Supplementary Figure S1. Designation of major (> 25 % intensity) and minor (< 25 % intensity of base peak) isobaric lipid molecular species from an HFD LD sample. Shown arerepresentative low resolution product ion spectra,data were acquired in the linear ion trap of an LTQ-FT in data dependent acquisition mode. (A) TG 52:2: The peak at m/z 859.6 reflects the initial loss of NH3. Mass peaks at m/z 575.5, m/z 577.5, m/z 579.5 and m/z 603.5 correspond to the respective secondary neutral losses of FA 18:0, 18:1, 18:2 and 16:0. Fragmentation of FA 18:1 and 16:0 result in major product ions and FA 18:0 and 18:2 in minor product ions. This indicates major TG 16:0_18:1_18:1 and minor TG 16:0_18:0_18:2 molecular species. (B) PC 36:4: The zoomed part of the spectrum shows information relevant for determination of FA composition. Peaks at m/z 478.5, 502.5 and 526.5 reflect the neutral loss of FA 20:4, 18:2 and 16:0, respectively. Peaks at m/z 496.5, m/z 520.5 and m/z 544.5 are the corresponding ketene losses of FA 20:4, 18:2 and 16:0. These fragments result in PC 16:0_20:4 as major and PC 18:2_18:2 as minor molecular species.

A

B

Supplementary Figure S2. Principal component analysis of FED (red), FAS (blue) and HFD (yellow) samples based on their TG profile (112 species in ‰ relative to total amount TG). Each group consists of 1sttrial (open symbols) and 2ndtrial (filled symbols) with n=3 per trial. (A) The groups are clearly separated by principal components 1 and 2. The cummulative variance explained by the first two proncilpe components is over 96 % indicating that the TG profiles have a high discriminatory power to separate the three sample groups. The FED samples are located between the FAS and HFD samples, based on the distinct changes in the profiles of the two latter groups. (B)Principle component 3 explains 2.3 % of the variance in the profiles only, but separates FED trial 1 samples from FED trial 2 (batch effect) and from the FAS and HFD samples.

A

B

Supplementary Figure S3. Principal component analysis of FED (red), FAS (blue) and HFD (yellow) samples based on their PC profile (24 species in ‰ relative to total amount of PC). Each group consists of 1sttrial(open symbols) and 2ndtrial (filled symbols) with n=3 per trial. (A) The groups are clearly separated by principal component 1 but not by 2. The cummulative variance explained by the first two principle components is close to 96 % indicating that the PC profiles have a discriminatory power to separate the three sample groups. Opposite to TG species, for PC species FED samples are located between the HFD and FAS samples, the molecular basis for it is explained in the paper. (B)Principle comoponent 3 explains 2.4 % of the variance in the profiles only, a clear batch effect is visible.

Supplementary Figure S4. Scope ofTG species detected in LD lipidomes of hepatocytes from FED animals by LC-MS/MS in combination with LDA. Data for each species are presented as ‰ relative to amount of total TG species identified and are means ± SD (n=6). A total of 112 TG species (C28-C62) were identified, even down to 0.1 ‰.