Gut microbiome in neuroendocrine and neuroimmune interactions: The case of genistein

Abstract

The healthy and diverse microbes living in our gut provide numerous benefits to our health. It is increasingly recognized that the gut microbiome affects the host’s neurobehavioral state through production of metabolites, modulation of intestinal immunity (e.g., cytokines) and other mechanisms (e.g., gut neuropeptides). By sending the sensed information (e.g., metabolic and immunologic mediators) about the state of the inner organs to the brain via afferent fibers, the vagus nerve maintains one of the connections between the brain and GI tract, and oversees many critical bodily functions (e.g., mood, immune response, digestion and heart rate). The microbiome-gut-brain axis is a bidirectional communication between the gut, its microbiome, and the nervous system. In the present review, the roles of microbiome in neuroendocrine and neuroimmune interactions have been discussed using naturally occurring isoflavones, particularly the phytoestrogen genistein, as there are sex differences in the interactions among the microbiome, hormones, immunity and disease susceptibility. A deep understanding of the mechanisms underlying the interactions among the endocrine modulators, brain, endocrine glands, gut immune cells, vagus nerve, enteric nervous system and gut microbiome will provide important knowledges that may ultimately lead to treatment and prevention of debilitating disorders characterized by deficits of microbiome-neuroendocrine-neuroimmune relationships.

1. Introduction

It is well known that the gastrointestinal (GI) tract and central nervous system (CNS) interact, and mechanisms underlying the bidirectional gut–brain interactions have gradually been revealed. The microbiome-gut-brain axis is a dynamic interaction among various tissues and organs, including the brain, endocrine glands, gut immune cells, vagus nerve, enteric nervous system and gut microbiome (GMB) that communicate in a multidirectional manner to maintain organism homeostasis. This interaction is now recognized as a regulator of mood, fear, cognition, pain, sleep and behaviors (Chu et al., 2019). The GMB is a dynamic ecosystem formed by thousands of distinct bacterial species in the gut. The first evidence associating GMB disturbances (dysbiosis) and neurobehavioral disorders originated from germ-free mice. These mice exhibited abnormalities in the GI tract as well as the hypothalamic-pituitary-adrenal gland axis by showing more anxieties and fear-associated behaviors, and less exploratory, cognitive and social behaviors (Chu et al., 2019). These deficits could be reversed with bacterial reconstitution or fecal transplantation, suggesting a critical role for GMB in postnatal development of the enteric nervous system. In addition, other experimental paradigms including treatment with antibiotics or pre-/probiotics have demonstrated that GMB influence many facets of CNS physiology e.g., neurotransmitter signaling, synaptic plasticity, myelination and neurogenesis.

Gut houses 70% of the body’s immune system along with 80% of plasma cells (Vighi, Marcucci, Sensi, Di Cara, & Frati, 2008), and intestinal inflammation and imbalance of GMB (dysbiosis) and associated metabolic activities are linked to many diseases, including neurological symptoms (e.g., depression), diabetes and obesity (metabolic), malnutrition and inflammatory bowel disease (IBD; immune) (Day, 2018; Lazar et al., 2019; Valles-Colomer et al., 2019). The objective of this manuscript was to review the roles of GMB in neuroendocrine and neuroimmune interactions by focusing on naturally occurring isoflavones, particularly the phytoestrogen genistein. There are sex differences in the interactions among the GMB, hormones, immunity and disease susceptibility. In addition, there are pathophysiological differences between the microbiome of humans and animals. For example, all rodent guts contain equol-producing bacteria, while only 30–50% of humans harbor such bacteria (Cross et al., 2017). To this end, we have discussed human and lab animal microbiome separately for the phytoestrogen genistein.

