Late-life dietary folate restriction reduces biosynthesis without compromising healthspan in mice

Abstract

Folate is a vitamin required for cell growth and is present in fortified foods in the form of folic acid to prevent congenital abnormalities. The impact of low-folate status on life-long health is poorly understood. We found that limiting folate levels with the folate antagonist methotrexate increased the lifespan of yeast and worms. We then restricted folate intake in aged mice and measured various health metrics, metabolites, and gene expression signatures. Limiting folate intake decreased anabolic biosynthetic processes in mice and enhanced metabolic plasticity. Despite reduced serum folate levels in mice with limited folic acid intake, these animals maintained their weight and adiposity late in life, and we did not observe adverse health outcomes. These results argue that the effectiveness of folate dietary interventions may vary depending on an individual's age and sex. A higher folate intake is advantageous during the early stages of life to support cell divisions needed for proper development. However, a lower folate intake later in life may result in healthier aging.

Figure 1.. Inhibitors of 1C metabolism extend the lifespan of yeast and worms.

(A) Schematic of 1C enzymatic reactions. The reactions inhibited by methotrexate (MTX) or ATIC Dimerization Inhibitor are indicated. (B) Survival curves on rich undefined media for S. cerevisiae MATα (strain BY4742) cells (shown in black), compared with experiment-matched cells mock-treated with DMSO (shown in gray) or three different doses of MTX (as indicated, in different shades of green). The number of mother cell divisions (replicative lifespan) is on the x-axis. (C) Survival curves for C. elegans (strain N2) exposed to the indicated doses of MTX. Survival probability is on the y-axis, and time (in days) is on the x-axis. Mean lifespans and the number of animals assayed in each case are shown in parentheses. The indicated P-value was based on the log-rank test. (C, D) Survival curves, as in (C), for animals exposed to the indicated doses of ATIC Dimerization Inhibitor. Source data are available for this figure.

Figure S1.. Methotrexate increases the size of yeast cells.

Cell size histograms of yeast cells (BY4742 strain background) treated with the indicated doses of methotrexate. Size (in fL) is on the x-axis, and the cell number on the y-axis. Source data are available for this figure.

Figure S2.. Survival curves of female Swiss mice on chronic, low-dose methotrexate.

Methotrexate was administered in the food at the indicated dose every other week, starting at 7 wk of age. Raw data were from Rustia & Shubik (1973). Survival probability is on the y-axis, and time (in weeks) is on the x-axis. Mean lifespans and the number of animals assayed in each case are shown in parentheses. The indicated P-value was based on the log-rank test.

Figure 2.. Evaluating healthspan in mice under dietary folate restriction late in life.

(A) Schematic of the study design. See the Materials and Methods section for a detailed description of each assay. (B) Serum folate levels were measured at 120w, using an established microbiological assay (Cooper & Jonas, 1973) and shown on the y-axis. The different diets are on the x-axis, as indicated for the folate/choline-replete (F/C+) or -limited (F/C−) groups. The boxplot graphs were generated with R language functions. Each box is drawn from the first to the third quartile, with a horizontal line denoting the median. The whiskers show the interquartile range, and they were drawn at 1.5 x interquartile range. The replicates were all biological ones from different animals. (C) The weight of the animals (y-axis) was measured every month (x-axis). The red horizontal lines were drawn to help visualize the weight changes over time in female and male mice. Loess curves and the std errors at a 0.95 level are shown. A mixed effects regression model was applied with the lme4 and lmer R language packages to evaluate the effects of the fixed variables (diet, time, sex) on the observed weight, taking into account the repeated longitudinal measurements on each mouse (see the Materials and Methods section, and description in the text). A negative association with the F/C+ diet was significant (P = 0.0313). Source data are available for this figure.

Figure S3.. Evaluating assumptions of a linear regression model for longitudinal weight measurements.

(A) Plot of the standardized residuals on the y-axis and the fitted values on the x-axis, from the data shown in Fig 2C. (B) Q-Q plot of the residuals, with the sample quantiles from the measurements in Fig 2C (y-axis), against the theoretical normally distributed ones (x-axis).

Figure S4.. Number and size of blood cells from mice placed on dietary folate restriction late in life.

(A) Blood cell numbers were measured at 108 wk of age from animals of the indicated sex and diet group. (A, B) Cell size (y-axis; in fL) was measured from the same samples shown in (A) from animals of the indicated sex and diet group. The boxplots were drawn as in Fig 2. Source data are available for this figure.

Figure S5.. Survival curves of mice placed on dietary folate restriction late in life.

Survival probability is on the y-axis, and time (in weeks) on the x-axis, from female (right panel) and male (left panel) animals of each diet test group. The indicated P-value was based on the log-rank test. Source data are available for this figure.

Figure 3.. No adverse healthspan metrics in mice placed on dietary folate restriction late in life.

