A TFEB-TGFβ axis systemically regulates diapause, stem cell resilience and protects against a senescence-like state

Abstract

Diapause is a long-lived state of resilience that allows organisms to outlast adversity. Caenorhabditis elegans can endure months in a fasting-induced adult reproductive diapause (ARD) and, upon refeeding, regenerate and reproduce. Here we find that mutants of ARD master regulator hlh-30/TFEB arrest in a senescence-like state during ARD and refeeding, in which germline stem cells are characterized by DNA damage, nucleolar expansion, cell cycle arrest and mitochondrial dysfunction, alongside dysregulated immune and growth metabolic signatures, elevated senescence-associated β-galactosidase and premature aging at the organismal level. Forward genetic screens reveal a TFEB-TGFβ signaling axis that systemically controls diapause, stem cell longevity and senescence, aligning nutrient supply to proper metabolism and growth signaling. Notably, TFEB's vital role is conserved in mouse embryonic and human cancer diapause. Thus, ARD offers a powerful model to study stem cell longevity and senescence in vivo, directly relevant to mammals.

Fig. 1. HLH-30 protects against a senescence-like state, enabling dormancy and stem cell longevity.

a, ARD lifespan of hlh-30(tm1978) and N2 WT animals. b, Brood size of self-fertilizing worms refed after 48 h of ARD. Genotypes N2, hlh-30(tm1978). Each circle represents the total progeny per worm. One representative experiment. Mann–Whitney test (two sided). Biological replicates (BR) = 3. c, Representative photomicrographs of germ cell nucleoli of N2, hlh-30(tm1978) at 96 h of ARD. Dashed circles denote germ cell nucleoli; arrowhead indicates a DTC. Scale bars, 10 μm. d, Distal gonad arms dissected from worms, refed at 48 h of ARD for 24 h, stained with DAPI (DNA, turquoise) and anti-phospho-histone H3 (M-phase chromosomes, magenta, arrows). Representative photomicrographs. Genotypes N2, hlh-30(tm1978). Scale bars, 10 μm. Arrowhead indicates a DTC. e, Photomicrographs of distal gonad arms of sygl-1::3xFlag worms in N2 and hlh-30(tm1978) backgrounds, stained with DAPI (nucleus, gray), anti-RAD-51 (DNA damage foci, magenta) and anti-3xFLAG (SYGL-1, GSC zone, yellow) antibodies. Following 48 h of ARD, worms were dissected after 24 h of refeeding. Scale bars, 10 μm. f, Three-dimensional (3D) quantification of mitochondrial sphericity in GSCs at 48 h of ARD and 48 h of refeeding. Genotypes N2, hlh-30(tm1978). Each dot represents the mitochondrial sphericity of one individual worm. Pooled data of two independent biological replicates. Two-way analysis of variance (ANOVA) followed by Tukey’s post hoc test. g, Representative images of mitochondrially produced ROS using CM-H₂TMROS staining after 48 h in ARD. Scale bars, 10 μm. h, Representative images of SA-β-gal staining after 48 h of ARD. Dashed circles show the two pharyngeal bulbs. Scale bars, 10 μm. i,j, Pharyngeal pumping (i) and body bending (j) rates of 48-h ARD worms. 30-s intervals. Each dot represents the pumping rate or body bends of one animal. Mann–Whitney test. BR = 3, one representative experiment. k, Percentage reproductive worms refed after 10 days of ARD. The genotypes are shown in Table 1. Each circle represents the percentage from one biological repeat. Mann–Whitney test (two sided). Survival curves depict one experiment. All data and statistics are presented in Supplementary Table 2. If not stated otherwise, data are the mean ± s.d. NS, not significant. Source data

Fig. 2. Selection against ARD senescence-like state reveals longevity and stem cell resilience pathways.

