Polyamines sustain epithelial regeneration in aged intestines by modulating protein homeostasis

Abstract

Ageing dampens the regenerative potential of intestinal epithelium across species including humans, yet the underlying causes remain elusive. Here we characterized the temporal dynamics of regeneration following injury induced by 5-fluorouracil, a commonly used chemotherapeutic agent, using proteomic and metabolomic profiling of intestinal tissues together with functional assays. The comparison of regeneration dynamics in mice of different ages revealed the emergence of proteostasis stress and increased levels of polyamines following injury exclusively in old epithelia. We show that delayed regeneration is an intrinsic feature of aged epithelial cells that display reduced protein synthesis and the accumulation of ubiquitylated proteins. The inhibition of the polyamine pathway in vivo further delays regeneration in old mice, whereas its activation by dietary intervention or supplementation of polyamines is sufficient to enhance the regenerative capacity of aged intestines. Our findings highlight the promising epithelial targets for interventions aimed at tackling the decline in tissue repair mechanisms associated with ageing.

Fig. 1. Identifying proteome dynamics during regeneration of young small intestines.

a, A schematic of 5-FU induced regeneration in young mice. i.p., intraperitoneal. b, The relative body weight of young mice treated with a single dose of 5-FU or PBS as control. The body weight of each mouse was normalized to its body weight at the time of injection. n = 18 PBS, n = 17 5-FU mice. The red and blue lines indicate the median body weight for each treatment group on each day. The P values were calculated using the two-tailed Welch’s t-test by comparing the body weights of 5-FU-treated mice with PBS-treated controls on the indicated day and by two-way ANOVA for overall day and treatment comparisons. c, The representative pictures of haematoxylin and eosin (H&E) staining of the small intestine from indicated treatments and timepoints. Scale bars, 50 µm. d, The quantification of the number of crypts per millimetre of small intestine in the indicated groups. The PBS-treated mice from different days were combined. Each dot represents one mouse. n = 12 PBS, n = 4 5-FU mice per group. The data are presented as mean ± s.d. The P value was calculated using the two-tailed Welch’s t-test. e, The quantification of the number of cells per micrometre of villus in the indicated groups. PBS-treated mice from different days were combined. Each dot represents one mouse. n = 11 PBS, n = 4 5-FU mice per group. The data are presented as mean ± s.d. The P value was calculated using the two-tailed Welch’s t-test. f, Left: the representative pictures of pH3 staining from indicated treatment and timepoint. Scale bar, 100 µm. The asterisks indicate the pH3⁺ crypts. Right: the percentage of pH3⁺ crypts in the indicated groups. The PBS-treated mice from different days were combined. Each dot represents one mouse. n = 14 PBS; n = 5 day 2, n = 6 day 5, n = 11 day 7 5-FU. The data are presented as mean ± s.d. The P value was calculated using the two-tailed Welch’s t-test. g, The classification of protein groups quantified by proteomics according to their abundance changes relative to the PBS controls. h, The abundance profiles for the four clusters of proteins affected by 5-FU. The representative REACTOME gene sets significantly enriched in each cluster are shown. Stand., standardized; SLC, solute carrier; ENOS, endothelial nitric oxide synthase. i, The abundance profile of ribosome (67 proteins) and MCM (6 proteins) complexes. The protein abundances of individual complex members were normalized to the median complex abundance of PBS-treated mice, and profiles plotted using a locally estimated scatterplot smoothing function. The shaded area around the regression line represents the 95% confidence interval. n = 16 PBS, n = 5 day 2, n = 5 day 5, n = 4 day 7 5-FU. j, A scheme of puromycin incorporation assay performed on freshly isolated crypts. k, Left: a representative immunoblot for puromycin incorporation. Right: the quantification of puromycin incorporation relative to Ponceau staining (loading control). Each dot represents one mouse. n = 3 mice per group. The data are presented as mean ± s.d. The P value was calculated using the two-tailed Welch’s t-test. Panels a and j created with BioRender.com. Source data

Fig. 2. Proteostasis stress delays intestinal regeneration following 5-FU.

