Background
Cancer is characterized by dysregulated growth and proliferation; in proliferating malignant cells there is an enhanced requirement for building blocks, including amino acids, nucleic acids and lipids. In addition to modulating glucose metabolism and energy production [
1,
2], neoplastic cells also alter lipid metabolic pathways [
3,
4] factoring net biosynthesis over energy production [
5]. In various cancers, lipogenesis and cholesterol synthesis pathways are upregulated and several of these over expressed genes correlate with poor prognosis [
6,
7]. In contrast to carbohydrate metabolism, little is known about the role of fatty acid metabolism in promoting cancer cell growth and metastasis [
8,
9].
Recent studies have shown that cancer cells not only use fatty acids as a building blocks but also use them preferentially for ATP production through fatty acid oxidation [
10,
11]. Neoplastic cells alter lipid metabolizing enzymes, triggering oncogenic signaling to promote growth [
12]. Dysregulated lipid metabolism also promotes aberrant cancer cell-stromal cell communication, contributing to disease progression. In some cancer types, neoplastic cells derive energy from supporting host cells by modulating their metabolic activity [
13,
14]. In several cancers dysregulated fatty acid (FA) synthesis, storage, uptake transport and degradation are associated with disease outcome. Some of these cancer cells are known to upregulate FA synthesis which in turn supports rapid proliferation and decreased drug sensitivity [
12,
13,
15,
16]. Cancer cells tend to alter FA synthesis by increasing production of fatty acid precursors glutamine and citrate; alternately they also uptake extracellular FA for use as building blocks, energy production and storage [
17‐
19]. Knockdown studies on FA synthesis genes show poor prognosis and decreased overall survival in several cancers including prostate [
13,
18,
20,
21] hence FA synthesis genes have been implicated as therapeutic targets [
15].
Our recent studies demonstrate that cancer cells tend to uptake FA and store them as lipid droplets which can be used later to aid proliferation [
17,
22‐
24]. The preferential uptake of lipids over glucose in prostate cancer circulating tumor cells has been assessed for potential therapeutic targeting [
25]. Upon entering the circulation, CTCs uptake lipid, storing them in the form of lipid droplets that may be used subsequently for growth and proliferation at the metastatic site. As the neoplastic cells uptake increasing amount of FA, size and number of the lipid droplets increase [
26]. The increase of lipid droplet size is an indication of increased TG mass which is catalyzed by several enzymes present on the lipid droplet monolayer in collaboration with ER which plays a major role in lipid droplet dynamics [
27,
28].
The enzymes involved in the synthesis of TG from FA aid in the increase of size and number of lipid droplets whereas lipolysis enzymes metabolize TG for energy production and membrane synthesis for cell proliferation. The major enzymes involved in TG synthesis and storage are diglyceride acyltransferase (DGAT), monoacylglycerol acyltransferase (MGAT), glycerol-3-phosphate acyltransferase (GPAT) and enzymes involved in cholesterol metabolism like
ACAT (acyl-CoA cholesterol acyl transferase) [
29‐
32]. Enzymes involved in lipolysis include hormone sensitive lipase (HSL), monoacylglycerol lipase (MGL) and adipose triglyceride lipase (ATGL). Additionally there are set of lipid droplet associated proteins which regulate cellular lipid stores namely-perilipin, adipose fatty acid binding protein, caveolin, alpha/beta- hydrolase domain containing protein 5(ABHD5), tail interacting protein 47 (TIP47), OXPAT and fat-specific protein (FSP27) [
29,
31,
33]. All the above mentioned proteins regulate and maintain a balance between TG synthesis and lysis eventually deciding the size and number of fatty acid droplets in the cells. Based on the observation that neoplastic cells show increased accumulation of large lipid droplets we hypothesize that this results by controlling the balance of the above mentioned proteins.
The focus of this study is to understand the differences in FA uptake and storage between malignant epithelial cells versus peripheral blood mononuclear cells (PBMCs), with the broader goal of identifying and therapeutically targeting the tumor-specific metabolic changes. As fatty acids in malignant epithelial cells are generally stored as lipid droplets, we analyzed differential gene expression relevant to lipid droplet formation and processing between PBMCs and LNCaP cells. We then analyzed the individual differentially expressed genes in prostate cancer cell growth.
