FABP5 coordinates lipid signaling that promotes prostate cancer metastasis

Prostate cancer (PCa) is defined by dysregulated lipid signaling and is characterized by upregulation of lipid metabolism-related genes including fatty acid binding protein 5 (FABP5), fatty acid synthase (FASN), and monoacylglycerol lipase (MAGL). FASN and MAGL are enzymes that generate cellular fatty acid pools while FABP5 is an intracellular chaperone that delivers fatty acids to nuclear receptors to enhance PCa metastasis. Since FABP5, FASN, and MAGL have been independently implicated in PCa progression, we hypothesized that FABP5 represents a central mechanism linking cytosolic lipid metabolism to pro-metastatic nuclear receptor signaling. Here, we show that the abilities of FASN and MAGL to promote nuclear receptor activation and PCa metastasis are critically dependent upon co-expression of FABP5 in vitro and in vivo. Our findings position FABP5 as a key driver of lipid-mediated metastasis and suggest that disruption of lipid signaling via FABP5 inhibition may constitute a new avenue to treat metastatic PCa.

FABP5 thus represents a key transport protein delivering cytosolic lipids to nuclear receptors to promote a metastatic PCa phenotype. Given the robust increase in fatty acid metabolism and upregulation of FABP5 in metastatic PCa, we hypothesized that FABP5 may represent a central mechanism linking cytosolic lipid biosynthesis to pro-metastatic nuclear signaling. Here, using FASN and MAGL as prototypical examples, we show that the ability of these lipid-metabolizing enzymes to enhance PCa metastasis in vitro and in vivo is critically dependent upon FABP5, thus positioning FABP5 as a key node in a lipid signaling network that promotes PCa metastasis.

FASN and MAGL enhance the metastatic potential of PCa cells only in the presence of FABP5.
LNCaP cells are weakly metastatic and androgen-dependent. Overexpression of human FABP5 enhanced the migratory and invasive potential of LNCaP cells relative to empty-vector controls ( Fig. 1A; Supplementary  Fig. S1). FABP5 is a lipid chaperone and we reasoned that cytosolic enzymes such as FASN or MAGL provide FABP5 with a source of ligands to promote PCa metastasis. LNCaP cells robustly express FASN while MAGL is expressed at low, albeit detectable levels ( Supplementary Fig. S1). To determine whether FASN activity is necessary for FABP5 to enhance the metastatic potential of LNCaP cells, we treated cells with the FASN inhibitor C75 27 , which does not appreciably inhibit FABP5 or adversely impact normal cellular proliferation over the time course of our studies ( Supplementary Fig. S2). While C75 (40 µM) had no significant effect upon the migratory and invasive capacity of control LNCaP cells, it reduced migration and invasion of FABP5-expressing cells to a level comparable to vector alone (Fig. 1A). Similar effects were observed upon shRNA-mediated knockdown of FASN ( Fig. 1B; Supplementary Fig. S1), indicating that FASN provides a source of lipids that enhance migration/ invasion via FABP5. We next assessed whether FABP5 is similarly required for FASN to increase cellular migration and invasion. Overexpression of FASN in LNCaP cells failed to increase their metastatic potential compared to vector controls ( Fig. 1C; Supplementary Fig. S1). In contrast, concomitant overexpression of FASN and FABP5 increased cellular migration and invasion ( Fig. 1C; Supplementary Fig. S1) to a level greater than introduction of FABP5 alone. Collectively, these results indicate that FASN activity is required for FABP5 to enhance the metastatic potential of LNCaP cells and conversely that FABP5 is essential for FASN to increase cellular migration and invasion.
We next assessed whether the ability of FABP5 to enhance the metastatic potential of PCa cells is unique to FASN-derived lipids or agnostic of the fatty acid origin. MAGL is an enzyme that cleaves 2-monoacylglycerols to generate free fatty acids. Similar to FASN, overexpression of MAGL in LNCaP cells had no effect upon their migratory and invasive potential ( Fig. 2; Supplementary Fig. S1), consistent with previous work 7 . However, co-expression of MAGL and FABP5 increased cellular metastatic potential to an extent greater than that observed in cells overexpressing FABP5 alone ( Fig. 2; Supplementary Fig. S1). To further explore the interplay between MAGL and FABP5, we treated LNCaP cells expressing FABP5 with the selective MAGL inhibitor JZL184 28 , which also did not negatively affect normal cellular proliferation ( Supplementary Fig. S2). JZL184 (10 µM) had no effect upon the metastatic potential of FABP5 expressing cells (Fig. 1A), which we attribute to their robust expression of FASN ( Supplementary Fig. S1). However, JZL184 reduced migration and invasion in LNCaP cells co-expressing MAGL and FABP5 (Fig. 2), confirming that the enhancement of their metastatic potential is dependent upon MAGL activity. In contrast, incubation of cells co-expressing MAGL and FABP5 with C75 did not significantly reduce migration or invasion (Fig. 2), indicating that MAGL-derived lipids are able to promote metastasis in the absence of FASN activity. Interestingly, in contrast to the reported changes in long chain free fatty acids observed upon acute FASN or MAGL inhibition 2,7,29 , we did not detect changes in fatty acids in our cell-lines ( Supplementary Fig. S3). Taken together, our results demonstrate that FASN and MAGL enhance the metastatic potential of LNCaP cells only in the presence of FABP5.
