Intestinal circadian disruption in an IBD-relevant mouse model

The development of various GI diseases, including IBD [5], has been linked to circadian disruption, such as that occurring during shift work and frequent travel across time zones [2]. IL-10-deficient mice on a Sv129 background (IL-10−/− Sv129) exhibit a normal circadian behavioral phenotype (Fig. 1A, Supplementary Fig. 1A). Central circadian clock functions, such as rhythmic wheel-running activity and food intake during a light–dark (LD) cycle or constant darkness (DD), were indistinguishable from those of the controls, although total activity was slightly reduced in the IL-10−/− Sv129 mice (Fig. 1A, Supplementary Fig. 1B–D).

Fig. 1

Restricted feeding restores the disrupted colonic clock in IL-10−/−Sv129 mice and ameliorates colitis. A Representative actogram of wild type (IL-10+/+Sv129) and IL-10−/−Sv129 mice exposed to a light (yellow)-dark (gray) cycle (LD), constant darkness (DD), and nighttime-restricted feeding (RF, blue boxes). Tick marks represent running wheel activity. B Clock gene expression in colonic tissues from IL-10−/−Sv129 mice (Control AD and RF: n = 24; IL-10−/−Sv129AD: n = 24; IL-10−/−Sv129RF: n = 19; n = 3-4 mice/time points for each group) C Clock gene and D inflammatory genes expression in colonic tissues. E Quantification of the amount of CD4 + T cells in the colon LP from wild type and IL-10−/−Sv129 mice under AD and RF conditions. F Organ weights, G representative colonic cross section stained with H&E and H histopathological scores of colonic tissues from wild-type and IL-10−/−Sv129 mice under AD and RF conditions (top) and I quantification of individual mice with different degrees of inflammation (bottom). Tissues with scores between 0 and 2 were classified as noninflamed, those with scores between 3 and 5 were classified as mildly inflamed, and those with scores >5 were classified as highly inflamed. The data points represent individual mice (n = 3–4/time point for each group). Significant rhythms are illustrated with fitted cosine-wave regression using a line (significance: cos-fit p value ≤ 0.05). Two-way ANOVA with Tukey’s correction was also used. The data are presented as the means ± SEMs. Asterisks indicate significant differences; *p < 0.05, ***p < 0.001, ****p < 0.0001. Details on the number of mice per time point are summarized in the Supplementary Table 6

Recently, we demonstrated an imbalance in GI immune homeostasis in mice lacking a functional intestinal clock [11]. This prompted us to characterize the molecular intestinal clock in inflamed IL-10−/− Sv129 mice. Robust circadian rhythms were found in the expression of the clock genes Per1/2 (Period 1/2), Rev-erbα (NR1D1) and Dbp in the colonic epithelial cells of wild-type mice. This rhythmicity was lost, and gene expressions were strongly suppressed in IL-10−/− Sv129 mice (Fig. 1B), indicating a dysfunctional colonic clock. The circadian origin of these results was validated in colonic tissues collected at 6 different circadian times from IL-10−/− Sv129 mice kept in constant darkness and in cultured colonic organoids (Supplementary Fig. 1E, F). This effect was restricted to the colon since the major clock genes remained rhythmic in the liver, cecum and cultured jejunal organoids (Supplementary Fig. 2). Loss of rhythmicity in IL-10−/− Sv129 mice was also observed for inflammatory genes, e.g., Tnf, Ifnγ, Cxcr4, and Vcam-1, and T-cell recruitment to the lamina propria (LP) (Fig. 1D, E, Supplementary Fig. 1G, H). Furthermore, the peak expression of most clock genes was strongly correlated with the colonic histological score; the frequency of immune cells in the lamina propria, including macrophages, neutrophils, CD4+ cells and dendritic cells; and the expression of genes involved in barrier function (Supplementary Fig. 1I, J). These results indicate a causal relationship between colon clock dysfunction in IL-10−/− Sv129 mice and their inflammatory phenotype.

