Peripheral blood mononuclear cells (PBMCs) were isolated from blood of healthy volunteers by density gradient centrifugation using Ficoll-plaque (GE Healthcare, Diegem, Belgium). PBMCs were washed in cold phosphate buffered saline (PBS, Braun Melsungen, Germany) and resuspended in Roswell Park Memorial Institute (RPMI) 1640 Dutch modification culture medium (Life Technologies, Carlsbad, California, USA) supplemented with gentamycin 50 µg/mL, pyruvate 1 mM, glutamax 2 mM and counted using a Coulter particle counter (Beckman Coulter, Fullerton, California, USA). Monocytes were isolated within the PBMC fraction using Percoll as described previously.21

The coculture experiments were performed using the TC cell lines TPC-1 (papillary, RET/PTC rearrangement) and FTC-133 (follicular, PTEN deficient) cell lines22 and the NB cell lines SK-N-AS and IMR-32 (MYC-amplification). The cancer cell lines and cocultures were cultured as described previously.18 A total of 1.0×10⁶ Percoll-enriched monocytes in 500 µL were added to the lower compartment of the trans-well system in coculture medium: RPMI (Life Technologies, Carlsbad, California, USA) supplemented with glucose 5 mM, pyruvate 1 mM, glutamine 2 mM, gentamicin 50 ug/mL and HEPES 10 mM (Life Technologies, Carlsbad, California, USA). After 1-hour adhesion in the 24-well plates, non-adherent cells comprising mainly lymphocytes were discarded, and the adherent monocytes were incubated with the tumor cells in the upper compartments. Medium alone or the same number of monocytes was used as controls.

In some experiments, the FASN inhibitor C75 (2 µM; Sigma-Aldrich, St. Louis, Missouri, USA), the CD36 inhibitor Sulfo-N-succinimidyl oleate (25 µM; Sigma-Aldrich) or lactate receptor antagonist α-cyano-4-hydrocycinnamioic acid (1 mM; Sigma-Aldrich) were added to the medium.

After 24 hours of incubation, the cell culture inserts were discarded and the adherent monocytes were stimulated for 24 hours with RPMI or 10 ng/mL lipopolysaccharide (LPS) (E. coli strain O55:B5, Sigma-Aldrich). At the end of the incubation period, supernatant was collected and stored at −20°C until further analysis.

RNA isolation, deep sequencing and analysis were performed as described by Arts et al.18 The detailed data have been deposited in the GEO database with accession number GSE76445.

After coculture, RNA from monocytes/macrophages was extracted in 500 µL of TRIzol reagent (Invitrogen, Carlsbad, California, USA). RNA was reverse transcribed into complementary-DNA using oligo(dT) primers and MMLV reverse transcriptase. PCR was performed using 7300 PCR system (Applied Biosystems). The primer sequences are as follows: FASN AAGGACCTGTCTAGGTTTGATGC (forward) and TGGCTTCATAGGTGACTTCCA (reverse), ACACA TCGCTTTGGGGGAAATAAAGTG (forward) and GTGTGACCATGACAACGAATCTA (reverse), FADS2 GGATGGCTGCAACATGATTATGG (forward) and GCAGAGGCACCCTTTAAGTGG (reverse), CD206 CTACAAGGGATCGGGTTTATGGA (forward) and TTGGCATTGCCTAGTAGCGTA (reverse), CD163 TGAAGACTCTGGATCTGCTGA (forward) and GAACTGGTGACAAAACAGGCA (reverse), MerTK GCCCCATCAGTAGCACCTTT (forward) and GCTCTTTGGAAATCCCTGCAC (reverse). β2M was used as a reference gene, for which the primers were ATGAGTATGCCTGCCGTGTG (forward) and CCAAATGCGGCATCTTCAAAC (reverse). PCR conditions were as follows: 2 min at 50°C, followed by 40 cycles of PCT at 95°C for 15 s and 60°C for 1 min.

Cells were harvested by scraping in cold PBS, washed in PBS and incubated with the viability dye eFluor 780 (eBioscience, San Diego, California, USA) at room temperature for 20 min. After washing with PBS, blocking was performed with PBS containing 2% human serum (Sanquin, Nijmegen, The Netherlands) at 4°C for 10 min. Next, cells were washed with PBA (PBS containing 0.5% bovine serum albumin (Roche, Indianapolis, Indiana, USA) and 0.01% sodium azide) and incubated with the following directly conjugated monoclonal antibodies at 4°C for 20 min: CD163-APC (eBioscience), CD206-AF488 (eBioscience), CD86-PECy7 (Becton Dickinson, Franklin Lakes, New Jersey, USA) or MerTK-PECy7 (Biolegend, San Diego, California, USA), MHCII-BV421 (Becton Dickinson). After washing in PBS twice, cells were resuspended in PBA and measured on Cytoflex (Beckman Coulter) flow cytometer. Data were analyzed in Kaluza 2.1 (Beckman Coulter) where events were gated on single cells for further analysis.

