Colitis-associated intestinal microbiota regulates brain glycine and host behavior in mice

Abstract

Inflammatory bowel diseases (IBD) are chronic and relapsing inflammatory disorders of the gastrointestinal tract with complex etiology and no strategies for complete cure. IBD are often complicated by mental disorders like anxiety and depression, indicating substantial shifts in the microbiota gut-brain axis. However, the mechanisms connecting IBD to mental diseases are still under debate. Here we use Muc2 knockout mouse model of chronic colitis to uncouple the effects of the intestinal microbiota on host behavior from chronic inflammation in the gut. Muc2 knockout male mice exhibit high exploratory activity, reduced anxiety-related behaviors, impaired sensorimotor gating, and altered social preference towards males and females. Microbial transfer to wild-type mice via littermate co-housing shows that colitis-associated microbiota rather than inflammation per se defines behavioral features in Muc2 colitis model. Metagenomic profiling and combination of antibiotic treatments revealed that bacterial species Akkermansia muciniphila is associated with the behavioral phenotype in mutants, and that its intestinal abundance correlates with social preference towards males. Metabolomic analysis together with pharmacological inhibition of Gly and NMDA receptors helped us to determine that brain glycine is responsible for the behavioral phenotype in Muc2 mice. Blood and brain metabolic profiles suggest that microbiota-dependent changes in choline metabolism might be involved in regulation of central glycine neurotransmission. Taken together, our data demonstrates that colitis-associated microbiota controls anxiety, sensorimotor gating and social behavior via metabolic regulation of the brain glycinergic system, providing new venues to combat neurological complications of IBD.

Figure 1

Behavioral traits of Muc2^−/− animals. (A) Behavior in home cage (n = 9–12). Motor activity: t = 2.46, p = 0.024; water consumption rate: t = − 3.59, p = 0.002, Student's t-test. (B) Open field test (n = 18–20). Distance: t = − 1.91, p = 0.06, Student's t-test; rearings: t = − 2.326, p = 0.026, Student's t-test; time in the center: Z = − 3.333, p < 0.001, Mann–Whitney u test vs. C57BL/6. (C) Light–dark test (n = 17–20). Distance: t = − 3.067, p = 0.004; time: t = − 3.08, p = 0.004; entries: t = − 3.66, p < 0.001, Student's t-test. (D) Marble burying test (n = 17–20). Number of buried marbles: Z = 2.93, p = 0.003; Mann–Whitney u test. (E) Habituation and startle reflex (n = 16–18). Startle reflex: t = 3.246, p = 0.003, Student's t-test; habituation: Z = 2.17, p = 0.03; Mann–Whitney u test. (F) Two intruders test (n = 10–11). There was a statistically significant interaction between the resident genotype and the intruder gender (number of contacts: F(1, 38) = 21.990, p < 0.001; duration: F(1, 38) = 26.045, p < 0.001, two-way ANOVA). Number of contacts (C57BL/6, male vs. female): p < 0.001, Fisher’s LSD test; duration (C57BL/6, male vs. female): p < 0.001, Fisher’s LSD test. Number of contacts with a male/female intruder vs. C57BL/6: p < 0.01, Fisher’s LSD test; duration of contact with a male/female intruder vs. C57BL/6: p < 0.001, Fisher’s LSD test. (G) Odor preference test (n = 12–21). There was a statistically significant influence of the gender of animal whose odor was presented to the tested male F(1, 62) = 34.17 p < 0.001 two-way ANOVA. Duration of sniffing (male vs. female): p < 0.001, Fisher’s LSD test. * = p < 0.05, ** = p < 0.01, *** = p < 0.001, vs. C57BL/6. ### = p < 0.001, male vs. female.

Figure 2

Behavioral traits are associated with microbiota of Muc2^−/− mice. (A) The scheme of the littermate co-housing method. (B) Open field test (n = 18–20). Distance: t = − 3.895, p < 0.001; rearings: t = − 4.97, p < 0.001, Student's t-test; time in the center: Z = − 2.997, p < 0.001, Mann–Whitney u test. (C) Light–dark test (n = 17–20). Distance: t = − 4.019, p < 0.001; time: t = − 3.584, p < 0.001; entries: t = − 4.26, p < 0.001, Student's t-test. (D) Marble burying test (n = 18–20). Number of buried marbles: Z = 2.22, p = 0.026; Mann–Whitney u test. (E) Two intruders test (n = 10–11). Two-way ANOVA revealed a statistically significant interaction between the resident group and the intruder gender (number of contacts: F(1, 38) = 15.490, p < 0.001; duration: F(1, 38) = 28.306, p < 0.001 Number of contacts (C57BL/6, male vs. female): p < 0.001, Fisher’s LSD test; duration (C57BL/6, male vs. female): p < 0.001, Fisher’s LSD test. Number of contacts with a male/female intruder vs. C57BL/6: p = 0.008, Fisher’s LSD test; duration of contact with a male/female intruder vs. C57BL/6: p < 0.001, Fisher’s LSD test. * = p < 0.05, ** = p < 0.01, *** = p < 0.001, vs. C57BL/6. ### = p < 0.001, male vs. female.

