Triacylglycerol mobilization underpins mitochondrial stress recovery

Abstract

Mitochondria are central to myriad biochemical processes, and thus even their moderate impairment could have drastic cellular consequences if not rectified. Here, to explore cellular strategies for surmounting mitochondrial stress, we conducted a series of chemical and genetic perturbations to Saccharomyces cerevisiae and analysed the cellular responses using deep multiomic mass spectrometry profiling. We discovered that mobilization of lipid droplet triacylglycerol stores was necessary for strains to mount a successful recovery response. In particular, acyl chains from these stores were liberated by triacylglycerol lipases and used to fuel biosynthesis of the quintessential mitochondrial membrane lipid cardiolipin to support new mitochondrial biogenesis. We demonstrate that a comparable recovery pathway exists in mammalian cells, which fail to recover from doxycycline treatment when lacking the ATGL lipase. Collectively, our work reveals a key component of mitochondrial stress recovery and offers a rich resource for further exploration of the broad cellular responses to mitochondrial dysfunction.

Fig. 1. A multiomic mass spectrometry screen to identify requirements for overcoming mitochondrial dysfunction.

a, A schematic of the collection timeline for multiomic mass spectrometry screening. All 14 strains (2 WT and 12 experimental) were inoculated in YPG respiratory media and incubated for 24 h until the first early collection timepoint after the diauxic shift (E, green dots). Strains with an appreciable growth defect were collected a second time (L, blue dots) when each respective strain had reached the optical density (OD) of the WT at the E timepoint. b,c, The OD (600 nm wavelength) at collection for each of the 14 strains at the E timepoint (n = 3) (b) or the six growth deficient strains at the second (L) timepoint (n = 3) (c). d, Breakdown of the >8,500 biomolecules quantified in each strain by class. e, Hierarchical clustering of the experimental strains in the screen (group 1 (G1), group 2 (G2), group 3 (G3), group 4 (G4)). Strains were clustered based on the average abundance FC (n = 3) for all biomolecules quantified (proteins, dark green; lipids, light green; metabolites, yellow)). f, Relative protein abundances for all group 3 strains (n = 9) collected at the E timepoint compared with WT (n = 3) versus statistical significance. g, Relative protein abundances for all group 3 strains (n = 9) collected at the L timepoint compared with all group 3 strains (n = 9) collected at the E timepoint versus statistical significance. In f and g, non-mitochondrial proteins are coloured light grey and mitochondrial proteins are coloured dark grey. Mitochondrial proteins that are significantly changed (|FC| >0.7, P < 0.05; two-sided Student’s t-test) are highlighted black. h, The relative abundance of mtDNA in group 3 and ssc1 strains at both in the E and L timepoints as measured by a ratio of mtDNA to nuclear DNA (nuDNA). Strains were grown in growth conditions identical to the original screen (n = 3, *P = 0.01, **P = 6.37 × 10⁻⁴, ***P = 1.21 × 10⁻³; two-sided Student’s t-test). For all experiments, the error bars are the s.d., centre values represent the mean and n is the number of independent biological replicates. Source data.

Fig. 2. Yeast strains that recover from mitochondrial stress mobilize TAG stores to facilitate cardiolipin biosynthesis.