2. Possible Pathways in the Microbiome-Gut-Brain Axis

It has been realized a long time ago that intestinal dysfunctions, such as IBD, and psychiatric disease (e.g., depression) might share the same pathogenesis and biological mechanisms, including alterations in the hypothalamic-pituitary-adrenal gland axis mediated by corticotropin releasing factor in response to stress, cytokine secretion and immunological modulations. The microbiome-gut-brain axis is a bidirectional communication between the gut, its microbiome, and the nervous system. The efferent brain-gut signaling includes neuroendocrine, neuroimmune and autonomic regulations (Mayer, Tillisch, & Gupta, 2015). The afferent gut–brain signaling involves the enteroendocrine system, cytokines, sensory epithelial cells, and GMB (Figure 1). Both the vagus nerve and palatine nerve can relay cytokine-induced signals to the brain, and neurotransmitters, such as serotonin, play a key role in the activation of immune cells to produce proinflammatory cytokines (Banks, 2008; Herr, Bode, & Duerschmied, 2017). In addition, cytokines alter the concentrations of several neurotransmitters that regulate the communications in brain, including serotonin, dopamine, and glutamate. Cytokines, together with neurotransmitters and hormones, are critical in the maintenance of neuro-immune-endocrine system homeostasis. Crosstalk among the intestinal epithelium, intestinal immune system and GMB can modulate systemic immunity and affect the interaction between GMB and CNS-restricted immune cells, the microglia (Chu et al., 2019).

Figure 1.

Illustrated are three major ways for gut microbiome to send signals to the brain. (1) Gut microbes may signal the brain through the vagus nerve using neurotransmitters. (2) Gut microbes stimulate gut immune cells to secrete cytokines for brain signaling. (3) The gut-brain communication occurs through metabolites produced by gut microbes. SCFAs = short chain fatty acids, LPS = lipopolysaccharides, ENS = enteric nervous system.

The CNS can alter GMB composition and behavior via the autonomic nervous system, and neuroendocrine and neuroimmune pathways (Wasilewska & Klukowski, 2015). For example, reducing hypothalamic inflammation improves leptin sensitivity (Milanski et al., 2012), which is under GMB control (Cani & Knauf, 2016). The present review focuses more on the enteric-afferent pathways than on the complicated neuro-efferent events. There exist at least three major ways for GMB to send signals to the brain (Figure 1). Firstly, GMB may signal the brain through the vagus nerve, which connects networks of nerves in the gut to various brain regions (e.g., hypothalamus) using neurotransmitters. It is also possible for GMB to stimulate gut immune cells to secrete cytokines that travel to the brain via the bloodstream. The third possible way for gut-brain communication to occur is through metabolites produced by GMB, e.g., short chain fatty acids (SCFAs), which may stimulate enteric neurons and enteroendocrine cells to produce gut neuropeptides. Metabolites such as SCFAs can also travel to the brain through the bloodstream and directly modulate microglia density, morphology and maturity (Erny et al., 2015). However, these three pathways are not mutually exclusive. There exist interactions and overlaps among them that allow for these processes to amplify each other. For example, the vagus nerve has immunomodulatory properties (Breit, Kupferberg, Rogler, & Hasler, 2018), and SCFAs can stimulate free fatty acid receptors in enterochromaffin cells to trigger serotonin biosynthesis (Reigstad et al., 2015).

2.1. Vagus nerve and neurotransmitters in the microbiome-gut-brain axis

The vagus nerve bridges the direct communication between GMB and the brain (Figure 1). In the gut, the sensed information, such as metabolic mediators from GMB and immunologic mediators, is integrated at the vagal nuclei and then transmitted to different brain regions to alter behavioral responses. Furthermore, GMB can alter the neurochemical levels in the vagus nerve. GMB produces a range of neurotransmitters through the metabolism of indigestible fibers (i.e., cellulose, hemicellulose, lignin, pectin and beta-glucans) (Lattimer & Haub, 2010). These include dopamine and noradrenaline by members of the Bacillus family, GABA (γ-aminobutyric acid) by the Bifidobacteria family, serotonin (5-hydroxytryptamine) by the Enterococcus and Streptococcus families, noradrenaline and serotonin by the Escherichia family, and GABA and acetylcholine by the Lactobacilli family (Sarkar et al., 2016). These neurotransmitters stimulate the vagus nerve, and it may in turn alter the activities in hypothalamus and other brain regions. It is also possible that some of these neurotransmitters reach the brain via blood and circumventricular organs. However, more studies are required to determine what are the physiological implications of GMB-mediated alterations of these neurotransmitters, although it has been proposed that some of these neurotransmitters may reach the brain through the vagus nerve (Bonaz, Bazin, & Pellissier, 2018; Klarer et al., 2014).