Isolated cells were washed with PBS and 150 mM ammonium acetate and rapidly pelleted. To the pellet, 500 µL of extraction mix (CHCl₃/MeOH 1/5 containing internal standards: 210 pmol PE (31:1), 396 pmol PC (31:1), 98 pmol phosphatidylserine (31:1), 84 pmol phosphatidylinositol (34:0), 56 pmol phosphatidic acid (PA) (31:1), 51 pmol phosphatidylglycerol (28:0), 28 pmol cardiolipin (56:0), 39 pmol LPA (17:0), 35 pmol lysophosphatidylcholine (LPC) (17:1), 38 pmol lysophosphatidylethanolamine (LPE) (17:0), 32 pmol Cer (17:0), 99 pmol sphingomyelins (SM) (17:0), 55 pmol GlcCer (12:0), 14 pmol GM3 (18:0-D3), 359 pmol TG (47:1), 111 pmol CE (17:1), 64 pmol DG (31:1), 103 pmol MG (17:1), 724 pmol Chol(d6), 45 pmol Car(15:0)) were added and the sample was sonicated for 2 min followed by centrifugation at 20 000 g for 2 min. The supernatant was collected into a new tube and 200 µL chloroform and 800 µL 1% AcOH in water were added, the sample briefly shaken and spun for 2 min at 20 000 g. The upper aqueous phase was removed and the entire lower phase transferred into a new tube and evaporated in a speed vac (45°C, 10 min). Spray buffer (500 µL of 8/5/1 2-propanol/MeOH/water, 10 mM ammonium acetate) was added, the sample was sonicated for 5 min and infused at 10 µL/min into a Thermo Q Exactive Plus spectrometer equipped with the HESI II ion source for shotgun lipidomics. MS1 spectra (resolution 280 000) were recorded in 100 m/z windows from 250 to 1200 m/z (pos.) and 200–1700 m/z (neg.) followed by recording MS/MS spectra (res. 70000) by data independent acquisition in 1 m/z windows from 200 to 1200 (pos.) and 200–1700 (neg.) m/z. Raw files were converted to .mzml files and imported into and analyzed by LipidXplorer software using custom mfql files to identify sample lipids and internal standards. For further data processing, absolute amounts were calculated using the internal standard intensities followed by calculation of mol% of the identified lipids.

Interleukin-10 (IL-10) (Sanquin, Amsterdam, Netherlands), IL-6 and tumor necrosis factor α (TNF-α) (R&D, the Netherlands) concentrations in the cell culture supernatant were measured by commercial ELISA kits according to the instruction of the manufacturer. Lactate was measured by a Lactate Fluorometric Assay Kit (Biovision, California, USA).

TPC-1-induced macrophages were cultured as described above and Brefeldin A (1 µg/mL) was added 1 hour after LPS stimulation. After 4-hour LPS stimulation, cells were lysed. Total protein amounts were determined by BCA assay (Thermo Fisher Scientific, Massachusetts, USA). Samples were mixed and boiled at 95°C for 5 min with Laemmli buffer and Dithiothreitol. Equal amounts of proteins were loaded on precasted 4%–15% gels (Biorad, California, USA). The separated proteins were transferred to a PVDF membrane (Merck Millipore, The Netherlands), which was blocked with 5% milk. Incubation overnight at 4°C with recombinant rabbit monoclonal anti-TNF alpha antibody (abcam, Cambridge, UK) and mouse monoclonal anti-β-actin antibody (Santa Cruz Biotechnology, Texas, USA) was used to determine the protein expression, which was visualized using secondary antibodies goat-anti-rabbit IRDye 680 and goat-anti-mouse IRDye 800, respectively.

Control monocytes and tumor-induced macrophages were detached and a total of 100 000 cells were exposed to 50 ng/mL Phorbol 12-myristate 13-acetate (PMA) (Sigma-Aldrich) or 3 mg/mL of plasma-opsonized Zymosan A (Sigma-Aldrich). Reactive oxygen species (ROS) formation was detected by a chemiluminescence assay using 0.1 mM 5-amino-2,3-dihydro-1,4-phthalazinedione (luminol) (Sigma-Aldrich). The luminometer measured chemiluminescence in the integration mode at 37°C every 142 s for 1 hour after luminol had been added.

