Extracellular Fatty Acids Are the Major Contributor to Lipid Synthesis in Prostate Cancer Free

Metabolic changes during malignant transformation are recognized as a major hallmark of cancer (1). In general, cancer cells exhibit increased glucose uptake and enhanced rates of glycolysis resulting in lactate and energy production, termed the Warburg effect (2). Other cancer-related alterations in intracellular metabolic pathways include elevated synthesis of nucleotides, proteins, and fatty acids (FA) to support increased rates of growth and division (3). Significant attention has centered on altered glucose and glutamine metabolism, including their roles as precursors for de novo FA synthesis/lipogenesis (4–6), yet the contribution of FAs to cancer cell biology remains elusive. FAs are the main structural components of biological membranes and are building blocks for complex lipids such as triacylglycerols (TAG) and membrane phospholipids, and signaling intermediates including diacylglycerol, phosphoinositols, sphingosine, and phosphatidic acid (7). The intracellular FA pool that acts as precursors for complex lipid synthesis is supplied by extracellular sources such as lipoprotein-contained TAGs and adipose-derived free FAs, as well as through de novo synthesis. Importantly, obese patients tend to have higher stage and a high mortality rate from a range of cancers (see ref. 8). These patients typically have increased adiposity and dyslipidemia, resulting in a lipid-rich extratumoral environment.

Prostate cancer is the most common cancer in men and the second leading cause of male cancer-related death. The mainstay of treatment for advanced prostate cancer is androgen deprivation therapy; however, this treatment is not curative, and patients inevitably develop a lethal form of the disease termed castration-resistant prostate cancer (9). Unlike most other carcinomas, prostate cancer is characterized by a slow glycolytic rate and may be more reliant on FA oxidation to provide ATP for cell proliferation and growth, due to increased rates of citrate oxidation (10). Prostate cancer also exhibits a number of lipid-specific features, including increased lipid droplet number and size in high-grade carcinomas compared with low-grade prostate carcinomas and normal prostate tissue (11), and enhanced rates of de novo FA synthesis (12–14). While the functional significance of these phenotypes is yet to be elucidated, they likely arise from aberrant activation of SREBPs and enhanced expression of FASN (14–16). The resultant increase in de novo FA synthesis, which uses acetyl-CoA as a major substrate, is the basis for evaluating ¹¹C-acetate-PET for diagnosis and staging (17); however, the influence of factors other than tumor metabolism on acetate uptake results in high false-positive results (17). One potential driving factor that influences ¹¹C-acetate uptake and metabolism could be the extratumoral lipid environment. It is known that high extracellular lipid levels influence glucose metabolism in type II diabetes (18) and in breast cancer (19–21), and can directly influence the rate of de novo FA synthesis from nonlipid sources (i.e. glucose- or glutamine-derived acetyl-CoA; ref. 22). It has also been reported that dietary fat and metabolic disorders, including dyslipidemia and obesity, are adverse prognostic factors influencing disease behavior (see ref. 23). Recently, we identified a prognostic three-lipid signature that is associated with poor prognosis in men with lethal metastatic castration-resistant prostate cancer (24). Collectively, these observations suggest that the extracellular lipid environment may influence prostate cancer FA metabolism and behavior, yet this relationship remains to be characterized.

The aim of this study was to assess FA metabolism in patient-derived explants and in a range of prostate cancer cell lines, and elucidate the role of FA metabolism in prostate cancer cell survival. Insights into these situations may provide a greater understanding into the potential role that elevated extracellular FA levels may play in obesity-induced prostate cancer progression (see refs. 23, 25).

The human prostate epithelial cell line PNT1 and the human prostate carcinoma cell lines LNCaP (androgen receptor positive, androgen sensitive), C4-2B, 22Rv1 (androgen receptor positive, androgen independent), and PC-3 (androgen receptor negative, androgen resistant) were obtained from the ATCC. Cell lines are validated annually by Garvan Molecular Genetics using a test based on the Powerplex 18D Kit (DC1808, Promega) and tested for Mycoplasma every 3 months (MycoAlert Mycoplasma Detection Kit, Lonza). All cell lines were cultured in RPMI1640 medium (Life Technologies Australia Pty Ltd.) supplemented with 10% FCS (HyClone, GE Healthcare Life Sciences) and 100 IU/mL penicillin and 100 IU/mL streptomycin (Life Technologies Australia Pty Ltd.). The passage numbers of all cell lines were below 20 between thawing and use in the experiments.

To create the spheroids, approximately 1 × 10⁶ cells were obtained by trypsinization from growing monolayer cultures. Cells (2 × 10⁴/well) were seeded in 96-well, ultra-low attachment, round-bottom plates (Costar, Corning). These cells were then centrifuged at 1,500 × g for 10 minutes and incubated at 37°C in a humidified 5% CO₂ incubator for 7 days. The cells were kept without agitation except when fresh growth medium was administered every 72 hours.

To inhibit DGAT-1 activity, cells were treated with 60 nmol/L of AZD3988 (Tocris Bioscience, Invitrogen; ref. 26) for 24 hours in RPMI, 10% FCS, and no antibiotics. After treatment, cells were washed and sensitivity to palmitate-induced apoptosis or cell growth in fresh RPMI containing 10% FCS was assessed. To inhibit FA oxidation, cells were treated with 100 μmol/L etomoxir (Sigma) in RPMI, 10% FCS, and no antibiotics.

