Herbal terpenoids activate autophagy and mitophagy through modulation of bioenergetics and protect from metabolic stress, sarcopenia and epigenetic aging

Abstract

Small molecular food components contribute to the health benefits of diets rich in fruits, vegetables, herbs and spices. The cellular mechanisms by which noncaloric bioactives promote healthspan are not well understood, limiting their use in disease prevention. Here, we deploy a whole-organism, high-content screen in zebrafish to profile food-derived compounds for activation of autophagy, a cellular quality control mechanism that promotes healthy aging. We identify thymol and carvacrol as activators of autophagy and mitophagy through a transient dampening of the mitochondrial membrane potential. Chemical stabilization of thymol-induced mitochondrial depolarization blocks mitophagy activation, suggesting a mechanism originating from the mitochondrial membrane. Supplementation with thymol prevents excess liver fat accumulation in a mouse model of diet-induced obesity, improves pink-1-dependent heat stress resilience in Caenorhabditis elegans, and slows the decline of skeletal muscle performance while delaying epigenetic aging in SAMP8 mice. Thus, terpenoids from common herbs promote autophagy during aging and metabolic overload, making them attractive molecules for nutrition-based healthspan promotion.

Fig. 1. Development of high-content imaging-based assay to quantify autophagic flux in zebrafish larvae.

a, Schematic of the transgenic Tg(actc1b:ZsGreen-map1lc3;cryaa:TdTomato) zebrafish line for live monitoring of autophagy. b, Representative fluorescence image of transgenic ZsGreen-LC3 zebrafish larvae at 3 dpf, acquired at 4× (top) and 20× (bottom) magnifications. c, Detection of LC3-positive vesicles through semiautomated high-content imaging in transgenic larvae at 3 dpf (top) and 6 dpf (bottom). After acquisition, images were processed to locate objects corresponding to LC3 puncta (segmentation). Arrowheads indicate ZsGreen accumulation in autophagosomes in response to autophagy activation. The resulting objects were identified and counted as described in Extended Data Fig. 1b. d, Measurement of autophagic flux in zebrafish larvae by relative quantification of the total area of LC3 puncta normalized to control before and after 4 h of exposure to 100 mM NH₄Cl (− NH₄Cl: 3 dpf, n = 72; 4 dpf, n = 59; 5 dpf, n = 60; 6 dpf, n = 69; + NH₄Cl: 3 dpf, n = 75; 4 dpf, n = 61; 5 dpf, n = 68; 6 dpf, n = 72). e, Workflow for rapid testing of small molecules in 96-well microplates to identify modulators of autophagy. f, Representative images of transgenic zebrafish larvae at 3 dpf treated for 16 h with 1 µM rapamycin (top), followed by treatment with NH₄Cl (bottom). g, Quantification of autophagic flux in zebrafish larvae treated with rapamycin (3 dpf; vehicle −/+ NH₄Cl, n = 68; rapamycin − NH₄Cl, n = 67; rapamycin + NH₄Cl, n = 71) and normalized to control (vehicle, 1% DMSO). Violin plots in d and g report the median (solid lines) and quartiles (dotted lines). Data were pooled from three independent experiments and analyzed by two-way ANOVA followed by Sidak multiple comparisons. Scale bars, 100 µm (b top) and 20 µm (b bottom, c and f). Panels a and e created with BioRender.com. Source data

Fig. 2. Whole-organism screening identifies thymol and carvacrol as activators of autophagy.