a, Genetic screen for mutations that rescue hlh-30(tm1978) ARD survival and recovery. The red asterisk indicates a mutation induced by EMS. b, Gene structure of hits from TGFβ (daf-1/TGFβ receptor, daf-3/SMAD4), cGMP (tax-4/CNGA2) and insulin/IGF signaling (daf-2/Insulin/IGF receptor, pdk-1/PDPK1 and akt-1/AKT). Schematic intron–exon map based on WormBase gene structures. Positions of point mutation identified in this screen are marked by red arrowheads. Amino acid alterations are given in single-letter code. Reference alleles used to validate the gene hits are in black. c, TGFβ receptor mutation daf-1(m40) prolongs ARD survival of hlh-30(tm1978). d, Representative images of worms at 96 h of ARD. Genotypes N2, hlh-30(tm1978), hlh-30 daf-1(m40). Scale bars, 10 μm. e, Percentage of reproductive worms refed after 10 days in ARD. Genotypes N2, hlh-30(tm1978), hlh-30 daf-1(m40) and hlh-30;daf-2(e1370). Each dot represents one experiment. One-way ANOVA followed by Tukey’s post hoc test. f, Simplified TGFβ signaling schematic. Under ad libitum conditions, neuronal DAF-7/TGFβ induces a DAF-1/TGFβ receptor signaling cascade inhibiting repressor activity of DAF-3/SMAD4–DAF-5/SNO-SKI complex in DTC and other target tissues. Under diapause conditions, TGFβ signaling is low, and the DAF-3–DAF-5 repressor complex is activated. Below, dauer phenotypes at 25 °C of loss-of-function (lf) mutants; Daf-c, dauer formation constitutive; Daf-d, dauer formation defective. g, hlh-30 daf-1 ARD lifespan extension depends on daf-3, daf-5 and daf-16 encoded transcription factors. Genotypes hlh-30(tm1978), hlh-30 daf-1(m40), hlh-30 daf-1;daf-3(e1376), hlh-30 daf-1;daf-5(e1386) and hlh-30 daf-1;daf-16(mgDf50) are all loss of function. h, daf-3(syb2718) gain-of-function mutation extends hlh-30 ARD survival. Survival curves depict one representative experiment. Data and statistics are presented in Supplementary Table 2. If not stated otherwise, data are the mean ± s.d. Source data

Fig. 3. HLH-30 regulates TGFβ signaling in response to nutrient cues.

a, hlh-30::mNeonGreen (hlh-30::mNG) expression pattern in ASI sensory neurons at 2 h of ARD and upon 1 h of refeeding 48-h ARD worms. Differential interference contrast (DIC) and fluorescence images of the head region. The arrow indicates an ASI neuron. Ph, terminal pharyngeal bulb. Scale bars, 10 μm. b, Representative images of daf-7p::GFP expression in the ASI neuron at 96 h of ARD and 48 h of refeeding after 96 h ARD. Fluorescence and DIC images of head regions. The arrow indicates an ASI neuron. Scale bars, 10 μm. c, Quantification of daf-7p::GFP expression in ASI neurons at 48, 72 and 96 h of ARD (80-ms exposure time) and after 48 h of refeeding (40 ms exposure time). Each dot represents daf-7p::GFP expression of one ASI neuron. BR = 3, one representative experiment shown. d, Quantification of whole-body daf-1p::GFP at 48 h of ARD and 48 h of refeeding measured with the COPAS Biosorter. Mean ± s.e.m., each dot represents one experiment. Two-way ANOVA followed by Fisher’s least significant difference post hoc test. e, Representative images of daf-1p::GFP expression in the DTC at 48 h of ARD and 48 h of refeeding. Fluorescence and DIC images of the distal gonad. The arrow indicates the DTC nucleus. Scale bars, 10 μm. f, Percentage of DTC positive for daf-1p::GFP expression of N2 and hlh-30(tm1978) at 48 h of ARD and after 48 h refeeding. BR = 4. One-tailed Mann–Whitney to compare the percentage of daf-1::GFP-positive DTC within ARD or refeeding and one-way ANOVA followed by Tukey’s post hoc test to compare genotypes across ARD and refeeding conditions. g, ChIP enrichment of HLH-30 binding to promotor regions of canonical TGFβ pathway genes. For each gene, one HLH-30 binding site (green) and the corresponding 3′ untranslated region (UTR) control (gray) was tested in this analysis. Data were normalized to 3′ UTR binding (dashed line). Mann–Whitney statistics (two sided) were performed on the non-normalized ∆∆c_t values. One dot represents one biological repeat. BR = 4. h, Normalized mRNA expression from RNA-seq data of daf-14 and daf-5 in 48-h ARD comparing WT and hlh-30(tm1978) mutants. Data and statistics are presented in Supplementary Table 2. If not stated otherwise, data are the mean ± s.d. Two-tailed Mann–Whitney test. Source data