Cells were seeded two days before the experiment and transfected using RNAiMAX transfection reagent (#13778075; Thermo Fisher Scientific) and 25 pmol Pooled CPT1A siRNA (ON-TARGETplus SMARTpool L-009749-00-0005; Thermo Fisher Scientific; ref. 27) or ON-TARGETplus Non-targeting Pool (D-001810-10-05) according to the manufacturer's instructions. After 48 hours, cells were washed and sensitivity to palmitate-induced apoptosis and FA oxidation assessed.

Human ethical approval for this project was obtained from the University of Adelaide Human Research Ethics Committee, St Andrew's Hospital Research Ethics Committee (reference number 80; Adelaide, South Australia) or St Vincent's Hospital's Human Research Ethics Committee (12/231; Sydney, New South Wales, Australia). Fresh prostate cancer specimens were obtained with written informed consent through the Australian Prostate Cancer BioResource from men undergoing robotic radical prostatectomy at either the St Andrew's Hospital (Adelaide, South Australia) or St Vincent's Clinic (Sydney, New South Wales, Australia). Patient-derived explants (PDE) were prepared and cultured as reported previously (28, 29). Briefly, a 6-mm core of tissue was dissected into 1-mm³ pieces and cultured in triplicate on a presoaked gelatin sponge (Johnson & Johnson) in 24-well plates containing RPMI1640 with 10 % FBS, antibiotic/antimycotic solution (Sigma), 0.01 mg/mL hydrocortisone, and 0.01 mg/mL insulin (Sigma). PDEs were cultured at 37°C for up to 72 hours and then rinsed twice in ice-cold PBS prior to being snap frozen.

Cell lines, spheroids, and PDEs were incubated in RPMI medium supplemented with differing concentrations of either oleate only, or 1:2:1 palmitate:oleate:linoleate (FA Mix; Sigma) as indicated, plus 10% FCS (wt/vol) and no antibiotics for 24 hours.

Explants were cultured in assay media containing 2% FA-free BSA, 0.2 mmol/L cold oleate (Sigma), and 0.2 μCi/mL [1-¹⁴C]oleate (PerkinElmer) in the presence or absence of 100 nmol/L insulin. Snap-frozen tissues were homogenized in 100 μL PBS using a Precellys24 tissue homogenizer (Bertin Technologies) and lysate transferred to 900 μL Ultima Gold scintillation fluid for counting on a Tri-Carb 2810TR liquid scintillation analyzer (PerkinElmer).

Cells were maintained in FCS-containing media prior to experimentation. Cells were washed in warmed PBS, then incubated in assay medium containing 0.5 mmol/L cold oleate or palmitate, [1-¹⁴C]oleate or [1-¹⁴C]palmitate (0.5 μCi/mL; PerkinElmer) conjugated to 2% (wt/vol) FA-free BSA and 1 mmol/L l-carnitine in low-glucose DMEM for 4 hours. Mitochondrial oxidation was determined from ¹⁴CO₂ production as described previously (30). Cells were harvested on ice-cold PBS to determine ¹⁴C-oleate incorporation into intracellular lipid pools and protein content. FA uptake was calculated as the sum of ¹⁴CO₂, ¹⁴C activity in the aqueous phase, and ¹⁴C incorporation into lipid-containing organic phase of cell lysates.

Cells were maintained in FCS-containing media prior to experimentation. Cells were washed in warmed PBS, then pulsed overnight for 18 hours in assay media containing 2% FA-free BSA, with [1-¹⁴C]oleate (1 μCi/mL; PerkinElmer) and cold oleate (C4-2B: 20 or 150 μmol/L; PC-3: 80 or 300 μmol/L) to prelabel the endogenous TAG pool. Following the pulse, the specific activity of the TAG pool was determined in a cohort of cells by measuring the ¹⁴C activity in the TAG following lipid extraction and TLC as well as the biochemical assessment of the TAG pool (for details, see Biochemical measures). TAG-derived FA oxidation (endogenous FA oxidation) was determined by measuring ¹⁴CO₂ production in another cohort run in parallel where cells were chased for 4 hours in RPMI media containing 0.5% FA-free BSA and 1 mmol/L l-carnitine.

Cells were maintained in FCS-containing media prior to experimentation. Cells were washed in warmed PBS, then incubated in the same media for oleate metabolism but with the either U-[¹⁴C]-d-Glucose or 1-[¹⁴C]-l-Glutamine (0.5 μCi/mL, PerkinElmer) in place of [1-¹⁴C]oleate and placed on cells for 4 hours. Substrate uptake was calculated as the sum of ¹⁴CO₂, ¹⁴C activity in the aqueous phase and ¹⁴C incorporation into lipid-containing organic phase of cell lysates.

Cellular lipids were extracted using the Folch method (31). Lipids were separated by TLC using heptane-isopropyl ether-acetic acid (60:40:3, v/v/v) as developing solvent for TAG and phospholipids or by a two-step solvent system for ceramides where TLC plates were developed to one-third of the total length of the plate in chloroform: methanol: 25% NH₃ (20:4:0.2, v/v/v), dried, then rechromotographed in heptane: isopropyl ether: acetic acid (60:40:3, v/v/v). ¹⁴C activity in the TAG, phospholipid and ceramide bands was determined by scintillation counting.

