Computer prediction and genetic analysis identifies retinoic acid modulation as a driver of conserved longevity pathways in genetically diverse Caenorhabditis nematodes
- Institute of Ecology and Evolution, University of Oregon, United States
- Rutgers University, Department of Molecular Biology and Biochemistry, United States
- The Buck Institute for Research on Aging, United States
Figures
Figure 1 with 1 supplement
Summary of lifespan effects for candidate compounds.
(A) Compounds were selected for testing by filtering the top 10% of predicted hits from Fuentealba et al., 2019 and cross-referencing for compounds that also appeared in the top 10% of Barardo et al., 2017a, or Janssens et al., 2019, or had been shown to work in other model organisms via the DrugAge database. Candidate compounds were then filtered using the DrugAge database and a literature search to deprioritize compounds previously characterized as extending lifespan in C. elegans to generate a list of 16 compounds for screening using lifespan analysis. (B) Percent difference in median lifespan of individual trial plates compared to the median survival from their pooled carrier control (DMSO and H2O) for animals treated with one of 16 candidate compounds selected for preliminary analysis. The dot represents the mean of all plate replicates across two trials, and the bars represent the standard error. Shown are the results from the longest-lived concentration treatment (4–5 concentrations were tested) for each candidate compound. The shown p-values and error bars are taken from the hierarchical CPH model (see Materials and methods).
Figure 1—figure supplement 1
Longevity analysis screen in C. elegans N2 of 14 candidate compounds that were not selected for further study.
Four compounds (fisetin, berberine, aldosterone, and ritonavir) showed a positive but non-robust lifespan effect. Seven (decitabine, dasatinib, erlotinib, dexamethasone, temsirolimus, everolimus, and thalidomide) had no significant effect on survival, and three compounds (gefitinib, arecoline, and bortezomib) showed toxic effects. Each compound was assayed at 4–5 concentrations, with the upper limit defined by solubility of the compound or toxicity. The Kaplan–Meier curves presented consist of pooled replicates from two trials. Asterisks represent p-values from the CPH model such that ****p < 0.0001, ***p < 0.001, **p < 0.01, and *p < 0.05.
Figure 2 with 2 supplements
Dose-dependent lifespan effects of all-trans retinoic acid and propranolol across diverse Caenorhabditis species.
(A–C) Manual lifespan analysis of five concentrations of atRA (black – DMSO control, increasing levels of teal 1–150 µM) and (D–F) propranolol (black – H2O control, increasing levels of pink – 50–5000 µM) on three Caenorhabditis species. The upper limit tested was determined by compound solubility (atRA), or toxicity (propranolol). The Kaplan–Meier curves presented consist of pooled replicates from two trials. Asterisks represent p-values from the CPH model such that ****p < 0.0001, ***p < 0.001, **p < 0.01, and *p < 0.05.
Figure 2—figure supplement 1
Median lifespan dose response of all-trans retinoic acid and propranolol in C. elegans N2.
Percent difference in median lifespan of an individual trial plate (1–150 µM atRA, 50–5000 µM propranolol) compared to its specific control (DMSO control for atRA or H2O for propranolol) for C. elegans N2 (data also presented as survival curves in Figure 2A, D). Each point represents a single plate replicate with approximately 50 worms. The error bars represent the mean ± the standard error of the mean. Error bars and p-values are from the hierarchical CPH model, with ****p < 0.0001, ***p < 0.001, **p < 0.01, and *p < 0.05.
Figure 2—figure supplement 2
Metoprolol does not extend lifespan in C. elegans, nor does propranolol on PFA-treated OP50-1, consistent with its bacteriostatic activity.
(A) Manual lifespan analysis of five concentrations of metoprolol (black – H2O control, increasing levels of blue – 50–1500 µM) on C. elegans N2. (B) Manual lifespan analysis of C. elegans N2 on propranolol (black – H2O control, increasing levels of pink – 500 and 1000 µM) using paraformaldehyde treated OP50-1 E. coli. Kaplan–Meier curves consist of pooled replicates from two trials. Asterisks represent p-values from the CPH model such that ****p < 0.0001, ***p < 0.001, **p < 0.01, and *p < 0.05. (C) Colony forming units of E. coli cultures (100 µl at a 1 × 10–6 dilution) on LB plates treated with H2O, 500, 1000, and 5000 µM propranolol.
Figure 3 with 3 supplements
The vitamin A derivative atRA extends life in a species-specific manner.
