Triglyceride cycling enables modification of stored fatty acids

Triglyceride/fatty acid (TG/FA) cycling entails partial or complete degradation of stored triacylglycerols (TGs) to release free fatty acids (FAs) that are subsequently re-esterified into new TGs. It can occur systemically (adipose-to-liver re-esterification) and intracellularly as a futile substrate cycle that consumes energy without net biomass synthesis. Conceptually, whether benefits of such cycles (regulation, thermogenesis) outweigh their energetic cost is debated. Experimentally, intracellular TG/FA cycling is hard to quantify directly because conventional isotope tracing often renders educts and products indistinguishable after one cycle. Prior approaches have provided indirect estimates of esterification/re-esterification but not direct molecular-species-resolved cycling. The authors previously developed an alkyne-FA tracing and click-chemistry mass spectrometry method with high sensitivity and the ability to discern single- and multi-labelled lipid species. Here they apply a multiplexed multi-label strategy to directly demonstrate TG cycling in adipocytes, estimate TG half-life in the cycle, and show how cycling enables remodeling of the stored FA pool via elongation and desaturation.

Earlier studies emphasized regulatory roles of TG/FA cycling in adapting to energy demand and more recently implicated it in thermogenesis in adipose tissue. Conventional isotope labelling methods (e.g., ratiometric [3H]FA vs [14C]glucose incorporation with glycerol release; [3H]H2O or [18O]H2O enrichment) infer total TG synthesis/turnover and re-esterification indirectly. However, direct tracing showing that a defined TG pool gives rise to a new pool by reusing the same FAs has been lacking due to analytical challenges distinguishing single- vs multi-labelled species across many possible TG molecular species. The authors’ alkyne-FA tracers with click-chemistry reporters and MS overcome these limitations by providing high sensitivity, specificity, multiplexing, and molecular-species resolution, including discrimination of single- and multi-labelled TGs.

• Tracer design and calibration: Combined three distinct alkyne-labelled FA tracers to span chain length and unsaturation while enabling unambiguous simultaneous discrimination: odd-chain medium FA 11:0;Y, even-chain polyunsaturated FA 18:2;Y, and 13C-labelled long-chain saturated FA 13C-16:0;Y. Odd/even chain lengths and stable isotope provided orthogonal identifiers. Established calibration by labelling 3T3-L1 adipocytes with individual and all pairwise/triple combinations to assign MS signals to each FA tracer, including single- and multi-labelled lipid species, using LipidXplorer mfql files. Achieved clean separation of single-, double-, and triple-labelled TG;Yn signals. Extended analysis to eight lipid classes (CE, Cer, DG, PA, PC, PE, PI, TG). Quantified up to ~1,600 labelled species per multiplexed sample using 16 internal standards. - Cell systems: Differentiated 3T3-L1 adipocytes; primary murine white and brown adipocytes isolated from C57BL/6 mice with defined induction/maintenance protocols. - Pulse-chase experiments: Pulse with 50 µM of each tracer FA for 1 h; chase in fresh medium for 0, 2–6, 24, and 48 h (higher-resolution experiments up to 24 h). Performed both alkyne-FA and heavy-isotope FA labelling combinations, including 11:0;Y/16:0;Y/18:2;Y and 16:0;Y/18:2;Y/19:1;Y; and isotope analogues 11:0[D3]/16:0[13C16]/18:2[13C18] and 16:0[13C16]/18:2[13C18]/19:1[D8]. - Sample processing: Lipid extraction (MeOH/CHCl3), addition of alkyne-labelled and non-alkyne internal standards, phase separation, drying. Click chemistry with C175-7x reporters (trialkylamine variants enabling 4-plex multiplexing), pooling of multiplexed samples, back-extraction, resuspension in spray buffer. - Mass spectrometry: Thermo Q-Exactive Plus with HESI; direct infusion; MS1 segmented scans (m/z 300–1,400); DIA MS2 with inclusion lists covering lipid mass defects; high resolution (280k) for MS1/MS2; targeted scans for double/triple-charged species. For alkyne-labelled species, neutral loss (NL) of 73.1 Da used to selectively visualize clicked alkyne-labelled neutral lipids and provide sample multiplex separation (NL 73/75/76/77). For isotope-labelled species, used FA-specific NL (e.g., NL m/z 206.2 for FA 11:0[D3]+NH2) due to lack of an alkyne-equivalent reporter. - Data analysis: Conversion to mzML; identification/quantification via LipidXplorer with 131 optimized mfql search files covering label assignments for each FA tracer and lipid class, including multi-labelled TG;Yn. Quantified labelled species and summed into lipid-class and label-count pools (TG;Y1, TG;Y2, TG;Y3). For species-level modification analysis, interrogated MS2 FA fragment series from TG;Y1 to identify elongated/desaturated metabolites (e.g., FA 20:4;Y from 18:2;Y). - Kinetic and modeling analyses: Tracked redistribution of label across TG;Y1/Y2/Y3 over chase. Compared observed label distributions to binomial predictions under stochastic TG→DG→TG cycling with re-acylation from a common acyl-CoA pool. Estimated TG half-life in the cycling process by fitting decay of TG;Y3 and redistribution kinetics. - Modulation of beta-oxidation: Applied CPT1 inhibitor teglicar (30 µM; IC50 ~40 µM) during chase in primary adipocytes to assess effects on apparent cycling rates.