Nighttime-restricted feeding restored the colonic clock and ameliorated pathological changes in IL-10 −/− mice

Nighttime-restricted feeding (RF) is a widely used approach in mice to influence the rhythmicity of peripheral clocks [15]. We tested whether RF in IL-10−/− Sv129 mice restores the colonic clock and impacts their Crohn’s disease-like phenotype. Therefore, RF was gradually introduced to IL-10−/− Sv129 mice and controls, which were euthanized after 4 weeks of RF on 6 different circadian times (Fig. 1A, Supplementary Fig. 1A). Food intake and nighttime activity levels were comparable between genotypes and feeding conditions (Supplementary Fig. 1B–D). Importantly, RF substantially improved the circadian rhythmicity of clock gene expression in IL-10−/− Sv129 colonic tissue to similar levels as those observed in control mice under ad libitum (AD) conditions (Fig. 1C, Supplementary Table 1). Moreover, a restored rhythmicity of Tnf and reduced levels of Tnf and Ifn-y as well as a restored rhythmicity in the number of CD3+CD4+ cells isolated from the colonic LP and an overall reduced number of CD4 + T lymphocytes were observed in IL-10−/− Sv129 mice during RF in comparison to those in AD-fed IL-10−/− Sv129 mice (Fig. 1D, E, right). In contrast, CD3+CD8+ cell recruitment remained arrhythmic (Supplementary Fig. 1H). Consistently, the weights of the spleen and mesenteric lymph nodes (MLNs) and the colon density, which were much heavier in the AD-fed IL-10−/− Sv129 mice than in the control mice, were significantly lower during RF and were undistinguishable from those of the controls (Fig. 1F). Histological analysis of colon cross-sections further revealed that more than 58% of AD-fed IL-10−/−Sv129 mice were highly inflamed (Histoscore>5), which was dramatically reduced to 5% by RF (Fig. 1G–I). These data demonstrate that RF restores the disruption of colon clock function and completely ameliorates the immune phenotype in IL-10−/−Sv129 mice.

Loss of microbial rhythmicity during colonic inflammation in IL-10 −/− mice is restored by RF

Gut microbiota dysbiosis has been associated with the development of IBD in humans and mouse models [10]. Previously, we identified a novel link between the intestinal clock, microbiome rhythms and GI homeostasis [11]. Consistently, the microbiota composition differed significantly between genotypes despite the feeding conditions (Fig. 2A). Circadian rhythmicity in community diversity (normalized species richness), the abundance of both major phyla, Bacteroidota and Firmicutes, and highly abundant families, including Lachnospiraceae and Oscillospiraceae, as well as zOTUs, followed circadian oscillation in the wild type, whereas these rhythms were abolished in AD-fed IL-10−/−Sv129 mice (Fig. 2B–E). In accordance with the results obtained from IBD patients [17], alterations in the abundance of Lachnospiraceae, Oscillospiraceae and Erysipelotrichaceae were observed in IL-10−/− mice (Supplementary Fig. 3A). The abundances of the genera Oscillibacter, Eubacterium, Clostridium and Pseudoflavonifractor significantly differed among the genotypes (Supplementary Fig. 3Bleft, Supplementary Table 2). Moreover, we detected significant correlations between disease markers, such as histological score, inflammatory marker gene expression and zOTUs belonging to Lachnospiraceae and Oscillospiraceae (Supplementary Fig. 3Bright), which is consistent with our recent findings showing a correlation between Lachnospiraceae and active disease in IBD patients [18]. Importantly, RF not only restored the differences in abundance between genotypes (Supplementary Fig. 3B, Supplementary Table 3) but also restored microbial rhythmicity in IL-10−/−Sv129, as illustrated by the community diversity and at the levels of the major phyla and families (Fig. 2B–E). Enhanced microbial rhythmicity was also found among the 237 identified zOTUs visualized by heatmaps (Fig. 2D, Supplementary Fig. 3C). More than 50% of the zOTUs that gained rhythmicity during RF in IL-10−/− mice belonged to the Lachnospiraceae family (Supplementary Table 2), which includes bacterial taxa important for SCFA production and secondary bile acid (BA) conversion and thus plays crucial roles in IBD progression [19, 20].