Cell viability was determined using a lactate dehydrogenase cytotoxicity assay (CytoTox 96 Non-Radioactive Cytotoxicity Assay, Promega, Minnesota, USA) according to the manufacturer’s instructions using fresh supernatants.

All analyses were performed in Graphpad Prism 5 (California, USA). Differences in gene expression levels and surface expression were analyzed using two-way ANOVA followed by Bonferroni’s multiple comparison test. Differences in cytokine production were analyzed using two-tailed paired t-test. Data are shown as means±SEM.

In a previous study, we generated genome-wide transcriptome data by RNA-sequencing of TC-induced macrophages, using the TPC-1 cell line in an indirect transwell coculture system for 24 hours. The transcriptome of control monocytes was compared with that of tumor-induced macrophages. Cells were cultured in serum-free medium with low glucose concentrations (5 mM) to avoid the effects of exogenous metabolites on intracellular metabolism. After 24 hours, both were stimulated with the toll-like receptor 4 (TLR4)-agonist LPS or medium as control for 4 or 24 hours (see experimental setup in figure 1A), simulating the release of endogenous TLR ligands from injured cells, such as dying tumor cells. As a first analysis of this dataset, we reported that mTOR-dependent aerobic glycolysis and inflammation-related pathways were enriched in the TC-induced macrophages.18

In the current study, we focus on the strong enrichment of the hallmark pathway ‘FA metabolism’, which was enriched at the 4-hour time point. Using reactome pathways to compute enrichment, we found that the pathways ‘synthesis of very long chain FAs’ and ‘peroxisomal lipid metabolism’ were enriched at the 4-hour time point, and that the lipid-related pathways ‘sphingolipid metabolism’, ‘HDL-mediated lipid transport’ and ‘phospholipid metabolism’ were enriched at the 24-hour time point (figure 1B). Interestingly, after stimulation with the TLR4-agonist LPS for 24 hours, we found even more lipid-related pathways to be enriched, mainly related to de novo biosynthesis of lipids or FAs (figure 1 and online supplemental table 1).

To validate the transcriptome data, qPCR was performed for the most significantly upregulated lipid-related genes. We confirm that key enzymes in FA biosynthesis (FASN and ACACA) and FA desaturation (FADS2) were upregulated in TC-induced (TPC-1 and FTC-133 cell line) macrophages. We also extended our study to NB-induced (SK-N-AS and IMR-32 cell line) macrophages, as NB is another tumor type known to be highly infiltrated by macrophages. Interestingly, two independent NB cell lines (SK-N-AS and IMR-32 cell line) also induced macrophages to express higher levels of ACACA and higher levels of FADS2 after LPS stimulation (figure 1D). Stimulation with LPS affected the expression levels only mildly.

These results indicate that FA synthesis is upregulated in tumor-induced macrophages at the 4-hour and 24-hour time point. After 24-hour LPS stimulation, FA synthesis is the most apparent transcriptional characteristic of tumor-induced macrophages. This transcriptional signature is present in TC-induced macrophages and observed in NB-induced macrophages.

To further characterize the tumor-induced macrophages, cell-surface expression of the macrophage polarization markers MHC-II and CD86 (M1-associated) as well as CD206, CD163 and MerTK (M2-associated) were measured by flow cytometry (figure 1E). Overall, expression levels were low, with only 5%–40% of cells showing positivity, indicating a broad heterogeneity in the cell population at this quite early timepoint (48 hours). Cell viability was 80%–90% in all conditions, as determined by a viability dye (online supplemental figure 1). Interestingly, expression of the M1 markers MHC-II and CD86 was not affected by the presence of tumor cells, whereas a significant upregulation of the M2-associated markers CD206, CD163 and MerTK was observed in both TPC-1-induced and SK-N-AS-induced macrophages (figure 1E). These findings were validated by qPCR showing an increased expression of the M2-associated markers on the transcriptional level (online supplemental figure 2). This pattern was present in both the unstimulated and the LPS-stimulated condition, but was most prominent in the unstimulated condition.

