Proximity proteomics reveals a mechanism of fatty acid transfer at lipid droplet-mitochondria- endoplasmic reticulum contact sites

Abstract

Membrane contact sites between organelles are critical for the transfer of biomolecules. Lipid droplets store fatty acids and form contacts with mitochondria, which regulate fatty acid oxidation and adenosine triphosphate production. Protein compartmentalization at lipid droplet-mitochondria contact sites and their effects on biological processes are poorly described. Using proximity-dependent biotinylation methods, we identify 71 proteins at lipid droplet-mitochondria contact sites, including a multimeric complex containing extended synaptotagmin (ESYT) 1, ESYT2, and VAMP Associated Protein B and C (VAPB). High resolution imaging confirms localization of this complex at the interface of lipid droplet-mitochondria-endoplasmic reticulum where it likely transfers fatty acids to enable β-oxidation. Deletion of ESYT1, ESYT2 or VAPB limits lipid droplet-derived fatty acid oxidation, resulting in depletion of tricarboxylic acid cycle metabolites, remodeling of the cellular lipidome, and induction of lipotoxic stress. These findings were recapitulated in Esyt1 and Esyt2 deficient mice. Our study uncovers a fundamental mechanism that is required for lipid droplet-derived fatty acid oxidation and cellular lipid homeostasis, with implications for metabolic diseases and survival.

Fig. 1. Identification of LD-mitochondria interface proteins using proximity proteomic screens.

a BioID targeted to the surface of LDs, containing the transmembrane domain of AAM-B, RFP (mScarlet), HA, and BirA* (AAM_TMD-RFP-HA-BirA*). b BioID targeted to the surface of the outer membrane of mitochondria, containing the transmembrane domain of FIS1, CFP (mTurquoise2), FLAG, and BirA* (FIS1_TMD-CFP-FLAG-BirA*). c Mass spectrometric analysis of biotinylated proteins purified from HepG2 cells stably expressing AAM_TMD-RFP-HA-BirA*chimeras (LD-BioID). Plot compares BirA* to the negative control in HepG2 cells supplemented with biotin. Colored dots indicate proteins enriched using the three independent constructs (a–e). Proteins known to be enriched at LDs (ArtGAP1, SNX, FASN), peroxisomes (GOSAR1, ABCD3), endoplasmic reticulum (ESYT1, ESYT2, RTN4) and mitochondria (DNM1L, CSDE1, OCIAD1) are highlighted. d Mass spectrometric analysis of biotinylated proteins purified from HepG2 cells stably expressing Fis1_TDM-CFP-FLAG-BirA* (Mito-BioID). Experiment and data analysis as indicated in (c). e Design of the Split-BioID approach, containing the transmembrane domain of FIS1, CFP, FLAG, and half of BirA*(BirA_N*) targeted to the surface of the outer membrane of mitochondria (Fis1_TMD-CFP-FLAG-BirA_N*). The transmembrane domain of AAM-B anchors RFP, HA, and the other half of BirA*(BirA_C*) targeted to the surface of LDs (AAM_TMD-RFP-HA-BirA*). f Airyscan imaging of HepG2 cells stably expressing the Split-BioID. Merged image showing Fis1_TMD-CFP-FLAG-BirA_N* (magenta) and AAM_TMD-RFP-HA-BirA_C*(green). (representative of 4 independent experiments, scale bar, 2 μm). g Mass spectrometric analysis of biotinylated proteins purified from HepG2 cells stably expressing Split-BioID and negative controls (non-BirA*-transfected HepG2 cells). Colored dots indicate commonly enriched proteins in panels C & D. h Venn diagram showing the number of proteins enriched using Split-BioID, LD-BioID and mito-BioID approaches. i Dot-plot analysis of commonly enriched proteins using Split-BioID, LD-BioID and mito-BioID approaches. j Validation of proteins identified by BioID using an organelle coprecipitation assay of HepG2 cell lysate. LD-mitochondria fraction (LD+Mito), mitochondrial fraction (Mito), and cytosol fraction (Cyto) were examined by immunoblot. Representative of n = 3 biological replicates. For c, d, g and i, data analyzed with unpaired two-tailed Student’s t-test with FDR < 0.05, n = 4 biological replicates. See also Supplementary Fig. 1 and Supplementary Data 1. Created in BioRender. Keenan, S. (2025) https://BioRender.com/n75s805.