a, The relative body weight of young (left) and old (right) mice treated with a single dose of 5-FU or PBS as control. The body weight of each mouse was normalized to its body weight at the time of injection. n = 18 PBS, n = 17 5-FU young mice and n = 12 PBS, n = 14 5-FU old mice. The data are presented as mean ± s.d. The P values were calculated using the two-tailed Welch’s t-test by comparing the body weights of 5-FU treated mice with PBS-treated controls on the indicated day. b, Left: the representative pictures of pH3 staining from indicated treatments and timepoints. Scale bar, 100 µm. The asterisks indicate the pH3⁺ crypts. Right: the percentage of pH3⁺ crypts relative to PBS. At each timepoint, 5-FU values were normalized to the average PBS control mice and expressed as percentage. Each dot represents one mouse. Young mice: n = 5 day 2, n = 6 day 5, n = 11 day 7; old mice: n = 6 day 2, n = 6 day 5, n = 7 day 7. The data are presented as mean ± s.d. The P value was calculated using the two-tailed Welch’s t-test for timepoint comparison and two-way ANOVA for overall day and age comparisons. c, The workflow for the comparison of protein abundance profiles between young and old mice. PBS-treated young mice: n = 5 day 2, n = 5 day 5, n = 4 day 7. 5-FU-treated young mice: n = 5 day 2, n = 5 day 5, n = 4 day 7. PBS-treated old mice: n = 5 day 2, n = 5 day 5, n = 5 day 7. 5-FU-treated old mice: n = 5 day 2, n = 5 day 5, n = 3 day 7. d, The fold change profiles (old versus young, log₂) for the four clusters of proteins that display different dynamics in young and old mice upon 5-FU treatment. Stand., standardized. e, The network of proteostasis-related proteins from cluster 1. The protein–protein interactions were derived from STRING using a cut-off of 0.7. f, The protein abundance profiles for the components of the 19S proteasome (24 proteins), TriC (7 proteins) complexes, ATG7 and ribosomal proteins (67 proteins). The protein profiles are plotted using a locally estimated scatterplot smoothing function. The shaded area around the regression line represents the 95% confidence interval. n = 14 PBS young, n = 5 day 2, n = 5 day 5, n = 4 day 7 5-FU young; n = 15 PBS old, n = 5 day 2, n = 5 day 5, n = 3 day 7 5-FU old. g, Left: a representative immunoblot for the K48-polyubiquitylated proteins. Right: the quantification of K48-polyubiquitylated proteins relatively to Ponceau staining (loading control). Each dot represents one mouse. n = 3 mice per group. The data are presented as mean ± s.d. The P value was calculated using the two-tailed Welch’s t-test for timepoint comparison and two-way ANOVA for overall day and age comparisons. h, A schematic of the effect of CHX on intestinal regeneration in young mice. i, The body weight changes of young mice after 5-FU ± CHX. Each mouse’s body weight was normalized to its weight on the day of the PBS or 5-FU injection. At each timepoint, the average body weight change in PBS-treated controls was subtracted from that of 5-FU-treated mice (with or without CHX). This analysis isolates the effect of 5-FU on body weight dynamics between CHX and control groups. The averages were obtained from: n = 3 PBS–PBS, n = 3 PBS–CHX, n = 3 5-FU–PBS, n = 3 5-FU–CHX young mice. The data are presented as mean ± s.d. The P value was calculated using two-way ANOVA for comparisons. j, Left: the representative pictures of IHC staining for pH3 at day 4 following 5-FU (±CHX). Scale bars, 100 µm. The arrows indicate the pH3⁺ cells. Right: the change in number of pH3⁺ cells per crypt. n = 3 mice per group. The data are presented as mean ± s.d. The P value was calculated using the two-tailed Welch’s t-test for group comparison. Panel h created with BioRender.com. Source data

Fig. 3. Delayed regeneration of organoids from old mice upon 5-FU treatment.

