The gut microbiota regulates autism-like behavior by mediating vitamin B6 homeostasis in EphB6-deficient mice

Abstract

Background: Autism spectrum disorder (ASD) is a developmental disorder, and the effective pharmacological treatments for the core autistic symptoms are currently limited. Increasing evidence, particularly that from clinical studies on ASD patients, suggests a functional link between the gut microbiota and the development of ASD. However, the mechanisms linking the gut microbiota with brain dysfunctions (gut-brain axis) in ASD have not yet been full elucidated. Due to its genetic mutations and downregulated expression in patients with ASD, EPHB6, which also plays important roles in gut homeostasis, is generally considered a candidate gene for ASD. Nonetheless, the role and mechanism of EPHB6 in regulating the gut microbiota and the development of ASD are unclear.

Results: Here, we found that the deletion of EphB6 induced autism-like behavior and disturbed the gut microbiota in mice. More importantly, transplantation of the fecal microbiota from EphB6-deficient mice resulted in autism-like behavior in antibiotic-treated C57BL/6J mice, and transplantation of the fecal microbiota from wild-type mice ameliorated the autism-like behavior in EphB6-deficient mice. At the metabolic level, the disturbed gut microbiota in EphB6-deficient mice led to vitamin B₆ and dopamine defects. At the cellular level, the excitation/inhibition (E/I) balance in the medial prefrontal cortex was regulated by gut microbiota-mediated vitamin B₆ in EphB6-deficient mice.

Conclusions: Our study uncovers a key role for the gut microbiota in the regulation of autism-like social behavior by vitamin B₆, dopamine, and the E/I balance in EphB6-deficient mice, and these findings suggest new strategies for understanding and treating ASD. Video abstract.

Keywords: ASD; Dopamine; E/I balance; EphB6; Gut microbiota; Vitamin B6.

Fig. 1

The deletion of EphB6 led to autism-like behavior and gut microbial disturbance in mice. a The 8-week-old male KO mice spent more time self-grooming than WT mice. n = 17 mice for each group. b In social partition test, KO mice spent less time sniffing the partition than WT mice. n = 10, 11 mice respectively. c–h In three-chambered social approach task, time spent in chambers during different 10-min trials (c–e), trajectory diagram during the second 10-min trial (f) were showed. KO mice showed less preference for the social mouse over the object (g) and less preference for the novel social mouse over the familiar social mouse (h) than WT mice. n = 10, 9 mice respectively. i In olfactory habituation/dishabituation test, KO mice spent less time sniffing social odors than WT mice. n = 11 mice for each group. j In elevated plus maze test, KO mice spent less time in open arm and more time in closed arm than WT mice. n = 12, 13 mice respectively. k The intestinal permeability of 8-week-old WT and KO mice was detected using FITC-dextran. n = 4 mice for each group. l The mRNA expressions of tight junction molecules were detected in colon of 8-week-old WT and KO mice. n = 7, 6 mice respectively. m The mRNA expressions of cytokines were detected in colon of 8-week-old WT and KO mice. n = 4, 5 mice respectively. n–r 16S rRNA gene sequencing of gut microbiota of 8-week-old WT and KO mice. The species richness (n) and diversity (o) of gut microbiota were similar, while the microbial composition (p) was different between the two groups. Relative abundance of different bacteria in phylum level was showed (q). At genus level, the relative abundance of Mucispirillumn was decreased in KO mice (r). n = 8 mice for each group. Data shown are mean ± SEM or median ± IQR. Two-tailed unpaired student’s t test (a, c–e, g–h, j–m), Mann-Whitney test (n–o, q, r), mixed design ANOVA with genotype as independent factor and stimuli/trials as repeated-measure factor (b, i), anosim analysis (p). *p < 0.05; **p < 0.01; ***p < 0.001. WT EphB6^+/+ mice, KO EphB6^−/− mice, FITC fluorescein isothiocyanate. Statistical values are presented in Additional file 3: Table S2

