Vagal nerve

The most direct route between the gut and the brain is the vagal nerve, roughly translated as the wandering nerve, and also known as the 10th cranial nerve. In reports from 1953 and 1961, ablation of the vagal nerve, formerly used for the treatment of peptic ulcers, resulted in an increase in the incidence of psychiatric disorders (Browning & Houseworth, Reference Browning and Houseworth1953; Whitlock, Reference Whitlock1961). Interestingly, this procedure recently turned out to reduce the incidence of Parkinson's disease (Svensson et al., Reference Svensson, Horváth-Puhó, Thomsen, Djurhuus, Pedersen, Borghammer and Sørensen2015). In animals, Sgritta et al. (Reference Sgritta, Dooling, Buffington, Momin, Francis, Britton and Costa-Mattioli2019) observed that the effects of Lactobacillus reuteri on social behaviour were dissipated after vagotomy in a genetic mouse model of autism (Shank3B^−/− mouse).

Short-chain fatty acids

SCFAs are capable of signalling to the brain indirectly via nerve activation and can therefore influence behaviour (Sherwin, Sandhu, Dinan, & Cryan, Reference Sherwin, Sandhu, Dinan and Cryan2016). Ninety five per cent of SCFAs consist of acetate, propionate and butyrate. The primary source of SCFAs is microbial fermentation of dietary fibres in the gut, which is supported by the fact that germ-free animals and antibiotic treatment results in low SCFA levels (Cryan et al., Reference Cryan, O'riordan, Cowan, Sandhu, Bastiaanssen, Boehme and Dinan2019). SCFAs are capable of modulating neurotransmission. For example, propionic acid increases tryptophan hydroxylase expression which can reduce indoleamine serotonin, thereby influencing serotonergic neurotransmission (El-Ansary, Bacha, & Kotb, Reference El-Ansary, Bacha and Kotb2012; Nankova, Agarwal, MacFabe, & La Gamma, Reference Nankova, Agarwal, MacFabe and La Gamma2014). So far, only a few studies have investigated relations between neuropsychiatric disorders and SCFAs. Skonieczna-żydecka et al. (Reference Skonieczna-żydecka, Grochans, Maciejewska, Szkup, Schneider-Matyka, Jurczak and Stachowska2018) reported lower median content of acetate and higher isocaproic acid concentration in depressed women as compared to healthy women. Negative correlations were yielded between acetic, propionic and isocaproic acids and severity of depression as assessed with the Beck's Depression Inventory (BDI) scores. Szczesniak, Hestad, Hanssen, and Rudi (Reference Szczesniak, Hestad, Hanssen and Rudi2016) reported significantly higher isovaleric levels in the depressed patients as compared to healthy subjects. However, Kelly et al. (Reference Kelly, Borre, O’ Brien, Patterson, El Aidy, Deane and Dinan2016) found no differences in acetate, propionate, isobutyrate and butyrate between depressed patients and healthy controls. The limited sample sizes in all three studies could explain these inconsistent results. In animal models of mania, sodium butyrate reversed mania-like, e.g. behavioural hyperactivity, and depressive-like behaviour (Moretti et al., Reference Moretti, Valvassori, Varela, Ferreira, Rochi, Benedet and Quevedo2011; Resende et al., Reference Resende, Valvassori, Réus, Varela, Arent, Ribeiro and Quevedoa2013; Valvassori et al., Reference Valvassori, Resende, Budni, Dal-Pont, Bavaresco, Reus and Quevedo2015)

Immune mediators

Immune mediators are important intermediaries between the gut microbiota and the brain. Cytokines can signal to the brain from the periphery via the vagal nerve or via the circumventricular organs (Sherwin et al., Reference Sherwin, Sandhu, Dinan and Cryan2016). In the blood, bacteria or fragments from bacteria can bind lipopolysaccharide (LPS) binding protein (LBP), which is then connected to TLR-4, expressed on monocytes, macrophages and microglia, by soluble cluster of differentiation 14 (sCD14) (Kiecolt-Glaser et al., Reference Kiecolt-Glaser, Wilson, Bailey, Andridge, Peng, Jaremka and Belury2018; Kitchens & Thompson, Reference Kitchens and Thompson2005; Lim, Chang, & Wu, Reference Lim, Chang and Wu2019), which can lead to nuclear factor kappa-light-chain enhancer of activated B cells (NFκB) activation, inducing pro-inflammatory cytokine production (Genedi et al., Reference Genedi, Janmaat, Haarman and Sommer2019; Sherwin et al., Reference Sherwin, Sandhu, Dinan and Cryan2016). Thus, the presence of bacteria or small parts of bacteria can activate the immune system which also affects the brain. Blood levels of LBP, sCD14 and NFκB can reflect activity of this route. The group of Robert Yolken showed that these levels are increased in patients with BD and SSD, when patients had comorbid gastro-intestinal complaints (Severance et al., Reference Severance, Gressitt, Stallings, Origoni, Khushalani, Leweke and Yolken2013; Severance, Dickerson, & Yolken, Reference Severance, Dickerson and Yolken2020). Moreover, Foster, Baker, and Dursun (Reference Foster, Baker and Dursun2021) recently reviewed the relation of the GM and the immune system in MDD, where they draw attention to the importance of the immune system as an important player in the neurobiology combined with the GM in subtypes of depression.

