Neurotransmitter Profiles Are Altered in the Gut and Brain of Mice Mono-Associated with Bifidobacterium dentium

Background: Accumulating evidence indicates that the gut microbiota can synthesize neurotransmitters as well as impact host-derived neurotransmitter levels. In the past, it has been challenging to decipher which microbes influence neurotransmitters due to the complexity of the gut microbiota. Methods: To address whether a single microbe, Bifidobacterium dentium, could regulate important neurotransmitters, we examined Bifidobacteria genomes and explored neurotransmitter pathways in secreted cell-free supernatant using LC-MS/MS. To determine if B. dentium could impact neurotransmitters in vivo, we mono-associated germ-free mice with B. dentium ATCC 27678 and examined fecal and brain neurotransmitter concentrations. Results: We found that B. dentium possessed the enzymatic machinery to generate γ-aminobutyric acid (GABA) from glutamate, glutamine, and succinate. Consistent with the genome analysis, we found that B. dentium secreted GABA in a fully defined microbial media and elevated fecal GABA in B. dentium mono-associated mice compared to germ-free controls. We also examined the tyrosine/dopamine pathway and found that B. dentium could synthesize tyrosine, but could not generate L-dopa, dopamine, norepinephrine, or epinephrine. In vivo, we found that B. dentium mono-associated mice had elevated levels of tyrosine in the feces and brain. Conclusions: These data indicate that B. dentium can contribute to in vivo neurotransmitter regulation.


Bifidobacteria Genome Mining
The ability of Bifidobacteria members to generate neuro-active metabolites was estimated using the Integrated Microbial Genomes (IMG) database (http://img.jgi.doe.gov (accessed on 17 January 2021)). The IMG annotates publicly available sequence data and the total number of Bifidobacteria genomes queried are shown in Supplemental Table S1.

LC-MS/MS Equipment
The liquid chromatography-tandem mass spectrometry (LC-MS/MS) system was comprised of a Shimadzu Nexera X2 MP Ultrahigh-Performance Liquid Chromatography (UH-PLC) system (Kyoto, Japan) coupled to a Sciex 6500 QTrap hybrid triple-quadrupole/linear ion trap MS system from Danaher (Washington, DC, USA). Operational control of the LC-MS/MS was performed with Analyst ® (Ver. 1.6.2), and quantitative analysis was performed using MultiQuant™ (Ver. 3.0.1).

LC-MS/MS Methods
Fecal and tissue homogenization and extraction procedures, and targeted LC-MS/MSbased metabolomics methods used for the quantitative analysis of the glutamate cycle and tyrosine pathway metabolites are described in the Supplemental Section. See Tables S2 and S3 in Supplemental Materials for the molecule-specific selected reaction monitoring (SRM) parameters used to acquire the chromatographic data for the targeted metabolites.

Calibration Standard Preparations
Calibration standards (Calibrators) for the glutamate cycle and tyrosine pathway methods were prepared at concentrations of 0.98, 3.9, 15.6, 62.5, and 1000 ng/mL for all metabolites measured in each method. In each instance, a consistent concentration of method-specific deuterated IS compounds were added to each calibrator and unknown sample-the assay-specific concentration levels for each deuterated IS compound is specified in the Supplemental Materials section.

Statistics
GraphPad Software (GraphPad Software, Inc. La Jolla, CA, USA) was used to generate graphs and analyze data for statistical significance. Comparisons between the groups were made with one-way or two-way Analysis of Variance (ANOVA), using the Holm-Sidak post-hoc test comparisons, * p < 0.05.