Fig. 2

Transplantation of the fecal microbiota from EphB6-deficient mice caused autism-like behavior in 3-week-old SPF C57BL/6J mice. a–i Schematic of the fecal microbiota transplantation (a). The 3-week-old SPF male C57BL/6J mice were orally gavaged with the fecal microbiota from 8-week-old male WT or KO mice (each contained eight healthy mice from at least three cages) for 1 week. After 3 weeks, the fecal microbiota of the treated C57BL/6J mice were sequenced (b–d, eight treated C57BL/6J mice of each group were selected randomly from at least three cages) and self-grooming test (e), olfactory habituation/dishabituation test (f), three-chambered social approach task (g–i), open field test, and elevated plus maze test were conducted with an interval of at least 2 days (e–i, n = 13, 19 mice respectively). j–r Schematic of the fecal microbiota transplantation (j). The 3-week-old SPF male C57BL/6J mice were orally gavaged with antibiotics (ampicillin, vancomycin, neomycin, metronidazole) for 5 days and then orally gavaged with the fecal microbiota from 8-week-old male WT or KO mice (each contained eight healthy mice from at least three cages) for another 5 days. After 19 days, the fecal microbiota of the treated C57BL/6J mice were sequenced (k–m, six treated C57BL/6J mice of each group were selected randomly from at least two cages) and self-grooming test (n), olfactory habituation/dishabituation test (o), three-chambered social approach task (p–r), open field test, and elevated plus maze test were conducted with an interval of at least 2 days (n–r, n = 10, 16 mice respectively). Data shown are mean ± SEM or median ± IQR. Two-tailed unpaired student’s t test (e, h, i, n, q, r), Mann-Whitney test (b, c, k, l), mixed design ANOVA with genotype as independent factor and stimuli/trials as repeated-measure factor (f, o), anosim analysis (d, m). *p < 0.05. WT col or KO col colonized with the fecal microbiota from EphB6^+/+ or EphB6^−/− mice, Abx pretreated with antibiotics (ampicillin, vancomycin, neomycin, metronidazole). Statistical values are presented in Additional file 3: Table S2

Fig. 3

Fecal microbiota transplantation from EphB6-deficient mice partially induced social deficits in 6-week-old SPF C57BL/6J mice. a–i Schematic of the fecal microbiota transplantation (a). The 6-week-old SPF male C57BL/6J mice were orally gavaged with fecal microbiota from 8-week-old male WT or KO mice for 1 week. After 1 week, the fecal microbiota of the treated C57BL/6J mice were sequenced (b–d, n = 6 mice for each group) and self-grooming test (e), olfactory habituation/dishabituation test (f), three-chambered social approach test (g–i) were conducted with an interval of at least 2 days (e–i, n = 12, 16 mice respectively). j–m Fecal metabolites from 8-week-old male WT and KO mice were orally gavaged to 6-week-old SPF male C57BL/6J mice for 1 week (j). After 1 week, olfactory habituation/dishabituation test (k) and three-chambered social approach task (l, m) were conducted with an interval of at least 2 days. n = 9, 10 mice respectively. Data shown are mean ± SEM or median ± IQR. Two-tailed unpaired student’s t test (e, h, i, m), Mann-Whitney test (b, c), mixed design ANOVA with genotype as independent factor and stimuli/trials as repeated-measure factor (f, k), anosim analysis (d). *p < 0.05; ***p < 0.001. WT col or KO col, colonized with fecal microbiota or fecal metabolites from EphB6^+/+ mice or EphB6^−/− mice. Statistical values are presented in Additional file 3: Table S2

Fig. 4

Transplantation of the fecal microbiota from wild-type mice ameliorated autism-like behavior in adult EphB6-deficient mice. a Schematic of the the fecal microbiota transplantation. The 8-week-old male WT and KO mice were orally gavaged with the fecal microbiota from 8-week-old male WT mice (eight healthy mice from at least three cages) or sterile PBS for 1 week. After 1 week, the fecal microbiota of the treated WT and KO mice were sequenced (b–d) and behavioral tests were conducted with an interval of at least 2 days (e–i). b–d 16S rRNA gene sequencing of the fecal microbiota from mice. Principal coordinates analysis of Bray-Curtis distance (b), the relative abundance of Deferribacteres (c) at phylum level, and the relative abundance of Mucispirillum (d) at genus level were showed. At phylum level, the range of 0–0.8 on x axis was used for the relative abundance of p_Bacteroidetes, p_Firmicutes, and p_Proteobacteria and the range of 0–0.05 on x axis was used for other bacteria. n = 6, 6, 5, 5 mice respectively. e–i Self-grooming test (e), olfactory habituation/dishabituation test (f), and three-chambered social approach task (g–i) were performed. n = 15, 15, 18, 20 mice respectively. Data shown are mean ± SEM or median ± IQR. One-way ANOVA (e, h, i), Kruskal-Wallis test (c, d), mixed design ANOVA with genotype as independent factor and stimuli/trials as repeated-measure factor (f), anosim analysis (b). *p < 0.05; **p < 0.01; ***p < 0.001. WT EphB6^+/+ mice, KO EphB6^−/− mice, FMT fecal microbiota transplantation, PBS phosphate-buffered saline. Statistical values are presented in Additional file 3: Table S2