To validate the changes in the lipid metabolism identified at the level of the transcriptome, and to enable a comprehensive and complete analysis of the metabolic programming of tumor-induced macrophages, non-targeted mass-spectrometry-based metabolomics measurements, identifying 1121 metabolites, were performed on TPC-1-induced macrophages. Monocytes that were cocultured with monocytes in the upper compartment of the transwells were used as a control. Following 24 hours of coculture, cells were stimulated with medium or LPS for 24 hours (see experimental setup in figure 2A). Using KEGG pathway analysis to compute enrichment, we found that several lipid-related pathways and the pathway ‘Glycolysis’ were among the most highly enriched pathways (figure 2B). Although the p values were low, due to a relatively limited sample size (n=3), these data confirmed an increase of mainly lipid-related metabolites. Principal component analysis (PCA) based on all metabolites did not reveal a clear separation of control monocytes from tumor-induced macrophages (figure 2C), whereas a PCA based selectively on all lipid metabolites detected within the metabolomics dataset did reveal a clear separation (figure 2D). This lipid metabolite signature was most pronounced in the unstimulated (RPMI) condition (figure 2E), with 99 lipids being significantly enriched, compared with LPS-restimulated cells, where changes in lipids did not reach significance (data not shown). Interestingly, the most highly upregulated lipids in TPC-1-induced macrophages were mainly phospholipids, such as phosphatidylcholines (PC) and phosphatidylethanolamines (PE) (figure 2E). A table of all identified significantly upregulated lipid metabolites is provided in online supplemental table 2. Altogether, these data confirm that lipid metabolism is one of the key metabolic process affected in tumor-induced macrophages. Importantly, one should note that the metabolome analysis can only serve as an indication for a broader group of metabolites, after which further validation is necessary.

To validate our combined transcriptome and metabolome findings and in order to be able to identify which specific lipid classes or species are specifically enriched in tumor-induced macrophages, we performed lipidomics. A dedicated lipidome analysis with a higher coverage and specificity for lipid species was performed by mass spectrometry on TPC-1-induced macrophages and control monocytes (see experimental setup in figure 2A). Lipidomics provides molecular lipid species concentrations of several lipid classes and complete FA composition of the lipid species. We found that TPC-1-induced macrophages had a higher total lipid content, both with and without LPS stimulation (p=0.0075 after LPS stimulation) compared with control monocytes (figure 3A). Analysis of the lipid classes showed that especially concentrations of the SM (unstimulated p=0.072) and the phospholipid class PE (unstimulated p=0.014, LPS-stimulated p=0.015) were increased, whereas PA (unstimulated p=0.008, LPS-stimulated p=0.015), the lysophospholipids LPE (unstimulated p=0.036, LPS-stimulated p=0.012) and LPC (unstimulated p=0.062, LPS-stimulated p=0.011) and ceramides (Cer) (unstimulated p=0.019, LPS-stimulated p=0.045) were decreased (figure 3B). Interestingly, analysis of the chain length and saturation revealed that lipids in TPC1-induced macrophages contained significantly shorter FA chains after LPS stimulation (p=0.016) and a trend to less double bonds and thus more saturated lipids in TPC-1-induced macrophages (figure 3C). Differentially expressed lipid species, based on a principal component analysis, as well as chain length and saturation per lipid class are accessible in online supplemental figure 3. Similar findings were obtained in a second, independent experiment even with a lower cell number input (n=3) (data not shown).

Taken together, lipidome analysis confirmed upregulation of FA synthesis at the transcriptional level in tumor-induced macrophages by revealing an increase in total lipid content, with glycerophospholipids and SM being the main contributors. In parallel, the building blocks for these lipid species, such as PA, lysophospholipids and ceramide, are decreased (figure 3D). The observation that mainly phospholipid species are upregulated is in line with the metabolome analysis.

In a following set of experiments, we set out to elucidate the functional relevance of the enhanced FA biosynthesis. Cellular metabolism of macrophages is known to be crucial for their inflammatory function, and many lipid species modulate inflammation.26 27 To assess the inflammatory function of tumor-induced macrophages, we measured cytokine production of macrophages after coculture with TC (TPC-1 and FTC-133) and NB (IMR-32 and SK-N-AS) cell lines. All these tumor cell lines induced higher production of TNF-alpha, IL-6 and IL-10 in macrophages compared with control monocytes after LPS stimulation (figure 4A). One of the inflammatory metabolites known to be released by tumor cells is lactate, the end-product of glycolysis in the Warburg effect.8 This prompted us to investigate lactate concentrations in tumor-conditioned medium. TC cell lines as well as NB cell lines secrete lactate, although the NB cells secrete slightly lower amounts (figure 4B). In line with earlier findings by our group,18 blockade of the lactate receptor in the coculture setting significantly reduced cytokine response by tumor-induced macrophages, in TC-induced as well as NB-induced macrophages (figure 4A).