a, The relative abundances (log₂ FC) of select lipid species for group 3 and ssc1 strains at the E timepoint compared with WT from the multiomic screen (n = 3). CL, cardiolipin; PA, phosphatidic acid; PC, phosphatidylcholine; PE, phosphatidylethanolamine; TAG, triacylglycerol. b, Representative confocal microscopy images of BODIPY (green) stained yeast strains grown in respiratory YPG media taken at 100× magnification. Scale bar, 10 µm. c, Relative fluorescent signal from Nile red staining of select group 3 and ssc1 yeast strains grown in respiratory media (n = 3, **P = 8.61 × 10⁻⁴, ***P = 5.85 × 10⁻⁴; two-sided Student’s t-test). n.s., not significant. d,e, The normalized abundance of triolein (d) or de novo synthesized [¹³C₃]triolein (e) in cells treated with 400 µM DOX or DMSO vehicle control (n = 3, in d, **P = 1.62 × 10⁻³ and in e, **P = 5.01 × 10⁻⁴; two-sided Student’s t-test). Where indicated, cells were treated with 0.002% (v/v) of [¹³C₁]oleic acid. f, A schematic for the acute DOX labelling experiment. WT cells grown in YPG respiratory media were inoculated and incubated for 23 h and treated for 1 h with [¹³C₁]oleic acid, followed by a 3 h acute DOX treatment and collection. g, The normalized abundance of de novo synthesized [¹³C₄]72:4 CL (red), [¹³C₂]36:2 PC (purple) and [¹³C₂]36:2 PE (orange) in cells after pulse-chase growth. Abundance is normalized to the average T₀ values (n = 3). h, The normalized abundance of the four most abundant species of CL in group 3 strains at both the E (solid) and, when specified, L (striped) timepoints from the multiomic screen. CL species are denoted by colour. Abundance is normalized to the average WT values (n = 3). i, The normalized abundance of the four most abundant species of CL in crd1Δ yeast compared with WT in the presence or absence of DOX. Cells were grown for 24 h in respiratory YPG media. Abundance is normalized to the average WT values (n = 3). j,k, Growth assay of WT and crd1Δ yeast treated with DMSO vehicle control (j) or 400 µM DOX (k) (n = 3, **P = 5.91 × 10⁻⁴; two-sided Student’s t-test). Growth was measured after 24 h (j and k, timepoint E) or 48 h (k, timepoint L) in respiratory YPG media. l, The relative lipid abundance versus statistical significance for crd1Δ yeast treated with DOX compared with DMSO vehicle control. Cells were grown for 24 h in respiratory YPG media. Quantified TAG species are coloured light blue, with TAG species that are significantly changed (|FC| >0.7, P < 0.05; two-sided Student’s t-test) highlighted dark blue. All other lipid species are coloured grey. For all experiments, the error bars are the s.d., centre values represent the mean and n is the number of independent biological replicates. For samples with more than four comparisons, *P < 0.05, **P < 0.01; two-sided Student’s t-test. For all lipid experiments, abundances were normalized to CoQ₈ internal standard. Source data.

Fig. 3. Lipid dropletss are linked to recovery from mitochondrial stress.

a, Fluorescent microscopy-derived cellular volume fractions of mitochondria for cells treated with vehicle or 200 µM DOX. Cells were grown in YPG respiratory media and imaged at 24 h (prerecovery) and again when the OD of the treated strain had reached the OD of the untreated (postrecovery). b,c, The percentage of cells from a that contained normal volume of mitochondria (mitos) (b) or no detectable mitochondria (c). d, Cellular volume fractions of endoplasmic reticulum (ER). e, The percentage of cells from c and d that contained a normal volume of ER. f, Cellular volume fractions of lipid droplets. g, The percentage of cells from c and d that contained a normal volume of lipid droplets (LDs). In a, d and f, density is plotted as the estimated kernel density based on the counts of the measured population (n = 90, 90 and 100 for untreated mitochondria, ER and lipid droplets, respectively; n = 98, 90 and 97, for treated prerecovery mitochondria, ER and lipid droplets, respectively; n = 186, 186 and 185 for postrecovery mitochondria, ER and lipid droplets, respectively). In b, e and g, volume fractions were considered ‘normal’ if they fell within 1.5 s.d. of the mean volume fraction in WT cells. h, Correlation analysis among all individual, quantified biomolecules from multiomic screen and decreased TAG species. Biomolecules were rank ordered according to their average Spearman correlation coefficient between themselves and each individual TAG species. The inset shows the top 50 most positively correlated biomolecules with the decreased TAG species. i, Representative confocal microscopy images of BODIPY (green) stained WT or pln1Δ yeast strains containing either an empty vector (GPD-EV) or the PLN1 gene (GPD-PLN1). Cells were grown for 24 h in SDG respiratory media treated with either 400 µM DOX or DMSO vehicle control. Images were taken at 100× magnification. Scale bar, 10 µm. j, Relative fluorescent signal from Nile red staining of WT or pln1Δ yeast strains expressing either empty vector or GPD-PLN1. Cells were grown for 24 h in SDG respiratory media treated with either 400 µM DOX or DMSO vehicle control (n = 6, *P = 4.66 × 10⁻⁴, **P = 2.19 × 10⁻⁵, ***P = 1.83 × 10⁻³, ****P = 5.92 × 10⁻⁶; two-sided Student’s t-test). k, A schematic for the TAG mobilization experiment. WT cells, expressing empty vector or GPD-PLN1, were inoculated and incubated for 23 h in YPG respiratory media then treated for 1 h with [¹³C₁]oleic acid, followed by an acute 400 µM DOX and 10 mg l⁻¹ cerulenin treatment. Cells were collected at 1, 3 and 5 h after DOX/cerulenin inoculation. l, Normalized abundance of de novo incorporated [¹³C₃]triolein after pulse treatment at T₀ (n = 3). m, Normalized abundance of [¹³C₃]triolein at 1, 3 and 5 h timepoints post-DOX/cerulenin treatment to track TAG mobilization. The abundance is normalized to the average corresponding T₀ samples (n = 3, *P = 0.08, **P = 8.19 × 10⁻³; two-sided Student’s t-test). n, Growth assay of WT yeast strains expressing empty vector or GPD-PLN1. Cells were treated with 400 µM DOX or DMSO control and growth was measured after 24 h (E timepoint) or 44 h (L timepoint) in YPG respiratory media (n = 3, **P = 3.97 × 10⁻⁵, ***P = 8.85 × 10⁻⁵; two-sided Student’s t-test). For all experiments, error bars are the s.d., centre values represent the mean and n is the number of independent biological replicates. For all lipid experiments, abundances are given as peak area normalized to a CoQ₈ internal standard. Source data.

