LAP-tTA and TRE-MYC mice were previously described and MYC expression in the liver was activated by removing doxycycline treatment (100 μg ml−1) from the drinking water of 4-week-old double transgenic mice for both TRE-MYC and LAP-tTA as previously described9, 13. C57BL/6 mice were obtained from NCI Frederick. Chemically induced HCC was established by intraperitoneal injection of diethylnitrosoamine (DEN) (Sigma) into 2-week-old male pups at a dose of 20 μg g−1 body weight13. Twelve-week-old male B6.Cg-Lepob/J (ob/ob) mice or wild-type control mice were obtained from Charles River. Foxp3–GFP mice were previously described31. NAFLD was induced by feeding mice with a methionine–choline-deficient (MCD) diet (catalogue number 960439, MP biomedical), a choline-deficient and amino-acid-defined (CDAA) diet (catalogue number 518753, Dyets) or a high-fat diet (catalogue number F3282, Bio Serv) for the indicated time10, 11, 32. The MCD diet was supplied with corn oil (10%, w/w), and no fish oil was added. Control diet was purchased from MP Biomedical (catalogue number 960441). Custom-made high- or low-linoleic-acid mouse diets were purchased from Research Diets. The modified diets were based on AIN-76A standard mouse diet, and are isocaloric (4.45 kcal g−1) and contained the same high-fat content (23%, w/w). Linoleic-acid-rich safflower oil and saturated fatty-acid-containing coconut oil were supplied at different ratios to yield 2% (w/w) for the low-linoleic-acid diet or 12% (w/w) for the high-linoleic-acid diet. C57BL/6 mice were fed with the high- or low-linoleic-acid diet for 4 weeks. MYC mice were injected i.p. with 50 μg CD4 antibody (clone GK1.5, BioXcell) every week for the indicated time period to deplete CD4+ T cells33. N-acetylcysteine (NAC) was given in drinking water (10 mg ml−1)34 for the indicated time period to prevent excess ROS production. Mitochondrial-specific antioxidant mitoTEMPO was purchased from Sigma. Mice received mitoTEMPO at a dose of 0.7 mg kg−1 per day25 by osmotic minipumps (ALZET). At the experimental end points, mice were killed. For flow cytometry analysis, single-cell suspensions were prepared from spleen, liver and blood as described previously. Red blood cells were lysed by ACK Lysis Buffer (Quality Biologicals). Parts of live tissue were fixed by 10% formaldehyde and subjected to H&E staining. Free fatty acids were purchased from Sigma. Lipid accumulation was detected by Oil Red O staining in frozen liver sections using the custom service of Histo Serv. Cells were surface-labelled with the indicated antibodies for 15 min at 4 °C. Flow cytometry was performed on BD FACSCalibur or BD LSRII platforms and results were analysed using FlowJo software version 126.96.36.199 (TreeStar). The following antibodies were used for flow cytometry analysis: anti-CD3-FITC (clone 17A2, BD Pharmingen), anti-CD4-PE (clone RM4–4, Biolegend), anti-CD4-APC (clone RM4–5, eBioscience), anti-CD8-Alexa Fluor 700 (clone 53–6.7 Biolegend), anti-CD45, anti-CD44-PE (clone IM7, eBioscience), anti-CD62L-PerCP/Cy5.5 (MEL-14, Biolegend), anti-CD69-Pacific blue (clone H1.2F3, Biolegend), PBS57/CD1d-tetramer-APC (NIH core facility). To determine cytokine production, cells were stimulated with PMA and ionomycine for 30 min, and then were fixed and permeabilized using cytofix/cytoperm kit (BD Pharmingen) followed by anti-IFN-γ-PE (clone XMG1.2, BD Pharmingen), anti-IL-17-PerCP/Cy5.5 (clone TC11-18H10.1, Biolegend) staining. Cell death and apoptosis were detected with annexin V-PE (BD Pharmingen) and 7-AAD (BD Pharmingen) staining according to the manufacturer’s instructions. Intrahepatic CD4+ lymphocytes were gated on the CD3hiCD4+ population from total live hepatic infiltrating mononuclear cells. Absolute numbers were calculated by multiplying frequencies obtained from flow by total live mononuclear cell count, then divided by liver weight. The antibodies used for human peripheral blood mononuclear cell (PBMC) staining are the following: anti-CD3-PE (clone SK7, BD Pharmingen), anti-CD4-FITC (clone RPA-T4, BD Pharmingen), anti-CD8-APC (clone RPA-T8, BD Pharmingen). Murine T assays were performed as described31. Briefly, liver T cells were isolated as CD4+GFP+ by flow-cytometry-assisted cell sorting from Foxp3–GFP mice kept on an MCD or control diet for 4 weeks. CD4+GFP− T effector (T ) cells (5 × 104) were stimulated for 72 h in the presence of irradiated T-depleted splenocytes (5 × 104) plus CD3ε monoclonal antibody (1 μg ml−1), with or without T cells added at different ratios. 