Mitochondria Bound to Lipid Droplets Have Unique Bioenergetics, Composition, and Dynamics that Support Lipid Droplet Expansion

Abstract

Mitochondria associate with lipid droplets (LDs) in fat-oxidizing tissues, but the functional role of these peridroplet mitochondria (PDM) is unknown. Microscopic observation of interscapular brown adipose tissue reveals that PDM have unique protein composition and cristae structure and remain adherent to the LD in the tissue homogenate. We developed an approach to isolate PDM based on their adherence to LDs. Comparison of purified PDM to cytoplasmic mitochondria reveals that (1) PDM have increased pyruvate oxidation, electron transport, and ATP synthesis capacities; (2) PDM have reduced β-oxidation capacity and depart from LDs upon activation of brown adipose tissue thermogenesis and β-oxidation; (3) PDM support LD expansion as Perilipin5-induced recruitment of mitochondria to LDs increases ATP synthase-dependent triacylglyceride synthesis; and (4) PDM maintain a distinct protein composition due to uniquely low fusion-fission dynamics. We conclude that PDM represent a segregated mitochondrial population with unique structure and function that supports triacylglyceride synthesis.

Keywords: brown adipose tissue; lipid droplet; mitochondria; mitochondrial dynamics; peridroplet mitochondria.

Figure 1. Isolation of peridroplet mitochondria by differential centrifugation

A. Schematic representation of peridroplet (PDM) and cytoplasmic (CM) mitochondrial isolation from interscapular brown adipose tissue (BAT). BAT was dissected from mice and homogenized with glass-Teflon dounce homogenizer. Low-speed centrifugation separated the fat layer containing PDM from supernatant containing CM. High-speed centrifugation stripped PDM from lipid droplets (LDs) and pelleted CM mitochondria from the supernatant. Note that some BAT mitochondrial isolation protocols discard the fat layer and/or begin with high-speed centrifugation step. B–E. PDM are stripped from LDs by high-speed centrifugation. B–C. Super-resolution confocal images of the fat layer before and after high-speed centrifugation. LDs were marked by the neutral BODIPY 493/503 fluorescent dye (BODIPY) and mitochondria by MitoTracker deep red dye (MitoTracker). Note the tubular structures staining positively for MitoTracker on LDs. D. Quantification of LDs with MitoTracker staining in the fat layer pre- and post-stripping by high-speed centrifugation. 11,744 LDs were assessed in total. *** p < 0.0001. E. Super-resolution confocal image of PDM pellet separated from fat layer by high-speed centrifugation. F–G. Low-magnification (20×) images of the fat layer, PDM pellet, and CM pellet. LD content was assessed by BODIPY staining. 5–6 technical replicates per group. N = 3 independent isolations. ns p>0.05, *** p < 0.0001. One-way ANOVA with Tukey post-test. H. Mass spectrometry analysis of relative mitochondrial protein content of CM and PDM preparations. I–J. Analysis of CM and PDM membrane potential by fluorescence microscopy of CM and PDM double-stained with the membrane potential-sensitive dye MitoTracker Red and the mitochondrial protein dye MitoTracker Green. N = 15–22 images per group from 3 independent isolations. ns p>0.05.

