This study demonstrates, for the first time, that the gut microbiome is compositionally and functionally altered in people with social anxiety disorder (SAD) compared with healthy controls. Moreover, it increases the growing evidence linking social brain function and the microbiome [27]. Firstly, we show that beta diversity, an indicator of overall microbiota composition, was significantly different between the two groups. The relative abundance of three genera, Anaeromassilibacillus, Gordonibacter and Parasutterella, and two corresponding species, Anaeromassilibacillus sp An250 and Parasutterella excrementihominis differed significantly between SAD patients and controls. Additionally, functional differences were evident with the microbiome of SAD patients enriched for the gut metabolic pathway, aspartate degradation I.
Strikingly, we found Anaeromassilibacillus sp An250 to be present in almost half of our SAD group but in only one healthy control. Anaeromassilibacillus is a newly-discovered genus, which was first isolated in 2017 from the faecal sample of a 1-yo Senegalese patient with kwashiorkor [61]. Several strains of Anaeromassilibacillus, including sp An250 have since been identified from the caecal microbiome of chickens [62, 63]. Anaeromassilibacillus is a member of the Clostridiales order and Clostridiaceae family of bacteria [61], taxonomic groups which are increased in the gut microbiota of patients with autism spectrum disorder (ASD) [64], depression [65] and schizophrenia [66]. Conversely, various genera from the Clostridiales order were found to be reduced in the faecal microbiota of people with GAD [21] although Clostridiales was positively correlated with anxiety scores in a study analysing serum microbial DNA composition in patients with mood disorders [67]. Despite such inconsistencies, significant shifts in the abundance of Clostridiales taxa appears to be common to many psychiatric disorders and may represent disease-shared microbial responses [68]. Furthermore, preclinical studies suggest a link between Clostridiales and social behaviour. In a recent study, mice subjected to early life social isolation stress showed a significantly increased abundance of Clostridiales. These mice subsequently demonstrated reductions in sociability and social novelty behaviours, which negatively correlated with Clostridiales abundance [69]. In another study, mice exposed to a social stressor had increased relative abundance of the genus Clostridium [70]. However, it is difficult to extrapolate findings from animal studies to humans [71] especially with regards to a process as complex as social behaviour.
Given the very recent addition of Anaeromassilibacillus to human microbiome databases, there is little in the existing literature about its role in human health and disease. It was one of several genera found to be enriched in untreated patients with MDD compared to those receiving antidepressant treatment, suggesting that it could be altered by psychotropic medication or be an indicator of treatment response [72]. Additionally, the relative abundance of Anaeromassilibacillus reduced significantly in the faeces of children with ASD after guar gum prebiotic supplementation, which was associated with reduced irritability and improved constipation [73], again suggesting that reduction of this genus may be associated with improved psychopathology. Thus, although the literature is sparse, Anaeromassilibacillus appears be of relevance in ASD and depression, psychiatric conditions which are highly comorbid with SAD [74, 75] and which may involve shared pathophysiological processes. Of note, we did not see a difference in the relative abundance of Anaeromassilibacillus in medicated, compared to, unmedicated SAD patients; although given the small sample size, this should be interpreted with caution.
Gordonibacter is another genus about which relatively little is known. It is a member of the Eggerthellaceae family and Coriobacteriia class [76] and is notable in its ability to produce urolithins from the metabolism of polyphenols [77], which may have an impact on mental health [78]. Parasutterella has been more extensively studied. It is a member of the Sutterellaceae family and in humans, is largely represented by a single species, Parasutterella excrementihominis [79]. Similar to our finding of lower Parasutterella levels in SAD, members of this genus have also been found to be reduced in ASD [80]. Weight and dietary factors appear to be important influences. Parasutterella is negatively associated with BMI and waist circumference [81] and conversely, can be induced by high sugar [82] and high-fat diets [83]. Our SAD group had elevated sugar intake and did not differ in terms of fat intake but, although the group difference for Parasutterella excrementihominis remained after adjusting for BMI, increased BMI in the SAD group could contribute to its reduced abundance in SAD patients.