The contribution of oleate, glucose and glutamine to lipid synthesis was calculated by summing the ¹⁴C activity from the organic phase following lipid extraction for each substrate, expressed as pmol/min/mg, then calculating the percent contribution for each substrate.

Monolayer cultured cell and spheroid TAGs were extracted using the method of Folch and colleagues (31) and quantified using an enzymatic colorimetric method (GPO-PAP reagent, Roche Diagnostics). Cell protein content was determined using Pierce Micro BCA protein assay (Life Technologies Australia Pty Ltd.).

PDEs were washed twice in ice-cold PBS and snap frozen. Frozen sections of 5-μm thickness were dried then washed with propylene glycol prior to staining with warm (60°C) Oil Red O for 10 minutes. Tissues were differentiated in 85% propylene glycol, rinsed twice in distilled water, and counterstained with hematoxylin for 30 seconds. Tissues were rinsed in distilled water and mounted with aqueous mounting medium. Stained tissues were visualized using a Panoramic 250 Digital Slide Scanner (3D Histech) and staining quantitated using ImageJ software (v1.49t).

Spheroids were washed in PBS and fixed with 4% PFA. Spheroids were then washed with 60% isopropanol and stained with Oil Red O for 15 minutes at room temperature. Spheroids were washed with isopropanol, distilled water, and then embedded in optimal cutting temperature, before cryosectioning, and counterstained with hematoxylin. The stained spheroid droplets were observed using a Leica DM4000 microscope.

ATGL staining was performed on PDEs from 10 patients on the Leica BOND RX platform. Sections of 4-μm–thick paraffin-embedded blocks were dewaxed and rehydrated with Bond wash solution (ref: AR9222). The slides underwent antigen retrieval using a standardized heat-induced epitope retrieval protocol (HIER 20 minutes with ER2). The primary antibody against ATGL (#2138S, Cell Signaling Technology) was applied and incubated for 15 minutes, followed by application of a post-primary mouse antibody for 8 minutes, followed by a secondary rabbit antibody for 8 minutes (both part of Leica Bond Polymer Refine Detection System). Bound antibody was stained with 3, 3-diaminobenzidine tetrahydrochloride (Mixed DAB Refine) and then counterstained with hematoxylin. Optimal primary antibody concentration was determined by serial dilutions, optimizing for maximal signal without background interference. The final ATGL antibody dilution used was 1:500. ATGL staining was quantified only in areas of tumor as determined by a clinical pathologist. Only 5 of the 10 PDEs that were sectioned and stained for ATGL contained tumor.

Protein extraction from monocultures was performed as described previously (32). Cell lysates were subjected to SDS-PAGE, transferred to polyvinylidene difluoride (PVDF) membranes (Merck Millipore), and then immunoblotted with antibodies for anti-ATGL (2138S), anti-PARP (9532S), anti-ATF4 (11815S), and anti-GAPDH (2118S) obtained from Cell Signaling Technology, anti-CPT1A (ab128568), anti-DGAT-1 (ab54037), and anti-DGAT-2 (ab59493) from Abcam. Chemiluminescence performed using Luminata Crescendo Western HRP Substrate (Merck Millipore) and imaged using the Bio-Rad ChemiDoc MP Imaging System (Bio-Rad Laboratories) using Image Lab software 4.1 (Bio-Rad Laboratories).

PC-3 and C4-2B cells were plated in triplicate in 96-well plates (3 × 10³ cells/well) and a group of cells were then lipid loaded for 24 hours, with ethanol used as a vehicle control. The following day, the media were removed, cells were washed, and fresh RPMI media containing 10 % FCS supplemented with 250 μmol/L palmitate (Sigma-Aldrich) or ethanol as a vehicle control added. In separate experiments, a palmitate dose response was performed using fresh RPMI media containing 10 % FCS supplemented with 62.5, 125, 250, or 500 μmol/L palmitate or ethanol as a vehicle control. At defined time points stated in figure legends, MTT assays were performed as described previously (33), cells counted, and viability assessed by Trypan blue dye exclusion at indicated time points. In another cohort, cells were lysed for protein content determination after 4 days of treatment or immunoblot analysis after 24 hours of palmitate treatment.

Statistical analyses were performed with GraphPad Prism 7.03 (Graphpad Software). Differences among groups were assessed with appropriate statistical tests noted in figure legends. P ≤ 0.05 was considered significant. Data are reported as mean ± SEM of at least three independent determinations.

First, we assessed the influence of the extracellular lipid environment on neutral lipid levels in clinical prostate tumors cultured as patient-derived explants (PDE). PDEs incubated in FCS-containing media supplemented with 500 μmol/L of the monounsaturated FA oleate for 72 hours displayed increased intracellular neutral lipid levels (as determined by Oil Red-O staining; Fig. 1A), and a trend for increased protein levels (P = 0.09) of the lipid droplet lipase adipose triglyceride lipase (ATGL; Fig. 1B and C). PDEs also accumulated radiolabeled oleate in a time-dependent manner (Fig. 1D). FA uptake is stimulated by insulin in many tissues (34) and here this was also evident in PDEs (Fig. 1D). Collectively, these data demonstrated that castrate-sensitive prostate cancer was sensitive to the extracellular lipid environment and accumulates FAs as neutral lipids.