Fig. 5

Gut microbiota regulated vitamin B₆ in EphB6-deficient mice. a–d In non-targeted metabolomics analysis, the metabolites in PFC of 8-week-old male WT and KO mice were differently clustered by orthogonal partial least squares discriminant analysis (a). The enriched KEGG pathways associated with differential metabolites (b), the relative abundance of pyridoxamine (PM, c), and pyridoxal 5′-phosphate (PLP, d) were showed. n = 7, 6 mice respectively. e–k The fecal microbiota from 8-week-old WT mice or PBS were gavaged to 8-week-old WT or KO mice for 1 week (e). One week later, the level of PM (f), PLP (g), and pyridoxine (PN, h) in feces of mice were detected. n = 3, 4, 4 mice respectively. The level of PM and PLP in plasma (i, j, n = 5, 5, 6 mice respectively) and level of PLP in PFC (k, n = 4, 5, 6 mice respectively) of mice were also detected. Data shown are mean ± SEM. R (b–d), one-way ANOVA (f–k). *p < 0.05; **p < 0.01; ***p < 0.001. WT, EphB6^+/+ mice; KO, EphB6^−/− mice; FMT fecal microbiota transplantation, PBS phosphate-buffered saline, PFC prefrontal cortex, PM pyridoxamine, PLP pyridoxal 5′-phosphate, PN pyridoxine. Statistical values are presented in Additional file 3: Table S2

Fig. 6

Gut microbiota-mediated vitamin B₆ homeostasis regulated social behavior in EphB6-deficient mice. a–g Schematic of the injection of PLP (a). The 8-week-old male KO mice were injected with 1 mg PLP or saline intraperitoneally. After 1 h, mice were either sacrificed to detect PLP in plasma (b, n = 4 mice for each group) or subjected to self-grooming test (c, n = 10 mice for each group), olfactory habituation/dishabituation test (d, n = 8, 9 mice respectively), and three-chambered social approach task (e–g, n = 14 mice for each group). Different mice were used for different behavioral tests. h–j Schematic of the injection of PLP (h). Saline or PLP (1 mg, 2 mg, 5 mg per 0.2 mL) were injected intraperitoneally to 8-week-old male C57BL/6J mice. After 1 h, three-chambered social approach task was conducted (i, j, n = 10, 6, 10, 10 mice respectively). k–n Schematic of deficiency of vitamin B₆ in 6-week-old SPF male C57BL/6J mice (k). Normal diet contained 12 mg vitamin B₆ and vitamin B₆-deficient diet were provided for 6-week-old C57BL/6J mice for 2 weeks. The level of PLP in plasma of C57BL/6J was detected (l, n = 4 mice for each group) and three-chambered social approach task (m, n, n = 8, 8 mice respectively) were conducted. Data shown are mean ± SEM. Two-tailed unpaired student’s t test (b, c, f, g, l, n), one-way ANOVA (j), mixed design ANOVA with genotype as independent factor and stimuli/trials as repeated-measure factor (d). *p < 0.05; **p < 0.01; ***p < 0.001. WT EphB6^+/+ mice, KO EphB6^−/− mice, PLP pyridoxal 5′-phosphate, VB6 vitamin B₆. Statistical values are presented in Additional file 3: Table S2

