Alterations in lipid metabolism in cancer cells impact cell structure, signaling, and energy metabolism, making lipid metabolism a potential diagnostic marker and therapeutic target. In this study, we combined PET, desorption electrospray ionization-mass spectrometry (DESI-MS), nonimaging MS, and transcriptomic analyses to interrogate changes in lipid metabolism in a transgenic zebrafish model of oncogenic RAS-driven melanocyte neoplasia progression. Exogenous fatty acid uptake was detected in melanoma tumor nodules by PET using the palmitic acid surrogate tracer 14(R,S)-18F-fluoro-6-thia-heptadecanoic acid ([18F]-FTHA), consistent with upregulation of genes associated with fatty acid uptake found through microarray analysis. DESI-MS imaging revealed that FTHA uptake in tumors was heterogeneous. Transcriptome and lipidome analyses further highlighted dysregulation of glycerophospholipid pathways in melanoma tumor nodules, including increased abundance of phosphatidyl ethanolamine and phosphatidyl choline species, corroborated by DESI-MS, which again revealed heterogeneous phospholipid composition in tumors. Overexpression of the gene encoding lipoprotein lipase (LPL), which was upregulated in zebrafish melanocyte tumor nodules and expressed in the majority of human melanomas, accelerated progression of oncogenic RAS-driven melanocyte neoplasia in zebrafish. Depletion or antagonism of LPL suppressed human melanoma cell growth; this required simultaneous fatty acid synthase (FASN) inhibition when FASN expression was also elevated. Collectively, our findings implicate fatty acid acquisition as a possible therapeutic target in melanoma, and the methods we developed for monitoring fatty acid uptake have potential for diagnosis, patient stratification, and monitoring pharmacologic response.

Significance:

These findings demonstrate the translational potential of monitoring fatty acid uptake and identify lipoprotein lipase as a potential therapeutic target in melanoma.

Because of Otto Warburg's pioneering work, it has been recognized that cancer cells undergo metabolic changes to sustain their unbridled growth and dissemination. Lipids are major cell membrane components, signaling molecules, and also an energy source for cells. Altered lipid metabolism is essential for cancer progression, with lipids playing a key role in cancer cell proliferation and metastasis. Moreover, lipidomics is generating insight into disease mechanisms and identifying new therapeutic targets (1). While still sparse, emerging evidence implicates altered lipid metabolism in the development of melanoma too, with alterations in the lipidome differentiating benign skin neoplasms from melanoma (2).

Fatty acids are building blocks for complex lipids incorporated into membranes as well as substrates for β-oxidation, a major energy-yielding reaction. How cancer cells obtain fatty acids is altered compared with healthy tissue, although the exact mechanisms remain to be elucidated. Fatty acids can be synthesized de novo by the enzyme fatty acid synthase (FASN). FASN is overexpressed in several cancers including melanoma (3, 4) and has been the target of drug development programs over decades (5). Although FASN inhibitors can antagonize the proliferation of cancer cell lines including melanoma cell lines (6, 7), they have failed in the clinic due to insufficient efficacy and excessive toxicity (5). Fatty acids can also be scavenged from the blood supply, or potentially acquired through close contact with adipocytes (8, 9). Fatty acid scavenging by fatty acid–binding protein 7 (FABP7) has been associated with enhanced proliferation and invasive capability of melanoma cells (10–12), while CD36-mediated fatty acid scavenging is prominent in metastasis-initiating melanoma cells (13). Furthermore, it was recently shown that FATP1/SLC27A1, which transports long-chain fatty acids into cells, is enriched in melanoma compared with other cancer types (9), thereby highlighting the importance of fatty acid uptake for melanoma progression.

The small domesticated fish, Xiphophorus, medaka, and zebrafish, have been used for modeling melanoma since (in the case of Xiphophorus) the early decades of the last century. Melanocyte tumors arising in these models show comparable histopathology and molecular signatures to human melanoma due to the conservation of skin structures, melanocyte development, and signaling pathways controlling cell growth and invasion, all underpinned by the extensive conservation of vertebrate genomes (14). Previously, we have used transgenic zebrafish to establish a critical role for ERK MAPK in initiating melanocyte neoplasia and both PI3K and RAC in its malignant progression (15, 16). More recently, transgenic zebrafish were used to reveal cooperation between FATP1 and BRAFV600E in stimulating melanocyte neoplasia, while melanoma cell transplantation into zebrafish revealed preferential metastasis to adipose tissue (9).

PET is a highly sensitive in vivo imaging technique for visualizing noninvasively in 3D the distribution of an administered radiolabeled pharmaceutical (or tracer). Fluorodeoxyglucose-PET that detects glucose uptake capacity has become a well-established technique to visualize tumor locations. However, there have been relatively few PET-based studies investigating the role of lipid metabolism in cancer. PET tracers such as [11C]-choline and [11C]-acetate have emerged as useful tracers to study lipogenesis in tumors (17–20); but they cannot be used to investigate uptake of exogenous lipids. The tracer 14(R,S)-(18)F-fluoro-6-thia-heptadecanoic acid (herein [18F]-FTHA), an analogue of palmitic acid (a saturated fatty acid with a 16-carbon chain) has previously been used to study free fatty acid uptake in tissues (21) but has not yet been used to investigate fatty acid uptake in tumors.

Tumors are not uniform, and this spatial heterogeneity complicates the study of metabolism. Tumor heterogeneity also poses a major problem for therapy, particularly when the disease progresses to the metastatic stage. PET is able to image metabolic heterogeneity in vivo somewhat, however it is limited by low spatial resolution—in the low millimeter range, but exact resolutions vary depending on the scanner model and radioisotope used—and its targeted nature (defined by the tracer used). In contrast to PET, desorption electrospray ionization mass spectrometry (DESI-MS) imaging is an ambient ex vivo technique in which a stream of charged microdroplets is directed onto a tissue section to generate gaseous ions from species on the section surface. These ions are then taken up into a mass spectrometer inlet for analysis. Fatty acids and phospholipids ionize well by DESI-MS, making it an ideal lipid imaging tool (22). DESI-MS has the advantage of being an untargeted label-free technique, revealing the distribution of large numbers of lipids within a tissue section in one analysis at higher spatial resolution than achievable by PET (<100 μm). Furthermore, tandem mass spectrometry (DESI-MS/MS) can be performed postimaging to identify individual lipid species.

Here, transcriptome profiling and nonimaging mass-spectrometry–based metabolomics have been combined with PET and DESI-MS imaging modalities to investigate fatty acid uptake and lipid metabolism in oncogenic RAS-driven zebrafish melanocyte neoplasia. Transcriptome profiling identified a gene expression program associated with disease progression, which included genes involved in fatty acid uptake and glycerophospholipid metabolism. Enhanced fatty acid uptake was confirmed by both PET and DESI-MS imaging of FTHA. DESI-MS and conventional mass spectrometry also confirmed enhanced glycerophospholipid metabolism. DESI-MS provided insight into tumor heterogeneity in lipid metabolism. The product of one disease progression–associated gene, lipoprotein lipase (LPL), implicated in fatty acid scavenging (23), was found to be expressed in a majority of human melanoma cell lines and tumor samples. LPL depletion or antagonism suppressed melanoma cell growth, although high concurrent FASN expression could compensate for reduced LPL activity.

Zebrafish genetic models

All regulated procedures involving zebrafish were ethically approved by The University of Manchester Animal Welfare and Ethical Review Body and carried out under license in accordance with the UK Home Office Animals (Scientific Procedures) Act (1986), the guidelines of the Committee of the National Cancer Research Institute and the University's Policy on the Use of Animals in Research. Zebrafish were housed at the Biological Services Unit at The University of Manchester at approximately 28°C under a 14-hour light/10-hour dark cycle. Transgenic zebrafish expressing BRAFV600E or HRASG12V in the melanocyte lineage have been described previously (15). Generation of the NRASG12D construct was described previously (24) and breeding from founders resulted in the creation of a transgenic line. The HRASG12V or NRASG12D lines were crossed onto a mitfaw2/w2 (mitfa−/−) background to suppress melanocyte development and the NRASG12D transgene further onto a tp53M214K/M214K background to promote tumorigenesis. Melanocyte restoration and simultaneous overexpression of LPL, CCND1, mCherry, or eGFP were then achieved by injection of embryos with a mitfa-minigene–containing plasmid as described previously (25). Briefly, mCherry cDNA amplified from a vector and LPL cDNA amplified from zebrafish tumors using primers (Table 1) were cloned into pDONR221 using Gateway assembly (Gateway Cloning Kit, Thermo Fisher Scientific) to create pME mCherry and pME LPL. These were then incorporated into the mitfa-minigene–containing destination vector together with p5′E mitfa promoter and p3′E-polyA by Gateway cloning. The p5′E mitfa promoter, and p3′E-polyA entry clones, as well as the destination vector, and CCND1 and eGFP control expression plasmids were kindly provided by Dr. Craig Ceol (University of Massachusetts Medical School, Worcester, MA). Expression plasmid was injected into zebrafish zygotes along with Tol2 mRNA. pCS2-TP plasmid for Tol2 mRNA generation was a kind gift from Dr. Koichi Kawakami (National Institute of Genetics, Shizuoka, Japan). Only zebrafish embryos with near complete melanocyte rescue at 5 days postfertilization were retained and followed weekly for tumor development. Tumor nodules arising on fins, where the full mass could be evaluated, were measured in two directions using calipers (Expert Dual Reading Ip65 Digital Vernier Caliper 46611, Draper) when they were first detected and again 2 weeks later to determine growth rate. The cross-sectional area (A) was calculated using the formula π((d1+d2)/4)⁁2 and growth rate per week as (Aweek2-Aweek1)/2. Animals were sacrificed before clinical symptoms arose.

Table 1.