Methods
Cell lines, kits and antibodies
LNCaP (CRL-1740), HeLa (CCL-2) and OP9 (CRL-2749) cells were purchased from ATCC (Manassas, VA) and maintained in RPMI, MEM and alpha MEM media respectively from Invitrogen, (Carlsbad, CA). Differentiation in OP9 cells is induced by replacing MEM alpha medium with knockout serum at 15% concentration. PBMCs were isolated from fresh volunteer blood samples or were purchased along with plasma from Bioreclamation IVT, USA. PBMCs were growth in X-vivo15 media with or without phenol red (for MTS assay) from Lonza (Allandale, NJ). The siRNA transfection reagents, oligofectamine and OPTIMEM were from Invitrogen; the siRNAs were from Dharmacon (Thermo scientific). DGAT1 (diglyceride acyltransferase, ab181180), ABHD5 (alpha/beta- hydrolase domain containing protein 5, ab183739), ATGL (adipose triglyceride lipase, ab99532) and ACAT (Acyl-CoA cholesterol acyl transferase, ab168342) antibodies were from Abcam (Cambridge, MA). Cyclin antibodies (cyclin antibody sampler kit #9869S), ACC (acetyl CoA carboxylase) and AMPK (AMP activated protein kinase) antibodies (AMPK and ACC sampler kit #9957S), raptor (regulatory-associated protein of mTOR) and ULK (serine/threonine-protein kinase) antibodies (ULK1 sampler kit #8359S), LC3B (microtubule-associated protein 1A/1B–light chain 3) antibody (autophagosome marker antibody sampler kit #8666 s) PARP (Poly ADP-ribose polymerase, #9542) and pP70S6/P70S6 (ribosomal protein S6 kinase, #9205 and #2708) were from Cell Signaling Technology (Danvers, MA). GAPDH antibody was from Fitzgerald (Acton, MA). MTT, crystal violet, oil red and propidium iodide was from Sigma-Aldrich (St. Louis, MO). The secondary antibodies were from LI-COR (Lincoln, NE). MTS cell titer reagent was from Promega (Madison,WI). CellLite BacMam 2.0 lysosome-RFP and bodipy dye was from Molecular Probes, (part of Thermo Fisher Scientific, San Deigo, CA). Ficoll paque (density 1.077) was from GE health care (Milwaukee, WI). Apoptosis/ necrosis detection kit was from Abcam (Cambridge, MA).
Western blotting
Cells were lysed using RIPA buffer supplemented with protease and phosphatase inhibitors after the desired treatment was performed. GAPDH was used as an internal control. Infrared fluorescent-labeled secondary antibodies (IR dye 680 and IR dye 800) from LICOR Biosciences (Lincoln, NE) were used for detection using Odyssey CLx, which reduces background substantially. The bands were quantified using Odyssey software, which calculates pixel density and automatically takes an area adjacent to each band for background corrections.
siRNA inhibition
Cells were plated in complete media without antibiotics on a poly D-lysine-coated plate (80,000 cells per 6 well). After 48 h. of growth the cells were transfected using RNAimax according to the manufacturer’s instructions. The smart pool non-target (NT) siRNA was used as a transfection control with the experimental target gene siRNAs. A pool of four siRNA against the target genes were used to block the expression. The final concentration of the siRNA (NT and targets) was 30 nM. For cell growth assays, cells were trypsinized 24 h after transfection and replated in 96 well plates (2000 cells/well).
Growth assay
Growth of the cells was measured using MTT assays and/or clonogenic assays. For MTT (3-(4,5-Dimethylthiazol-2-yl)-2,5-Diphenyltetrazolium bromide) assay, 2000 cells/ 96 well or 13,000 cells/ 24 well were plated then treated for required time periods as mentioned and incubated with MTT (4 mg/ml) for 2 h. at 37 °C. Cells were centrifuged at 2000 g for 10 min and the supernatant was discarded. The cell pellet was dissolved in 100/500 μl of DMSO. A plate reader was used to read the absorption at 540 nm. Experiments were performed in octuplet/quadruplet and repeated at least three times. For clonogenic assay, cells were trysinized 48 h after siRNA transfection and dilution (1:500 and 1:1000) plated in 6 well plates. After 2 to 3 weeks, the colonies were fixed, stained (0.75% crystal violet, 50% ethanol and 1.75% formaldehyde) and counted. The assay was repeated three times.
Lipid staining assays
Lipids were stained using either oil red O (ORO) or bodipy. For staining cells with ORO they were washed with DPBS and fixed in 10% formaldehyde for 30 min at RT. Prior to staining a stock solution of 3 mg/ml was prepared in isopropanol. The stock solution was diluted in water (3:2) and incubated at room temperature for 10 min to prepare the working solution. The working solution was filtered using Whatman filter prior to staining the cells. The fixed cells were washed with water, and then incubated with 60% isopropanol for 5 min. After this the filtered oil red working solution was added for 5 min and then washed with water before viewing under the microscope. For staining with bodipy the cells didn’t need to be fixed, as the dye is cell permeable. For staining with bodipy cells were washed with 0.1% BSA in DPBS. The stock of bodipy dye was also prepared in the same solution. The cells were stained in bodipy solution (0.5 μg/ml, final) for 5 min and then again washed with DPBS containing BSA. Cells were visualized under the fluorescent/ confocal microscope in the FITC channel.
