Association of lipidome remodeling in the adipocyte membrane with acquired obesity in humans

Abstract

Identification of early mechanisms that may lead from obesity towards complications such as metabolic syndrome is of great interest. Here we performed lipidomic analyses of adipose tissue in twin pairs discordant for obesity but still metabolically compensated. In parallel we studied more evolved states of obesity by investigating a separated set of individuals considered to be morbidly obese. Despite lower dietary polyunsaturated fatty acid intake, the obese twin individuals had increased proportions of palmitoleic and arachidonic acids in their adipose tissue, including increased levels of ethanolamine plasmalogens containing arachidonic acid. Information gathered from these experimental groups was used for molecular dynamics simulations of lipid bilayers combined with dependency network analysis of combined clinical, lipidomics, and gene expression data. The simulations suggested that the observed lipid remodeling maintains the biophysical properties of lipid membranes, at the price, however, of increasing their vulnerability to inflammation. Conversely, in morbidly obese subjects, the proportion of plasmalogens containing arachidonic acid in the adipose tissue was markedly decreased. We also show by in vitro Elovl6 knockdown that the lipid network regulating the observed remodeling may be amenable to genetic modulation. Together, our novel approach suggests a physiological mechanism by which adaptation of adipocyte membranes to adipose tissue expansion associates with positive energy balance, potentially leading to higher vulnerability to inflammation in acquired obesity. Further studies will be needed to determine the cause of this effect.

Figure 1. Adipose tissue lipidome in acquired obesity.

Lipidomic analysis covered 314 molecular lipids in adipose tissue biopsies from 44 subjects (13 twin pairs discordant for BMI, nine twin pairs concordant for BMI). Thirty-four lipids were differentially regulated when comparing obese and lean weight-discordant twins (FDR q<0.05). (A) Lipidomic profiles of the 34 differentially regulated lipids in acquired obesity, shown for all subjects included in the study. The detected ether phospholipids are marked in green. The twin status for each subject is shown at the bottom. (B) Concentration, shown as fraction of the total phospholipid concentration as measured by lipidomics (Tot. PL), of the 11 most abundant differentially regulated lipids from (A), shown separately across the four groups corresponding to lean and heavy twins discordant for BMI, and the concordant twin pairs divided into two groups: high BMI (BMI≥25 kg/m²) and low BMI (BMI<25 kg/m²). p-Values are shown for pairwise t test comparison of discordant twins. *, p<0.05; **, p<0.01; ***, p<0.001. Error bars are ± SEM. SM, sphingomyelin.

Figure 2. Fatty acid composition of adipose tissue in acquired obesity.

Fatty acid profile was measured in adipose tissue biopsies in each of the 44 subjects. Complete results are shown in Table S2. (A) Selected fatty acid relative amounts in 13 twin pairs discordant for BMI. Lines connect the pairs of twins. (B) Schematic representation of fatty acid compositional changes when comparing heavy and lean obesity-discordant twins. Significant changes (p<0.05; pairwise t test) are color-coded. The activities of specific fatty acid elongation or desaturation steps are estimated by appropriate fatty acid concentration ratios.

Figure 3. Large-scale molecular dynamics simulations of different lipid membrane systems.

Six single-lipid component systems were studied, representing major groups of lipids found to be differentially regulated in acquired obesity (Figure 1A). Two lipid mixes, corresponding to observed concentration changes of most abundant lipids (Figure 1B), were also studied to represent lipid membrane composition in heavy and lean weight-discordant twins. (A) Structures of two PUFA-containing PCs included in the simulations, one plasmalogen and one ester-bonded. (B) Average area per lipid of the eight simulated bilayer systems, indicating the packing or fluidity of the bilayer (smaller area means decreased fluidity). (C) The molecular order parameter (S _mol) profiles along selected unsaturated acyl chains. Segment number means the carbon position starting from ester-vinyl linkage and ending at the methyl group in the end of the chain. (D) Snapshots from the end of four simulations, with the lipids ordered from top to bottom according to decreased fluidity (area per lipid parameter from [B]).

Figure 4. Regulation of lipid remodeling in adipose tissue.

A dependency network was constructed from selected gene expression, clinical, and lipidomic data from twin pairs discordant for BMI. Node shapes represent different types of variables and platforms (L, UPLC-MS lipidomics; FA, fatty acid gas chromatography; GE, gene expression), node color corresponds to significance and direction of regulation (full data in Table S4), and line width is proportional to strength of dependency. The variables are connected by an edge if and only if their partial correlation is significantly non-zero. PUFA percent and the nearest network hub (Elovl6) are highlighted with green squares. The cutoff for the presence of edge was set at 0.55 by the average non-rejection rate, i.e., an edge in the graph tested positive in 55% of the 500 samplings.

Figure 5. Regulation of lipidomic profiles by Elovl6 in differentiated adipocytes.

3T3-L1 preadipocytes were differentiated in wild-type (control) cell line, and following 50% and 70% Elovl6 knockdown (KD) (three biological replicates for each group). Lipidomic analysis was performed in preadipocytes and in mature adipocytes. One-way ANOVA was performed to test for the significance of time (mature adipocytes versus preadipocytes) and group-specific lipid changes. Error bars are ± SEM. Tot. PL, total phospholipids as measured by lipidomics. (A) The upregulated phospholipids in Elovl6 knockdown cells are mainly shorter chain and saturated. Selected lipids are shown. (B) The diminished phospholipids are mainly PUFA containing, predominantly ether lipids. Selected lipids are shown. All differentially regulated phospholipids in Elovl6 knockdown cells are listed in Table S5.

Figure 6. Adipose tissue in morbid obesity is characterized by diminished levels of PUFA-containing ether lipids.

Comparison of the amounts of abundant phospholipids differentially regulated in acquired obesity in the subcutaneous adipose tissue (Figure 1B), shown as a fraction of the total phospholipid concentration, with the mean amounts in intra-abdominal (Ia fat) and subcutaneous (Sc fat) tissue biopsies of eight morbidly obese subjects from an earlier study . Tot. PL, total phospholipids as measured by lipidomics. (A) There are no marked differences for abundant shorter-chain and saturated phospholipids. *, p<0.05; **, p<0.01. (B) The most abundant ether phospholipids are markedly diminished in morbid obesity. (C) FCS and fasting serum insulin (fS-Insulin) in morbidly obese subjects and twins discordant for obesity. (D) FCS correlates with fasting serum insulin in twins (all 44 subjects included in the study), but not in morbidly obese subjects. The two subjects diagnosed with T2D are marked with filled circles.

Figure 7. Model of physiological regulation of lipid membrane composition in obesity.

MUFA, monounsaturated fatty acid; PL, phospholipid; SFA, saturated fatty acid.