Fig. 4. TAG mobilization during mitochondrial stress requires tgl lipases.

a, A schematic illustrating the two main pathways for TAG mobilization in yeast. TAGs stored in lipid droplets are accessed through either lipolysis using the Tgl3-5p lipases (top) or lipophagy involving direct consumption of the lipid droplet into the vacuole and digestion of TAGs requiring Ldo45/16 and Atg14 (bottom). DAG, diacylglycerol; FFA, free fatty acid. b, The relative fluorescent signal from Nile red staining of WT, TGL triple deletion (tglΔΔΔ), atg14Δ, and ldoΔΔ) yeast grown in YPG respiratory media (n = 6). Cells were treated with either 400 µM DOX or a DMSO vehicle control. The error bars represent the s.d., centre values represent the mean (**P = 1.37 × 10⁻⁷, ***P = 5.05 × 10⁻⁶, ****P = 5.67 × 10⁻⁶; two-sided Student’s t-test). c,d, The relative lipid abundance versus statistical significance for WT (c) or tglΔΔΔ (d) yeast treated with either 400 µM DOX or a DMSO vehicle control (n = 3). Cells were grown for 24 h in YPG respiratory media. In c and d, quantified TAG species are coloured light blue, with TAG species that are significantly changed (|FC| >0.7, P < 0.05; two-sided Student’s t-test) highlighted dark blue. All other lipid species are coloured grey. For all experiments, n is the number of independent biological replicates. Source data.

Fig. 5. Tgl-dependent TAG mobilization is essential for yeast to overcome mitochondrial stress.