Fig. 7

The modulated dopamine by gut microbiota-mediated vitamin B₆ regulated social behavior of EphB6-deficient mice. a–c The 8-week-old male WT and KO mice were orally gavaged with the fecal microbiota of 8-week-old male WT mice or sterile PBS for 1 week (a). After 1 week, the level of amino acid neurotransmitters (b, n = 7, 7, 5 mice respectively) and monoamine neurotransmitters (c, n = 5, 5, 7 mice respectively) in PFC of mice were detected. d One hour after the injection of 1 mg PLP intraperitoneally, the level of DA in PFC of 8-week-old male KO mice was increased compared with that of KO mice injected with saline. n = 5 mice for each group. e The level of DA in PFC of SPF male C57BL/6J mice fed without vitamin B₆ was decreased compared with that of C57BL/6J mice fed with normal diet. n = 3, 4 mice respectively. f The mRNA expression of dopamine receptors and Th in mPFC or VTA of 8-week-old WT and KO mice were detected by qRT-PCR. n = 4, 3 mice respectively. g–j Thirty minutes after the injection of D1R agonist (SKF38393, 0.0625 μg/0.3 μL) in mPFC of 8-week-old male KO mice, olfactory habituation/dishabituation test (g) or three-chambered social approach task (h–j) was conducted with an interval of 1 week. n = 9, 11 mice respectively. k–m The 8-week-old SPF male C57BL/6J mice that injected with D1R agonist (SKF38393, 0.0625 μg/0.3 μL) in mPFC were performed with olfactory habituation/dishabituation test (k) or three-chambered social approach task (l, m) with an interval of 1 week. n = 7, 8 mice respectively. n–q The 8-week-old SPF male C57BL/6J mice were injected with D1R antagonist (SCH23390) in mPFC, and olfactory habituation/dishabituation test (n, n = 8, 8, 9, 8 mice respectively) or three-chambered social approach task (o–q, n = 9, 5, 9, 5 mice respectively) were conducted with an interval of 1 week. Data shown are mean ± SEM. Two-tailed unpaired student’s t test (d–f, i, j, m), one-way ANOVA (b, c, p, q), mixed design ANOVA with genotype as independent factor and stimuli/trials as repeated-measure factor (g, k, n). *p < 0.05; **p < 0.01. WT EphB6^+/+ mice, KO EphB6^−/− mice, FMT fecal microbiota transplantation, PBS phosphate-buffered saline, PFC prefrontal cortex, PLP pyridoxal 5′-phosphate, VB6 vitamin B₆, Glu glutamic acid, GABA gamma-aminobutyric acid, Gly glycine, Asp aspartic acid, Ser serine, Gln glutamine, NE norepinephrine, EP epinephrine, DA dopamine, 5-HT 5-hydroxytryptamine, DOPAC dihydroxy-phenyl acetic acid, mPFC middle prefrontal cortex, VTA ventral tegmental area, Th tyrosine hydroxylase, ACSF artificial cerebrospinal fluid. Statistical values are presented in Additional file 3: Table S2

Fig. 8

Gut microbiota and dopamine modulated E/I balance in mPFC of EphB6-deficient mice. a The 8-week-old male WT and KO mice were orally gavaged with the fecal microbiota from 8-week-old male WT mice or sterile PBS for 1 week. After 1 week, we recorded sEPSCs (b–f) and sIPSCs (g–k) of pyramidal neurons in mPFC of mice. b–f Representative sEPSCs traces from pyramidal neurons in mPFC of mice (b) were presented, scale bars: 3 s, 20 pA. Cumulative distribution of sEPSCs amplitudes (c), average amplitude of sEPSCs (d), cumulative distribution of sEPSCs frequencies (e), and average frequency of sEPSCs (f) were showed. n = 22, 21, 15, 18 cells from at least four mice respectively. g–k Representative sIPSCs traces from pyramidal neurons in mPFC of mice (g) were presented, scale bars: 3 s, 20 pA. Cumulative distribution of sIPSCs amplitudes (h), average amplitude of sIPSCs (i), cumulative distribution of sIPSCs frequencies (j), and average frequency of sIPSCs (k) were showed. n = 18, 14, 25, 20 cells from at least four mice respectively. l–o The mPFC slices of 8-week-old male KO mice were treated with 10 μM D1R agonist, then sEPSCs and sIPSCs of pyramidal neurons were recorded. Average amplitude (l) and frequency (m) of sEPSCs and average amplitude (n) and frequency (o) of sIPSCs were showed. n = 15, 15 or 14, 14 cells from at least four mice respectively. Data shown are mean ± SEM or median ± IQR. Two-tailed unpaired student’s t test (l, m, o), Mann Whitney test (n), Kruskal-Wallis test (d, f, i, k). *p < 0.05; **p < 0.01. WT EphB6^+/+ mice, KO EphB6^−/− mice, FMT fecal microbiota transplantation, PBS phosphate-buffered saline, mPFC middle prefrontal cortex, sEPSCs spontaneous excitatory postsynaptic currents, sIPSCs spontaneous inhibitory postsynaptic currents. Statistical values are presented in Additional file 3: Table S2

Fig. 9

Working model of the modulated social deficits in EphB6-deficient mice by gut microbiota. a Deletion of EphB6 induced gut microbial dysbiosis and decreased vitamin B6 in plasma and PFC, which led to the decreased dopamine, E/I imbalance and social deficits in mice.