Tryptophan metabolites

Along with influencing other metabolites, gut-bacteria influence tryptophan metabolism (Carlessi, Borba, Zugno, Quevedo, & Réus, Reference Carlessi, Borba, Zugno, Quevedo and Réus2021). Tryptophan is an essential amino acid, which is metabolized by two main pathways, namely, the serotonin (5-HT) pathway and the kynurenine pathway. According to a meta-analysis of 101 studies tryptophan and kynurenine are decreased across MDD, BD and SSD (Marx et al., Reference Marx, McGuinness, Rocks, Ruusunen, Cleminson, Walker and Fernandes2020). Conventional antidepressants enhance levels of central serotonin to produce a therapeutic effect. Most of the tryptophan is metabolized into kynurenine in the liver by tryptophan-2,3-dioxygenase for energy production, or following an inflammatory stimulus by indoleamine-2,3-dioxygenase. The presence of gut bacteria such as Clostridium perfringens can modulate gut production of 5-HT (Beaver & Wostmann, Reference Beaver and Wostmann1962; Yano et al., Reference Yano, Yu, Donaldson, Shastri, Ann, Ma and Hsiao2015). Studies in rats have shown that probiotic treatment can influence tryptophan, kynurenine and serotonin levels (Sherwin et al., Reference Sherwin, Sandhu, Dinan and Cryan2016). Rats treated with Bifidobacterium infantis 35624 had increased circulating tryptophan levels (Desbonnet, Garrett, Clarke, Bienenstock, & Dinan, Reference Desbonnet, Garrett, Clarke, Bienenstock and Dinan2009). Furthermore, rats treated with Lactobacillus johnsonii had lower circulating kynurenine levels with increased serotonin levels in the ileum and in the circulation (Valladares et al., Reference Valladares, Bojilova, Potts, Cameron, Gardner, Lorca and Gonzalez2013). Moreover, one study in MDD patients found decreases in the kynurenine/tryptophan ratio after treatment with probiotics (Lactobacillus helveticus and Bifidobacterium longum) (Kazemi, Noorbala, Azam, Eskandari, & Djafarian, Reference Kazemi, Noorbala, Azam, Eskandari and Djafarian2019). A meta-analysis studying the effect of probiotic supplementation on the tryptophan–kynurenine pathway observed significantly lower serum kynurenine and a decreased kynurenine/tryptophan ratio after treatment (Purton et al., Reference Purton, Staskova, Lane, Dawson, West, Firth and Marx2021). The results so far provide preliminary evidence that probiotics can modulate the tryptophan–kynurenine pathway.

Gastrointestinal barrier

The gastrointestinal barrier is a dynamic, multilayer system which consists of a physical barrier and a biochemical barrier. The main components of the physical barrier are the epithelial cells sealed by tight junctions and the gut mucosa. The biochemical barrier consists of the gut microbiota and the mucosal immune system. Its main function is to keep pathogens out of the host's internal milieu, while at the same time facilitate the absorption of nutrients, water and electrolytes (Bischoff et al., Reference Bischoff, Barbara, Buurman, Ockhuizen, Schulzke, Serino and Wells2014). For the homoeostasis of the organism, the permeability of the gastrointestinal barrier needs to be maintained within a narrow equilibrium.

Diet and lifestyle factors such as energy-dense food and alcohol can disturb gut permeability and lead to translocation of luminal components and reactivity of the intestinal immune system (Bischoff et al., Reference Bischoff, Barbara, Buurman, Ockhuizen, Schulzke, Serino and Wells2014). Increased intestinal permeability can also be a result of changes in gut microbiota, infections or reduced perfusion of the gut (Bischoff et al., Reference Bischoff, Barbara, Buurman, Ockhuizen, Schulzke, Serino and Wells2014).

