There are rich connections between the CNS and intestinal microbes. Research has shown that intestinal microorganisms can modify the CNS, affecting behavior, mood, cognition, and even causing anxiety and depression. The gut–brain axis refers to the interaction between the gut microbes and the CNS, involving multiple systems such as nerves and endocrine (Collins et al., 2012; Holmes et al., 2012, 2020; Nicholson et al., 2012; Montiel-Castro et al., 2013). Intercommunication between intestinal microbes and the CNS involves two neuroanatomical pathways, and one is the direct exchange of information between the brain and the gut, including the vagus nerve in the spinal cord and the autonomic nervous system. The other is a bidirectional communication in the spinal cord through the enteric nervous system, autonomic nervous system, and vagus nerve of the gut (Wang and Wang, 2016).

Although the structure of the microorganism is relatively stable, factors including infection, diet, or genetics can also affect the intestinal flora, resulting in the growth of pathogenic bacteria, the loss of normal commensal bacteria, and the decline in diversity (Levy et al., 2017), which is called dysbiosis, and can further lead to diseases such as obesity (Gérard, 2016), autoimmune disease (Knip and Siljander, 2016), neurological disorders (Tremlett et al., 2017), and inflammatory bowel disease (Wlodarska et al., 2015). A study showed that using mass spectrometry to detect chemicals in the peripheral circulation in germ-free (GF) mice found that the synthesis of most substances depends on the gut microbiome (Wikoff et al., 2009). Substances produced by intestinal microbial metabolism can act on the CNS. For example, short-chain fatty acids (SCFAs) are the final product produced by microorganisms in the lower intestinal tract through fermentation, including acetic acid, propionic acid, and butyric acid (Martin-Gallausiaux et al., 2021), are important substrates for keeping the integrity of the epithelial barrier (Morrison and Preston, 2016), and are also involved in regulating human immune function and exerting anti-inflammatory effects (Martin-Gallausiaux et al., 2021). SCFAs can also penetrate the gut–blood barrier and the blood–brain barrier (BBB) (Macfabe, 2012), alter neurotransmitter and hormone concentrations (Alexander et al., 2019), reduce the permeability of the BBB (Braniste et al., 2014), and regulate the formation and function of microglia (Hu et al., 2020).

In addition, entheogenic GABA, 5-HT, can affect microglial activation in the brain (Abdel-Haq et al., 2019). Diaz Heijtz et al. (2011) showed that GF mice exhibited less anxiety-like behavior than conventional mice fed in a specific pathogen-free (SPF) environment. However, after the adult GF mice were transferred to the SPF environment, there was no significant increase or decrease in anxiety-like behavior, but the anxiety-like behavior of their offspring returned to normal (Diaz Heijtz et al., 2011). It has also been shown that ACTH and corticosterone levels under restraint stress are different in GF mice and SPF mice (Sudo et al., 2004). All of these studies illustrate the ability of gut microbes to regulate behavior.

Although the BBB separates the brain and peripheral tissues, gut microbiota can influence CNS function (Bercik et al., 2011; Diaz Heijtz et al., 2011). By comparing GF and SPF rats, Erny et al. (2015) showed that gut microbiota could affect microglial function, including reducing cell-activated gene transcription and increasing transcriptional repressor expression, and Mosher and Wyss-Coray (2015) observed that entheogenic SCFAs can promote microglial maturation and related gene expression in SPF mice. Microglia are located in the CNS and are the primary immune cells in the brain, but they are closely related to the mononuclear macrophage system (Erny et al., 2015) and can generate immune responses to peripheral inflammation. Combrinck et al. (2002) used an intraperitoneal injection of lipopolysaccharide (LPS) to induce peripheral systemic inflammation and observed further activation of activated microglia in the brain in animals with chronic inflammation, suggesting that there is an interaction between peripheral systemic inflammation and the CNS.

Gut microbial surface substances and their metabolites can positively or negatively affect the CNS through the peripheral circulation or enteric nervous system, and microbial-derived neurotransmitters have regulatory effects on neurons or nerve cells after entering the brain. The metabolites can not only play an anti-inflammatory effect and protect the BBB but also play a pro-inflammatory effect, activate microglia to secrete inflammatory mediators, and then damage the BBB, leading to leakage.