3H-Thymidine was added to the culture for the last 6 h and incorporated radioactivity was measured. Freshly isolated splenocytes from MYC-ON MCD mice were incubated with 5 μg ml−1 of mouse α-fetoprotein protein (MyBioSource) for 24 h. Golgiplug was added for the last 6 h. Then, cells were fixed and permeabilized using cytofix/cytoperm kit (BD Pharmingen) followed by anti-IFN-γ-PE (clone XMG1.2, BD Pharmingen) staining. Primary mouse hepatocytes were isolated from MYC mice and cultured according to a previous report35. Briefly, mice were anaesthetized and the portal vein was cannulated under aseptic conditions. The livers were perfused with EGTA solution (5.4 mM KCl, 0.44 mM KH PO , 140 mM NaCl, 0.34 mM Na HPO , 0.5 mM EGTA, 25 mM Tricine, pH 7.2) and Gey’s balanced salt solution (Sigma), and digested with 0.075% collagenase solution. The isolated mouse hepatocytes were then cultured with complete RPMI media in collagen-I-coated plates. Hepatic fatty acid composition was measured at LIPID MAPS lipidomics core at the University of California (San Diego) using an esterified and non-esterified (total) fatty acid panel. Briefly, liver tissues were homogenized and lipid fraction was extracted using a modified Bligh Dyer liquid/liquid extraction method. The lipids were saponified and the hydrolysed fatty acids were extracted using a liquid/liquid method. The extracted fatty acids were derivatized using pentaflourylbenzylbromine (PFBB) and analysed by gas chromatography (GC) using an Agilent GC/mass spectrometry (MS) ChemStation. Individual analytes were monitored using selective ion monitoring (SIM). Analytes were monitored by peak area and quantified using the isotope dilution method using a deuterated internal standard and a standard curve. Isolated primary hepatocytes from MYC mice fed with MCD or control diet were cultured in complete RPMI for 24 h. Supernatant were harvested and FFAs were identified by GC/MS. Splenocytes from MYC mice were cultured with or without 50 μM C18:2 for 24 h. CD4+ and CD8+ T lymphocytes were sorted and total RNA was extracted using miRNeasy mini kit (Qiagen). Array analysis was performed in the Department of Transfusion Medicine, clinical centre at NIH. Mouse gene 2.0 ST array (Affymetrix) was used and performed according to the manufacturer’s instruction. Data were log-transformed (base 2) for subsequent statistical analysis. The Partek Genomic Suite 6.4 was used for the identification of differentially expressed transcripts. The Ingenuity Pathway Analysis tool (http://www.ingenuity.com) was used for analysis of functional pathways. RNA was extracted from frozen tissues with RNeasyMini Kit (Qiagen). Complementary DNA was synthesized by iScriptcDNA synthesis kit (BioRad). Sequence of primers used for quantitative RT–PCR can be obtained from the authors. The reactions were run in triplicates using iQSYBR green supermix kit (BioRad). The results were normalized to endogenous GAPDH expression levels. CD4+ T lymphocytes were isolated from the spleen of MYC mice by negative autoMACS selection using a CD4+ T lymphocytes isolation kit (Miltenyi Biotec) or flow cytometry cell sorting. Human CD4+ T lymphocytes were prepared from PBMCs by autoMACS using a CD4+ T lymphocytes isolation kit (Miltenyi Biotec). The purity of CD4+ T lymphocytes was above 90% after autoMACS separation and above 95% after flow cytometry cell sorting. C16:0, C18:0, C18:1,and C18:2 were purchased from Sigma. Fatty acids were dissolved in DMEM with 2% fatty-acid-free bovine serum albumin (BSA; Sigma, catalogue number A8806) after solvent was evaporated, then followed by two rounds of vortexing and 30 s of sonication. Isolated CD4+ T lymphocytes or splenocytes were incubated with different fatty acids or conditioned medium from hepatocyte culture for 3 days. Unless specifically described, fatty acids were used at 50 μM concentration. For fatty acid depletion, active charcoal (catalogue number C-170, Fisher) was used as described before36. Briefly, 0.5 g of active charcoal was added into every 10 ml of conditioned medium. Then pH was lowered to 3.0 by addition of 0.2 N HCl. The solution was rotated at 4 °C for 2 h. Charcoal was then removed by centrifugation, and the clarified solution was brought back to pH 7.0 by addition of 0.2 N NaOH. NAC (10 mM), catalase (1,000 U ml−1) or mitoTEMPO (10 μM) was used to inhibit ROS production, mitochondrial ROS levels were determined by mitoSOX staining 24 h after treatment, cell death and apoptosis were measured by annexin V and 7-AAD staining 3 days after treatment. Caspase activity assay was measured by caspase-Glo 3/7 assay kit (Promega) according to the manufacturer’s protocol. Fresh prepared liver-infiltrating mononuclear cells were washed and resuspended in 500 μl of BODIPY 493/503 at 0.5 μg ml−1 in PBS. Cells were stained for 15 min at room temperature. Then cells were subjected to flow cytometry analysis. Two pZIP lentiviral shRNA vectors