Primers for mCherry and LPL amplification

mCherry attB1F: GGGGACAAGTTTGTACAAAAAAGCAGGCTCGCCACCATGGTGAGCAAGGGCGAGGAGG 
 attB2R: GGGGACCACTTTGTACAAGAAAGCTGGGTACCTTACTTGTACAGCTCGTCCATG 
LPL attB1F: GGGGACAAGTTTGTACAAAAAAGCAGGCTTCGCCACCATGATGTTTAATAAGGGGAGAG 
 attB2R: GGGGACCACTTTGTACAAGAAAGCTGGGTACTTTACTCGTTGTTCTGTTTG 
mCherry attB1F: GGGGACAAGTTTGTACAAAAAAGCAGGCTCGCCACCATGGTGAGCAAGGGCGAGGAGG 
 attB2R: GGGGACCACTTTGTACAAGAAAGCTGGGTACCTTACTTGTACAGCTCGTCCATG 
LPL attB1F: GGGGACAAGTTTGTACAAAAAAGCAGGCTTCGCCACCATGATGTTTAATAAGGGGAGAG 
 attB2R: GGGGACCACTTTGTACAAGAAAGCTGGGTACTTTACTCGTTGTTCTGTTTG 

Transcriptome analysis

The microarray slides, custom designed by Agilent Technologies, have been described previously (26) and the design submitted to the Gene Expression Omnibus (GEO) database (www.ncbi.nlm.nih.gov/geo) with accession number GPL7735. For RNA isolation, three pools of 10 caudal fins or three pools of 6 tumors within caudal fins were snap-frozen in liquid nitrogen and subsequently stored at −80°C. Samples were homogenized in 1 mL of TRIzol reagent (Thermo Fisher Scientific), and subsequently total RNA was extracted according to the manufacturer's instructions. RNA samples were incubated for 20 minutes at 37°C with 10 U of DNase I (Roche Applied Science) to remove residual genomic DNA before purification using the RNeasy MinElute Cleanup Kit (Qiagen). The integrity of the RNA was confirmed by lab-on-chip analysis using the 2100 Bioanalyzer (Agilent Technologies). Samples used for microarray analysis had an average RNA integrity number value of 9 and a minimum RNA integrity number value of 8.

RNA labeling was performed as described previously (26). For comparison, aRNA of wild-type fish were coupled with Cy3 and aRNA of transgenic lines were coupled with Cy5. Hybridization and scanning were performed according to standard Agilent protocols. The feature extraction software version 9.5, protocol ge2_V5_95 from Agilent was used to generate the feature extraction data. For the background subtraction the option “No background subtraction and spatial detrend” was used. The arrays were scanned twice with 10% PMT and 100% PMT laser power and the XDR function was used to extend the dynamic range by 10-fold. Microarray data were imported into Rosetta Resolver 7.0 (Rosetta Biosoftware) and subjected to default ratio error modeling. Data analysis was performed for unigene clusters (Unigene Build 105) and significance cutoffs for the ratios of wild-type versus a particular transgenic line were set at 2-fold change and P ≤ 10−5. The two-dimensional hierarchical cluster analysis was performed with Rosetta Resolver settings for agglomerative algorithm (average link) with cosine correlation. The microarray data discussed in this publication have been deposited in GEO with Series accession number GSE76791 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE76791).

Human orthologs of the differentially expressed genes were determined using BioDBnet (27) and BioMart (28). Gene Ontology (GO) enrichment analysis for the human orthologs was performed using PANTHER Overrepresentation Test (Fisher Exact test, Bonferroni correction for multiple testing) through the Gene Ontology Consortium enrichment analysis platform (29). For the GO analysis, gene function and gene ontology terms were based on the PANTHER Classification System (GO Ontology database Released 2018-11-15; doi: 10.1038/nprot.2013.092).

[18F]-FTHA PET scanning

Production of [18F]-FTHA was carried out via nucleophilic [18F] fluorination of a benzyl-14-(R,S)-tosyloxy-6-thiaheptadecanonate precursor (ABX advanced biochemical compounds), based on the method by DeGrado and colleagues (30) and was fully automated on a TRACERlab FXFN Radiochemistry System (GE Healthcare). Animals were not fed the evening prior to scanning (to limit background signal in the gut). Fish were immersed in 200 mL of water containing 100 MBq of [18F] FTHA for 2 hours. Animals were then anesthetized with MS222 and placed in cuvettes (held in place with sponge) and cuvettes placed in falcon tubes filled with ice water. Static PET scans were then performed for 5 and 10 minutes on a preclinical PET Scanner (Inveon). Fish were culled immediately after scanning without recovery from anesthetic, and flash-frozen in isopentane. PET images were reconstructed using the 3d-OSEM/MAP algorithm (4 OSEM3D iterations and no MAP iterations, with a resolution of 1 mm). Images were visualized using Inveon Research Workplace Software (Siemens). Regions of interest (ROI) were drawn manually over tumors and the SUVmean (an average of all voxels) was calculated. Normalization was performed using the dose of [18F]-FTHA (average dose was 4 MBq) and the weight of the fish (average weight was 0.5 g) to give standardized uptake values.

DESI-MS imaging

Fish were immersed in 100 mL of water containing 0.5 mg of FTHA for 30 minutes, culled, then flash-frozen in isopentane. Fish were sectioned to 12 μm onto glass slides. Imaging experiments were carried out on a 2D DESI System (Prosolia) mounted on a Xevo-G2-XS quadrupole-time of flight (Q-TOF; Waters Corporation). The solvent spray consisted of 95% methanol and 5% water. Spray conditions used were a flow rate of 2 μL/minutes (using a syringe pump from Harvard Apparatus, Inc.), with a nebulizing gas of nitrogen at 4 bar pressure. Typical positions of the sprayer were used (sprayer 1.5 mm above surface, 6 mm sprayer to capillary distance, 75° sprayer impact angle, 5° collection capillary angle). The source temperature was 100°C and the capillary voltage was set between 4.82 and 5 kV. Images were acquired using a scan rate of 1–10 scans/second, and a mass range of 50–1,200 Da. The spatial resolution varied from 40–200 μm. Images were processed and normalized to total ion current using the High Definition Imaging v1.4 (Waters). MS/MS experiments were carried out on a Xevo-G2-XS Q-TOF mass spectrometer using argon as the collision gas and a collision energy range of 20 V–40 V. Spray conditions were the same as DESI imaging. Interpretation of spectra was carried out with the help of LIPID MAPS Lipidomics Gateway database (http://www.lipidmaps.org; ref. 31).

Hematoxylin and eosin staining

Hematoxylin and eosin (H&E) staining was carried out on sections after DESI-MS imaging, and also on adjacent sections. Slides were immersed in Xylene (Sigma-Aldrich) for two minutes, and then transferred to industrial Methylated Spirit (Sigma-Aldrich) for 4 × 2 minutes. Sections were then immersed in tap water followed by 3 minutes in Harris Hematoxylin (Sigma-Aldrich), a rinse in hot water, and a one second dip in Eosin (Leica). Slides were transferred to industrial methylated spirit for (4 × 2 minutes), then to xylene (4 × 2 minutes), and left to dry. Aqueous Mounting Media (Abcam) was used to mount coverslips (VWR) onto slides and left overnight to dry. Images were acquired using a [20×/0.80 Plan Apo] objective using the 3D Histech Pannoramic 250 Flash II slide scanner, and processed using Panoramic Viewer (3D HISTECH).

Nile red staining

Nile Red (Sigma-Aldrich) stock solution was prepared at 0.5 μg/mL in acetone. For staining, the stock was diluted 1:1,000 in 80% glycerol. Cryosections were allowed to defrost for 40 minutes at room temperature. Samples were briefly washed with distilled water before 100 μL of Nile Red was added to each slide. Samples were incubated for 5 minutes in the dark before being coverslipped and immediately imaged. Images were acquired on an Olympus BX51 upright microscope using a 10×/0.30 Plan Fln objective using a Coolsnap ES Camera (Photometrics) through MetaVue Software (Molecular Devices). Specific band pass filter sets for FITC and Texas Red were used to prevent bleed-through from one channel to the next. Images were then processed and analyzed using ImageJ (http://rsb.info.nih.gov/ij).

Lipid extraction and mass spectrometry analysis of metabolic pathways

Wild-type and V12RAS fish skin and tumor nodules from V12RAS fish were pooled to a mass between 25 and 35 mg. At least six biological pools were gathered for each sample. Extraction of tissue metabolites was performed as described previously by lysing tissue in a mixture of chloroform, methanol, and water to separate polar (methanol/water phase) and nonpolar (chloroform phase) metabolites. Nonpolar fractions were dried under vacuum for 16 hours. Ultra-high performance liquid chromatography mass spectrometry (UHPLC-MS) analysis of the nonpolar extracts was performed using an [Accela UHPLC system coupled to an Orbitrap Velos MS. Univariate statistical tests (mean ratio with confidence intervals, Mann–Whitney U test] and multivariate principal component analysis (PCA) were applied in R. Metabolites were annotated applying the software PUTMEDID_LCMS (32); all metabolites were annotated to level 2 as defined in the reporting standards for chemical analysis constructed by the Metabolomics Standards Initiative (33).

Integrating transcriptomic and lipidomic analyses with MetScape

The MetScape (34) plugin for Cytoscape was used for integration. The analyzed transcriptome data and mass spectrometry data were entered with their respective fold change and P values. The gene data were added with the human homologue Entrez gene identifier; the metabolites were added with Kyoto Encyclopedia of Genes and Genomes compound identifiers. All mapping was performed with unmodified, standard settings, and MetScape workflow.

Cell culture and LPL depletion by siRNA

All human melanoma cell lines were maintained in DMEM with l-glutamine, pyruvate, and sodium bicarbonate (Sigma-Aldrich) supplemented with 10% FBS (Life Technologies) and 1% penicillin/streptomycin solution (Sigma-Aldrich). Cell lines were authenticated using short tandem repeat analysis. Normal human melanocyte (NHM) cells (Life Technologies) were maintained in medium 254 (Life Technologies) supplemented with 1% human melanocyte growth supplement (Life Technologies). All cells were maintained under standard conditions at 37°C with 5% CO2 and routinely screened for Mycoplasma infection.

For LPL depletion, siRNA was purchased from Dharmacon (sequences in Table 2). All siRNAs were diluted to 20 μmol/L in 5× siRNA Resuspension Buffer (GE Dharmacon). The cells were plated 24 hours prior to transfection in 6-well plates and at a cell number to allow the cells to be at 60% confluence (200,000 cells per well). Cells were transfected with 20 pmol of siRNA using the Lipofectamine RNAiMAX Transfection Reagent (Thermo Fisher Scientific) as instructed by the manufacturer. A nontargeting control was included in every experiment. Fresh media was added after 24 hours or cells were collected.