Fig. 2

Impaired microbiome rhythms in IL-10−/−SV129 mice are restored after RF. A Beta-diversity illustrated by MDS plots of fecal microbiota based on generalized UniFrac distances (GUniFrac) in wild type and IL-10−/−Sv129 mice under AD and RF conditions. Comparison of genotype and feeding effects was performed by PERMANOVA (adj.p = 0.001). Data points represent individual sample collected from different mice at different time points (n > 5 mice/time points for each group). B Circadian profile of normalized richness and C relative abundance of phyla in wild type and IL-10−/−Sv129 mice under AD and RF conditions. D Heatmap depicting the relative abundance of zOTUs (mean relative abundance>0.1%; prevalence > 10%). Data from IL-10−/−Sv129 mice under AD or RF conditions are normalized to the peak of each zOTU and ordered by the peak phase in wild type mice under AD or RF conditions, respectively. Yellow and blue indicate high and low abundance, respectively. Each column represents the group average for each time point. E Relative abundance of highly abundant families and representative zOTUs from wild type and IL-10−/−Sv129 mice under AD and RF conditions. F Classification of the top 5 fully annotated super classes of metabolites that lost rhythmicity in IL-10−/−Sv129 mice under AD. Circadian profiles of bile acids (BAs) G measured by untargeted or H targeted metabolomics of wild type and IL-10−/−Sv129 mice under AD conditions. At some time points (CT 1, 4, 7, 16, 19, and 22) the levels for the IL-10−/− group were under the detection limit and thus not illustrated in the graph. I Heatmap of pathways (calculated by PICRUST 2.0) with restored rhythms in IL-10−/−Sv129 mice under RF (left) and the quantification of the superclass (right). Rhythmicity identified by JTK_Cycle (Bonferroni adj. p value ≤ 0.05). Significant rhythms are illustrated with fitted cosine-regression; data points connected by straight lines indicate no significant cosine fit curves (p > 0.05) and thus no rhythmicity. Comparisons of groups were performed by two-way ANOVA followed by Sidak correction. The results are summarized in Supplementary Table 3. Data are represented as the mean ± SEM. Details on mouse numbers per time point are summarized in Supplementary Table 6

To address the physiological relevance of disturbed microbial oscillations in IL-10−/−Sv129 mice, (un)targeted metabolite analysis was performed on feces. Indeed, 55 fully annotated metabolites were clustered according to genotype (Supplementary Fig. 3D) in IL-10−/−Sv129 mice, most of which were lipids and lipid-like metabolites, including fatty acids and BAs (Fig. 2F). For example, the rhythmicity of bile acids, including muricholic acid and 7-sulfocholic acid, as well as the long-chain fatty acid oleic acid and linoleic acid, was lost and suppressed (Fig. 2G, Supplementary Fig. 3E). Targeted metabolomics further validated these genotype differences (Fig. 2H, Supplementary Fig. 3F). For example, in IL-10−/−sv129 mice, a highly suppressed amplitude or lost time difference was observed for primary and secondary bile acids (BAs), such as deoxycholic acid (DCA), hyodeoxycholic acid, 7-sulfocholic acid (7-sulfo-CA) and α-muricholic acid (Fig. 2H). Most of these secondary BAs are key mediators involved in gut dysbiosis, especially in IBD [21]. Additionally, reduced SCFA levels, such as those of acetate, butyrate and propionate, were found in IL-10−/−sv129 mice (Supplementary Fig. 3F), similar to observations in IBD patients and related mouse models [19]. The concentrations of valeric acid and desaminotyrosin were also suppressed in IL-10−/−Sv129 mice (Supplementary Fig. 3F). Importantly, PICRUSt2 analysis of samples obtained from IL-10−/−Sv129 mice under RF conditions revealed restored rhythmicity of the assigned pathways involved in fatty acid synthesis and SCFA fermentation (Fig. 2I). Taken together, these results indicate that the changes in microbial rhythmicity and composition observed in IL-10−/−Sv129 mice can be reversed by RF; thus, microbial rhythmicity might be involved in the beneficial effect of RF on colonic inflammation.