The expressions of neurotransmitters, such as GABA and serotonin, can be regulated by GMB (Martin et al., 2019; Strandwitz et al., 2019). GABAergic transmission plays a key role in controlling emotional state and participates in the regulation of various psychophysiological phenomena. There are GABA-producing bacteria found in the stool samples from healthy people, e.g., Bacteroides, Parabacteroides and Escherichia species (Strandwitz et al., 2019), and more have been identified from various dietary sources. In patients with depressive disorders, the relative abundance of fecal GABA-producing Bacteroides is decreased and negatively correlates with the depressive signatures in the brain (Strandwitz et al., 2019). In adult male BALB/c mice, administration of the potential psychobiotic Lactobacillus rhamnosus (JB-1) over 28 days lowered the level of stress-induced corticosterone, and decreased the anxiety and depression-like behaviors in the forced swim test (Bravo et al., 2011). At the same time, JB-1-treated mice exhibited region-dependent alterations in the expression of GABA_B1b and GABA_Aα2 mRNAs, which are related to the modulations of memory and anti-depression, respectively (Bravo et al., 2011). Further studies in vagotomized mice did not show either behavioral or neurochemical changes in the same tests, suggesting an indispensable role of the vagus nerve in the communication between JB-1 and brain via regulating inhibitory neurotransmitter GABA (Bravo et al., 2011).

Serotonin has been used in the form of drugs and nutraceuticals for vagus nerve stimulation and for sleep and feelings of well-being. However, about 95% of the body’s serotonin locates in the gut but not in the brain (Fung et al., 2019). Germ-free mice display lower serotonin levels in cecum and colon, and lower percentage of unconjugated serotonin (bioactive form) than the germ-free mice recolonized with specific pathogen-free fecal flora (Hata et al., 2017). One possible explanation is that some bacterial species, such as lactic acid bacteria (e.g., Streptoccocus thermophilus) and E. coli, produce serotonin. In addition, indigenous spore-forming bacteria from GMB promote the serotonin biosynthesis in enterochromaffin cells through secreting metabolites, e.g., α-tocopherol, butyrate, cholate, deoxycholate, p-aminobenzoate, propionate and tyramine (Yano et al., 2015). In spite of a plethora of information showing that serotonin is vital for emotional and basic physiological functions, and that GMB can regulate the serotonin levels, comprehensive evidence is missing to directly link GMB-regulated serotonin level to emotion and behavior. It is also unclear whether the vagus nerve is mediating this communication.

2.2. Cytokines and gut neuropeptides in the microbiome-gut-brain axis

Although the interactions between GMB and intestinal immune system (Figure 2) have been well studied (Guo et al., 2018), the interplays among cytokine production, neuroendocrine regulation and GMB have not been investigated extensively. In a study to determine the effect of probiotics on chronic stress induced by maternal separation during perinatal stages in C57BL/6 mice, the introduction of the potential probiotic Bifidobacterium pseudocatenulatum CECT 7765 downregulated maternal separation-induced intestinal inflammation by reducing IFN-γ and intestinal hypercatecholaminergic activity, e.g., dopamine and adrenaline, at postnatal day 21 (Moya-Perez, Perez-Villalba, Benitez-Paez, Campillo, & Sanz, 2017). In a rat study, modulation of GMB by a multi-species probiotic treatment significantly reduced depressive-like behavior by 34% in the forced swim test, and altered the cytokine production by the stimulated blood mononuclear cells towards IFN-γ, IL-2 and IL-4 at the expense of TNF-α and IL-6 (Abildgaard, Elfving, Hokland, Wegener, & Lund, 2017). Interestingly, the probiotic use lowered the transcript levels of factors involved in the regulation of hypothalamic-pituitary-adrenal gland axis, including corticotropin releasing hormone receptor-1, −2 and mineralocorticoid in hippocampus (Abildgaard et al., 2017). In a mechanistic study, it was demonstrated that depression was associated with a decreased GMB richness (Kelly et al., 2016): Fecal microbiota transplantation from depressed patients to microbiota-depleted rats induced behavioral and physiological features related to depression in the recipient animals, including anhedonia and anxiety-like behaviors, as well as alterations in tryptophan metabolism. In addition, the depressed rats showed an elevated IL-8 and TNF-α (Kelly et al., 2016).

