a–c, Time course of the amounts of the three input FA;Ys in single- (a), double- (b) and triple-labelled (c) TG;Yn in the pulse-chase experiment. d, Relative percentage distribution of the total FA;Y over the labelled classes of TG;Yn. e, Illustration of TG cycling and its effect on the distribution of label between single- and multi-labelled pools. The black numbers are pmol of the respective molecular forms at timepoint 0 derived from the data in panels a–d and represent the sum of all three FA;Ys. The amount of unlabelled TG;Y0 was also measured in the same samples. The blue numbers are predicted values for one cycle of TG de-acylation followed by stochastic re-acylation. Note that this calculation assumes discrete steps during the cycle, which in reality is not the case since degradation and resynthesis are simultaneous. f, The curves show the theoretical calculated binominal label distribution within the pools of TG;Y1, TG;Y2 and TG;Y3 as a function of the fraction of FA;Y of all FAs (unlabelled and labelled; three per TG molecule) in the total pool of TG;Yn. The symbols are the measured values derived from the experiment. The position of the symbols and the corresponding numbers 3.5%, 5%, 12% and 26% indicate the point of best fit of the measured data to the calculated curves. All measured data values were normalized to total FA;Y in TG;Y to compensate for its variations. g–k, Using the data representation as in panel d, TG cycling in separate pulse-chase experiments is compared. All experiments were performed with shorter maximal chase time of 24 h and shorter minimal chase times of 2 h and 4 h. FA label combinations are FA 11:0/16:0/18:2 with alkyne (g) or isotope labeling (h) and 16:0/18:2/19:1 with alkyne (i) or isotope labeling (j), as indicated in the panels. All data are average ± s.d., n = 11–12 (g–j, n = 4). Error bars smaller than symbol size are omitted for clarity.