a, Growth assay of WT, TGL triple deletion (tglΔΔΔ), dga1Δ and pox1Δ yeast treated with 400 µM DOX. Growth was determined after 24 h (E timepoint) or 48 h (L timepoint) in YPG respiratory media (n = 3, **P = 2.17 × 10⁻⁴, ***P = 3.35 × 10⁻⁴; two-sided Student’s t-test). b, Normalized abundance of the four most abundant species of cardiolipin in WT (solid colours) or TGL triple deletion (tglΔΔΔ, striped colours) yeast treated with 400 µM DOX or DMSO control. Cells were grown for 24 h (E timepoint) or 48 h (L timepoint) in YPG respiratory media (n = 3). The abundance is normalized to the average WT values. c,d, Relative protein abundances for WT yeast treated with 400 µM DOX or DMSO control versus statistical significance: cells were grown for 24 h (WT early) (c) or 48 h (WT late) (d) in YPG respiratory media. The abundance is given by a log₂ transformed FC compared with the WT DMSO-treated vehicle at 24 h (n = 3). e,f, The relative protein abundances for TGL triple deletion (tglΔΔΔ) yeast treated with 400 µM DOX of DMSO control versus statistical significance: cells were grown for 24 h (tglΔΔΔ early) (e) or 48 h (tglΔΔΔ late) (f) in YPG respiratory media. The abundance is given by a log₂-transformed FC compared with the tglΔΔΔ DMSO-treated vehicle control at 24 h (n = 3). In c–f, non-mitochondrial proteins are coloured light grey and mitochondrial proteins are coloured dark grey. Mitochondrial proteins that are significantly changed (|FC| >0.7, P < 0.05; two-sided Student’s t-test) are highlighted black. g, The number of significantly decreased mitochondrial proteins (FC <−0.7, P < 0.05; two-sided Student’s t-test) as quantified in c–f with calculated enrichment. The enrichment is calculated using the Fisher exact test. For all experiments, the error bars are the s.d., centre values represent the mean and n is the number of independent biological replicates. For analyses with more than four comparisons, *P < 0.05, **P < 0.01; two-sided Student’s t-test. Source data.

Fig. 6. TAG mobilization is required for mammalian cells to overcome mitochondrial stress.

a, A growth assay of WT HAP1 cells treated with 10 µM DOX (dark blue) or DMSO control (light blue). Cells were plated into glucose-containing media for 24 h before being swapped into galactose media (day 0) to measure respiratory growth over 4–6 days (n = 3). b, The relative protein abundances for DOX-treated WT HAP1 cells from a collected on day 2 or 4 versus statistical significance. The abundance is the log₂ transformed FC compared with the day 2 DOX-treated sample. Non-mitochondrial proteins (grey), mitochondrial proteins (light orange), OXPHOS (dark orange) and mtDNA-encoded proteins (dark brown) are highlighted by colour (n = 3; two-sided Student’s t-test). c, The relative lipid abundance versus statistical significance for DOX-treated WT HAP1 cells from a collected on day 2 or 5 versus statistical significance. d, The relative lipid abundance versus statistical significance for ATGL knockout HAP1 cells (ATGL^KO) grown in galactose media and treated with 10 µM DOX for 5 days compared with 5-day-treated WT cells. In c and d, the quantified TAG species are coloured light blue with TAG species that are significantly increased (FC >0.7, P < 0.05; two-sided Student’s t-test) (d) or decreased (FC <−0.7, P < 0.05; two-sided Student’s t-test) (c) highlighted dark blue. All other lipid species are coloured grey (n = 3). e, Growth assay of WT HAP1 (from a) or ATGL^KO cells treated with 10 µM DOX (dark blue or green, respectively) or DMSO vehicle control (light blue or green, respectively) grown in the same conditions as a (n = 3). f, Growth assay of CPT2 knockout HAP1 cells (CPT2^KO) treated with 10 µM DOX (tan) or DMSO vehicle control (peach) grown in the same conditions as a (n = 3). g, The growth phenotypes at day 5 of WT, ATGL^KO and CPT2^KO HAP1 cells under vehicle or DOX treatment as depicted in a, e and f. Growth was calculated as the number of population doublings after 5 days (n = 3, **P = 1.14 × 10⁻⁴; two-sided Student^’s t-test). For all experiments, the error bars are the s.d., centre values represent the mean and n is the number of independent biological replicates. Source data.