We further explored these observations using 3D spheroid models to assess the time- and dose-dependent effects of extracellular FAs on neutral lipid levels. Incubating androgen-sensitive LNCaP spheroids in 500 μmol/L oleate increased Oil Red-O staining, with stronger staining observed following 72 hours of culture (Fig. 1E). Increased Oil Red-O staining in LNCaP spheroids was associated with an increase in the neutral lipid TAG in a dose- and time-dependent manner (Fig. 1F). Similar patterns were observed in androgen-insensitive C4-2B spheroids (Fig. 1G and H).

We next assessed the effects of the extracellular lipid environment on intracellular TAG levels in a range of prostate cancer cell lines and normal prostate epithelial cells. Nonmalignant prostate epithelial PNT1 cells (Fig. 1I), AR-positive and androgen-sensitive LNCaP cells, AR-positive and androgen-independent C4-2B and 22Rv1 cells (Fig. 1J–L), and AR-negative PC-3 cells (Fig. 1M) accumulated intracellular TAG in a dose-dependent manner when cultured in FCS-containing media supplemented with increasing extracellular levels of oleate. Importantly, this pattern was also observed in media containing a 1:2:1 palmitate: oleate:linoleate mixture of FAs, representing a physiologic mixture that reflects the most prominent circulating free FAs (35). Collectively, these experiments demonstrated that increasing extracellular FA levels enhanced intracellular TAG levels in a range of prostate cancer model systems, and this was not restricted to a specific FA species.

The esterification of extracellular FAs into TAG for storage in lipid droplets is only one intracellular fate for these FAs. Using established radiometric approaches (36), we defined in detail the intracellular handling of extracellular FAs across a range of prostate cancer cells. We observed that FA uptake occurred in all cell lines analyzed, but was faster in LNCaP, C4-2B, and PC-3 cells (P < 0.05), and tended to be slower in 22Rv1 cells (P = 0.06) compared with PNT1 cells (Fig. 2A). While it has been long assumed that prostate cancer cells activate FA oxidation (10), little direct evidence of catabolism of FAs in prostate cancer cells has been reported. The generation of CO₂ from extracellular oleate was approximately 11-fold higher in all prostate cancer cells compared with PNT1 cells (Fig. 2B). Finally, the incorporation of extracellular FAs into TAG was increased in LNCaP and C4-2B, but lower in 22Rv1 cells compared with PNT1 cells (Fig. 2C).

The assessment of cancer cell FA metabolism is often limited to the generation of new FAs from nonlipid sources such as glucose and glutamine (i.e., de novo lipogenesis), which has been proposed to meet the increased requirement for phospholipid (37). As such, we next assessed the contribution of glucose and glutamine carbons to total lipid synthesis in a range of prostate cancer cells and nonmalignant prostate epithelial cells and compared these to the contribution of extracellular FAs. First, we observed that all cells tested took up glucose and glutamine at a greater rate compared with oleate (Fig. 2D–H). Only a very small proportion of the glucose (∼5%) and glutamine (∼2%) was partitioned to cellular lipids via de novo lipogenesis, whereas the vast majority of oleate was incorporated into lipids (∼95%; Fig. 2D–H). We further examined the relative contributions of glucose, glutamine, and oleate as substrates for total lipid synthesis by summing the absolute rates of lipid synthesis for each. Oleate contributed an average of 83% of carbons to the total lipid pool in all cell lines with glucose providing approximately 13% and glutamine contributing approximately 4% (Fig. 2I). However, LNCaP cells tended to have a greater contribution to lipid carbons from glucose compared with PNT1 cells (PNT1: 11%; LNCaP: 17%, P = 0.1), evidence of increased de novo lipogenesis as reported previously (12–14). Collectively, these data clearly demonstrated that lipid synthesis from glucose and glutamine carbons contributed only a minor fraction (∼17%) of the total lipid synthesis in the basal state.

A major destination for extracellular FAs is TAG stored in cytosolic lipid droplets (Fig. 1). This TAG is not a terminal destination for FAs and can be mobilized via the actions of neutral lipases (38, 39). As such, we next assessed the catabolism of intracellular-derived FAs. Overnight exposure to media containing higher amounts of oleate and 0.2 μCi/ml ¹⁴C-oleate increased TAG levels in both C4-2B and PC-3 cells compared with cells treated with lower amounts of oleate and 0.2 μCi/mL ¹⁴C-oleate (data not shown, similar to Fig. 1). The oxidation of intracellular TAG-derived FAs was increased in cells that were incubated overnight in media containing more oleate (Fig. 2J and K). Collectively, these experiments demonstrated that a larger intracellular FA store not only promoted FA oxidation to generate ATP, NADH, and other metabolites, but also increased mobilization of FAs, potentially for provisioning into phospholipid synthesis.