• Direct evidence of intracellular TG/FA cycling: After pulse, labelled TGs were predominantly multi-labelled; during chase there was a marked increase in single-labelled TG;Y1 and a concomitant decrease in double- and triple-labelled TG;Y2/Y3 across >200 TG species. - Kinetics and quantitative redistribution: After 6 h chase, 46.7 ± 4.1% of initial TG;Y2 and 36.5 ± 3.5% of initial TG;Y3 remained. Redistribution from TG;Y3 and TG;Y2 to TG;Y1 matched binomial expectations for a stochastic TG→DG→TG cycle fed by a common FA-CoA pool. At pulse end, labelled TG pools contained TG;Y3 94 pmol, TG;Y2 1,017 pmol, TG;Y1 3,030 pmol, against an endogenous unlabelled TG pool ~138,000 ± 7,000 pmol. The fraction of labelled FA in the effective TG pool diluted from ~26% at 0 h to ~12% (6 h), ~5% (24 h), and ~3.5% (48 h). - Estimated TG half-life in the cycle: Redistribution kinetics (especially TG;Y3 decay) indicate a cycling half-life t1/2 ~2–4 h; analysis suggests ~4 h (6 h ≈ 1.5 t). - Generality across tracer sets and labels: Similar cycling kinetics observed with 11:0;Y/16:0;Y/18:2;Y and 16:0;Y/18:2;Y/19:1;Y, and with heavy-isotope tracers, demonstrating independence from alkyne tags and medium-chain FA presence. - FA modification coupled to cycling: Released FAs underwent elongation/desaturation before re-esterification. Notably, 18:2;Y converted to 18:3;Y, 20:3;Y, and 20:4;Y (arachidonate equivalent). 16:0;Y converted to 16:1;Y and 18:1;Y. 11:0;Y was partially degraded (7:0;Y, 9:0;Y) and elongated (13:0;Y, 15:0;Y). 19:1;Y partially degraded to 17:1;Y with minimal elongation to 21:1;Y. Phospholipids showed increasing average chain length and double bonds over time, supporting de novo modification rather than exchange from phospholipids. - Differential fate of input FAs (48 h): ~82% of medium-chain 11:0;Y cleared (likely oxidation); of the remainder, ~10% modified (notably 15:1;Y, 17:1;Y), ~16% remained as 11:0;Y. For 16:0;Y, ~100 ± 13% remained with ~88% unmodified; modifications produced 16:1;Y and 18:1;Y. For 18:2;Y, little loss; 20–30% converted to 20:3;Y and 20:4;Y. - Primary adipocytes and beta-oxidation: White and brown primary adipocytes exhibited TG cycling at rates comparable to 3T3-L1. Inhibiting CPT1 with teglicar increased the apparent cycling rate (more TG;Y1 accumulation), consistent with less diversion of released FAs into oxidation and more re-esterification. - Energetic estimate: With ~4 h half-life in 3T3-L1, each FA in TG would turn over roughly once per day; energetic cost approximated to ~1 ATP per FA per day (~1% of palmitate’s energy content per day).