Gut microbes are a microbial ecosystem that exists in the human gastrointestinal tract. The average adult’s gut microbes weigh as much as the human brain, and these microbes encode genes that are more than 100 times more abundant than the human genome (Dinan et al., 2015), and there are more than 10¹⁴ kinds of microorganisms. In addition to bacteria, archaea, viruses, bacteriophages, yeasts, and fungi coexist in the gut (Cani, 2018). This complex microbial population is increasingly becoming an important role in affecting human health and is closely related to the stability of the human body’s internal environment. For example, (1) the formation of a bacterial barrier in the gut prevents pathogenic bacteria from penetrating the intestinal mucosal barrier and invading the host (Chopyk and Grakoui, 2020), (2) the synthesis of nutrients and supplements, such as SCFAs and vitamins (Bishehsari et al., 2018; Li et al., 2020), (3) detoxification of ingested dietary toxicants (Letertre et al., 2020), (4) modulation of host immune system function (Takiishi et al., 2017), and (5) gut–brain communication (Morais et al., 2021).

The characteristics of a healthy gut microbiome can be summed up in four words, namely, diversity, stability, resistance, and resilience. These four words correspond to the richness within the microbiota, the ability to adapt to environmental changes, and the ability of the microbiota to recover (Lozupone et al., 2012). To investigate the impact of gut microbes on humans, methods such as 16S ribosomal RNA gene sequencing (Johnson et al., 2019), metagenomic sequencing (Riesenfeld et al., 2004), and Illumina genome analyzers (Qin et al., 2010) are used to explore the composition of the microbiota. Alpha diversity (single site, such as the human gut) and beta diversity (multiple sites, such as different parts of the human body) were applied to evaluate sequencing data, and the human gut microbiota could be roughly divided into four phyla, namely, Bacteroidetes, Firmicutes, Proteobacteria, and Actinobacteria (Arumugam et al., 2011; Greenhalgh et al., 2016; Binda et al., 2018). Firmicutes and Bacteroidetes accounted for 90% of the intestinal flora abundance, and Proteobacteria and Actinobacteria accounted for 10%. The abundance of Bacteroidetes is associated with a high-fat, high-protein diet, and malnutrition (Johnson et al., 2017), while Firmicutes is increased in obese mice (Turnbaugh et al., 2006).

In healthy people, the composition of the gut microbiome is all about the same, but there is a marked difference in the microbial structure compared with epilepsy patients. According to Peng et al. (2018), an abnormally increased abundance of rare flora was observed in the patients with drug-resistant epilepsy, while patients with drug-sensitive epilepsy have an intestinal flora composition similar to healthy controls. Recently, a 22-year-old patient with drug-resistant epilepsy who received gut microbiota transplantation from a healthy individual remained seizure-free for 20 months despite discontinuation of anti-seizure medications (ASMs) (He et al., 2017). Therefore, it is inferred that normal gut microbiota can control epileptic seizures.

Diet is the most important factor in modifying the structure and function of intestinal flora (Zmora et al., 2019). The ketogenic diet (KD) and the modified Atkins diet are commonly used for epilepsy control. They are characterized by high fat and low carbohydrates, making the body switch from consuming glucose to consuming ketone bodies produced by fat metabolism as an energy source (Ułamek-Kozioł et al., 2019). A study shows that children with drug-resistant epilepsy treated with a KD for seizures had lower gut microbial diversity than healthy infants (Xie et al., 2017). The KD also had different effects on the intestinal microbiota composition of healthy infants and children with drug-resistant epilepsy. Firmicutes did not change significantly in healthy infants before and after KD treatment, but the proportion of Bacteroidetes increased. In addition, Proteobacteria was more enriched in children with drug-resistant epilepsy, and the proportion decreased after KD treatment (Xie et al., 2017). In the study by Wells et al. (2020), KD had a good effect on seizure control. This study has shown that elevated levels of ketone bodies (through a KD, exogenous ketonic supplements, or weight loss) can reduce airway hyperresponsiveness in obese asthmatic rats, which may be related to the inhibition of inflammatory responses (Mank et al., 2022).

The possible mechanism of the KD for epilepsy is not fully understood, but it may exert antiepileptic effects by increasing endogenous adenosine (Masino et al., 2009), opening ATP-sensitive potassium channels (Juge et al., 2010), acting on adenosine A1 receptors (Kadowaki et al., 2017), and inhibiting lactate dehydrogenase (Sada et al., 2015). Furthermore, according to Braakman and van Ingen (2018), six drug-resistant epilepsy patients acquired seizure-freedom during antibiotic use, suggesting that antibiotics may be a potential treatment for epilepsy.

Recently, probiotics represented by Bifidobacteria and Lactobacilli exist in our daily diets, such as yogurt, nutritional supplements, and fermented biscuits. One study showed that after oral administration of 8 mixed probiotics (mainly Lactobacillus) as a supplement to antiepileptic drugs in epilepsy patients, 28.9% of patients had more than 50% reduction in seizures, which was similar to other new ASMs; meanwhile, the quality of life (QoL) in the effective group of probiotics was significantly improved (Gómez-Eguílaz et al., 2018), and it is suggested that probiotic supplementation therapy may be a new method to control epilepsy.