Cell cycle analysis
Cell cycle analysis was performed after staining the cells with propidium iodide (PI). The cells were trypsinized washed with PBS and fixed in 70% ethanol (optional) and then stained with PI in nicoletti buffer (propidium iodide 50 μg/ml, 0.1% sodium citrate, 0.1% triton X-100, RNase 1 mg/ml, in DPBS). Cells were analyzed using C6 Accuri flow cytometer (Becton Dickinson, Mountainview, CA). ModFit LT software was used to analyze and assign percentage values to cells present in the different cell cycle stages and remove the debris and aggregates.
Apoptosis assay
To test apoptosis; apoptosis/necrosis detection kit from Abcam (Cambridge, MA) was used. The kit uses apopxin green indicator dye which binds to phosphatidylserine which is flipped outside during apoptosis. 7-AAD dye in the kit is able to detect late apoptotic and necrotic cells as it is a membrane impermeable dye and stains cells with loss of plasma integrity. Cells were trypsinized 72 h after the siRNA treatment and stained with apopxin and 7-AAD for 30 min at RT in assay buffer as described in the manufacturers protocol. The cells were analyzed using C6 Accuri flow cytometer. The apopxin dye is generally read in FL1 channel (Ex/Em = 490/525) whereas 7-AAD is read in the FL3 channel (Ex/Em = 550/650). Alternatively cells were grown on glass plates and after staining with apopxin and 7-AAD were analyzed using the Nikon Elipse Ti confocal microscope.
Autophagy detection assays
We looked at prevalence of lysosomes for detection of autophagy and confirmed it with western blotting of LC3BII (microtubule-associated protein 1A/1B–light chain 3) and immunocytochemistry studies. For lysosomal detection we used cell light reagent BacMam 2.0 (RFP), it is a baculovirus based fluorescent protein-signal peptide fusion targeting lysosomes. The transduction method was followed as recommended by the manufacturer. Cells (40,000) were plated in the glass bottom petri plates, after 48 h cells were transduced with the reagent (24 μl). Cells were visualized under confocal microscope after overnight incubation with the reagent. Western blotting procedure using LC3B antibody was performed as described earlier. The ratio (less than one) of band intensity of LC3BI and II (faster) was used to confirm autophagy. LC3B antibody was also used in an immunocytochemistry study and intense punctate staining confirmed autophagy.
Immunofluorescence staining and analysis
Cells were either plated on glass bottom petriplates or were transferred to glass slides by cytospin after the appropriate specified treatment. The cells were fixed in 3.7% paraformaldehyde for 10 min followed by permeabilization for 5 min using 0.2% Triton X-100 in PBS. The cell were then washed twice and blocked with 10% goat serum and 1% BSA in PBS for 30 min at 37 °C followed by staining with the primary antibody (1:100 dilution, 1% BSA in PBS) for 2 h at 37 °C. After the incubation, the cells were washed three times (5 min) with PBS and incubated with secondary-TRITC labelled antibody for 1 h at 37 °C, washed three times with PBS, and mounted with DAPI (40,6-Diamidino-2-phenylindole dihydrochloride) for nuclear staining. Cells were visualized under a Nikon Elipse Ti confocal microscope and captured using NIS-elements imaging software. The experiments were repeated three to four times and the figures are representative maximum image projections with the same laser power settings.
Discussion
Cancer cells uptake FA rapidly and store them as lipid droplets suggesting that these cells have modified gene expression leading to upregulation of lipid droplet formation. Lipid droplets can be used to facilitate cancer cell growth by providing precursors essential for cell proliferation and/or by FA oxidization for energy production. Our results indicate that four enzymes ACAT, ATGL ABHD5 and DGAT1 are differentially overexpressed in prostate cancer cells as compared to PBMCs. These differentially expressed genes are involved both in the formation of the lipid droplets and in the processing of the FAs. ACAT is involved in the esterification of cholesterol and is known to promote accumulation of cholesterol ester in fat droplets. ATGL on the other hand is a lipase and is involved in the degradation of triglyceride (TG), consistent with our observation that ATGL siRNA leads to accumulation of lipid droplets. That we do not observe any effect of ACAT inhibition on cell growth suggests lipid accumulation in cancer cells is a cholesterol-independent process. As inhibition of both ACAT and ATGL did not affect the growth of the LNCaP cells we focused on the other two differentially identified genes DGAT1 and ABHD5 which effected growth. Whereas DGAT1 is involved in the synthesis of triacylglycerol, ABHD5 is involved in the degradation of triglycerides, similar to ATGL, and is known to be responsible for maintaining lipid homeostasis [
36]. Consistent with these mechanisms of action we observe that DGAT1 siRNA leads to absence of TG formation and shows absence of lipid droplets where as ABHD5 siRNA leads to accumulation of lipid droplets. As ABHD5 activates ATGL a critical component in lipolysis initiation it is possible that the lipid accumulation observed with ABHD5 siRNA treatment is due to blocking of lipolysis initiation. ATGL siRNA treatment did not result in as much lipid droplet accumulation as ABHD5 siRNA treatment and did not affect the growth of LNCaP cells. This possibly may be because ABHD5 is regulating triacylglycerol metabolism independently of ATGL in LNCaP, similar to prior reports in hepatocytes [
37]. ATGL-independent function of ABHD5 is reported but the mechanism not well understood [
38‐
40].
