Effect of Gut Microbiotas Diversity During 32-39 Weeks of Gestation on Postpartum Depression

Background: Patients with major depression are accompanied by intestinal ora occulation; however, the relationship between the composition of gut microbiota in pregnancy and postpartum depression (PPD) has not been established. In this study we determined the effect of the gut microbiota in pregnant women during 32-39 weeks of gestation on PPD. Methods: Participants (n = 74) were enrolled between 2016–2017 from the Guangzhou Women and Children’s Medical Centre (GWCMC). Stool samples were collected during 32-39 weeks of gestation, and the relative abundance of fecal microbiota was characterized by 16S rRNA sequencing. The parturients completed the mainland Chinese version of the Edinburgh Postnatal Depression Scale (EPDS) 42 days postpartum to detect PPD. The linear discriminant analysis (LDA) effect size (LEfSe) method was used to identify bacterial population differences between the PPD and control groups. Results: The top three bacteria phyla in the PPD and control groups were Firmicutes, Bacteroidetes, and Actinobacteria. Compared with healthy pregnant women, the alpha diversity index of the PPD group was lower. Beta diversity analysis was performed by PCoA showing that no signicant differences in bacterial community structures between the two groups (R 2 = 0.013, P = 0.549). The composition of gut microbiota during 32-39 weeks of gestation of the two groups was different. At the genera level, Acinetobacter, Plesiomonas, Enterococcus, Olsenella, Alloscardovia, and Anaerotruncus were increased in the PPD group, while Lactococcus, Adlercreutzia, Clostridium, Coprococcus, and unclassied-Clostridiales were decreased. At the species level, hypermegale, uli, casseliavus, and hathewayi were increased in the PPD group, and celatum was increased in the control group. Conclusions: During 32-39 weeks of gestation, a reduction in diversity of gut microbiota and anti-inammatory bacteria, and an increase in opportunistic pathogenic bacteria are more likely to cause PPD.


Introduction
Postpartum depression (PPD) refers to a non-psychotic depressive episode lasting > 2 weeks after delivery. Women with PPD were more likely to cry, more irritable, more emotionally unstable, attention decreased, sleep disturbances, loss of appetite, and lack of interest in recreational activities than usual. Suicidal thoughts are extremely common, affecting about 20% of women with PPD symptoms [1][2][3][4][5]. PPD is one of the most common complications of childbirth and a major public health problem. The high prevalence rate is ranging from 6.9-12.9% in high-income countries to more than 20% in some low-or middle-income countries [6]. When untreated, it has the potential for a profound negative impact on mothers, children, and families. Case identi cation and accurate diagnosis are important which needs global attention.
There are 10 13 -10 14 microbes in the human gut, which is thought to be 10 times the number of human cells and 150 times the size of the human genome [7]. Indeed, the gut microbiota is considered to be another "organ" in human beings [7]. There is a two-way communication network between the gut microbiota and the brain, which is referred to as the "brain-enteric-axis." Changes in the composition and number of gut microbiotas can disturb the "brain-enteric-axis," which has an impact on the central nervous system and is closely related to neuropsychiatric diseases, including anxiety, cognitive decline, autism spectrum disorder, schizophrenia, bipolar disorder, and depression [8]. Previous studies have revealed different bacterial populations, including Blautia, Clostridium, Faecalibacterium, Ruminococcus Prevotella, and Roseburia, between people with depression and the general population [9]. Of note, the research subjects with depression have mostly involved patients with major depression, and little attention has been paid to patients with PPD. Because the course and physiological state of PPD and severe depression are different, we cannot directly infer the relationship between PPD and gut microbiota from the study results involving major depression. The interrelationships between PPD, intestinal ora, and stress during pregnancy, hormone levels, and immune regulation indicate an association between PPD and gut microbiota [10]. Supplementation of probiotics during pregnancy can reduce the PPD score, further con rming the association between intestinal bacteria and PPD [11]. Indeed, the association between gut microbiota composition and PPD has not been established. In this study we demonstrated that the gut microbiota is associated with postpartum depression which will be helpful to the case identi cation and accurate diagnosis.