Germ-free intestinal clock-deficient mice develop an increased inflammatory response following the transfer of IL-10 −/−-associated microbiota

Recently, we demonstrated that a functional intestinal clock is required to maintain GI homeostasis by driving the microbiome, including Lachnospiraceae and microbiota-derived DCA and 7-sulfo-CA [11]. To determine whether a dysfunctional intestinal clock in IL-10−/−Sv129 mice or the arrhythmicity of the microbiome directly affects the severity of inflammation, we performed two distinct cecal microbiota transfer experiments. First, colonialization of cecal content from intestine-specific clock-deficient (Bmal1IEC−/−) and littermate (Bmal1flox/flox) donors was performed to transfer rhythmic and arrhythmic microbiota into germ-free (GF) IL-10−/−BL6 recipients (Fig. 3A). Surprisingly, the weight of the immune organs and the amount of immune cell infiltration to the LP did not differ among the recipients (Fig. 3B). Consistently, histological staining, scoring, and inflammatory marker gene expression revealed no pathological differences between recipients (Fig. 3C), indicating that microbial rhythmicity might not directly induce intestinal inflammation. However, when GF Bmal1IEC−/− mice received disease-associated microbiota from inflamed IL-10−/−BL6 mice, colon weight, immune cell recruitment to the colonic LP and Tnf gene expression were severely elevated in Bmal1IEC−/− recipients compared to controls, although the histological score obtained from a single colon cross-section showed no clear difference between the recipients (Fig. 3D–G, Supplementary Fig, 4A–C). Interestingly, clustering between circadian time points and recipients was found for beta diversity in cecal samples (Fig. 3H). Alterations in the abundance of Lachnospiraceae and Erysipelotrichaceae in Bmal1IEC−/− recipients were similar to those in IL-10−/−Sv129 mice (Fig. 3I, Supplementary Fig. 3A). Accordingly, the cecal concentrations of α-MCA, DCA, HDCA and MDCA were lower than those in control recipients (Fig. 3J, Fig. 2H). Procrustes analysis (PA) revealed an association between zOTUs and BA production; in particular, Lachnospiraceae and Oscillospiraceae were correlated with DCA and aMCA concentrations (Fig. 3K, L). Taken together, these results reflect an early stage of inflammation following colonization with disease-associated microbiota when recipients lack a functional intestinal clock. Consequently, the intestinal clock represents a functional link between the microbiome and GI inflammation.

