Gut Microbiota Composition and Functional Prediction in Recurrent Spontaneous Abortion


 Objective The changes of microbial community in pregnant women, let alone those of patients with recurrent spontaneous abortion (RSA), remain unclear. We analyzed the differences of gut mircobiota (GM) between RSA patients and pregnant women to find the possible mechanism of RSA. MethodsWe enrolled 30 RSA patients (RSA group) and 30 pregnant women who terminated their pregnancy and did not have a history of spontaneous abortion (NR group) in our hospital from June 2020 to August 2020, and fecal samples were obtained to analyze the GM using 16S rDNA V3–V4 sequencing.ResultsAt the phylum level, we found that there is no significant difference in composition of GM between RSA and NR. But at the genus level, compared with NR, Roseburia significantly decreased (P<0.01), and Ruminococcus significantly increased in RSA patients (P<0.05). Further analysis indicated that Klebsiella (P<0.05) was significantly increased, Prevotella.9 (P<0.05) and Roseburia (P<0.05) were significantly decreased in RSA2 group (BMI>23.9 in RSA). Moreover, Agathobacter (P<0.01) was significantly increased in NR2 group (no delivery in NR). Functional prediction indicated that GM may interfere with RSA through membrane transport, carbohydrate metabolism, amino acid metabolism and other pathways.ConclusionDecreased Roseburia in GM of pregnant women maybe related to RSA. Our results provide the basis for in-depth studies of the composition of gut microbial communities in RSA.

In view of the important in uence of microbiota on the mother and fetus, some scholars have studied the microbiota in natural cavities of pregnant women, such as the oral cavity, vagina, uterus and gut, to observe their in uence on the pregnancy outcome. Recently, based on 16SrDNA sequencing, several studies have shown that the structure and stability of maternal microbiota changed during pregnancy, for example, Lactobacillus in the maternal vagina and intestinal tract have a potentially important impact on fetal development and may affect the outcome of pregnancy. But the results of these studies varied widely, which possibly due to the differences in race, diet, antibiotic use and education level [8][9][10][11][12]. The research on the gut microbiota (GM) of pregnant women focuses on the aspects of pregnant women with diabetes mellitus, pre-eclampsia and [13][14].
Recurrent spontaneous abortion (RSA) is a common gynecological disease in China, de ned by two or more failed pregnancies, its etiology remains unclear, genetics, autoimmune abnormalities, endocrine, anatomy, or pre-thrombotic state are thought to be related to its pathogenesis [15,16]. Previous proteomic study had shown that there is a signi cant changes occurred in the pathway of Fc gamma R-mediated phagocytosis in RSA, this result indicated that autoimmunity is invloved in its pathogenesis [17]. A study suggested that RSA may be caused by an imbalance of vaginal ora (especially that of Pseudomonas) in RSA patients [18]. In this study, we focused on the relationship of maternal GM and RSA. So far, no such study has been reported.
Through this study, we want to explain the following questions: 1) Whether there is a difference in GM between RSA patients and normal abortion patients, whether there is statistical signi cance; 2) Is the maternal gut microbiota related to RSA, and if so, which bacteria play an important role? 3) How these bacteria invloved in mechanisms of RSA and whether it is related to body mass index (BMI).

Study design
From June 2020 to August 2020, in the case group (RSA), we recruited 30 RSA patients who met the diagnostic criteria of Royal College of Obstetricians and Gynecologists (RCOG) consensus [16]. Meanwhile, in the control group (NR), we recruited 30 women who terminated their pregnancy and did not have a history of spontaneous abortion. Exclusion criteria for all subjects included: 1) Complicate with chromosomal abnormalities, anatomical abnormalities, surgery and trauma, etc; 2) taking antibiotics, probiotics, or other treatments, within 4 weeks; 3) Complicate with endocrine diseases, infectious diseases, immune diseases; 4) any psychiatric comorbidity; 5) excessive physical exercise.
Characteristics of participants were summarized in Table 1.

Sample collection and DNA Extraction
A single fecal sample was collected by each participant at home, and immediately stored at -20℃, then transferred to -80℃ for longer-term storage. Fecal bacteria genomic DNA was extracted with Cetyltrimethyl Ammonium Bromide (CTAB)/ Sodium dodecylsulfate (SDS).

16S rDNA gene sequencing
The following steps and methods refer to the implementation method of Lundberg DS [19].
Amplicon Generation: Primer: 16S V3-V4: 341F-806R. 16S rDNA genes were ampli ed used the speci c primer with the barcode. All PCR reactions were carried out in 30μL reactions with 15μL of Phusion®High-Fidelity PCR Master Mix (New England Biolabs).