Gastrointestinal permeability and brain structure and function

Gastrointestinal permeability and brain structure and function are governed by a bidirectional interaction. On one hand, pre-clinical studies (Keita, Söderholm, & Ericson, Reference Keita, Söderholm and Ericson2010; Kiliaan et al., Reference Kiliaan, Saunders, Bijlsma, Cecilia Berin, Taminiau, Groot and Perdue1998; Söderholm et al., Reference Söderholm, Yang, Ceponis, Vohra, Riddell, Sherman and Perdue2002; Vicario et al., Reference Vicario, Guilarte, Alonso, Yang, Martínez, Ramos and Santos2010) and studies in healthy volunteers (Vanuytsel et al., Reference Vanuytsel, Van Wanrooy, Vanheel, Vanormelingen, Verschueren, Houben and Tack2014) suggest a causal effect of psychosocial stress on gut permeability, likely through corticotropin-releasing hormone-mediated mast cell activation and decreased blood flow to the gut during stressful periods. On the other hand, it is hypothesized that increased gut permeability and abnormal influx of food- and bacteria-derived antigens drives systemic low-grade immune dysregulation, which in turn affects brain structure and function (Genedi et al., Reference Genedi, Janmaat, Haarman and Sommer2019). Lastly, psychiatric comorbidity is common in diseases with known structural and functional abnormalities of the gastrointestinal barrier, such as Crohn's disease and colitis ulcerosa (Bennett, Tennant, Piesse, Badcock, & Kellow, Reference Bennett, Tennant, Piesse, Badcock and Kellow1998; Faresjö et al., Reference Faresjö, Grodzinsky, Johansson, Wallander, Timpka and Åkerlind2007; Nicholl et al., Reference Nicholl, Halder, Macfarlane, Thompson, O'Brien, Musleh and McBeth2008). It is therefore hypothesized that abnormal intestinal permeability is involved both as an effect and as a cause of severe psychiatric disorder, yet, this is still a research area in its infancy.

Evidence of gastrointestinal barrier dysfunction in MDD, BD and SSD

In MDD, BD and SSD, gastrointestinal permeability has been assessed with markers of structural barrier integrity and paracellular permeability such as zonulin and intestinal-type fatty acid-binding protein (I-FABP) (Table 1). In a study in MDD or anxiety disorder, authors reported significantly higher levels of zonulin and I-FABP in patients compared to controls and this was associated with gut dysbiosis (Stevens et al., Reference Stevens, Goel, Seungbum, Richards, Holbert, Pepine and Genomics2018). Alvarez-Mon et al. (Reference Alvarez-Mon, Gómez, Orozco, Lahera, Sosa, Diaz and Alvarez-Mon2019) also reported significantly higher levels of I-FABP in MDD patients compared to controls but no significant difference was observed for zonulin. In another study, patients of different psychiatric diagnoses who had recently attempted suicide had higher levels of I-FABP but lower levels of zonulin compared to both MDD patients without a history of suicide attempt and healthy controls (Ohlsson et al., Reference Ohlsson, Gustafsson, Lavant, Suneson, Brundin, Westrin and Lindqvist2019). In BD, patients in full remission showed higher zonulin and claudin-5 compared to controls, while there were no differences between active manic episode and remission (Kılıç, Işık, Demirdaş, Doğuç, & Bozkurt, Reference Kılıç, Işık, Demirdaş, Doğuç and Bozkurt2020). Interestingly, this observation might suggest that zonulin and claudin-5 are ‘trait’ rather than ‘state’ markers of BD. However, the small sample size of this study precludes firm conclusions, before the findings are confirmed in larger populations. Moreover, there is a certain level of uncertainty surrounding the measurement of zonulin with commercially available ELISA kits. The results should be interpreted with caution until an in depth understanding of the biomarker is acquired (Massier, Chakaroun, Kovacs, & Heiker, Reference Massier, Chakaroun, Kovacs and Heiker2021). In schizophrenia, Maes, Kanchanatawan, Sirivichayakul, and Carvalho (Reference Maes, Kanchanatawan, Sirivichayakul and Carvalho2019a) and Maes, Sirivichayakul, Kanchanatawan, and Vodjani (Reference Maes, Sirivichayakul, Kanchanatawan and Vodjani2019b) showed that immunoglobulin M (IgM) responses to zonulin were higher in patients compared to controls, while IgM response to occludin was significantly associated with deficit schizophrenia. Authors also reported an increased ratio of IgM towards components of the paracellular route (zonulin + occludin)/components of the transcellular route (talin + actin + vinculin) in deficit v. non-deficit schizophrenia and in schizophrenia v. controls. This ratio was significantly associated with increased IgA responses to Gram-negative bacteria. Schizophrenia patients with deficit syndrome (i.e. severe negative and cognitive symptoms) can be expected to have unhealthy dietary habits, which could explain the association with IgM response to occludin. Barber et al. (Reference Barber, Sturgeon, Fasano, Gascella, Eaton, McMahon and Kelly2019) showed that 42.9% of the patients had higher levels of zonulin than the cut-off for elevated levels (>2.33 mg/dl).

Table 1. Main findings of studies assessing gut permeability in MDD, BD and SSD

sCD14, soluble cluster of differentiation 14; LBP, lipopolysaccharide (LPS) binding protein; I-FABP, fatty acid-binding protein.