(A) Frailty Index scores are shown on the y-axis. Measurements were taken at the indicated times from female and male animals of each diet test group, as described in the Materials and Methods section. (B) The total, fat, and lean mass of each mouse in the study was measured by MRI (see the Materials and Methods section), and shown on the y-axis. The boxplots were drawn as in Fig 2. Source data are available for this figure.

Figure S6.. No significant gait changes in mice placed on dietary folate restriction late in life.

As described in the Materials and Methods section, step width variance and gait symmetry values (shown on the y-axis) were measured with the Digigait system. As indicated, measurements were taken at the indicated times from female and male animals of each diet test group. The boxplots were drawn as in the previous figures. Source data are available for this figure.

Figure S7.. Open field and novel object recognition assays in mice placed on dietary folate restriction late in life.

In (A), the inner zone and moving times during open field evaluation are on the y-axis. In (B), the discrimination ratio values, reflecting the ability of the mice to recognize a new object, are on the y-axis. Detailed descriptions of the assays are in the Materials and Methods section. The boxplots were drawn as in the previous figures. Source data are available for this figure.

Figure S8.. Normal cardiac function of mice placed on dietary folate restriction late in life.

Several parameters of cardiac function (x-axis) were measured by echocardiography as described in the Materials and Methods section, and the corresponding values are on the y-axis. Measurements were taken at the indicated times from female and male animals of each diet test group, as indicated. The boxplots were drawn as in the previous figures. Source data are available for this figure.

Figure 4.. Improved metabolic activity of mice placed on dietary folate restriction late in life.

(A) Photographs of male mice on the indicated diet were taken at 85 wk of age. Signs of graying were visible on the coat of mice on the F/C+ diet (left) but not on the coat of mice on the F/C− diet (right). (B) The respiratory exchange rate values (y-axis) were from six to eight mice in each indicated group at 108 wk of age. The measurements were taken after the animals were acclimated for 2 d in the metabolic cages. The period between the red vertical lines corresponds to night-time when mice are active. Loess curves and the std errors at a 0.95 level are shown. Source data are available for this figure.

Figure 5.. Metagenomic profiling of the gut microbiome of mice placed on dietary folate restriction late in life.

(A) Beta-diversity principal component analysis plots were based on Bray-Curtis dissimilarity indices (Knight et al, 2018), of the DNA from the fecal microbiome sampled and sequenced at 90 wk of age, from five mice in each test group. Three principal components (PC1,2,3; shown at the top) accounted for ∼70% of the dataset variance. (B) Metabolic pathway biomarker changes associated with folate limitation late in life, from metagenomic data of the fecal microbiome. The LEfSe computational pipeline was used to determine the features most likely to explain the observed differences. The computed linear discriminant analysis scores (Log₁₀-transformed) are on the x-axis. Linear discriminant analysis scores incorporate effect sizes, ranking the relevance of the identified biomarkers and enabling their visualization (Segata et al, 2011). Gray arrows indicate pathways involved in amino acid synthesis and black arrows indicate pathways involved in IMP synthesis. Source data are available for this figure.

Figure S9.. Species diversity within the gut microbiome of mice placed on dietary folate restriction late in life.

Shannon’s diversity index (y-axis) is shown for each sex and diet as a metric of the alpha diversity of the fecal microbiome sampled and sequenced at 90 wk of age, from five mice in each test group. The boxplots were drawn as in the previous figures. Source data are available for this figure.

Figure S10.. Cytokine levels of mice placed on dietary folate restriction late in life.

Serum cytokine levels were measured at 120 wk of age using a mouse multiplex cytokine assay service by Eve Technologies (see the Materials and Methods section). The measured amounts (in pg/ml; Log₁₀-transformed) are on the y-axis. The different diets are on the x-axis for each cytokine, as indicated. Significant differences in the measured values within a sex and diet group are indicated with a red asterisk (P < 0.05, based on the Wilcoxon rank sum test). The boxplots were drawn as in the previous figures. Source data are available for this figure.

Figure S11.. Disease load of mice placed on dietary folate restriction late in life.

The pathology of the indicated tissues, collected at 120 wk of age, was scored on a 0–4 scale (with four reflecting the highest degree of pathological changes; see the Materials and Methods section) and shown on the y-axis. The different diets are on the x-axis. Differences in kidney abnormalities in male mice were significant (indicated with a red asterisk; P = 0.0165, based on the Wilcoxon rank sum test). The boxplots were drawn as in the previous figures. Source data are available for this figure.

Figure S12.. No significant changes in methylation status or uracil misincorporation in the genome of mice placed on dietary folate restriction late in life.