We have demonstrated that a range of prostate cancer model systems accumulate lipid droplets, which is supported by recent in vivo data (16). Lipid accumulation is associated with increased tumor burden (11, 16) and therefore it is possible that these higher intracellular lipid levels provide a survival advantage. Next we assessed the responsiveness of lipid-loaded castration-resistant C4-2B and PC3 cells to apoptotic stimuli by challenging these cells to either high levels of palmitate in FCS-containing media (21, 40, 41) or with serum-free media. Both C4-2B and PC3 cells are androgen-independent metastatic prostate cancer cell lines that model CRPC, which is currently incurable. The addition of 250 μmol/L palmitate to FCS-containing media reduced MTT absorbance within 2 days and this effect was further enhanced by 4 days compared with cells cultured in FCS-containing media (Fig. 3A). This reduced metabolic activity was consistent with a striking reduction in cell number after 4 days (Fig. 3B) and preceded by activation of PARP and ATF4 signaling after 1 day of palmitate treatment (Fig. 3C). Interestingly, PC-3 cells lipid-loaded with either oleate alone (Fig. 3A), or FA mixture (Supplementary Fig. S1A), were partly protected from palmitate-induced apoptosis (Fig. 3B) and displayed blunted activation of PARP and ATF4 signaling (Fig. 3C). Lipid-loaded PC-3 cells were similarly, but less strikingly, protected from serum starvation–induced reduction in viable cells (Supplementary Fig. S1B) and cell number (Supplementary Fig. S1C).

PC-3 and C4-2B cells are castrate-resistant prostate cancer cells that accumulate similar amounts of TAG following incubation in FA-rich media (Fig. 1), but only C4-2B cells express AR. C4-2B cells incubated in FCS-containing media supplemented with palmitate had only a mild attenuation of MTT absorbance compared with control cells grown in FCS-only media (Fig. 3D). This response to palmitate supplementation by C4-2B cells was different from PC-3 cells (Fig. 3A), and this difference was also observed in response to dose-dependent palmitate supplementation (Supplementary Fig. S2A–S2D). Specifically, the MTT absorbance for PC-3 cells was nearly zero after 2 and 4 days of 250 μmol/L and 500 μmol/L palmitate supplementation (Supplementary Fig. S2A), whereas 250 μmol/L palmitate supplementation attenuated MTT absorbance of C4-2B cells (D2: 120%, D4: 223% of day 0; Supplementary Fig. S2B) and 500 μmol/L palmitate supplementation reduced MTT absorbance (D2: 77%, D4: 63% of day 0; Supplementary Fig. S2B). The blunted MTT signal was due to reduced cell number after 4 days exposure to FCS-containing media supplemented with 250 μmol/L palmitate relative to control, but the number of cells were still greater than the number of cells at the start of the experiment (Fig. 3E). Similar to PC-3 cells, C4-2B cells pretreated with either oleate or FA mix were protected from palmitate-induced reduction in viable cells (Fig. 3D; Supplementary Fig. S1D) and cell number (Fig. 3E). Collectively, this demonstrated that PC-3 cells were highly sensitive to palmitate-induced apoptosis, whereas C4-2B cells were much less sensitive while lipid-loading both cell lines protected from this palmitate insult.

Altered intracellular palmitate metabolism is one potential explanation for the striking differences in the sensitivity to palmitate-induced lipotoxicity between PC-3 and C4-2B cells, both in the basal state and after lipid loading. To test this, PC-3 and C4-2B cells were incubated in ¹⁴C-palmitate, in the basal state or following overnight exposure to oleate, and the fate of radiolabeled FAs determined. Despite being more sensitive to palmitate-induced apoptosis, PC-3 cells had lower total palmitate uptake compared with C4-2B cells and there was no effect of overnight oleate exposure on palmitate uptake (Fig. 4A). Interestingly, C4-2B cells had significantly greater rates of palmitate oxidation compared with PC-3 cells (Fig. 4B), which was likely due to increased CPT1 protein levels (Fig. 4C). CPT1 catalyzes the rate-limiting step in FA oxidation (42). Overnight treatment with oleate induced a modest increase in CPT1 levels in both C4-2B and PC-3 cells (Fig. 4C), but this did not change the rate of palmitate oxidation (Fig. 4B).

The incorporation of palmitate into TAG for storage was also greater in C4-2B cells compared with PC-3 cells (Fig. 4D). This difference was not explained by differences in the amount of DGAT-1 or DGAT-2 (Fig. 4E), which catalyze the final reaction in TAG synthesis (43, 44). Pretreatment with oleate increased TAG synthesis in PC-3 cells up to rates equivalent to C4-2B cells, which did not change in response to oleate (Fig. 4D). In addition, ATGL protein levels were greater in C4-2B cells compared with PC-3 cells, and ATGL expression was increased in both cell lines with overnight oleate treatment (Fig. 4E), consistent with increased expression in oleate-treated PDEs (Fig. 1B and C) and enhanced lipolysis seen in oleate-treated cells (Fig. 2J and K). Overall, the partition of FAs between mitochondrial oxidation and storage was similar in C4-2B and PC-3 cells, but this intracellular partitioning of FA was shifted toward storage relative to oxidation in PC-3 cells following pretreatment with oleate (Fig. 4F).

Palmitate is a critical substrate for de novo ceramide synthesis (45) and one hypothesis to explain the enhanced sensitivity to palmitate in PC-3 cells compared with C4-2B cells was enhanced ceramide synthesis (46). However, the rate of palmitate incorporation into ceramide was lower in PC-3 cells compared with C4-2B cells (Fig. 4G). Interestingly, overnight oleate exposure reduced ceramide synthesis in C4-2B cells, but not in PC-3 cells. There were no differences in the rate of palmitate incorporation in the phospholipid pool (Fig. 4H).

From these observations, the most striking differences in palmitate metabolism between C4-2B and PC-3 cells were higher palmitate oxidation rates in C4-2B cells and the increase in palmitate incorporation into TAG following overnight oleate treatment in PC-3 cells. These differences in intracellular palmitate handling may explain the differential sensitivity to palmitate-induced apoptosis in C4-2B and PC-3 cells (Fig. 3).

