henhaobio
Sign in

NatureGeneticsPeer-reviewed journal article

Regulators of health and lifespan extension in genetically diverse mice on dietary restriction

Andrea Di FrancescoiD, Andrew G. DeighaniD, Lev LitichevskiyiD, Zhenghao CheniD, Alison Luciano, Laura L. Robinson, Gaven Garland, Hannah Donato, Matthew J. VincentiD, Will Schott, Kevin M. WrightiD, Anil RajiD, G.V. PrateekiD, Martin N. MullisiD, Warren G. HilliD, Mark L. ZeideliD, Luanne L. PetersiD, Fiona HardingiD, David BotsteiniD, Ron KorstanjeiD, Christoph A. ThaissiD, Adam FreundiD, Gary A. ChurchilliD

PDFDOIPreprintCodeData
In-depth full-text digest available · Jump to digest ↓

Abstract

Caloric restriction (CR) delays aging and extends healthy lifespan in multiple species. Alternative forms of dietary restriction (DR) such as intermittent fasting (IF) have drawn significant interest as a more sustainable regimen, but the landscape of longevity-promoting dietary interventions remains largely unexplored. Identifying the most robust, efficacious, and experimentally tractable modes of DR is key to better understanding and implementing effective longevity interventions for human healthspan. To that end, we have performed an extensive assessment of DR interventions, investigating the effects of graded levels of CR (20% and 40%) and IF (1 day and 2 days of fasting per week) on the health and survival of 960 genetically diverse female mice. All interventions extended lifespan, although only CR significantly reduced the mortality doubling time. Notably, IF did not extend lifespan in mice with high pre-intervention bodyweight. We carried out extensive phenotyping to determine the health effects of long-term DR and to better understand the mechanisms driving within-diet heterogeneity in lifespan. The top within-diet predictor of lifespan was the ability of mice to maintain bodyweight through periods of handling, an indicator of stress resilience. Additional predictors of long lifespan include specific changes in immune cells, red blood cell distribution width (RDW), and retention of adiposity in late life. We found that lifespan is heritable (h2 = 0.24), and that genetic background has a larger influence on lifespan than dietary interventions. We identified a significant association for lifespan and RDW on chromosome 18 that explained 4.3% of the diet-adjusted variation in lifespan. Diet-induced changes on metabolic traits, although beneficial, were relatively poor predictors of lifespan, arguing against the long-standing notion that DR works by counteracting the negative effects of obesity. These findings indicate that improving health and extending lifespan are not synonymous and that metabolic parameters may be inappropriate endpoints for evaluating aging interventions in preclinical models and clinical trials.

In-depth summary

Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 2 (29 Nov 2023) · CC BY 4.0

In 937 genetically diverse female mice, caloric restriction and intermittent fasting extended lifespan, yet bodyweight resilience, immune and red-cell traits, not metabolic improvements, predicted longevity.

Responses to dietary restriction (DR) vary between individuals, caloric restriction (CR) is hard for humans to sustain, and how CR compares with intermittent fasting (IF), or which markers forecast individual benefit, was poorly known. This lifelong study in genetically diverse mice compares graded CR and IF directly, links lifespan to longitudinal phenotypes and genotype, and questions metabolic endpoints as surrogates for aging interventions.

Key results

  1. ShownAll four restriction regimens lengthened lifespan in proportion to restriction severity, but only CR, not intermittent fasting, slowed the Gompertz rate of aging relative to ad libitum (AL) feeding.

    Overall log-rank p < 2.2e-16 (4 df); order 40% CR > 20% CR > 2-day IF > 1-day IF > AL; 40% CR median lifespan ~9 months (36.3%) above AL; mortality doubling time showed a significant slowing of aging for CR only; survival curves separated only after ~18 months of age. · DR extends lifespan in female DO mice; Effects of body weight and composition on…; Survival Analysis; Tables

  2. ShownIntermittent fasting failed to extend lifespan in mice that were heavier before the intervention, whereas CR benefited both lighter and heavier mice.

    Kaplan-Meier analysis split at median 6-month (pre-DR) bodyweight: CR extended lifespan in both strata with no significant difference in extension; lighter mice gained from IF, while heavier IF mice lived about as long as AL mice. · Bodyweight and lifespan prediction; Effects of body weight and composition on…; Conclusions

  3. ShownWithin diets, retaining bodyweight was the strongest lifespan predictor: weight resilience during stressful phenotyping, less post-DR weight loss and higher adiposity all tracked longer life, opposite to the direction DR pushes these traits.

    Bodyweight change from 10 to 11 months over the phenotyping period (PhenoDelta 10) was the top lifespan predictor among all traits (p < 2.2e-16); weight retained from 6 to 20 months, p = 1.77e-10; higher adiposity predicted longer life at 10 months (p = 5.28e-15) and 22 months (p < 2.2e-16). · Bodyweight and lifespan prediction; Effects of body weight and composition on…

  4. ShownDR lowered fasting glucose and weight-adjusted energy expenditure and, under CR, widened day-night respiratory quotient (RQ) swings, yet none predicted lifespan within diets; lymphocyte fraction, naive T cells and erythroid traits, notably red cell distribution width (RDW), did.