The BBB acts as the dividing line between the central and peripheral parts, preventing blood cells, pathogens, and poisons from entering the brain (Montagne et al., 2017). The phenomenon that neurons regulate the BBB function puts forward the neurovascular unit (NVU) concept. NVU is formed by the mutual coupling of vascular-related cells, glial cells, and neurons (Zlokovic, 2011), which regulates not only cerebral blood flow (Lecrux and Hamel, 2011) but also the tight junction (TJ) protein between endothelium can limit the paracellular permeability of the BBB, reducing the substance transport between cerebrospinal fluid and blood (Zlokovic, 2008), thereby maintaining central environmental homeostasis. The basis of this barrier function depends on the structure of TJs between endothelial cells, which are mainly composed of claudin, occludin, and ZO proteins (Berndt et al., 2019; Lochhead et al., 2020; Yuan et al., 2020). Among them, claudin-5 is the main protein constituting TJs (Greene et al., 2019). Research shows that the CNS of claudin-5 knockout mice displays higher permeability to macromolecules (Lochhead et al., 2020). At the same time, in cultured cerebral vascular endothelial cells, claudin-5 overexpression showed decreased paracellular permeability and increased tightness (Ohtsuki and Terasaki, 2007). The study by Li et al. (2016) showed that sodium butyrate could improve the neurological deficit after traumatic brain injury, upregulate the expression of TJs, and reduce the permeability of the BBB. Some studies have also pointed out that the local leakage of plasma proteins caused by the increased permeability of the BBB may play a critical role in epileptogenesis (Bankstahl et al., 2018).

In the CNS, astrocytes are one of the most abundant glial cells. They have an important role in the CNS and actively participate in the composition of the BBB or promote or limit the development of diseases (Linnerbauer et al., 2020). Furthermore, the role of astrocytes in epileptogenesis is increasingly well understood (Ivens et al., 2007; Seifert et al., 2010; Heinemann et al., 2012).

Glutamine synthase is an enzyme specifically expressed in astrocytes in the CNS, which can catalyze the synthesis of glutamine from ammonia and glutamate (Rose et al., 2013), thus maintaining CNS homeostasis, but decreased or lost glutamine synthase expression was found in patients with temporal lobe epilepsy (Sandhu et al., 2021), and this may be one of the possible mechanisms for inducing seizures. In addition, astrocytes also express a variety of potassium ion channels on the surface of astrocytes (Seifert et al., 2018) to maintain intracellular and extracellular potassium balance through potassium ion buffering and potassium ion uptake (Seifert and Steinhäuser, 2013). During the neuronal activity, extracellular potassium concentrations can rapidly fluctuate to upper levels, sufficient to induce seizures if the hyperkalemic environment is not properly corrected. Niday and Tzingounis (2018) suggested that extracellular potassium rises to depolarize neurons in the absence of Kir 4.1 channels, leading to the inactivation of sodium channels, thereby prolonging neuronal firing time or increasing the frequency, which causes neuronal damage hyperexcitability. Furthermore, in the research of Snowball et al. (2019), using a lentiviral vector packaging the engineered potassium channel (EKC) gene significantly reduced the number of seizures in focal neocortical epilepsy.

Microglia are macrophages in the brain parenchyma that derive from primitive hematopoiesis in the yolk sac (Aguzzi et al., 2013) and play an important role in neuronal health, apoptosis, and synapse formation (Nayak et al., 2014). Many studies show that microglia are closely related to neurological diseases (Hansen et al., 2018; Ho, 2019; Jia et al., 2021). The possible mechanisms include inflammation, gliosis, and stress. Microglia-derived inflammation is important in epileptogenesis, and ASMs with antiglial inflammatory properties benefit seizure control in a previous study (Dambach et al., 2014). Studies have also focused on the connection between microglia and neurons as the main factor for microglia activation, and the inflammatory proteins released by glial activation can increase excitability and contribute to epilepsy (Alyu and Dikmen, 2017).