Figure 2.

Interactions and overlaps among the three pathways described in Figure 1. AM = adrenomedullin, α-MSH = α-melanocyte stimulating hormone, CCK = cholecystokinin, CGRP = calcitonin gene-related peptide, GLP = glucagon-like peptide, ILC3 = type 3 innate lymphoid cells, NPY = neuropeptide Y, PYY = peptide YY, SP = substance P, Treg = regulatory T cells, VIP = vasoactive intestinal peptide, Mφ = macrophage, DCs = dendritic cells, IBD = inflammatory bowel disease.

In addition to immune mediators, gut neuropeptides originated from enteric neurons and enteroendocrine cells can serve as a mediator between GMB and host (Figure 2). Common gut neuropeptides include substance P, neuropeptide Y, α-melanocyte stimulating hormone, vasoactive intestinal peptide, calcitonin gene-related peptide and adrenomedullin, and they are likely to play important roles in the bidirectional gut-brain communication. The function of gut neuropeptide-releasing enteroendocrine cells is directly influenced by metabolites (e.g., SCFAs) generated by GMB from indigestible fiber, and gut neuropeptides may control the impact of GMB on inflammatory processes, pain, brain function and behavior. The effects of gut neuropeptides on GMB can be direct or indirect when the stimuli are sensed. Gut neuropeptides can cross the epithelial barrier and exert antimicrobial activity in the gut lumen by different mechanisms (direct) or induce immune responses (innate or adaptive), and subsequently result in microbial imbalance (indirect) (Aresti Sanz & El Aidy, 2019).

In the CNS, the microglia are the innate sentinel immune cells that can detect subtle changes in molecules in their locality. The proper functioning of microglia in brain regions (e.g., the hypothalamus) is critical for maintaining brain health and regulating metabolism (Figure 2). When activated, they perform functions such as removing damaged cells at a site of injury. A critical role for GMB in microglia maturation, morphology and immunological function has been shown (Erny et al., 2015), and a healthy and diverse GMB is essential for the continuous preservation of healthy microglia and proper brain function throughout host lifespans. Furthermore, it has also become clear that microglia have a crucial role in synaptic connectivity. By engulfing and degrading unwanted synapses, microglia can ensure that neuronal connections are pruned or maintained as needed, which have been shown to be critical for fear extinction (Chu et al., 2019). In addition, there exist a neuroimmune circuit involving microglia activation and an altered sympathetic neural tone to the peripheral immune system to recruit inflammatory monocytes to the brain (Wohleb, Mckim, Sheridan, & Godbout, 2015). Taken together, GMB closely interact with the body’s major neuroendocrine and neuroimmune systems that control various physiological processes in response to stress, metabolic dysfunction and infections (Figure 2).

2.3. Bacterial-derived metabolites in the microbiome-gut-brain axis

The microbial metabolism is seen as a complement to the host metabolism. Dietary metabolites derived from GMB play a critical role in the regulation of multiple neural behaviors (e.g., anxiety, depression) through the microbiome-gut-brain axis, and gut dysbiosis favoring pro-inflammatory microbial communities precedes depression development (Macedo et al., 2017). GMB can metabolize dietary compounds into metabolites (e.g., phenolic acids) with important biological activities. These small, gut-derived metabolites may be responsible for the health benefits of diets high in fruits and vegetables. Nuclear magnetic resonance and liquid chromatography-mass spectrometry-based metabolomic studies have shown that microbial metabolites are often the compounds most markedly altered in the disease state when compared to healthy individuals (L. S. Zhang & Davies, 2016). Importantly, many studies suggest that these metabolites may be effective anxiolytic, antidepressant, and/or anti-inflammatory agents. Furthermore, these metabolites may exert their biological effects using various pathways simultaneously rather than acting through a single mechanism (e.g., biological signature; Figure 2). Although GMB is highly variable, the summation of genomes composing it tends to be quite conserved when considering the microbial metabolic pathways.