a, A scheme of the 5-FU-induced regeneration experiment using intestinal organoids. b, The brightfield images of intestinal organoids from the indicated conditions, taken 3 days after PBS or 5-FU wash out. The images are representative of organoid cultures derived from n = 3 young and n = 3 old mice, independently cultured under the different conditions. Scale bars, 200 µm. c, The quantification of organoid morphology at day 3 post 5-FU treatment. The bars show the percentage of budded, cystic or non-regenerating organoids from independent organoid cultures derived from young and old mice (n = 3 young and n = 3 old mice). The data are presented as mean ± s.d. The P value was calculated by the two-tailed Welch’s t-test comparing the different conditions with young or old PBS control. d, Left: a representative immunoblot for the pH3 protein. Right: the quantification of the pH3 level relative to Ponceau staining (loading control). n = 5 mice per group. The data are presented as mean ± s.d. The P value was calculated using the two-tailed Welch’s t-test for timepoint comparison and two-way ANOVA for overall day and age comparisons. e, Left: a representative immunoblot for the puromycin incorporated proteins. Right: the quantification of the puromycin incorporation relative to Ponceau staining (loading control). n = 4 mice per group. The data are presented as mean ± s.d. The P value was calculated using the two-tailed Welch’s t-test for timepoint comparison and two-way ANOVA for overall day and age comparisons. f, Left: a representative FACS gating for annexin V assay on old organoids at day 3 after 5-FU wash out. Right: the quantification of the sum of early and late apoptotic cells at the indicated timepoints. n = 6 young; n = 5 old. The data are presented as mean ± s.d. The P value was calculated using the two-tailed Welch’s t-test for timepoint comparison and two-way ANOVA for overall day and age comparisons. g, Left: a representative immunoblot for the SQSTM1 protein. Right: the quantification of SQSTM1 protein level relative to Ponceau staining (loading control). n = 4 mice per group. The data are presented as mean ± s.d. The P value was calculated using the two-tailed Welch’s t-test for timepoint comparison and two-way ANOVA for overall day and age comparisons. h, Left: a representative immunoblot for the K48-polyubiquitylated proteins. Right: the quantification of K48-polyubiquitylated protein level relative to Ponceau staining (loading control). n = 5 mice per group. The data are presented as mean ± s.d. The P value was calculated using the two-tailed Welch’s t-test for timepoint comparison and two-way ANOVA for overall day and age comparisons. n.s., not significant. Source data

Fig. 4. Polyamine pathway elevation upon 5-FU in the aged intestinal epithelium.

a, A schematic representation of the dynamics of protein synthesis comparing our damage–regeneration model with the fasting–RF model from Imada et al. 2024. b, A schematic representation of the polyamine biosynthesis and ornithine metabolism pathways. c, The LC–MS quantification of polyamines from crypts lysate. Each dot represents one mouse. n = 4 mice per condition. The data are presented as mean ± s.d. The P value was calculated using the two-tailed Welch’s t-test for timepoint comparison and two-way ANOVA for overall day and age comparisons. d, Left: a representative immunoblot for hypusinated EIF5A and total EIF5A proteins. Right: the quantification of hypusinated EIF5A and total EIF5A relative to Ponceau staining (loading control). Each dot represents one mouse (young: n = 3; old: n = 5 per each indicated condition). The data are presented as mean ± s.d. The P value was calculated using the two-tailed Welch’s t-test for timepoint comparison and two-way ANOVA for overall day and age comparisons. e, The top 29 motifs that depend on hypusinated EIF5A for efficient translation according to ref. . Pro–Pro, proline–proline. f, The distribution of proteins enriched in hypusinated EIF5A-dependent motifs across clusters that show distinct dynamics in young and old mice (see Fig. 2c for cluster assignment and Fig. 2d for cluster profiles). The cluster numbers and their corresponding protein amounts (in brackets) are displayed on the x axis. Highlighted are the proteins that contain the highest number of hypusinated EIF5A-dependent motifs. The black line of each boxplot represents the median value, and the whiskers extend to 1.5× the interquartile range. The P value was calculated using Wilcoxon rank sum. g, The protein abundance profiles of the COL1A1, COL15A1 and FBN2 proteins. The protein profiles are plotted using a locally estimated scatterplot smoothing function. The shaded area around the regression line represents the 95% confidence interval. n = 14 PBS young, n = 5 day 2, n = 5 day 5, n = 4 day 7 5-FU young; n = 15 PBS old, n = 5 day 2, n = 5 day 5, n = 3 day 7 5-FU old. h, A schematic of young and old organoids in polyamine-free basal media. On day 12, young and old organoids were passaged to induce regeneration. The polyamine pathway activation was assessed both at baseline (on fully developed organoids, before passaging) and 3 days later. i, Left: a representative immunoblot for hypusinated EIF5A and total EIF5A proteins. Right: the quantification of hypusinated EIF5A and total EIF5A normalized to Ponceau staining. n = 4 mice per group. The data are presented as mean ± s.d. The P value was calculated using the two-tailed Welch’s t-test for comparisons. Source data

Fig. 5. Polyamine pathway deficiency impairs intestinal regeneration in young and old mice.