Extended Data Fig. 1. A multiomic mass spectrometry screen to identify requirements for overcoming mitochondrial disfunction.

a-d, Growth assays of WT and experimental yeast strains used in multiomic screen. Strains were grouped based on growth rate: WT-like growth (Group 1) (a), slight growth defect (Group 2) (b), large growth defect (Group 3) (c), and unrecoverable growth defect (Group 4, ssc1) (d). Growth was measured continuously in a 100 µl culture in YPG media (n = 3) e, log₂ transformed lag time calculated for WT and all experiment strains from growth assays (a-d) (n = 3). f, Principal component analysis (PCA) derived from the normalized abundances of all biomolecules for WT controls and all experimental strains used in multiomic screen (n = 3). g, Enriched GO-terms for proteins that had significantly decreased abundance (FC < -0.7, p < 0.05, two-sided Student’s t-test) in combined Group 3 strains (n = 9) compared to the WT (n = 3) at the early time point. Size of ball represents number of proteins containing the respective GO term, color represents the false discovery rate (FDR), and the bar size represents the fold enrichment of the GO term over the expected frequency. h,i, Number of significantly decreased (FC < -0.7, p < 0.05, two-sided Student’s t-test) mitochondrial proteins for Group 3 strains at the early time point (n = 9) (h) or increased mitochondrial proteins (FC > 0.7, p < 0.05, two-sided Student’s t-test) at the later time point (n = 9) (i) as quantified in Fig. 1f, g, with calculated enrichment. j, Oxygen consumption rate (OCR) of WT and Group 3 strains at the early and later time points. OCR was measured as a change in relative fluorescence over a linear period of either 60-, 90-, or 120-minutes depending on rate of change (n = 3, *P = 0.06, **P = 8.42×10⁻³, ***P = 1.92×10⁻², two-sided Student’s t-test). k, Number of significantly decreased (FC < -0.7, p < 0.05, two-sided Student’s t-test) mitochondrial proteins for the ssc1 strain with calculated enrichment (n = 3). l, Probability of respiratory deficiency in WT, ssc1, and Group 3 strains at either the early (E) or later (L) time point. Probability was calculated by a support vector machine model trained on data obtained from a previous large-scale multiomic screen containing both respiratory competent and deficient yeast strains. m,l, Relative protein abundance of common retrograde response proteins Cit2p (m) and Pdh1p (n) in WT and experimental yeast strains (n = 3). For all experiments error bars are standard deviation, centre values represent the mean and n= number of independent biological replicates. All enrichments are calculated using a Fisher exact test. Numerical data are available in Source data. Source data

Extended Data Fig. 2. Yeast strains that recover from mitochondrial stress mobilise TAG stores to facilitate cardiolipin biosynthesis.

a-d, Relative lipid abundance versus statistical significance for DOX-treated (a), rrf1 (b), hsp10 (c), and ssc1 (d) yeast strains collected at the early time point. Abundance is given by the log₂ transformed fold-change for each strain compared to the WT. Points are color-coded according to lipid class. (n = 3) e-g, Relative log₂ fold change abundance of all quantified TAG species of DOX-treated (e), rrf1 (f), or hsp10 (g) yeast strains compared to WT. Saturation and length of TAG species is given by the number of double bonds or total number of carbons in the acyl-tails, respectively (n = 3). h, Representative confocal microscopy images of WT and hsp10 BODIPY (green) stained yeast strains taken at 60X magnification. Cells were grown for 24 hours in YPG respiratory media. Scale bar represents 10 µm. i, Relative fluorescent signal from Nile red staining of hsp10 and WT yeast strains. Cells were grown for 24 hours in YPG respiratory media (n = 6, **P = 1.40×10⁻⁸, two-sided Student’s t-test). j, Normalized relative abundance of light Triolein (light blue, m + 0) or heavy [¹³C₃]-Triolein (dark blue) in WT cells (from Fig. 2f) pulsed with [¹³C₁]-oleic acid for one hour (n = 3, ***P = 1.70×10⁻⁶, two-sided Student’s t-test). k, Normalized relative abundance of light 36:2 Phosphatidylethanolamine (PE) (orange, m + 0), [¹³C₁]-36:2 PE (tan, m + 1), or [¹³C₂]-36:2 PE (beige, m + 2) in WT cells (from Fig. 2f) pulsed with [¹³C₁]-oleic acid for one hour (n = 3, *P = 5.55×10⁻⁶, **P = 1.12×10⁻⁵, two-sided Student’s t-test). l, Normalized relative abundance of light 36:2 Phosphatidylcholine (PC) (purple, m + 0), [¹³C₁]-36:2 PC (pink, m + 1), or [¹³C₂]-36:2 PC (peach, m + 2) in WT cells (from Fig. 2f) pulsed with [¹³C₁]-oleic acid for one hour (n = 3, *P = 6.39×10⁻⁵, **P = 1.35×10⁻⁴, two-sided Student’s t-test). m, Normalized relative abundance of light 72:4 Cardiolipin (CL) (dark red, m + 0), [¹³C₁]-72:4 CL (red, m + 1), [¹³C₂]-72:4 CL (orange, m + 2), [¹³C₃]-72:4 CL (light orange, m + 3), or [¹³C₄]-72:4 CL (light peach, m + 4) in WT cells (from Fig. 2f) pulsed with [¹³C₁]-oleic acid for one hour (n = 3, *P = 3.39×10⁻⁴, **P = 5.61×10⁻⁵, ***P = 3.85×10⁻⁵, ****P = 3.26×10⁻⁵, two-sided Student’s t-test). n, Relative lipid abundance versus statistical significance for crd1Δ versus WT yeast. Cells were grown for 24 hours in YPG respiratory media (n = 3). Quantified cardiolipin species are colored in red and phosphatidylglycerol species highlighted in pink. All other lipid species are colored in grey. For all experiments error bars are standard deviation, centre values represent the mean and n= number of independent biological replicates. Numerical data are available in Source data. Source data

