A correlation study of intestinal microflora and first‐episode depression in Chinese patients and healthy volunteers

Abstract Objective This research examines the intestinal‐associated flora of patients with depression compared with healthy volunteers to identify the characteristics and differences of flora associated with depression. It provides a theoretical basis for the prevention and treatment of depression through intestinal micro‐ecological regulation. Methods We recruited 30 patients with depression to participate in the patient group (PG), and 30 volunteers were recruited for the healthy control group (HG) from the Beijing Hui‐long‐guan Hospital. Thereafter, the 16S rRNA high‐throughput sequencing method, using the Hamilton Depression Scale, was applied to analyze patient and healthy groups. Results PG and HG microflora were different regarding phylum, Family, Genus, and Order. The results showed that Barnesiella was the dominant flora in depression patients, while Lachnospiraceae and Alloprevotella were the dominant bacteria in healthy participants. The proportion of Betaproteobateria (Proteobacteria), Alcaligenaceae (proinflammatory), Peptostreptococcaceae, Catenibacterium, Romboutsia, Sutterella, and Burkholderiales in the anxiety‐negative depressed group was significantly higher than in the anxiety‐positive group; and the proportion of Anaerostipes (inflammation) and Faecalibacterium (anti‐inflammatory) bacteria was significantly lower than that of patients with anxiety. Conclusion Results showed there were differences in intestinal micro‐ecology between patients with depression and healthy volunteers. We found that the level of inflammation‐related bacteria in anxiety‐positive patients was lower than that in anxiety‐negative patients. These results enrich the knowledge of relationships between depression and intestinal flora and provide a theoretical basis for probiotics to assist in the treatment of depression.


| INTRODUC TI ON
The intestinal flora is a dynamic and complex microbial ecosystem which contains a number of genes magnitudes higher than humans. The adults have a weight of about 1 kg (Dinan et al., 2015), including bacteria, fungi, and viruses (Collins et al., 2009). At present, more and more evidence demonstrates that there is an intricate interaction between the host and the gut flora at almost all levels, from direct communication between cells to a wide range of signals among different systems, which involves various organs and body systems, including the central nervous system (Arumugam et al., 2011;Stilling et al., 2014).
Intestinal flora affects emotion and cognition through the gutbrain axis (Cryan & Dinan, 2012;Heijtz et al., 2011). The gut-brain axis is a two-way neurohumoral conduction system. It contains a series of bidirectional pathways, including the hypothalamic-pituitary-adrenal axis (HPA), the vagus nerve, immune pathways, and metabolism (Bruce-Keller et al., 2017;Dinan & Cryan, 2017). These complementary pathways promote interaction between the gut flora and the brain.
First, the intestinal flora can produce many neuroactive compounds (Dinan et al., 2013). For example, gamma-aminobutyric acid (GABA) is secreted by lactic acid bacteria and Bifidobacteria, Candida, Streptococcus, and Escherichia coli flora all secrete 5-hydroxytryptamine (5-HT) in the intestinal (Barrett et al., 2012;Schousboe & Waagepetersen, 2007), and Bacillus and Escherichia coli secrete norepinephrine (NE) (Roshchina, 2010). Second, the vagus nerve plays an important role in signaling between the brain and the intestine. In addition, the immune system strengthens the relationship between the intestinal flora and the brain (Dinan et al., 2015).

Levels of various inflammatory biomarkers have been shown to be
elevated in depression patients (Raison & Miller, 2011). Additionally, infections of food-borne campylobacter jejuni increased c-reactive protein levels in brain regions associated with autonomic function and increased depression-like behavior in mice. The dysfunction of the intestine-brain axis may cause physiological and pathological consequences (Mayer, 2011), and the abnormal function of the endocrine system plays an important role in the development of depression. Related studies show that (Sudo, 2014) intestinal flora can affect development and behavioral regulation. Therefore, the intestinal flora is a key node in the gut-brain axis and can provide a new target for the treatment of depression (Dinan & Cryan, 2017). This study aims to explore the changes of intestinal flora in patients with depression, to better understand its cause and development; second, to improve depression treatment by regulating intestinal flora to achieve a better prognosis, and try to prevent recurrence.

| Participants
This study is a controlled trial that was conducted at Beijing Hui-Long-Guan Hospital to explore changes in the intestinal flora of patients with depression. From November 2017 to February 2018, we included 30 outpatients and inpatients with depression for the patient group (PG) and 30 healthy participants (HG) for the control group.
Before inclusion in the study, we assessed the physical condition of the participants. The patients with depression were assessed using the Hamilton 24-item depression scale and screened for the inclusion criteria in Table 1. We divided the experimental group into groups 1 and 2 according to the reference standard. See the results for details (Tables 2 and 3). Before the assessment scale, we conducted a scale consistency training for the researchers.