a, A schematic of 5-FU treatments in old mice with and without DFMO. b, The relative body weight of old mice treated twice with DFMO or vehicle control after a single dose of 5-FU (100 mg kg⁻¹). The body weight of each mouse was normalized to its body weight at the day of injection. n = 5 5-FU + DFMO, n = 5 5-FU − DFMO mice. The data are presented as mean ± s.d. The P value was calculated using the two-tailed Welch’s t-test for timepoint comparison and two-way ANOVA for overall comparisons. c, Left: the representative pictures of IHC staining for pH3 at day 12 following 5-FU (±DFMO). Scale bars, 100 µm. The arrows indicate the pH3⁺ cells. Right: the change in the number of pH3⁺ cells per crypt; n = 5 mice per group. The data are presented as mean ± s.d. The P value was calculated using the two-tailed Welch’s t-test for group comparison. d, Top: a schematic of ODC1 KO induction and 5-FU treatment. Bottom: the relative body weight of young ODC1 KO and WT mice treated with a single dose of 5-FU. The body weight of each mouse was normalized to its body weight at the day of injection. n = 4 ODC1 KO, n = 3 ODC1 WT mice. The data are presented as mean ± s.d. The P value was calculated using the two-tailed Welch’s t-test for timepoint comparison and two-way ANOVA for overall day, treatment and day x treatment comparisons. e, The representative H&E stainings of ODC1 KO and WT small intestines at 7 days post 5-FU injection. Scale bars, 50 µm. f, The quantification of the number of crypts per millimetre of small intestine in the indicated groups (7 dpi). Each dot represents one mouse. n = 3 5-FU ODC1 WT group; n = 4 5-FU ODC1 KO group. The data are presented as mean ± s.d. The P value was calculated by the two-tailed Welch’s t-test. g, The quantification of the number of cells per micrometre of villus in the indicated groups (7 dpi). Each dot represents one mouse. n = 3 5-FU ODC1 WT group; n = 4 5-FU ODC1 KO group. The data are presented as mean ± s.d. The P value was calculated by the two-tailed Welch’s t-test. h, A comparison of protein fold changes (FC) induced by 5-FU in ODC1 WT mice (x axis) and the effect of ODC1 KO on 5-FU injected mice (y axis). The proteomics data are from intestinal crypts collected at 7 dpi. The proteins significant (q < 0.05) in both comparisons are plotted. The colour code indicates the sum of the log₂FC from the two comparisons. The number of proteins in each quadrant is indicated. The correlation between the two treatments was assessed using Spearman’s correlation test. n = 3 PBS, n = 3 5-FU mice ODC1 WT group; n = 4 5-FU mice ODC1 KO group. The q values are from Spectronaut differential abundance analysis. i, The effect of ODC1 KO on proteostasis-related proteins. The volcano plots are based on the comparison of ODC1 KO versus WT in 5-FU injected mice. The significantly affected (absolute log₂FC >0.3 and q < 0.05) proteins are shown. n = 3 5-FU mice ODC1 WT group; n = 4 5-FU mice ODC1 KO group. The q values are from Spectronaut differential abundance analysis. Panels a and d created with BioRender.com. Source data

Fig. 6. Effect of DR followed by RF on intestinal epithelium regeneration.

a, A heat map of fold changes (log₂FC) for significantly affected proteins (absolute log₂FC >0.3 and q < 0.05) related to proteostasis from aged mice after DR or DR + RF treatments. UPS, ubiquitin proteasome system. n = 4 mice per group. b, The levels of total and hypusinated EIF5A from aged mice at different days after RF, quantified by immunoblot (Extended Data Fig. 6b). c, The correlation between hypusinated EIF5A and polyamines levels measured on the same intestinal crypt samples from aged mice that underwent DR followed by different days of RF. The R values represent Pearson correlation coefficient, and the P values are the result of the association test between the variables based on Pearson’s product moment correlation coefficient. d, The scheme of AL, DR and DR + RF treatments prior PBS or 5-FU injection. e, The relative average body weight changes after 5-FU. The body weight of each mouse was normalized to its body weight at the time of injection. For each timepoint, the average body weight of the vehicle treated group was subtracted from the average body weight of the 5-FU one. Each dot represents the daily average body weight change after 5-FU, n = 6. The averages were obtained from n = 16 PBS, n = 15 5-FU AL, n = 4 PBS, n = 6 5-FU DR, n = 5 PBS, n = 7 5-FU RF young mice and n = 17 PBS, n = 15 5-FU AL, n = 10 PBS, n = 16 5-FU DR, n = 7 PBS, n = 17 5-FU RF old mice. The P value was calculated by the two-tailed Welch’s t-test. f, The daily relative body weight changes induced by 5-FU in young (left) and old (right) mice. The body weights were calculated as in e. The P value was calculated using two-way ANOVA for overall day and treatment comparisons. The data in a are from ref. . Source data