The effect of adult exposure to 150 µM atRA on median survival in manual lifespan assays. Three strains were tested from each of three species: C. elegans strains N2, JU775, and MY16, C. briggsae AF16, ED3092, and HK104, and C. tropicalis strains JU1630, JU1373, and QG834. Each point represents the percent change in median survival for an individual trial plate relative to the vehicle control median. The bars represent the mean ± the standard error of the mean. Replicates were completed at the three CITP testing sites (circles – Oregon, squares – Buck Institute, and diamonds – Rutgers). Error bars and asterisks represent p-values from the hierarchical CPH model such that ****p < 0.0001, ***p < 0.001, **p < 0.01, and *p < 0.05.
Figure 3—figure supplement 1
Lifespan curves for the data presented in Figure 3 – manual survival of nine Caenorhabditis strains with 150 µM atRA treatment.
Lifespan analysis for DMSO control (black) and 150 µM (dark teal) atRA treatment starting at day 1 of adulthood. Three C. elegans (N2, JU775, MY16), three C. briggsae (AF16, ED3092, HK104), and C. tropicalis (JU1630, JU1373, QG834) strains were tested. Kaplan–Meier curves represent pooled replicates from multiple trials at each of the three CITP testing sites (Oregon, Buck Institute, Rutgers). Asterisks represent p-values from the CPH model such that ****p < 0.0001, ***p < 0.001, **p < 0.01, and *p < 0.05.
Figure 3—figure supplement 2
Survival of nine CITP strains with atRA treatment on the ALM.
Longevity analysis for control (black) and 150 µM (dark green) atRA. Three C. elegans (N2, JU775, MY16), three C. briggsae (AF16, ED3092, HK104), and C. tropicalis (JU1630, JU1373, QG834) strains were tested. The longevity analysis was performed using the Lifespan Machine technology (Stroustrup et al., 2013) in a single lab. Kaplan–Meier curves represent pooled replicates from at least three trials. Asterisks represent p-values from the CPH model such that ****p < 0.0001, ***p < 0.001, **p < 0.01, and *p < 0.05.
Figure 3—figure supplement 3
Swimming ability of the nine CITP strains with atRA treatment.
The C. elegans Swim Test (CeleST) was used to assess swimming ability in all nine strains as a locomotory measure of healthspan. Eight measures of swimming ability were combined to create a single composite measure, the adjusted swimming score. Each dot represents the mean adjusted swimming score for a single trial at one of the three CITP testing sites (circle – Oregon, square – The Buck Institute, diamond – Rutgers). Bars indicate the mean ± the standard error of the mean. Error bars and p-values are from the linear mixed model, with ****p < 0.0001, ***p < 0.001, **p < 0.01, and *p < 0.05.
Figure 4 with 1 supplement
Genetic analysis of atRA effects on lifespan.
Lifespan analysis under 150 µM atRA (green) or vehicle control (black). For wildtype, the canonical CITP N2 strain was used. This response was compared to loss of function/downregulation of the Akt/protein kinase B (PKB) homologs (A) akt-1 and (B) akt-2, (C) the AMP-activated protein kinase aak-2, (D) the heat shock transcription factor homolog hsf-1, (E) the Nrf transcription factor homolog skn-1, (F) the FOXO transcription factor homolog daf-16, (G) the p38 MAP kinase homolog pmk-1, (H) the toll-like receptor tol-1, and (I) the acetylcholine receptor eat-2. For skn-1, RNAi knockdown was used because of the lethality of the mutant (see Figure 4—figure supplement 1B for control RNAi experiment). For all other genes, loss of function mutants were used. Kaplan–Meier curves include pooled replicates from two trials, except for the RNAi experiment, which consisted of three trials. Asterisks represent p-values from the CPH model such that ****p < 0.0001, ***p < 0.001, **p < 0.01, and *p < 0.05.
Figure 4—figure supplement 1
SKN-1 regulation and skn-1 RNAi lifespan.
(A) Post-transcriptional regulation of SKN-1 via phosphorylation is mediated by the pmk-1/p38 MAPK, PI3K/Akt, and GSK kinase pathways. Each pathway regulates phosphorylation of pathway-specific serines of SKN-1, enabling additive regulation. (B) Kaplan–Meier lifespan curves of worms grown on E. coli HT115 expressing either an empty vector control (solid) or skn-1 RNAi (dashed) and treated with either 150 µM atRA (dark green) or the vehicle control (black). Each curve represents pooled replicates from three independent trials. Asterisks represent p-values from the CPH model such that ****p < 0.0001, ***p < 0.001, **p < 0.01, and *p < 0.05.
Figure 5 with 2 supplements
Altered transcriptome under atRA treatment.
(A) Volcano plot for gene expression from RNA-seq experiments performed on day 4 of adulthood exposed to 150 µM atRA or vehicle control. (B) Enrichment analysis using WormCat (Higgins et al., 2022; Holdorf et al., 2020).
Figure 5—figure supplement 1
Collagens are downregulated in response to atRA exposure.