Table 2.

siRNA sequences

Control siRNA AAUAUAAUCACUAUCAGGUGC 
LPL siRNA A GGCCUCUGCUUGAGUUGUA 
LPL siRNA B CCUACAAAGUCUUCCAUUA 
Control siRNA AAUAUAAUCACUAUCAGGUGC 
LPL siRNA A GGCCUCUGCUUGAGUUGUA 
LPL siRNA B CCUACAAAGUCUUCCAUUA 

Drug dose–response analysis and survival assays

For drug dose–response assays, cells were plated in 96-well plates and treated with serial dilutions of GSK264220A (Tocris) and C75 (Sigma Aldrich) for 72 hours. Cell survival was quantified by fixing and staining cells with 0.5% crystal violet in 4% formaldehyde and then measuring the absorbance of the solubilized dye (in 2% SDS in PBS) at an optical density of 595 nm.

RNA isolation and qRT-PCR analysis

RNA was isolated from cells using TRIzol (Qiagen). cDNA was synthesized from RNA using the Omniscript Reverse Transcriptase Kit (Qiagen) according to the manufacturer's instructions. Amplification of specific PCR products was detected using an Agilent Technologies Stratagene Mx3000P System, the SensiMix SYBR No-ROX Kit (Bioline), and the following cycling conditions: 95°C for 10 minutes and 40 cycles of 95°C for 30 seconds, 55°C for 45 seconds, and 72°C for 45 seconds. Primer (primers used in Table 3) efficiency was previously determined >90% and <110% and correlation (R2) over three log concentrations >0.97. Relative expression using β-actin (ACTB) expression to normalize and normal human melanocytes as a common reference was calculated using the ΔΔCt approach.

Table 3.

Primers used for qPCR analysis

LPL F: GACACAGCTGAGGACACTTG 
 R: TGGAGTCTGGTTCTCTCTTG 
FASN F: CTTCCGAGATTCCATCCTACGC 
 R: TGGCAGTCAGGCTCACAAACG 
CD36 F: GACCTGCTTATCCAGAAGAC 
 R: TTGCTGCTGTTCATCATAC 
FABP7 F: AGCTGACCAACAGTCAGAAC 
 R: ATGTGCTGAGAGTCCTGATG 
FATP1 F: AGTACAACTGCACGATCGTC 
 R: GCAGTCATAGAGCACGTCAG 
FATP2 F: AGTTGGTGCTGTTGGAAGAG 
 R: ACCAGAAGTCCAACTTCACC 
ACTB F: GCAAGCAGGAGTATGACGAG 
 R: CAAATAAAGCCATGCCAATC 
LPL F: GACACAGCTGAGGACACTTG 
 R: TGGAGTCTGGTTCTCTCTTG 
FASN F: CTTCCGAGATTCCATCCTACGC 
 R: TGGCAGTCAGGCTCACAAACG 
CD36 F: GACCTGCTTATCCAGAAGAC 
 R: TTGCTGCTGTTCATCATAC 
FABP7 F: AGCTGACCAACAGTCAGAAC 
 R: ATGTGCTGAGAGTCCTGATG 
FATP1 F: AGTACAACTGCACGATCGTC 
 R: GCAGTCATAGAGCACGTCAG 
FATP2 F: AGTTGGTGCTGTTGGAAGAG 
 R: ACCAGAAGTCCAACTTCACC 
ACTB F: GCAAGCAGGAGTATGACGAG 
 R: CAAATAAAGCCATGCCAATC 

LPL immunofluorescence

Cells were plated onto UV-treated cover slips contained in 6-well plates, washed with 1× PBS, and then placed in a 4% paraformaldehyde and 1× PBS solution for 30 minutes. Coverslips were then washed in 1× PBS and again in 1mol/L glycine (Sigma Aldrich). Coverslips were then incubated in 0.1% T×100/PBS (Sigma Aldrich) solution for 4 minutes to permeabilize cells before being incubated with a 1:100 dilution of mouse A4 anti-LPL monoclonal (ab21356, Abcam) in 0.5 μg/mL BSA/TBST for 20 minutes. Following this, coverslips were washed three times in PBS and then incubated with a secondary antibody conjugated to CY3 for a further 20 minutes. Coverslips were then washed in PBS and incubated with Cy3-labelled secondary antibody for 2 hours at room temperature and mounted using DAPi-containing vectashield. (Vectorshield Antifade with DAPI, Vector Laboratories). Slides were sealed with clear polish. Images were collected on a Olympus BX51 upright microscope using a 10×/0.30 Plan Fln objective and captured using a Coolsnap ES Camera (Photometrics) through MetaVue Software (Molecular Devices). Specific band pass filter sets for DAPI and Cy3 were used. Images were then processed and analyzed using ImageJ (http://rsb.info.nih.gov/ij).

Western blotting analysis

Laemmli buffer was added to cells for 1 minute before the well was scraped with a cell scraper and the buffer was transferred to a 0.5 mL tube and sonicated (VibraCell X130PB, Sonics Materials) for 30 seconds at 20% amplitude. Samples were then boiled for 5 minutes before being loaded onto precast gels in the SureLock Precast Gel System (Novex). Protein was then transferred to nitrocellulose membrane and blocked with 5% BSA in TBST or 5% skimmed milk in TBST dependent on the antibody (see table below). Primary antibodies (Table 4) were incubated at 4°C overnight, blots were washed in TBST, and secondary antibody was added at 1:5,000 for 1 hour at room temperature. The blots were then washed again before application of Western Lightning Plus ECL chemiluminescence for 1 minute. The blot was then exposed to X-ray Film (Biomax MR Film, Kodak) developed using a JP-33 Film Processor (JPI Healthcare) and subsequently scanned on an HP flat-bed scanner.

Table 4.

Antibodies used in Western blotting

AntibodyConcentration
Goat α h/m polyclonal LPL (R&D Systems) 1:1,000 (5% BSA/TBST) 
Mouse α CD36 (FA6-152; Abcam) 1:2,000 (5% BSA/TBST) 
Rabbit α FASN (C20G5; Cell Signaling Technology) 1:1,000 (5% BSA/TBST) 
Rabbit α ERK2 (Santa Cruz Biotechnology) 1:2,000 (5% milk/TBST) 
Rabbit α β-tubulin (Santa Cruz Biotechnology) 1:2,000 (5% milk/TBST) 
HRP-conjugated donkey α goat secondary (Santa Cruz Biotechnology) 1:5,000 (5% milk/TBST) 
HRP-conjugated donkey α rabbit secondary (GE Healthcare) 1:5,000 (mirrors primary) 
HRP-conjugated sheep α mouse secondary (GE Healthcare) 1:5,000 (mirrors primary) 
AntibodyConcentration
Goat α h/m polyclonal LPL (R&D Systems) 1:1,000 (5% BSA/TBST) 
Mouse α CD36 (FA6-152; Abcam) 1:2,000 (5% BSA/TBST) 
Rabbit α FASN (C20G5; Cell Signaling Technology) 1:1,000 (5% BSA/TBST) 
Rabbit α ERK2 (Santa Cruz Biotechnology) 1:2,000 (5% milk/TBST) 
Rabbit α β-tubulin (Santa Cruz Biotechnology) 1:2,000 (5% milk/TBST) 
HRP-conjugated donkey α goat secondary (Santa Cruz Biotechnology) 1:5,000 (5% milk/TBST) 
HRP-conjugated donkey α rabbit secondary (GE Healthcare) 1:5,000 (mirrors primary) 
HRP-conjugated sheep α mouse secondary (GE Healthcare) 1:5,000 (mirrors primary) 

IHC for LPL

Four micrometer sections obtained from formalin-fixed, paraffin-embedded archive samples were stained by IHC. Briefly, after deparaffinization, antigen retrieval was performed by heating the slides placed in a TRIS-Buffer pH 9 using a pressure cooker. A mouse mAb A4 against human LPL (ab21356, Abcam) diluted 1:300 was then applied for 120 minutes at room temperature. Secondary staining was performed using the Dako REAL Detection System, Alkaline Phosphatase/RED, Rabbit/Mouse Kit, based on the labeled streptavidin–biotin method. The diagnosis as well as the intensity of staining and the proportion of positive tumor cells were ascertained for each sample by a clinical dermatopathologist (J. Kamarashev, UniversitätsSpital Zürich, Zürich, Switzerland).

Transcriptome analysis reveals deregulated lipid metabolic pathways in advanced melanocyte neoplasia

Previously, we generated transgenic zebrafish displaying different grades of melanocyte neoplasia. Transgenic animals expressing BRAFV600E alone develop benign melanocyte hyperplasia, manifested as broader and darker stripes, while expression of HRASG12V alone results in malignant melanocyte neoplasia that initially expands horizontally in the epidermis and hypodermis enveloping the entire body other than the ventral surface (the radial growth phase or RGP), and then vertically, infiltrating the subcutaneous tissue (the vertical growth phase or VGP) and resulting in mainly raised tumor nodules (Fig. 1A; ref. 15). To minimize contamination from mRNA arising from nondiseased tissue that could add noise and drown out signal, transcriptome profiling was performed on RNA extracted from caudal fins of wild-type, BRAFV600E, or HRASG12V transgenic animals or tumor nodules located within caudal fins of HRASG12V transgenic animals, using a customized Agilent 44K zebrafish transcriptome microarray (26).

Figure 1.

Transcriptome profiling identifies a gene signature associated with malignant progression. A, The four models from which fin or tumor samples were excised and RNA extracted. RGP, radial growth phase; VGP, vertical growth phase. B, Bar chart depicting the number of highly significant (P < 0.0001, Kruskal–Wallis test) DEGs with at least a two-fold difference in expression from wild-type in the different models. C, A heatmap of gene expression following hierarchal clustering. D, Venn diagrams illustrating the extent of overlap in DEGs between the models. E, GO enrichment analysis reveals that 19% of DEGs in VGP samples is implicated in metabolism, of which lipid metabolism genes represent 28%.

Figure 1.