targeting human CPT1a and a control vector (NT#4) were purchased from TransOMIC Technologies. Lentivirus was packed in 293T cells. Jurkat cells were purchased from the German Collection of Microorganisms and Cell Cultures (DSMZ), and no authentication test was performed by us. Cells were cultured in complete RPMI medium and were tested to be mycoplasma free. Jurkat cells were infected with shRNA lentivirus. Puromycin was added to eliminate non-transduced cells. Doxycycline (100 ng ml−1) was added to induce shRNA and GFP expression for 3 days. Efficiency of shRNAs was confirmed by western blot. Jurkat cells with CPT1a knockdown were treated with 200 μM C18:2 for 24 h. Mitochondrial ROS production and cell survival were measured in GFP+-transduced cells. Fatty acid oxidation was measured according to a previous publication37. 1-14C-C18:2 and 1-14C-C16:0 were purchased from PerkinElmer. Briefly, isolated CD4+ or CD8+ T lymphocytes were pretreated with C18:2 or kept in regular media. After 24 h, cell media was changed to media containing 50 μM cold C18:2 plus 1 μCi 1-14C-C18:2 per ml or 50 μM cold C16:0 plus 1 μCi 1-14C-C16:0 per ml. After 2 h, medium was removed and mixed with concentrated perchloric acid (final concentration 0.3 M) plus BSA (final concentration 2%) to precipitate the radiolabelled fatty acids. Samples were vortexed and centrifuged (10,000g for 10 min). Radioactivity was determined in the supernatant to measure water-soluble β-oxidation products. Mitochondrial membrane potential was measured by TMRM (ImmunoChemistry Technologies) staining according to the manufacturer’s protocol. Briefly, cells were kept in culture medium with 100 nM of TMRM for 20 min in a CO incubator at 37 °C. After washing twice, cells were processed to flow cytometry analysis. Mitochondria-associated superoxide was detected by mitoSOX (Life Technologies) staining according to the manufacturer’s protocol. Briefly, cells were first subjected to surface marker staining. Then cells were stained with 2.5 μM mitoSOX for 30 min in a CO incubator at 37 °C. After washing twice, cells were processed for flow cytometry analysis. OCR was measured using an XFe96 Extracellular Flux Analyzer (Seahorse Bioscience) as previously described38. AutoMACS-sorted mouse CD4+ and CD8+ T lymphocytes were attached to XFe96 cell culture plates using Cell-Tak (BD Bioscience) in RPMI media with 11 mM glucose. Cells were activated with 1:1 CD3:CD28 beads (Miltenyi BioTech) and vehicle or 50 μM C18:2 was added. Twenty-four hours after activation, cells were incubated in serum-free XF Base Media (Seahorse Bioscience) supplemented with 10 mM glucose, 2 mM pyruvate and 2 μM glutamine, pH 7.4, along with 50 μM C18:2 if previously present, for 30 min at 37 °C in a CO -free cell culture incubator before beginning the assay. Five consecutive measurements, each representing the mean of 8 wells, were obtained at baseline and after sequential addition of 1.25 μM oligomycin, 0.25 μM trifluorocarbonylcyanide phenylhydrazone (FCCP), and 1 μM each of rotenone and antimycin A (all drugs from Seahorse Bioscience). OCR values were normalized to cell number as measured by the CyQUANT Cell Proliferation Assay Kit (Life Technologies). Human liver samples were stained as previously described8. For immunostaining, formalin-fixed, paraffin-embedded human liver tissue samples were retrieved from the archives of the Institute of Surgical Pathology, University Hospital Zurich. Fibrosis grade was analysed for NASH according to NAFLD activity score (NAS)39 and for others according to METAVIR score40. The study was approved by the local ethics committee (Kantonale Ethikkommission Zürich, application number KEK-ZH-Nr. 2013-0382). Human PBMCs from healthy donors were obtained on an NIH-approved protocol and prepared as described previously41. Informed consent was obtained from all subjects. The sample sizes for animal studies were guided by a previous study in our laboratory in which the same MYC transgenic mouse stain was used. No animals were excluded. Neither randomization nor blinding were done during the in vivo study. However, mice from the same littermates were evenly distributed into control or treatment groups whenever possible. The sample size for the patient studies was guided by a recent publication also studying NASH-induced HCC, but focused on different aspects8. Statistical analysis was performed with GraphPad Prism 6 (GraphPad Software). Significance of the difference between groups was calculated by Student’s unpaired t-test, one-way or two-way ANOVA (Tukey’s and Bonferroni’s multiple comparison test). Welch’s corrections were used when variances between groups were unequal. P < 0.05 was considered as statistically significant.