To investigate the role of lipid biosynthesis in the inflammatory function of TC-induced (TPC-1 cell line) macrophages, lipid biosynthesis was inhibited by the well-characterized FASN inhibitor C75, and cytokine production was assessed. Inhibition of lipid biosynthesis significantly reduced extracellular cytokine levels of TNF-alpha, IL-6 and IL-10 in the coculture setting (figure 4C,D, upper panels). In order to exclude that these findings are due to the direct effect of the inhibitor on the tumor cells and not on the macrophages, FA synthesis was inhibited in the monocytes or the tumor cells separately (see experimental setup in figure 4C, middle and lower panel, respectively). Conditioned medium from untreated tumor cells supplemented with the inhibitor also reduced cytokine responses in tumor-induced macrophages (figure 4D, middle panel). In contrast, when the inhibitor was solely added to the tumor cells and was removed before adding monocytes in the lower compartment of the coculture (figure 4C, lower panel), no reduction in cytokine levels was observed (figure 4D, lower panel). These results show that inhibition of lipid biosynthesis in the macrophages, and not in the tumor cells, affects cytokine responses of tumor-induced macrophages. Interestingly, blockade of CD36-mediated FA uptake did not alter cytokine responses (online supplemental figure 4A). At the inhibitor concentrations tested, no cytotoxic effects or lysis of the monocytes was observed (online supplemental figure 4B). Collectively, these results indicate that lipid biosynthesis in macrophages is critical for cytokine response to LPS stimulation.

Interestingly, at the transcriptional level, no obvious upregulation of cytokine production was seen in tumor-induced macrophages (after 4-hour LPS stimulation: TNF p=0.364, FC 1.149; IL6 p=0.033, FC −8.799; IL10 p=0.291, FC −1.121; after 24-hour LPS stimulation: TNF p=0.987, FC −1.116; IL6 p=0.991, FC −1.042; IL10 p=0.305, FC −1.328).18 We noted, however, that vesicle biogenesis was slightly upregulated in the tumor-induced macrophages (figure 1B). Furthermore, the identified enriched lipid species comprised mainly phospholipids that are main components of membranes. Together, these findings prompted us to investigate whether the enhanced lipid biosynthesis in tumor-induced macrophages may cause higher extracellular cytokine levels via enhanced secretion of cytokines. Therefore, intracellular cytokine levels of TPC-1-induced macrophages and control monocytes were determined by Western blot on stimulation with LPS in the presence or absence of Brefeldin A to block cytokine secretion. Interestingly, intracellular TNF-alpha levels were higher in TPC-1-induced macrophages compared with control monocytes 4 hours after stimulation with LPS including blockade of secretion by Brefeldin A (figure 5A), showing that there is an increased cytokine production in TPC-1-induced macrophages. In line with previous findings, extracellular cytokine levels were significantly higher in TPC-1-induced macrophages (figure 5B). On treatment with Brefeldin A, these differences in extracellular cytokine levels were no longer observed (figure 5B), showing that treatment with Brefeldin A successfully blocked cytokine secretion. In combination with the intracellular cytokine levels on Brefeldin A treatment, these data show that the higher extracellular cytokine levels in tumor-induced macrophages are most likely caused by (post)-transcriptional changes.

Next to cytokines, other inflammatory mediators produced by macrophages are the ROS. ROS are considered potent DNA-damaging agents, which are believed to be neoplasia-preceding events in thyroid cells and are also suggested to contribute to maintenance of genomic instability during the subsequent phases of tumorigenesis.28 29 Furthermore, in NB, ROS are associated with a more aggressive tumor phenotype.30 Therefore, we investigated whether ROS production was enhanced in tumor-induced macrophages. After coculture, tumor-induced macrophages and control monocytes were harvested to determine ROS production. To induce ROS production, cells were treated with the ROS-inducers PMA or opsonized zymosan during a 1-hour time period. On treatment with ROS-inducers, TPC-1-induced macrophages showed significantly higher ROS levels compared with control monocytes, although we do observe donor variation, with some donors showing low levels of ROS (figure 6A). Interestingly, when lipid biosynthesis was inhibited during coculture, ROS levels in TPC-1-induced macrophages and IMR-32–induced macrophages were strongly reduced to the levels of control monocytes (figure 6B). These results indicate that lipid biosynthesis is critical for the capacity of tumor-induced macrophages to produce ROS.