Transcriptome profiling identifies a gene signature associated with malignant progression. A, The four models from which fin or tumor samples were excised and RNA extracted. RGP, radial growth phase; VGP, vertical growth phase. B, Bar chart depicting the number of highly significant (P < 0.0001, Kruskal–Wallis test) DEGs with at least a two-fold difference in expression from wild-type in the different models. C, A heatmap of gene expression following hierarchal clustering. D, Venn diagrams illustrating the extent of overlap in DEGs between the models. E, GO enrichment analysis reveals that 19% of DEGs in VGP samples is implicated in metabolism, of which lipid metabolism genes represent 28%.

Close modal

Considering only genes with highly significant (P < 0.0001) differential expression to minimize false positives and a fold change of at least 2 compared with wild-type (Supplementary Table S1), the VGP tumor model had the greatest number of differentially expressed genes (DEG) totaling 4,429 (Fig. 1B). Hierarchical clustering of DEGs suggested a closer relationship between RGP and VGP profiles than any other two-way comparison (Fig. 1C). Indeed, a sizeable proportion of DEGs were shared in common between RGP and VGP demonstrating potentially step-wise progression, but much less so between BRAF and RAS models reflecting their very different progression pathways (Fig. 1D). Several groups have shown that melanoma develops independently from benign neoplasia and this extends to gene expression programs (35). Furthermore, melanocytes expressing BRAFV600E alone have likely undergone a course of proliferation and senescence (25) that would mean samples are divergent both from wild-type skin as well as from the V12RAS model. Many unique DEGs were detected in V12RAS-driven VGP melanocyte neoplasia. This is to be expected as melanomas increase their genetic diversity as they progress, while hypoxia in larger lesions also generates significant changes in gene transcription (36).

DEGs were converted to human orthologues using BioBDnet and BioMart, which were then subjected to GO enrichment analysis for biological processes. GO enrichment analysis of the VGP upregulated genes highlighted metabolism as an altered process, relating to 19% of upregulated genes, of which lipid metabolism genes constituted 28% (Fig. 1E; see also Supplementary Table S2 for a list of genes comprising this ontology that are also exclusive to the VGP samples). Over-represented GO terms associated with RGP upregulated genes included cell communication, signal transduction, developmental process, and MAPK signaling cascade, while development and signal transduction–related GO terms were enriched for BRAF upregulated genes. In contrast to VGP, lipid metabolism GO terms were not significantly over-represented in upregulated genes in RGP or VBRAF models (Supplementary Fig. S1).

Melanocyte tumors scavenge exogenous free fatty acids

Two gene products, LPL and FABP7, encoded by the VGP DEGs lpl (upregulated 9.5-fold exclusively in VGP) and fabp7a (upregulated 24.5-fold), are implicated in fatty acid scavenging in cancer (23). We also detected 40.5-fold upregulation of the fatty acid transporter FATP2/SLC27A2 in VGP samples, mRNA encoding a long-chain fatty acid transporter highly related to FATP1/SLC27A1, recently described as being enriched in melanoma compared with other cancer types and linked to melanoma development (9). LPL secreted by cells into the interstitial space and presented on the vasculature glycocalyx hydrolyzes dietary triglycerides transported as chylomicrons or very low density lipoprotein to release free fatty acids, which can then be transported across the plasma membrane by fatty acid transporters such as CD36, FABP7, FATP1, and FATP2. LPL and FABP7 are overexpressed in several cancers (23, 37). Increased expression of FABP7 has been observed in human melanoma, and expression correlates with aggressive tumors and poor prognosis (10–12). These findings suggest melanoma tumors increase their fatty acid scavenging to deal with an increased demand for fatty acids.

To image free fatty acid uptake by tissues in vivo, we employed [18F]-FTHA, a radiolabeled palmitic acid analogue, as a PET tracer (21). A PET imaging protocol was developed for zebrafish (see methods) that did not require modification to the PET scanner. PET scans of wild-type fish 2 hours post administration revealed [18F]-FTHA accumulation predominantly in the anterior abdomen (Fig. 2A). However, due to the limited spatial resolution, it was difficult from the PET scans alone to specify where [18F]-FTHA had been taken up. Therefore, DESI-MS imaging was carried out after PET scans, to utilize its high spatial resolution to determine a more precise location for the tracer. The concentration of FTHA used for PET (∼8 nmol/L) was, however, below the limit of detection of DESI-MS. Therefore, unlabeled FTHA was added at a higher concentration. DESI-MS imaging of wild-type fish with higher FTHA dosing (8 μmol/L) revealed that FTHA accumulated mainly in the liver (Fig. 2B), where dietary fatty acids are converted into triglycerides for incorporation into lipoproteins for distribution around the body. DESI-MS/MS was then carried out on sections to confirm that the negative-ion mode peak at m/z 305.30 was indeed FTHA, and not an endogenous species. This yielded a diagnostic ion with m/z 205.18 (Supplementary Fig. S2). Previous studies have shown that although FTHA will predominantly enter the β-oxidation pathway, it can also be incorporated into phospholipids via lipogenesis with the pathway it is used by being tissue dependent (38). DESI-MS/MS was performed for lipid peaks colocalizing with FTHA in the liver. One of these peaks (m/z 861) yielded small amounts of FTHA that had been incorporated into this lipid after 30 minutes exposure (Supplementary Fig. S3). However, identification of m/z 861 was not possible, as spectra were indicative of more than one unresolved lipid at m/z 861. Nevertheless, this finding indicates that FTHA can enter the lipogenesis pathway, at least in liver cells.

Figure 2.

PET and DESI-MS imaging of ([18F])-FTHA in wild-type zebrafish and zebrafish bearing melanoma tumors. A, [18F]-FTHA PET scans of wild-type zebrafish. B, Top, negative ion mode DESI-MS image (120-μm spatial resolution) of FTHA distribution in a wild-type zebrafish, overlaid with an optical image. Bottom, an adjacent section was stained with H&E. C, [18F]-FTHA PET scans of three tumor-bearing zebrafish. D, An optical image of one of the zebrafish that underwent PET scanning. E, DESI-MS image of FTHA distribution in the whole fish (50-μm resolution). F, DESI-MS image of FTHA distribution in the tail tumor (40-μm resolution). Images E and F have been overlaid, with an ion appearing only in the skeletal muscle (m/z 154.1). G and H, Corresponding H&E images. Arrows, tumors in the fish. This zebrafish has an obvious tumor in its tail and a tumor in its snout, which is apparent in the DESI-MS and H&E image.

Figure 2.

PET and DESI-MS imaging of ([18F])-FTHA in wild-type zebrafish and zebrafish bearing melanoma tumors. A, [18F]-FTHA PET scans of wild-type zebrafish. B, Top, negative ion mode DESI-MS image (120-μm spatial resolution) of FTHA distribution in a wild-type zebrafish, overlaid with an optical image. Bottom, an adjacent section was stained with H&E. C, [18F]-FTHA PET scans of three tumor-bearing zebrafish. D, An optical image of one of the zebrafish that underwent PET scanning. E, DESI-MS image of FTHA distribution in the whole fish (50-μm resolution). F, DESI-MS image of FTHA distribution in the tail tumor (40-μm resolution). Images E and F have been overlaid, with an ion appearing only in the skeletal muscle (m/z 154.1). G and H, Corresponding H&E images. Arrows, tumors in the fish. This zebrafish has an obvious tumor in its tail and a tumor in its snout, which is apparent in the DESI-MS and H&E image.

Close modal

Next, PET scans of V12RAS transgenic zebrafish bearing tumor nodules revealed concentration of [18F]-FTHA in 8/8 tumors (Fig. 2C and D), consistent with fatty acid uptake being elevated in V12RAS-transformed melanocytes as predicted from transcriptomics. Previously it was shown that enhanced fatty acid uptake through increased expression of FATP1 in zebrafish melanoma cells resulted in increased fatty acid storage as lipid droplets, while FATP1 expression was shown to correlate with lipid droplets in patient melanoma biopsies (9). Using Nile Red staining we detected abundant lipid droplets in V12RAS-driven melanoma nodules, again consistent with increased fatty acid uptake (Supplementary Fig. S4A and S4B). As in wild-type animals, FTHA was not detected by DESI-MS in the tumor-bearing fish using the concentrations used in PET. Therefore, fish were treated with a higher concentration of unlabeled FTHA (8 μmol/L) prior to DESI-MS imaging. In the representative example shown, FTHA localized in both snout and tail tumors as well as in the liver (Fig. 2E–H). Whereas uptake of FTHA was homogenous in the snout tumor, uptake was heterogeneous in the fin tumor (a finding that would not be apparent by PET scans alone) and did not colocalize with areas of the most densely packed cells, as may have been expected. Both PET and DESI-MS indicated that FTHA accumulation was markedly higher in tumor nodules than flat areas of skin containing only RGP melanoma, consistent with our transcriptomic analysis identifying upregulation of fatty acid scavenging genes selectively in VGP melanoma.

Glycerophospholipid metabolism is deregulated in malignant melanocyte neoplasia

UHPLC-MS metabolomics was performed on nonaqueous solvent extracts of wild-type skin, skin containing RGP melanoma, and on VGP tumor nodules excised from the body wall to determine changes in metabolite concentration during melanocyte neoplasia progression. A number of species from all lipid metabolite classes changed in abundance compared with wild-type skin, increasing with disease progression and mirroring the diversity seen in the transcriptome (see Supplementary Table S3; Supplementary Fig. S5A and S5B). PCA of metabolites detected by UHPLC-MS indicated a clear separation between wild-type, RGP, and VGP samples with a more significant separation observed for negative-ion mode data (Fig. 3A). The transcriptome and metabolome data were integrated using MetScape, which highlighted the glycerophospholipid metabolism pathway in both the VGP (Fig. 3B) and RGP (Supplementary Fig. S5C) samples. Major components of the cell membrane include phosphatidyl choline (PC) and phosphatidyl ethanolamine (PE) species, which increased in relative abundance in zebrafish V12RAS-driven melanocyte neoplasia, beginning in the RGP model and becoming more prominent in the VGP model. This change was also accompanied by a depletion of phosphatidylserine (PS), potentially driven by increased expression of phosphatidylserine decarboxylase (PISD), as revealed by the transcriptome analysis (upregulated 3.7-fold exclusively in VGP), which catalyzes the formation of PE by decarboxylation of PS. DESI-MS was carried out to visualize phospholipid species in VGP melanoma tumors. As well as revealing clear differences in glycerophospholipid abundance between the tumor and healthy tissue, imaging revealed heterogeneity of glycerophospholipid species, including PE species in the tumor (Fig. 3C and D; Supplementary Fig. S6A and S6B).

