Lipid catabolism inhibition sensitizes prostate cancer cells to antiandrogen blockade

Abstract

Prostate cancer (PCa) is the most common malignancy among Western men and the second leading-cause of cancer related deaths. For men who develop metastatic castration resistant PCa (mCRPC), survival is limited, making the identification of novel therapies for mCRPC critical. We have found that deficient lipid oxidation via carnitine palmitoyltransferase (CPT1) results in decreased growth and invasion, underscoring the role of lipid oxidation to fuel PCa growth. Using immunohistochemistry we have found that the CPT1A isoform is abundant in PCa compared to benign tissue (n=39, p<0.001) especially in those with high-grade tumors. Since lipid oxidation is stimulated by androgens, we have evaluated the synergistic effects of combining CPT1A inhibition and anti-androgen therapy. Mechanistically, we have found that decreased CPT1A expression is associated with decreased AKT content and activation, likely driven by a breakdown of membrane phospholipids and activation of the INPP5K phosphatase. This results in increased androgen receptor (AR) action and increased sensitivity to the anti-androgen enzalutamide. To better understand the clinical implications of these findings, we have evaluated fat oxidation inhibitors (etomoxir, ranolazine and perhexiline) in combination with enzalutamide in PCa cell models. We have observed a robust growth inhibitory effect of the combinations, including in enzalutamide-resistant cells and mouse TRAMPC1 cells, a more neuroendocrine PCa model. Lastly, using a xenograft mouse model, we have observed decreased tumor growth with a systemic combination treatment of enzalutamide and ranolazine. In conclusion, our results show that improved anti-cancer efficacy can be achieved by co-targeting the AR axis and fat oxidation via CPT1A, which may have clinical implications, especially in the mCRPC setting.

Keywords: CPT1A; INPP5K; enzalutamide; prostate cancer; ranolazine.

Figure 1. CPT1A expression is increased in advanced prostate cancer

A.-B. Representative images of serial sections of benign and cancer tissue (arrows) from the same RRP specimen stained with H&E (A) or CPT1A (B) specific stain. C. Quantification of CPT1A stain in 39 RRP specimens grouped by Gleason Score (GS). ANOVA: p < 0.0001, post hoc tests for each GS group compared to benign: GS7, n = 15, *p = 0.02; GS8, n = 12 *p = 0.018; GS9, n = 12, *p = 0.02. AU = arbitrary units. D. Graphical representation of Oncomine data (Setlur dataset) showing increased expression of CPT1A with advanced Gleason score. E. Graph from cBioPortal showing CPT1A gene amplification in neuroendocrine (NEPC) and adenocarcinoma (SUC2C) samples in 2 recent datasets (Trento /Cornell/Broad 2016 and stand-up-2-cancer/PCF projects, respectively).

Figure 2. CPT1A is needed to maintain viability and invasion of prostate cancer cell lines

A. Western blot of CPT1A- knockdown (KD) LNCaP clones (sh-1 and sh-2). B. Increased accumulation of lipid in CPT1A-KD clones. Red stain shows lipid droplets. C. Decreased clonogenicity of CPT1A-KD clones, *p < 0.01 compared to control (NT). D. Decreased invasion of KD clones compared to controls (NT), *p < 0.01. E. Western blot of CPT1A-KO CRISPR-edited LNCaP cells (CPT1A-KO). F.-G. Representative clonogenic assay (F) and live cell photographs (G) of CPT1A-KO cells. H. Decreased invasion of CPT1A-KO cells compared to controls (CTRL), *p < 0.01. I. Decreased palmitate oxidation in CPT1A-KO cells compared to control clones, *p < 0.01.

Figure 3. Gene expression results from CPT1A-KD LNCaP cells. A

Results from RNAseq analysis of CPT1AKD cells compared to controls (NT). Upregulated (red) and downregulated (blue) genes were used to generate a protein interaction network. B. Protein interaction diagram of the significant genes from (A). The figure is drawn using the STRING database through Cytoscape (http://cytoscape.org) to identify interactions of the up- or down-regulated genes from CPT1A knockdown. AR and UBC genes have a non-colored background because they are not significantly changed in the RNAseq analysis. C. Semi-quantitative RTPCR of significant AR-regulated genes: ACPP (acid phosphatase, *p ≤ 0.01) KLK3 (PSA, *p < 0.001), AZGP1 (*p ≤ 0.01) and CLU (*p ≤ 0.001). D. RTPCR of AR-full length (AR-FL, *p < 0.01), AR variant 7 (AR-v7, *p ≤ 0.001) compared to NT controls. E. Representative western blot of full length AR in CPT1AKD clones. N = N-terminus specific antibody (N-20). C = C-terminus specific antibody (C-19). F. Clonogenic assay of CPT1AKD clones in the presence of DHT (100 pM), ANOVA p = 0.006, *p ≤ 0.04 compared to NT control.