PCR Products treatment:
PCR products were mixed with same volume of 1X loading buffer (contained SYB green), electrophoresis were operated on 2% agarose gel fordetection, mixture PCR products were puri ed with AxyPrepDNA Gel Extraction Kit (AXYGEN).

Library preparation and sequencing:
Following manufacturer's recommendations, sequencing libraries were generated using NEB Next®Ultra™DNA Library Prep Kit for Illumina (NEB, USA) and index codes were added. The quality of library was assessed on the Qubit@ 2.0 Fluorometer (Thermo Scienti c) and Bioanalyzer 2100 system (Agilent). Then, paired-End (PE) amplicon library was constructed using a TruSeq® DNA PCR-Free Sample Preparaion Kit (Illumina, US) and quanti ed by Qubit, then sequencing was performed using the Illumina Hiseq platform (APTBIO Technology, Shanghai, China).

Data and bioinformatics analysis
The following steps and methods refer to the implementation methods of Lundberg DS [19] and Avershina E [20].
Paired-end reads assemblies: Paired-end reads from the original DNA fragments were merged using FLASH (http://ccb.jhu.edu/software/FLASH/), Quality tering on the raw tags was performed under specifc tering conditions to obtain high-quality. clean tags according to the Fast QC (V 0.10.1).

Page 5/22
OTU cluster and Species annotation: Chimeric sequences were tered using Useach sofware (10 version). Sequences with 97% similarity were assigned to the same operational taxonomic units (OTUs) by Usearch (10 version). Representative sequences were chosen for each OTU, and taxonomic data were then assigned to each representative sequence using the RDP (Ribosomal Database Project) classifer. Sequences were processed with the sofware package of the QIIME (V1.9.0) toolkit. Taxonomy-based analyses were conducted by classifying each sequence using the SILVA database (https://www.arb-silva.de/). In order to compute Alpha Diversity, we rarify the OTU table and calculate seven indexes: Ace, Chao1, Goods coverage, Observed Species, PD whole tree, Shannon, Simpson.
Phylogenics distance and community distribution: Graphical representation of the relative abundance of bacterial diversity from phylum to species can be visualized using Krona chart. Cluster analysis was preceded by principal component analysis (PCA), which was applied to reduce the dimension of the original variables using the QIIME (Version 1.9.1) software package. QIIME calculates both weighted and unweighted unifrac distance, which are phylogenetic measures of beta diversity. We used unweighted unifrac distance for Principal Coordinate Analysis (PCoA) and Unweighted Pair Group Method with Arithmetic mean (UPGMA) Clustering.

Statistical analysis:
STAMP software was utilized to con rm differences in the abundances of individual taxonomy between the two groups. LDA Effect Size (LEfSe) was used for the quantitative analysis of biomarkers within two groups. To identify differences of microbial communities between the two groups, ANOSIM and ADONIS were performed based on the Bray-Curtis dissimilarity distance matrices.

The number of OTUs
The number of common OTUs between the patients of RSA and NR was 23672. Meanwhile, the number of proper OTUs in the patients of RSA was 17847, and that in NR was 43885, as shown in Table 1. Table 1 Characteristics and OTUs of two groups

Characterization of fecal microbiota
The phylum and genus level were choosed to show a histogram of relative abundance of species. As shown in Figure 1, at the phylum level, we found that there is no signi cant difference in composition of GM between RSA and NR. But at the genus level, the results were shown in Figure 2 that compared with NR, Roseburia signi cantly decreased (P<0.01), and Ruminococcus signi cantly increased in RSA patients (P<0.05).
The normal group was divided into two groups according to whether to give birth or not, those who did not give birth were included in NR2. According to BMI, uterine and ovarian lesions, patients with intrauterine adhesion, polycystic ovary syndrome (PCOS), and BMI > 23.9 among RSA patients were classi ed as RSA1 group, and the remaining patients were classi ed as RSA2 group. We analyzed the relative abundance of different groups again. As shown in Figure 3, the results indicated that at the genus level, Klebsiella (P<0.05) were signi cantly increased, Prevotella.9 (P<0.05) and Roseburia (P<0.05) were signi cantly decreased in RSA1 group. Moreover, Agathobacter was signi cantly increased in NR2 group (P<0.01).
NR1 was pregnant women who had given birth, NR2 was pregnant women who had not given birth, RSA1 was RSA patients with intrauterine adhesion, PCOS, and BMI > 23.9, RSA2 was other RSA patients.