As part of innate immunity, cytokines are also closely related to the BBB. For example, in Qin et al.’s (2019) study, it was found that interleukin-1β can induce the pericyte NF-κB/p65 pathway leading to upregulation of matrix metalloproteinase-9 expression to the destruction of vascular endothelial TJs, thereby increasing the permeability of the BBB. TGF-β plays a crucial role in human development. When TGF-β binds to TGF-βR2, downstream SMAD proteins are phosphorylated and translocated to the nucleus, ultimately regulating the transcription of target genes of TGF-β (Vander Ark et al., 2018). Albumin present in brain tissue is translocated into cells by astrocytes through binding to TGF-β receptors on astrocytes and by downregulating the Kir 4.1 potassium channel on the surface of astrocytes, and it leads to the decrease of extracellular buffer potassium, increasing N-methyl-D-aspartate (NMDA) receptor-mediated neuronal excitability (Ivens et al., 2007).

The IL-1 family consists of pro- and anti-inflammatory cytokines (Yazdi and Ghoreschi, 2016). Among them, interleukin-1β is the most characteristic pro-inflammatory interleukin. Description of IL-1β was initially thought to be an endogenous pyrogen that induces the expression of COX-2, iNOS, TNF-α, IL-6, chemokines, adhesion molecules, and matrix metalloproteinases (Dinarello, 2005).

Studies have shown that the genes for IL-1β, IL-1R1, and IL-1RA are overexpressed in rodent models of epilepsy (Mukhtar, 2020). IL-1R1 is a receptor for IL-1β, belonging to the Toll-like/IL-1 receptor, activating prostaglandins (PGs) and NF-κB through MyD88 protein, thereby inducing inflammation (Yazdi and Ghoreschi, 2016). IL-1RA is an endogenous inhibitor of IL-1R1 (Ferrara-Bowens et al., 2017). Vezzani et al. (2008) suggested that IL-1β activates IL-1R1 on neurons and induces tyrosine phosphorylation of the NR2B subunit of NMDA receptor through Src kinase, leading to increased NMDA receptor-mediated calcium influx, thereby enhancing neuronal excitability. IL-1β also promotes glutamate release and inhibits glutamate reuptake by astrocytes through TNF-α, thereby inducing epileptiform events (Vezzani et al., 2008). IL-6 is a cytokine with dual effects, and proper IL-6 expression is very important for host defense function. After infection or injury, IL-6 is rapidly secreted and produced by monocytes/macrophages (Tanaka et al., 2016). In the healthy CNS, IL-6 is lowly expressed, but astrocytes and microglia become important sources of IL-6 in the CNS (Gruol, 2015). When TGF-beta on astrocytes can transmit information to the cells, upregulation of IL-6 leads to increased cortical excitability and finally induces epileptiform discharges in vitro (Levy et al., 2015).

Cyclooxygenase (COX) is an enzyme present on the cell membrane that catalyzes arachidonic acid to PGs (Clària, 2003). COX has three isoenzymes, of which COX-2 is an inducible enzyme associated with inflammation (Zhu et al., 2020), and found the induction of COX-2 in a hippocampal kindled rat model, suggesting that COX-2 is a key factor in epileptogenesis (Rawat et al., 2019). Rojas et al., 2014. found that celecoxib has an antiepileptic effect on acute seizures, but the antiepileptic effect of NSAIDs appears to be related to the time of administration (Rojas et al., 2014). Besides, a previous study by Kovács et al. (2011) showed that intraperitoneal injection of LPS in WAG/Rij rats enhanced their spike-wave discharge (Coenen and Van Luijtelaar, 2003), they subsequently found that the number and duration of spike-wave discharge increased after intraventricular injection of LPS in WAG/Rij rats, NSAIDs (indomethacin) could eliminate this phenomenon, and this strongly validates the role of COX in epilepsy.

HMGB1 is a highly conserved and proteinaceous structure in cells and acts as a classic alarm protein due to its ability to activate DAMP receptors of the innate immune system when present extracellularly (Yang et al., 2020). There are many reported receptors of HMGB1, such as RAGE, TLR9, TLR4, CD24, and CXCR4 (Paudel et al., 2018), but only RAGE and TLR4 are not controversial (Andersson et al., 2018). Recently, the role of HMGB1 in epilepsy has gradually attracted researchers’ attention. Zhao et al. (2017) showed that anti-HMGB1 mAbs could antagonize seizures in various epilepsy models and TLR4 knockout mice. This antiepileptic effect was absent (Zhao et al., 2017). Fu et al. (2017) also found that anti-HMGB1 monoclonal antibodies can delay HMGB1 translocation and downregulate the expression of inflammation-related inflammation-related factors.

Lipopolysaccharide, albumin, and SCFAs in the peripheral circulation can act on various receptors, resulting in increased calcium conductance, downregulated potassium channels, promotion of intracellular transcription, and release of immune molecules to cause BBB damage, which in turn lowers the seizure threshold or promotes epileptiform discharges, which eventually lead to seizures.