Fig. 2. ESYT1, ESYT2 and VAPB form homo- and hetero-oligomeric complexes.

a Volcano plot showing proteins associated with ESYT1. Data are from a label-free proteomic analysis of anti-HA immunoprecipitates from HepG2 cells stably expressing 3HA-eGFP-ESYT1 versus 3HA-EGFP. p-values of two-tailed Student’s t-test, n = 4 biological replicates. b, c Volcano plot showing proteins associated with ESYT2 and VAPB using HepG2 cells stably expressing 3HA-eGFP-ESYT2 and 3HA-eGFP-VAPB. Experiments and statistical analysis as indicated in (a). d–f Western blot analysis of ESYT1, ESYT2 and VAPB interactions following anti-HA immunoprecipitants from HepG2 cells stably expressing 3HA-eGFP-ESYT1, 3HA-eGFP-ESYT2, 3HA-eGFP-VAPB or empty vector (EV; 3HA-EGFP) and probed for endogenous proteins. MBOAT9 was used as a negative control. Representative of n = 3 biological replicates. g, h Lifetime maps (g) of the FLIM data acquisitions between eGFP-ESYT1, eGFP-ESYT2, and eGFP-VAPB (donor) with VAPB-mCH (acceptor) pseudo-colored according to the FRET palette defined in the phasor plot (h) (i.e., teal pixels = 0 % FRET while red pixels = 23% FRET) that reports hetero protein-protein interaction. i Hetero protein-protein interaction of ESYT1, ESYT2 and VAPB in HepG2 cells. From left to right columns: n = 21, 15, 16, 17, 18, 19 and 14 cells. j Fraction of heterocomplex between eGFP-ESYT1, eGFP-ESYT2, and eGFP-VAPB with VAPB-mCH. From left to right columns: n = 16, 13, 13, 15, 15, 18, and 17 cells. k Structure of ESYT1 in complex with ESYT2. Arrows- hydrophobic side chain residues at the interacting interface that are identified via in silico saturation mutagenesis and selected for further experimental validation of heterodimer formation. l Quantification of the fraction of pixels exhibiting FRET between eGFP-ESYT2 and wild type and mutant ESYT1-mCH (i.e., hetero protein-protein interaction) across multiple HepG2 cells, from left to right columns n = 5, 6, 8, 4, 6, 5, 4, 6 and 6 cells. Data analyzed using unpaired two-tailed t-tests. *P < 0.05. The box and whisker plots in 2i, j and l show the minimum, maximum and sample mean. In 2i, 2j and 2 l, *p < 0.05, unpaired t-test, two-sided. See also Supplementary Fig. 2 and Supplementary Data 2.

Fig. 3. ESYT1/2-VAPB form a complex at the LD-Mitochondria-ER interface.

a Left panel: Single plane view of Airyscan imaging of HepG2 cells stably expressing eGFP-ESYT, eGFP-ESYT2 or eGFP-VAPB (blue). HepG2 cells were fixed and stained for LDs with HCS LipidTox red Neutral Lipid Stain (green), mitochondria with TOMM20 antibody (yellow), and ER with Calnexin antibody (magenta). White arrows highlight tripartite interfaces of LD, mitochondria, and ER. A total of 16 cells from four independent experiments were imaged and a representative of image is indicated. Large scale bars, 2 μm and inset scale bars 0.5 μm. Right panel: Line scan analysis of images presented to the left. A and B on x-axis corresponds to the line scan in panel a inset. b Left panel: Donor lifetime maps of the FLIM data acquisitions between donor eGFP-ESYT1, eGFP-ESYT2, and eGFP-VAPB with acceptor VAPB-mCH FRET pair pseudo-colored according to the palette defined in the phasor plot in (Fig. 2h) spatially map the hetero protein-protein interaction (red). Right panel: Identification of the LDs and mitochondrial interface based on PLIN2-AF405 and TOMM20-AF647 immunofluorescence intensity mask analysis. c Quantification of the fraction of ESYT1/VAPB (upper panel), ESYT2/VAPB (middle panel), and VAPB/VAPB interaction inside the LDs and mitochondrial interface defined by PLIN2-AF405 and TOMM20-AF647 intensity masks (panel B, right) versus whole cell protein interaction fraction with and without Forksolin stimulation across multiple HepG2 cells. ESYT1/VAPB -FSK n = 10; ESYT1/VAPB + FSK n = 7; ESYT2/VAPB -FSK n = 10; ESYT2/VFSK+fork n = 9; VAPB/VAPB -FSK n = 9; VAPB/VAPB + FSK n = 7 cells. *P < 0.05 paired t-test, one-tailed. See also Supplementary Fig. 3.