a, Heat map of autophagy induction in response to bioactives tested at four ascending doses with and without treatment with NH₄Cl in zebrafish larvae at 3 dpf. The colors represent the relative area of LC3 puncta normalized to the vehicle control, ranging from low (blue) to high (red) activation. White crossed cells represent data excluded due to treatment toxicity (3 dpf; n = 4–21 per condition). FC, fold change. b, Representative images of ZsGreen-LC3 zebrafish larvae treated for 16 h with the indicated molecules, either alone or in combination with NH₄Cl. Scale bar, 20 µm. c–e, Quantification of autophagic flux in transgenic zebrafish larvae treated with 1 µM rapamycin, 50 µM thymol, 50 µM carvacrol and 50 µM oregano essential oil (OEO) (3 dpf; − NH₄Cl: vehicle, n = 59; rapamycin, n = 61; thymol, n = 61; carvacrol, n = 60; OEO, n = 60; + NH₄Cl: vehicle, n = 58; rapamycin, n = 61; thymol, n = 60; carvacrol, n = 60; OEO, n = 42) (c); 50 µM thymol sulfate (3 dpf; vehicle − NH₄Cl, n = 68; vehicle + NH₄Cl, n = 70; thymol sulfate − NH₄Cl, n = 67; thymol sulfate + NH₄Cl, n = 68) (d); and 50 µM thymol glucuronide (3 dpf; vehicle − NH₄Cl, n = 70; vehicle + NH₄Cl, n = 73; thymol glucuronide − NH₄Cl, n = 65; thymol glucuronide + NH₄Cl, n = 70) (e). Violin plots report the median (solid lines) and quartiles (dotted lines). Data were pooled from three independent experiments and analyzed by two-way ANOVA followed by Sidak multiple comparisons. Source data

Fig. 3. Thymol transiently dampens mitochondrial respiration and protects against mitochondrial toxicity.

a, LC3 signals measured by flow cytometry in Jurkat cells treated with rapamycin or thymol (blue data points: native; green data points: lysosomal (Lys) inhibitor cotreatment). Each data point represents the mean of a biological replicate (n = 2). MFI, mean fluorescence intensity. b, Autophagic flux calculated using the LC3-II mean fluorescence intensity: ((+ Lys inhibitor) − (− Lys inhibitor))/(− Lys inhibitor). Data points represent the means of biological replicates (n = 2). c, MMP in Jurkat cells treated with vehicle (0.1% DMSO), thymol (100 µM) or FCCP (100 nM), measured by flow cytometry as the ratio between the monomeric form (emission wavelength (Em) = 590 nm) and the aggregated form (Em = 525 nm) of JC-10 (n = 3 wells per condition). Data are presented as means ± s.e.m. Ordinary one-way ANOVA followed by Sidak multiple comparisons. d, ADP-stimulated oxygen consumption rate (OCR) of mitochondria isolated from livers treated with thymol (green solid line). Values were normalized to the vehicle (dashed blue line) and are presented as means ± s.e.m. (n = 4). e, Microscopy images of MAFs treated with vehicle (0.1% DMSO) or thymol (100 µM) in the presence of MitoTracker (green) and TMRM (red). Scale bar, 50 µm. f, Quantification of the TMRM/MitoTracker area to calculate the mitochondrial membrane potential (MMP). Each data point represents one cell (control, n = 50; thymol, n = 55). Two-tailed Student’s t test. g, Gene set enrichment analysis in MAFs ranked by level of significance. Dotted lines indicate the significance threshold. FDR, false discovery rate. h, Twenty-four significantly enriched pathways related to autophagy and cellular energy metabolism. Experiments (e–h) were conducted on MAFs with a 24-h exposure to vehicle (0.1% DMSO) or thymol (100 µM) (control, n = 6; thymol, n = 6). i, Setup for OCR analysis. j, OCR in zebrafish continuously treated with thymol compared to the washout and control groups (vehicle, 1% DMSO). FCCP stimulates maximal OCR. Rotenone and antimycin A (R + AA) were used to determine nonmitochondrial respiration (3 dpf; control, n = 18; thymol, n = 21; washout, n = 21 pools of larvae). k, Basal and maximal OCR calculated as the average of the last three points before the addition of FCCP and R + AA, respectively (3 dpf; control, n = 18; thymol, n = 20; washout, n = 21 pools of larvae). Data were pooled from three independent experiments. Ordinary one-way ANOVA followed by Sidak multiple comparisons. Box plots show the median, 25th and 75th percentiles, and minima and maxima. l, Survival analysis of zebrafish larvae treated with thymol or vehicle and exposed to mitochondrial stress through rotenone treatment. Data were pooled from two independent experiments (3 dpf; n = 48 per group). Pairwise comparisons to control using a log-rank test. P values were adjusted using Bonferroni correction. Flow cytometry gating schemes are shown in Supplementary Fig. 1. Panel i created with BioRender.com. Source data

Fig. 4. Thymol induces mitophagy in cells and mice.