Fig. 3

Dysfunction of the host intestinal clock promotes microbiota-induced colonic inflammation. A Schematic illustration of the experimental design. Cecal contents from control (rhythmic) and Bmal1IEC−/− (arrhythmic) donors were transferred to germ-free IL-10−/−BL6 mice. B Organ weights (top) and number of immune cells recruited into the colonic LP (bottom) of the recipients. C Representative H&E staining of colonic cross sections from the recipients (left) and Tnf gene expression, as well as histopathological scores (right). D Schematic illustration of the experimental design. Disease-associated microbiota from IL-10−/− donors was transferred to germ-free control (Bmal1flox/flox) and Bmal1IEC−/− recipients. E Colon weight, F number of immune cells recruited into the colonic lamina propria and G Tnf expression in colon tissues from recipients. H Beta diversity illustrated by MDS plots of cecal microbiota based on generalized UniFrac distances (GUniFrac) in recipients at circadian times (CTs) 1 and 13. I Relative abundance of families and J amounts of α-muricholic acid (α-MCA), deoxycholic acid (DCA), hyodeoxycholic acid (HDCA) and murideoxycholic acid (MDCA) in the cecal contents of control and Bmal1IEC−/− recipients. K Procrustes analysis (PA) of the cecal microbiota and bile acid (BA) levels. The length of the line is proportional to the divergence between the data from the same mouse. L Representative correlation plot between zOTUs and BAs. The data points represent individual mice (n > 5/time points for each genotype). Mann–Whitney U tests and two-way ANOVA followed by Benjamini–Hochberg correction were used. Asterisks indicate significant differences; *p < 0.05, **p < 0.01, ****p < 0.0001. The data are presented as the mean ± SEMs. Details on the number of mice per time point are summarized in Supplementary Table 6

Genetic intestinal clock dysfunction alters colonic immune functions

To determine whether intestinal clock dysfunction influences the colonic immune response, we used Bmal1IEC−/− mice. Colonic IEC clock dysfunction was validated by arrhythmic and suppressed clock gene expression (Fig. 4A). Furthermore, bulk RNA-sequencing analysis of colonic tissue samples obtained during the circadian day revealed differential expression between genotypes (fold change>1.5, adj. p < 0.05), such as Retnlb (Resistin-like beta) and Tff2 (Trefoil factor 2) (Fig. 4B), which are involved in colitis development [22, 23]. Consistently, gene ontology enrichment analysis highlighted pathways relevant for ‘humoral immune response,’ ‘defense response to bacterium’ and ‘T-cell apoptotic process’ (Fig. 4C). Clustering according to the genotype was observed by principal component analysis (PCA), but clustering according to the collection time point was less pronounced in Bmal1IEC−/− (Fig. 4D). Consistently, 55% of rhythmic transcripts in the controls were arrhythmic in Bmal1IEC−/− mice (Fig. 4Eleft, Supplementary Table 4). The majority of the resilient rhythmic transcripts showed a highly suppressed amplitude in Bmal1IEC−/− mice (Fig. 4E, right). DODR analysis further confirmed that >200 genes either lost or changed rhythmicity (114/104, respectively) in Bmal1IEC−/− mice and were enriched in pathways related to “circadian rhythm” and “inflammatory response” (Fig. 4F). In particular, genes involved in the inflammatory and immune responses [24,25,26], including Ffar2, Hc, Abcc2 and Rorc, goblet cell function [27] and immune cell migration [28, 29], such as Cxcr4 and Vcam-1, lost rhythmicity in Bmal1IEC−/− mice (Fig. 4G, Supplementary Fig. 5A, B, Supplementary Table 4). Accordingly, immune cell frequency (CD4+ and CD8+ T cells and dendritic cells) lost rhythmicity in the colonic LP of Bmal1IEC−/− mice (Fig. 4H), similar to the results obtained for IL-10−/−Sv129 mice (Fig. 1E). Although a lack of the intestinal clock did not result in a pathophysiological phenotype (Supplementary Fig. 5C–E), neither the colon weight nor the concentration of complement 3 in the stools of Bmal1IEC−/− mice was significantly increased (Fig. 4I). These findings are consistent with the results obtained from IL-10−/− mice (Fig. 1F, Supplementary Fig. 5F), suggesting that intestinal clock dysfunction might indeed magnify inflammatory processes in IL-10−/− mice.