| Evaluation tool
The Hamilton Anxiety Rating Scale (HAMA) is one of the earliest commonly used scales in psychiatric clinical practice, including 14 projects. It is often used clinically for the diagnosis and classification of anxiety disorders. The Hamilton Depression Rating Scale (HAMD) is the most commonly used scale for clinically assessed depression.
This study used the 24-item version of the scale to evaluate the severity of the condition and the treatment effect. The assessment criteria are based on relevant information. Before conducting the measurement scale, we trained the researchers on consistency of the scale.

| Sample collection and sequencing
Approximately 2 g of a stool sample was collected from PG; it was placed immediately in a stool collection tube with 2 ml of preservation solution and frozen in a refrigerator at −80°C and standby application; the samples for HG were collected under the same conditions. Then, the 16S rRNA high-throughput sequencing technology was used to analyze the differences in flora between the PC and HC groups (Claesson et al., 2011;Dethlefsen et al., 2008).

| Bioinformatic analysis
The raw fastq files were demultiplexed based on the barcode. PE reads for all samples were run through Trimmomatic (version 0.35) to remove low quality base pairs using these parameters (SLIDING

| Ethical statement
The study has been approved by the Ethics Committee of Hui-Long-Guan Clinical Medical College, Peking University. All participants signed an informed consent form and a complete and comprehensive introduction for them or their Family which included the purpose, procedure, and possible risks of the study. Volunteers had the right to withdraw from the study at any time.

| RE SULT
The HAMD scores in the patient group and the division of the experimental group according to anxiety scores have been outlined in Tables 2 and 3, respectively.

| Demographic data
The data represented by mean value, standard deviation, and comparison among groups, using independent sample t test, have been outlined in Table 4.

| Sequencing data
After 16S rRNA sequencing all sixty samples, a total of 1,777,341 raw gene sequences were selected. Then, we obtained 1,496,472 high-quality gene sequences after optimization, with an average of 24,532 per sample. To decrease the quantity of gene sequences and prevent the sequence diversity being overestimated, we clustered the high comparability sequences into one OTU. After clustering, a total of 477 units of OTU were obtained. There were no significant differences between the two groups.
Mothur software was used to draw a rarefaction curve to compare the richness of the flora in the sample (Figure 1). The rarefaction curve is a curve in which a certain number of individuals were randomly selected from the overall sample, and the number of Species represented by these individuals was counted and constructed by the number of individuals and Species. When the curve tends to be flat, this indicates that the sequencing is reasonable.

| Abundance difference analysis
Microbial diversity is studied in community ecology.

| Differential analysis of intestinal microecology
We found that the sequence of fecal bacteria mainly belongs to four phyla, Bacteroidetes ( Due to the complicated Species relationship between the two groups, the detailed results are shown in Figure 2 because the accuracy of the data is extremely low for bacterial groups with the relative abundance of less than 0.1% in MiSeq analysis. Therefore, we summarized the flora with abundance greater than 0.1%, as shown in Table 5.

| The LDA Effect Size analysis of community differences between groups
LEfSe is a software for discovering high-dimensional biomarkers and revealing genomic characteristics, which includes genes, metabolism, and classification; it is used to distinguish two or more bio-

| Abundance difference analysis
According to the above principle of abundance analysis, we found that after the alpha statistical analysis on the PG, each index of PG 2 was larger than PG 1, but there was no statistical significant difference (p > .05).

| Based on metastas analysis
The metastas method was used to analyze the two groups at the Class, Family, Genus, and Order levels. We concluded that Betaproteobateria (2.4805% vs. 5.0373%, p = .007) had significant differences between the two groups in Class. F I G U R E 1 Rarefaction curve in the sample F I G U R E 2 Differences at the level of Class, Family, Genus, Order, and Species; the abscissa is the Species of bacteria, and the ordinate is mean ± SD. * means it is relevant In Order, Burkholderiales (2.4805% vs. 5.0373%, p = .02) and Thermoanaerobacterales (<0.1% vs. <0.1%, p = .03) had significant difference between two groups.
The Species results are not listed here; the detailed results are shown in Figure 4. As above mentioned, because the accuracy of the data is extremely low for bacterial groups with the relative abundance of less than 0.1% in MiSeq analysis, we summarized the flora with abundance greater than 0.1%, as shown in Table 6.