Extended Data Fig. 3. Pln1p has a role TAG mobilization and recovery from mitochondrial stress.

a, Spearman correlation of log₂ transformed Pln1p abundance and all quantified species of triacylglyceride (TAG) from multiomic screen. Saturation and length of TAG species is given by the number of double bonds or total number of carbons in the acyl-tails respectively. b-c, Relative protein abundance at the early time point of multiomic screen (b) or mRNA (c) abundance of Pln1p in WT and experimental yeast strains. Red bars indicate Group 3 strains. for c, cells were grown for 24 hours in YPG respiratory media (n = 3, *P = 0.03, **P = 0.02, two-sided Student’s t-test). d, Relative lipid abundance versus statistical significance for pln1∆ yeast overexpressing GPD-PLN1 treated with DOX or a DMSO control. Cells were grown for 24 hours in SD (Ura-) respiratory media. Abundance is given as a log₂ transformed fold-change compared the vehicle control. Quantified TAG species are colored in light blue with TAG species that are significantly changed ( | FC | > 0.7, p < 0.05, two-sided Student’s t-test) highlighted in dark blue. All other lipid species are colored in grey (n = 3). e-f, Full growth assay of WT expressing the GPD-PLN1 gene as depicted in Fig. 3n. or an empty vector control. Cells were treated with either DMSO control (e) or DOX (f) (n = 3). Growth was determined in cells grown in YPG + G418 respiratory media over the course of 24 (e) to 48 (f) hours. For all experiments error bars are standard deviation, centre values represent the mean and n= number of independent biological replicates. Numerical data are available in Source data. Source data

Extended Data Fig. 4. Triacylglyceride mobilisation during mitochondrial stress requires tgl lipases.

a-d, Relative lipid abundance versus statistical significance for ldo45Δldo16Δ (ldoΔΔ) a), atg14Δ (b), atg15Δ (c), or atg1Δ (d) yeast. Cells were grown for 24 hours in YPG respiratory media. Abundance is given as a log₂ transformed fold-change compared to the same strain treated with vehicle control. For (a-d), quantified TAG species are colored in light blue with TAG species that are significantly changed ( | FC | > 0.7, p < 0.05, two-sided Student’s t-test) highlighted in dark blue. All other lipid species are colored in grey (n = 3). e, Relative fluorescent signal from Nile red staining of WT, TGL triple deletion (tgl3Δtgl4Δtgl5Δ, tglΔΔΔ), atg1Δ, and atg15Δ yeast (n = 6, *P = 2.20×10⁻⁴, **P = 2.15×10⁻⁷, ***P = 1.31×10⁻³, n.s.= not significant, two-sided Student’s t-test). Cells were grown for 24 hours in YPG respiratory media and were treated with either DOX or a DMSO vehicle control. For all experiments error bars are standard deviation, centre values represent the mean and n= number of independent biological replicates. Numerical data are available in Source data. Source data