Fig. 4. Regulation of growth signaling protects against a senescence-like state.

a, Hermaphrodite distal gonad schematic. Germ cells in gray. The progenitor zone contains a distal pool of stem cells (area of sygl-1 expression, yellow) and a proximal pool of differentiating cells. Forward growth signaling induces LAG-2 secretion from the DTC to GSCs, initiating mitotic cell division and sygl-1 expression via GLP-1. SYGL-1 abundance declines with increasing distance to the DTC. b, Quantification of lag-2p::mKate2::PH expression intensity in the DTC cap at 48 h of ARD. Each dot represents lag-2p::mKate2::PH expression in one DTC per worm. c, lag-2p::mKate2::PH expression in DTC at 48 h of ARD and after 48 h refeeding. Fluorescence images. Contrast and intensity were adjusted to visualize DTC processes. Scale bars, 10 μm. d, Quantification of SYGL-1 abundance in the distal gonad of 48-h-ARD worms and of 24-h refed worms, based on intensity of α-FLAG staining. Average intensity values (y axis) were plotted against distance (μm) from the DTC (x axis). Lines indicate mean intensity; shaded areas are the s.e.m.; BR = 3, one representative experiment. e, Quantification of the area under the curve (AUC) of SYGL-1 expression. AUCs were normalized to the N2 AUC for each biological replicate. One-tailed Mann–Whitney test. BR = 3. f, Quantification of germ cell nucleolar area at 96 h of ARD. Each dot represents the nucleolar area in μm² of one germ cell per worm. BR = 3, one representative experiment. g, Number of M-phase cells in the distal gonad arms (progenitor zone) at 48 h of ARD and 24 h of refeeding. BR = 3, one representative experiment. One dot indicates the number of M-phase-positive cells per gonad arm. h, Quantification of GSCs positive for RAD-51 foci in the sygl-1-positve area per gonad. One dot represents the number of positive GSCs per gonad arm per worm. Pooled from three BRs. i, Biological age prediction (BiT) from transcriptomes at 48 h of ARD. Each point represents one replicate. j, Correlation plot of hlh-30(tm1978)/N2 and hlh-30 daf-1(m40)/hlh-30 DEGs at 48 h of ARD. Significantly regulated genes (adjusted P < 0.05) highlighted in red (genes down in hlh-30 and reversed by daf-1, 2,347 DEGs, quadrant 2) or blue (genes up in hlh-30 and reversed by daf-1, 2,083 DEGs, quadrant 4). Simple linear regression line in gray. Equation Y = −0.48X + 0.13; R² = 0.27. k, Working model. In WT ARD HLH-30/TFEB is active and downregulates TGFβ signaling at multiple levels. Consequently, Notch signaling is inhibited via LAG-2/DSL downregulation in the DTCs, resulting in GSC quiescence. Upon refeeding, HLH-30 becomes inactive, resulting in reactivation of TGFβ and downstream signaling. Subsequently, stem cell niche remodeling (DTC outgrowth) is initiated, GSCs proliferate and worms reproduce. cGMP and insulin–IGF signaling (IIS) work upstream or in parallel to TGFβ signaling. Data and statistics are presented in Supplementary Table 2. If not stated otherwise, data are the mean ± s.d. One-way ANOVA followed by Tukey’s post hoc test. Genotypes N2, hlh-30(tm1978), hlh-30 daf-1(m40). Source data

