Colonic Epithelial Circadian Disruption Worsens Dextran Sulfate Sodium-Induced Colitis

Abstract

Background: Disruption of central circadian rhythms likely mediated by changes in microbiota and a decrease in gut-derived metabolites like short chain fatty acids (SCFAs) negatively impacts colonic barrier homeostasis. We aimed to explore the effects of isolated peripheral colonic circadian disruption on the colonic barrier in a mouse model of colitis and explore the mechanisms, including intestinal microbiota community structure and function.

Methods: Colon epithelial cell circadian rhythms were conditionally genetically disrupted in mice: TS4Cre-BMAL1lox (cBMAL1KO) with TS4Cre as control animals. Colitis was induced through 5 days of 2% dextran sulfate sodium (DSS). Disease activity index and intestinal barrier were assessed, as were fecal microbiota and metabolites.

Results: Colitis symptoms were worse in mice with peripheral circadian disruption (cBMAL1KO). Specifically, the disease activity index and intestinal permeability were significantly higher in circadian-disrupted mice compared with control animals (TS4Cre) (P < .05). The worsening of colitis appears to be mediated, in part, through JAK (Janus kinase)-mediated STAT3 (signal transducer and activator of transcription 3), which was significantly elevated in circadian-disrupted (cBMAL1KO) mice treated with DSS (P < .05). Circadian-disrupted (cBMAL1KO) mice also had decreased SCFA metabolite concentrations and decreased relative abundances of SCFA-producing bacteria in their stool when compared with control animals (TS4Cre).

Conclusions: Disruption of intestinal circadian rhythms in colonic epithelial cells promoted more severe colitis, increased inflammatory mediators (STAT3 [signal transducer and activator of transcription 3]), and decreased gut microbiota-derived SCFAs compared with DSS alone. Further investigation elucidating the molecular mechanisms behind these findings could provide novel circadian directed targets and strategies in the treatment of inflammatory bowel disease.

Keywords: circadian disruption; metabolomics; microbiota; peripheral circadian disruption; ulcerative colitis.

Figure 1.

Dextran sulfate sodium (DSS) induces changes on the intestine that are sex and genotype dependent. These analyses were conducted in both female and male mice. A, Disease activity indices (DAIs) for male mice (top) and female mice (bottom). Multiple comparison analysis revealed that cBMAL1KO + DSS male mice had significantly higher DAI on days 4 to 8 than TS4Cre + DSS male mice, and cBMAL1KO + DSS female mice had significantly higher DAI on days 4 to 6 than TS4Cre + DSS female mice: ****q ˂ 0.0001. Additionally, mortality is demonstrated on these graphs: ⊗. Mortality rate for cBMAL1KO + DSS male mice was 7 (28%) of 25, while no TS4Cre + DSS male mice died (chi-square: P = .025). The mortality rate for cBMAL1KO + DSS female mice was 4 (16%) of 25 and 1 (4.8%) of 21 for TS4Cre + DSS female mice (chi-square: P = .22). B, Percent change in weight by group. Multiple comparison analysis indicated TS4Cre + DSS male mice had significant weight loss compared with TS4Cre control male mice on days 6 to 9 (P < .05). cBMAL1KO + DSS male mice had significant weight loss compared with cBMAL1KO control male mice on days 3 to 9 (P < .05). TS4Cre + DSS female mice had significant weight loss compared with TS4Cre control female mice on days 7 to 9 (P < .05). cBMAL1KO + DSS female mice had significant weight loss compared with cBMAL1KO control female mice on days 3 to 9 (P < .05). C, Colon lengths for male mice (top) and female mice (bottom). Multiple comparison analysis indicated that both male and female TS4Cre and cBMAL1KO mice had significant DSS-induced colon shortening compared with their respective control animals (q < 0.0001). Both cBMAL1KO + DSS male mice and female mice had more colon shortening, as compared with TS4Cre + DSS male mice (q < 0.0001) and female mice (q < 0.004): **q ˂ 0.01; ****q ˂ 0.0001; ^ϕP ˂ .05 for treatment effect; ^ΨP < .05 for genotype effect. Male and female TS4Cre DSS-induced mice did not differentiate between colon lengths (independent t test: P = .456), whereas male cBMAL1KO + DSS indicated a significant shorter colon length as compared with female cBMAL1KO + DSS (independent t test: P < .0001). D, Histology for TS4Cre and cBMAL1KO male mice and female mice with and without DSS. A 3-way analysis of variance revealed that there was a treatment effect (P < .0001) and genotype effect (P < .0007) but there was no difference in sex effect for histology scores.