Alpha diversity
Alpha diversity is used to analyze the microbial community diversity within samples or groups. Observed species is the number of OUT, Shannon and Simpson were the bacterial diversity index, Chao1 and ACE are the bacterial abundance index. Coverage was the depth index of sequencing. PD whole tree is the phylogenetic diversity index. Comparison of alpha diversity difference indexes between RSA and NR was shown in Table 2. The results of difference analysis of alpha diversity between two groups were represented by a boxplot (Figure 4). Beta diversity difference index The t-test was used to analyze whether the difference of species diversity among groups was signi cant. Based on Weighted Unifrac and Unweighted Unifrac distances, the results of difference analysis of beta diversity between two groups were represented by a boxplot (The upper two gures in Figure 5). In order to study the similarity of community composition, we performed Principal Co-ordinates Analysis (PCoA) based on weighted unifrac distance and unweighted unifrac distance (The lower two gures in Figure 5).
The closer the sample distance is, the more similar the species composition.

Functional prediction
Based KEGG and COG datashops, we performed functional prediction of GM in each sample or group.
The community structure component map shows the functional composition and relative abundance of each sample or group. TOP10 KEGG function items were selected to generate the relative abundance column accumulation diagram, as shown in gure 6. Membrane transport, carbohydrate metabolism, amino acid metabolism were rst three pathway in function of GM. The COG functional prediction results were analyzed by Heatmap, as shown in gure 7. Moreover, STAMP results of COG was shown in gure 8.
Different colors represent different functional items, corresponding to the legend on the right; The horizontal axis represents different samples or groups, and the vertical axis represents the relative abundance of each functional item.
The horizontal axis represents different samples, and the vertical axis represents different functional items. The shade of color is related to the abundance of functional items. The darker the color, the higher the abundance.
The gure on the left shows the abundance ratio of different functions in the two samples or groups; the gure in the middle shows the difference ratio within the 95% con dence interval; the value on the far right is the P value, P value < 0.05, indicating signi cant difference.

Discussion
In the last decades, using 16S rDNA sequences, there have been an increasing amount of research into GM, and it seems that most systemic diseases in the body are related to them [21].In our previous study, we found that there were differences in GM between irritable bowel syndrome patients and normal population in Nanchang, China [22]. In this study, we focused on the RSA patients and wanted to nd the difference in GM between RSA patients and normal pregnant women in Nanchang, China. It is the rst to analyze the differences in GM between RSA patients and normal pregnant women.
The structure of gut microbiota changes during pregnancy The changes of GM in different stages of pregnancy have attracted much attention [23]. Studies from different regions have shown clear differences in the characteristics of GM between pregnant women and normal women. A US study reported that GM changed dramatically from rst to third trimesters, and showed vast expansion of diversity between mothers, an overall increase in Proteobacteria and Actinobacteria [24]. A Sandi Arabia study indicated that bacterial diversity decreased in pregnant woman, whereas phylum Bacteroidetes declined signi cantly (p<0.05) in the rst trimester, and a relatively high abundance of butyrate-producing bacteria (eg, Faecalibacterium spp. and Eubacterium spp.) in the gut of pregnant women [25]. A Japan research focused the differences in GM between early and late pregnancy, and they suggested that there were no obvious differences among four major phyla (Firmicutes, Bacteroidetes, Actinobacteria and Proteobacteria) between early and late pregnancy, although the proportion of the TM7 phylum decreased in late pregnancy compared with that in early pregnancy [26]. A study in China reported that at genus level, Akkermansia, Bacteroides, Subdoligranulum, Oscillospira, Ruminococcacea UCG-004), and Alistipes showed higher abundance during pregnancy [27].

