Fecal Transplant and Bifidobacterium Treatments Modulate Gut Clostridium Bacteria and Rescue Social Impairment and Hippocampal BDNF Expression in a Rodent Model of Autism

Abstract

Autism is associated with gastrointestinal dysfunction and gut microbiota dysbiosis, including an overall increase in Clostridium. Modulation of the gut microbiota is suggested to improve autistic symptoms. In this study, we explored the implementation of two different interventions that target the microbiota in a rodent model of autism and their effects on social behavior: the levels of different fecal Clostridium spp., and hippocampal transcript levels. Autism was induced in young Sprague Dawley male rats using oral gavage of propionic acid (PPA) for three days, while controls received saline. PPA-treated animals were divided to receive either saline, fecal transplant from healthy donor rats, or Bifidobacterium for 22 days, while controls continued to receive saline. We found that PPA attenuated social interaction in animals, which was rescued by the two interventions. PPA-treated animals had a significantly increased abundance of fecal C. perfringens with a concomitant decrease in Clostridium cluster IV, and exhibited high hippocampal Bdnf expression compared to controls. Fecal microbiota transplantation or Bifidobacterium treatment restored the balance of fecal Clostridium spp. and normalized the level of Bdnf expression. These findings highlight the involvement of the gut-brain axis in the etiology of autism and propose possible interventions in a preclinical model of autism.

Keywords: BDNF; Bifidobacterium; Clostridium cluster IV; Clostridium perfringens; autism spectrum disorder; fecal transplant; hippocampus; microbiota; propionic acid.

Figure 1

Study experimental design. Animals either received propionic acid (PPA) or saline for three days, after which PPA animals were treated with either saline, fecal transplant (FT), or Bifidobacterium (BF) for 22 days. Later, animals were evaluated for sociability using the 3-chamber social test and sacrificed a day later, their brains collected, and the hippocampi dissected out.

Figure 2

Social interaction in the three-chamber social test following saline (n = 6) and propionic acid (PPA) (n = 7) treatments, and fecal transplant (FT) (n = 7) and Bifidobacterium (BF) interventions (n = 7): (A) time spent by focal animals interacting with the conspecific animal and the empty holding box during the three-chamber social test; (B) percentage of time spent in the social chamber relative to other chambers; (C) time spent immobile during the test. Data presented are means ± standard error. Significant two-way and one-way ANOVAs were followed by multiple comparisons by Fisher’s least significance difference ** p ≤ 0.01, * p ≤ 0.05, # p = 0.06.

Figure 3

Levels of fecal Clostridium bacteria following saline (n = 5) and propionic acid (PPA) (n = 3–5) treatments, and fecal transplant (FT) (n = 3–5) and Bifidobacterium (BF) (n = 4–5) interventions at day 11 and at the last day of the intervention (day 22): (A) Clostridium perfringens; (B) Clostridium cluster IV; (C) Clostridium cluster XIVa at 11 days; (D) Clostridium perfringens; (E) Clostridium cluster IV; (F) Clostridium cluster XIVa at 22 days. Data presented are means ± standard error of 2^−ΔΔCt values. Significant one-way ANOVA was followed by multiple comparisons by Fisher’s least significance difference or Kruskal–Wallis followed by uncorrected Dunn’s test. ** p ≤ 0.01, * p ≤ 0.05.

Figure 4

Radar plots of the patterns of fecal Clostridium bacteria levels following propionic acid (PPA) treatment and fecal transplant (FT) and Bifidobacterium (BF) interventions at day 11 and 22 of the intervention: (A) levels of Clostridium perfringens, Clostridium cluster IV, and Clostridium cluster XIVa at 11 days; (B) levels of Clostridium perfringens, Clostridium cluster IV, and Clostridium cluster XIVa at 22 days. Data presented as log-transformed means of 2^−ΔΔCt values.

Figure 5

Fold change in hippocampal transcript levels following saline (n = 5) and propionic acid (PPA) (n = 5) treatments, and fecal transplant (FT) (n = 5) and Bifidobacterium (BF) (n = 5) interventions: (A) brain-derived neurotrophic factor (Bdnf); (B) methyl-CpG binding protein 2 (Mecp2); (C) solute carrier family 17, member 7 (Slc17a7); (D) glutamate decarboxylase 1 (Gad1). Data presented are means ± standard error of 2^−ΔΔCt values. Significant one-way ANOVA was followed by multiple comparisons by Fisher’s least significance difference ** p ≤ 0.01, * p ≤ 0.05, # p = 0.06.