Figure 3.

Altered glycerophospholipid metabolism accompanies melanocyte neoplasia progression. A, PCA analysis from UPLC-MS negative and positive ion modes demonstrates the significant difference between samples based on separation by principal component 1 and 2 (PC1 and PC2). QC refers to a pooled biological quality control sample run before and every fifth injection during analysis. B, Integration of lipid metabolite and lipid metabolism genes for tumor nodules (VGP) using Metscape reveals glycerophospholipid metabolism to be the most enriched network. Metabolites are indicated as hexagons and genes as circles. A darker color demonstrates a greater degree of deregulation, with the relative size of the symbol indicating up or downregulation (see also http://ndexbio.org/#/network/e22027c9-6dbb-11e8-a4bf-0ac135e8bacf?accesskey=679ae449f48ae05d9116df1983f201390c539b285cc525ab149f2e178e55717e). C, Negative ion mode DESI-MS images of the whole zebrafish, including an overlay with H&E stain where tumors have been circled (MS/MS spectra for m/z 720.50 and 734.53 species shown in Supplementary Fig. S6A and S6B, respectively). D, Positive ion mode DESI-MS images of unidentified lipids in the tail tumor with corresponding H&E stain (left).

Figure 3.

Altered glycerophospholipid metabolism accompanies melanocyte neoplasia progression. A, PCA analysis from UPLC-MS negative and positive ion modes demonstrates the significant difference between samples based on separation by principal component 1 and 2 (PC1 and PC2). QC refers to a pooled biological quality control sample run before and every fifth injection during analysis. B, Integration of lipid metabolite and lipid metabolism genes for tumor nodules (VGP) using Metscape reveals glycerophospholipid metabolism to be the most enriched network. Metabolites are indicated as hexagons and genes as circles. A darker color demonstrates a greater degree of deregulation, with the relative size of the symbol indicating up or downregulation (see also http://ndexbio.org/#/network/e22027c9-6dbb-11e8-a4bf-0ac135e8bacf?accesskey=679ae449f48ae05d9116df1983f201390c539b285cc525ab149f2e178e55717e). C, Negative ion mode DESI-MS images of the whole zebrafish, including an overlay with H&E stain where tumors have been circled (MS/MS spectra for m/z 720.50 and 734.53 species shown in Supplementary Fig. S6A and S6B, respectively). D, Positive ion mode DESI-MS images of unidentified lipids in the tail tumor with corresponding H&E stain (left).

Close modal

LPL is a potential therapeutic target, especially when de novo fatty acid synthesis is not enhanced by FASN upregulation

To evaluate the potential for fatty acid scavenging in human melanoma, we quantified expression of LPL, CD36, FABP7, FATP1, and FATP2 mRNA by reverse-transcription coupled RT-PCR in a panel of established human melanoma cell lines (Supplementary Table S4) and contrasted this to levels in NHMs. This revealed elevated LPL transcript in all the melanoma cell lines, both NRAS and BRAF mutant, across a wide range of fold changes compared with NHM (Fig 4A). Elevation of LPL mRNA reached more than 40-fold, whereas fewer instances of significant mRNA upregulation for the fatty acid transporters were observed, and fold change was only ever modest (less than 5-fold). Western blotting analysis for LPL on protein lysates extracted from melanoma cell lines in this panel revealed elevated protein compared with melanocytes consistent with the extent of mRNA upregulation (Supplementary Fig. S7A), increased LPL expression was also confirmed by immunofluorescence staining for LPL (Supplementary Fig. S7B). Moreover, positive IHC staining revealed LPL expression in 90% or more of primary and metastatic melanoma tumors (Fig. 4B). LPL was also detected in 77% of naevi, but was largely restricted to nests of naevocytes occupying the junction between the epidermis and the dermis (Fig. 4C). LPL staining appeared specific, as outside the tumor a positive signal was detected, as expected (39), only in adipocytes and capillary lumens (Supplementary Fig. S7C). There were significantly more examples of strong LPL staining (++ and +++) in melanoma cases (primary and metastatic) compared with naevi (P = 0.005, χ2 test).

Figure 4.

LPL expression in human melanoma cell lines and tumors. A, RT-PCR quantitation for LPL, CD36, FATP1, FATP2, and FABP7 mRNA expression in a panel of melanoma cell lines relative to NHMs. B, Positive IHC signal (pink/red) reveals LPL expression in a panel of naevi, primary tumors, and metastases. −, absent staining; +, weak staining; ++, strong heterogeneous staining; +++, strong homogeneous staining. C, Representative image illustrating LPL expression (arrowheads) in nests of naevocytes at the epidermal/dermal junction in a compound naevus.

Figure 4.

LPL expression in human melanoma cell lines and tumors. A, RT-PCR quantitation for LPL, CD36, FATP1, FATP2, and FABP7 mRNA expression in a panel of melanoma cell lines relative to NHMs. B, Positive IHC signal (pink/red) reveals LPL expression in a panel of naevi, primary tumors, and metastases. −, absent staining; +, weak staining; ++, strong heterogeneous staining; +++, strong homogeneous staining. C, Representative image illustrating LPL expression (arrowheads) in nests of naevocytes at the epidermal/dermal junction in a compound naevus.

Close modal

By investigating mRNA expression for LPL and the fatty acid transporters listed above in cutaneous melanoma cases collected and analyzed by the Cancer Genome Atlas network (40), we found no significant effect from NRAS and BRAF mutation status (Supplementary Fig. S8A). LPL mRNA expression was most highly correlated with CD36 mRNA followed by FABP7 mRNA, and there was no correlation between LPL expression and either FATP1 or FATP2 mRNA expression (Supplementary Fig. S8B). Furthermore, although fatty acid scavenging was shown to enhance metastasis of melanoma cells (13), mRNA, for LPL and the fatty acid transporters listed above, was not more abundant in metastatic samples in the TCGA collection than primary lesions (Supplementary Fig. S8C), and there was no difference evident in LPL expression in metastases versus primary from IHC (Fig. 4B; P = 0.91, χ2 test). Therefore enhanced fatty acid scavenging is most likely a common feature of both primary and secondary tumors.

Upregulation of LPL mRNA was only detected in VGP zebrafish melanocyte neoplasia (not in RGP or benign neoplasia models). To determine whether in this V12RAS-driven model upregulation of LPL is sufficient to enhance progression from RGP to VGP, we turned to a transient transgenesis assay originally devised to detect cooperation between oncogenic BRAF and other candidate melanoma oncogenes (25). The V12RAS transgene was crossed into a mitfa-mutant background to suppress melanocyte development. One-cell stage eggs from these animals were then injected with a plasmid containing a wild-type mitfa minigene to rescue melanocyte development while expressing a second mitfa-promoter–driven cassette encoding either mCherry (as a negative control), LPL, or CCND1 (a positive control; Supplementary Fig. S9A and S9B). Tumor nodule formation was then followed in animals where pigmentation had been nearly completely restored while embryos. All pigment-rescued animals developed the characteristic RGP appearance of the parent V12RAS line (Fig. 5A). Compared with mCherry, however, coexpression of LPL accelerated tumor nodule formation comparable with CCND1 (Fig. 5B). Moreover, coexpression of LPL increased tumor growth rate compared with either mCherry or CCND1 (Fig. 5C). Using the same transient transgenesis approach, a weaker but still significant cooperation was also observed with NRASG12D (Supplementary Fig. S9C and S9D).

Figure 5.

LPL contributes to melanoma cell growth. A, A transgenic system for coexpression of LPL and V12RAS. Top, uninjected V12RAS; mitfa−/− animal; bottom, V12RAS; mitfa−/− animal injected with mitfa-encoding plasmid displaying complete pigment rescue and a tumor nodule (arrow). B, LPL coexpression accelerates tumor nodule appearance to the same extent as CCND1. ****, P < 0.0001, log-rank (Mantel–Cox) test. C, LPL coexpression accelerates tumor nodule growth. **, P < 0.01, Kruskal–Wallis test. D, Knockdown of LPL using siRNA suppresses cell growth. **, P < 0.01; ***, P < 0.001, ANOVA. E, Expression of FASN mRNA in a panel of melanoma cells relative to NHMs detected by RT-PCR. F, Cotreatment of cells with GSK264220A (LPLi) and C75 (FASNi).

Figure 5.

LPL contributes to melanoma cell growth. A, A transgenic system for coexpression of LPL and V12RAS. Top, uninjected V12RAS; mitfa−/− animal; bottom, V12RAS; mitfa−/− animal injected with mitfa-encoding plasmid displaying complete pigment rescue and a tumor nodule (arrow). B, LPL coexpression accelerates tumor nodule appearance to the same extent as CCND1. ****, P < 0.0001, log-rank (Mantel–Cox) test. C, LPL coexpression accelerates tumor nodule growth. **, P < 0.01, Kruskal–Wallis test. D, Knockdown of LPL using siRNA suppresses cell growth. **, P < 0.01; ***, P < 0.001, ANOVA. E, Expression of FASN mRNA in a panel of melanoma cells relative to NHMs detected by RT-PCR. F, Cotreatment of cells with GSK264220A (LPLi) and C75 (FASNi).

Close modal

Turning to human melanoma cells, knockdown of LPL using two independent siRNA oligonucleotides reduced cell growth by as much as 70% in WM852 cells and 50% in WM266-4 cells, but did not affect A375 cells (Fig. 5D). To address the difference in sensitivity to suppressing LPL in melanoma cells, we evaluated expression of FASN required for de novo fatty acid synthesis by reverse transcription coupled RT-PCR and Western blotting analysis. FASN mRNA and protein levels were significantly higher in A375 cells and WM266-4 cells relative to primary melanocytes but also compared with WM852 cells (Fig. 5E; Supplementary Fig. S9E), implying de novo fatty acid synthesis may complement fatty acid scavenging in A375 and WM266-4 cells but not WM852 cells. WM852 cells were also more sensitive to a small-molecule inhibitor of LPL (GSK264220A; ref. 41) than either A375 cells or WM266-4 cells (Fig. 5F). Significantly, combined treatment with FASN inhibitor C75 (42) and LPL inhibitor synergized to suppress the growth of A375 cells and WM266-4 cells but not WM852 cells (Fig. 5F). Moreover, FASN expression is relatively low in WM1361 cells and high in WM35 cells (Fig. 5E; Supplementary Fig. S9E) and consistent with predictions from above WM1361 cells were sensitive to LPL inhibitor with no added effect of FASN inhibitor, while WM35 were less sensitive to LPL inhibitor but could be sensitized by addition of FASN inhibitor (Fig. 5F).