Markers of bacteria translocation can also reflect abnormal gastrointestinal permeability. LPS and LBP were up-regulated in MDD or anxiety disorder (Alvarez-Mon et al., Reference Alvarez-Mon, Gómez, Orozco, Lahera, Sosa, Diaz and Alvarez-Mon2019; Stevens et al., Reference Stevens, Goel, Seungbum, Richards, Holbert, Pepine and Genomics2018). In BD, sCD14, a co-receptor for the LBP, used as a marker for bacterial translocation was significantly higher in patients compared to controls (Severance et al., Reference Severance, Gressitt, Stallings, Origoni, Khushalani, Leweke and Yolken2013). In SSD, sCD14 was significantly higher in multiple studies (Dzikowski et al., Reference Dzikowski, Juchnowicz, Dzikowska, Rog, Próchnicki, Kozioł and Karakula-Juchnowicz2020; Severance et al., Reference Severance, Gressitt, Stallings, Origoni, Khushalani, Leweke and Yolken2013; Tanaka et al., Reference Tanaka, Matsuda, Hayes, Yang, Rodriguez, Severance and Eaton2017; Weber et al., Reference Weber, Gressitt, Cowan, Niebuhr, Yolken and Severance2018) but not all (Morch et al., Reference Morch, Dieset, Færden, Reponen, Hope, Hoseth and Andreassen2019). In regard to antibodies against gut commensal bacteria, in MDD, Maes et al. (Reference Maes, Kubera, Leunis, Berk, Geffard and Bosmans2013) showed increased IgM/IgA against Gram-negative bacteria. More recently, Simeonova et al. (Reference Simeonova, Stoyanov, Leunis, Carvalho, Kubera, Murdjeva and Maes2020) showed increased immune responses to Gram-negative bacteria in both MDD and BD, especially in the presence of melancholia, compared to healthy controls. Interestingly, IgA/IgM response profiles differ among the two diagnosis, as well as between BD I and II subtypes and patients with melancholia. This might reflect a common underlying disruption of the gut barrier across diagnoses, accompanied however by distinct microbiome profiles and immune susceptibilities, though this remains to be confirmed in future studies. In SSD IgM/IgA against gut commensal bacteria were associated with negative symptoms, neurocognitive impairments and the deficit phenotype of SSD (Maes et al., Reference Maes, Sirivichayakul, Kanchanatawan and Vodjani2019b).

Alpha and beta diversity

α-Diversity is the mean species diversity in sites or habitats at a local scale (Whittaker, Reference Whittaker1972). α-Diversity is often measured by the Fisher, Ace, Chao, Simpson and/or Shannon indices. β-Diversity shows differentiation among habitats (Whittaker, Reference Whittaker1972). Multiple studies analysed diversity metrics; the findings however were mixed.

For MDD studies (Table 2), α-diversity was not different from controls in three studies (Chung et al., Reference Chung, Chen, Chou, Chen, Lee, Chuang and Kuo2019; Zheng et al., Reference Zheng, Zeng, Zhou, Liu, Fang, Xu and Fan2016, Reference Zheng, Yang, Li, Wu, Liang, Yin and Wang2020). Two studies found lower α-diversity in MDD (Huang et al., Reference Huang, Shi, Li, Shen, Shi, Wang and Liang2018; Liu et al., Reference Liu, Zhang, Wang, Wang, Zhang, Jiang and Duan2016). Four studies found inconsistent findings across indices (Jiang et al., Reference Jiang, Ling, Zhang, Mao, Ma, Yin and Ruan2015; Kelly et al., Reference Kelly, Borre, O’ Brien, Patterson, El Aidy, Deane and Dinan2016; Lai et al., Reference Lai, Deng, Xu, Zhao, Xu, Liu and Rong2021; Rong et al., Reference Rong, Xie, Zhao, Lai, Wang, Xu and Liu2019). β-Diversity was significantly different between MDD and controls in four studies (Huang et al., Reference Huang, Shi, Li, Shen, Shi, Wang and Liang2018; Kelly et al., Reference Kelly, Borre, O’ Brien, Patterson, El Aidy, Deane and Dinan2016; Lai et al., Reference Lai, Deng, Xu, Zhao, Xu, Liu and Rong2021; Zheng et al., Reference Zheng, Zeng, Zhou, Liu, Fang, Xu and Fan2016) and not different in three studies (Chung et al., Reference Chung, Chen, Chou, Chen, Lee, Chuang and Kuo2019; Jiang et al., Reference Jiang, Ling, Zhang, Mao, Ma, Yin and Ruan2015; Rong et al., Reference Rong, Xie, Zhao, Lai, Wang, Xu and Liu2019).

