Effects of caloric restriction on the gut microbiome are linked with immune senescence

Abstract

Background: Caloric restriction can delay the development of metabolic diseases ranging from insulin resistance to type 2 diabetes and is linked to both changes in the composition and metabolic function of the gut microbiota and immunological consequences. However, the interaction between dietary intake, the microbiome, and the immune system remains poorly described.

Results: We transplanted the gut microbiota from an obese female before (AdLib) and after (CalRes) an 8-week very-low-calorie diet (800 kcal/day) into germ-free mice. We used 16S rRNA sequencing to evaluate taxa with differential abundance between the AdLib- and CalRes-microbiota recipients and single-cell multidimensional mass cytometry to define immune signatures in murine colon, liver, and spleen. Recipients of the CalRes sample exhibited overall higher alpha diversity and restructuring of the gut microbiota with decreased abundance of several microbial taxa (e.g., Clostridium ramosum, Hungatella hathewayi, Alistipi obesi). Transplantation of CalRes-microbiota into mice decreased their body fat accumulation and improved glucose tolerance compared to AdLib-microbiota recipients. Finally, the CalRes-associated microbiota reduced the levels of intestinal effector memory CD8+ T cells, intestinal memory B cells, and hepatic effector memory CD4+ and CD8+ T cells.

Conclusion: Caloric restriction shapes the gut microbiome which can improve metabolic health and may induce a shift towards the naïve T and B cell compartment and, thus, delay immune senescence. Understanding the role of the gut microbiome as mediator of beneficial effects of low calorie diets on inflammation and metabolism may enhance the development of new therapeutic treatment options for metabolic diseases.

Trial registration: NCT01105143 , "Effects of negative energy balance on muscle mass regulation," registered 16 April 2010. Video Abstract.

Keywords: Adaptive immune system; Caloric restriction; Gut microbiota; Immune senescence; Obesity.

Conflict of interest statement

The authors have no conflict of interest to declare.

© 2022. The Author(s).

Figures

Fig. 1

Caloric restriction changes the gut microbiome composition in humanized gnotobiotic mice. A Number of observed amplicon sequence variants (ASVs) and Shannon index indicate microbial richness and diversity for 3 weeks following colonization. P-values are given for repeated measures non-parametric ANOVA-type testing for differences between groups and time points (n = 104 stool samples across 23 mice and 5 time points). B Principal coordinates of Bray-Curtis dissimilarity between stool samples. Color indicates groups. Numeric labels denote the 5 time points. Ellipses enclose group masses (99% CI) for day 21. P-values are given for adonis permutational analysis of variance testing for group × time differences. C Heatmap representation of day 21 differentially abundant ASVs (n = 20). ASVs are shown at the most specific assigned taxonomy with FDR-adjusted P-values < 0.01 in DESeq2 two-sided Wald test. Bubbles indicate log-fold change (LFC) between the two groups. Heatmap colors indicate relative abundances as variance-stabilizing (log2) transformed counts ranging from 4 (dark blue, corresponding to 0 raw counts) to 14 (yellow, corresponding to 14K raw counts). Color of bubbles and species labels denote phylum. Bubble sizes are proportional to − log10 FDR-adjusted P-values. Dendrogram leaf numbers indicate individual housing cages

Fig. 2

Colonization with CalRes-associated gut microbiota alters body fat and glucose clearance. Metabolic analysis of germ-free (GF) mice and mice inoculated with AdLib and CalRes human gut microbiota. A–C Spleen (A), epigonadal white adipose tissue (eWAT) (B), and liver (C) weights from GF and colonized mice. D Feces weight was measured using bomb calorimetry in GF and colonized mice. E Caecum weights from GF and colonized mice were analyzed with and without fecal contents. F Fasting male adult GF or colonized mice maintained on normal chow diet were challenged with oral glucose and blood was sampled for glucose at times indicated. G Area under the OGTT glucose-time curve (AUC). * P < 0.05, ** P < 0.01, **** P < 0.0001 as determined using ANOVA with Bonferonni’s post-test correction for multiple comparisons; error bars = SEM; ns = not significant