Bifidobacteria Harbor the Enzymatic Machinery to Generate the Neurotransmitter GABA
Bifidobacteria are known to colonize the mammalian gastrointestinal tract, but the characterization of Bifidobacteria-specific neuro-active compounds remains incomplete. We first focused on the neurotransmitter GABA, which serves as the major inhibitory neurotransmitter of the ENS and CNS [59]. Select microbes have been demonstrated to generate GABA from glutamate or succinate as a mechanism to decrease intracellular pH [60]. In this process, glutamate is taken into the bacteria via the transporter GadC and can be converted to GABA by the enzyme glutamate decarboxylase (GAD; Enzyme Commission number (EC) 4.1.1.15). Glutamate can also be converted to glutamine (EC 6.3.1.2) or glutamine can be converted to glutamate (EC 1.4.1.13). Alternatively, in some bacteria succinate can be converted into GABA via succinate-semialdehyde dehydrogenase (EC: 1.2.1.16) and 4-aminobutyrate-2-oxoglutarate transaminase (EC 2.6.1.19). To address whether B. dentium harbored the genes for the ECs associated with GABA production, we queried the B. dentium Bd1 genome in the KEGG database (https://www.genome.jp/kegg (accessed on 17 January 2021)) and found that this genome contained the complete enzymatic pathway to generate GABA from glutamate/glutamine and succinate ( Figure 1A). Four of the six B. dentium genomes in the Integrated Microbial Genomes (IMG) database (http://img.jgi.doe.gov (accessed on 17 January 2021)), contained all five genes for GABA production from glutamine/glutamate and succinate ( Figure 1B). B. dentium ATCC 27679 and JCVIHMP022 did not contain the gene to express succinate-semialdehyde dehydrogenase (EC 1.2.1.16) needed to convert succinate to GABA, although these species could still produce GABA from glutamate.
To see how common this pathway was among other Bifidobacteria members, we examined the genomes of 83 Bifidobacteria species (total 498 genomes) ( Figure 1C, Supplemental  Figure 1C). Four Bifidobacteria species, B. actinocoloniiforme (2 genomes), B. aemilianum (1 genome), B. catulorum (1 genome), and B. xylocopae (1 genome) did not have the ability to generate GABA from glutamate (EC 4.1.1.15) or succinate (EC 2.6.1. 19), suggesting that GABA production may not be essential for these strains. In this genome comparison, we found that B. dentium (6 genomes) and 10 other Bifidobacteria species contained all genes required to convert glutamine, glutamate, and succinate into GABA. These findings indicate that B. dentium is well equipped to produce GABA.
To address whether B. dentium could generate GABA in vitro, we grew B. dentium ATCC 27678 in a fully defined bacterial media, termed ZMB1, for 18 h and examined GABA, glutamine, and glutamate levels by LC-MS/MS ( Figure 1D). ZMB1 has high concentrations of glutamate (787 µg/mL) and we observed a slight decrease in glutamate concentrations in ZMB1 after the growth of B. dentium. While glutamate concentrations were relatively unchanged, we did observe increased concentrations of GABA and glutamine in B. dentiumconditioned ZMB1, indicating that B. dentium can generate GABA in ZMB1 in vitro.

Bifidobacterium Dentium ATCC 27678 Influences Fecal GABA Levels in Gnotobiotic Animals
After confirming in vitro production of GABA by B. dentium ATCC 27678, we monoassociated adult mice to determine if B. dentium could modulate GABA in vivo. We examined stool and whole brain homogenates in both B. dentium mono-associated mice and germ-free mice. Colonization was confirmed by Gram staining (Figure 2A). As expected, bifid-shaped microbes were found in the mucus layer above the colonic epithelium in the B. dentium mono-associated mice, while no microbes were present in the germ-free controls ( Figure 2A). Next, we assessed the relative concentrations of fecal GABA, glutamate, and glutamine cycle compounds by LC-MS/MS ( Figure 2B,C). We found that B. dentium monoassociation had no effect on glutamine or glutamate concentrations, both of which were high in the rodent diet (Rodent 50 IF/6F auto 5V0F chow cat# 3002875-703). However, we did observe a significant increase in colonic GABA. We also examined GABA, glutamate, and glutamine in the whole brain homogenate ( Figure 2D). Interestingly, no changes were observed in glutamine, glutamate, or GABA concentrations in B. dentium mono-associated brains compared to germ-free controls. This is consistent with the notion that GABA is unable to cross the blood-brain barrier. These data indicate that B. dentium can impact the concentration of GABA in the colon. . Analysis of glutamine, glutamate, and GABA measured from a known mass of stool (ng/mg wet feces weight) by LC-MS/MS reflecting in vivo levels of these neurotransmitters from the germ-free and B. dentium mono-associated mice. (D). Concentrations of glutamine, glutamate, and GABA in a known mass of whole brain homogenate (ng/mg brain). n = 3-5 mice/group; two-way ANOVA, * p < 0.05.