Fig. 7. Effect of polyamine supplementation on intestinal epithelium regeneration in old mice.

a, Left: a schematic of polyamines supplementation in old mice. Right: the LC–MS quantification of polyamines from crypts lysate. Each dot represents one mouse. n = 4 (polyamines gavage) n = 5 (PBS gavage) mice per condition. The data are presented as mean ± s.d. The P value was calculated using the two-tailed Welch’s t-test for group comparison. b, A schematic of polyamines supplementation before 5-FU injury in old mice. c, The relative body weight of old mice orally gavaged with PBS prior 5-FU (5-FU) or PBS (vehicle control) injection. The body weight of each mouse was normalized to its body weight at the time of injection. n = 6 for vehicle control group, n = 6 for 5-FU group. The data are presented as mean ± s.d. The P value was calculated using the two-tailed Welch’s t-test for timepoint comparison and two-way ANOVA for overall comparisons. d, The relative body weight of old mice orally gavaged with polyamines prior 5-FU injection (5-FU + polyamines supplementation) or PBS prior PBS injection (vehicle control). The body weight of each mouse was normalized to its body weight at the time of injection. The data are presented as mean ± s.d. n = 6 for vehicle control group, n = 7 for 5-FU + polyamines supplementation group. The P value was calculated using the two-tailed Welch’s t-test for timepoint comparison and two-way ANOVA for overall comparisons. e, Left: the representative pictures of IHC staining for pH3 at day 7 following 5-FU (± polyamines gavage). Scale bar, 200 µm. The asterisks indicate the pH3⁺ crypts. Right: the change in % of pH3⁺ crypt; n = 3 mice 5-FU group; n = 4 mice 5-FU + polyamines group. The data are presented as mean ± s.d. The P value was calculated using the two-tailed Welch’s t-test for group comparison. f, A comparison of protein fold changes (FC) induced by 5-FU in old mice (x axis) and the effect of polyamines supplementation on old, 5-FU injected mice (y axis); proteomics data from intestinal crypts collected at 7 dpi. The proteins plotted in both comparisons were filtered for significance on the basis of q values (q < 0.05) calculated by Spectronaut using the unpaired t-test between replicates followed by correction for multiple testing. The colour code indicates the difference of the log₂FC between 5-FU and polyamines supplementation effects. The number of proteins in each quadrant is indicated. The correlation between the two treatments was assessed using Spearman’s correlation test. n = 6 vehicle control, n = 6 5-FU; n = 7 5-FU + polyamines supplementation. Source data

Extended Data Fig. 1. Proteome dynamics during young small intestinal epithelium regeneration.

a Quantification of crypt length in the indicated groups. PBS-treated mice from different days were combined. Each dot represents one mouse. n = 16 PBS, n = 5 day 2, n = 6 day 5, n = 8 day 7 5-FU. Data are presented as mean ± SD. p-value was calculated using two-tailed Welch’s t test. b Number of quantified protein groups in the indicated treatments and time points. PBS-treated young mice: n = 5 day 2, n = 5 day 5, n = 4 day 7. 5-FU-treated young mice: n = 5 day 2, n = 5 day 5, n = 4 day 7. c Schematic of experimental design, proteomics data analysis and visualization strategy. d Representative abundance profiles for proteins affected by 5-FU treatment and belonging to different clusters. Protein profiles are plotted using a loess smooth function. The shaded area around the regression line represents the 95% confidence interval. n = 14 PBS, n = 5 day 2, n = 5 day 5, n = 4 day 7 5-FU. e Protein abundance profiles of 40S (26 proteins) and 60S (41 proteins) ribosome complexes. Protein profiles are plotted using a loess smooth function. The shaded area around the regression line represents the 95% confidence interval. n = 14 PBS, n = 5 day 2, n = 5 day 5, n = 4 day 7 5-FU. f Protein abundance profiles of components of the MCM complex. Protein profiles are plotted using a loess smooth function. The shaded area around the regression line represents the 95% confidence interval. n = 14 PBS, n = 5 day 2, n = 5 day 5, n = 4 day 7 5-FU. g Left: Schematic of colony forming assay. Middle: Representative pictures of organoids in the different groups. Right: quantification of the growing organoids in the indicated groups normalized to the average of PBS, scale bar, 200 µm. n = 4 PBS; n = 3 day 2, n = 4 day 5, n = 4 day 7 5-FU. Data are presented as mean ± SD. p-value was calculated using two-tailed Welch’s t-test. Source data