Fig. 4

The intestinal clock regulates transcripts involved in intestinal immune responses. A Clock gene expression and B volcano plot of transcriptional changes (upregulated: adjusted p < 0.05, FC > 1.5, red; downregulated: adjusted p < 0.05, FC > 1.5, blue) in colon tissues from control and Bmal1IEC−/− mice. C Gene Ontology enrichment analysis of differentially expressed genes. D Principal component analysis of 46 colon samples obtained from control (red) and Bmal1IEC−/− (blue) mice. Different shapes indicate different circadian time (CT) points. E Heatmap of transcripts that lost, maintained or gained rhythmicity in colon tissues from Bmal1IEC−/− mice (left) and amplitude comparison of the transcripts that maintained rhythmicity (right) (JTK_cycle adj.p < 0.05 as rhythmic). Transcripts are ordered according to the peak phase of the control groups. Each column represents the group average for each time point. F Number of transcripts whose rhythmic changes were lost/changed according to comparison (left) and GO analysis (top 5, right). Circadian profiles of G gene expression and immune cell frequency H in the colonic LP of control and Bmal1IEC−/− mice. I Colon weight (top) and complement 3 levels in the feces of control and Bmal1IEC−/− mice. The data points represent individual mice (n = 4/time points for each genotype). Mann–Whitney U test. Asterisks indicate significant differences; *p < 0.05, ****p < 0.0001. Significant rhythms are illustrated with fitted cosine-wave regression using a line (significance: cos-fit p value ≤ 0.05). Details on the number of mice per time point are summarized in Supplementary Table 6

Bmal1 IEC−/− mice are susceptible to acute and chronic colon inflammation

To directly determine the cause‒effect relationship between intestinal clock dysfunction and GI inflammation, Bmal1IEC−/−mice and controls were released into DD condition and received drinking water supplemented with 2% DSS (Fig. 5A) to induce tissue damage and create a model of acute colitis [30]. DSS treatment of Bmal1IEC−/− mice caused a significant reduction in body weight, an increase in the disease activity index (DAI) [31], and a decrease in colon length compared to DSS treated controls (Fig. 5B–D). Accordingly, the expression of inflammatory markers, such as Tnf, and neutrophil and macrophage recruitment to the colonic LP, were strongly enhanced in Bmal1IEC−/− mice (Fig. 5E, F). Consistently, in the same mice, histopathological evaluation of colon cross-sections revealed significant differences between genotypes after DSS treatment, which was reflected by increased histological scores in Bmal1IEC−/−mice (Fig. 5G, H). Additionally, intestinal clock dysfunction caused the loss of goblet cells following DSS treatment (Fig. 5I), which represents a critical factor during DSS-induced colitis [31]. These results demonstrated increased sensitivity to DSS-induced colitis in Bmal1IEC−/− mice.

Fig. 5

Bmal1IEC−/− mice are more susceptible to DSS-induced colitis. A Schematic illustration of the DSS treatment procedure. Body weight changes B and disease activity indices C of control and Bmal1IEC−/− mice during DSS/water administration. D Representative images of the colon (left) and colon length measurements (right). E Gating strategy for macrophages and neutrophils in the colon LP and quantification. Tnf expression in colon tissues F, histopathological scores G, representative H&E staining H and PAS-AB staining quantification of the number of goblet cells based on a minimum of 20 crypts for each section I of colon sections from control and Bmal1IEC−/− mice with/without DSS treatment. The data points represent individual mice (n > 8 for each group). Two-way ANOVA and Benjamini‒Hochberg correction were used. Asterisks indicate significant differences between genotypes after DSS treatment; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Octothorpes indicate significant differences between Bmal1flox/flox mice after treatment (#p < 0.05, ##p < 0.01, ###p < 0.001). Details on the number of mice per time point are summarized in Supplementary Table 6