We next assessed whether the interplay between FASN, MAGL, and FABP5 extends to the more aggressive PCa cell-line PC3, which expresses FABP5, FASN, and MAGL ( Supplementary Fig. S1). FABP5 knockdown reduced migration and invasion in PC3 cells ( Fig. 3A; Supplementary Fig. S1), and co-incubation of cells with both JZL184 and C75 reduced migration and invasion to a larger extent than observed with either inhibitor alone (without affecting normal cellular proliferation over the time course of the study) ( Fig. 3A; Supplementary  Fig. S2), indicating that lipid pools originating from FASN and MAGL both contribute to the metastatic potential of PC3 cells. Consistent with this notion, overexpression of FASN or MAGL increased the migratory and invasive potential of PC3 cells ( Fig. 3B; Supplementary Fig. S1). However, this increase was attenuated upon simultaneous knockdown of FABP5 (Fig. 3B). These results confirm that similar to LNCaP cells, FASN and MAGL enhance the metastatic potential of PC3 cells only in the presence of FABP5.

Nuclear translocation and PPARγ activation are essential to produce a metastatic phenotype.
Our results thus far indicate that FASN and MAGL produce lipids that converge upon FABP5 to enhance the metastatic potential of PCa cells in vitro. Given that cellular free fatty acid levels are not altered upon FABP5, FASN, and/or MAGL overexpression ( Supplementary Fig. S3), we hypothesized that FABP5 promotes metastasis by regulating lipid signaling. Previous work has established that the nuclear receptor PPARγ mediates the pro-metastatic effects of FABP5 in vitro 8,23,30 . Thus, we assessed whether FASN and MAGL enhance the metastatic potential of PCa cells via PPARγ activation. We first generated an FABP5 variant that is excluded from the nucleus by fusing FABP5 with a nuclear export signal (NES) as we previously described (Supplementary Fig. S1; Supplementary Fig. S2) 31 . Expression of NES-FABP5 in LNCaP cells failed to increase migration and invasion even in cells simultaneously overexpressing MAGL or FASN (Fig. 4A). These results suggest that nuclear entry of FABP5 is critical to promote the metastatic potential of LNCaP cells. To determine whether activation of PPARγ mediates the effects of FABP5, FASN, and MAGL, we incubated LNCaP cells with the PPARγ antagonist GW9662 (10 µM). Treatment of cells co-expressing FABP5 and FASN or MAGL with GW9662 completely suppressed the metastatic potential of the cells to levels seen in vehicle treated LNCaP cells (Fig. 4B). Similar effects www.nature.com/scientificreports www.nature.com/scientificreports/ and observed a direct interaction between PPARγ and FABP5 (Fig. 5B). Following densitometry analysis, there was significantly greater interaction between PPARγ and FABP5 in PC3 cells overexpressing FASN or MAGL compared to PC3 vector control cells (Fig. 5C). These findings were substantiated following the extraction of separate cytosolic and nuclear protein fractions (Fig. 5D) wherein PC3 cells overexpressing FASN or MAGL displayed greater accumulation of nuclear FABP5 relative to vector-expressing PC3 cells (Fig. 5E). Taken together, our results indicate that FABP5 mediates PPARγ activation upon FASN and MAGL overexpression to promote a metastatic phenotype.

FABP5 is critical for FASN-and MAGL-mediated PCa metastasis in vivo.