In the process of lipid droplet formation DGAT1 and ABHD5 genes seem to maintain a balance between the FA storage and usage. Blocking of DGAT1 leads to blockage of storage while blockage of ABHD5 leads to blockage of usage; hence both the siRNA treatments lead to unavailability of FA for cancer cell proliferation and decreased growth (Fig.
2d and e). The seemingly opposing functions suggest that lipid droplet homeostasis is tightly regulated and may be modulated according the needs of the cancer cells.
Our results indicate that both the siRNA treatments promote a G0/G1 population increase resulting from slow passage of cells through the G1/S check point associated with decrease in cyclins A2, D1 and D3. Further, our results show that both DGAT1 and ABHD5 results in cell death, the mechanism of cell death is very different, one triggers autophagy and the other trigger apoptosis. ABHD5 siRNA shows an increase in AMPK phosphorylation (Thr 173) and decrease in P70S6 phosphorylation, decrease in P70S6 phosphorylation is known to result in increased phosphorylation of eEF2 and inhibition of protein synthesis [
41‐
43] leading to apoptosis (cleaved PARP). From our experimental results we conclude that the cell death with ABHD5 inhibition involves AMPK/P70S6 axis leading to blockage of protein synthesis and apoptosis (Fig.
6). On the other hand although DGAT1 siRNA also increases AMPK phosphorylation it does not affect the P70S6/apoptosis pathway, instead it shows inhibition of ACC phosphorylation leading to reduced fatty acid synthesis and oxidation. Simultaneously inhibiting DGAT1 also leads to phosphorylation of raptor, inhibiting the m-TOR pathway and triggering autophagy [
44,
45]. The initiation of autophagy correlates with increased phosphorylation of ULK1 (S555), both m-TOR inhibition and AMPK phosphorylation can directly phosphorylate ULK1 [
46]. The triggering of autophagy and cell death is also further confirmed with LC3 punctate staining and the decrease in the ratio of LC3B I and II bands [
47‐
49]. Collectively our data indicate that DGAT1 inhibition leads to decreased energy metabolism, and cell death by autophagy by inhibition of m-TOR pathway via AMPK/ACC and AMPK/raptor/ULK1 axis respectively (Fig.
6).
Our results strongly suggest that inhibition of both DGAT1 and ABHD5 promote prostate cancer cell death. Since both DGAT1 and ABHD5 are overexpressed in cancer cells and CTCs vigorously uptake lipid as compared to PBMCs, targeting them may represent a new oncolytic approach. Targeting lipogenic enzymes to block cancer growth have been the subject of several studies and their efficacy as anticancer agents have been proven [
20,
50]. In colorectal cancer it has been shown that ABHD5 expressed in tumor associated macrophages promote cancer growth [
51]. ABHD5 knockdown is lethal in animal models and ABHD5 deficient mice die within hours of being born. There are no known inhibitors of ABHD5 in development, likely due to these global off target effects. As ABHD5 is known to interact with perilipin and activate ATGL [
52‐
54], another alternate approach can be to use small molecule ATGL inhibitors. ATGL deficient mice develop hepatic steatosis but are viable [
39]. Our results show that LNCaP cells treated with ATGL siRNA do not accumulate as much lipid droplets as compared with ABHD5 siRNA treatment and do not cause as much cell death (Fig.
2). This may be due to alternate regulation of triacylglycerol metabolism by ABHD5, independently of ATGL. Thus targeting ABHD5 will further require understanding of this alternate function and presently limits a dual inhibition approach of both ABHD5 and DGAT1. However several DGAT1 inhibitors are being developed for treating obesity and other metabolic diseases [
55,
56], and some of these are in clinical trials [
57‐
59]. Utilizing these available small molecule DGAT1 inhibitors in future studies will help assess their therapeutic feasibility in prostate cancer.
Acknowledgements
We acknowledge Dr. Kenneth Rosenthal (Roseman University) for helping with editing the manuscript.