The link between changes in the gut microbiome and MDD has been supported by several studies [14, 17]. Here, we showed that these changes could reflect the severity of MDD. In this study, we found that the gut microbiota of moderate and severe MDD patients were characterized by the enrichment of Bacteroidetes, while Ruminococcus and Eubacterium were depleted in the severe patients. Consistently, the major enterotype of HCs was Faecalibacterium while which of MDD subgroups were Bac and Bac 2. In addition, we also identified a microbial marker panel which is capable of distinguishing MDD patients with different severity.
It is generally believed that high diversity of gut microbiome is a sign of healthy status [31, 32]. Some studies have found that the gut microbial diversity reduced in depression, bipolar disorder and schizophrenia [33]. In addition, reduced diversity has been associated with disease severity and higher risk of death in patients with diseases such as bronchiectasis [34], cystic fibrosis [35] and ulcerative colitis [36]. Consistently, we found that the simpson index decreased only in the moderate and severe MDD groups at the genus level. In addition, the Venn diagram showed that the number of shared species between moderate and severe is higher than that of other groups. Bar plot showed that Bacteroides were significantly enriched in moderate and severe subgroups, while there was no change in the mild group. These findings suggested that the gut microbial composition remains relatively stable during the early stages of MDD; but with increase in disease severity, the disturbances of gut microbiota become inevitable. Furthermore, based on the LDA effect size analysis, the differentially expressed microbiota in the MDD subgroups were identified. We constructed the co-occurrence network of perturbed microbiota of MDD with mild to severe. We found that the depletion of Blautia and Eubacterium were common features of MDD patients; enriched Bacteroides were characteristic of moderate and severe MDD. Consistently, our previous study proved that increase of Bacteroides was a signature of MDD [17]. Bacteroides was reported involved in immune system maturation, tumor formation and activation of autoimmune disease by affecting T cell’s function and promote cognitive impairment disease pathologies through activating microglia [37,38,39,40]. Bacteroides faecis helps maintaining the epithelial barrier integrity and increasing the gut IgA level to reduce inflammatory bowel disease [41, 42]. Because of related closely, Bacteroides dorei and Bacteroides vulgatus are often researched as a group, they were been reported playing an important role in brown adipose tissue metabolism and suppressing proinflammatory immune responses [43, 44]. Similarly, researched revealed that Bacteroides uniformis and Bacteroides eggerthii were involved in pathological mechanism of obesity, metabolism, and colitis [45,46,47]. In some studies, Blautia have been reported as potential probiotics that can induce anti-inflammatory peripheral immune response, alleviate obesity-related disease and regulate metabolism through cross-feeding with other bacteria [48,49,50]. Also, it has been reported that Eubacterium can produce short-chain fatty acids, which play an important role in regulating cell metabolism, immune and endocrine response [51, 52]. In addition, unlike mild and moderate patients, the highly concentrated clusters in severe MDD were dominated by decreased 7 Ruminococcus, 8 Eubacterium, 5 Clostridium and 7 Clostridium. Depletion of these potential probiotics may contribute to the development of depression. Compared with patients with mild and moderate depression, the severe individuals need more active physical and drug therapy. Therefore, combining a probiotic intervention strategy with conventional treatment will be helpful in promoting the improvement of both disease recovery and quality of life. In addition, we found that K21572 (susD) and K12373 (HEXA) may be key node connecting Blautia and Bacteroides, showing completely opposite correlation with these two microbial clusters. susD is an outer membrane protein. It is the main starch binding protein on the surface of Bacteroides, and can effectively use polysaccharides as a source of carbon and energy [53]. Further work characterizing the interaction between Blautia and Bacteroides could elucidate its role in MDD severity. K12373 (HEXA) is a hexose kinas, play an important role in sugar metabolism in Bacteroides fragilis [54]. Likewise, K03088 (rpoE) is an important part of transcription activation factor that binds to RNA polymerase complex to regulate gene expression in bacteria [55]. In sight of the positive association of K03088 and Bacteroides, the increased expression of K03088 might indicate the function enhancement of Bacteroides.
Motivated by the results that showed that there were alterations in the gut microbiota in moderate and severe MDD, we constructed a random forest model with 37 bacteria species. The AUC value of the classification of MDD subgroups were 0.992 to 0.998; suggesting a high diagnostic value. Overall, this finding provided evidence that gut microbiota-targeted biomarkers may become potential non-invasive tools for MDD stratification.
Additionally, we initially explored the gender bias in the structure of gut microbiota. The prevalence of MDD in women was twice of in men [56, 57], we wondered if there is a potential relationship between gender preference of disease and gut microbiota. Generally, we found that the changes of microbiota were similar in female and male. Previous study declared that Bacteroides and Prevotella had higher abundance in male [58]. In our study, male group owned more differentially enriched bacteria and most of them were belonged Bacteroides. Further research was required to figure out if this was related to gender differences in MDD. In order to obtain consolidation evidence, it is necessary to further expand the study cohort to determine the gender bias of gut microbiota in MDD.
The limitations of this study are: (i) The sample size of the three MDD subgroups is relatively small, and the samples were collected from a clinical center, therefore regional variations cannot be ruled out; (ii) All patients were not under medication; therefore, follow-up studies are required to explore whether this biomarker panel can be used to monitor treatment response; (iii) Given that fecal bacteria transplantation could transfer depressive phenotypes from humans to mice, it will be important to determine the correlation between the disturbance of gut microbiota and disease severity in animal models, and uncover the underlying mechanisms.
Taken together, we analyzed and found the unique and common alterations in gut microbiota across different disease severities. Microbiota may affect physiological functions through mutual synergy or antagonism, and the status may shift from balance to imbalance as symptoms get worse. Furthermore, we identified a novel combined biomarker that could discriminate different severity subgroups with high accuracy. In conclusion, our study provides a new direction for understanding the progression of MDD, and a potential promising strategy for developing a novel method for objectively assessing the severity of MDD.