| Based on LEfSe analysis
As can be seen in Figure 5, based on the LDA SCORE, we found that the dominant bacterial communities were obtained between the two groups according to the HAMA anxiety SCORE: Alcaligenaceae, Burkholderiales, Betaproteobacteria, and Facecalibacterium in the two groups.

| Differences between PG and HG
Through sequencing data and an alpha diversity analysis, we found that there was no statistical difference in the abundance of each group. The results were the same as those observed by Naseribafrouei et al. (2014), but different from those of Jiang et al. (2015). Now most scholars believe that the higher the diversity of intestinal flora, the better the health (Matsuoka & Kanai, 2015). people. Therefore, the specific mechanism of microbial diversity in depressed patients still needs to be further explored. In Jiang's study (Jiang et al., 2015), all the participants recruited were <40 years old and had no history of hypertension. In this study, we limited the age to 65 years old but did not count evaluate hypertension history. In addition, there were also large differences in regional dietary habits, which may also be the reason for difference in the results.
In this study, the imbalance of intestinal flora is mainly reflected by changes in five levels: Class, Family, Genus, Order, and Species. Compared with healthy individuals, the differences are mainly reflected in Rikenellaceae, Alistipes (Alistipes belongs to the Rikenellaceae Family) in PG, there was a significant difference between the PG and the HG in the bacterial group with a relative abundance of 0.1% or more at least one group (Table 5), the abundance of which were increased significantly. Alistipes is an indole-positive bacterium, which can break down tryptophan to produce indole, and affects human tryptophan metabolism. Since tryptophan is a precursor of 5-HT, the rise of Alistipes may affect the intestine tryptophan metabolism, which affects brain neurotransmitter signaling (Song et al., 2006). On the other hand, some studies indicate that the abundance of Alistipes is detected in patients with chronic fatigue syndrome, and chronic fatigue syndrome is often accompanied by the development of anxiety and depression. Therefore, the effect of Alistipes on 5-HT metabolism in patients with depression is worth studying.
The abundance of Lachnospiraceae, Succinivibrionaceae, Alloprevotella, Anaerostipes, Aeromonadales, and Succinivibrio, which belong to the thick-walled bacteria with a relative abundance of 0.1% or more at least one group, was significantly reduced (Table 5). Prior studies (Wong et al., 2016) have pointed out that cysteine-1 inhibitors may have antidepressant effects. Studies have found that the changes of Lachnospiracea abundance are consistent with changes of the microbiome in cysteine-1-deficient mice. In addition, some studies (Duncan et al., 2007) suggest that Lachnospiraceae is involved in the metabolism of short-chain fatty acids (SCFAs). SCFAs are an important source of intestinal epithelial cell energy, which can also affect the permeability of intestinal epithelial cells and various biochemical reactions (Vince et al., 1990;Wong et al., 2006). The abundance of Lachnospiraceae may affect intestinal permeability, which may induce neurotoxic substances, neurotransmitters, etc., and affect the brain's function through the vagus nerve more easily. In addition, SCFAs can also speed up the secretion of 5-HT and accelerate gastrointestinal motility. Therefore, we suspect that the decrease of Lachnospiraceae abundance may also be related to gastric motility disorders in patients with depression.
By LEfSe software analysis, we found that Barneslella (inflammatory) was the dominant flora in PG, which is different from the results of Jiang's study (Jiang's results were Porphyromonadaceae, Alistipes (inflammation)), but it may be related to inflammatory bacteria. In HG, Lachnospiraceae and Alloprevotella are the dominant flora. The abundance of Lachnospiraceae in depression patients is reduced. We also demonstrated the importance of Lachnospiraceae in the human body.