Another explanation for the observed resistance of C4-2B cells to palmitate-induced apoptosis compared with PC-3 cells may be higher palmitate oxidation and higher CPT1A protein levels (Fig. 4B and C). Therefore, we tested whether inhibiting CPT1-mediated palmitate oxidation sensitized C4-2B cells to palmitate-induced apoptosis. Treating cells with 100 μmol/L of the CPT1 inhibitor, etomoxir, lowered palmitate oxidation (Fig. 5A). The addition of 250 μmol/L palmitate to growth media reduced cell viability (Fig. 5B); however, the combination of etomoxir and palmitate completely abolished MTT metabolic activity (Fig. 5B), cell number (Fig. 5C), and activated PARP signaling (Fig. 5D). Similar, but less striking, patterns were observed in siRNA-mediated CPT1A knockdown in C4-2B cells. Knockdown of CPT1A (Fig. 5E) lowered palmitate oxidation (Fig. 5F), but this was associated with a mild reduction in MTT metabolic activity (Fig. 5G) and no effect on cell number (Fig. 5H). The addition of 250 μmol/L palmitate to growth media attenuated cell viability (Fig. 5G); however, the combination of CPT1A knockdown and palmitate further lowered MTT metabolic activity (Fig. 5G) and cell number (Fig. 5H). Collectively, these results indicate that inhibition of FA oxidation sensitized C4-2B cells to palmitate-induced apoptosis.

Pretreatment with either oleate alone or the FA mixture protected PC-3 cells from palmitate-induced and serum starvation–induced apoptosis (Fig. 3; Supplementary Fig. S1). Radiometric analysis of palmitate metabolism pointed to an increase in TAG synthesis to shunt palmitate into lipid droplets (Fig. 4D) as a potential mechanism by which pretreatment with FAs protected PC-3 cells from palmitate-induced apoptosis. We directly tested this by inhibiting TAG synthesis through the addition of a DGAT-1 inhibitor only during the oleate preincubation of PC-3 cells prior to palmitate treatment. As expected, DGAT-1 inhibition blunted oleate-induced increase in PC-3 TAG content (Fig. 6A) due to reduced incorporation of radiolabeled oleate into TAG (Fig. 6B). As previously observed, palmitate treatment induced apoptosis, as determined by reduced MTT (Fig. 6C), PARP activation (Fig. 6D), and reduced cellular protein content (Fig. 6E). Preincubation with oleate blunted this effect (Fig. 6C–E); however, lowering intracellular TAG levels by DGAT inhibition in the presence of oleate restored PC-3 cell sensitivity to palmitate (Fig. 6C–E). Importantly, preincubation with DGAT-1 inhibitor did not affect subsequent PC-3 (Supplementary Fig. S3) and C4-2B (data not shown) cell growth in FCS-containing media. Therefore, TAG synthesis is required for the protective effects of pretreating PC-3 cells with FAs to palmitate-induced apoptosis.

Cancer cells require adaptive alterations in intermediary metabolism to fulfil the energy requirements and biochemical needs of their uncontrolled growth capacity (47). Unlike many other solid tumors, prostate cancer exhibits a low rate of glucose utilization and an increased dependence on lipids as a major energy source (48). While enhanced de novo fatty acid synthesis in prostate cancer has been firmly established (see ref. 23), the crucial importance of extracellular FAs in prostate cancer progression has been underappreciated and less well-studied. Using a range of models, we demonstrate that prostate cancer cells take up FAs and incorporate them into intracellular TAG for storage in a dose-dependent manner. Furthermore, these extracellular FAs are a greater contributor to intracellular synthesis compared with glucose and glutamine, which are used as substrates for de novo fatty acid synthesis. We also show for the first time, striking heterogeneity in the intracellular handling of FAs in castration-resistant C4-2B and PC-3 prostate cancer cells and in their response to palmitate-induced apoptosis. Specifically, C4-2B cells have increased FA oxidative capacity that underpins their resistance to palmitate-induced apoptosis. On the other hand, PC-3 cells are highly sensitive to palmitate-induced apoptosis, which was inhibited by prior lipid-loading to stimulate TAG synthesis. These observations highlight the diversity of intracellular FA metabolism in prostate cancer cells.

Storage of lipids is an evolutionarily conserved phenomenon, which can buffer energy fluctuations and promote survival in all cells and organisms (49). In this scenario, lipid storage predominantly refers to the partitioning of excess FAs into TAG for storage in cytosolic lipid droplets. Lipid droplets are closely localized with most intracellular organelles, notably mitochondria and endoplasmic reticulum (ER) and are highly conserved in yeast through to mammals (50) but the size and number of lipid droplets varies between cell types. Prostate cancer cells have detectable levels of TAGs and lipid droplets (11, 51) and the amounts of these correlate with disease grade (11). Here, we show that PDEs, LNCaP, and C4-2B spheroids and a range of prostate cancer cells can respond to changing levels of extracellular FAs and accumulate these as TAG in a dose-dependent manner, independent of FA species. Collectively, these in vitro observations suggest that the lipid biology of prostate cancer is likely to be influenced by the local lipid environment of the host, including local adipose tissue that may be expanded in obese patients and/or the circulating lipid profile, which itself is influenced by diet and body composition (52, 53).