To further assess the impact of the intestinal clock on IBD development, we evaluated IBD pathology in a newly generated genetic mouse model prone to developing chronic colitis combined with a dysfunctional intestinal clock (Bmal1IEC−/−xIL-10−/−BL6). IL-10 deficiency on the BL6 background causes less severe colitis symptoms than on the SV129 background [16], which is likely a result of less severe alterations in clock gene expression (Fig. 6A, Fig. 1B, Supplementary Fig. 6F). The additional loss of the intestinal clock dramatically reduced the survival rates of Bmal1IEC−/−xIL-10−/−BL6 mice (Fig. 6B). In particular, all Bmal1IEC−/−x IL-10−/−BL6 mice were euthanized before the end of the experiment due to their severe disease burden, while approximately 30% of IL-10−/−BL6 mice survived. More severe tissue inflammation was observed in Bmal1IEC−/−xIL-10−/−BL6 mice, which was reflected by increased spleen, colon and MLN weights and increased disease scores (Fig. 6C–E). Similar to our previous results obtained from Bmal1IEC−/− mice [11], the microbiota diversity differed between genotypes, and circadian rhythmicity in terms of species richness and the abundance of the major phyla was dramatically disrupted in Bmal1IEC−/−xIL-10−/−BL6 mice, whereas only mild circadian changes were observed in IL-10−/−BL6 mice (Supplementary Fig. 6A–C). Moreover, a heatmap showing the peak relative abundances of zOTUs and their quantification confirmed the disruption of microbial rhythmicity in Bmal1IEC−/−xIL-10−/−BL6 mice, whereas intermediate phenotypes were found in IL-10−/−BL6 mice (Supplementary Fig. 6D, E). Taken together, these data demonstrate that a lack of the intestinal clock promotes sensitivity to acute and chronic colon inflammation.

Fig. 6

Restricted feeding ameliorates the IBD-like symptoms in mice by targeting the intestinal clock. Colon clock genes expression A and survival B in control, Bmal1IEC−/−, IL-10−/−BL6 and Bmal1IEC−/−xIL-10−/−BL6 mice under AD and RF conditions. Disease scores C, organ weights D, and representative H&E staining images of colon sections E from IL-10−/−BL6 and Bmal1IEC−/−xIL-10−/−BL6 mice under AD conditions. F Schematic illustration of the experimental RF design. Survival analysis G, representative H&E staining of colon sections H, histopathological scores I, immune cell frequency in the colonic LP J and Tnf expression K in IL-10−/−BL6 and Bmal1IEC−/−xIL-10−/−BL6 mice under AD and RF conditions. The data points represent individual mice (n > 6/time points for each group). One-way ANOVA, two-way ANOVA following Benjamini‒Hochberg correction, and three-way ANOVA followed by Benjamini–Krieger-–Yekutieli correction were used. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. The data are presented as the mean ± SEM. Details on the number of mice per time point are summarized in Supplementary Table 6

Restricted feeding requires a functional intestinal clock to ameliorate colitis symptoms

To evaluate whether the beneficial effect of RF on the severity of GI inflammation observed in IL-10−/−Sv129 mice is caused directly by restoration of the gut clock function, Bmal1IEC−/−xIL-10−/−BL6 mice, IL-10−/−BL6 mice and controls were exposed to RF conditions (Fig. 6F). RF in IL-10−/−BL6 mice restored clock gene expression to control levels at CT13, whereas its expression was not restored in Bmal1IEC−/− mice or Bmal1IEC−/−xIL-10−/−BL6 mice due to a lack of intestinal Bmal1 (Fig. 6A, Supplementary Fig. 6F). Assessment of the humane endpoint of the experiment revealed that RF failed to improve the survival rate of Bmal1IEC−/−xIL-10−/−BL6 mice (9/15, 60%), but RF significantly improved the survival rate of IL-10−/−BL6 mice (9/9, 100%) (Fig. 6G). Consistently, according to the pathohistological evaluation, CD4+ T-cell recruitment and colonic Tnf expression were greater than the control levels in IL-10−/−BL6 mice under RF conditions (histology: p = 0.0138, CD4 cell recruitment: p = 0.0069, TNF expression: p = 0.0404) but remained indistinguishable between AD and RF conditions in Bmal1IEC