Fatty acids are used by cells for both energy production (β-oxidation) and lipogenesis to form membrane phospholipids. As cancer cells have higher energy demands and proliferate at higher rates than healthy cells, increased acquisition of fatty acids can be expected. However, tumors can acquire fatty acids by different mechanisms: uptake from outside the cell or de novo synthesis (5). VGP upregulated DEGs included lpl, fabp7a, and fatp2/slc27a2, whose gene products are implicated in the uptake of exogenous fatty acid. Kamphorst and colleagues have shown that oncogenic RAS-driven tumors source fatty acids by scavenging them as lysophospholipids (43), demonstrating also that RAS-driven tumors are dependent on fatty acid scavenging. It has also recently been reported that the upregulation of the scavenger fatty acid receptor, CD36, is correlated to metastases and poor prognosis in melanoma and is a marker of tumor-initiating cells (13). Of interest, culturing melanoma cells with adipocytes stimulates melanoma cell proliferation and this effect could be reproduced with palmitate alone. Furthermore, the transfer of fatty acid from adipocytes to melanoma cells could be directly visualized (8). Indeed, proximity to adipose has been demonstrated to facilitate the development of zebrafish melanocyte neoplasia through the supply of exogenous fatty acid and this could be inhibited using a FATP inhibitor (9). It is notable that relative to normal skin we observed reduced triglyceride levels in both RGP and VGP V12RAS–driven melanocyte neoplasia, in keeping with depletion of adipose lipid reserves.

Tumor cells can also synthesize fatty acids de novo, and FASN and SCD-1 are crucial enzymes in this pathway. FASN synthesizes palmitate from acetyl-CoA and malonyl-CoA, and palmitate can undergo subsequent changes such as elongation and desaturation. SCD-1 generates unsaturated fatty acids such as oleic acid (18:1) and palmitoleic acid (16:1), by inserting a double bond into a saturated fatty acid at the Δ9 position. FASN and SCD-1 have been found to be overexpressed in many cancers including melanoma (3, 4, 44), implying that de novo fatty acid production is also increased. Scda (6.5-fold) and -b (4.1-fold) overexpression but not fasn overexpression was also observed in zebrafish V12RAS-driven VGP melanocyte neoplasia. Because SCD1 could just as well desaturate fatty acids that are scavenged as synthesized de novo, its upregulation here does not necessarily indicate increased de novo synthesis. Cells can also break down phospholipids to generate free fatty acids. Expression of several PLA2 isoforms (pla2g6, pla2g7, and pla2g10) appeared to be upregulated (2.0–5.3 fold) in zebrafish V12RAS-driven VGP melanocyte neoplasia. PLA2 enzymes are a diverse family of phospholipases that hydrolyze phospholipids to generate free fatty acids (commonly arachidonic acid) and lysophospholipids. Lysophospholipids and arachidonic acid are both signaling molecules, with arachidonic acid being a precursor of several bioactive eicosanoids, which in turn are involved in inflammatory mechanisms. PLA2 deregulation in cancer is increasingly recognized (45) with the PLA2G6 gene being associated with heightened risk of malignancy developing in naevi (46, 47), and the PLA2 family being investigated as a source of potential drug targets (45).

PET and DESI-MS imaging using the palmitic acid analogue FTHA were used as complementary imaging modalities to reveal enhanced fatty acid scavenging in zebrafish melanocyte neoplasia, implied by the transcriptomic data. Melanocyte tumors may be scavenging palmitate (and hence FTHA) to be used in β-oxidation or in lipogenesis. β-oxidation provides an alternative energy source to glucose and has been linked with the metastatic phenotype of melanoma cell lines (48). Lipogenesis rates will also be increased in cancer, and so it is plausible that FTHA may also be incorporated into more complex lipids. Indeed, a previous study found that exogenous palmitate was incorporated into glycerophospholipids, sphingolipids, and ether lipids in a range of human cancer cells (49). However, no lipids were found that colocalized exactly with the FTHA distribution in tumors, making targeted MS/MS experiments difficult. It is possible that administering the FTHA for a longer period of time would give rise to more lipids containing FTHA as a fatty acid chain.

To understand changes to lipid metabolism more precisely, transcriptome data were integrated with UHPLC-MS metabolome data. This integrated approach, further corroborated by DESI-MS imaging, highlighted the glycerophospholipid metabolism pathway, which is critical to the synthesis of plasma membrane phospholipids, playing a role in melanocyte neoplasia progression. Aberrant glycerophospholipid metabolism has previously been implicated in melanoma metastatic potential (50). Phospholipids, including PC and PE species, were upregulated in VGP tumors compared with the surrounding healthy tissue, while PS species were downregulated. Of note, the mitochondrial protease lactamase beta (LACTB) has been shown to exert its tumor-suppressive function through regulating the stability of PISD, which we observed to be upregulated in zebrafish V12RAS-driven melanocyte VGP neoplasia. Loss of LACTB expression in oncogene-expressing breast epithelial cells resulted in elevation of PISD protein accompanied by a conversion of mitochondrial PS to PE and lysoPE that was necessary for cancer cell proliferation and dedifferentiation (51). In addition, lipidomics comparing lipid species in human and zebrafish BRAF-mutant melanoma cells cocultured with adipocytes to those cultured alone, or those recovered from tumors in vivo contrasted to those cultured in vitro also revealed increased PC and PE species (9), indicating that our V12RAS-driven zebrafish melanocyte malignancies resemble oncogenic BRAF-driven malignant melanocytes stimulated by adipocytes. Choline-containing biomolecules are increased in human primary and metastatic melanoma when measured by MRI (52, 53) as in other cancers such as prostate cancer (54). Indeed, altered anabolism but also catabolism of PC (to produce important signaling molecules such as diacylglyceride) is now considered a pervasive feature of cancer cell metabolism and intracellular signaling (55). Phospholipase D1, upregulated (2.1-fold) in zebrafish V12RAS-driven VGP melanocyte neoplasia, which removes the choline head group, is essential for HRASG12V-mediated transformation and tumor formation (56) and is upregulated also in human melanoma (57). Thus, choline metabolism too emerges as a potential target for therapeutic intervention in melanoma.

Although both the DESI-MS and UHPLC-MS use electrospray as the ionization source, different solvents were used for each experiment therefore giving rise to different adducts. In this respect, both similar and different lipid species were expected to be detected between the two techniques. Fig. 3C shows DESI-MS images of two of the species also seen upregulated in tumors in the UHPLC-MS analysis (m/z 714.52 and 776.54) as well as two PEs that were not detected by UHPLC-MS. The identified species both contain a monounsaturated fatty acid, indicating the action of SCD-1, which was revealed to be overexpressed in the transcriptome analysis. UHPLC-MS is able to give a more quantitative analysis compared with DESI-MS, but UHPLC-MS cannot give information regarding spatial distribution. Interestingly, DESI-MS imaging revealed heterogeneous distributions of phospholipid species in VGP tumors. Positive-ion mode images revealed lipids with interesting heterogeneity, and are potential PCs, based on database searches using accurate mass. However, identification of these lipids is not trivial due to multiple ionization mechanisms (i.e., the addition of a proton, sodium, or potassium), and therefore have not been identified in this instance. It is clear from this study that lipid metabolism is not a uniform process across the whole tumor, and further studies will be needed to elucidate the reasons behind such complexity in melanoma metabolism.

RT-PCR, Western blotting, and IHC revealed LPL expression in a majority of melanoma samples, albeit to varying amounts and with varying fractions of positive cells. We hypothesize that at one extreme (represented in culture by WM852 and WM1361 cells and potentially in vivo by tumors with homogeneous strong LPL expression), melanoma cells depend entirely on LPL for fatty acid supply; while at the other end of the spectrum tumor cells are largely independent of LPL for fatty acid provision and may even not express LPL (a few primary tumors were identified seemingly lacking LPL expression). These latter cells (or even whole tumors) may rely entirely on de novo synthesis, or may express alternative lipases to carry out lipolysis. In between are cell populations (represented in culture by WM266-4, WM-35, and A375 cells) that express LPL for lipolysis and also undertake fatty acid synthesis, either representing a heterogeneous mixture of cell types, or with cells possessing both capabilities simultaneously. Coincident lipolysis and fatty acid synthesis have previously been described in breast carcinoma, liposarcoma, and prostate carcinoma cells (58). Proximity to adipose and the availability of exogenous fatty acid have been shown to decrease dependency on lipogenesis in melanoma cells, concomitant with decreased sensitivity to FASN inhibition (9). Given the capacity of melanoma cells (and other cancer cells) to profit from both lipolytic and lipogenic mechanisms, it is desirable to explore strategies to target simultaneously both fatty acid scavenging and de novo synthesis. We demonstrated here the potential of an LPL small-molecule inhibitor in combination with FASN inhibitor; lipid transporters such as FATP family members, FABP7, and CD36 might also make for appropriate targets.