Fig. 5. TFEB controls resilience in mammalian diapause models.

a, sgRNAs counts for mammalian TFEB in the four analyzed conditions: control proliferating mES cells, control proliferating mES cells with doxycycline to induce the Cas9 activity, mES cells in diapause (by treatment with INK128) and mES cells in diapause treated with doxycycline. BR = 3. Two-way ANOVA followed by Tukey’s post hoc test. b, Effect of TFEB silencing on the viability of mES cells, when proliferating and in the diapause-like state. The diapause-like state was induced by treatment with INK128. Each column is normalized on its respective siRNA scrambled. BR = 3. Two-tailed unpaired t-test. c, Effect of TFEB silencing on the viability of SK-Mel-147 cells, when proliferating and in the diapause-like state. The diapause-like state was induced by treatment with INK128. Each column is normalized on its respective siRNA scrambled. BR = 3. Two-tailed unpaired t-test. d, TFEB expression levels from RNA-seq datasets of proliferating compared to diapause-like SK-Mel-147 cells. BR = 4. Two-tailed unpaired t-test. e, Gene-set enrichment analysis enrichment plot of the TGFβ signaling pathway genes in diapause-like SK-Mel-147 compared to proliferating SK-Mel-147. Statistical testing was performed using Kolmogorov–Smirnov-like statistics and adjusted for multiple testing using the Benjamini–Hochberg false discovery rate method. Data and statistics are presented in Supplementary Table 2. NES, normalized expression score. Source data

Extended Data Fig. 1. HLH-30 protects against cellular senescence, enabling dormancy and stem cell longevity.

(a) Number of M-phase cells in the distal gonad arms (progenitor zone) of wild type (N2) recovered for 24 hours from 4 or 40 days of ARD. The salmon color indicates worms without M-phase positive cells. Pooled data from 2 independent experiments. (b) Percentage of N2 gonad arms negative for M-phase positive cells (salmon color) upon recovery for 24 hours from 4 or 40 days of ARD. BR = 2. (c) Quantification of germ cell nucleolar areas of N2 worms at 4 or 40 days of ARD. BR = 3. One representative experiment. Each dot represents one nucleolus. (d) Photomicrographs of distal gonad arms of sygl-1::3xFlag worms in N2 background, stained with DAPI (nucleus), anti-3x FLAG (SYGL-1, GSC zone, yellow) and anti-RAD-51 (DNA damage foci, magenta) antibodies. Following 1 hour post 60 Gy irradiation, ad libitum worms were dissected and stained. Scale bar 10 μm. The distal end of the gonad is facing the left side of the image. (e) Representative images of mKate2 labeled mitochondria in GSCs at 48 hours of ARD and 48 hours of refeeding. Images show a single z-layer of the germline, fluorescence and DIC images. Genotypes pie-1p::tomm-20mKate in N2 and hlh-30(tm1978) background. Dashed lines mark the most distal GSC area (30 μm) used for image analysis. Arrowhead show the distal end of the gonad. Scale bar 10 μm (f) Representative photomicrographs of head regions of 48 hour ARD worms stained with SA-β-gal. Genotypes N2, hlh-30(tm1978). 4 worms per genotype. Scale bar 10 μm. (g) SA-β-gal RGB color intensity of head regions of young adult hlh-30(tm1978) worms normalized to N2. Staining control is RNAi again bgal-2. Each dot represents the SA-β-gal intensity of one worm. Pooled data of three biological repeats. One-way ANOVA followed by Turkey’s post-hoc test. (h) Precent mean brood size change of autophagy and lysosomal mutants under ad libitum (grey) or refeeding after 10 days in ARD (blue). Percent change was determined in respect to N2 or him-5 controls. Genotype alleles can be found in the reagent table. Each dots reflect one biological replicate. Kruskal-Wallis followed by Dunn’s post-hoc test. If not stated otherwise: Mean & s.d., Mann-Whitney test (two-sided). Source data

Extended Data Fig. 2. Selection against ARD senescence reveals ARD longevity and stem cell resilience pathways.