SCFAs such as acetate (C2), propionate (C3), butyrate (C4) and pentanoate (valerate, C5) are mainly produced by bacterial fermentation of dietary fiber or glycosylated host proteins such as mucins in the colon. Bacteroidetes (gram-negative) and Firmicutes (gram-positive) are the most abundant phyla in the intestine, with members of the Bacteroidetes mainly producing acetate and propionate, while Firmicutes mostly produce butyrate in the human gut (Parada Venegas et al., 2019). SCFAs are not only able to protect host from mucosal inflammation and colorectal tumorigenesis, but may also act in a systemic manner to ameliorate T cell-driven autoimmunity in the brain (Luu et al., 2019). Systemic sodium butyrate injections in rats produce antidepressant effects, and increase central serotonin neurotransmission and brain-derived neurotrophic factor expression (Sun et al., 2016). Dysbiosis in patients with multiple sclerosis, an autoimmune disease affecting the CNS, is characterized by a reduction of species belonging to Clostridia XIVa and IV clusters (Miyake et al., 2018). These species produce SCFAs by the fermentation of soluble fiber contained in the diet. However, unfavorable health effects of SCFAs have also been described. For example, butyrate has been shown to act on the locus of enterocyte effacement pathogenicity island of enterohemorrhagic E. coli, which enables this pathogen to efficiently colonize the host epithelium (Luu et al., 2019).

Children, especially newborns and fetuses, are more sensitive to environmental toxicants compared to adults (Kamai, McElrath, & Ferguson, 2019). There is a clear association between alterations of GMB and metabolites in children and the risk of developing depression in adulthood (Frye et al., 2015; Petra et al., 2015). In the fetus, early-life gut bacterial colonization plays an important role in metabolic tissue development and in influencing the risk of immune related diseases because intestinal immune system development starts as early as 11 weeks of gestation in humans (Romano-Keeler & Weitkamp, 2015; Younge et al., 2019). Modulation of GMB by probiotics (Lactobacillus rhamnosus or Bifidobacterium lactis) during pregnancy alters infant immune responses (Prescott et al., 2008). The mode of delivery, antibiotic use after birth and infant formula consumption could all help shape the infant GMB and further modulate the immune system.

3. Phytoestrogens and the Gut-Brain Axis

The complex symbiotic interaction between GMB species can be perturbed by endocrine modulators. Furthermore, GMB can interact with the endocrine modulators by altering their processes of absorption, disposition, metabolism and excretion (Lai et al., 2018). Dietary isoflavones, especially the phytoestrogen genistein (GEN, Formula: C₁₅H₁₀O₅, CAS ID: 446-72-0), have been proposed as possible preventive or complementary medicines for depression, and they might improve the overall quality of life and decrease self-rating depression scores (Atteritano et al., 2014). The widely used dietary supplement GEN has been explored for its potential effects in cognitive function, cancer therapy, and bone and cardiovascular health. GEN presents as glycosides (genistin; Formula : C₂₁H₂₀O₁₀, CAS Number : 529-59-9; Figure 3) in intact soybeans. Orally administered glycoside form is hydrolyzed by β-glucosidase to aglycones in the GI tract. The aglycone form is either absorbed intact or further metabolized by GMB. Only a small fraction of dietary GEN is absorbed in the small intestine, and large proportions of that reach the colon where they undergo modifications by GMB. It has been estimated that at least 30% of metabolites have a bacterial origin. GEN and gut microbe-derived GEN metabolites (MGMs) can interact with estrogen receptors (ERs), and function as either antagonist or agonist depending on the estrogen level (Hwang et al., 2006). GEN is first converted by GMB to dihydrogenistein (Formula : C₁₅H₁₂O₅, CAS Number 21554-71-2; Figure 3), which can bind to ERs and exert biological effects (e.g., antioxidative). Dihydrogenistein has been detected in human urine and plasma at high concentrations, which may act as bioactive component of GEN (Kobayashi, Shinohara, Nagai, & Konishi, 2013). Dihydrogenistein is further metabolized to 6’-hydroxy-O-desmethylangolensin (C₁₅H₁₄O₅), with unknown health effects, through absorption and enterohepatic circulation (Kobayashi et al., 2013). A peak with molecular weight (257.0819 g/mol) identical to 5-hydroxy-equol was also found, suggesting that the production of this compound could be more common than equol. Interestingly, 5-hydroxy-equol showed an antioxidant activity superior to that of GEN (Gaya, Medina, Sanchez-Jimenez, & Landete, 2016). The 5-hydroxy-equol is also expected to bind to ERs, preferably to ER-β. Complete cleavage of the C-ring can also produce 2,4,6-trihydroxybenzoic acid and p-ethyl phenol. C-ring fission may also generate 2-(4-hydroxyphenyl)-propionic acid and trihydroxybenzene (Figure 3).