In conclusion, zebrafish V12RAS-driven melanocyte tumors were seen to undergo substantial changes to their lipid metabolism. Multimodality imaging combined with lipidome and transcriptome analyses have shown that zebrafish V12RAS-driven melanocyte tumors scavenge free fatty acids and upregulate the production of certain glycerophospholipids. Key lipid metabolism pathways are well-conserved between zebrafish and human and several overlaps between altered lipid metabolism genes in the zebrafish melanocyte neoplasia models studied here and human melanoma have been highlighted, supporting the preclinical significance of our zebrafish models. This study demonstrated the complementarity of PET and DESI-MS as in vivo and ex vivo imaging modalities, respectively, which differ in spatial resolution as well as their targeted nature. Potentially, detection of fatty acid uptake by PET might be used to locate tumors in patients, as well as to stratify patients for treatment with inhibitors of fatty acid scavenging, or to monitor therapeutic intervention. The feasibility of [18F]-FTHA PET in humans has already been demonstrated (59), albeit not yet for tumor imaging. UHPLC-MS and DESI-MS could be used on biopsy samples, and so also have the potential to inform treatments in the clinic. Combining lipidomics and transcriptomics reveals a more complete picture of altered tumor lipid metabolism than either alone. While deregulated lipid metabolism pathways are promising targets for future therapeutic intervention, the heterogeneous nature in which lipid metabolism is altered must also be taken into consideration and combinatorial approaches are likely to be more effective than monotherapies.

P. Lorigan reports receiving a commercial research grant and other commercial research support from BMS; has received speakers bureau honoraria from BMS, MSD, and Novartis; and is a consultant/advisory board member for BMS, MSD, Pierre Fabre, and Novartis. No potential conflicts of interest were disclosed by the other authors.

Conception and design: F. Henderson, H.R. Johnston, A.P. Badrock, H.P. Spaink, M.P. Smith, P. Lorigan, K.J. Williams, A.W. McMahon, A. Hurlstone

Development of methodology: F. Henderson, H.R. Johnston, D. Forster, R.T. Nagaraju, J. Kamarashev, I. B.-R. Ramirez, H.P. Spaink, A.W. McMahon, A. Hurlstone

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): F. Henderson, H.R. Johnston, A.P. Badrock, E.A. Jones, D. Forster, R.T. Nagaraju, J. Kamarashev, M. Fairclough, I. B.-R. Ramirez, S. He, B.E. Snaar-Jagalska, K. Hollywood, W.B. Dunn, H.P. Spaink, M.P. Smith, E. Claude, K.J. Williams

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): F. Henderson, H.R. Johnston, A.P. Badrock, E.A. Jones, D. Forster, R.T. Nagaraju, C. Evangelou, J. Kamarashev, I. B.-R. Ramirez, S. He, K. Hollywood, W.B. Dunn, H.P. Spaink, M.P. Smith, E. Claude, K.J. Williams, A.W. McMahon, A. Hurlstone

Writing, review, and/or revision of the manuscript: F. Henderson, H.R. Johnston, A.P. Badrock, E.A. Jones, C. Evangelou, J. Kamarashev, K. Hollywood, W.B. Dunn, H.P. Spaink, P. Lorigan, K.J. Williams, A.W. McMahon, A. Hurlstone

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M. Green, M. Fairclough, S. He, H.P. Spaink

Study supervision: P. Lorigan, K.J. Williams, A.W. McMahon, A. Hurlstone

The authors like to thank Paul Begley for his technical support. The work was supported by the European Research Council (grant no. ERC-2011-StG-282059 PROMINENT to A. Hurlstone) and by the Cancer Research UK (CRUK) and Engineering and Physical Sciences Research Council (EPSRC) Manchester and Cambridge Cancer Imaging Centre (grant no. C8742/A18097 to K.J. Williams and A.W. McMahon). H.R. Johnston was funded by a BBSRC Ph.D studentship and Christie Charity Award. F. Henderson was funded by a BBSRC PhD studentship.

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.