Bifidobacteria Harbor the Enzymatic Machinery to Generate the Neuroactive Compoung Tyrosine
Next, we examined neurotransmitters in the dopamine pathway, including tyrosine, L-dopa, dopamine, norepinephrine, and epinephrine. These neurotransmitters have been shown to act on the enterochromaffin cells and modulate the ENS. We queried the B. dentium Bd1 genome using the KEGG database for the ECs associated with the production of tyrosine, L-dopa, dopamine, norepinephrine, and epinephrine. Analysis of the KEGG pathway revealed that B. dentium could convert chorismate into tyrosine through a three-step process ( Figure 3A). However, no genes were identified which could generate L-dopa, dopamine, norepinephrine, or epinephrine. We next searched within the IMG database for the presence of these genes within several B. dentium genomes and found that all six B. dentium genomes, including the B. dentium ATCC 27678 strain used in our gnotobiotic studies, harbored the genes to convert chorismate into tyrosine ( Figure 3B). To determine if tyrosine pathway genes were conserved in the Bifidobacteria genomes, we examined the genomes of the 83 Bifidobacteria species available in the IMG database. Interestingly, 31 (49%) Bifidobacteria species did not have tyrosine aminotransferase (EC 2.6.1.1), which converts 4-hydroxy-phenylpyruvate into tyrosine, indicating that tyrosine production is species dependent. We did not observe any genes related to L-dopa, dopamine, norepinephrine, or epinephrine production in our genome analysis. Next, we examined tyrosine concentrations in our fully defined ZMB1 media ( Figure 3C). ZMB1 at baseline contains high concentrations of tyrosine (103 µg/mL). Although we observed that B. dentium had the molecular machinery to produce tyrosine, we found that B. dentium did not generate tyrosine in ZMB1 using LC-MS/MS analysis. As expected, we found no production of L-dopa, dopamine, norepinephrine, and epinephrine, consistent with the absence of the required genes in the B. dentium genome.

Bifidobacterium Dentium ATCC 27678 Influences Fecal and Neuronal Tyrosine Levels in Gnotobiotic Animals
To address whether B. dentium could modulate tyrosine concentrations in vivo, we examined the tyrosine pathway in the stool of B. dentium mono-associated mice by LC-MS/MS and compared them with germ-free controls (Figure 4). We observed elevated levels of tyrosine in the feces of B. dentium-treated mice compared to the germ-free controls ( Figure 4B). Tyrosine can be transported through epithelial cells and enter the circulation. Alternatively, tyrosine can enter cells such as enteric neurons and be converted to Ldopa and dopamine, and subsequently to norepinephrine and epinephrine ( Figure 4A). Interestingly, no changes were observed in the other neurotransmitters in the stool samples. Since tyrosine can be transported through the blood-brain barrier, we examined the whole brain homogenate by LC-MS/MS ( Figure 4C). Similar to our stool, we found that tyrosine was elevated in the brain of B. dentium mono-associated mice compared to the germ-free controls; without changes to the other neurotransmitters. Collectively, these data indicate that B. dentium colonization influences the levels of select gut neurotransmitters ( Figure 5). . Analysis of tyrosine, L-dopa, dopamine, norepinephrine, and epinephrine measured from a known mass of stool (ng/mg wet feces weight) by LC-MS/MS reflecting in vivo levels of these neurotransmitters from germ-free and B. dentium mono-associated mice. (C). Concentrations of tyrosine, Ldopa, dopamine, norepinephrine, and epinephrine in a known mass of whole brain homogenate (ng/mg brain). For all experiments, n = 5 mice/group; two-way ANOVA, * p < 0.05.