Study design and participants
Between July 2016 and October 2017, a prospective cohort study including 14 women with PPD and 60 gestational healthy control women was performed in Guangzhou Women and Children's Medical Centre (GWCMC), a large modern city located in Guangdong Province in South China. The pregnant women were recruited into the study at the third trimester, and followed-up to 42 days postpartum. The exclusion criteria included participants who had taken antibiotic treatment or probiotic supplements in the four weeks prior to sample collection, strict vegetarians, individuals with alcoholism or with other unusual dietary habits or with diseases, such as gestational diabetes mellitus, gestational hypertension, thyroid disease, gastroenteritis, and major mental illnesses.
The information regarding the regular prenatal examinations, basic personal information, past medical history, psychiatric history, pregnancy history, family history, co-existing conditions during pregnancy, and other clinical data were entered by trained healthcare workers in the medical records. The protocol for this study was approved by the Ethics Review Committees of Guangzhou Medical University, and all participants provided voluntary signed informed consent. All procedures performed in studies involving human participants were in accordance with the requirements of the ethics committee and with the 1964 Helsinki Declaration.

Stool sample collection
Stool samples were collected from all participants during the third trimester. The collection time of fecal samples from the 74 subjects included in this study was 32-39 weeks gestation. Stool samples were self-collected by the participants. Brie y, the subjects were instructed on how to self-collect the samples, and all materials were provided in a convenient specimen collection kit. Participants used polypropylene spoons (three scoops of approximately 10 g) to collect stool samples at home or in the hospital, transfer the samples to sterile sampling containers, and immediately stored at 4 °C. The specimens were transported to the laboratory within 12 h of collection at a refrigerated temperature. Containers were immediately stored at − 80 °C. The 16S rRNA sequencing analysis Quantitative Insights into Microbial Ecology (QIIME2) was used to process 16S rRNA amplicon sequences. All reads were truncated at the 150th base with a median Q score > 20 to avoid sequencing errors at the end of the reads. Noisy sequences, chimeric sequences, and singletons in the sequence data were removed using DATA. Denoised paired-end reads were joined, setting a maximum mismatch parameter of two bases. The representative sequences (i.e., the features) were de ned as 100% similarmerged sequences. We have used the term "operational taxonomic unit" (OTU) instead of "feature" in the current study for convenience. Then, the taxonomy of the features was identi ed using the classifysklearn classi cation method based on the Greengenes 13.8 database (https://data.qiime2.org/2018.11/common/gg-13-8-99-515-806-nb-classi er.qza) via the q2-featureclassi er plugin. The phylogenetic analysis was performed in QIIME2 with "qiime alignment mafft," "qiime alignment mask," and "qiime phylogeny fasttree" commands, based on the tutorials at https://docs.qiime2.org/2019.1/tutorials/moving-pictures/. The phylogenetic tree of the core OTUs was visualized using iTOL v4. To measure the gut microbiota diversity and quantify the taxonomic composition of the samples, all samples were rare ed to an even sampling depth of 20,000 sequences.

Edinburgh Postnatal Depression Scale (EPDS)
The participants completed the mainland Chinese version of the EPDS on postpartum day 42, which has been validated for Chinese women in other studies [11]. The EPDS is a 10-item self-reporting instrument for evaluating depressive symptoms over the past 7 days. Participants rated the severity or frequency of each item based on four levels, where 0 indicates the most favorable condition and 3 indicates the least favorable condition, for each item. The total score ranged from 0-30. A score ≥ 10 was diagnosed as postpartum depression [12].

Statistical methods
Descriptive statistics are presented as the mean ± standard deviation (SD) for continuous variables and as a frequency (percentage) for categorical variables. Anthropometric parameters between the two groups were compared using a t-test/chi square test. The stacked graph was drawn using the ggplot2 package in the R statistical program. The alpha diversity index (a measure of richness and evenness) was calculated using the vegan package in the R statistical program. The Wilcoxon test was used to compare the alpha diversity index between the two groups. Beta diversity analysis was performed by principal coordinates analysis (PCoA) based on Bray-Curtis distances at the OTU level. The analysis of similarities (Adonis) based on Bray-Curtis distances was conducted to compare different groups. Microorganism features distinguishing fecal microbiota between the PPD and non-PPD groups were identi ed using the linear discriminant analysis (LDA) effect size (LEfSe) method (http://huttenhower.sph.harvard.edu/lefse/). LEfSe used the Kruskal-Wallis rank-sum test with a normalized relative abundance matrix to detect features with signi cantly different abundances between assigned taxa and performs LDA to estimate the effect size of each feature. An alpha signi cance level of 0.05 and an effect-size threshold of 2 were used for all biomarkers. All analyses were conducted in R statistical package (version 3.5.1) and gures were produced using RStudio. All tests for signi cance were two-sided, and a P < 0.05 was considered to indicate statistical signi cance.