To determine whether FABP5 controls FASN and MAGL driven metastasis in vivo, we implanted PC3 cells expressing luciferase (PC3-Luc), whose migratory and invasive potential is comparable to PC3 cells, into the ventral lobe of the prostate gland of male BALB/c nude mice ( Fig. 6A; Supplementary Fig. S2). We assessed the luciferase signal of the Overexpression of MAGL in LNCaP cells (blue bars) increases migration and invasion only when FABP5 is co-expressed (blue and grey bars). Cells overexpressing FABP5 and MAGL were also treated with vehicle, C75 (40 µM), or JZL184 (10 µM). Data are presented as means ± SEM. *p < 0.05; **p < 0.01; ****p < 0.0001; (n ≥ 5).
primary tumor, whole mouse, and metastatic sites (femurs and lungs) for up to 7 weeks after implantation in vivo (when mice began to display morbidity) (Fig. 6). At the conclusion of week 7, the tumors, femurs, and lungs of all extant mice were excised for ex vivo analysis. Tumors were weighed (Fig. 7A) and also subjected to histological immunostaining with Ki-67, which confirmed that the tumors of all experimental cohorts were still actively proliferating at the conclusion of experimentation (Fig. 7B) 32 . Luciferase signal in femurs and lungs was quantified ex vivo (Fig. 8). Data are presented as means ± SEM. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001; (n ≥ 5).
Vector-expressing PC3-Luc cells (Vector) developed tumors that gradually metastasized to femurs while knockdown of FABP5 (shFABP5) reduced signal at the primary tumor, whole mouse, and femurs ( Fig. 6B), confirming that FABP5 inhibition reduces tumor growth and metastasis. Compared to vector-expressing cells,  The PPARγ IPs were western blotted using antibodies directed against FABP5 and PPARγ. The control antibody IPs (IgG1) were negative for FABP5. Samples were derived from the same experiment and blots were processed in parallel. (C) Following coimmunoprecipitation with PPARγ, FABP5 expression in vector-expressing PC3 cells and PC3 cells overexpressing FASN or MAGL was quantified using densitometry analysis of western blots. The signals were normalized to PPARγ. Data are presented as means ± SEM. ****p < 0.0001; (n = 3). (D) Western blot of vector-expressing PC3 cells and PC3 cells overexpressing FASN or MAGL following extraction of cytosolic and nuclear protein fractions. The purity of the collected fractions was corroborated using GAPDH and histone H3 as cytosolic and nuclear markers, respectively. Samples were derived from the same experiment and blots were processed in parallel. (E) Left, Western blot of FABP5 expression in the extracted nuclear fractions of PC3 vector control cells and PC3 cells overexpressing FASN or MAGL. Right, FABP5 expression was quantified using densitometry analysis of western blots. The signals were normalized to histone H3. Data are presented as means ± SEM. ****p < 0.0001; (n = 3). (2019) 9:18944 | https://doi.org/10.1038/s41598-019-55418-x www.nature.com/scientificreports www.nature.com/scientificreports/ overexpression of FASN (FASN) did not increase signal in the primary tumor in vivo but did lead to significantly higher signal in the whole mouse and femurs (Fig. 6C), indicating that FASN overexpression promotes metastasis of PC3-Luc cells. Ex vivo analysis confirmed higher total flux in the femurs and lungs of FASN-overexpressing  www.nature.com/scientificreports www.nature.com/scientificreports/ FABP5 in cells overexpressing MAGL (MAGL/shFABP5) significantly reduced luciferase signal in the whole mouse and femurs (Fig. 6D), which was confirmed ex vivo (Fig. 8B,C). Similar to FASN-overexpressing cells, we did observe larger tumor mass of MAGL-overexpressing cells relative to vector-expressing cells ex vivo, which was suppressed upon FABP5 knockdown (Fig. 7A).

Discussion
Converging evidence implicates dysregulated lipid metabolism and signaling as major drivers of PCa metastasis [1][2][3] . Accordingly, modulation of lipid metabolism holds therapeutic promise in the treatment of PCa, although the redundancy in lipid-metabolizing enzymes and nuclear receptors likely necessitates targeting multiple proteins in parallel. Our study provides evidence that FABP5 links cytosolic lipid metabolism and nuclear signaling, thus positioning FABP5 as a critical node in lipid signaling networks that drive PCa metastasis. Using FASN and MAGL as prototypical examples of fatty acid metabolizing enzymes commonly overexpressed in PCa, our study demonstrates that the pro-metastatic capabilities of these enzymes are critically dependent upon FABP5 expression in vitro and in vivo.