Extended Data Fig. 2. Old mice show signs of delayed regeneration following 5-FU.

a Relative body weight of old mice treated with a single dose of 5-FU or PBS as control. Body weight of each mouse was normalized to its body weight at the day of injection. n = 12 PBS, n = 14 5-FU mice. Orange and light blue lines indicate the median body weight for each treatment group on each day. p-values were calculated using two-tailed Welch’s t-test by comparing the body weights of 5-FU treated mice with PBS-treated controls on the indicated day, and by two-way ANOVA for overall day and treatment comparisons. b Body weight changes of young and old mice after 5-FU. Each mouse’s body weight was normalized to its weight on the day of the PBS or 5-FU injection. At each time point, the average body weight change in PBS-treated controls was subtracted from that of 5-FU-treated mice. This analysis isolates the effect of 5-FU on body weight dynamics between young and old groups. (Averages obtained from: n = 18 PBS, n = 17 5-FU young mice and n = 12 PBS, n = 14 5-FU old mice). Data are presented as mean ± SD. p-values were calculated using two-tailed Welch’s t-test and two-way ANOVA. c Representative pictures of H&E staining of the aged small intestine (PBS and 2 dpi). Scale bars, 50 µm. d Percentage of crypts number relative to PBS. At each time point 5-FU values were normalized to the average PBS control mice and expressed as percentage. Young mice n = 4 per time point; old mice: n = 5 per time point. Data are presented as mean ± SD. p-value was calculated using two-tailed Welch’s t-test for time point comparison and two-way ANOVA for overall day and age comparisons. e Percentage of villus cellular density relative to PBS. At each time point 5-FU values were normalized to the average of PBS and expressed as percentage. Young mice n = 4 per time point; old mice: n = 5 day 2, n = 5 day 5, n = 4 day 7. Data are presented as mean ± SD. p-value was calculated using two-tailed Welch’s t-test for time point comparison and two-way ANOVA for overall day and age comparisons. f Percentage of crypts length relative to PBS. At each time point 5-FU values were normalized to the average of PBS and expressed as percentage. Young mice: n = 5 day 2, n = 6 day 5, n = 8 day 7; old mice: n = 5 day 2, n = 5 day 5, n = 6 day 7. Data are presented as mean ± SD. p-value was calculated using two-way ANOVA for overall day and age comparisons. Source data

Extended Data Fig. 3. Perturbation of proteostasis after injury in aged intestinal tissue.