Figure 3.

Possible gut microbe-derived GEN metabolites.

The actions of GEN have been studied for more than 20 years, and a great deal has been learned, but the research up to date has not led to the significant clinical successes. More than 30 clinical trials of GEN with various disease indications have been conducted to evaluate its clinical efficacy, and ambiguous therapeutic effects and large interindividual variations have been observed (Yang, Kulkarni, Zhu, & Hu, 2012). The discrepancy between clinical studies of GEN could be attributed to a failure to distinguish between MGM producers and nonproducers in the metabolism of GEN and sex difference in GMB (Vemuri et al., 2019). The solubility of GEN in water and most aqueous buffers is low, e.g., 0.9 μg GEN/ml in water, and the oral availability of GEN is only 23.4% for the dose administered (Yang et al., 2012). In contrast, various MGMs have been detected in plasma and urine. These MGMs are more bioavailable than GEN per se, and they have increased biological activities, e.g., estrogenic or antiestrogenic, antioxidant, anti-inflammatory, antiproliferative and apoptosis-inducing (Gaya et al., 2016). The clinical effectiveness of GEN in depression may be attributed to MGMs. On one hand, GMB that is altered through GEN intake modulates active estrogen in the serum by secreting β-glucuronidase that deconjugates estrogen (Plottel & Blaser, 2011). On the other hand, the metabolites of isoflavones and estrogen from GMB can modulate the immune responses. They may be transmitted through the vagal nerve or systemic circulation to affect neural function. As the gut-associated lymphoid tissue represented 70% of entire immune system, the mechanism of GEN affecting GMB also needs to be studied further from the perspective of metabolome. In addition, as an endocrine disrupting chemical, GEN has been linked to some detrimental health effects, especially during developmental exposure (discussed later). For example, mice treated neonatally with GEN developed cancer of the uterus later in life (Newbold, Banks, Bullock, & Jefferson, 2001). Therefore, understanding the mechanisms underlying GEN’s beneficial and detrimental actions (e.g., depending on the dose and windows of exposure) will help form a more targeted therapy that have fewer side effects.

Human microbiome studies - Adult exposure.

Because limited human studies are available specifically for GEN, this section has considered both in vivo and in vitro gut microbial profile changes following either GEN exposure or soy consumption (Table S1). Soy intake can modulate GMB, estrogen metabolism and immunity. Isoflavone administration in the human GMB-associated mice led to a significant increase in fecal Clostridia (Tamura, 2004), and modified numbers of key bacterial species in the gut in vitro (Vazquez, Florez, Guadamuro, & Mayo, 2017). By culturing human feces in reactor vessels and introducing soy powder upon stabilization, an increase of several bacterial strains (Lactobacillus sp.) together with a 30% increase of SCFAs were found (De Boever, Deplancke, & Verstraete, 2000). In postmenopausal women, supplementation of isoflavones aglycon stimulated dominant microorganisms of the Clostridium coccoides-Eubacterium rectale cluster, Lactobacillus-Enterococcus group, Faecalibacterium prausnitzii subgroup and Bifidobacterium genus (Bolca et al., 2007; Clavel et al., 2005). Similarly, a week of diet supplementation with soy bars containing isoflavones (160 mg soy isoflavones/day) significantly increased Bifidobacterium (Nakatsu et al., 2014). In overweight and obese men, consuming soymilk altered the microbiome including a potentially beneficial alteration of the Firmicutes to Bacteroidetes ratio (Fernandez-Raudales et al., 2012). In athymic nude mice transplanted with human microbiome, GEN at the dose of 0.25 g/kg modulates the microbiome and contributes to its effects on increasing the latency of breast tumor and reducing tumor growth (Paul et al., 2017). Thus, adult exposure to GEN seemed to produce an overall beneficial effect.