The multiplexed alkyne-FA MS tracing method enables direct, molecular-species-resolved demonstration of intracellular TG cycling in adipocytes, overcoming limitations of conventional isotopic methods. The observed rapid shift from multi- to single-labelled TGs, with quantitative agreement to binomial redistribution under stochastic TG→DG→TG cycling, rules out alternative explanations; the increase in TG;Y1 is largely accounted for by the decline in TG;Y2 and TG;Y3. While the minimal cycle involves TG hydrolysis to DG and re-acylation, the pathway could include further deacylation to MG or free glycerol; current data cannot discriminate among these routes or identify specific enzymes. The cycling appears stochastic and fast (t1/2 ~4 h in 3T3-L1), implying nontrivial energetic costs, though likely lower in large-droplet white adipose tissue in vivo. Biologically, beyond regulatory flexibility and potential roles in UCP1-independent thermogenesis, a key advantage is dynamic remodeling of the stored FA pool: saturated FAs are slowly desaturated to monounsaturated species, linoleate is converted towards arachidonate, and medium-chain FAs are cleared. Such remodeling supports membrane homeostasis, mitigates palmitate-associated lipotoxicity risk, sustains arachidonate supply for signaling, and allows clearance of damaged (e.g., peroxidized) FAs from TG stores by making them accessible as FA-CoAs for modification or degradation.

This study introduces a robust, multiplexed alkyne-FA tracing platform with mass spectrometry that directly quantifies intracellular TG cycling at molecular-species resolution. In adipocytes, TGs undergo rapid cycling (t1/2 ~2–4 h), driving redistribution of labels from multi- to single-labelled TGs. Cycling couples to FA remodeling via elongation and desaturation, gradually converting saturated FAs to monounsaturated FAs and linoleate to arachidonate. The phenomenon is general across tracer sets, seen with isotope labels, and present in primary white and brown adipocytes; limiting FA oxidation increases apparent cycling. These findings establish TG cycling as a mechanism that maintains and adapts the composition of stored FA pools. Future work should define the precise enzymatic steps (e.g., roles of specific lipases and acyltransferases), resolve whether MG/glycerol steps are involved, incorporate early time points with inhibitors and gene knockouts, and develop mathematical models that integrate TG;Yn with DG;Yn, MG;Y, PA;Y, and phospholipid pools.

• Pathway resolution: Current data cannot distinguish whether the cycle is limited to TG↔DG or proceeds through MG and/or free glycerol; enzymes responsible are not identified. - Temporal granularity: Additional very early chase time points and perturbations (inhibitors, gene ablations) are needed for mechanistic dissection. - Analytical constraints: Isotope-labelled FA analyses have lower sensitivity/specificity than alkyne tracers and require FA-specific NL quantification, potentially introducing quantitative bias. - System scope: Experiments were in differentiated 3T3-L1 and primary murine adipocytes ex vivo; in vivo rates in white adipose tissue may differ (e.g., larger droplets). - Assumptions in modeling: Binomial/stochastic re-acylation assumptions ignore positional specificity and potential enzyme preferences. - Phospholipid tracing: Direct identification of labelled FA fragments in phospholipids is limited (no labelled FA fragment release in MS2), necessitating inference from aggregate properties (chain length, unsaturation).