a Left: Protein abundance profiles of the proteins related to the autophagy lysosome pathway. Right: Protein abundance profiles of the ubiquitin proteasome pathway. Proteostasis-related proteins list was obtained from https://www.proteostasisconsortium.com/pn-annotation/. Protein profiles are plotted using a loess smooth function. The shaded area around the regression line represents the 95% confidence interval. n = 14 PBS young, n = 5 day 2, n = 5 day 5, n = 4 day 7 5-FU young; n = 15 PBS old, n = 5 day 2, n = 5 day 5, n = 3 day 7 5-FU old. b–d Protein abundance profiles of ribosomal proteins (b; 67 proteins), 20S proteasome (c; 17 proteins) and MCM complex (d; 6 proteins). Protein profiles are plotted using a loess smooth function. The shaded area around the regression line represents the 95% confidence interval. n = 14 PBS young, n = 5 day 2, n = 5 day 5, n = 4 day 7 5-FU young; n = 15 PBS old, n = 5 day 2, n = 5 day 5, n = 3 day 7 5-FU old. e Left: representative immunoblot for the SQSTM1 protein in the indicated groups. Right: Quantification of SQSTM1 protein level relative to ponceau staining (loading control). Each dot represents one mouse. n = 3 mice per group. Data are presented as mean ± SD. p-value was calculated using two-tailed Welch’s t-test for time point comparison and two-way ANOVA for overall day and age comparisons. f Protein abundance profiles of the RPS6 protein. Protein profiles are plotted using a loess smooth function. The shaded area around the regression line represents the 95% confidence interval. n = 14 PBS young, n = 5 day 2, n = 5 day 5, n = 4 day 7 5-FU young; n = 15 PBS old, n = 5 day 2, n = 5 day 5, n = 3 day 7 5-FU old. g Left: representative immunoblot for the RPS6 protein. Right: Quantification of RPS6 protein level relative to ponceau staining (loading control). Each dot represents one mouse. n = 3-4 mice per group. Data are presented as mean ± SD. p-value was calculated using Welch’s t-test for time point comparison and two-way ANOVA for overall day and age comparisons. h Relative mRNA expression level of the p62 gene in the indicated groups. Data are presented as mean ± SD. p-value was calculated using Welch’s t-test for time point comparison and two-way ANOVA for overall day and age comparisons. i Left: representative immunoblot for the total ubiquitylated conjugates. Right: Quantification of total ubiquitylated proteins relative to ponceau staining (loading control). Each dot represents one mouse. n = 3 mice per group. Data are presented as mean ± SD. p-value was calculated using two-tailed Welch’s t-test for time point comparison and two-way ANOVA for overall day and age comparisons. j Left: Representative immunoblot for the puromycin incorporation. Right: Quantification of puromycin incorporation relative to ponceau staining (loading control). Each dot represents one mouse. n = 3 mice per group. Data are presented as mean ± SD. p-value was calculated using two-tailed Welch’s t-test for time point comparison and two-way ANOVA for overall day and age comparisons. Source data

Extended Data Fig. 4. CHX-induced perturbation of proteostasis in young mice.

a Schematic of the effect of CHX on intestinal regeneration in young mice, collection at day 4. b Representative immunoblot for puromycin incorporation on 5-FU injected mice (± CHX). c Left: Schematic of the effect of CHX on the small intestine of young mice (PBS injected). Right: Relative body weight of young mice treated with a single dose of PBS and injected twice with CHX or PBS as control. Body weight of each mouse was normalized to its body weight at the day of PBS injection (day 1). n = 3 PBS-PBS, n = 3 PBS-CHX young mice. Data are presented as mean ± SD. p-value is calculated with two-way ANOVA for overall group comparisons. d FACS gating strategy for annexin V assay used to analyze apoptotic cells in young and old organoids treated with 2.5 µg/ml of 5-FU (related to Fig. 3f). Panel a created with BioRender.com. Source data

Extended Data Fig. 5. Polyamines pathway during intestinal regeneration.

a Protein abundance profiles of the enzymes involved in polyamines biosynthetic pathways. Protein profiles are plotted using a loess smooth function. The shaded area around the regression line represents the 95% confidence interval. n = 14 PBS young, n = 5 day 2, n = 5 day 5, n = 4 day 7 5-FU young; n = 15 PBS old, n = 5 day 2, n = 5 day 5, n = 3 day 7 5-FU old. b Protein abundance profiles of the enzymes involved in urea cycle metabolism. Protein profiles are plotted using a loess smooth function. The shaded area around the regression line represents the 95% confidence interval. n = 14 PBS young, n = 5 day 2, n = 5 day 5, n = 4 day 7 5-FU young; n = 15 PBS old, n = 5 day 2, n = 5 day 5, n = 3 day 7 5-FU old. c, Quantification of proline and L-citrulline. d Quantification of arginosuccinic acid and arginine amino acids by LC-MS. Each dot represents one mouse. n = 4 mice per condition. Data are presented as mean ± SD. p-value was calculated using two-tailed Welch’s t-test for time point comparison and two-way ANOVA for overall day and age comparisons. e Protein abundance profiles of the Eukaryotic Translation Initiation Factor 5 A (EIF5A). Protein profiles are plotted using a loess smooth function. The shaded area around the regression line represents the 95% confidence interval. n = 14 PBS young, n = 5 day 2, n = 5 day 5, n = 4 day 7 5-FU young; n = 15 PBS old, n = 5 day 2, n = 5 day 5, n = 3 day 7 5-FU old. f Left: Schematic of young and old organoids in polyamines free basal media. On day 12 young and old organoids were passaged to induce regeneration. Right: representative pictures of fully developed organoids before passaging (baseline) and at day 3. Scale bars, 200 µm. Source data

Extended Data Fig. 6. Polyamines pathway deficiency in young, uninjured mice.