Figure 4. Decreased phosphatidylinositol (PI) and increased INPP5K expression result in decreased activation of AKT and increased expression of AR in CPT1A KD cells

A. Increased mRNA of INPP5K (inositol phosphatase, *p < 0.01) in LNCaP CPT1A KD cells. B. Phospholipid analysis of the CPT1AKD cells. The x-axis represents phosphatidic acid, phosphatidylcholine, phosphatidylethanolamine, phosphatidylglycerol, phosphatidylinositol and phosphatidylserine, respectively. The graph represents the sum of the area ratios of every molecular species in each phospholipid class, normalized to the total phospholipid content in each extract of cells, and it shows decreased total abundance of PI species and increased PC species (*p ≤ 0.05 compared to NT control). The inset represents the area ratio of specific PI(18:0/20:4) species (*p < 0.05 compared to NT control). C. Representative array blot (Cell Signaling) of CPT1A-KD cells showing decreased p-AKT signal. D. Signal array quantification, *p < 0.001 compared to control (NT clones). E.-F. Representative western blots showing CPT1A-KD and double CPT1A+INPP5K KD cell lysates probed for p-AKT, INPP5K, CPT1A and AR in the double KD cell lysates. G. Diagram of the site of action of INPP5K, PI and its potential effect on AR expression via decreased activation of p-AKT. The decreased expression of CPT1A (1) leads to increased phospholipid degradation and increased INPP5K phosphatase activity, (2) resulting in decreased AKT expression and activation (3). This consequent decrease in p-AKT leads to increased AR protein expression and action, increasing PSA (KLK3) and resulting in a more differentiated phenotype.

Figure 5. Combinatorial effects of CPT1A inhibition and anti-androgen therapy in human PCa cells

A. CPT1A-KD cells show increased sensitivity to the anti-androgen enzalutamide (MDV) at 48 hours, a p < 0.001, *p < 0.01. B. 22RV1 CPT1A-KD cells show decreased protein expression of the ligand-independent AR variant 7 (ARv7). C. 22Rv1 cells deficient in CPT1A show decreased growth and increased sensitivity to enzalutamide, ANOVA, p < 0.001. Post hoc Tukey tests: *p < 0.001 compared to control clone (NTshRNA) and vehicle treatment, #p = 0.035 compared to vehicle treatment. D. LNCaP-enzalutamide resistant cells can grow in the presence of enzalutamide (MDV), ANOVA for MDV-resistant cells p < 0.001, Post hoc *p < 0.001 compared to parental cell line for each drug dose. E. Increased sensitivity to the combination of ranolazine (Rano), etomoxir (Etom) or perhexiline (PMS) with enzalutamide (MDV) in LNCaP-enzalutamide resistant cells, post hoc tests *p < 0.05 compared to vehicle. F.-H. Increased sensitivity to the combination of etomoxir (F), perhexiline (G) or ranolazine (H) with enzalutamide in 22Rv1 cells, post hoc tests *p < 0.001 compared to vehicle treatments.

Figure 6. Combinatorial effects of beta-oxidation inhibition and anti-androgen therapy in mouse TRAMPC1 cells and human 22Rv1 xenografts

A.-C. Clonogenic assay showing the effects of the combination of ranolazine (A), etomoxir (B) or perhexiline (C) with enzalutamide (MDV) in mouse TRAMPC1 cells, post hoc tests *p ≤ 0.03 compared to individual treatments. D. Tumor growth in mice treated with vehicle (0.5% methyl cellulose with 0.1% tween 80), enzalutamide (20 mg/kg), ranolazine (40 mg/kg) or a drug combination (enza+rano) via gavage over 21 days. Five mice with 2 tumors each were used for each treatment group. Arrow indicates beginning of treatment. Repeated measures ANOVA showed significant changes due to the effect of treatment over time (p < 0.001, F = 2.5 df = 24). *p < 0.05 (paired t-test comparing control to combination group). E. Effect of treatments on mouse body weights. No significant changes and no signs of toxicity were observed throughout the study.