Several studies have reported that exogenous FA availability can influence prostate cancer cell growth in vitro (54–60). Monounsaturated FAs, including oleate (C18:1), stimulate LNCaP cell proliferation (54), but activate apoptosis in DU145 cells (60) and retard growth in PC-3 cells (58, 59). However, another study reported that oleate stimulates PC-3, but not LNCaP cell growth (55), which cannot be explained by differences in concentration. Similar inconsistent observations have been made using other FA species including linoleate (C18:2), arachidonic acid (C20:4), and eicosapentaenoic acid (C20:5; refs. 54, 55, 59, 60). Despite these inconsistencies, overall these observations suggest that FAs can have both pro- or antiproliferative effects. Surprisingly, little is known about the effects of extracellular palmitate (C16:0), which accounts for approximately 20%–22% of FA species of complex lipids (i.e. phospholipids, TAGs etc.; ref. 61) or approximately 30% of free FAs (62) in human plasma, on prostate cancer cells. The saturated FA palmitate can induce apoptosis in a range of cells including 3T3 fibroblasts (41), peripheral blood mononuclear cells (63), human cardiac progenitor cells (64), pancreatic β cells (65, 66), macrophages (67), breast cancer cells (21, 40), and hepatocytes (68). We observed that palmitate indeed activates apoptosis in PC-3 cells but attenuates C4-2B cell growth. This is similar to our recent observations in breast cancer cells, where palmitate induced apoptosis in triple-negative MDA-MB-231 cells but only attenuated growth in estrogen receptor α–positive MCF-7 cells (21). The precise mechanism by which palmitate induces apoptosis remains to be elucidated but several mechanisms have been proposed. These include ER stress (65), impaired autophagy (63), altered NAD metabolism (68), and ceramide synthesis (66).

Prostate cancer is initially sensitive to hormonal manipulation; however, resistance to androgen deprivation therapy ultimately occurs, which results in the development of lethal metastatic castration-resistant prostate cancer. Here, we identified that the differential responsiveness of castration-resistant C4-2B and PC-3 cells to palmitate was associated with differences in palmitate handling. Specifically, attenuation of C4-2B cell growth by palmitate treatment, rather than activation of apoptosis, was due to enhanced FA oxidation compared with PC-3 cells. FA oxidation has been proposed to be a dominant bioenergetic pathway in prostate cancer cells (10). While the relative contribution of various substrates to ATP turnover in prostate cancer cells is yet to be reported, we provide clear evidence that prostate cancer cells have increased FA oxidation compared with prostate epithelial PNT-1 cells. Our data compliment a previous observation that LNCaP and VCaP prostate cancer cells have greater palmitate oxidation compared with the benign prostatic hyperplasia epithelial cell line BPH-1 (69). We also show that the differences between C4-2B and PC-3 cells may be related to CPT1A expression and function. Inhibition of FA oxidation by etomoxir and CPT1A knockdown in C4-2B cells attenuated growth, consistent with a previous observation in castration-sensitive LNCaP and VCaP cells and xenografts (69). Interestingly, the attenuation of C4-2B cell growth by the CPT1A inhibitor etomoxir was similar to palmitate treatment, with the combination of FA oxidation inhibition and palmitate treatment leading to cell death. Our results shed new light on the contribution of FA oxidation in castration-resistant C4-2B cells where pharmacologic and genetic inhibition of FA oxidation sensitized C4-2B cells to palmitate-induced apoptosis and in combination with other observations (69) suggests that targeting FA oxidation is an attractive therapeutic strategy in prostate cancer.

The saturated FA palmitate can activate apoptosis in a range of cells (21, 40, 41, 63–68, 70), via a range of proposed mechanism, and we show that this also occurs in PC-3 prostate cancer cells. Another common observation is that the addition or presence of oleate ameliorates the cytotoxic effects of palmitate (67, 68, 71–74). Interestingly, oleate also prevents arachidonic acid and linoleic acid induced cell death in DU-145 prostate cancer cells (60). Several mechanisms have been proposed including restoring insulin stimulated protein kinase B (Akt) signaling (73), attenuating palmitate-induced ER stress (71), preventing activation of the unfolded protein response (75), activating prosurvival pathways of ER stress (72), and activation of AMP-activated protein kinase and mTOR signaling (74). Cotreatment with oleate can also modulate palmitate metabolism to increase TAG synthesis and prevent diacylglycerol accumulation (73, 76) and block mitochondrial dysfunction and production of reactive oxygen species (76). We observed that pretreatment of PC-3 cells with either oleate or a FA mixture prevented palmitate-induced apoptosis, as well as serum starvation, by increasing palmitate partitioning into TAG synthesis for storage in lipid droplets. We also observed that pretreatment of C4-2B cells with either oleate or a FA mixture prevented palmitate-induced attenuation of growth. The ability of FA pretreatment to alter the response to palmitate treatment by PC-3 and C4-2B cells was not due to changes in palmitate uptake. We did observe a reduction in palmitate incorporation into ceramide in C4-2B cells; however, C4-2B cells have higher rates of ceramide synthesis from palmitate compared with PC-3 cells, despite C4-2B cells being less sensitive to the cytotoxic effects of palmitate. This suggests that ceramide synthesis does not activate apoptosis in these cells and that the preventative ability of FA pretreatment was due to altered TAG synthesis and oxidation.