Fig. 4. Impaired fatty acid metabolism in ESYT1/2-VAPB-deficient cells.

a Representative western blot showing ESYT1, ESYT2 and VAPB protein in HeLa cells following CRISPR-Cas9 mediated knockout of ESYT1, ESYT2, VAPB and ESYT1 and ESYT2. GAPDH, loading control. * non-specific immunoreactive band. b 3D rendering images of Control and KO cells stained for LDs with HCS LipidTox Deep Red Neutral Lipid Stain (various colors) and mitochondria with TOMM20 antibody (red). Prior to staining ESYT1/2/VAPB^KO and Control HeLa cells were treated with 400 μM fatty acids (oleic acid and palmitic acid; 2:1) for 4 h. Scale bars: 0.5 μm. c Quantification of volume occupied by LDs per cell from experiments shown in panel B. Control n = 35, ESYT1^KO n = 44, ESYT2^KO n = 42, VAPB^KD n = 34, DKO n = 35 cells. d Lipidomic analysis showing triacylglycerol content in HeLa cells treated with 400 μM oleic acid and palmitic acid (2:1) for 4 h. n = 6 biological replicates. e, f HeLa cells were pulsed for 6 h with radiolabeled fatty acid (¹⁴C oleic acid) conjugated to 1% BSA and chased for 4 h in low glucose medium without fatty acids (starvation medium). Data indicates rate of ¹⁴C- oxidation (LD-derived fatty acid oxidation) and ¹⁴C remaining in triglyceride following the chase period. Bar graphs represent the mean ± SEM from 3 (e) and 4 (f) biological replicate experiments. g BODIPY C₁₆ in LDs of cells that were pulsed with BODIPY C₁₆ for 6 h, washed, and incubated in starvation medium for 4 h. n = 6 (VAPB^KD, DKO) or 7 (Control, ESYT1^KO, ESYT2^KO) biological experiments. h Western blot showing steady state level of phosphorylated and total ATGL and HSL protein in cells. * denotes non-specific immunoreactive band. i Flux of ¹³C fatty acids into TCA cycle and non-mitochondria metabolites (n = 4 per group). The ¹³C-fatty acid mix contained myristic (0.2%), palmitoleic (9.4%), palmitic (38.9%), margaric (0.3%), linoleic (10.7%), oleic (26.9%), elaidic (1.6%), and stearic (1.6%) acid. For (c– i) bar graphs represent the mean ± SEM, *p < 0.05 vs. Control using one-way ANOVA with Bonferroni’s multiple comparison test. See also Supplementary Fig. 4 and Supplementary Data 3.

Fig. 5. SMP domain of ESYT1 and ESYT2 are required for fatty acid transfer from LDs to mitochondria.

(a) Apolar channel (yellow) formed by ESYT1 or ESYT2 SMP domain in ESYT1/2 dimer complex. b-c Ligand docking analysis showing oleic acid inside apolar channel of ESYT1 as a lateral (b) and frontal (c) view. (d) Enrichment of oleic acid in 3HA-EGFP-ESYT2 cells (compared with empty vector, EV, 3HA-EGFP) following HA immunoprecipitation and mass spectrometry lipidomic analysis. Cells were incubated in 400 μM oleic acid for 4 h prior to lysing. *p < 0.05 vs EV by unpaired two-tailed t-tests. n = 4 biological replicates. Bar graphs represent the mean ± SEM. e-f A comparison of SMP domain and key residues substituted by heavy-side chain amino acid of ESYT1 that protrude into the apolar channel to obstruct fatty acid trafficking. g Airyscan images of HeLa cells showing lipid droplets in ESYT1^KO cells without or with re-expression of wild type ESYT or mutations in the SMP domain. Cells were treated with 400 μM oleic acid and palmitic acid (2:1) for 4 h prior to staining. LD staining with HCS LipidTox Green (green) and DAPI (Blue). Scale bar: 3 μm. h Fatty acid oxidation in HeLa cells as described in G. n = 6 biological replicate experiments. i, j A comparison of WT SMP domain (i) and a residue substituted by heavy-side chain amino acid (I353Y; j) of ESYT2 that protrudes into the apolar channel. k Airyscan images of HeLa cells showing lipid storage in ESYT2^KO cells in comparison with ESYT2^KO cells with re-expression of ESYT2^WT-eBFP or a mutation in the SMP domain (ESYT2^I354Y-eBFP). Prior to staining cells were treated as in g for LD staining with HCS LipidTox Green Neutral Lipid (green) and nuclei staining with SYTOX Deep Red (Blue). Scale bar: 5 μm. (l) Fatty acid oxidation in HeLa ESYT2^KO cells in comparison with rescued cell lines with expression of wild type or mutant ESYT2 on SMP domain. n = 4 biological replicates. In (h, i) * P < 0.05 using one-way ANOVA with Bonferroni’s multiple comparison test. Bar graphs represent the mean ± SEM. See also Supplementary Fig. 5.