Using a previously published in silico analysis of the C. elegans matrisome (i.e., all proteins that make up the extracellular matrix; Teuscher et al., 2019), we extracted all collagen and collagen-related genes with expression data from our RNA-seq dataset. For easier presentation, we used the in silico analysis grouping of core matrisome genes (e.g., collagens, proteoglycans, glycoproteins) and matrisome associate genes (ECM-affiliated proteins like C-type lectins, galectins, annexins, and ECM regulators like MMPs, ADAMs, and crosslinking enzymes, and secreted factors like BMPs, FGFs, and chemokines [see Teuscher et al., 2019 and supplemental data, Caenorhabditis Intervention Testing Program, 2025 for a full list]). The genes were further divided into genes for which mammal to C. elegans orthology could be established, and those that were nematode specific.
Figure 5—figure supplement 2
Effects of atRA on transcription of sphingolipid metabolism genes.
(A) The C. elegans sphingolipid metabolism network with genes significantly (FDR <0.05) downregulated (red) or upregulated (blue), noted by color. (B) The log fold change for the genes and a (C) volcano plot showing the genes with their respective FDRs. All data available in supplemental information (Caenorhabditis Intervention Testing Program, 2025).
Figure 6 with 1 supplement
Analysis of the role of hsf-1 and aak-2 in atRA transcriptional response.
(A) Comparison of DE genes with FDR <0.5 and |LFC| >1 in the hsf-1 background to the changes observed in the wildtype background. The gray dashed line shows the expected relationship if the mutation had no effect on atRA response, while the orange dashed line shows the fit to the observed data. (B) Comparison of the genome-wide atRA-induced change in expression for all genes detected in both the N2 and hsf-1(sy441). (C) Plot of the FDR for the WaRGs detected in the hsf-1(sy441) background by the difference in expression changes between WT and mutant background normalized to the change observed in WT animals. (D) Classification of WaRGs as maintaining, weakening, losing, or flipping their response in hsf-1(sy441) animals. (E) Comparison of the genome-wide atRA-induced change in expression for all genes detected in both the N2 and aak-2(ok524) datasets. The gray dashed line shows the expected relationship if the mutation had no effect on atRA response, while the blue dashed line shows the fit to the observed data. (F) Comparison of DE genes with FDR <0.5 and |LFC| >1 in the aak-2(ok524) background to the changes observed in the wildtype background. (G) Plot of the FDR for the WaRGs detected in the aak-2(ok524) background by the difference in expression change between WT and mutant background normalized to the change observed in WT animals. (H) Classification of WaRGs has maintaining, weakening, losing, or flipping their response in aak-2(ok524) animals.
Figure 6—figure supplement 1
The WormCat enrichment analysis for the WaRGs whose response was maintained, weakened, or lost in hsf-1(sy441) or aak-2(ok524) animals.
Tables
Table 1
CITP tested compounds that meet computational prediction selection criteria of this study.
| Candidate | Predictions | CITP publication | Beneficial effect on median survival in C. elegans in CITP? | Pathway/mode of action |
|---|---|---|---|---|
| Bortezomib | F + B | This study | NP | Proteasome inhibitor (Chen et al., 2011) |
| Metformin | F + J | Onken et al., 2022 | 41% increase at 70 mM | Anti-diabetes |
| 17-Alpha estradiol | F + J | Banse et al., 2024b | NP | Estrogen receptor agonist |
| Rapamycin | F + J | Banse et al., 2024b | NP | mTOR inhibitor |
| Aspirin | F + J | Lucanic et al., 2017 | NP | NSAID |
| Imatinib | F + J + B | Coleman-Hulbert et al., 2019 | NP | Tyrosine kinase inhibitor |
| atRA | F + J | This study | 29% extension at 150 µM | Collagen formation, activates xenobiotic metabolism |
| Dasatinib | F + B | This study | NP | Tyrosine kinase inhibitor |
| Temsirolimus | F + B | This study | NP | mTOR inhibitor |
| Gefitinib | F + B | This study | NP | EGFR inhibitor (tyrosine kinase inhibitor) |
| Resveratrol | F + J | Lucanic et al., 2017 | 12% extension at 100 µM | Sirtuin activator |
| Berberine | F + B | This study | 16.3% extension at 100 µM | AMPK activator |
| Decitabine | F + B | This study | NP | Nucleic acid synthesis inhibitor |
| Erlotinib | F + B | This study | NP | EGFR inhibitor (tyrosine kinase inhibitor) |
| Valproic acid | F + J | Lucanic et al., 2017 | NP | Blocks sodium-gated ion channels, increases GABA |
| Aldosterone | F + B | This study | Significant at 50 µM due to late life effects, no change in median lifespan | Steroid hormone |
| Dexamethasone | F + B | This study | NP | Anti-inflammatory corticosteroid |
| Propranolol | F + J | This study | 44% extension at 1 mM* | Beta-blocker |
| Thalidomide | F + J | This study | NP | TNF-a inhibition |
| Arecoline | F + J | This study | NP | Muscarinic agonist (inhibits pharyngeal pumping) |
| Ritonavir | F + B | This study | 4.1% extension at 20 µM | HIV protease inhibitor: inhibits enzymes that normally metabolize other protease inhibitors (primarily in intestines, liver, etc.) |
| Fisetin | F + D | This study | 11.7% extension at 50 µM | Sirtuin activator |
| Everolimus | F + D | This study | NP | mTOR inhibitor |
-
*
May be an indirect effect. NP = no positive effect detected. F – in the top 10% of Fuentealba et al., 2019, B – in the top 10% of Barardo et al., 2017a, J – in the top 10% of Janssens et al., 2019, and D – a positive listing in DrugAge Barardo et al., 2017b. Compounds denoted in bold were tested by the CITP for this study, while other listed compounds were tested previously by the CITP.