a, Representative confocal microscopy images of mito-QC MAFs treated with vehicle (0.1% DMSO), oligomycin (1 nM), thymol (100 µM) or thymol (100 µM) + oligomycin (1 nM) for 24 h. Scale bar, 20 µm. b, Quantification of mitophagy activation as the mitophagy index, which represents the relative mitolysosomal area in mito-QC MAFs. Each data point in the box plots represents one cell (control, n = 22; oligomycin, n = 25; thymol, n = 23; thymol + oligomycin, n = 23). Ordinary one-way ANOVA followed by Dunnett multiple comparisons. c, Study scheme for the acute exposure of transgenic mito-QC reporter mice to vehicle or thymol. Mice received two doses of thymol by oral gavage at a dose of 20 mg per kg body weight, or vehicle, 16 and 2 h before liver and muscle tissue collection (10 weeks old, n = 5 per group). d, Representative confocal microscopy images of gastrocnemius skeletal muscle 2 h after the second treatment. Scale bar, 50 µm. e, Quantification of the mitophagy index as the relative mitolysosomal area. f, Number of mCherry-positive mitophagy foci. Ten fields were analyzed for each mouse, with each data point in the box plots in e and f representing the value from an individual microscopic field (control, n = 50; thymol, n = 50). Two-tailed Student’s t test. Box plots show the median, 25th and 75th percentiles, and minima and maxima. Panel c created with BioRender.com. Source data

Fig. 5. Thymol prevents liver fat accumulation in mice fed a HFD.

a, Schematic overview of acute intervention with thymol in C57BL/6 mice. Mice received either thymol 20 mg kg⁻¹ or vehicle by two gavages 16 and 2 h before they were killed, and proteins or mitochondria were isolated from liver tissues after animal killing (10 weeks old, n = 5 per group). b, Western blot analysis of total lysates and mitochondrial extracts from liver of mice treated with thymol or vehicle using the indicated antibodies. c, Densitometric quantification of the western blots. One-tailed Student’s t test. d, Schematic overview of chronic intervention study in wild-type mice. Ten-week-old mice were fed standard chow (control) or a HFD. Thymol or vehicle was administered per os to mice fed a HFD 5 days per week for 8 weeks. The control group was treated with vehicle only. e, Body weight changes throughout the intervention. Data are presented as means ± s.e.m. with pairwise comparisons at the end of the treatment (n = 15 mice per group). Two-tailed Student’s t test. f, Representative histological images of liver tissue stained with Oil Red O (ORO) and hematoxylin and eosin (H&E). Scale bar, 100 µm. g, Quantification of mean lipid droplet size from ORO staining. Each data point represents the value quantified from one mouse (control + vehicle, n = 14; HFD + vehicle, n = 15; HFD + thymol, n = 14). h, Liver triglyceride (TG) content (n = 13 per group). i, Western blot analysis of liver tissues using the indicated antibodies for the quantification of LC3-II/LC3-I and P62 protein contents. j, Densitometric quantification of the western blots (n = 15 per group). Box plots in c, g, h and j show the median, 25th and 75th percentiles, and minima and maxima. Ordinary one-way ANOVA followed by Sidak multiple comparisons. Panels a and d created with BioRender.com. Source data

Fig. 6. Thymol increases pink-1-dependent stress resistance, improves motility and reduces age-related myofibril damage in C. elegans.