Fig. 6. ESYT1, ESYT2 or VAPB depletion sensitizes cells to palmitic acid-induced lipotoxicity and cellular stress.

a Schematic representation of fatty acid trafficking at LD, mitochondria and ER contacts in the presence of protein-mediated fatty acid transfer (efficient fatty acid trafficking) and with inefficient fatty acid trafficking mediated by disruption to the ESYT/VAPB complex. Created in BioRender. Keenan, S. (2025) https://BioRender.com/n75s805. b–f Lipidomic analysis showing abundance of lipids associated with lipotoxicity including diacylglycerol, ceramide, sphingomyelin (SM), cardiolipin, and lysophosphatidylcholine (LPC). Bar graphs represent the mean ± SEM from 6 biological replicate experiments. *p < 0.05 vs. Control using one-way ANOVA with Bonferroni’s multiple comparison test. g Representative Western blot and quantitation of ER stress marker proteins. Bar graphs represent the mean ± SEM from n = 4 biological replicate experiments of BIP, ATF5, p-PERK (Thr980), PERK and PDI. ATF5_s/ATF6_FL indicates ratio of protein content of short and full length ATF6. GAPDH was used as loading control. Data analysed using one-way ANOVA and a Kruskal-Wallis post hoc test. p < 0.05 vs. Control. (h) Representative Western blot and quantitation of p-ERK (Thr202/Tyr204) and ERK, p-JNK (Thr183/Tyr185) and JNK. GAPDH was used as a loading control. Bar graphs represent the mean ± SEM. (p-JNK/JNK n = 4 and p-Erk1/2 / ERK1/2 n = 5 independent experiments). Data analysed using one-way ANOVA and a Kruskal-Wallis post hoc test. *p < 0.05 vs. Control. i Representative Western blot and quantitation of pyroptosis markers with palmitic acid-induced stress in control and KO cells. In all experiments in this panel, HeLa cells were treated with palmitate (500 µM) conjugated to 1% BSA for 4 h. Bar graphs represent the mean ± SEM protein content of HMGB1 (n = 6 for Control, ESYT2^KO, VAPB^KD and DKO, n = 5 ESYT1^KO), cleaved caspase 1 (n = 4), ratio of full length and splice variant of Gasdermin (n = 4) from independent experiments. GAPDH was used as loading control. Data analysed using one-way ANOVA and a Kruskal-Wallis post hoc test. *p < 0.05 vs. Control.

Fig. 7. ESYT1 and ESYT2 regulate fatty acid oxidation in vivo.

a Schematic outlining strategy for CRISPR-Cas9 mediated deletion of Esyt1 and Esyt2 in hepatocytes of mice. b Esyt1 and Esyt2 peptide abundance in liver of mice. Data are mean ± SEM. *P < 0.05 vs. WT using one-way ANOVA and a Kruskal-Wallis post hoc test. n = 5 Control, n = 3 ESYT1^KO, n = 3 ESYT2^KO (n = 5 ESYT1 peptide; n = 3 ESYT2 peptide), ESYT1^KO (n = 3 for both ESYT1 and ESYT2 peptide), ESYT2^KO (n = 3 for both ESYT1 and ESYT2 peptide). c Fatty acid oxidation in precision-cut liver slices from Control, Esyt1^KO and Esyt2^KO mice. Data are mean ± SEM. *P < 0.05 vs. WT using one-way ANOVA and a Kruskal-Wallis post hoc test. n = 8 Control, n = 5 ESYT1^KO, n = 7 ESYT2^KO. d dEsyt2 mRNA in WT and dEsyt^KO and dEsyt^FB-KO Drosophila. Data are mean ± SEM. *p < 0.05 vs. WT by unpaired two-tailed t-test. n = 3 per group. e Breeding strategy for the generation of fat body-specific dEsyt2 inhibition in Drosophila (left) and survival response of WT ^FB and dEsyt^FB-KO males under starvation (right), n = 100 flies per genotype. Data analyzed using Log-rank (Mantel-Cox) test. (f) Breeding strategy for the generation of global dEsyt inhibition in Drosophila (left) and survival response of WT and dEsyt^KO males under starvation (right), n = 100 flies per genotype. Data analyzed using Log-rank (Mantel-Cox) test. See also Supplementary Fig. 6. Created in BioRender. Keenan, S. (2025) https://BioRender.com/n75s805.

Fig. 8. Proposed Model of ESYT1-ESYT2-VAPB Mediated Fatty Acid Transfer at the LD-Mitochondria-ER Interface.

Left panel: TEM image of HeLa cells showing an exploratory tripartite interaction among LDs, mitochondria and the ER. Right panel: Summarizing model of complex formation of ESYT1, ESYT2 and VAPB at the interface of these organelles. Created in BioRender. Keenan, S. (2025) https://BioRender.com/n75s805.