It is difficult to interpret the importance and relevance of specific bacterial taxa differences in a patient group. The gut microbiome is a highly complex and dynamic ecosystem where microbes continuously interact with, and impact, one another and the host [84]. Attempts are underway to characterise microbial community structures and gain insights into the many complex microbe-microbe and host-microbe networks and interactions [85, 86]. Some human gut microbial groups appear to be highly influential and exert a metabolic impact on a substantial number of other microbial entities, so-called ‘network influencers’ [85]. None of our differentially expressed genera or species have been reported as being such core taxa or ‘influencers’ and it is unclear what these over- and under-represented taxa mean in the overall context of the gut microbial environment of SAD patients. With this in mind, exploring microbial function may offer deeper insights than relying on composition alone in an ever-changing ecosystem.
Using GBMs and GMMs, which are microbiome-related functional pathways that have been manually curated from existing literature [32, 33], we identified one functional pathway that was enriched in the SAD group – aspartate degradation I. According to MetCyc, a comprehensive reference database of metabolic pathways and enzymes [87], this pathway involves the conversion of the amino acid, L-aspartate to the corresponding keto acid, oxaloacetate, by the enzyme, aspartate aminotransferase (AspAT). Several bacteria and archaea have demonstrated this enzymatic ability including Haloferax mediterranei [88], Pseudoalteromonas translucida TAC125 [89], Saccharolobus solfataricus [90] and Escherichia coli [91, 92]. Interestingly, bacterial AspAT enzyme activity may represent a link between gut microbiome function and the tryptophan-kynurenine pathway, a key physiological system in psychiatric disorders. There are significant interactions between tryptophan metabolism and the MGB axis [93,94,95,96] and the gut microbiome may influence host diet selection behaviour by mediating the availability of essential amino acids such as tryptophan [97]. Tryptophan metabolism involves the downstream conversion of kynurenine to kynurenic acid (KYNA) by the enzyme kynurenine aminotransferase (KAT). KYNA is an important neuroactive substance, which is elevated by chronic stress in animal models [98, 99] as well as in psychiatric conditions such as schizophrenia [100, 101] and SAD [102]. Notably, KAT activity has been detected in E. coli cells in vitro, and authors suggested that the source of KYNA detected in the rat small intestine could be the gut bacteria [103]. This bacterial KAT enzyme protein has been identified as being identical to the bacterial AspAT enzyme [104] and thus, an elevation in the ‘aspartate degradation I’ functional pathway may represent increased KAT, as well as AspAT potential, by the microbiome. While currently a hypothetical supposition, it is possible that the elevated peripheral KYNA which we previously reported in SAD patients [102] may be linked with the key functional difference seen in the microbiome of this group. In support of this hypothesis is the fact that D-cycloserine, an orally-administered, broad-spectrum antibiotic, has been found to enhance the treatment response to exposure therapy for SAD [105, 106], an effect which could plausibly be related to its ability to inhibit KAT activity and lower KYNA [107].
This is, to our knowledge, the first study to investigate the composition and function of the gut microbiome in patients with SAD and has several notable strengths. Firstly, our sample consisted of carefully selected patients with a pre-existing clinical diagnosis of SAD who had sought treatment from a mental health professional. Secondly, we used a whole genome shotgun sequencing method, providing information on the functional capacity of the microbiome, as well as offering greater resolution of bacterial species identification, than the more commonly used 16 S rRNA gene sequencing [108, 109]. Thirdly, we took into account many of the important host variables known to confound gut microbiota studies in human disease [110]. Stool quality is a particularly strong source of gut microbiota variance [110, 111] which has often been neglected in psychiatric microbiota studies. Stool consistency, as measured by the BSC, was matched between our groups, as were other important variables, including age, sex, birth delivery mode, smoking status and alcohol. Our groups were not matched in terms of BMI and exercise levels, variables which may be of relevance to the gut microbiome [112, 113]. Although adjusting for these variables did not affect group differences, it would, of course, be preferable to have samples with equivalent BMI and exercise scores. Additionally, we collected detailed dietary information, which has often been lacking in studies of the microbiota in psychiatric conditions. Our groups were well-matched in terms of overall dietary intake. The only difference seen was in relation to carbohydrate consumption, driven by higher sugar intake in the SAD group, and this was adjusted for in our statistical analyses.