(a), (b), (c) Insulin/IGF signaling mutants daf-2(e1370), daf-2(e1368) and pdk-1(sa680) extend ARD lifespan and partially rescue hlh-30(tm1978) ARD short survivorship. (d) TGFβ signaling mutant daf-7(e1372) extends ARD survival and partially rescues hlh-30(tm1978) ARD short survivorship. (e) Body length of ARD worms at 96 hours. Genotypes N2, hlh-30(tm1978), hlh-30 daf-1(m40), and hlh-30;daf-2(e1370). Each dot represents one animal. BR = 3, one representative experiment. One-way ANOVA followed by Turkey’s post-hoc test. (f) Predicted protein structure (Alphafold) of daf-3(syb2718) showing mutation E853K in the MH2 domain in red. Conserved Mad homology 1 and 2 domains (MH1 and MH2) in blue and yellow, respectively. (g) daf-3(syb2718) gain-of-function mutation restores progeny production of hlh-30 worms refed at days 10 ARD. daf-3(e1376) null mutant abolishes progeny production of hlh-30(tm1978) daf-1(m40) double mutants. Each dot represents the progeny number of one worm. One representative replicate. BR = 3. One-way ANOVA followed by Turkey’s post-hoc test. (h) daf-3(syb2718) exhibits a dominant gain-of-function dauer constitutive (Daf-c) phenotype at 25 °C. Genotypes N2, daf-3(syb2718), and daf-3(syb2718)/+ heterozygous. Each dot represents one experiment. One-way ANOVA followed by Turkey’s post-hoc test. (i) daf-12(rh61rh411) slightly enhances ARD survival of hlh-30(tm1978) daf-1(m40). Survival curves depict one representative experiment. Lifespan data and statistics (log-rank tests) are presented in Supplementary Table 2. If not stated otherwise: Mean & s.d. Source data

Extended Data Fig. 3. HLH-30 regulates TGFβ signaling in response to nutrient cues.

(a) Nuclear/cytosolic hlh-30::mNeonGreen expression in ASI neurons under ARD (2 and 48 hours) and recovery from 48 h ARD (1, 4, 6, and 24 hours of refeeding). Each dot represents one experiment. BR = 3. Mean ± SEM (b) Quantification of daf-7p::GFP fluorescence in OLQ neurons of N2 and hlh-30(tm1078) at 96 hours of ARD. Exposure time 400 ms. Expression in OLQ is 5–10-fold lower than in ASI. Each dot represents the daf-7p::GFP expression in one OLQ neuron. BR = 3, one representative experiment. Mann-Whitney test (two-sided). (c) Schematic of ChIP-qPCR experimental design. (d) Control western blot ChIP-qPCR experiment showing specific elution of HLH-30::FLAG (BR = 2). ‘+’ indicates the use of an Anti-FLAG antibody during IP, ‘-‘ indicates no use of Anti-FLAG antibody (bait). Tubulin control staining was performed on the same gel after washing off the Anti-FLAG antibody. If not stated otherwise: Mean & s.d., One-way ANOVA followed by Turkey’s post-hoc test. Source data

Extended Data Fig. 4. Regulation of growth signaling protects against senescence.