Human microbiome studies - Developmental exposure (Table S1).

In humans, a critical period influencing lifelong health is the period from conception to 24 months (the first 1000 days) when GMB composition and eating patterns are established (Schwarzenberg, Georgieff, & Committee On, 2018). Infants who had their cow’s milk-based formula replaced with soymilk were associated with a decrease in the intestinal bifidobacterial population (Piacentini, Peroni, Bessi, & Morelli, 2010). A study in Australian children of 2 to 3 years old found soy intake was positively associated with the relative abundance of bacteria related to Bacteroides xylanisolvens (Smith-Brown, Morrison, Krause, & Davies, 2016). A cross-sectional study found that the urinary concentration of soy isoflavone GEN in infants consuming soy-based formula was 500 times higher than in those consuming cow’s milk-based formula (Cao et al., 2009). However, human studies have shown that twice as many children with type 1 diabetes consumed soy-based formula in infancy as compared to controls (Fort et al., 1986; Strotmeyer et al., 2004). In addition, soy milk formula consumption during infancy was associated with a significant increase in the use of asthma or allergy drugs in young women (Strom et al., 2001), and possible increases in autistic behaviors (Westmark, 2013). These conditions are associated with an overactive immune system, suggesting that GEN might have some adverse effects on children, especially newborns and fetuses.

Lab animal microbiome studies - Adult exposure (Table S2).

It was reported that soy milk could rescue cholesterol-disturbed GMB in male Sprague–Dawley rats (S. M. Lee, Han, & Yim, 2015), which was supported by additional four studies: (1) Soy protein isolate modulated the effects of prebiotic oligosaccharides on gut fermentation and microbiota in female Wistar rats (Bai, Ni, Tsuruta, & Nishino, 2016), (2) Dietary soy exerts a beneficial shift in gut microbial communities in ovariectomized rats with low-running capacity (Cross et al., 2017), (3) Soy exposure resulted in a lower Firmicutes:Bacteroidetes ratio in ovariectomized rats with low-running capacity (Vieira-Potter et al., 2018), and (4) In male BALB/c mice, Odoribacter (Bacteroidales family), Lactobacillus (Lactobacillales order), and Alistipes (Rikenellaceae family) were enriched in soymilk while bacterial taxa from Bacteroides and Lactobacillus were enriched in L. rhamnosus-fermented soymilk (Dai et al., 2019). For phytoestrogen GEN, it was shown that GMB alteration by ovariectomy may affect GEN bioavailability in C57BL/6 mice (D. H. Lee et al., 2017). Our in vivo studies showed that GEN could modulate GMB and immune homeostasis in adult non-obese diabetic (NOD) mice (Huang et al., 2017). In NOD male mice, it was found that GEN treatment during adulthood induced decreases in GM-CSF (75.6%), IFN-γ (22.9%), IL-5 (36.3%), IL-10 (45.3%), and MCP-1 (72.2%), suggesting an anti-inflammatory effect (Huang et al., 2017). These cytokines/chemokines have a strong association with depressive responses by interacting with GMB. When the composition of GMB at the genus level was compared, GEN treatment induced an increased Prevotella, and a decreased Alistipes and Blautia in terms of relative abundance, suggesting an anti-inflammatory response (Huang et al., 2017). Prevotella can also help maximize energy harvest from a plant-based diet (Y. J. Zhang et al., 2015), while Alistipes show a significant association with depressive symptoms as they are overly represented in patients with depression (Jiang et al., 2015; Naseribafrouei et al., 2014). Overall, consistent with human studies, adult exposure to GEN produced beneficial effects in lab animals (López et al., 2018).

Lab animal microbiome studies - Developmental exposure (Table S2).