Figure 2. Peridroplet mitochondria have enhanced bioenergetic capacity

A–I. Peridroplet (PDM) and cytosolic (CM) mitochondria isolated from brown adipose tissue (BAT). A. Fluorescence microscopy images of seahorse respirometry plate containing isolated CM and PDM stained with MitoTracker Red. B. Quantification of MitoTracker Red fluorescence intensity (F.I.) in Seahorse wells loaded with CM or PDM. 5–7 wells quantified per condition. C. Representative traces of oxygen consumption rates (OCRs) of isolated PDM and CM driven with pyruvate+malate. ADP, Oligomycin, FCCP, and Antimycin were sequentially injected to assess mitochondrial respiratory states. 4–6 technical replicates per group. D. Quantification of OCR at different mitochondrial respiratory states in representative experiment. State II quantifies respiration driven proton leak (no ATP synthesis), State III quantifies respiration driven by ATP synthesis, and maximal respiration quantifies maximal electron transport activity induced by the chemical uncoupler FCCP. 6 technical replicates per group. E. Quantification of mitochondrial respiratory states in N = 8 independent experiments. For each individual experiment, average OCR values of CM and PDM were normalized to the average OCR of all mitochondria (see Quantification and Statistical Analysis for complete equations). F–G. Cytochrome C oxidase activity in PDM and CM isolated from BAT. F. Representative traces of oxygen consumption rate (OCR) of isolated PDM and CM driven with the cytochrome c oxidase-specific substrates TMPD+Ascorbate. Rotenone and Antimycin were injected in the beginning of the assay to extinguish cytochrome c reduction by Complex I and Complex III. The COX-specific inhibitor sodium azide was injected at the end of the assay to control for non-COX respiration. 5 technical replicates per group. G. Quantification of COX activity in N = 4 independent isolations. Data were normalized as in E. H–I. ATP synthase activity in PDM and CM isolated from BAT. H. Representative traces of luciferase luminescence assay in isolated mitochondria normalized to baseline. ATP synthesis rates were determined by the rate of luminescence gain. I. Quantification of ATP synthase activity in N = 4 independent isolations. Data were normalized as in E. J–K. Confocal imaging of living cultured brown adipocytes stained with membrane potential-sensitive dye TMRE. J. Confocal imaging before and after addition of the ATP synthase inhibitor Oligomycin. Bright field image was used to identify LDs. TMRE images were pseudo-colored for quantitative display (see calibration bar in top left). Note that PDM had higher fluorescence than CM after oligomycin treatment. White dashed circles denote LDs, white N denotes the nucleus, and white square denotes zoomed region. K. Quantification of TMRE fluorescence intensity in oligomycin-treated brown adipocytes. 159 mitochondria were assessed in total. N = 33 cells collected in 6 independent experiments. Data are expressed as means ± SEM. ns p>0.05, * p< 0.05, ** p < 0.001, *** p < 0.0001. See also Figure S1.

Figure 3. Peridroplet mitochondria have increased levels of cytochrome c oxidase, ATP synthase and super complex I+III assembly

A–C. Western blot analysis of Peridroplet (PDM) and cytosolic (CM) mitochondria isolated from brown adipose tissue (BAT). A. Western blot of isolated CM and PDM stained with the dye Ponceau S for total protein loading. B. Western blot probed with antibodies of OXPHOS complex subunits I–V (CI-CV) and Tom20 as a loading control. C. Quantification of OXPHOS complex subunits normalized to Tom20 loading control in N = 4–7 independent isolations. For each individual experiment, average values of CM and PDM were normalized to the average OCR of all mitochondria (see Quantification and Statistical Analysis for complete equations). D–E. Western blot of Blue Native PAGE of PDM and CM isolated from BAT. D. Western blot of assembled Complex I and Complex III in isolated mitochondria. E. Quantification of Complex III assembled into I+III supercomplexes relative to total Complex III. N = 5 independent isolations. Data were normalized as in C. F–I. Super-resolution confocal imaging of fixed cultured brown adipocytes (no adrenergic stimulation). F. Brown adipocytes immunostained for COX4. Bright field image was used to identify LDs and TOM20 immunostaining was used to mark the mitochondrial network. White dashed circles denote LDs, white N denotes the nucleus, and white square denotes zoomed region. G. Quantification of COX4 distribution in brown adipocyte mitochondria. 490 mitochondria were assessed in total. N = 22 cells collected in 3 independent experiments. H. Brown adipocytes immunostained for ATP Synthase. Bright field image was used to identify LDs and TOM20 immunostaining was used to mark the mitochondrial network. White dashed circles denote LDs, white N denotes the nucleus, and white square denotes zoomed region. I. Quantification of ATP Synthase distribution in brown adipocyte mitochondria. 507 mitochondria were assessed in total. N = 20 cells collected in 4 independent experiments. Data are expressed as means ± SEM. * p< 0.05, *** p < 0.0001. See also Figure S2.

Figure 4. Peridroplet mitochondria have decreased fatty acid oxidation capacity and mitochondria-LD contact is decreased upon activation of thermogenic fatty acid oxidation in vivo