Figure 5.
Model diagram of B. dentium colonization. Our data indicate that B. dentium can generate glutamine, GABA, and tyrosine. We speculate that GABA has a local effect on the enteroendocrine cells and enteric neurons in the gut, while tyrosine can be transported through the epithelium, enter the circulation, and be transported into the brain.

Discussion
Accumulating evidence indicates that the gut microbiota has a significant impact on the host physiology. One route by which microbes can influence the host is through the production or regulation of neurotransmitters. Herein, we characterize the production of important host-related neurotransmitters by the gut commensal B. dentium. In contrast to other microbes [59], we found that B. dentium was unable to synthesize or modulate dopamine, epinephrine, and norepinephrine. However, B. dentium was able to modulate intestinal concentrations of GABA and tyrosine in vivo. Additionally, we found that B. dentium colonization was associated with an increase in tyrosine in the whole brain homogenate. These findings demonstrate new pathways by which B. dentium can modify the host.
Gnotobiotic models serve as a simple exploratory technique to determine how individual microbes influence the host. Using this approach, we previously found that B. dentium metabolites stimulated the release of serotonin from intestinal enterochromaffin cells [52]. However, the complete profile of intestinal and neuronal neurotransmitters associated with B. dentium colonization was not examined. Herein, we attempted to address this knowledge gap by systemically characterizing stool and whole brain homogenates from germ-free and B. dentium mono-associated mice. We selected the stool as a simplified approach to examine microbial-produced or apically-released neurotransmitters. Because B. dentium was able to produce GABA in vitro in ZMB1 media, we believe that the increased levels of GABA in B. dentium mono-associated stool samples reflect microbially-produced GABA. According to our genome analysis, only 15% of Bifidobacteria harbor all the enzymatic machinery to generate GABA from all sources. We interpret this data to indicate that B. dentium is a more interesting Bifidobacteria candidate for the manipulation of GABA. In vitro, we also observed that B. dentium was able to secrete glutamine, which can enter intestinal epithelial cells via the ASCT2 transporters. Glutamine can be converted to glutamate and subsequently to GABA. Therefore, it is possible that host-derived GABA could contribute to the elevated fecal GABA levels. In the future, it would be interesting to examine how quickly B. dentium produces GABA in vitro and in vivo. This information may inform therapeutic approaches. Our KEGG and IMG pathways analysis revealed that B. dentium could generate glutamate and glutamine, but we did not see any changes in these compounds in our stool samples. Glutamine and glutamate are abundant in the diet, particularly in the diet of germ-free mouse chow, and we speculate that the high availability of these compounds in the intestinal lumen limited the production of these compounds by B. dentium. Additionally, it is possible that B. dentium may be consuming dietary glutamate and glutamine at the same rate as secretion; resulting in no net change in these amino acids. Further studies using stable-isotope labeled substrates would help shed light on the direct contribution of B. dentium to glutamine and glutamate production in vivo. Interestingly, although GABA, glutamate, and glutamine were found in high concentrations in the brain (~1000 ng/mg brain tissue), these levels were unchanged in response to B. dentium colonization. These data suggest that B. dentium colonization in the gut does not influence neuronal GABA concentrations.
Another interesting finding was that B. dentium elevated tyrosine concentrations in the gut and brain. Tyrosine is a non-essential amino acid that can be transported from the intestine into the circulation, cross the blood-brain barrier, and enter neurons, where it is metabolized into catecholamine neurotransmitters. We found that B. dentium harbored the genes to generate tyrosine. We did not observe elevated levels of tyrosine in our ZMB1, which had incredibly high tyrosine levels at baseline. We speculate that the high levels of tyrosine in the ZMB1 limited the production of tyrosine by B. dentium in vitro. Future work optimizing the concentration of tyrosine for growth and secretion would provide an interesting baseline for confirming bacterial tyrosine production. Although we did not observe tyrosine production in vitro, we observed elevated concentrations of tyrosine in the feces of B. dentium-associated mice when compared to the germ-free controls. Since epithelial cells do not appear to have a mechanism to apically secrete tyrosine, we believe that the elevated tyrosine levels in B. dentium-associated mouse stool samples reflect B. dentium synthesis of this amino acid. Although select facultative anaerobes such as Bacillus, Escherichia, and Proteus have been found to produce dopamine and norepinephrine [59], we did not observe detectable levels of dopamine, epinephrine, or norepinephrine in B. dentium-conditioned ZMB1 or in our B. dentium mono-associated mouse feces. These findings are consistent with our KEGG pathway analysis and indicate that B. dentium is unable to synthesize these neurotransmitters.