Fig. 3

The CalRes microbiota reduces levels of intestinal effector memory CD8+ T cells and memory B cells. A, B The heatmaps show cluster phenotypes based on the expression of canonical lineage markers on pre-gated TCRß+ T cells (A) and on pre-gated TCRß−CD19−NK1.1− innate immune cells (B). The differential expression of each selected surface marker (rows) is shown for each immune cell cluster (columns). The significance levels of the comparison between the three mouse groups for each immune cell cluster are depicted by semi-supervised hierarchical clustering. The top bubbles denote clusters with significantly different abundances between the three groups. Bubble colors indicate one of the two groups being compared with higher average cellular frequencies; bubble size indicates the -log2 FDR-adjusted p-values. C Relative proportions of CD4+ (left panel) and CD8+ (right panel) naïve (Tnaïve, CD44−CD62L+), central memory (TCM, CD44+CD62L+), effector memory (TEM, CD44+CD62L−) and terminally differentiated effector memory T cells (TEMRA, CD44−CD62L−) measured by mass cytometry. T cell populations were manually gated according to established lineage markers. D–H Relative proportions of total B cells (D), naïve B cells (E), “switched” memory B cells (F), NK cells (G) and activated CD62L+CD11b+ NK cells (H). n = 9 or more mice per group. * P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001 as determined using Student’s t-test or Mann-Whitney test, dependent on the distribution of the data

Fig. 4

Colonization with human gut microbiota induces alterations of splenic immune cell subsets. A, B The heatmaps show the distribution of splenic immune lineages based on the expression of canonical lineage markers by t-SNE on pre-gated TCRß+ T cells (A) and on pre-gated TCRß−CD19−NK1.1− innate immune cells (B). The differential expression of each selected surface marker (rows) is shown for each immune cell cluster (columns). The significance levels of the comparison between the three mouse groups for each immune cell cluster are depicted by semi-supervised hierarchical clustering. The top bubbles denote clusters with significantly different abundances between the three groups. Bubble colors indicate one of the two groups being compared with higher average cellular frequencies; bubble size indicates the − log2 FDR-adjusted P-values. C Relative proportions of CD4+ (left panel) and CD8+ (right panel) naïve (Tnaïve, CD44−CD62L+), central memory (TCM, CD44+CD62L+), effector memory (TEM, CD44+CD62L−) and terminally differentiated effector memory T cells (TEMRA, CD44−CD62L−) measured by mass cytometry. T cell populations were manually gated according to established lineage markers. D–H Relative proportions of B cells (D), naïve B cells (E), “switched” memory B cells (F), NK cells (G), and activated CD62L+CD11b+ NK cells (H). n = 9 or more mice per group. * P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001 as determined using Student’s t-test or Mann-Whitney test, dependent on the distribution of the data

Fig. 5

Caloric restriction-associated gut microbiota reduces levels of hepatic effector memory CD4+ and CD8+ T cells. A, B The heatmaps show the distribution of hepatic immune lineages based on the expression of canonical lineage markers by t-SNE on pre-gated TCRß+ T cells (A) and on pre-gated TCRß−CD19−NK1.1− innate immune cells (B). The differential expression of each selected surface marker (rows) is shown for each immune cell cluster (columns). The significance levels of the comparison between the three mouse groups for each immune cell cluster are depicted by semi-supervised hierarchical clustering. The top bubbles denote clusters with significantly different abundances between the three groups. Bubble colors indicate the one of the two groups being compared with higher average cellular frequencies; bubble size indicates the − log2 FDR-adjusted p-values. C Relative proportions of CD4+ (left panel) and CD8+ (right panel) naïve (Tnaïve, CD44−CD62L+), central memory (TCM, CD44+CD62L+), effector memory (TEM, CD44+CD62L−) and terminally differentiated effector memory T cells (TEMRA, CD44−CD62L−) measured by mass cytometry. T cell populations were manually gated according to established lineage markers. D–H Relative proportions of B cells (D), naïve B cells (E), “switched” memory B cells (F), NK cells (G), and activated CD62L+CD11b+ NK cells (H). n = 9 or more mice per group. * P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001 as determined using Student’s t-test or Mann-Whitney test, dependent on the distribution of the data

Fig. 6

Immunologic changes correlate with gut microbial alterations. Heatmaps show latent correlation matrices between abundances of amplicon sequence variants (ASVs) detected in stool samples and all immune parameters analyzed in colon (A) and spleen (B) of mice 21 days after inoculation with AdLib and CalRes human gut microbiota. Immune parameters are expressed as frequencies, i.e., percent of parent, except those labeled # which were quantified as absolute cell counts. Heatmaps were ordered according to rows and columns first principal components to highlight cross-correlation structures. Asterisks indicate variables that were selected in L1-penalized sparse canonical correlation analysis (CCA). Circular chord plots display latent correlation between frequencies of manually defined immune subsets and L1-selected ASVs including the top ten taxa that either positively (upper) or negatively (lower) associate with the immunological dataset. Blue to red color scale in heatmaps and chords indicates negative and positive correlation values. Color of row-legend bar and species labels denotes the phylum level. Colors of column legend bars indicate parental lineage and differentiation level (antigen-experience) of lymphocyte subsets, respectively. Boxplot insets show how experimental groups as a latent variable are explained by the canonical covariate