Figure 3

Intestinal microbiota defines behavioral phenotype in the Muc2 mouse model of colitis. (A) PCA of metagenomic data based on top 70 abundant genera (n = 4). There was a significant effect of the group in PC1 (p = 0.044, Kruskal–Wallis test). There was a significant difference between C57BL/6 and Muc2^−/− and C57BL/6 and Muc2^+/+ groups (Muc2^−/−: Z = 2.16, p = 0.03; Muc2^+/+: Z = 2.16, p = 0.03, Mann–Whitney u-test). (B) Qualitative changes in average relative abundance of top 10 abundant genera. (C) Open field test after AMC treatment (n = 10/group). Distance: t = 2.34, p = 0.03; rearings: t = − 1.82, p = 0.086; time in the center: Z = − 1.97, p = 0.049; Mann–Whitney u test. (D) Light–dark test after AMC treatment (n = 10/group). In the lit compartment: time: t = 2.30, p = 0.034; distance: t = 2.05, p = 0.055; entries: t = 2.008, p = 0.059, Student's t-test. (E) Open field test after VR treatment (n = 9/group). Distance: t = − 1.79, p = 0.09; rearings: t = − 2.47, p = 0.025; time in the center: Z = − 1.97, p = 0.049; Mann–Whitney u test. (F) Light–dark test after VR treatment (n = 9/group). In the lit compartment: time: t = − 2.553, p = 0.022; distance: t = − 2.09, p = 0.052; entries: t = − 0.643, p = 0.529, Student's t-test. (G) Two intruders test after VR treatment (n = 9–10). Two-way ANOVA revealed a statistically significant interaction between the treatment and the intruder gender (number of contacts: F(1, 34) = 28.181, p < 0.001; duration: F(1, 34) = 56.854, p < 0.001. Number of contacts (C57BL/6, male vs. female): p < 0.001, Fisher’s LSD test; duration (C57BL/6, male vs. female): p < 0.001, Fisher’s LSD test; number of contacts with a male/female intruder vs. treatment: p < 0.001, Fisher’s LSD test; duration of contact with a male/female intruder vs. treatment: p = 0.001, Fisher’s LSD test. (H) Correlation analysis between social interaction with a female and relative abundance of A. muciniphila in the intestinal contents in the control and VR-treated animals: r = − 0.595, p = 0.007, Pearson’s correlation analysis. * = p < 0.05, ** = p < 0.01, *** = p < 0.001, vs. treatment. ### = p < 0.001, male vs. female.

Figure 4

Metabolomic profiling of blood and brain of Muc2^−/− and Muc2^+/+ mice. (A) PCA analysis of blood NMR metabolic profiles (n = 10/group). There was a significant effect of a group on PC1, which accounted for about 26.6% of variance (F (2, 27) = 8.84, p < 0.001, one-way ANOVA). Metabolomic profile of the C57BL/6 was significantly different from the two other groups (Muc2^−/−: p = 0.002; Muc2^+/+: p < 0.001, Fisher’s LSD test). (B) Volcano plots of blood metabolites as revealed by NMR. Horizontal line depicts a cut-off at p = 0.05. Vertical line depicts the ratio of 1. Metabolites with differences at p < 0.01 are shown in red. (C) Volcano plots of brain metabolites as revealed by NMR (n = 5–6). (D) The schemes of glycine biosynthesis from serine and betaine. (E) The levels of metabolites related to glycine biosynthesis. The data of the Kruskal–Wallis test followed by Mann–Whitney u-test are presented in Supplementary tables 3 and 4. * = p < 0.05, ** = p < 0.01 vs. C57BL/6. ## = p < 0.01 vs. Muc2^+/+.

Figure 5

Glycine neurotransmission mediates behavioral abnormalities in Muc2 model of colitis. (A) Open field test after strychnine treatment (n = 10–13). Distance: t = 3.53, p = 0.001; rearings: t = 2.897, p = 0.009, Student's t -test. (B) Light–dark test, (n = 7–8). Distance in light compartment: t = 2.21, p = 0.045; time in light compartment: t = 2.08, p = 0.058, Student's t-test. (C) Startle reflex and habituation (n = 10–12; startle: t = − 3.16, p = 0.005; habituation: t = − 3.09, p = 0.006, Student's t-test). (D) Two intruders test (n = 10/group). Two-way ANOVA revealed a statistically significant interaction between the treatment and the intruder gender (number of contacts: F(1, 46) = 21.39, p < 0.001; duration: F(1, 46) = 6.95, p = 0.011. Number of contacts (Muc2^−/− + strychnine, male vs. female, p < 0.001, Fisher’s LSD test; duration (Muc2^−/− + strychnine, male vs. female): p < 0.001, Fisher’s LSD test. Number of contacts with a male/female intruder vs. treatment, p = 0.003, Fisher’s LSD test; duration of contact with a male/female intruder vs. treatment, p = 0.07, Fisher’s LSD test. (E) Two intruders test (n = 10/group). Total duration of social contacts (not significant), duration of aggression: t = 0.75, p = 0.022, Student's t-test. (F) Open field test after L-701,324 treatment (n = 5–11). Distance: t = 3.303, p = 0.005; rearings: t = 2.71, p = 0.017, Student's t -test. (G) Light–dark test, (n = 5–11). Distance in light compartment: t = 2.36, p = 0.033; time in light compartment: t = − 2.56, p = 0.023, Student's t-test. (H) Marble burying test (n = 9/per group, Z = − 1.99, p = 0.047; Mann–Whitney u test). * = p < 0.05, *** = p < 0.001 vs. C57BL/6. ### = p < 0.001, male vs. female.