In vivo, tyrosine is an important regulator of the neuronal catecholamine neurotransmitters, including dopamine, epinephrine, and norepinephrine. Although we saw elevated tyrosine, we did not observe increased concentrations of these catecholamine neurotransmitters in the stool. We speculate that this is because dopamine-producing cells are not commonly found in the intestinal epithelium. Consistent with this notion, we have previously examined a panel of neurotransmitters in human jejunum organoids, also known as enteroids, transduced to increase enteroendocrine cells and we did not detect dopamine, epinephrine, or norepinephrine neurotransmitters [61]. Previous studies have shown that dopamine is largely synthesized in the brain [62]. We found moderate dopamine levels (~1 ng/mg brain tissue) in our germ-free and gnotobiotic mouse brain samples. Epinephrine and norepinephrine are largely produced in sympathetic nerve fibers or adrenal medulla [62]. We also observed moderate concentrations of epinephrine and norepinephrine in the whole brain homogenate. Although we observed increased tyrosine concentration in the B. dentium mono-associated mouse brains, we surprisingly did not see any changes in dopamine, epinephrine, or norepinephrine. This observation could be due to the fact that we examined whole brain neurotransmitter levels. Another possibility is that dopamine concentrations could be elevated in specific brain compartments, but these distinct levels are averaged in a whole brain homogenate. It is also possible that the neuronal cells which generate dopamine require additional metabolite signals generated by other gut microbes to convert tyrosine to dopamine. Finally, since our mice were colonized as adults, it may be that early-life colonization is necessary for the proper wiring of these circuits. Additional studies are necessary to fully tease out this information.
In addition to neurotransmitter production, there are several methods by which the gut microbiota can influence the gut-brain axis, including vagal nerve stimulation, activation of neuropods/direct stimulation of afferent nerves, altering the activity of the stress-associated hypothalamic-pituitary-adrenal (HPA) axis, regulation of the immune cells, and modulating the permeability of the blood-brain barrier [8,24,[63][64][65][66][67][68][69][70][71][72][73][74]. Although we have not examined these pathways in our gnotobiotic mouse model, we believe that these routes may be important contributors to the Bifidobacteria-host modulation and should be explored in the future.
In conclusion, this work has provided a unique glimpse into neurotransmitter profiles using germ-free and mono-associated gnotobiotic animals. Our data is among the first to demonstrate that a single microbe is sufficient to modulate select gut and neuronal neurotransmitters in vivo. Previous work with gut microbes has revealed that many microbial effects are strain-specific. As a result, it is important to delineate which microbes are associated with certain effects. We speculate that mono-associated animals and LC-MS/MS analyses may allow researchers to dissect the essential pathways by which commensal microbes influence the gut-brain axis. A better understanding of these pathways may provide novel adjuvant strategies for both gastroenterological and neurologic disorders.

Acknowledgments:
The Texas Children's Hospital Department of Pathology and Immunology provides salary support to Texas Children's Microbiome Center-Metabolomics Lab staff, and purchased all of the reagents, the consumables and durable supplies, and the LC-MS/MS equipment described.
Conflicts of Interest: J.V. serves on the scientific advisory board of Seed, a USA-based probiotics/prebiotics company, Biomica, an Israeli informatics enterprise and Plexus Worldwide, a USAbased nutrition company. J.K.S. and J.V. receive unrestricted research support from BioGaia AB, a Swedish probiotics company. The remaining authors have no commercial or financial relationships to declare.