RSA patients have abnormal gut microbiota
There has been no report on the correlation between RSA and GM, but there are many reports focused on the correlation between abortion and vaginal or oral bacteria. A China study was aimed at analyzing the changes in gut microorganism of patients with positive immune antibody-associated recurrent abortion using the 16s rRNA gene sequencing microbiome assay, and found that Bacteroides had the highest relative abundance in the positive group, Bacteroides, Erysipelotrichaceae_UCG-003, Faecalibacterium, and Prevotella_9 had high relative abundance in the negative group [28]. First trimester miscarriage associated with reduced prevalence of Lactobacillus spp.-dominated vaginal microbiota classi ed using hierarchical clustering analysis [29]. A study of vaginal microbiota showed that at the genus level, the relative abundance of Fam_Finegoldia and Lac_Roseburia signi cantly differed in the embryonic miscarriage group [30]. In the present study, although there was no signi cant difference in the abundance of dominant bacteria (Firmicutes, Bacteroidetes, Actinobacteria and Proteobacteria) between the two groups, but Actinobacteria and Proteobacteria tended to a higher in RSA patients at the phylum level. GM dysbiosis can cause insulin resistance (IR), which is closely linked to the occurrence of polycystic ovary syndrome (PCOS), and is associated with chronic in ammation, hormonal changes, follicular dysplasia, endometrial receptivity changes, and abortion or infertility [31].
Roseburia and Agathobacter may have a protective effect against RSA In our study, one of the most striking result was a signi cant decrease in the abundance of genera Roseburia in the RSA group (P<0.05). In addition, Agathobacter was signi cantly increased in NR2 group, while Klebsiella (P<0.05) were signi cantly increased, Roseburia (P<0.05) and Prevotella.9 (P<0.05) were signi cantly decreased in RSA1 group. These results suggest that these bacteria play an important role in pregnancy, and the following studies can provide certain support for our research.
A Japan study in GM of infertile women showed that the abundance of the genera Roseburia and Phascolarctobacterium were decreased in patients with infertility [32]. A mice study indicated that the microbiota of conventional mice were signi cantly different at the end of pregnancy (day 18) as compared with pre-pregnancy (p < 0.05), and the abundance of Roseburia faecis was signi cantly different at day 18 compared with pre-pregnancy [33].
Moreover, autism spectrum disorder (ASD) children with sleep disorder exhibited declines in the abundance of Agathobacter, decreased levels of 3-hydroxybutyric acid and melatonin [34]. Jin M et al found that Prevotella.9 were signi cantly decreased in patients with positive immune antibodyassociated recurrent abortion [28]. The enrichment of Faecalibacterium, Agathobacter and Roseburia were related to geriatric depression [35].
Gut mircobiota mechanism associated with RSA Functional prediction analysis indicated that GM may play their role through membrane transport, carbohydrate metabolism, amino acid metabolism and other mechanisms.

Dietary composition:
The composition of the GM can be in uenced by dietary composition. In our study, Roseburia (P<0.05) were signi cantly decreased in RSA group; furthermore, compared with RSA2 group, Roseburia (P<0.05) and Prevotella.9 (P<0.05) were signi cantly decreased in RSA1 group (patients with BMI>23.9 in RSA). The human gut Firmicute Roseburia intestinalis is a primary degrader of dietary β-mannans, gut Roseburia spp. metabolize dietary components that stimulate their proliferation and metabolic activities. They are part of commensal bacteria producing short-chain fatty acids, especially butyrate, affecting colonic motility, immunity maintenance and anti-in ammatory properties [36,37]. In early pregnancy, the relative abundances of Roseburia and Lachnospiraceae increased in GM of vegetarian compared with omnivorous diet, and sub-analysis of GM showed an alterations in fermentation end products from a mixed acid fermentation towards more acetate/butyrate [38].
Immune maintenance and anti-in ammatory: Roseburia is one of the most important microorganisms in human intestinal tract, and its main metabolite is butyric acid. Of note, the GMs that we found to differ signi cantly at the genus level, such as Roseburia, Prevotella.9, Agathobacter, were butyrate producing bacteriums. A ulcerative colitis (UC) study suggested that butyrate-producing species like Roseburia hominis, involved in the development of UC [39]. Several animal studies reported that abundance of Roseburia intestinalis (R.I), decreased signi cantly in patients with in ammatory bowel disease (IBD) and exerted an anti-in ammatory function in dextran sulfate sodium (DSS)-induced colitis [40,41,42]. Another animal study on UC indicated that R.I agellin plays an important role in the treatment of UC by inhibiting activation of the NLRP3 in ammasome and pyroptosis, [43]. In a murine model study suggested that Roseburia, a prominent gutassociated butyrate-producing bacterial genus, may provide a importion protection against atherosclerosis [44]. These studies suggest that R.I may be involved in the mechanism of immune maintenance and anti-in ammatory, which is related to butyrate.
Butyrate-producing bacteria may prevent RSA by regulating BMI Previous evidence from animal studies suggests that butyrate-producing bacteria prevents high fat dietinduced obesity [45]. In the present study, the patients with BMI>23.9 in RSA have decreased Roseburia (P<0.05) and Prevotella.9 (P<0.05), both well-known butyrate-producing bacteria. Study suggested that chronic butyrate supplementation can prevent diet-induced obesity, hyperinsulinaemia, hypertriglyceridaemia and hepatic steatosis [46]. Abundances of butyrate-producing bacteria, such as R.I and Faecalibacterium prausnitzii, were lower in patients with type 2 diabetes (T2D). This result support that butyrate and other short-chain fatty acids are able to exert profound immunometabolic effects [47]. The abundance and butyrate-producing bacteria and butyrate production of overweight and obese women at 16 weeks gestation were signi cantly negatively correlated with blood pressure and plasminogen activator inhibitor-1 levels. These results suggest that increased butyrate production may help obese pregnant women maintain normal blood pressure.