(a) Distal tip cell (DTC) extension measured from the cap to the end of the most distal process containing lag-2p::mKate2::PH. Each dot represents the mean DTC length of one BR. (b) Representative images of SYGL-1 stained and DAIP labeled germline stem cells (GSCs) at 48 hours of ARD and 24 hours of refeeding. Images show a single z-layer of the germline. Genotypes N2 and hlh-30(tm1978). Arrowhead show the distal end of the gonad. Scale bar 10 μm. (c) Mean SA-β-gal RGB color intensity in the head region of 48 h ARD worms normalized to N2. Genotypes N2, hlh-30(tm1978), hlh-30 daf-1(m40). Each dot represents the SA-β-gal mean intensity of one BR. (d) Mean pharyngeal pumping and (e) Mean body bending rates upon 48 hours of ARD. Genotypes: N2, hlh-30(tm1978), hlh-30 daf-1(m40). 30 second intervals. Each dot represents the mean pumping rate or mean body bends of one BR. (f) Quantification of mitochondrial ROS in N2, hlh-30(tm1078), hlh-30 daf-1(m40) at 48 hours of ARD. Pooled data from three BRs. Kruskal-Wallis followed by Dunn’s post-hoc test. If not stated otherwise: Mean & s.d., One-way ANOVA followed by Turkey’s post-hoc test. Source data

Extended Data Fig. 5. Downregulation of growth signaling partially reverses senescence transcriptome.

(a) Correlation plot of differentially expressed genes (DEGs) of hlh-30/N2 and hlh-30 daf-1/hlh-30 at 12 hours of refeeding (REF). DEGs (padj < 0.05) highlighted in red (genes down in hlh-30 and reversed by daf-1, quadrant 2) or blue (genes up in hlh-30 and reversed by daf-1, quadrant 4). Other DEGs, gray. Simple linear regression line in gray. Equation Y = −0.35X - 0.06; R² = 0.27. (b) Gene ontology enrichment analysis of 2347 DEGs down in hlh-30(tm1978)/N2 ARD and up in hlh-30 daf-1(m40)/hlh-30 ARD (padj < 0.05, quadrant 2). Top 10 biological processes of the DAVID GO BP DIRECT database. Complete GO term list is shown in Supplementary Table 3. Modified Fisher Exact p value, EASE Score. (c) Gene ontology enrichment analysis of 2083 DEGs up in hlh-30/N2 ARD and down in hlh-30 daf-1/hlh-30 ARD (quadrant 4). Top 10 biological processes. Complete GO term list is shown in Supplementary Table 3. Modified Fisher Exact p value, EASE Score. (d) Gene ontology enrichment analysis of 1499 DEGs down in hlh-30/N2 and up in hlh-30 daf-1/hlh-30 (padj<0.05, quadrant 2) at 12 hours of refeeding (REF). Top 10 biological processes. Modified Fisher Exact p value, EASE Score. (e) Gene ontology enrichment analysis of 2173 DEGs up in hlh-30/N2 and down in hlh-30 daf-1/hlh-30 (padj<0.05, quadrant 4) at 12 hours of refeeding. Top 10 biological processes. Modified Fisher Exact p value, EASE Score. (f) Heat map of log2(FC) values of differentially expressed Insulin-like peptides in hlh-30/N2 at 48 hours of ARD and upon 12 hours of refeeding. padj< 0.05. Source data

Extended Data Fig. 6. TFEB controls resilience in mammalian diapause models.

(a) Representative images of SK-Mel-147 cells treated with siTFEB or siNT control in diapause, or 6 days post INK128 wash out (w/o) either at day 0 of siRNA treatment or 72 hours post siRNA transfection. Scale bar: 100 µm (b) Heat map of wild type killifish tfeb, smad2, smad3b, and smad4b expression from published RNA seq dataset at 9 months of diapause or developing embryos 6 days post fertilization. z-score. BR = 4. (c) Heat map of wild type killifish tfeb, smad4(1 and 2 of 3) at 3 and 6 days and 1 months of diapause or developing embryos before and without diapause. z-score. BR = 3. (d) Heat map of wild type killifish tfeb, smad2, smad3b and smad4 expression from adipose tissues of young and old fasted and refed fish. RNA seq samples. z-score. BR = 4. ns = not significant. Source data