Analysis of 16S data
Accounting for 70% of the valid sequences, we obtained 1,620,000 high-quality sequences from 54 fecal samples of all participants: the HC group contains 630,000 sequences, the MDD group contains 690,000 sequences, and the GAD group contains 300,000 sequences. In particular, we obtained 10,996 species-level OTUs from the HC group, 14,406 OTUs from the MDD group, and 15,010 OTUs from the GAD group (Table 2). Venn analysis results show that 5069 OTUs were common to all three groups (Fig. 1).
Table 2 Comparison of phylotype coverage and diversity estimation of 16S rRNA gene libraries at 97% similarity from sequencing analysis
Group
|
Number of reads
|
Number of OTUs†
|
Coverage (%)‡
|
Richness estimator
|
Diversity index
|
ACE
|
95% CI
|
Chao
|
95% CI
|
Shannon
|
Simpson
|
Evenness§
|
HC
|
630000
|
10996
|
97.19
|
4257.84
|
4031.74–4506.03
|
3090.89
|
2840.24–3395.07
|
4.733331714
|
0.032098
|
0.354556683
|
MDD
|
690000
|
14406
|
97.47
|
3762.74
|
3557.59–3989.13
|
2790.49
|
2562.74–3069.66
|
4.647938217
|
0.036133
|
0.34454694
|
GAD
|
300000
|
15010
|
96.15
|
5978.93
|
5689.91–6291.70
|
4170.2
|
3857.99–4538.97
|
5.017976
|
0.025108
|
0.397936241
|
† The operational taxonomic units (OTUs) were defined with 97% similarity level.
‡ The coverage percentage, richness estimators (ACE and Chao), and diversity indices (Shannon and Simpson) were calculated using Good’s method and the mothur package, respectively.
§ The Shannon index of evenness was calculated with the formula E = H/ln(S), where H is the Shannon diversity index and S is the total number of sequences in that group.
The richness of gut bacterial communities in all three groups was estimated by ACE and Chao, and the diversity was estimated by using the Shannon diversity index and Simpson diversity index. ACE and Chao analysis showed that most of the gut microbial diversity in each sample had been captured with the current sequencing depth. After rarefying the sequencing depth among all of the samples using a bootstrap method (30,000 reads per sample), Shannon diversity index and Simpson diversity index estimates were calculated. There was no significant difference in richness and diversity between HC and MDD in this study. However, GAD showed significant difference in richness and diversity of the microbiota compared to HC (Fig. 2).
Analysis of fecal bacterial community
We analyzed the gut bacterial composition and relative abundance at the phylum level. Compared with the HC group, we found a considerable increase in the relative abundance of Proteobacteria and Actinobacteria in MDD and a considerable decrease in the relative abundance of Bacteroidetes. Compared with the HC group, the relative abundance of Fusobacteria, Tenericutes, and Verrucomicrobia increased and that of Firmicutes decreased in GAD; there was no obvious change in phylum levels. Compared with the MDD group, we found an increase in the relative abundance of Fusobacteria, Tenericutes, Verrucomicrobia, and Bacteroidetes, but a decrease in the relative abundance of Proteobacteria, Actinobacteria, and Firmicutes in the GAD group (Fig. 3a). At the family level, compared with the HC group, we found that the counts of some bacteria, such as Desulfovibrionaceae, were considerably reduced in the GAD and MDD patients. However, levels of bacteria, such as Enterobacteriaceae, were considerably higher in MDD patients, but showed no change in GAD patients. The levels of nine species, including Fusobacteriaceae, did not change in MDD patients, but were higher in GAD patients (Fig. 3b). At the genus level, compared with the HC group, we found that the levels of Megamonas were remarkably reduced in GAD and MDD patients. The counts of bacteria, such as Faecalibacterium, were higher in MDD patients, but showed no change in GAD patients. Compared with the HC group, the levels of Bacteroides did not change in MDD patients, but increased in GAD patients (Fig. 3c).
In summary, we found that patients with MDD or GAD showed considerable changes in the gut microbiota, and there were differences in the relative abundance of gut microbiota in patients with MDD and GAD.
Analysis of the differences in gut microbiota
Analysis of the differences among the three groups at the genus level revealed significant differences in the gut microbiota (P<0.05) (Fig. 4a). Compared with HC individuals, we found that levels of Butyricimonas, Megamonas, Fusicatenibacter, Sutterella, Coprococcus_3, Bergeyell , Bilophila, Acinetobacter, Ruminococcus_2, and Rothia were significantly different in MDD patients. To identify key significant differences between MDD and HC individuals, we analyzed the metagenome data by using the LEfSe method (Fig. 4b). We found that Butyricimonas, Megamonas, Fusicatenibacter, Sutterella, Coprococcus_3, Bergeyella, and Bilophila were abundant in HC, and Acinetobacter, Ruminococcus_2, and Rothia were abundant in MDD. Additionally, we found that Butyricimonas, Christensenellaceae_R_7_group, Megamonas, Sutterella, Fusicatenibacter, Haemophilus, Coprococcus_3, and Lachnospira were abundant in the HC, and Lactobacillus, Hungatella, Coprobacter and Flavonifractor were abundant in the GAD (Fig. 4c). Of note, we found that levels of Faecalibacterium, Sutterella, some Bacteroides, and some Carnobacterium showed clear differences between MDD and GAD patients. To identify key significant differences between MDD and GAD, we used the LEfSe method to analyze the metagenome data (Fig. 4d). We found that Sutterella, Burkholderiaceae, and betaproteobacteriales were abundant in GAD patients, whereas Faecalibacterium, Klebsiella, Ruminococcus, Rothia, and Subdoligranulum were abundant in MDD patients. These were the dominant phylotypes that contributed to the differences between the intestinal microbiota of GAD and MDD patients.
Relationship between gut microbiota and clinical parameters
We evaluated correlations among the relative abundance of bacterial genera, the hormones (including PTC, ACTH, FT3, FT4, TT3, TT4, and TSH), the total and factor scores of HAM-D (Hopelessness, Sleep disturbance, Block, Diurnal/variation, Cognitive impairment, Weight and Anxiety/somatic) in the MDD group and the total and factor scores of HAM-A (Psychic anxiety and Somatic anxiety) in the GAD group. Compared with the GAD group, we found significant differences at the genus level forFusicatenibacter, LachnospiraceaeND3007group, Coprococcus3, Faecalibacterium, norank_Firmicutes, ChristensenellaceaeR7group, Gutmetagenome, and norank_Proteobacteria in MDD patients (Fig .5a). We observed that ChristensenellaceaeR7group negatively correlated with the total score and Hopelessness factor of HAM-D and ACTH (P<0.05, Fig. 5b, c, d), Coprococcus3 and LachnospiraceaeND3007group negatively correlated with PTC (P<0.05, Fig. 5e g), LachnospiraceaeND3007group positively correlated with Sleep disturbance factor of HAM-D (P<0.05, Fig. 5f), Fusicatenibacter negatively correlated with FT4 (P<0.05, Fig. 5h), and other key phylotypes showed no strong correlation. Fusicatenibacter, LachnospiraceaeND3007group, Coprococcus3, Faecalibacterium, norank_Firmicutes, ChristensenellaceaeR7group, Gutmetagenome, and norank_Proteobacteria showed significant interindividual variability in GAD (Fig. 5i). In the GAD group, we found that ChristensenellaceaeR7group, LachnospiraceaeND3007group, and Fusicatenibacter negatively correlated with PTC (P<0.05, Fig. 5j, k, l), whereas other key phylotypes showed no strong correlation.