Alteration of the Gut Microbiota in Missed Abortion

There is a symbiotic relationship between gut microbiota and human beings. Imbalance of the gut microbiota will cause pathological damages to humans. Although many risk factors are associated with missed abortion (MA), the pathological mechanism of it is still unclear. Here, we analyzed gut flora of the patients with MA by S16 high-throughput sequencing. The possible pathogenic mechanisms of the MA were explored. Fecal samples from 14 healthy controls and 16 MA patients were collected to do 16S rRNA gene high-throughput sequencing analysis. The abundance of the Bacteroidetes, Proteobacteria, Actinobacteria, Escherichia, Streptococcus_ Salivarius, and Lactobacillus was significantly reduced in the MA group, while, the abundance of the Klebsiella was significantly increased in the MA patients. The Ruminococcaceae and [Eubacterium]_coprostanoligenes_group were found only in the specimens of the MA patients. The Fabrotax function prediction analysis showed that four photosynthesis function bacteria (cyanobateria, oxygenic_photoautotrophy, photoautotrophy, and phototrophy) only existed in the MA group. In the analysis of the BugBase microbiome function prediction, the Escherichia of the MA group is significantly reduced compared to that of the healthy controls in the items of that Contains_Mobile_Elements, Facultatively_Anaerobic, Forms_Biofilms, Potentially_Pathogenic.png, Gram_Nagative, and Stress_Tolerant_relabundance. These alterations may affect the stability of the host's immune, neural, metabolic and other systems by interfering with the balance of the gut microbiota or by the metabolites of those bacteria, causing the MA. This study explored the possible pathogenic factors of the gut microbiota of the MA. The results provide evidence to figure out the pathogenesis of the MA.


Introduction
Missed abortion (MA) is defined as a spontaneous death of the embryo or fetus [1,2]. MA has a great impact on pregnant women and her families. It may cause anxiety, depression or stress in the relevant personnel. Many who experienced MA have a sad process [3]. About 80% of the MA occurs in the first three months of pregnancy [1,4]. The risk of the MA within five to 20 weeks is 11% to 22%. The MA rate increases with the age of both parents [5,6]. The MA rate was 11%, increased from 9% at the age of 22 to 84% at the age of 48 in Denmark [7]. The etiological cause of about half of the MA involves chromosomal abnormalities. Although scientists believe there are many risk factors, however, not all of them can be determined [8][9][10]. Therefore, it is still a challenge for scientists to clarify the pathological mechanism of the MA.
Gut flora is the bacteria living in the human digestive tract [11,12]. The gut flora has a wide range of effects and perform many useful functions, such as fermenting unused energy substrates, training the immune system through metabolic end products (such as propionate and acetate, preventing the growth of harmful species), maintaining the intestinal epithelium, synthesizing vitamins (such as biotin and vitamin K) for the host, and producing hormones to guide the host to store fat, metabolize dietary and pharmaceutical compounds, control immune function and even influence behavior through gut-brain-axis. The relationship between gut flora and humans is not only symbiotic (harmless coexistence), but also reciprocal. Some human gut florae benefit the host by fermenting dietary fiber into short chain fatty acids (SCFAs) [13][14][15]. It has been proven that the imbalance of the gut flora will bring many diseases to humans. For example, ulcerative colitis [16], allergy, asthma and diabetes [14,17], cirrhosis, nonalcoholic fatty liver [18], obesity [19], and even increase the risk of cancer [20]. Bacteria constitute the majority of the flora in the colon, accounting for 60% of the dry weight of feces [20]. It makes feces as an ideal source of gut flora for any test and experiment. Through nucleic acids extraction from fecal samples and 16S rRNA gene sequencing, the gut bacteriome information could be easily obtained. This method is also generally preferable to the invasive techniques such as biopsy.
About 15% of recurrent abortion is related to immune factors [21]. Autoimmune disorder is a risk factor for MA. Abnormal immune status will influence the developing fetus, resulting in abnormal embryos, leading to MA [22,23]. On the other hand, the gut flora also has a profound impact on the host's immune system. Gut-brain-axis is a biochemical signal that occurs between the gastrointestinal tract and the central nervous system [24]. Gut flora can affect the host's neuroimmune system via the gut-brain-axis [24,25]. Gut flora also plays a direct role in defending against pathogens, secreting cytokines to initiate an inflammatory response against infection by utilizing all available nutrients [26]. In this study, we analyzed gut flora of the patients with MA by S16 high-throughput sequence analysis. The possible pathogenic mechanisms of the MA were explored from the analysis of the differential gut flora between the patients with MA and healthy controls (EPs). The results provided a proof in gut microflora factors for the study of the pathogenic mechanism of MA.