Table 2. Studies of the GM in major depressive disorder

HC, healthy controls; MDD, major depressive disorder; PCoA, principal coordinates analysis; PLS-DA, partial least-squares discriminant analysis; HDRS, Hamilton Depression Rating Scale; MADRS, Montgomery–Åsberg Depression Rating Scale; HAMDS, Hamilton's Depression Scale; PROMIS Depression Score: Patient-Reported Outcomes Measurement Information System Depression Score; OTU, operational taxonomic unit.

^AZheng et al. (Reference Zheng, Zeng, Zhou, Liu, Fang, Xu and Fan2016); ^BJiang et al. (Reference Jiang, Ling, Zhang, Mao, Ma, Yin and Ruan2015); ^cKelly et al. (Reference Kelly, Borre, O’ Brien, Patterson, El Aidy, Deane and Dinan2016); ^DChung et al. (Reference Chung, Chen, Chou, Chen, Lee, Chuang and Kuo2019); ^EHuang et al. (Reference Huang, Shi, Li, Shen, Shi, Wang and Liang2018); ^FLai et al. (Reference Lai, Deng, Xu, Zhao, Xu, Liu and Rong2021); ^GLiu et al. (Reference Liu, Zhang, Wang, Wang, Zhang, Jiang and Duan2016); ^HZheng et al. (Reference Zheng, Yang, Li, Wu, Liang, Yin and Wang2020); ^IRong et al. (Reference Rong, Xie, Zhao, Lai, Wang, Xu and Liu2019); ^JLiu et al. (Reference Liu, Rowan-Nash, Sheehan, Walsh, Sanzari, Korry and Belenky2020).

For α-diversity in the BD studies (Table 3), three had inconsistent findings (Hu et al., Reference Hu, Li, Huang, Lai, Li, Sublette and Xu2019; Painold et al., Reference Painold, Mörkl, Kashofer, Halwachs, Dalkner, Bengesser and Reininghaus2018; Rong et al., Reference Rong, Xie, Zhao, Lai, Wang, Xu and Liu2019) and one found no differences (Zheng et al., Reference Zheng, Yang, Li, Wu, Liang, Yin and Wang2020). β-Diversity was different in two studies (Hu et al., Reference Hu, Li, Huang, Lai, Li, Sublette and Xu2019; Zheng et al., Reference Zheng, Yang, Li, Wu, Liang, Yin and Wang2020) and two found no differences (Painold et al., Reference Painold, Mörkl, Kashofer, Halwachs, Dalkner, Bengesser and Reininghaus2018; Rong et al., Reference Rong, Xie, Zhao, Lai, Wang, Xu and Liu2019).

Table 3. Studies of the GM in individuals with BD

HC, healthy controls; BD, bipolar disorder; PLS-DA, partial least-squares discriminant analysis; PCoA, principal coordinates analysis; HCL-32, Hypomania Check List-32; HAMD, Hamilton's Depression Scale; BDI, Beck Depression Inventory; MADRS, Montgomery–Åsberg Depression Rating Scale; HDRS-17, 17-item Hamilton Depression Rating Scale.

^THu et al. (Reference Hu, Li, Huang, Lai, Li, Sublette and Xu2019); ^UZheng et al. (Reference Zheng, Yang, Li, Wu, Liang, Yin and Wang2020); ^VPainold et al. (Reference Painold, Mörkl, Kashofer, Halwachs, Dalkner, Bengesser and Reininghaus2018); ^WRong et al. (Reference Rong, Xie, Zhao, Lai, Wang, Xu and Liu2019).

In SSD (Table 4), three studies reported lower α-diversity in the SSD groups compared to healthy controls (Ma et al., Reference Ma, Asif, Dai, He, Zheng, Wang and Chen2020; Xu et al., Reference Xu, Wu, Liang, He, Gu, Li and Wang2020; Zheng et al., Reference Zheng, Zeng, Liu, Chen, Pan, Han and Xie2019), four reported no differences (He et al., Reference He, Kosciolek, Tang, Zhou, Li, Ma and Chen2018; Li et al., Reference Li, Zhuo, Huang, Huang, Zhou, Xiong and Wu2020; Nguyen et al., Reference Nguyen, Kosciolek, Maldonado, Daly, Martin, McDonald and Jeste2018; Shen et al., Reference Shen, Xu, Li, Huang, Yuan, Wang and Liang2018), two did not report on α-diversity (Schwarz et al., Reference Schwarz, Maukonen, Hyytiäinen, Kieseppä, Orešič, Sabunciyan and Suvisaari2018; Yuan et al., Reference Yuan, Zhang, Wang, Liu, Li, Kumar and Song2018) and one reported higher α-diversity in the SSD group (Zhu et al., Reference Zhu, Ju, Wang, Wang, Guo, Ma and Ma2020b). β-Diversity was different between groups in five studies (He et al., Reference He, Kosciolek, Tang, Zhou, Li, Ma and Chen2018; Nguyen et al., Reference Nguyen, Hathaway, Kosciolek, Knight and Jeste2021; Shen et al., Reference Shen, Xu, Li, Huang, Yuan, Wang and Liang2018; Xu et al., Reference Xu, Wu, Liang, He, Gu, Li and Wang2020; Zheng et al., Reference Zheng, Zeng, Liu, Chen, Pan, Han and Xie2019), from those five, two studies reported tighter clustering in the healthy control group (Nguyen et al., Reference Nguyen, Kosciolek, Maldonado, Daly, Martin, McDonald and Jeste2018; Shen et al., Reference Shen, Xu, Li, Huang, Yuan, Wang and Liang2018).