Key resources table
| Reagent type (species) or resource | Designation | Source or reference | Identifiers | Additional information |
|---|---|---|---|---|
| Strain, strain background (C. elegans) | See strain list in Methods | This paper | Available at Caenorhabditis Genetics Center | |
| Strain, strain background (C. briggsae) | See strain list in Methods | This paper | Available at Caenorhabditis Genetics Center | |
| Strain, strain background (C. tropicalis) | See strain list in Methods | This paper | Available at Caenorhabditis Genetics Center | |
| Software, algorithm | Lifespan (R script) – ATRA | 10.6084/m9.figshare.26308177 | ||
| Software, algorithm | Lifespan (R script) – compound screen | 10.6084/m9.figshare.26308153 | ||
| Software, algorithm | Lifespan (R script) – propranolol PFA-killed OP50-1 | 10.6084/m9.figshare.26308159 | ||
| Software, algorithm | Lifespan (R script) – pathway mutants | 10.6084/m9.figshare.26308165 | ||
| Software, algorithm | Lifespan (R script) – ATRA automated lifespan | 10.6084/m9.figshare.26308186 | ||
| Software, algorithm | CeleST (R script) | 10.6084/m9.figshare.26308198 | ||
| Software, algorithm | Lifespan (R script) – C. briggsae and C. tropicalis | 10.6084/m9.figshare.26308171 | ||
| Software, algorithm | RNA-seq (R script) | 10.6084/m9.figshare.26314531 | ||
| Software, algorithm | Transcriptomic alignments and feature counts | 10.6084/m9.figshare.26314591 | ||
| Chemical compound/drug | Bortezomib | Sigma-Aldrich | ||
| Chemical compound/drug | Tretinoin | Sigma-Aldrich | ||
| Chemical compound/drug | Fisetin | Tocris Bioscience | ||
| Chemical compound/drug | Temsirolimus | Cayman Chemical | ||
| Chemical compound/drug | Everolimus | Cayman Chemical | ||
| Chemical compound/drug | Dasatinib | Cayman Chemical | ||
| Chemical compound/drug | Decitabine | Selleck Chemicals | ||
| Chemical compound/drug | Gefitinib | Sigma-Aldrich | ||
| Chemical compound/drug | Metoprolol | Sigma-Aldrich | ||
| Chemical compound/drug | Berberine | Cayman Chemical | ||
| Chemical compound/drug | Erlotinib | Cayman Chemical | ||
| Chemical compound/drug | Dexamethasone | Sigma-Aldrich | ||
| Chemical compound/drug | Aldosterone | Sigma-Aldrich | ||
| Chemical compound/drug | Propranolol | Sigma-Aldrich | ||
| Chemical compound/drug | Metoprolol | Sigma-Aldrich |
Additional files
-
Supplementary file 1
Supplementary tables providing additional quantitative detail for the results.
Tables containing (a) sources of experimental variation across trials, genes significantly (b) up- and (c) downregulated by atRA treatment, and (d) WaRG comparisons across hsf-1 and aak-2 gene expression analyses.
- https://cdn.elifesciences.org/articles/104375/elife-104375-supp1-v1.docx
-
MDAR checklist
- https://cdn.elifesciences.org/articles/104375/elife-104375-mdarchecklist1-v1.docx
Download links
A two-part list of links to download the article, or parts of the article, in various formats.
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
Computer prediction and genetic analysis identifies retinoic acid modulation as a driver of conserved longevity pathways in genetically diverse Caenorhabditis nematodes
eLife 13:RP104375.
https://doi.org/10.7554/eLife.104375.3