Patients and Sample Collection
In this study, from June to September of 2021, 16 hospitalized early pregnant patients with the MA (average age: 29.07 ± 4.428 years old) were recruited in our hospital. 10-20 g of their fresh feces on the day of diagnosis were collected in a sterilized 50 ml covered test tube from them, kept in a − 80 °C freezer until it was submitted for gut flora detection. During the same period, the feces of 14 normal pregnant women (average age: 28.75 ± 4.235 years old) were also collected as EPs. Inclusion and diagnosis criteria of the MA were the patients whose pregnacy period were within three months and who matched the MA diagnosis of the Society of Radiologists in Ultrasound in America [27,28]. Exclusion criteria of MA were those who had hereditary diseases, diabetes, infection, tumor, malnutrition, thyroid disease, or autoimmune diseases, who recently used antibiotics or other drugs, or who has ovarian and uterine malformations, uterine fibroids, or obesity. This study was approved by the Ethic Committee of the First Hospital of Putian (No. 2022-025). All patients and healthy controls signed the informed consents.

Gut Flora Detection
All the gut flora were detected by Biomarker (Beijing, China). The specific primer with barcode was synthesized according to the full-length primer sequence. PCR amplification was carried out using the primers with barcode. The product was purified, quantified and homogenized to form corresponding sequencing library (Single Molecule Real Time Bell), and sequenced by PacBio Sequel RS II (Pacific Biosciences, Menlo Park, CA, USA). Cutadapt method [29] was applied to recognize the primer sequences. Trimmatic means [30] was employed to filter the CodeCharge Studio (CCS) documents. USEARCH [31] and UCHIME [32] measures were used to remove chimeras. Finally highquality sequences were obtained for subsequent analysis.

Operational Taxonomic Unit (OTU)/Amplification Sequence Variants (ASVs) Analysis
The sequences were clustered below the level 97% similarity using USEARCH to filter out the OTUs. The DADA2 [33] method in QIIME2 [34] was used to denoise the data after quality control. The reserved total sequences number was equal and above 0.005% Features (i.e. OTU) of the total number of sequences.

Diversity Analysis
Alpha diversity indexes of the gut flora of the MA and the EP groups were analyzed and evaluated by QIIME2 (https:// qiime2. org/) to get indexes of Chao 1 richness estimator, ACE richness estimator, Shannon Wiener diversity index, Simpson diversity index, and PD_whole_tree, respectively. The beta diversity analysis was processed with binary Jaccard, Bray Curtis and (UN) weighted UniFrac algorithms to present the species diversity matrix. R language platform was used to prepare Principal component analysis (PCA), principal coordinate analysis (PCoA) and correlation analysis between environmental factors and sample composition (RDA/CCA).

Species Annotation and Taxonomic Analysis
The feature sequence was taxonomically annotated with Silva as the reference database, and naive Bayes classifier combined with alignment method. The species classification information corresponding to each feature and the composition of each sample colony at each level (phylum, class, order, family, genus, species) were obtained. QIIME software was applied to generate species abundance tables at different classification levels. Following that, the R language tool was used to draw the community structure map of the samples at various taxonomic levels.

Analysis of Significance of Difference Between Groups
Significant difference analysis between groups, which can be called biomarkers analysis, at genus and species levels were performed by Metastats analysis, ANOVA and Willcox rank sum test.

Correlation Analysis
Network graph correlation analysis was employed to perform Spearman rank correlation analysis. Screen data with correlation greater than 0.1 and p value less than 0.05 were used to build up a correlation network according to the abundance and change of each genus or species in each sample. The genus correlation network diagram was also drawn based on Python.
Based on the fact that the accumulation of metabolites of the microbiota may cause changes in the host microenvironment, thereby affecting the physiological and pathological activities of the host, it is necessary to study the correlation between gut flora and host metabolomic functions. On account of the analysis of network diagram, the coexistence relationship of species in environmental samples can be obtained, and the interaction and important mode information of species in the same environment can be obtained, further explaining the formation mechanism of phenotypic differences between samples.

Function Prediction Analysis
BugBase was applied to predict the biological level coverage of functional pathways within the complex microbiome and the biologically interpretable phenotypes. Kyoto Encyclopedia of Genes and Genomes (KEGG) and FAPROTAX [35] were also employed to predict metabolic pathways and ecological relevant functions.
BugBase first normalizes OTUs by predicting 16S copy numbers, and then predicts microbial phenotypes using the provided precomputed files. First, for each sample in the biological dataset, the relative abundance of traits was estimated in the whole range of coverage thresholds (0 to 1, in increments of 0.01). Then, BugBase selects the coverage threshold with the highest variance in all samples for each feature in the data. After setting the threshold, BugBase will generate the final organism level trait prediction table, which contains the predicted trait relative abundance of each sample. The predicted phenotype types included gram positive, gram negative, biofilm forming, genetic, mobile element containing, oxygen utility, and oxidative stress tolerance.