Table 4. Studies of the GM in schizophrenia-spectrum disorder

HC, healthy controls; FPE, first psychotic episode patients; SSD, schizophrenia-spectrum disorder patients; FSCZ, antipsychotic-free first episode patients; TSCZ, chronically antipsychotic-treated patients; OTU, operational taxonomic unit; PLS-DA, partial least-squares discriminant analysis; NMDS, non-metric multidimensional scaling; PCoA, principal coordinates analysis; BPRS, Brief Psychiatric Rating Scale; PANSS, Positive and Negative Syndrome Score.

^KZheng et al. (Reference Zheng, Zeng, Liu, Chen, Pan, Han and Xie2019); ^LXu et al. (Reference Xu, Wu, Liang, He, Gu, Li and Wang2020); ^MNguyen et al. (Reference Nguyen, Hathaway, Kosciolek, Knight and Jeste2021); ^NShen et al. (Reference Shen, Xu, Li, Huang, Yuan, Wang and Liang2018); ^OHe et al. (Reference He, Kosciolek, Tang, Zhou, Li, Ma and Chen2018); ^PLi et al. (Reference Li, Zhuo, Huang, Huang, Zhou, Xiong and Wu2020); ^QYuan et al. (Reference Yuan, Zhang, Wang, Liu, Li, Kumar and Song2018); ^RZhu et al. (Reference Zhu, Ju, Wang, Wang, Guo, Ma and Ma2020b); ^SSchwarz et al. (Reference Schwarz, Maukonen, Hyytiäinen, Kieseppä, Orešič, Sabunciyan and Suvisaari2018); ^XMa et al. (Reference Ma, Asif, Dai, He, Zheng, Wang and Chen2020).

In the present review more studies reported inconsistent findings (n = 7) or no differences (n = 8) than studies who reported lower α-diversity in the psychiatric disorders (n = 5). These results are in line with other studies (Sanada et al., Reference Sanada, Nakajima, Kurokawa, Barceló-Soler, Ikuse, Hirata and Kishimoto2020; Simpson et al., Reference Simpson, Diaz-Arteche, Eliby, Schwartz, Simmons and Cowan2021), suggesting that host–microbe interactions are more complex than can be modelled by α-diversity.

Findings at different taxonomic levels

A large number of bacterial taxa were significantly different in their relative abundance between control and psychiatric groups [MDD groups (Table 2), BD groups (Table 3) and SSD groups (Table 4)]. Interestingly, multiple bacterial taxa abundances were similar between psychiatric disorders (Fig. 2). All investigations reported taxonomic differences between the neuropsychiatric disorders and healthy control groups; in the next paragraphs the most important differences, similarities and findings are discussed.