Demographic characteristics
This study involved 74 participants, including 14 participants with PPD and 60 healthy participants. The mean ages of the PPD and control groups were 29.29 ± 4.30 and 30.78 ± 4.60 years, respectively. The average height and weight before pregnancy of the PPD group was 159.79 ± 7.20 cm and 51.89 ± 8.33 kg, respectively. The height of the control group was less than the PPD group (158.57 ± 5.42 cm), but the weight before pregnancy was greater (52.19 ± 7.03 kg). The pre-pregnancy body mass index (BMIs) of the PPD and the control group were 20.30 ± 2.78 and 20.79 ± 2.89 kg/m 2 , respectively. The cesarean section rates in the PPD and control groups were 35.71% and 35.00%, respectively. Additional general demographic information is shown in Table 1.  Fig. 1.
Alpha diversity analysis of the gut microbiota during 32-39 weeks of gestation between the PPD and control groups Alpha (α) diversity was used to evaluate the variety of species in the samples, which re ects OTU richness and evenness using several different indices. The alpha diversity index of the PPD group was lower than the control group (Supplementa Table 1 Beta diversity analysis of the gut microbiota during 32-39 weeks of gestation between the PPD and control groups The similarity of the bacterial community structures among the two groups was evaluated by PCoA (Fig. 3). NO clear segregation of gut microbiota was observed between samples of PPD group and control group. Adonis analysis also demonstrated no signi cant distinction between PPD and control groups (R 2 = 0.013, P = 0.549). This nding indicated that PPD patients and healthy controls have similar gut bacterial community structures during 32-39 weeks of gestation. PCoA1 explained 9.835% of the variation observed, and PCoA2 explained 9.576% of the variation.
Altered gut microbiota composition during 32-39 weeks of gestation between the PPD and control groups To determine different taxa between the non-PPD and PPD groups, the LEfSe algorithm on the Galaxy browser was used. LEfSe detected 31 bacterial taxonomic clades showing statistically signi cant differences (16 increased and 15 decreased) in the PPD group compared to the control group ( Fig. 4a and  b). At the class level, Bacilli were more abundant in the control group. At the order level, Pseudomonadales were more abundant in the PPD group, while Lactobacillales were more abundant in the control group. At the family level, Moraxellaceae were increased in the PPD group, whereas Clostridiaceae and unclassi ed-Clostridiales were decreased. Among the different predominant genera, Acinetobacter, Plesiomonas, Enterococcus, Olsenella, Alloscardovia, and Anaerotruncus were more abundant in the PPD group compared to the non-PPD group. Lactococcus, Adlercreutzia, Clostridium, Coprococcus, and unclassi ed-Clostridiales were more abundant in the non-PPD group compared to the PPD group. Among the identi ed species, compared with the non-PPD group, the abundance of more species (hypermegale, uli, casseli avus, and hathewayi) were increased in the PPD group, while celatum were decreased.