Using both the weakly metastatic LNCaP and the moderately metastatic PC3 cell-lines, we demonstrate that overexpression of FASN or MAGL in the absence of FABP5 fails to increase cellular migration and invasion. This is consistent with a previous report demonstrating that MAGL overexpression does not increase these parameters in LNCaP cells but does so in more aggressive PCa cell-lines 7 , which we now attribute to the lack of FABP5 expression in LNCaP cells. FABP5 overexpression enhances PCa metastasis 18,26 and our results demonstrate that the ability of FABP5 to enhance metastasis is critically dependent upon FASN and/or MAGL activity. Notably, these effects are not dependent upon the activity of the androgen receptor ( Supplementary Fig. S2). This interplay between FABP5, FASN, and MAGL extended to the in vivo setting as evidenced by suppression of metastasis in FABP5 knockdown cells despite the concomitant overexpression of FASN or MAGL.
FABP5 transports lipid ligands to nuclear receptors including PPARγ to promote metastasis 33,34 . Our results extend these findings by demonstrating that products of FASN, MAGL, and likely other lipid-metabolizing enzymes, rely upon FABP5 for PPARγ activation. From a therapeutic perspective, the noted redundancy in lipid-metabolizing enzymes and nuclear receptors suggests that targeting these proteins alone may not efficiently suppress metastasis. For example, many tumors weakly express FASN, while FABP5 also transports ligands to other nuclear receptors including PPAR β/δ and estrogen-related receptor α 3,10,11,14 . It is noteworthy that dietary lipids enhance PCa progression and a high fat diet could subvert some of the beneficial effects exerted by blocking lipogenesis/lipolysis 35 . In addition to endogenously synthesized lipids, FABP5 translocates exogenous fatty acids to nuclear PPAR receptors 36,37 and its inhibition would be expected to suppress metastasis induced by endogenously-synthesized as well as dietary lipids. Collectively, our study demonstrates that FABP5 plays a critical role in gating lipid-mediated metastasis and may represent a druggable node in a PCa lipid signaling network that drives metastasis. www.nature.com/scientificreports www.nature.com/scientificreports/ human FABP5, the hFABP5-eGFP-N1 plasmid was established. Briefly, human FABP5 was amplified from the pET28a-hFABP5 plasmid using the following primers: 5′-GCCTCGAGGCCACAGTTCAGCAGCTG-3′ (forward) and 5′-GCGGTACCTTCTACTTTTTCATAGATCCGAGTACA-3′ (reverse). Next, mouse FABP5 was excised from the mFABP5-eGFP construct (via Xho1 and Kpn1 digestion), utilized by Kaczocha et al. 2012, and human FABP5 was subcloned in its place using Xho1 and Kpn1 restriction enzymes 31 . Similarly, to drive the expression of GFP-tagged human NES-FABP5, the NES-FABP5-eGFP-N1 plasmid was constructed. Briefly, human FABP5 was amplified from the pET28a-hFABP5 plasmid using the previously utilized primers: 5′-GCCTCGAGGCCACAGTTCAGCAGCTG-3′ (forward) and 5′-GCGGTACCTTCTACTTTTTCATAG ATCCGAGTACA-3′ (reverse). Next, mouse FABP5 was excised from the NES-FABP5-eGFP-N1 construct (via Xho1 and Kpn1 digestion), utilized by Kaczocha et al. 2012, and human FABP5 was subcloned in its place using Xho1 and Kpn1 restriction enzymes 31 . To knockdown the expression of FABP5, a previously utilized lentiviral plasmid containing shRNA corresponding to human FABP5 was employed (shRNA clone V3LHS_402771; GE Dharmacon, Lafayette, CO) 31 . To knockdown the expression of FASN, a lentiviral plasmid containing shRNA corresponding to human FASN was purchased (shRNA clone V3LHS_332906; GE Dharmacon, Lafayette, CO). A non-silencing shRNA control for human FASN expression was also purchased (shRNA clone V3LHS_173006; GE Dharmacon, Lafayette, CO). Lentiviral packaging and transduction. All lentiviral plasmids utilized to alter protein expression levels in LNCaP, PC3, and PC3-Luc cells (pLenti-FABP5-Blast, pBK649-pLenti-FASN-Puro, pLenti-MAGL-Puro, hFABP5-eGFP-N1, NES-FABP5-eGFP-N1, V3LHS_402771, V3LHS_332906, and V3LHS_173006) were packaged into functional lentiviruses and used to infect host cells. HEK293T cells were sub-cultivated in a 10 cm cell culture plate to reach a confluency of 70-80% in complete DMEM (containing 10% FBS and 100 units/mL of penicillin/streptomycin). Next, the HEK293T cells were co-transfected with the desired lentiviral plasmid and the third-generation lentiviral packaging plasmids p-RSV-Rev (Addgene, Watertown, MA; Plasmid #: 12253, RRID: Addgene_12253), pCMV-VSV-G (Addgene, Watertown, MA; Plasmid #: 8454, RRID: Addgene_8454), and pCgpV (Cell Biolabs Inc., San Diego, CA; Plasmid #: 320024), at a 3:1:1:1 ratio using GenJet Plus transfection reagent (SignaGen, Rockville, MD) according to manufacturer's instructions. After 24 hours, the media of transfected HEK293T cells was refreshed, and functional packaged lentiviruses were harvested from the media at 48 hours. LNCaP cells were infected at a multiplicity of infection (MOI) of 10, and PC3/PC3-Luc cells were infected at a MOI of 50. Twenty-four hours following transduction, cells were refreshed with complete media, and subsequently split at 48 hours. At 72 hours, cells were subjected to puromycin-selection (1 µg/mL) and/or blasticidin-selection (5 µg/mL) (MilliporeSigma, Burlington, MA). Appropriate overexpression/knockdown of proteins was assayed and confirmed via Western blotting. Infected cells were kept under selective pressure for at least one week and passaged at least once in the absence of selection prior to experimentation.