We and others have demonstrated that TAG synthesis can protect cells from palmitate-induced lipotoxicity (21, 41, 77). For example, oleate supplementation promotes TAG synthesis in CHO cells that can prevent palmitate-induced apoptosis in mouse embryonic fibroblasts but this protection was not seen in DGAT1^−/− mouse embryonic fibroblasts, which lack the ability to synthesize TAG (78). Recently, we demonstrated that TAG synthesis was required for oleate to protect MDA-MB-231 breast cancer cells from palmitate-induced apoptosis (21) and here we show that this also occurs in PC-3 cells. The final step of TAG synthesis is catalyzed by DGAT1 and DGAT2 and knockdown of DGAT1 in LNCaP cells reduced cell growth and colony formation (79). Collectively, these observations demonstrate that TAG synthesis supports cancer cell progression. Interestingly, the inability to breakdown TAG to liberate FAs also impairs cell growth and invasion in LNCaP cells (51, 79). TAG hydrolysis is catalyzed by ATGL and knockdown of ATGL in LNCaP cells ablated cell invasion and growth (51). We observed that pretreatment with oleate increased ATGL protein levels as well as oxidation of TAG-derived FAs in both C4-2B and PC-3 cells. Collectively, these observations suggest that TAG-derived FAs may support prostate cancer cell progression and therefore suggest that FAs that traverse lipid droplet contained TAG pool play an important role in prostate cancer biology.

In conclusion, we report for the first time that prostate cancer cells can use extracellular FAs as both fuel for oxidation and the primary substrates for complex lipid synthesis such as TAG, and that high extracellular lipid availability further enhances FA flux in these cells. Furthermore, we identified distinct differences in palmitate handling and sensitivity between castration-resistant C4-2B and PC-3 cells. The reduced sensitivity of C4-2B cells to palmitate-induced apoptosis was due to the high rates of FA oxidation, thus suggesting a potential therapeutic vulnerability for AR-positive prostate cancer types with high levels of CPT1A. Oleate pretreatment, which stimulated TAG synthesis, prevented palmitate-induced apoptosis in AR-negative PC-3 cells and adds to a growing body of evidence suggesting that proteins regulating intracellular TAG homeostasis may be further therapeutic targets. The outcomes from these experiments inform the potential role that obesity-associated dyslipidemia (see ref. 23) or the host circulating lipidome (24) may play in influencing prostate cancer progression and therefore, these metabolic traits might be the basis for novel, targeted treatment interventions in prostate cancer.

A.Y. Zhang has received speakers bureau honoraria from AstraZeneca. L.G. Horvath reports receiving a commercial research grant from Astellas. No potential conflicts of interest were disclosed by the other authors.

Conception and design: M. Schreuder, D.N. Saunders, L.G. Horvath, L.M. Butler, A.J. Hoy

Development of methodology: A.Y. Zhang, E. Hosseini-Beheshti, M. Schreuder, R.-A. Hardie, J. Holst, A.J. Hoy

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Z.D. Nassar, A.Y. Zhang, E. Hosseini-Beheshti, M.M. Centenera, M. Schreuder, H.-M. Lin, A. Aishah, B. Varney, L.S. Lee, R.F. Shearer, R.-A. Hardie, M.S. Kakani, L.G. Horvath

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Balaban, Z.D. Nassar, A.Y. Zhang, E. Hosseini-Beheshti, M.M. Centenera, M. Schreuder, H.-M. Lin, F. Liu-Fu, L.S. Lee, R.F. Shearer, R.-A. Hardie, N.L. Raftopulos, D.N. Saunders, J. Holst, L.M. Butler, A.J. Hoy

Writing, review, and/or revision of the manuscript: A.Y. Zhang, H.-M. Lin, S.R. Nagarajan, R.F. Shearer, R.-A. Hardie, D.N. Saunders, J. Holst, L.G. Horvath, L.M. Butler, A.J. Hoy

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A.Y. Zhang, A. Aishah, S.R. Nagarajan, A.J. Hoy

Study supervision: L.G. Horvath, L.M. Butler, A.J. Hoy

Other (performed experiments with guidance): M.S. Kakani

The authors thank the Bosch Institute Molecular Biology Facility for technical support. L.M. Butler, A.J. Hoy, and J. Holst acknowledge grant support from The Movember Foundation/Prostate Cancer Foundation of Australia (MRTA3 and MRTA1). A.J. Hoy is supported by a University of Sydney Robinson Fellowship and was supported by Helen and Robert Ellis Postdoctoral Research Fellowship from the Sydney Medical School Foundation and funding from the University of Sydney. R.-A. Hardie and A.J. Hoy received support from the Sydney Medical School. S. Balaban was a recipient of a University of Sydney Australian Postgraduate Award. Z.D. Nassar is supported by an Early Career Fellowship from the National Health and Medical Research Council of Australia and John Mills Young Investigator Award from the Prostate Cancer Foundation of Australia. L.M. Butler is supported by a Principal Cancer Research Fellowship produced with the financial and other support of Cancer Council SA's Beat Cancer Project on behalf of its donors and the State Government of South Australia through the Department of Health and was supported by a Future Fellowship from the Australian Research Council (FT130101004). D.N. Saunders was supported by the National Health and Medical Research Council (project grant GNT1052963). M. Schreuder was supported by funding from the Dutch Cancer Institute KWF.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.