GABA-producing Bifidobacterium dentium modulates visceral sensitivity in the intestine

Abstract

Background: Recurrent abdominal pain is a common and costly health-care problem attributed, in part, to visceral hypersensitivity. Increasing evidence suggests that gut bacteria contribute to abdominal pain perception by modulating the microbiome-gut-brain axis. However, specific microbial signals remain poorly defined. γ-aminobutyric acid (GABA) is a principal inhibitory neurotransmitter and a key regulator of abdominal and central pain perception from peripheral afferent neurons. Although gut bacteria are reported to produce GABA, it is not known whether the microbial-derived neurotransmitter modulates abdominal pain.

Methods: To investigate the potential analgesic effects of microbial GABA, we performed daily oral administration of a specific Bifidobacterium strain (B. dentiumATCC 27678) in a rat fecal retention model of visceral hypersensitivity, and subsequently evaluated pain responses.

Key results: We demonstrate that commensal Bifidobacterium dentium produces GABA via enzymatic decarboxylation of glutamate by GadB. Daily oral administration of this specific Bifidobacterium (but not a gadB deficient) strain modulated sensory neuron activity in a rat fecal retention model of visceral hypersensitivity.

Conclusions & inferences: The functional significance of microbial-derived GABA was demonstrated by gadB-dependent desensitization of colonic afferents in a murine model of visceral hypersensitivity. Visceral pain modulation represents another potential health benefit attributed to bifidobacteria and other GABA-producing species of the intestinal microbiome. Targeting GABAergic signals along this microbiome-gut-brain axis represents a new approach for the treatment of abdominal pain.

Keywords: Bifidobacterium; GABA; brain gut axis; microbiome; neuromodulation.

Figure 1

Microbial glutamate decarboxylase gene (gadB) in the human microbiome. (A) The relative abundance of the gadB gene among different body sites in 96 healthy adult individuals is depicted as a bar graph. The vertical bars indicate the mean relative abundances (and body site) of gadB (±SEM). The chemical structure of glutamate and GABA and conversion of glutamate to GABA by GadB are shown within the graph. The colored horizontal bars indicate body sites. (B) Bacterial genera/species of the human gut microbiome harboring putative glutamate decarboxylases are depicted as a pie chart. The prevalence of glutamate decarboxylases among the members of the healthy human gut microbiome was estimated from data deposited at the Integrated Microbial Genomes/Human Microbiome Project (IMG/HMP) database (http://img.jgi.doe.gov). Percentages displayed represent genus‐level distribution among these genomes, with species‐level distribution shown for the Bifidobacterium species

Figure 2

GABA production by commensal intestinal strains and Bifidobacterium dentium in the human gut microbiome. (A) Screening of 16 different intestinal commensal and/or probiotic isolates identified B. dentium ATCC 27678 as a major GABA producer. GABA concentrations were measured using LC‐MS after 48 hours of anaerobic growth at 37°C in either regular MRS (dark green) or MRS medium supplemented with 1% w/v glutamate (light green). Error bars represent standard error of 3 independently performed experiments. B. dentium was the only species that produced significantly more GABA (P<.0001; two‐way ANOVA with Bonferroni correction for multiple comparisons). (B) B. dentium in the healthy human gut microbiome, as detected by Metaphlan profiling of shotgun metagenomic sequence libraries. Different colors represent subject cohorts, while bars show proportion of sequences with hits to B. dentium found in each individual. The horizontal bars represent the average relative abundance (as determined by the proportion of sequences with hits to B. dentium) in each cohort

Figure 3

In vitro and in vivo activity of glutamate decarboxylase (GadB) from Bifidobacterium dentium. (A) GadB 3D structure with its proposed active site highlighted. Dark blue color depicts catalytic lysine at position 289 (K289), while threonine (T225) and aspartate (D256) are colored in light blue and red, respectively. Position of co‐factor pyridoxal phosphate, PLP, is shown in green. (B) Site directed mutagenesis effect on recombinant GadB activity. pQE60—negative control of crude extract from Escherichia coli strain harboring empty pQE60 vector. GadB‐WT—crude extract with recombinant wild‐type GadB overexpressed in E. coli. GadB‐T, GadB‐D, and GadB‐K are crude extracts of recombinant GadB with mutated amino acids from T225, D256, and K289 to alanine, respectively. Bars demonstrate GABA (dark green) or l‐glutamate (light green) concentration. Error bars represent standard error of 3 independently performed experiments (*P<.01, **P<.05; one‐way ANOVA of log transformed data, Bonferroni correction for multiple comparisons). (C) Expression of GadB from B. dentium in B. breve by complementation. gadB from B. dentium ATCC 27678 (BD) was cloned into the pESH46 (pESHgadB) expression vector and transformed into B. breve NCIMB8807 (BB) allowing for constitutive expression (BBgadB). GABA was measured via LC‐MS method (Data S1). Error bars represent standard error of 3 independently performed experiments (****P<.0001, ns=not significant; one‐way ANOVA of log transformed data, Bonferroni correction for multiple comparisons). (D) Six‐week‐old male Swiss Webster mice (n=6–8 per group) were orally administered 1% glutamate plus B. dentium ATCC 27678 (BD), B. breve NCIMB8807 (BB), B. breve NCIMB8807 pESHgadB (BBgadB), or saline (PBS) for 5 days. Mice administered B. breve pESHgadB had significantly more GABA in their cecal content as measured by ELISA (**P<.01; one‐way ANOVA with Bonferroni correction)

Figure 4

Neuromodulatory effects of GABA‐producing Bifidobacterium dentium ATCC 27678 administration on colonic sensory neuron activity. Control (Sham) and fecal retention (FR) rats were gavaged daily with GABA‐producing, gadB‐positive B. dentium or gadB‐negative B. breve strains (n=4 rats and 20–24 neurons per treatment group). Colon‐specific DRG neurons were isolated and used for the measurements of cell excitability by patch clamp recordings. The following parameters are displayed: (A) Resting membrane potential (RMP), (B) rheobase, (C) action potential spikes at 2× rheobase, (D) action potential spikes at 3× rheobase, (E) cell diameter, (F) membrane capacitance, (G) input resistance, (H) action potential threshold, (I) action potential amplitude, (J) action potential overshoot, (K) action potential duration, (L) action potential latency. Bars represent mean values with standard error (*P<.05, **P<.01, ***P<.001; Kruskal‐Wallis with Dunn correction for multiple comparison)