(A) DNA methylation at multiple sites in the genome was measured with the epigenetic clock assay by ZymoResearch (see the Materials and Methods section) from liver samples at 120 wk of age. Based on the measured against the predicted changes, a biological age estimate was compared with the actual age (ΔDNAage; shown on the y-axis). The different diets are on the x-axis. The boxplots were drawn as in the previous figures. (B) Uracil levels in the DNA were measured as described in Materials and Methods from liver tissue collected at 120 wk of age (y-axis; in pg/μg). The different diets are on the x-axis. The boxplots were drawn as in the previous figures. Source data are available for this figure.

Figure 6.. Metabolite changes in mice placed on dietary folate restriction late in life.

(A) Steady-state serum amino acid levels were measured at 120 wk of age using an HPLC-based assay (see the Materials and Methods section). The measured amounts (in nmol/ml) are on the y-axis. The different diets are on the x-axis for each amino acid, as indicated. The boxplots were drawn as in the previous figures. Differences in glutamine (Gln) levels in male mice were significant (marked with a red asterisk; P = 2.6 × 10⁻⁵, based on the Wilcoxon rank sum test). (B) Steady-state levels of primary and biogenic amine metabolites from liver tissue collected at 120 wk of age were measured by GC-TOF MS and HILIC-QTOF MS/MS, respectively. Changes in the indicated pairwise comparisons were identified from the magnitude of the difference (x-axis; Log₂-fold change) and statistical significance (y-axis; based on robust bootstrap ANOVA tests), as indicated by the red lines (see the Materials and Methods section). (B, C) Metabolite enrichment analysis based on the MetaboAnalyst platform (Chong et al, 2019) for male mice from the data in (B), for metabolites present at significantly lower levels under folate limitation (P < 0.05 and >1.5-fold change). The corresponding metabolic pathways are on the y-axis, and the P-values are on the x-axis. The size of each bubble on the chart reflects the relative number of “hit” metabolites in each pathway. The color of each bubble reflects the enrichment ratio, which is the number of hits within a metabolic pathway divided by the expected number of hits. Only pathways with enrichment >2 and FDR values < 0.05 are shown. Source data are available for this figure.

Figure 7.. Transcriptomic changes in mice placed on dietary folate restriction late in life point to reduced protein synthesis in both sexes.

(A) Steady-state levels of mRNAs from liver tissue collected at 120 wk of age were measured by RNAseq. Changes in the indicated pairwise comparisons were identified from the magnitude of the difference (x-axis; Log₂-fold change) and statistical significance (y-axis; based on robust bootstrap ANOVA tests), as indicated by the red lines (see the Materials and Methods section). The number of transcripts whose levels changed within these thresholds is shown in each case. (A, B) Transcript enrichment analysis based on the PANTHER platform (Ashburner et al, 2000; Thomas et al, 2022; Gene Ontology Consortium et al, 2023) for male mice from the data in (A), for transcripts present at significantly lower levels under folate limitation (P < 0.05 and >1.5-fold change) that could be assigned to a specific gene ID (n = 165). The corresponding biological processes are on the y-axis, and the P-values are on the x-axis. The size of each bubble on the chart reflects the relative number of “hit” transcripts in each process. The color of each bubble reflects the enrichment ratio, which is the number of hits within a process divided by the expected number of hits. Only pathways with enrichment >2 and FDR values < 0.05 are shown. (A, C) Transcript enrichment analysis for female mice from the data in (A), for transcripts present at significantly lower levels under folate limitation (P < 0.05 and >1.5-fold change) that could be assigned to a specific gene ID (n = 490). The plots were drawn as in (B). Source data are available for this figure.

Figure S13.. RPS6 phosphorylation levels in liver tissue are not significantly affected by dietary folate restriction late in life.

Phosphorylated RPS6 (P-RPS6) levels (y-axis) were measured with a phospho-specific antibody and normalized against the signal from an antibody against RPS6 (detecting the total amount), as described in the Materials and Methods section. The different diets are on the x-axis. The boxplots were drawn as in the previous figures. Source data are available for this figure.

Figure S14.. 4EBP1 phosphorylation levels in liver tissue are not significantly affected by dietary folate restriction late in life.

Phosphorylated 4EBP1 (P-4EBP1) levels (y-axis) were measured with a phospho-specific antibody and normalized against the signal from an antibody against 4EBP1 (detecting the total amount), as described in the Materials and Methods section. The different diets are on the x-axis. The boxplots were drawn as in the previous figures. Source data are available for this figure.

Figure S15.. Lower IGF-1 levels in the serum of female mice placed on dietary folate restriction late in life.

IGF-1 levels (in pg/ml; shown on the y-axis) were measured with a commercial mouse IGF-1 ELISA Kit (see Table 1). The different diets are on the x-axis. The boxplots were drawn as in the previous figures. In female mice, the difference was statistically significant, indicated with a red asterisk (P = 0.028, based on the Wilcoxon rank sum test). Source data are available for this figure.