Discussion
In this study we determined the composition of intestinal micro ora in gravidas (who subsequently developed PPD) during 32-39 weeks of gestation. The results showed that the alpha diversity index and the anti-in ammatory bacteria of gut microbiota in gravidas (who subsequently developed PPD) were decreased during 32-39 weeks of gestation compared with healthy pregnant women.
Based on the literature involving human clinical microbiota [9], we identi ed very few studies with consistent results for depression. In our study patients with PPD had a lower alpha diversity compared with the normal population. Similar to our results, Huang [13] reported that the alpha diversity indices of major depressive disorder were lower than healthy controls. In other studies [14,15], different results have been presented. Jiang [14] found that patients with active depression had a higher alpha diversity index compared with the normal population. Naseribafrouei [15] did not detect any signi cant differences in the alpha diversity index.
Acinetobacter is a genus of Moraxellaceae, both of which belong to Pseudomonadales. We observed that all three were increased in PPD. Acinetobacter is a conditionally pathogenic bacteria that is usually associated with respiratory tract infections, accounting for approximately 33.7% of the pathogens causing respiratory tract infections [16]. The researchers also found that the abundance of Acinetobacter is elevated in the gut of patients with neuroin ammatory diseases, such as multiple sclerosis [17]. This association is closely related to the fact that Acinetobacter can induce a proin ammatory response in human peripheral blood mononuclear cells [17]. In a murine study the abundance of Enterococcus was increased in a depression model induced by chronic unpredicted mild stress [18]. This suggests that stress may induce depression by increasing the abundance of Enterococcus. Alloscardovia is a newly discovered genus of Bi dobacterium. The link between Alloscardovia and disease has not been established, but high concentrations of Alloscardovia have been detected in the feces of patients with intrahepatic cholangiocarcinoma, and Alloscardovia may be involved in the metabolism of bile acids [19]. Previous studies have shown that patients with Parkinson disease have a higher abundance of Anaerotruncus in their intestines than healthy individuals [20]. In addition, Anaerotruncus is enriched in the feces of the patients with gestational diabetes mellitus, and negatively correlated with insulin sensitivity [21]. This nding suggests that Anaerotruncus may play a negative role in disease.
Like our results, a reduction in the anti-in ammatory gut microbiota was observed in depressed patients, including Clostridium [22]. Two other studies arrived at an opposite conclusion, with higher levels of clostridia in patients with major depression [14,23]. The difference in observations may be due to diet. Clostridium can metabolize carbohydrates to produce short-chain fatty acids (SCFAs), and when the intestinal protein is rich, Clostridium will metabolize proteins to produce harmful substances [9]. Patients with major depression may have relatively rich protein remaining in the intestine due to poor appetite, causing Clostridium to grow [9]. We did not include diet in the current study, which was a limitation. In a murine study it has been reported that Clostridium butyricum can modulate in ammatory factors and microglial activation to prevent depression-like behavior [24], which also supports our ndings. Butyrate-producing Coprococcus bacteria are consistently associated with higher quality of life indicators. Coprococcus spp. are also depleted in depression, even after correcting for the confounding effects of anti-depressants [25]. Adlercreutzia is an equol-producing bacterium isolated from human feces that contains a single species (Adlercreutzia equolifaciens) [26]. Equol attenuates microglial activation and potentiates neuroprotection in vitro [27]. Animal studies have also shown that equol is bene cial in mitigating depression and anxiety disorders [28]. Lactobacillus and Lactococcus are two common genera of Lactobacillus. Lactobacillus can reduce oxidative stress markers and in ammatory cytokines in the brain and serum to prevent depression [11,29]. Certain species of Lactococcus can improve depression and anxiety through antioxidant effects [30]. Among the identi ed species, Clostridium hathewayi can be used as a biomarker for colorectal adenoma and cancer [31], and Enterococcus casseli avus is a relatively rare enterococcus that causes human infectious diseases, including bacteremia, endocarditis, and meningitis [32]. No studies have found a direct link between Clostridium hathewayi, casseli avu and depression. Among patients with Behcet's disease, Megamonas hypermegale may synthesize SCFAs in the intestine and reduce the concentration of SCFAs. These changes will affect the function of the immune system, and thus affect the nervous system [33].
The mechanism underlying the brain-gut axis consists of the interaction of the nervous, endocrine, and immune systems [34]. The in ammation caused by the disorder of intestinal ora and the neuroendocrine hormones produced work together on the intestinal wall, changing the intestinal permeability, and interact with the central system through the vagus nerve. Metabolites produced by intestinal ora (such as SCFAs and tryptophan) can interact with the immune system, changing the body's immune state and thus changing the brain's behavior and mood. In addition, an imbalance of neurotransmitters and neuropeptides produced by gut bacteria, such as para-aminobutyrate (GABA), serotonin, and brain-derived neurotrophic factor (BDNF), which act as nerve signaling messengers, can also affect the central nervous system [34].

Conclusions
During 32-39 weeks of gestation, the reduction of the diversity of gut microbiota and anti-in ammatory bacteria and the increase in opportunistic pathogenic bacteria were more likely to cause PPD. This nding provides the basis for further exploring the relationship between intestinal ora and PPD.  Figure 1 Composition of bacterial phyla in gravidas during 32-39 weeks of gestation (PPD and control groups). (a)

Figures
The composition of the phylum abundance for each sample. (b) The composition of the phylum abundance for the two groups.
Pielou index between the PPD and control groups; (f): comparison of the Richness index between the PPD and control groups; (h): comparison of the Shannon index between the PPD and control groups; (i): comparison of the Simpson index between the PPD and control groups.

Figure 3
PCoA based on Bray-Curtis distances at the OTU level. Each sample is represented by a dot. Circles in different colors represent different groups. Figure 4