a, Mean number of GFP-positive puncta per individual in treated compared to control animals (day 1: control, n = 149; carvacrol, n = 193; thymol, n = 132; day 10: control, n = 109; carvacrol, n = 138; thymol, n = 114; 72-h pretreated: carvacrol, n = 79; thymol n = 61). Ordinary one-way ANOVA followed by Sidak multiple comparisons. b, Epifluorescence microscopy images of C. elegans expressing an mtRosella biosensor in body wall muscle cells to visualize mitophagy in response to control (0.1% DMSO) and thymol (25 µM) treatment in young (day 1) and old (day 10) wild-type and pink-1 (RNAi)-treated animals. Scale bar, 8 µm. c,d, Mitophagy quantified on day 1 (c) (wild-type (WT): vehicle, n = 80; thymol, n = 63; pink-1 RNAi: vehicle, n = 52; thymol, n = 66) and day 10 (d) (WT: vehicle, n = 15; thymol, n = 16; pink-1 RNAi: vehicle, n = 16; thymol, n = 18) as the ratio of DsRed to GFP fluorescence intensity. Violin plots show the median (solid lines) and quartiles (dashed lines). Two-way ANOVA followed by Sidak multiple comparisons. e,f, Activity score (arbitrary units) in response to 4 h of thermal shock at 37 °C. Before heat exposure, worms were treated with thymol (vehicle, n = 2,319; thymol, n = 1,728) (e) or carvacrol (vehicle, n = 2,266; carvacrol, n = 1,980) (f) for 72 h at 50 μM from days 1 to 3 of adulthood. g–j, Activity quantification upon genetic loss-of-function of daf-16 (vehicle, n = 1,677; thymol, n = 1,791) (g), pink-1 (vehicle, n = 1,275; thymol n = 1,295) (h), aak-2 (vehicle, n = 2,500; thymol, n = 2,234) (i) or skn-1 (vehicle, n = 1,800; thymol n = 1,497) (j) in response to 50 μM thymol following the same conditions as in e. k,l, Activity score on days 4, 7 and 10 of adulthood in worms exposed to thymol at 25 μM (vehicle, n = 1,068; thymol n = 1,029) (k) and 50 μM (vehicle, n = 1,068; thymol, n = 1,044) (l) or vehicle (1% DMSO) from the egg stage to adulthood. Ordinary one-way ANOVA followed by Dunnett multiple comparisons. In e–l, data were pooled from three independent experiments and are presented as means ± s.e.m. m, Percentage of damaged myofibrils in 10-day-old C. elegans in response to lifelong treatment with thymol. Box plots show the median, 25th and 75th percentiles, and minima and maxima. Ordinary one-way ANOVA followed by Sidak multiple comparisons. Data were pooled from two independent experiments, with 200 muscle fibers analyzed for each condition. NS, no significance. Source data

Fig. 7. Thymol delays biological aging, prevents sarcopenia and improves performance in mice with accelerated aging.

a, Schematic overview of chronic interventions with thymol in SAMP8 mice (n = 15 per group). b, Biological age measured in skeletal muscle through the quantification of epigenetic clock-specific methylation sites, expressed as biological age minus chronological age (BioAge − age) in SAMP8 mice (control, n = 10; thymol, n = 13). c, Representative immunofluorescence images of the tibialis anterior muscle stained with laminin. Scale bar, 20 µm. d, Frequency distribution of skeletal muscle fibers according to their cross-sectional area. e, Mean cross-sectional area (CSA) of tibialis muscle fibers. The violin plot shows the median (solid lines) and quartiles (dashed lines). f, Treadmill analysis at baseline and after 12 weeks of treatment (control, n = 11 mice; thymol, n = 13 mice). g, Treadmill analysis results expressed as the ratio of running distance at the end of the study to that at baseline (control, n = 11; thymol, n = 13). h, Grip strength measured at baseline and after 12 weeks of treatment (control, n = 12 mice; thymol, n = 13 mice). i, Grip strength expressed as the ratio of force between the end of the study and baseline (control, n = 12 mice; thymol, n = 13 mice). j, Gene set enrichment barcode plot for the TNF signaling via NF-κB gene set from the 3′ QuantSeq analysis of the gastrocnemius. k, Western blot analysis of skeletal muscle from mice treated with thymol or vehicle using the indicated antibodies. l, Densitometric quantification of the western blots. Each data point represents the densitometric value obtained from individual muscle lysates (control, n = 9; thymol, n = 10). Data points in the before–after plots in f and h represent values from each mouse at baseline and the end of the study (12 weeks). Box plots show the median, 25th and 75th percentiles, and minima and maxima. One-tailed Student’s t test. Panel a created with BioRender.com. Source data

Extended Data Fig. 1. Zebrafish autophagy reporter validation and high content screening.