Study limitations include the small sample size and the single-time point nature of the study, which prevents the establishment of any causal relationships. Additionally, two thirds of our SAD patients were taking psychotropic medication, which may have had an impact on microbiota composition [114, 115]. The majority of our medicated patients were taking an SSRI antidepressant, escitalopram being the most commonly prescribed. Escitalopram has antibacterial activity against some gut commensal strains in vitro [116, 117] although this effect did not translate to an in vivo animal model [117]. Other prescribed SSRIs in our patient group included fluoxetine, citalopram, sertraline, paroxetine and vortioxetine, all of which have shown varying levels of antibacterial activity in vitro [117,118,119,120], with in vivo evidence available for fluoxetine [121,122,123]. The SNRI venlafaxine, conversely, does not appear to impact common gut commensals in vitro [116, 117], although an influence on the microbial richness and on the abundance of certain genera were seen in a mouse model [122]. Thus, the translatability of studies using isolated in-vitro strains to animal models is unclear, with even more uncertainty in relation to their applicability to the human gut microbiome. Limited human data in relation to the effect of antidepressants on the microbiome is available. A small study of 17 depressed patients commenced on escitalopram, found no significant differences in beta-diversity or in any taxa levels between pre-treatment and 6-week post-treatment time-points, although increased alpha diversity was evident [124]. Furthermore, a longitudinal study of 40 patients with depression and/or anxiety revealed no difference in beta diversity between those taking, and not taking, antidepressant medications and no change in alpha diversity in antidepressant-treated patients between baseline and endpoint timepoints. Antipsychotic medications, conversely, did appear to exert an effect on the gut microbiome [125], consistent with previous findings [126]. Two of our patients were prescribed low-dose Quetiapine, a second-generation antipsychotic which thus may have had an impact.
Aside from impacting microbiota composition, it is also of course possible that psychotropic medications could have influenced functional pathways. A recent study demonstrated that oral intake of fluoxetine or amitriptyline by rats exposed to chronic unpredictable mild stress resulted in alterations in KEGG metabolic pathways, particularly those pathways concerning carbohydrate metabolism, membrane transport, and signal transduction [127]. However, no such alterations in KEGG pathways were seen in a longitudinal follow-up of psychiatric patients taking antipsychotics, antidepressants and/or anxiolytic medications [125]. In an approach similar to ours, these authors also chose to analyse GBMs in the psychiatric group, although they specifically investigated only 6 of the 56 available GBMs, namely those involving GABA and tryptophan synthesis or degradation. They found alterations in certain GBMs in patients taking antipsychotics and antidepressants but not in those taking anxiolytic medications. All in all, although there is clear evidence that many antidepressants have antibacterial effects, this evidence is based primarily on in-vitro and animal studies, and the impact of these medications on the human gut microbiome structure and function remain largely unknown. Although we did not find any differences in the relative abundance of any taxa between medicated and unmedicated patients, we cannot rule out a potential influence.
Finally, some of our SAD patient group had a past history of depression and/or a comorbid anxiety disorder. However, patients with a current depressive episode were excluded and in all, the primary diagnosis was SAD with any comorbid anxiety disorder representing a secondary diagnosis. Although we did not find a difference in the relative abundance of any taxa in those SAD patients with or without a past history of MDD, this must be interpreted with caution given the small numbers of such sub-groups. While it is not possible to disentangle the currently reported observations from the past psychiatric history of study participants, this was a clinically representative sample and we believe that including such patients makes our findings more generalizable considering the significant overlap between depression and anxiety disorders. Given the paucity of studies exploring the gut microbiome in any clinical anxiety disorders, our findings, despite the limitations, are important in generating a foundational base for larger, prospective and interventional microbiome studies in these highly prevalent and disabling psychiatric conditions. Additionally, future preclinical studies and secondary validation experiments would offer a complementary approach to confirm the presence and role of these differentially expressed bacterial taxa and functional pathways. Earlier compositional microbiota studies in depression [60, 128] and ASD [129, 130] have paved the way for a variety of successful interventional studies utilising probiotics [25], dietary change [131] and faecal microbiota transplantation (FMT) [132] in these conditions. It is hopeful that microbiota-based therapeutic interventions may also be realised for patients with clinical anxiety disorders. Indeed, a previous cross-sectional study in university students has suggested that consumption of fermented foods may be protective against the development of social anxiety [31].
In conclusion, the gut microbiome of patients with SAD differs in composition and function to that of healthy controls, raising the possibility that the MGB axis may represent a biomarker reservoir and potential therapeutic target for this early-onset, chronic disorder. Further preclinical studies focussing on mechanistic pathways and larger, longitudinal studies in SAD patients are needed to validate our preliminary results, understand the clinical implications (if any) and investigate the impact of psychotropic medication and treatment on the gut microbiome in SAD.