Orthotopic implantation. Following lentiviral infection, PC3-Luc cells were propagated in culture and
implanted orthotopically into the ventral lobe of the prostate gland of male BALB/c nude mice. Mice were anesthesized under 2.0% isoflurane anesthesia (Henry Schein, Melville, NY) and received continuous anesthesia throughout the entirety of the procedure. Additionally, mice received a subcutaneous injection of buprenorphine (0.1 mg/kg) (Henry Schein, Melville, NY). The surgical area (lower abdomen) was alternately swabbed 3 times with 70% ethanol and betadine. A low midline abdominal incision of 3-4 mm was then made utilizing sterile surgical scissors. Using sterile forceps, the bladder of the mice was lifted (without disturbing other organs or musculature) to expose the ventral lobe of the prostate gland found directly beneath the bladder (if necessary, fat was moved away using a sterile cotton swab). Using a 0.5 cc syringe with a 28 G needle, 2.5 × 10 5 cells were resuspended in 20 µL of sterile PBS and implanted directly into the ventral lobe of the prostate gland. Following implantation, the bladder was replaced and the muscle layer closed using 4-0 absorbable vicryl monofilament sutures in an uninterrupted pattern. The skin layer was closed using sterile 9 mm staples. The animal was then removed from isoflurane anesthesia, placed on a heating pad during the recovery period, and monitored until awake and ambulatory. Mice were administered buprenorphine 4 hours post-surgery, followed by an additional administration every 12-24 hours for the next 48 hours. Staples were removed from the mice 7-10 days post-surgery once the incision wound had healed. Mice were continually monitored for weight and food consumption. Humane endpoints for all animals that underwent surgery were as follows: body weight decreasing by >15%, tumor/incision ulcerations, failure to groom, paralysis, respiratory distress, and/or bleeding. Sample size analysis was carried out using Russ Lenth's Java Applets for Power and Sample Size computer software which determined 8 animals/group (one-way ANOVA with Tukey multiple comparisons test; SD = 20%; Power = 80%; Contrast difference detection = 30%).
Animal imaging. Measurement of PC3-Luc cellular growth and metastasis was carried out utilizing a Caliper IVIS Spectrum imaging system (PerkinElmer, Waltham, MA). Beginning at 24 hours after orthotopic implantation, mice were imaged weekly for 7 weeks (when mice began to display morbidity). Briefly, mice received an intraperotineal injection of luciferin (150 mg/kg) (PerkinElmer, Waltham, MA) and were imaged 10 minutes later under 2.0% isoflurane anesthesia. At the conclusion of week 7, extant mice received a luciferin injection and femurs, lungs, and primary tumors were expediently excised, weighed, and imaged in a 10 cm cell culture plate. Luminescence was quantified as total flux (photons/second) using LivingImage software (PerkinElmer, Waltham, MA).
Western blotting. 20  The blots were then washed at least three times with PBST, followed by development using the Immun-star HRP substrate (Bio-Rad, Hercules, CA) and exposure to film. Densitometry analysis of western blots was carried out using ImageJ software (NIH, Bethesda, MD).