a Graphical representation of the construct used to generate transgenic Tg(actc1b:ZsGreen-map1lc3;cryaa:TdTomato) zebrafish to monitor autophagy. b Schematic workflow for image analysis of transgenic zebrafish larvae. 20X (0.75 NA) objective Z stacks (26 images, 2 µm apart) of the bodies were acquired in FITC (536 nm) channel using automated microscope (ImageXpress, Molecular Devices) and a 60 µm pinhole spinning disk. c Image acquisition of transgenic zebrafish larvae in 96 microwell plate. Scale bars, 10 mm (top); 1 mm (bottom-left); 100 µm (bottom-right) d Western blot images showing time-course experiment of inhibition of mTORC1 activity in 3 dpf zebrafish larvae treated with 1 µM rapamycin. e Densitometric quantification of western blot. f Scatterplot showing autophagy flux in zebrafish larvae treated with 240 selected chemicals from TOCRIS library at 10 µM (3 dpf, n = 12 per compound). Dashed lines indicate 1-fold log2 change of LC3 relative area in presence of NH₄Cl. Green and orange dots represent known and novel hits respectively. Source data

Extended Data Fig. 2. Dose-response effects of thymol and carvacrol on autophagy.

a Representative images of transgenic zebrafish larvae at 3 dpf treated for 16 h with vehicle (1% DMSO), carvacrol and thymol at the indicated concentrations, followed by treatment with NH₄Cl for 4 h. Scale bar, 50 µm. b,c Quantification of autophagic flux in transgenic zebrafish larvae (3 dpf, n = 17–21). Data are shown as mean ± SEM. Two-way ANOVA followed by Sidak multiple comparisons. d Western blot analysis of zebrafish larvae at 3 dpf treated with indicated doses of thymol or 2 µM of rapamycin for phospho-S6 signal activation e Western blot analysis of P-AMPK immunoreactivity with same treatments as in (d). Source data

Extended Data Fig. 3. Characterization of thymol and carvacrol bioactivities in mammalian cells.

a Cytofluorimetric analysis of autophagy flux in Jurkat cells treated with vehicle (ctrl) or thymol for two hours and stained with LC3-II antibody. Representative LC3-II plots show fluorescence intensities in presence (upper panel) or absence (lower panel) of an inhibitor of lysosomal activity. b Time-course and dose-response cytofluorimetric analysis of cell viability in Jurkat cells treated with thymol and rapamycin (n = 1). c Quantification of mitochondrial membrane potential (MMP) in Jurkat cells treated with control (DMSO 0.1%), carvacrol or FCCP (100 nM) measured as ratio between monomeric (Em = 590 nm) and aggregate forms (Em = 525 nm) of JC-10 (n = 3 wells/condition). d Stimulation of ROS generation by thymol (100 µM), carvacrol (100) µM) and FCCP (5 µM) quantified by MitoSox red fluorescence signals (n = 3 wells/condition). Data in c and d are shown as means ± SEM. Ordinary one-way ANOVA followed by Sidak multiple comparisons. e Representative confocal microscopy images of Mouse Adult Fibroblasts (MAFs) treated with vehicle (0.1% DMSO), carvacrol (100 µM) and the uncoupler valinomycin (1 µM) in the presence of MitoTracker (green) and TMRM (red) to visualize changes in MMP. Scale bar, 50 µm. f Quantification of TMRM/MitoTracker area. Each data point in the boxplots represents one cell (control, n = 50, carvacrol, n = 56, valinomycin n = 50). One-way ANOVA followed by Sidak multiple comparisons. Box plots show median, the 25th and 75th percentiles, and minima and maxima. g Scatterplot of QuantSeq 3’ mRNA sequencing analysis in MAFs treated for 24 h with thymol (100 µM) or control (0.1% DMSO). Differentially regulated genes are highlighted in red (up-regulated) and blue (down-regulated); (control, n = 6; thymol, n = 6). h Unbiased gene set enrichment analysis in MAFs treated with vehicle and thymol. Gene expression analysis shows remaining significantly enriched gene sets following the top seven presented in Fig. 3g. Dotted lines indicate significance threshold.