    No significant within-diet lifespan association for fasting glucose, energy expenditure or Delta RQ at any age; diet-adjusted lymphocyte percentage positively and RDW negatively associated with lifespan; covariance path decomposition on a reduced network of 21 exemplar traits, derived from the 194-trait network, ranked bodyweight/composition and immune/hematologic paths highest. · Regulators of DR-mediated lifespan extension; Impact of DR on health and metabolism; Impacts of DR on immune and erythroid cell…; Network modeling

  5. ShownLifespan was moderately heritable and genetics outweighed diet; a CAST-derived chromosome 18 quantitative trait locus (QTL) shortened lifespan, and mediation analysis suggests much of this effect runs through RDW, whose own QTL mapped to the same position.

    h^2 = 0.236 (95% bootstrap CI [0.106, 0.360]) vs 7.4% of variance from diet; QTL at 21.55Mb: 164 of 929 mice carrying CAST lived 3.7 months (12.5%) less, p = 6.66e-07, 4.34% of diet-adjusted variance; an RDW QTL co-localized, and RDW accounted for 66.0% of the QTL effect on lifespan (p = 1.85e-7 unconditional vs 0.00225 given RDW). · Genetic analysis of lifespan; Conclusions

Methods, data and reproducibility

Female Diversity Outbred (DO) mice were randomized to ad libitum feeding, weekly 1- or 2-day fasting, or 20% or 40% caloric restriction from 6 months until death, with 194 traits phenotyped longitudinally; survival models, generalized additive mixed models, diet-adjusted lifespan correlations, a sparse partial correlation network and QTL mapping were applied.

  • Main lifespan cohort of female Diversity Outbred mice randomized to AL, weekly 1-day or 2-day IF, or 20% or 40% CR, enrolled in 12 waves (2016 to 2017) · mouse (Diversity Outbred, female) · n = 960 enrolled; 937 alive at DR onset (6 months)
  • Independent cohort on the same DR protocols for daily food intake and EchoMRI body composition before and after fasting · mouse (Diversity Outbred, female) · n = 160
  • Weekly lifelong bodyweight records, loess-smoothed within mouse · mouse (Diversity Outbred, female) · n = over 100,000 values from 937 mice
  • Longitudinal phenotyping of 194 traits across 11 categories (frailty, metabolic cages, body composition, echocardiography, blood counts, flow cytometry, behavior) · mouse (Diversity Outbred, female) · n = up to 689 single-timepoint measurements per mouse
  • Body composition by PIXImus densitometry at 10, 22 and 34 months of age · mouse (Diversity Outbred, female) · n = 895, 689 and 241 mice, respectively
  • Whole-genome genotypes used for heritability estimation and QTL mapping · mouse (Diversity Outbred, female) · n = 929 of 937 mice
  • Fecal microbiome samples collected every 6 months (analyzed in a companion paper) · mouse (Diversity Outbred, female)
  • Processed data, analysis scripts and supplementary files deposited on FigShare · mouse (Diversity Outbred, female) · doi:10.6084/m9.figshare.24600255.v1
Methods & tools
Kaplan-Meier survival curves and log-rank tests, flexsurv v2.2.2 (Gompertz log-linear hazard models), mgcv (generalized additive mixed models), qtl2 (QTL mapping, founder haplotype effects), ggLASSO (sparse low-rank Gaussian graphical model), Nonparanormal SKEPTIC covariance estimator (Kendall tau), KNN graph spectral clustering and covariance path decomposition, Siemens ADVIA 2120 hematology analyzer, BD LSRII flow cytometer, LUNAR PIXImus II densitometer and EchoMRI, Sable Systems metabolic cages; VisualSonics Vevo 770/2100 echocardiography, Climb (Rockstep) research data management; R 4.2.2
Software & environment
R 4.2.2 with RStudio 2022.12.0+353; flexsurv v2.2.2; mgcv and qtl2 R packages; ggLASSO Python package (lambda = 0.1, mu = 100); no container described
Reproducibility gaps
Only processed data and scripts are deposited; no raw genotype accession is given, microbiome data are deferred to a companion paper, and mgcv, qtl2 and ggLASSO versions are unstated.
Limitations (6)
  • Only one genome-wide significant locus was found; over 75% of the genetic contribution to lifespan remains unexplained, gene-by-gene and context-specific effects were not modeled, and no causal gene at the chromosome 18 locus was identified. · stated by the authors
  • The partial correlation network is estimated to capture only 43% of DR effects on lifespan; the direct diet-to-lifespan path (normalized score 0.573) represents effects not explained by any measured trait. · stated by the authors
  • Tumor incidence was not formally measured and specific neoplastic lesions were not identified, so less apparent but lethal cancers could have been missed; immune-longevity correlations might partly reflect tumor effects. · stated by the authors
  • Diet-induced shifts in circulating immune subsets may partly reflect fasting-related trafficking to tissues and blood-sampling timing, and limited flow markers in outbred mice forced coarse natural killer and naive/effector T cell definitions. · stated by the authors
  • Only female DO mice on a single standard chow were studied, so whether the IF failure in heavier animals and the predictor rankings hold in males or other diet compositions is unknown. · noted in this summary
  • Lifespan predictors and mediation paths come from observational correlations and partial-correlation path decomposition, and pairwise survival p-values were unadjusted for multiplicity (p < 0.01 cutoff), so causal roles of weight retention or RDW are not experimentally tested. · noted in this summary

Read this if you design dietary restriction or aging-intervention studies and want lifelong evidence from genetically diverse mice on which biomarkers actually track lifespan.

Citations
194
FWCI
23.40
Field percentile
top 1%

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISUnnikrishna '21Luciano '26Fleischman '26Zhang '21Roy '21Mosley '24
75% semantic matchMolecular Biology
Archana Unnikrishnan, Stephanie Matyi, Karla Pfeil Garrett+6 more·Aging Cell·Oct 29, 2021· 24 citations