| Difference of bacteria group in PG by HAMA classification
Because depression was accompanied by anxiety, we classified the depression group into anxiety group and nonanxiety group according to the HAMA score and as the PG1 and PG2. According to alpha diversity analysis, there was no significant difference in bacterial diversity and abundance between PG1 and PG2.
However, the metastas analysis revealed that the proportion of Betaproteobateria (Proteobacteria), Alcaligenaceae (Inflammatory), Peptostreptococcaceae, Catenibacterium, Sutterella, Burkholderiales in PG2 (which was significantly higher than PG1), and Anaerostipes, Anaerostipes-hadrus (inflammatory), and Faecalibacterium was significantly lower than in PG1 (Table 6). We found that the levels of inflammatory bacteria with anxiety symptoms in depression were significantly lower than those without anxiety. In addition, we also found that Faecalibacterium was higher in patients with anxiety.
Existing research shows that Faecalibacterium has a strong antiinflammatory effect and the lack of Faecalibacterium can lead to the onset of Crohn's disease (Barrett et al., 2012;Sokol et al., 2008).
When this bacterium was transplanted in animals, it was found to be able to fight against colitis, and if combined with human immune cells, it could produce anti-inflammatory effects (Barrett et al., 2012;Sokol et al., 2008).
In addition, by LEfSe analysis, the dominant groups of PG2 were Alcaligenaceae, Burkholderiales, Betaproteobacteria, and the PG1 was Facecalibacterium. Therefore, by means of grouping, it was found that the level of inflammation in patients without anxiety symptoms was higher than in patients with anxiety symptoms; the microflora varied between the two groups.
In summary, although we found a large number of changes in bacterial abundance in the study, a large amount of literature confirmed that not only did inflammatory factors increase, but also inflammatory responses increased in patients with depression. However, we found that there was neither an increase nor decrease in the antiinflammatory flora in our study. It is speculated that age may not be strictly controlled and segmented, and cardiovascular disease may not be strictly controlled in participants. In addition, the samples might not be strictly aseptically handled when collected, and there may be a risk of sample contamination.

F I G U R E 3
LEfSe tree graph and LDA fraction distribution. Onion parody of cluster tree, red and green areas represent different groups, the branches in the red nodes play an important role in the red group of microbial groups, green nodes play an important role in the green groups said the microbial groups, yellow nodes indicate microbial groups that have not played a significant role in two groups. Names of Species indicated by letters in the picture. This circle radiates from the inside out to the cluster tree, representing the classification levels of Phylum, Class, Order, Family, and Genus in turn. Each small circle on the different circle layers represents a classification at that level, and the diameter of the small circle is proportional to the relative abundance of the classification. The figure on the right is the LDA score obtained by LDA analysis (linear regression analysis) for the microbial groups with significant effects in the two groups The variability of the two groups at different levels; the abscissa is the Species of bacteria, and the ordinate is mean ± SD

| Limitations
This study is a cross-sectional study, which only evaluates the difference between the two groups, and there are constraints, such as the culturability of the bacteria, which have not further verified the importance of flora we selected. More experimental data are still needed. The score scale also has a certain degree of concealment during the assessment process. In addition, we also found that the influencing factors were relatively high using PCA measurement (PC1 37.67%, PC2 20.18%). Therefore, it is necessary to further improve the clinical observation indicators and related influential factors in subsequent studies to improve the accuracy of the experiment.

| CON CLUS I ON
In this study, a comprehensive analysis of the intestinal microecology of patients with depression was performed by highthroughput sequencing of 16sRNA. It was found that there was a significant difference in intestinal micro-ecology between patients and healthy volunteers, which was manifested in the proportion of pathogenic bacteria and Alistipes, which increased, Lachnospiraceae and other beneficial bacteria were significantly reduced. The analysis revealed that Barnesiella was the dominant group in PG, and Lachnospiraceae and Alloprevotella were the dominant group in HG. This study uses clinical samples of first-episode depression, and its findings provide theoretical and practical basis for the prevention and treatment of depression, and the regulation of bacterial flora (such as Alistipes and Lachnospiraceae found in the article) to cure or prevent depression in the future, which has a strong practical significance for the occurrence and development of clinical treatment of depression. In addition, we found that the patients with depression with anxiety symptoms and those without anxiety symptoms also have large differences in flora, which will be more useful for our clinical judgment of accompanying symptoms and it will be helpful for accompanying symptoms treatment. The experimental results have enriched the correlation between depression and intestinal flora.

CO N FLI C T S O F I NTE R E S T S
All authors report no conflicts of interest. F I G U R E 5 Distribution of dominant microflora in LDA SCORE SCORE (see Figure 3 for annotations)

AUTH O R CO NTR I B UTI O N
SJZ, YQW, and ZRW conceived the original idea for this study. The study design was planned by SJZ, ZRW, and FDY and with support from WDW and QZ. SJZ and WDW prepared the manuscript with repeated revisions commented on and amended by ZRW and YBZ. All authors were involved in the interpretation of the results. We would like to thank Dr.
ZR Wang and FD Yang for his guidance in this paper. We would also like to thank Editage (www.edita ge.cn) for English language editing.

Pe e r Rev iew
The peer review history for this article is available at https://publo ns.com/publo n/10.1002/brb3.2036.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data were made available to all interested researchers upon request.