The gut microbiome from patients with schizophrenia modulates the glutamate-glutamine-GABA cycle and schizophrenia-relevant behaviors in mice

Abstract

Schizophrenia (SCZ) is a devastating mental disorder with poorly defined underlying molecular mechanisms. The gut microbiome can modulate brain function and behaviors through the microbiota-gut-brain axis. Here, we found that unmedicated and medicated patients with SCZ had a decreased microbiome α-diversity index and marked disturbances of gut microbial composition versus healthy controls (HCs). Several unique bacterial taxa (e.g., Veillonellaceae and Lachnospiraceae) were associated with SCZ severity. A specific microbial panel (Aerococcaceae, Bifidobacteriaceae, Brucellaceae, Pasteurellaceae, and Rikenellaceae) enabled discriminating patients with SCZ from HCs with 0.769 area under the curve. Compared to HCs, germ-free mice receiving SCZ microbiome fecal transplants had lower glutamate and higher glutamine and GABA in the hippocampus and displayed SCZ-relevant behaviors similar to other mouse models of SCZ involving glutamatergic hypofunction. Together, our findings suggest that the SCZ microbiome itself can alter neurochemistry and neurologic function in ways that may be relevant to SCZ pathology.

Fig. 1. Gut microbial characteristics of SCZ.

(A) α-Phylogenetic diversity analysis showed that patients with SCZ were characterized by lower microbial richness (Chao, *P < 0.05) and diversity (Shannon, **P < 0.01) indices relative to HCs. (B) At the OTU level, partial least-squares discriminant analysis (PLS-DA) showed that gut microbiota composition of patients with SCZ was greatly different from that of HCs. (C) Heat map of the 77 discriminative OTU abundances between patients with SCZ and HCs; 23 up-regulated OTUs in SCZ are arranged on the left part of the image, and 54 decreased OTUs are arranged on the right part. The taxonomic assignment of each OTU is provided on the right column. (D) OTUs related to SCZ symptom severity (quantitation with PANSS). The red line designates negative correlation between PANSS and microbes, while the green line shows positive correlation. Lachnospiraceae and Ruminococcaceae are shown twice because they both had two different OTUs correlated with PANSS (see Results). (E) ROC analysis showed that the combination of Aerococcaceae, Bifidobacteriaceae, Brucellaceae, Pasteurellaceae, and Rikenellaceae can distinguish patients with SCZ from HCs with an AUC of 0.769.

Fig. 2. Behavioral comparisons between the SCZ microbiota recipient mice and the HC microbiota recipient mice.

(A and B) Open-field test. Compared to the HC microbiota recipient mice, the total distance (A) and proportion (B) of central distance traveled in 30 min were significantly increased in the SCZ microbiota recipient mice (HC, n = 24; SCZ, n = 25). (C) Forced swimming test. Compared to the HC microbiota recipient mice, the duration of immobility was significantly decreased in the SCZ microbiota recipient mice (n = 20 per group). (D) Y-maze test. There was no difference in the alteration rate between the two groups (n = 20 per group). (E and F) Sociability and social novelty preference test. In the sociability test (E) and social novelty preference tests (F), the time investigating the chamber containing a novel mouse versus the time investigating both chambers was indistinguishable between the two groups (HC, n = 24; SCZ, n = 23). (G and H) Prepulse inhibition (PPI) test. (G) The SCZ microbiota recipient mice displayed an exaggerated startle response to high-decibel tones (120 db) relative to the HC microbiota recipient mice. (H) Increasing prepulse intensity led to increased PPI magnitude in both microbiota recipient groups; however, the PPI magnitude did not differ between the two groups. All data were presented as means ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001 using nonparametric tests).

Fig. 3. Metagenomic analysis identified differential KEGG pathways between SCZ microbiota and HC microbiota recipient mice.

The altered differential KEGG Orthologs (KOs) were mainly involved in 33 disturbed metabolic pathways. Eight of 33 pathways were up-regulated, and the remaining pathways were down-regulated in the SCZ microbiota recipient mice compared to HC microbiota recipient mice (n = 8 per group). tRNA, transfer RNA.

Fig. 4. Altered metabolites in stool, serum, and hippocampus.

(A to C) Key metabolites glutamine (A), glutamate (B), and GABA (C) related to glutamatergic neurotransmission metabolism were significantly changed in the SCZ microbiota mice (n = 10 per group). All data were presented as means ± SEM. *P < 0.05 using Student’s t test. (D) A heat map shows the altered metabolites in stool, serum, and hippocampus. Functional clustering analysis showed that these differentially expressed fecal, serum, and hippocampal metabolites were consistently involved in amino acid and lipid metabolism (n = 10 per group).

Fig. 5. The workflow diagram for this study.

SCZ is associated with dysbiosis of gut microbial composition, which is distinct from that seen in major depressive disorder (MDD). This alteration can result in host gut-brain axis metabolic and neurobehavioral changes relevant to SCZ.