a Upper: Schematic of ODC1 KO induction and PBS treatment (TAM: tamoxifen). Lower: Relative body weight of young ODC1 KO and WT mice treated with a single dose of PBS. Body weight of each mouse was normalized to its body weight at the day of injection. Data are presented as mean. n = 2 ODC1 KO, n = 3 ODC1 WT mice. b Representative HnE stainings of ODC1 KO and WT small intestines at 7 days post PBS injection. Scale bars, 50 µm. c Quantification of number of crypts per millimeter of small intestine in the indicated groups (7 dpi). Data are presented as mean. Each dot represents one mouse. n = 3 PBS ODC1 WT group; n = 2 PBS ODC1 KO group. d Quantification of number of cells per one micrometer of villi in the indicated groups (7 dpi). Data are presented as mean. Each dot represents one mouse. n = 3 PBS ODC1 WT group; n = 2 PBS ODC1 KO group. Panel a created with BioRender.com. Source data

Extended Data Fig. 7. Effect of dietary restriction followed by re-feeding on intestinal epithelium regeneration.

a Quantification of EIF5A levels in intestinal crypts from old mice that underwent different dietary interventions. Based on proteomics data from Gebert et al. 2020. Data are presented as mean ± SD. p-value was calculated by two-tailed Welch’s t-test. n = 4 mice per group. b Immunoblot for hypusinated EIF5A and EIF5A proteins on freshly isolated crypts from old mice treated with PBS and re-feed at the indicated time points. c Relative body weight of young (left) and old (right) mice under DR for one month followed by two days of re-feeding or continuous DR prior to 5-FU or PBS injection. Body weight of each mouse was normalized to its body weight at the beginning of the dietary restriction period. n = 4 DR, n = 5 RF young mice and n = 4 DR, n = 3 RF old mice. Data are presented as mean ± SD. p-value was calculated using two-way ANOVA for overall day and treatment comparisons. d Left: representative immunoblot for the hypusinated EIF5A and total EIF5A proteins on aged mice in the indicated groups. Right: Quantification of the hypusinated EIF5A and total EIF5A proteins relative to ponceau staining (loading control). Data are presented as mean. Each dot represents one mouse. n = 2 mice per group. e Left: Representative immunoblot for the puromycin incorporated proteins on aged mice in the indicated groups. Right: Quantification of the puromycin incorporation relative to ponceau staining (loading control). Data are presented as mean. Each dot represents one mouse. n = 2 mice per group. f Left: Representative immunoblot for the pH3 protein on aged mice in the indicated groups. Right: Quantification of pH3 protein relative to ponceau staining (loading control). Data are presented as mean. Each dot represents one mouse. n = 2 mice per group. Source data

Extended Data Fig. 8. Effect of polyamines supplementation on uninjured intestinal epithelium.

a Schematic of polyamines supplementation in uninjured old mice. Mice were orally gavaged with polyamines (polyamines supplemented) or PBS (vehicle control) before PBS injection. b Relative body weight of old mice orally gavaged with polyamines (polyamines supplemented) or PBS (vehicle control) before PBS injection. The body weight of each mouse was normalized to its body weight at the time of injection. n = 6 vehicle control group, n = 8 polyamines supplemented group. Data are presented as mean ± SD. p-value was calculated using two-tailed Welch’s t-test for time point comparison and two-way ANOVA for overall comparisons. c Left: Representative pictures of immunohistochemistry (IHC) staining for pH3 at day 7 following polyamines supplementation. Scale bar, 200 µm. Asterisks indicate the pH3+ crypts. Right: Change in % of pH3⁺ crypt; n = 3 mice per group. Data are presented as mean ± SD. p-value was calculated using two-tailed Welch’s t-test for group comparison. d Model of intestinal regeneration dynamics. Left: in young mice, injury leads to rapid small intestine regeneration and body weight recovery within ~5 days without proteostasis imbalance. Middle: in old mice, injury induces proteostasis stress, leading to delayed regeneration and body weight recovery (~day 7). Right: old mice subjected to 30 days of dietary restriction followed by 2 days of re-feeding, or directly supplemented with polyamines prior to injury, exhibit elevated polyamine levels and proliferation. This leads to improved protein synthesis, reduced proteomic alterations induced by 5-FU, and ultimately results in rescued body weight loss and improved intestinal regeneration. Panel d created with BioRender.com. Source data