Extended Data Fig. 4. Lack of effects of thymol on mitophagy activation in liver tissue of mitoQC mice.

a Transgenic construct to quantify mitophagy in mito-QC mice. b Representative confocal microscopy images of liver tissue 2 h hours after the second treatment. Note mosaic expression of the transgenic reporter proteins in hepatic tissue. Scale bar, 50 µm. c Quantification of mitophagy as mitophagy index. d Number of mCherry-positive mitophagy foci. For each mouse, ten fields were analyzed, with each data point in the boxplots c and d representing the value from an individual microscopy field. (control, n = 50; thymol, n = 50). Two-tailed Student’s t test. Box plots show median, the 25th and 75th percentiles, and minima and maxima. Panel a created with BioRender.com.

Extended Data Fig. 5. Effects of high fat diet and thymol treatment on body mass and insulin resistance in mice.

a,b Whole body composition analysis in mice by multi-echo magnetic resonance imaging (MRI). Percentages of fat mass (a) and lean mass (b) were measured at the end of the treatment. Boxplots show media, the 25th and 75th percentiles, and minima and maxima (n = 15). One-way ANOVA followed by Sidak multiple comparisons. c Cumulative food intake of mice fed with control and high fat diet and treated with thymol or vehicle only. d Oral glucose tolerance test (OGTT) performed after 7 weeks of intervention. Blood glucose levels were measured after glucose challenge at the indicated timepoints. e Blood insulin concentrations after glucose challenge at the indicated timepoints. Data show mean ± SEM (n = 15) and are analyzed by one-way ANOVA followed by Sidak multiple comparisons. P values in d and e are color coded: orange for high fat diet versus control and green for high fat diet and thymol versus control.

Extended Data Fig. 6. Effects of thymol and carvacrol on healthspan and lifespan of C. elegans.

a Fluorescence microscopy images of transgenic GFP::LGG-1 reporter C. elegans visualizing the activation of autophagy through quantification of green fluorescent foci in body wall muscle cells. Young (Day 1) and aged (Day 10) worms were treated with thymol and carvacrol at 25 µM compared to vehicle (DMSO, 0.1%). Duration of administration was lifelong treatment (left panels) or 72 h (right panels) prior to imaging. Arrows indicate GFP::LGG-1 positive punctae, indicative of autophagosomes. b-e Motility assays in Day 10 C. elegans treated with thymol or carvacrol at the indicated doses. Mean speed (b,c) and total distance travelled (d,e) were quantified to assess age-related decline in motility functions. At least 92 worms were analyzed for each condition. f Percentage of damaged myofibrils in Day 10 old C. elegans in response to lifelong treatment with thymol at indicated doses. Boxplots show median, the 25^th and 75^th percentiles, minima and maxima. Ordinary One-way ANOVA followed by Sidak multiple comparisons. Data are pooled from two independent experiments, with at least 200 muscle fibers analyzed for each condition. g-i Quantification of damaged myofibrils as function of fluorescence intensity in response to thymol or carvacrol at indicated concentrations. Data are pooled from two independent experiments, with at least 300 muscle fibers analyzed for each condition. Ordinary One-way ANOVA followed by Sidaj multiple comparisons. Box plots show median, the 25th and 75th percentiles, and minima and maxima. Violin plot shows median (solid lines) and quartiles (dashed lines). j,k Survival analysis of C. elegans exposed to indicated doses of thymol (j) and carvacrol (k) throughout lifespan. At least 50 worms per condition have been analyzed and compared with vehicle using log rank test.

Extended Data Fig. 7. Effects of thymol on age-related outcomes in SAMP8 mice.

a Weekly body weight analysis of mice from start to end of treatment (n = 15 mice/group). Mice that died were excluded from subsequent analyses b, c Average weights of tibialis anterior (b) and gastrocnemius (c) muscles (Control, n = 10; thymol, n = 13) d Spatial memory performance measured by Y maze at the baseline and after 12 weeks of treatment (Control, n = 11; thymol, n = 13). e Motor performance measured as latency to fall in a rotarod test (Control, n = 11; thymol, n = 13). Datapoints in the before-after plots in panels d and e represent values from each mouse at the baseline and at the end of the study (12 weeks). f Western blot analysis of total lysate extracts of skeletal muscle from mice treated with thymol or vehicle using the indicated antibodies. g Gene enrichment for tnfα signaling via nfkb pathway from 3’ QuantSeq analysis of MAFs shown as barcode plot. Box plots show median, the 25^th and 75^th percentiles, minima and maxima. Two-tailed Student’s t test. Source data