A. PDM have lower fatty acid oxidation capacity. Representative quantification of maximal palmitoyl-carnitine driven oxygen consumption rate (Max OCR) in isolated peridroplet (PDM) and cytoplasmic (CM) mitochondria. 4–6 technical replicates per group. B. Quantification of palmitoyl-carnitine oxidation capacity. N = 5 independent experiments. For each individual experiment, average OCR values of CM and PDM were normalized to the average OCR of total mitochondria (see Quantification and Statistical Analysis for full equations). C. UCP1 protein is similarly abundant in PDM and CM. Western blot analysis of UCP1 in CM and PDM. N = 3 independent mitochondrial isolations. Data were normalized as in B. D–E. PDM have higher activity of the TCA cycle enzyme Citrate Synthase. D. Representative traces of citrate synthase DTNB absorbance assay in isolated mitochondria normalized to baseline. Citrate synthase activity was determined for CM and PDM by the rate of absorbance gain. E. Quantification of citrate synthase specific activity. N = 4 independent mitochondrial isolations. Data were normalized as in B. F–G. PDM have increased NAD(P)H content. F. Confocal image of NAD(P)H fluorescence in living cultured brown adipocytes. Image was pseudo-colored for quantitative display (see calibration bar in top left). Note the high level of NAD(P)H in PDM. G. Quantification of NAD(P)H level. N = 24 cells imaged in 6 independent experiments. CM and PDM fluorescent intensities (F.I.) were normalized to average cell F.I. for each individual cell. H. Electron micrographs (EMs) of BAT harvested from mice adapted to thermoneutral conditions (28°C), where fatty acids are stored in lipid droplets, and cold environment (6°C), where thermogenic fatty acid oxi dation is robustly increased. Blue lines highlight mitochondrial perimeter and red lines highlight overlap between mitochondria and LD border. I–K. Mitochondria in contact with LD quantified by count, % mitochondrial perimeter and % LD perimeter. N = 10 EMs per condition. Data are expressed as means ± SEM. ns p>0.05, * p< 0.05, ** p < 0.001, *** p < 0.0001. See also Figure S3.

Figure 5. Mitochondria-lipid droplet association promotes lipid droplet expansion

A. Schematic representation of Perilipin5 (Plin5) domains: the conserved perilipin domains PAT1 and PAT2, the ATGL-binding domain responsible for lipolysis regulation, and the mitochondrial recruiting sequence. B. Super-resolution confocal images of living brown adipocytes untransduced (control), transduced with the full-length Plin5 that contains mitochondrial recruiting sequence, and truncated Plin5 that lacks the mitochondria recruitment sequence (Plin5Δ399-463). Mitochondria are marked by TMRE staining and lipid droplets (LDs) by BODIPY 493/503. Note the increased lipid droplet (LD) mass and mitochondrial recruitment in Plin5. C. Quantification of mitochondrial recruitment to LDs assessed as the area of mitochondria within 0.5 µm of LD border. N = 14–24 cells analyzed per group from 4 independent experiments. D. Quantification of LD mass by cross-sectional area of BODIPY 493/503 normalized to cell area. N = 17–33 cells per group from 4 independent experiments. E. LD size distribution assessed by cross-sectional area of individual LDs. N = 302–489 LDs per group from 4 independent experiments. F. Quantification of lipolysis by glycerol release assay. N = 3 independent experiments. For each individual experiment, average values of CM and PDM were normalized to the average values of total mitochondria (see Quantification and Statistical Analysis for full equations). G. Super-resolution confocal images of living INS1 pancreatic beta cell line untransduced (control), transduced with Plin5, and Plin5Δ399-463 and stained with TMRE to mark mitochondria and BODIPY 493/503 to mark LDs. Note the increased LD mass and mitochondrial recruitment in Plin5. H. Quantification of mitochondrial recruitment to LDs assessed as the area of mitochondria within 0.5 µm of LD border. N = 12–20 cells analyzed per group from 3 independent experiments. I. Quantification of LD mass by cross-sectional area of BODIPY 493/503 normalized to cell area. N = 13–19 cells per group from 3 independent experiments. J. LD size distribution assessed by cross-sectional area of individual LDs. N = 226–344 LDs per group from 3 independent experiments. Data are expressed as means ± SEM. ns p>0.05, * p< 0.05, ** p < 0.001, *** p < 0.0001. See also Figure S4.

Figure 6. Mitochondria-lipid droplet association promotes triglyceride synthesis

A. Representative thin layer chromatography (TLC) of cellular lipids extracted from cultured brown adipocytes untransduced (control), transduced with the full-length Plin5 that contains mitochondrial recruiting sequence, and truncated Plin5 that lacks the mitochondria recruitment sequence (Plin5Δ399-463). Cells were incubated with BODIPY C12 558/568 (C12) overnight to assess triacylglyceride (TAG) synthesis. The mobility of fatty acids species from loading origin is determined by relative polarity, with TAG migrating the highest. B. Quantification of TAG from N = 3 independent experiments. Data were normalized to control for each individual experiment. C–D. TLC of cultured brown adipocytes incubated with C12 with or without the fatty acid esterification inhibitor Triacsin C (red). In histogram, note the decrease in TAG and increase in Free C12 induced by Triacsin C. E. Representative TLC of cultured brown adipocytes incubated with C12 with or without the mitochondrial ATP synthase inhibitor Oligomycin. F. Quantification of TAG synthesis dependent on mitochondrial ATP from N = 3 independent experiments. Mitochondrial ATP-dependent TAG synthesis was calculated as the difference in TAG between Oligomycin-treated and untreated cells. Data are expressed as means ± SEM. * p< 0.05, ** p < 0.001.