Boyden chamber migration and invasion assays. Cells were pre-treated for 2 hours before the initiation of the experiment with C75 (40 µM) and/or JZL184 (10 µM), GW9662 (10 µM), bicalutamide (10 µM) (Cayman Chemical, Ann Arbor, MI), or vehicle (0.1% DMSO). Briefly, 210 µL of cell media (±the appropriate pharmacological agent(s)/vehicle) was added to the bottom well of a Neuro Probe blind well chemotaxis chamber (Neuro Probe Inc., Gaithersburg, MD), and a porous hydrophilic polycarbonate membrane (8.0 µm pore size, 13 mm diameter) was placed on top (MilliporeSigma, Burlington, MA). When carrying out invasion assays, 100 µL of Matrigel (300 ng/µL) (Corning Inc., Corning, NY) was added to the top of the polycarbonate membrane and allowed to solidify in a humidified incubator set to 37 °C for 2 hours. Cells to be utilized in the assay were then harvested, counted, resuspended in the appropriate culture media, and added to the top well of the chemotaxis chamber (50,000 cells in 200 µL of media per chamber (±the pharmacological agent(s)/vehicle)). The completed chemotaxis chamber was then placed in a humidified incubator set to 37 °C containing 95% air and 5% CO 2 for 20 hours. After 20 hours, sterile cottons swabs were used on the top of the polycarbonate membrane to mechanically remove excess media, Matrigel (for invasion assays), and cells that had not migrated/invaded. Using forceps, the membrane was then removed from the chemotaxis chamber, placed in a 12-well plate, and the bottom fixed in 4% paraformaldehyde for 20 minutes at room temperature (Ted Pella Inc., Redding, CA). Following fixation, the paraformaldehyde was aspirated, and fixed cells were stained using Hoechst 33342 (1:1000) for 1 hour in the absence of light (Thermo Fisher Scientific, Gaithersburg, MD). After staining, the membrane was then mounted onto a slide, and all cells that had successfully migrated/invaded through the membrane were quantified using a Zeiss LSM 510 META NLO Two-Photon Laser Scanning Microscope (DAPI channel; 5x field of view) (Carl Zeiss Vision Inc., San Diego, CA). ppARγ transactivation assay. PC3 cells were seeded into 24-well plates to reach a confluency of 60-70%. Next, cells were co-transfected with plasmids encoding the PPARγ ligand binding domain fused to a GAL4 DNA binding domain (PPARγ LBD-GAL4-DBD), 4x upstream activation sequence for GAL4 upstream of luciferase (UAS 4x-TK Luc), and β-galactosidase as a transfection efficiency control (Kaczocha et al., 2012). Twenty-four hours later, the cells were lysed and processed using the Bright-Glo Luciferase Assay System (Promega Corporation, Madison, WI) and β-galactosidase Assay System (Promega Corporation, Madison, WI), according to manufacturer's instructions. PPARγ activation (measured via luciferase luminescence at 595 nm) and β-galactosidase activity (measured via absorbance at 405 nm following hydrolysis of ο-nitrophenyl-β-d-galactopyranoside to ο-nitrophenyl) were quantified using an F5 Filtermax Multi-Mode Microplate Reader (Molecular Devices, Sunnyvale, CA). Background luminescence from non-transfected PC3 cells was subtracted from all samples, and PPARγ activity was reported as relative luciferase activity (luciferase/β-galactosidase).
Cytoplasmic and nuclear protein extraction. Twenty-four hours prior to the initiation of extraction, PC3 vector control cells and PC3 cells overexpressing FASN or MAGL were sub-cultivated in 10 cm cell culture plates to reach a confluency of 70-80%. Cells were then harvested, and the cytosolic and nuclear protein fractions were isolated using the NE-PER Nuclear and Cytoplasmic Extraction Kit (Life Technologies -Thermo Fisher Scientific, Gaithersburg, MD) according to manufacturer's instructions. Following extraction, the purity of the collected fractions was assessed via western blot using antibodies directed against GAPDH and histone H3 as cytosolic and nuclear markers, respectively. Proliferation assays. Vector expressing LNCaP or PC3 cells were seeded into 24-well plates (50,000 cells/ well). Twenty-four hours later, JZL184 (10 µM) or C75 (40 µM) or JZL184 and C75 were added to the cells and subsequently incubated for 20 hours. Following incubation, cells were harvested and counted after the addition of trypan blue using a hemocytometer.
Ethics approval and consent to participate. The animal experiments conducted were approved by the Stony Brook University Institutional Animal Care and Use Committee (#850980).

Data availability
The datasets generated and analyzed during the current study are available from the corresponding author on reasonable request.