Extended Data Fig. 5. Tgl-dependent triacylglyceride mobilisation is essential for yeast to overcome mitochondrial stress.

a, Relative lipid abundance versus statistical significance for dga1Δ yeast. Cells were grown for 24 hours in YPG respiratory media. Abundance is given as a log₂ transformed fold-change compared to WT cells (n = 3). b, Representative confocal microscopy images of BODIPY (green) stained WT or dga1Δ yeast strains. Images were taken at 100X magnification. Scale bar represents 10 µm. c, Growth assay of WT, dga1Δ, lro1Δ, dga1Δlro1Δ double deletion, and dga1Δlro1Δare1Δare2Δ quadruple deletion (ΔΔΔΔ) yeast treated with DOX. Growth was determined after 24 hours (Time point E) or 48 hours (Time point L) in YPG respiratory media (n = 3, *P < 0.05, **P < 0.01, n.s.= not significant, two-sided Student’s t-test). d, Relative lipid abundance versus statistical significance for pox1Δ yeast compared to a WT strain (d) or treated with DOX and compared to pox1Δ yeast treated with a vehicle (e). Cells were grown for 24 hours in YPG respiratory media (n = 3). For (a), (c), and (d) quantified TAG species are colored in light blue with TAG species that are significantly changed ( | FC | > 0.7, p < 0.05, two-sided Student’s t-test) highlighted in dark blue. All other lipid species are colored in grey. f, Relative abundance of mitochondrial DNA (mtDNA) as measured by a ratio of mtDNA to nuclear DNA (nuDNA) for WT or TGL triple deletion (tgl3Δtgl4Δtgl5Δ, tglΔΔΔ) yeast treated with DOX or vehicle control after 24 hours (Time point E) or 48 hours (Time point L). Cells were grown in YPG respiratory media (n = 3, *P = 0.01, **P = 0.03, two-sided Student’s t-test). g, Normalized abundance of de-novo synthesized [¹³C₃]-Triolein (m + 3) in WT or TGL triple deletion (tgl3Δtgl4Δtgl5Δ, tglΔΔΔ) after 1 hr treatment with 0.0002% [¹³C₁]-oleic acid or vehicle (n = 3). Abundances were normalized to CoQ₈ internal standard. h, Growth assay of WT or TGL triple deletion (tgl3Δtgl4Δtgl5Δ, tglΔΔΔ) yeast treated with 0.0002% [¹³C₁]-oleic acid or vehicle. Growth was determined after 24 hours (Time point E) or 48 hours (Time point L) in YPG respiratory media, (n = 3, *P = 0.03, **P = 0.02, n.s.= not significant, two-sided Student’s t-test). For all experiments error bars are standard deviation, centre values represent the mean and n= number of independent biological replicates. Numerical data are available in Source data. Source data

Extended Data Fig. 6. Triacylglyceride mobilisation is required for mammalian cells to overcome mitochondrial stress.

a, Relative protein abundances for WT HAP1 cells grown in galactose media treated with 10 µM DOX collected on Day 2 versus statistical significance. Abundance is given by a log₂ transformed fold-change compared to the vehicle treated control with non-mitochondrial proteins (grey), mitochondrial proteins (light orange), OxPhos (dark orange) and mtDNA encoded proteins (dark brown) highlighted by colour. (n = 3) b, Relative lipid abundance versus statistical significance for WT HAP1 cells grown in galactose, treated with 10 µM DOX DOX for 2 days compared to a vehicle treated control. Quantified TAG species are colored in light blue with TAG species that are significantly increased (FC > 0.7, p < 0.05, two-sided Student’s t-test) or decreased (FC < −0.7, p < 0.05, two-sided Student’s t-test) highlighted in dark blue. All other lipid species are colored in grey (n = 3). For all experiments, n= number of independent biological replicates. Numerical data are available in Source data. Source data