Fig. 2. Taxonomic differences in neuropsychiatric disorders (at the phylum, family and genus levels), whereby ‘↑’ = higher relative abundance in the neuropsychiatric disorder group, ‘↓’ = lower relative abundance in the neuropsychiatric group and ‘I’ = no differences in abundance. The letters below the arrows refer to the studies the information was retrieved from and also can be connected to the letters in the tables. Studies: ^AZheng et al. (Reference Zheng, Zeng, Zhou, Liu, Fang, Xu and Fan2016); ^BJiang et al. (Reference Jiang, Ling, Zhang, Mao, Ma, Yin and Ruan2015); ^CKelly et al. (Reference Kelly, Borre, O’ Brien, Patterson, El Aidy, Deane and Dinan2016); ^DChung et al. (Reference Chung, Chen, Chou, Chen, Lee, Chuang and Kuo2019); ^EHuang et al. (Reference Huang, Shi, Li, Shen, Shi, Wang and Liang2018); ^FLai et al. (Reference Lai, Deng, Xu, Zhao, Xu, Liu and Rong2021); ^GLiu et al. (Reference Liu, Zhang, Wang, Wang, Zhang, Jiang and Duan2016); ^HZheng et al. (Reference Zheng, Yang, Li, Wu, Liang, Yin and Wang2020); ^IRong et al. (Reference Rong, Xie, Zhao, Lai, Wang, Xu and Liu2019); ^JLiu et al. (Reference Liu, Rowan-Nash, Sheehan, Walsh, Sanzari, Korry and Belenky2020); ^KZheng et al. (Reference Zheng, Zeng, Liu, Chen, Pan, Han and Xie2019); ^LXu et al. (Reference Xu, Wu, Liang, He, Gu, Li and Wang2020); ^MNguyen et al. (Reference Nguyen, Kosciolek, Maldonado, Daly, Martin, McDonald and Jeste2018); ^NShen et al. (Reference Shen, Xu, Li, Huang, Yuan, Wang and Liang2018); ^OHe et al. (Reference He, Kosciolek, Tang, Zhou, Li, Ma and Chen2018); ^PLi et al. (Reference Li, Zhuo, Huang, Huang, Zhou, Xiong and Wu2020); ^QRong et al. (Reference Rong, Xie, Zhao, Lai, Wang, Xu and Liu2019); ^RPainold et al. (Reference Painold, Mörkl, Kashofer, Halwachs, Dalkner, Bengesser and Reininghaus2018); ^SSchwarz et al. (Reference Schwarz, Maukonen, Hyytiäinen, Kieseppä, Orešič, Sabunciyan and Suvisaari2018); ^THu et al. (Reference Hu, Li, Huang, Lai, Li, Sublette and Xu2019); ^UZheng et al. (Reference Zheng, Yang, Li, Wu, Liang, Yin and Wang2020); ^VPainold et al. (Reference Painold, Mörkl, Kashofer, Halwachs, Dalkner, Bengesser and Reininghaus2018); ^WRong et al. (Reference Rong, Xie, Zhao, Lai, Wang, Xu and Liu2019).

Probiotics

Probiotics are defined as live organisms that exert a health benefit when ingested in an adequate amount. Probiotics contain living beneficial bacteria, traditionally from genera Lactobacilli and Bifidobacteria.

In MDD, Kazemi et al. (Reference Kazemi, Noorbala, Azam, Eskandari and Djafarian2019) conducted an 8 week randomized, double-blind, placebo-controlled study, in which 110 depressed patients were randomly assigned to receive probiotic (L. helveticus and B. longum), prebiotic (galactooligosaccharide) or placebo treatment. Probiotic supplementation resulted in a significant decrease in symptom severity compared to both the prebiotic and the placebo groups. Additionally, the serum kynurenine/tryptophan ratio was significantly decreased in the probiotic group compared to the placebo group. Another randomized, double-blind, placebo-controlled 8 weeks probiotics study by Akkasheh et al. (Reference Akkasheh, Kashani-Poor, Tajabadi-Ebrahimi, Jafari, Akbari, Taghizadeh and Esmaillzadeh2016) studied a probiotic (Lactobacillus acidophilus, Lactobacillus casei and Bifidobacterium bifidum) in 40 MDD patients. As a result symptom severity decreased significantly in the probiotic group. In addition, serum insulin levels, insulin resistance and serum high-sensitivity C-reactive protein concentrations were decreased.

In BD, Dickerson et al. (Reference Dickerson, Adamos, Katsafanas, Khushalani, Origoni, Savage and Yolken2018) found that the adjunctive probiotic treatment (Lactobacillus rhamnosus strain GG and Bifidobacterium animalis subsp. lactis strain Bb12) for 24 weeks prevented rehospitalization in patients with acute mania (n = 66). Probiotic treatment also resulted in fewer days of hospitalization. Another study (n = 38) found no effects of probiotics compared to placebo on symptom severity of both depression and mania (Shahrbabaki et al., Reference Shahrbabaki, Sabouri, Sabahi, Barfeh, Divsalar, Esmailzadeh and Ahmadi2020). In another longitudinal cohort study 20 euthymic individuals with BD received a probiotic called ‘OMNi-BiOTiC^® Stress Repair’ (B. bifidum W23, B. lactis W51, B. lactis W52, L. acidophilus W22, L. casei W56, Lactobacillus paracasei W20, L. plantarum W62, L. salivarius W24, L. lactis W19) over a time period of 3 months (Reininghaus et al., Reference Reininghaus, Wetzlmair, Fellendorf, Platzer, Queissner, Birner and Dalkner2020). Cognition concerning attention and psychomotor processing speed was improved as well as executive functioning.