Results
OTU Analysis of the S16 Sequences from the Gut Flora of Both the MA and the EP After 30 samples were sequenced by S16 high-throughput sequence and identified by Barcode, 389045 circular consensus sequencing (CCS) sequences were obtained. Each sample generated at least 12725 CCS sequences, with an average of 12968 CCS sequences.
Through S16 high-throughput sequence analysis, the similarity among different sequences was calculated and compared, and the fragments with 97% similarity were clustered to form different OTU. The OTU numbers found in each MA sample and EP sample were presented in Fig. 1A, respectively. In totally 269 OTUs were found among these samples. Amongst them, 235 OTUs were shared by both the MAs and EPs. 14 OTUs were exclusively owned by healthy controls and 20 OTUs were specifically shared by MAs (Fig. 1B).

Species Annotation and Taxonomic Analysis
SILVA (https:// www. arb-silva. de/) (Silva.138 version) is a comprehensive database containing rRNA gene sequences of three domain microorganisms (bacteria, archaea and eukaryotes). In order to annotate the obtained OTU data, through comparing to the SILVA database, the feature taxonomic sequence of the OTUs were annotated with the naive Bayes classifier method to obtain the species classification information corresponding to each feature. Thus, the community composition of each sample at various levels (phylum, class, order, family, genus, species) was gained. Subsequently, QIIME software was employed to generate species abundance tables at different taxonomic levels. The R language tools were used to draw community structure maps of samples at various taxonomic levels.
At phylum level, comparing to the EP group, the Firmicutes were comparatively enriched in the MA group, whereas the Proteobacteria and Bacteroidetes were relative enriched in the EP group ( Fig. 2A). Based on the suggestion of Turnbaugh et al. [36] that the elevation of the Firmicutes/ Bacteroidetes ratio is applied as a sign of pathological conditions. In our data, the ratio of the MA group was 2.83 (0.534198/0.188764) which is obviously higher than 1.883 (0.420116/0.223096) of the EP group, demonstrated that there was a pathological alteration in the MA patients.
Since the bacterial taxonomic at genus and species levels are the most abundant, in order to facilitate coherent Total 269 OTU numbers found from these samples were showed in the right bar. B Among the 269 OTUs, 235 were shared by them (MAs and Eps). 14 were exclusively owned by Eps. 20 OTUs were specifically occupied by Mas only analysis, we mainly focused on analyzing the alterations of the bacteria at the level of genus and species. In the distribution of bacterial abundance at the level of the top ten genera (Fig. 2B), compared with the EP, the relative abundance of Escherichia, Subdoligranulum, Prevotella, and Dialister in the MA group were remarkably decreased, while, the abundances of Bacteroides and Klebsiella were significantly increased in the MA group.
In the top ten distributions of bacterial abundance at specie level ( Fig. 2C), compared with the EP group, the abundances of Prevotella_copri and Escherichia_coli were significantly lower, however, the abundances of Klebsiella_ pneumoniae, uncultured_bacterium_g_Subdoligranulum, Bacteroides_uniformis, and Dialister_sp were obviously greater in the MA group.

Biological Diversity Analysis
Alpha-diversity indexes analysis of Shannon, Simpson, ACE, Chao1, and PD_ehole_tree did not showed any significant differences between the MA and EP groups (p > 0.05) ( Table 1).
Beta-diversity indexes analysis of Principal Component Analysis, Principal coordinates analysis, Non-MetricMulti-Dimensional Scaling, Unweighted Pair-group Method with Arithmetic Mean, combining UPGMA clustering tree with histogram, Sample thermographic analysis, and PER-MANOVA/Anosim analysis did not showed any obvious difference between the groups as well (p > 0.05)(data did not show).