In 21-day old DF508 mice, exposure to GEN at the dose of 600 mg/kg resulted in a lower within-sample diversity and significant differences in beta diversity when compared to control (Corrie Whisner, 2019). In California mice (Peromyscus californicus), early GEN exposure disrupted normal socio-communicative behaviors, which might be due to GEN-induced microbiota shifts and resultant changes in gut metabolites but might also be attributed to GEN disruptions on neural programming (Marshall et al., 2019). However, early-life GEN intake could also attenuate the harmful effects of maternal high fat diet in adult offspring, and the protective effects were associated with the alterations in GMB (Zhou, Xiao, Zhang, Zheng, & Deng, 2019). Our studies have also shown significant alterations of immune responses in mouse offspring exposed to GEN during in utero and lactation (Guo, Auttachoat, & Chi, 2005; Guo, Chi, Germolec, & White, 2005; Huang et al., 2018). In NOD mice, we have found that the effect of GEN in type 1 diabetes depended on sex and windows of exposure: perinatal exposure produced an exacerbation of type 1 diabetes in females and an anti-inflammatory effect in males (Huang et al., 2018), while adult exposure exerted a protection of type 1 diabetes in both sexes (Guo et al., 2015; Huang et al., 2017). RNA-seq analysis of gene expression in the ileum tissues in perinatal GEN-treated female offspring showed that intestinal α-defensin expression was decreased by 70% (Huang et al., 2018). Importantly, a case-control study using whole-genome copy number analysis showed that decreased dosage of defensin was a predisposing factor to idiopathic autism spectrum disorder (Cho et al., 2009).

Defensins, whose expression is modulated by estrogen, are 2–6 kDa, cationic, antibacterial peptides active against many Gram-negative and Gram-positive bacteria. We conducted GMB analysis in NOD offspring exposed to GEN during in utero and lactation and found that GMB from postnatal day (PND) 90 female offspring was significantly altered following perinatal GEN exposure with an increased level of Enterobacteriales (Genus), suggesting a pro-inflammatory response. Some members of the Enterobacteriaceae (e.g., E. coli) produce endotoxins that are an etiopathogenic agent of type 1 diabetes. In the NOD ileum, E. coli was the sole bacterium correlating with the insulitis score (Sane et al., 2018). Significantly, higher levels of Enterobacteriaceae genera and species were found in children with autism than healthy children (De Angelis et al., 2013). Moreover, compared with healthy subjects, serum levels of endotoxin were significantly higher in autistic patients, and inversely and independently correlated with socialization scores on the Vineland Adaptive Behavior Scales (VABS) and ADI-R Domain A score (social) (Emanuele et al., 2010). In addition, animal studies have shown that early exposure to GEN can lead to altered brain development and behavioral abnormalities (Ponti et al., 2017), and girls exposed during infancy to soy formula show reduced female-typical play behavior (Adgent, Daniels, Edwards, Siega-Riz, & Rogan, 2011). Taken together, developmental exposure to GEN might be harmful, which is dependent on sex.

4. Conclusions and Future Directions

The fundamental question driving the GMB field is - how do differences in microbial composition among individuals (i.e., interindividual diversity) affect human health and disease? It has been suggested that there are likely four major patterns of interindividual variability in GMB functions (Rosen & Palm, 2017). Type I functions show very low interindividual variability among humans. Type II functions are normally distributed among the population but show a wider range of variability, and the host can tolerate a broader range of activity. Type III functions are present in the majority of humans but absent in a small population. Type IV functions are present in a minority of individuals and absent in the majority. Future studies should focus on categorizing GMB and metabolite functions by type, which may not only help delineate the roles of specific microbial activities, or even microbes themselves, in health and diseases, but also act as a guide for which functions should be targeted therapeutically. A better understanding of the roles of GMB in neuroendocrine and neuroimmune interactions following exposure to endocrine modulating chemicals will permit individuals seeking dietary supplements to make better-educated choices. This is especially true for IBD, a chronic recurrent inflammatory disease in which sex hormones play an important role in its prevalence (Shah et al., 2018). If GMB contributes to neuroendocrine and neuroimmune interactions, therapeutics including prebiotics, probiotics, fecal transplants, isoflavone supplements or remediation strategies may be designed to induce/prevent such GMB alterations and thereby improve neurobehavioral outcomes.