Figure 7. Peridroplet mitochondria have unique structure, fusion-fission dynamics, and motility

A–C. Electron micrograph (EMs) of BAT harvested from mice adapted to thermoneutral conditions (28°C), where peridroplet mitochondria (PDM) are most abundant. Red lines highlight PDM and blue lines highlight cytoplasmic mitochondria (CM). Note the elongation of PDM. Mitochondrial size and shape were quantified in N = 22 – 34 mitochondria from 10 EMs per group. D–E. Confocal microscopy of living cultured brown adipocytes. D. Confocal images of brown adipocytes transduced with mitochondrially-targeted photo-activatable GFP (mtPAGFP) stained with TMRE to label the mitochondrial network. mtPAGFP in single mitochondria (white squares) were sequentially photo-converted and imaged immediately. Filled grey circles denote lipid droplets. E. Quantification of mitochondrial shape, as delineated by mtPAGFP. N = 47 mitochondria from 4 independent imaging experiments. F–G. Mitochondrial fusion assay in living cultured brown adipocytes. Brown adipocytes transduced mtPAGFP were stained with TMRE to label the mitochondrial network. White dashed circles denote LDs and white N denotes the nucleus. mtPAGFP was photo-converted in a small region of the cell (white squares) and its fluorescence intensity tracked over time. The dilution of mtPAGFP fluorescence intensity over time results from fusion between activated mitochondria with non-activated mitochondria. N = 5 cells per group imaged in 3 independent experiments. Data were normalized to baseline and statistically analyzed by Two-Way ANOVA for repeated measures with Bonferroni post-test. H–I. PDM have reduced motility compared to CM. H. Psudo-colored confocal images of brown adipocyte at two different time points (red and green). Merged image of two time points reveals immobile mitochondria (yellow) and mobile mitochondria that change position over time (red and green). White dashed circles denote LDs, white N denotes the nucleus, and white dashed square denotes zoomed region. Note the reduced mobility of PDM compared to CM. I. Quantification of mitochondrial motility. Mitochondrial motility was quantified in time-lapse images as the percent of area displaced over a period of 10 seconds. For each individual cell, CM and PDM motility values were normalized to the average motility value of all mitochondria in the cell. N = 15 cells images in 3 independent experiments. J. Confocal image of fixed cultured brown adipocytes immunolabeled for the mitochondrial fission protein Drp1. LDs were identified by bright field images and the mitochondrial network was marked with mitochondrially-targeted DsRed (mtDsRed). Note the low levels of Drp1 recruitment to PDM (white arrows). K. Quantification of Drp1 associated with CM and PDM. Drp1 association was quantified as puncta area divided by mitochondrial area. In each individual cell, Drp1 association to CM and PDM values were normalized to the average of the entire cell. N = 14 cells per group imaged in 3 independent experiments. L–O. Western blot analysis of the mitochondrial inner membrane protein OPA1 in isolated PDM and CM. Proteolytic cleavage of the long-forms OPA1 (L-OPA1) to short-OPA1 (S-OPA1) is associated with inner membrane fission. Densitometry of L-OPA1 and S-OPA1 in PDM and CM are shown in representative histogram. N–O. Quantification of total OPA1 and S-OPA1 in CM and PDM. N = 6–7 independent mitochondrial isolations. G. Confocal images of living primary brown adipocytes transduced with Drp1-dominant negative (Drp1DN) and transduction control. LDs were identified by bright field images and the mitochondrial network was marked with TMRE. P. Fission arrest by Drp1 dominant negative (Drp1DN) expression does not recruit mitochondria to LD surface compared to transduction control. Mitochondrial recruitment was assessed as the area of mitochondria within 0.5 µm of LD border. N = 10 cells analyzed per group. Data are expressed as means ± SEM. ns p> 0.05, * p< 0.05, ** p < 0.001, *** p <0.0001. See also Figure S5.