Figure 2.

Dextran sulfate sodium (DSS) induces changes on the intestinal barrier that are genotype dependent. These analyses were conducted only in male mice. A, In vivo permeability demonstrated by percent excretion of oral dose of sucralose. A significant treatment effect was observed (P < .0001). Multiple comparison analysis showed significantly increased sucralose permeability in both DSS-treated male mice groups compared with their respective control animals. B, In vivo permeability demonstrated by the ratio or percent excretion of oral dose of sucralose to lactulose. A significant treatment effect (P < .0001) and genotype effect (P = .0045) were found. Multiple comparison analysis showed significantly increased sucralose-to-lactulose ratio permeability in both DSS-treated male mice groups compared with their respective control animals, as well as that the cBMAL1KO + DSS male mice sucralose-to-lactulose ratio significantly enhanced when compared with TS4Cre + DSS male mice. C, Immunoblotting for ZO-1 revealed a significant effect of genotype (P = .003) and treatment (P = .016). D, Immunofluorescence staining for claudin-2 revealed a significant treatment effect (P = .048) but no genotype effect. E, Immunofluorescence staining for occludin demonstrated no effects of treatment or genotype. F, Immunofluorescence staining for E-cadherin revealed a significant effect of genotype (P = .043) and treatment (P = .024). Statistical analyses included 2-way analysis of variance corrected for multiple testing using the Benjamini-Hochberg method. *q ˂ 0.05; **q ˂ 0.01; ****q ˂ 0.0001; ^ϕP < .05 for treatment effect; ^ΨP < .05 for genotype effect.

Figure 3.

Systemic and intestinal inflammation are genotype dependent. These analyses were conducted only in male mice. A, Serum interleukin (IL)-6 did not indicate a significant difference for genotype or treatment effects. However, multiple comparison analysis was significantly different between cBMAL1KO control and cBMAL1KO + DSS male mice (P = .016) B, Serum tumor necrosis factor alpha (TNF-α) exhibited only a treatment effect (P = .004). Multiple comparison analysis showed that the serum TNF-α level was significantly elevated (q ˂ 0.02) in the TS4Cre + DSS compared with TS4Cre control male mice. C, Western blot analysis of phospho-STAT3 (pSTAT3) protein exhibited both a treatment effect (P = .032) and genotype effect (P = .038) Immunohistochemistry stains of pSTAT3 and DAPI in the colons of the mice with and without DSS. Statistical analyses included 2-way analysis of variance corrected for multiple testing using the Benjamini-Hochberg method. ^#P < .05; *q ˂ 0.05; ^ϕP < .05 for treatment effect; ^ΨP < .05 for genotype effect.

Figure 4.

Principal coordinate analysis with Bray-Curtis dissimilarity. Results revealed that fecal samples clustered separately indicating that the fecal community is significantly different from each male mice group. Orientations of: A, 4 male mice sample groups; B, TS4Cre and cBMAL1KO samples; C, TS4Cre + DSS and cBMAL1KO + DSS; D, TS4Cre and TS4Cre + DSS; and E, cBMAL1KO and cBMAL1KO + DSS.

Figure 5.

Microbial profiles of male TS4Cre and cBMAL1KO mice with and without dextran sulfate sodium (DSS). These analyses were conducted only in male mice. A, Stacked column plots depicting the percent mean relative abundance (>1%) of bacterial phyla. B, Stacked column plots depicting the percent mean relative abundance (>1%) of bacterial genera. Percent mean relative abundance of (C) total short chain fatty acid–producing taxa, (D) total butyrate-producing taxa, and (E) putative proinflammatory bacteria-to-total SCFA ratio examined are shown across mice groups. Statistical analyses included 2-way analysis of variance corrected for multiple testing using the Benjamini-Hochberg method. ^#P < .05; *q ˂ 0.05; ^ΨP < .05 for genotype effect.