate levels from the tumor samples are depicted from benign, PCa, and CRPC tissues samples.

Figure 6. Analysis of AR isoform-specific progression through glycolysis with [U-13C]-glucose metabolic flux analysis

A. Schematic of universally-labeled glucose with 13C (green circles) progressing through glycolysis and entering the TCA cycle with citrate having 2 carbons labeled and 4 unlabeled carbons (grey circles) in the first round. The mass isotopomers of the metabolites labeled in blue were analyzed. B. Mass isotopomers of glucose-6-phosphate and fructose-6-phosphate (G6P F6P) C. fructose-1,6-bisphosphate (FBP), and D. citrate were measured. First, LNCaP and LNCaP-AR-V7 cells were changed to 10% stripped serum and treated with vehicle (EtOH), 10 nM R1881, and/or 20 ng/mL Dox for 24 hours. The next day the medium was changed to starvation medium (i.e. without glucose for 6 hours). Then [U-13C]-glucose was added to each group for 3 hours. Cells were harvested and processed as described in the methods. *p < 0.05, **p < 0.01, n = 4.

Figure 7. Comparison of AR and AR-V7 Induced flux through the TCA cycle in LNCaP cells

A. LNCaP and LNCaP-AR-V7 cells were changed to 10% stripped serum and treated with vehicle (EtOH), 10 nM R1881, or 20 ng/mL Dox for 24 hrs and harvested for RNA. Metabolic genes, MDH1 and OGDH, were measured by q-PCR and normalized to 18S mRNA. B. Schematic of universally-labeled glucose with 13C (green circles) entering and progressing through the TCA cycle with citrate having 2 labeled carbons and 4 unlabeled carbons (grey circles) at the start of the first cycle. The mass isotopomers of the metabolites labeled in blue were analyzed. Mass isotopomers of C. fumarate and D. malate were measured after culture of LNCaP and LNCaP-AR-V7 as described in Figure 6. *p < 0.05, **p < 0.01, n = 4.

Figure 8. Analysis of AR isoform-specific utilization of glutaminolysis with [5-13C]-glutamine and [1-13C]-glutamine metabolic flux analysis

A. Schematic of [5-13C]-glutamine (i.e. only the fifth carbon of glutamine is labeled; 13C (green circles)) progressing through either the TCA cycle or reductive carboxylation leading to a production of citrate having 1 carbon labeled and 4 unlabeled carbons (gray circles) at the end of the first cycle. Mass isotopomers of B. citrate C. malate, and D. fumarate were measured. LNCaP and LNCaP-AR-V7 were changed to 10% stripped serum and then treated with vehicle (EtOH), 10 nM R1881, and/or 20 ng/mL Dox for 24 hours. The next day the medium was changed to starvation medium (i.e. without glutamine for 6 hours). Then [5-13C]-glutamine was added to each group for 3 hours. Cells were harvested and processed as discussed in the methods. E. Schematic of [1-13C]-glutamine (i.e. only the first carbon of glutamine is labeled; 13C (green circles)) progressing through reductive carboxylation with citrate having 1 carbon labeled and 4 unlabeled carbons (gray circles) at the end of the first cycle. Mass isotopomers of F. citrate G. malate, and H. fumarate were measured after culture of LNCaP and LNCaP-AR-V7 as described for [5-13C]-glutamine above. I. LNCaP-AR-V7 cells were changed to 10% stripped serum and then treated with vehicle (EtOH), 10 nM R1881, or 20 ng/mL Dox for 24 hours. Metabolic gene involved in glutaminolysis, GLUD1, was measured by q-PCR and normalized to 18S mRNA. *p < 0.05, **p < 0.01, and ***p < 0.001 compared to respective vehicle, n = 4.

Figure 9. AR induces lipid accumulation and beta-oxidation and regulates the fatty acid synthesis pathway

A. LNCaP-AR-V7 cells were changed to 10% stripped serum and then treated with vehicle (EtOH), 10 nM R1881, or 20 ng/mL Dox for 6 days. Media and treatments were changed out and replenished after 3 days. Oil Red O staining was used to stain for lipids in vitro. Red staining indicates the presence of lipid droplets. Images were taken at 20X magnification. B. Schematic of fatty acid synthesis pathway depicting the generation of lipids through conversion of Acetyl CoA through the rate-limiting step of ACACA, FASN, and the downstream long-chain fatty acid elongase, ELOVL7. C. LNCaP and LNCaP-AR-V7 cells were treated with vehicle (EtOH), 10 nM R1881, or 20 ng/mL Dox for 24 hours and harvested for RNA. ACACA, FASN, and ELOVL7 mRNAs were measured by q-PCR and normalized to 18S mRNA. D. LNCaP-AR-V7 cells were changed to 10% stripped serum and treated with vehicle (EtOH), 10 nM R1881, or 20 ng/mL Dox for 72 hours. Cells were harvested and AR, AR-V7, p-AMPK, AMPK, and Tubulin protein expression was detected with Western blot. (E-F) LNCaP and LNCaP-AR-V7 cells were changed to 10% stripped serum and then treated with vehicle (EtOH), 10 nM R1881, or 20 ng/mL Dox for 24 hours. Radiolabeled palmitate or oleate was added to the media. Oxidation of radiolabeled E. palmitate and F. oleate were measured using CO₂ trap assays and values normalized to DNA content. *p < 0.05 compared to respective vehicle, n = 3.

Figure 10. AR utilizes the TCA cycle, while AR-V7 preferentially enhances glutaminolysis

This model summarizes our findings of the metabolic profiles and functions of AR and AR-V7. Beginning with glycolysis, both AR and AR-V7 increased extracellular acidification rate measuring glycolytic activity. Glycolytic genes and intermediate metabolites were also increased by both AR isoforms. Metabolic flux analysis revealed that AR-V7 accelerated glycolysis more effectively than AR. Steady state metabolites, gene expression, and metabolic flux studies showed that AR progresses through the TCA cycle. Dashed arrows next to the metabolic genes indicate AR isoform specific action. AR can utilize both glucose and glutamine as starting material to progress through the TCA cycle to generate energy and fuel for cell growth and survival. Furthermore AR can utilize both the canonical the TCA cycle and reductive carboxylation for glutamine metabolism. Interestingly, AR-V7 has an increased dependence on reductive carboxylation to utilize glutamine as an energy source as seen with increased gene expression and metabolic flux analysis. Thus, AR-V7 can mimic AR's metabolic functions regulating glycolysis and also has unique metabolic roles preferentially enhancing glutaminolysis often seen in many cancer cells (i.e. a component of the Warburg effect).