Three studies investigated the effect of probiotics in SSD patients. One studied the combination three bacteria (L. acidophilus, B. bifidum, L. reuteri and L. fermentum) and vitamin D in a randomized placebo-controlled trial (n = 60) (Ghaderi et al., Reference Ghaderi, Banafshe, Mirhosseini, Moradi, Karimi, Mehrzad and Asemi2019). The combination was associated with significant improvement in symptom severity and had beneficial effects on metabolic profiles in reducing fasting plasma glucose levels, insulin concentrations, triglycerides and total cholesterol levels. Another study investigated a probiotic (L. acidophilus, B. lactis, B. bifidum and B. longum) and selenium co-supplementation in a randomized placebo-controlled trial with 60 people with chronic schizophrenia (Jamilian & Ghaderi, Reference Jamilian and Ghaderi2021). The probiotic and selenium co-supplementation resulted in a significant improvement in symptom severity compared to the placebo group. However, in both studies it is uncertain which component of the intervention was responsible for the changes. Only one study studied the effect of a probiotic alone (Dickerson et al., Reference Dickerson, Stallings, Origoni, Katsafanas, Savage, Schweinfurth and Yolken2014). Their 65 participants were subjected to a double-blind adjunctive probiotic (L. rhamnosus strain GG and B. animalis subsp. lactis strain Bb12) or placebo therapy for 14 weeks. No significant differences were found in symptom severity between probiotic and placebo supplementation.

Prebiotics

Prebiotics are non-digestible fibres that are selectively metabolized by the intestinal microbiome. Prebiotics mostly include fructans and glucans. So far, only one study has studied the effects of prebiotics in SSD, namely, by using the purified prebiotic Bimuno™ galactooligosaccharides (B-GOS^®) formulation. Thirty-nine non-hospitalized participants with non-affective psychosis received either B-GOS^® or a placebo for 24 weeks. Participants performed the Brief Assessment of Cognition in Schizophrenia (BACS, Keefe et al., Reference Keefe, Goldberg, Harvey, Gold, Poe and Coughenour2004) and the B-GOS^® group had significantly improved global cognitive performance, compared to the placebo group (Kao et al., Reference Kao, Safarikova, Marquardt, Mullins, Lennox and Burnet2019). While interesting, this study requires replication in larger studies.

Faecal microbiota transplantation

Faecal transplants from depressed patients to germ-free rats show interesting results. Kelly et al. (Reference Kelly, Borre, O’ Brien, Patterson, El Aidy, Deane and Dinan2016) showed that FMT from depressed patients to microbiota-depleted rats can induce behavioural and physiological features characteristic of MDD in the receiving animals, including anhedonia and anxiety-like behaviour, as well as an increased plasma kynurenine/tryptophan ratio. In two studies, FMT from schizophrenia patients to germ-free mice led to hyperactivity and reduced anxiety in the open-field test (Zheng et al., Reference Zheng, Zeng, Liu, Chen, Pan, Han and Xie2019; Zhu et al., Reference Zhu, Guo, Wang, Ju, Wang, Ma and Ma2020a). Zhu et al. (Reference Zhu, Guo, Wang, Ju, Wang, Ma and Ma2020a) also found that mice that received FMT from schizophrenia patients displayed impaired spatial learning and memory. Remarkable, however, was the heightened amount of social interaction these mice had compared to the control mice. In human, FMT has shown initial positive effects on autism spectrum disorder patients (Kang et al., Reference Kang, Adams, Coleman, Pollard, Maldonado, McDonough-Means and Krajmalnik-Brown2019, Reference Kang, Adams, Gregory, Borody, Chittick, Fasano and Krajmalnik-Brown2017).

The most traditional and widely used probiotics (consisting of Bifidobacterium spp. and Lactobacillus spp.) are safe, yet they are not disease-specific. Although different Lactobacillus and Bifidobacterium strains were used in the probiotic studies mentioned above, most of the GM studies in the psychiatric disorders already showed heightened Lactobacillus and Bifidobacterium levels compared to their control groups. Rather, F. prausnitzii could be a promising target for therapeutic purpose in psychiatric patients (Verhoog et al., Reference Verhoog, Taneri, Díaz, Marques-Vidal, Troup, Bally and Muka2019). Research in rats showed preventive and therapeutic effects of F. prausnitzii on chronic unpredictable mild stress-induced depression-like and anxiety-like behaviour (Hao, Wang, Guo, & Liu, Reference Hao, Wang, Guo and Liu2019). Supplementation of prebiotics consisting of inulin-oligofructose or inulin-type fructans and fructo-oligosaccharides have been demonstrated to increase F. prausnitzii levels (Dewulf et al., Reference Dewulf, Cani, Claus, Fuentes, Puylaert, Neyrinck and Delzenne2013; Hustoft et al., Reference Hustoft, Hausken, Ystad, Valeur, Brokstad, Hatlebakk and Lied2017; Ramirez-Fariaz et al., Reference Ramirez-Fariaz, Slezak, Fuller, Duncan, Holtrop and Louis2009), although this is not a consistent finding (Majid, Cole, Emery, & Whelan, Reference Majid, Cole, Emery and Whelan2014).