1.
Beloribi-Djefaflia
S
,
Vasseur
S
,
Guillaumond
F
. 
Lipid metabolic reprogramming in cancer cells
.
Oncogenesis
2016
;
5
:
e189
.
2.
Gniadecka
M
,
Philipsen
PA
,
Sigurdsson
S
,
Wessel
S
,
Nielsen
OF
,
Christensen
DH
, et al
Melanoma diagnosis by Raman spectroscopy and neural networks: structure alterations in proteins and lipids in intact cancer tissue
.
J Invest Dermatol
2004
;
122
:
443
9
.
3.
Innocenzi
D
,
Alo
PL
,
Balzani
A
,
Sebastiani
V
,
Silipo
V
,
La Torre
G
, et al
Fatty acid synthase expression in melanoma
.
J Cutan Pathol
2003
;
30
:
23
8
.
4.
Kapur
P
,
Rakheja
D
,
Roy
LC
,
Hoang
MP
. 
Fatty acid synthase expression in cutaneous melanocytic neoplasms
.
Mod Pathol
2005
;
18
:
1107
12
.
5.
Rohrig
F
,
Schulze
A.
The multifaceted roles of fatty acid synthesis in cancer
.
Nat Rev Cancer
2016
;
16
:
732
49
.
6.
Carvalho
MA
,
Zecchin
KG
,
Seguin
F
,
Bastos
DC
,
Agostini
M
,
Rangel
AL
, et al
Fatty acid synthase inhibition with Orlistat promotes apoptosis and reduces cell growth and lymph node metastasis in a mouse melanoma model
.
Int J Cancer
2008
;
123
:
2557
65
.
7.
Zecchin
KG
,
Rossato
FA
,
Raposo
HF
,
Melo
DR
,
Alberici
LC
,
Oliveira
HC
, et al
Inhibition of fatty acid synthase in melanoma cells activates the intrinsic pathway of apoptosis
.
Lab Invest
2011
;
91
:
232
40
.
8.
Kwan
HY
,
Fu
X
,
Liu
B
,
Chao
X
,
Chan
CL
,
Cao
H
, et al
Subcutaneous adipocytes promote melanoma cell growth by activating the Akt signaling pathway: role of palmitic acid
.
J Biol Chem
2014
;
289
:
30525
37
.
9.
Zhang
M
,
Martino
JSD
,
Bowman
RL
,
Campbell
NR
,
Baksh
SC
,
Simon-Vermot
T
, et al
Adipocyte-derived lipids mediate melanoma progression via FATP Proteins
.
Cancer Discov
2018
;
8
:
1006
25
.
10.
Goto
Y
,
Koyanagi
K
,
Narita
N
,
Kawakami
Y
,
Takata
M
,
Uchiyama
A
, et al
Aberrant fatty acid-binding protein-7 gene expression in cutaneous malignant melanoma
.
J Invest Dermatol
2010
;
130
:
221
9
.
11.
Slipicevic
A
,
Jørgensen
K
,
Skrede
M
,
Rosnes
AK
,
Trøen
G
,
Davidson
B
, et al
The fatty acid binding protein 7 (FABP7) is involved in proliferation and invasion of melanoma cells
.
BMC Cancer
2008
;
8
:
276
.
12.
Goto
Y
,
Matsuzaki
Y
,
Kurihara
S
,
Shimizu
A
,
Okada
T
,
Yamamoto
K
, et al
A new melanoma antigen fatty acid-binding protein 7, involved in proliferation and invasion, is a potential target for immunotherapy and molecular target therapy
.
Cancer Res
2006
;
66
:
4443
9
.
13.
Pascual
G
,
Avgustinova
A
,
Mejetta
S
,
Martin
M
,
Castellanos
A
,
Attolini
CS
, et al
Targeting metastasis-initiating cells through the fatty acid receptor CD36
.
Nature
2017
;
541
:
41
5
.
14.
van der Weyden
L
,
Patton
EE
,
Wood
GA
,
Foote
AK
,
Brenn
T
,
Arends
MJ
, et al
Cross-species models of human melanoma
.
J Pathol
2016
;
238
:
152
65
.
15.
Michailidou
C
,
Jones
M
,
Walker
P
,
Kamarashev
J
,
Kelly
A
,
Hurlstone
AF
. 
Dissecting the roles of Raf- and PI3K-signalling pathways in melanoma formation and progression in a zebrafish model
.
Dis Model Mech
2009
;
2
:
399
411
.
16.
Dalton
LE
,
Kamarashev
J
,
Barinaga-Rementeria Ramirez
I
,
White
G
,
Malliri
A
,
Hurlstone
A
. 
Constitutive RAC activation is not sufficient to initiate melanocyte neoplasia but accelerates malignant progression
.
J Invest Dermatol
2013
;
133
:
1572
81
.
17.
Hara
T
,
Inagaki
K
,
Kosaka
N
,
Morita
T
. 
Sensitive detection of mediastinal lymph node metastasis of lung cancer with 11C-choline PET
.
J Nucl Med
2000
;
41
:
1507
13
.
18.
Hara
T
,
Kosaka
N
,
Kishi
H
. 
PET imaging of prostate cancer using carbon-11-choline
.
J Nucl Med
1998
;
39
:
990
5
.
19.
Oyama
N
,
Akino
H
,
Kanamaru
H
,
Suzuki
Y
,
Muramoto
S
,
Yonekura
Y
, et al
11C-acetate PET imaging of prostate cancer
.
J Nucl Med
2002
;
43
:
181
6
.
20.
Tsuchida
T
,
Takeuchi
H
,
Okazawa
H
,
Tsujikawa
T
,
Fujibayashi
Y
. 
Grading of brain glioma with 1–11C-acetate PET: comparison with 18F-FDG PET
.
Nucl Med Biol
2008
;
35
:
171
6
.
21.
DeGrado
TR
,
Wang
S
,
Holden
JE
,
Nickles
RJ
,
Taylor
M
,
Stone
CK
. 
Synthesis and preliminary evaluation of (18)F-labeled 4-thia palmitate as a PET tracer of myocardial fatty acid oxidation
.
Nucl Med Biol
2000
;
27
:
221
31
.
22.
Takats
Z
,
Wiseman
JM
,
Gologan
B
,
Cooks
RG
. 
Mass spectrometry sampling under ambient conditions with desorption electrospray ionization
.
Science
2004
;
306
:
471
3
.
23.
Balaban
S
,
Lee
LS
,
Schreuder
M
,
Hoy
AJ
. 
Obesity and cancer progression: is there a role of fatty acid metabolism?
BioMed Res Int
2015
;
2015
:
17
.
24.
Casar
B
,
Badrock
AP
,
Jimenez
I
,
Arozarena
I
,
Colon-Bolea
P
,
Lorenzo-Martin
LF
, et al
RAS at the golgi antagonizes malignant transformation through PTPRkappa-mediated inhibition of ERK activation
.
Nat Commun
2018
;
9
:
3595
.
25.
Ceol
CJ
,
Houvras
Y
,
Jane-Valbuena
J
,
Bilodeau
S
,
Orlando
DA
,
Battisti
V
, et al
The histone methyltransferase SETDB1 is recurrently amplified in melanoma and accelerates its onset
.
Nature
2011
;
471
:
513
7
.
26.
Stockhammer
OW
,
Zakrzewska
A
,
Hegedus
Z
,
Spaink
HP
,
Meijer
AH
. 
Transcriptome profiling and functional analyses of the zebrafish embryonic innate immune response to Salmonella infection
.
J Immunol
2009
;
182
:
5641
53
.
27.
Mudunuri
U
,
Che
A
,
Yi
M
,
Stephens
RM
. 
bioDBnet: the biological database network
.
Bioinformatics
2009
;
25
:
555
6
.
28.
Kinsella
RJ
,
Kähäri
A
,
Haider
S
,
Zamora
J
,
Proctor
G
,
Spudich
G
, et al
Ensembl BioMarts: a hub for data retrieval across taxonomic space
.
Database
2011
;
2011
:
bar030
.
29.
The Gene Ontology Consortium
. 
Expansion of the gene ontology knowledgebase and resources
.
Nucleic Acids Res
2017
;
45
:
D331
D8
.
30.
Degrado
TR.
Synthesis of 14(R,S)-[F-18]fluoro-6-thia-heptadecanoic acid (FTHA)
.
J Labelled Compounds Radiopharm
1991
;
29
:
989
95
.
31.
Sud
M
,
Fahy
E
,
Cotter
D
,
Brown
A
,
Dennis
EA
,
Glass
CK
, et al
LMSD: LIPID MAPS structure database
.
Nucleic Acids Res
2007
;
35
:
D527
D32
.
32.
Brown
M
,
Wedge
DC
,
Goodacre
R
,
Kell
DB
,
Baker
PN
,
Kenny
LC
, et al
Automated workflows for accurate mass-based putative metabolite identification in LC/MS-derived metabolomic datasets
.
Bioinformatics
2011
;
27
:
1108
12
.
33.
Sumner
LW
,
Amberg
A
,
Barrett
D
,
Beale
MH
,
Beger
R
,
Daykin
CA
, et al
Proposed minimum reporting standards for chemical analysis Chemical Analysis Working Group (CAWG) Metabolomics Standards Initiative (MSI)
.
Metabolomics
2007
;
3
:
211
21
.
34.
Karnovsky
A
,
Weymouth
T
,
Hull
T
,
Tarcea
VG
,
Scardoni
G
,
Laudanna
C
, et al
Metscape 2 bioinformatics tool for the analysis and visualization of metabolomics and gene expression data
.
Bioinformatics
2011
;
28
:
373
80
.
35.
Rose
AE
,
Poliseno
L
,
Wang
J
,
Clark
M
,
Pearlman
A
,
Wang
G
, et al
Integrative genomics identifies molecular alterations that challenge the linear model of melanoma progression
.
Cancer Res
2011
;
71
:
2561
71
.
36.
Chi
JT
,
Wang
Z
,
Nuyten
DS
,
Rodriguez
EH
,
Schaner
ME
,
Salim
A
, et al
Gene expression programs in response to hypoxia: cell type specificity and prognostic significance in human cancers
.
PLoS Med
2006
;
3
:
e47
.
37.
Zaidi
N
,
Lupien
L
,
Kuemmerle
NB
,
Kinlaw
WB
,
Swinnen
JV
,
Smans
K
. 
Lipogenesis and lipolysis: the pathways exploited by the cancer cells to acquire fatty acids
.
Prog Lipid Res
2013
;
52
:
585
9
.
38.
Takala
TO
,
Nuutila
P
,
Pulkki
K
,
Oikonen
V
,
Gronroos
T
,
Savunen
T
, et al
14(R,S)-[18F]fluoro-6-thia-heptadecanoic acid as a tracer of free fatty acid uptake and oxidation in myocardium and skeletal muscle
.
Eur J Nucl Med Mol Imaging
2002
;
29
:
1617
22
.
39.
Davies
BSJ
,
Beigneux
AP
,
Barnes
RH
,
Tu
Y
,
Gin
P
,
Weinstein
MM
, et al
GPIHBP1 is responsible for the entry of lipoprotein lipase into capillaries
.
Cell Metabolism
2010
;
12
:
42
52
.
40.
Cancer Genome Atlas Network.
Genomic classification of cutaneous melanoma
.
Cell
2015
;
161
:
1681
96
.
41.
Keller
PM
,
Rust
T
,
Murphy
DJ
,
Matico
R
,
Trill
JJ
,
Krawiec
JA
, et al
A high-throughput screen for endothelial lipase using HDL as substrate
.
J Biomol Screen
2008
;
13
:
468
75
.
42.
Kuhajda
FP
,
Pizer
ES
,
Li
JN
,
Mani
NS
,
Frehywot
GL
,
Townsend
CA
. 
Synthesis and antitumor activity of an inhibitor of fatty acid synthase
.
Proc Natl Acad Sci U S A
2000
;
97
:
3450
4
.
43.
Kamphorst
JJ
,
Cross
JR
,
Fan
J
,
de Stanchina
E
,
Mathew
R
,
White
EP
, et al
Hypoxic and Ras-transformed cells support growth by scavenging unsaturated fatty acids from lysophospholipids
.
Proc Natl Acad Sci U S A
2013
;
110
:
8882
7
.
44.
von Roemeling
CA
,
Marlow
LA
,
Wei
JJ
,
Cooper
SJ
,
Caulfield
TR
,
Wu
K
, et al
Stearoyl-CoA desaturase 1 is a novel molecular therapeutic target for clear cell renal cell carcinoma
.
Clin Cancer Res
2013
;
19
:
2368
80
.
45.
Scott
KF
,
Sajinovic
M
,
Hein
J
,
Nixdorf
S
,
Galettis
P
,
Liauw
W
, et al
Emerging roles for phospholipase A2 enzymes in cancer
.
Biochimie
2010
;
92
:
601
10
.
46.
Bishop
DT
,
Demenais
F
,
Iles
MM
,
Harland
M
,
Taylor
JC
,
Corda
E
, et al
Genome-wide association study identifies three loci associated with melanoma risk
.
Nat Genet
2009
;
41
:
920
5
.
47.
Mario
F
,
Veronique
B
,
Nicholas
HK
,
David
DL
,
Julia
NBA
,
Tomi
P
, et al
Loci at 9p21 and 22q13 harbour alleles for development of cutaneous nevi and melanoma
.
Nat Genet
2009
;
41
:
915
9
.
48.
Rodrigues
MF
,
Obre
E
,
de Melo
FH
,
Santos
GC
 Jr
,
Galina
A
,
Jasiulionis
MG
, et al
Enhanced OXPHOS, glutaminolysis and beta-oxidation constitute the metastatic phenotype of melanoma cells
.
Biochem J
2016
;
473
:
703
15
.
49.
Louie
SM
,
Roberts
LS
,
Mulvihill
MM
,
Luo
K
,
Nomura
DK
. 
Cancer cells incorporate and remodel exogenous palmitate into structural and oncogenic signaling lipids
.
Biochim Biophys Acta
2013
;
1831
:
1566
72
.
50.
Kim
HY
,
Lee
H
,
Kim
SH
,
Jin
H
,
Bae
J
,
Choi
HK
. 
Discovery of potential biomarkers in human melanoma cells with different metastatic potential by metabolic and lipidomic profiling
.
Sci Rep
2017
;
7
:
8864
.
51.
Keckesova
Z
,
Donaher
JL
,
De Cock
J
,
Freinkman
E
,
Lingrell
S
,
Bachovchin
DA
, et al
LACTB is a tumour suppressor that modulates lipid metabolism and cell state
.
Nature
2017
;
543
:
681
6
.
52.
Guitera
P
,
Bourgeat
P
,
Stretch
JR
,
Scolyer
RA
,
Ourselin
S
,
Lean
C
, et al
Diagnostic value of 8.5 T magnetic resonance spectroscopy of benign and malignant skin lesion biopsies
.
Melanoma Res
2010
;
20
:
311
7
.
53.
Fedele
TA
,
Galdos-Riveros
AC
,
Jose de Farias e Melo
H
,
Magalhães
A
,
Maria
DA
. 
Prognostic relationship of metabolic profile obtained of melanoma B16F10
.
Biomed Pharmacother
2013
;
67
:
146
56
.
54.
Giskeødegård
GF
,
Hansen
AF
,
Bertilsson
H
,
Gonzalez
SV
,
Kristiansen
KA
,
Bruheim
P
, et al
Metabolic markers in blood can separate prostate cancer from benign prostatic hyperplasia
.
Br J Cancer
2015
;
113
:
1712
9
.
55.
Glunde
K
,
Bhujwalla
ZM
,
Ronen
SM
. 
Choline metabolism in malignant transformation
.
Nat Rev Cancer
2011
;
11
:
835
48
.
56.
Buchanan
FG
,
McReynolds
M
,
Couvillon
A
,
Kam
Y
,
Holla
VR
,
Dubois
RN
, et al
Requirement of phospholipase D1 activity in H-RasV12-induced transformation
.
Proc Natl Acad Sci U S A
2005
;
102
:
1638
42
.
57.
Oka
M
,
Kageshita
T
,
Ono
T
,
Goto
A
,
Kuroki
T
,
Ichihashi
M
. 
Protein kinase C alpha associates with phospholipase D1 and enhances basal phospholipase D activity in a protein phosphorylation-independent manner in human melanoma cells
.
J Invest Dermatol
2003
;
121
:
69
76
.
58.
Kuemmerle
NB
,
Rysman
E
,
Lombardo
PS
,
Flanagan
AJ
,
Lipe
BC
,
Wells
WA
, et al
Lipoprotein lipase links dietary fat to solid tumor cell proliferation
.
Mol Cancer Ther
2011
;
10
:
427
36
.
59.
Iozzo
P
,
Turpeinen
AK
,
Takala
T
,
Oikonen
V
,
Solin
O
,
Ferrannini
E
, et al
Liver uptake of free fatty acids in vivo in humans as determined with 14(R, S)-[18F]fluoro-6-thia-heptadecanoic acid and PET
.
Eur J Nucl Med Mol Imaging
2003
;
30
:
1160
4
.