Analysis of Significance of Difference Between Groups
The ANOVA and Wilcox rank sum tests revealed that the abundance of the Synechococcus by ANOVA analysis at genus level and the Synechococcus_ruvescers by Wilcox rank sum test at species level are only found in the MA group. However, the abundance of the Acinetobacter by Wilcox rank sum test at genus level and the Prevotella_copri by ANOVA analysis, the Acinetobacter_lwoffii and the Syn-echococcus_salivarius by Wilcox rank sum test at species  level were significantly lower in the MA group compared to those in the EP group (p < 0.05) ( Table 2). The Metastats analysis at genus level indicated that the Brochothrix, the Chryseobacterium, the Dielma, and the Mitsuokella only existed in the EP group. The Rumi-nococcaceae_UCG_008, the Alloprevotella, the uncul-tured_bacterium_o_Bacteroidales, and the Anaerotignum were only found in the MA group. The abundances of the[Eubacterium] _coprostanoligenes_group, the Brevundimonas, the Prevotella, the [Eubacterium] _ruminantium_ group, and the Lactobacillus were clearly lower in the MA group (p < 0.05). However, the abundances of the Gemella and the Lachnospiraceae_ND3007_group were significant higher in the MA group (p < 0.05) ( Table 3).
The Metastats analysis at species level suggested that the Brochothrix_thermosphacta, the Chryseobacterium_ haifense, the Dielma_fastidiosa, the Lactobacillus_iners, the Mitsuokella_sp, the Mobilibacterium_massiliense, and the Paraprevotella_xylaniphila were only observed in the EP group. The Intestinibacillus_massiliensis, the Fusobacterium_varium, the uncultured_bacterium_g_Alloprevotella, the uncultured_bacterium_o_Bacteroidales, and the Anaerotignum_lactatifermentans only existed in the MA group. The abundances of the Prevotella_copri, the uncul-tured_bacterium_g_[Eubacterium]_coprostanoligenes_ group, the Brevundimonas_mediterranea, the uncultured_ bacterium_g_ [Eubacterium]_ruminantium_group, and the Acinetobacter_lwoffii were notably lower in the MA group compared to those in the EP group (p < 0.05). The abundances of the Eubacterium_sulci, the Alistipes_finegoldii, the Gemella_sanguinis, the Eubacterium_ramulus, and the uncultured_bacterium_g_Lachnospiraceae_ND3007_group were specifically higher in the MA group compared to those in the EP group (p < 0.05) ( Table 4).

Correlation Network Analysis at Genus Level
According to the abundance and change of each species in each sample, Spearman rank correlation analysis was conducted and data with correlation greater than 0.1 and Table 2 Analysis of significance of difference between groups at genus and species levels (u: relative abundance; n = mean ± SE)  Table 5). The top 50 genera correlation network diagram with the highest correlation based on Python was prepared in Fig. 3.

Prediction and Analysis of Functional Genes at Genus Level
BugBase is a method to predict the biological functional pathways in complex microbiomes and the biologically interpretable phenotype. BugBase selectively carries out automatic hypothesis testing and visualization for different traits, and generates nine phenotypes (Aerobic, Anearobic,

Discussion
Consisting with trillions of symbiotic microflorae, the gut microbiota provides the essential materials to maintain the host's health [37,38]. In our analyzed data, Firmicutes, Proteobacteria, Bacteroidetes, Actinobacteria, Verrucomicrobia, Tenericutes, and Fusobacteria were the major dominant  phyla, which consists of more than 90% of the total microbial population ( Fig. 2A), being consistent with the findings by Jethwani and Grover [39]. Bacteroidetes, Firmicutes, Proteobacteria and Actinobacteria are the four major phyla in human adults. The Bacteroidetes and the Firmicutes possess the highest proportion, occupying ~ 48% or ~ 51%, respectively. The Proteobacteria and the Actinobacteria are of relatively lower proportions (1%) [40]. Our results revealed that the proportion of the Bacteroidetes, the Proteobacteria, and the Actinobacteria in the MA group are 19%, 22%, and 2.46%, respectively, being relatively lower than those in the EP group. However, the relative ratio of the Firmicutes (53%) of the MA group was relatively higher than that of the EP group. (Fig. 2A) The Firmicutes/Bacteroidetes ratio is suggested to be an indicator of a pathological conditions [36]. In this study, the Firmicutes/Bacteroidetes ratio was relatively high in the MA group (2.83) compared to the EP group (1.883). The tendence of these alterations are similar to Liu's findings, although the exact ratios are not the same [41]. Firmicutes, Bacteroidetes and some anaerobic gut microorganisms can metabolize indigestible carbohydrates such as hemicellulose, cellulose, pectin, resistant starch, oligosaccharides and lignin into short chain fatty acids (SCFA), such as acetic acid, propionic acid and butyric acid [42]. The interference of the SCFA biosynthesis can cause many pathological consequences to the host [43]. Thus, the higher proportion of the Firmicutes (53%) may participate in the pathogenesis of the MA. The abnormal increase of the Firmicutes has been proven to be related to diabetes and obesity [44,45]. In fact, obesity and diabetes are both the risk factors for recurrent abortion [46]. Both the observation in this experiment and Liu's study [41] clearly showed the abnormal elevation of the Firmicutes in the MA patients. Its related pathological mechanism needs further study. On the other hand, both the EP and MA groups exhibited higher rate of the Proteobacteria (31-22%) in our observation, which might be caused by different geography and eating habits between the Spain (< ~ 1%) and Chinese [40].
In the analysis results of the top ten microbiotas at both the genus and species levels, Escherichia (or Escherichia_ coli) were significantly lower in the MA group (Fig. 2B, C). Although some Escherichia_coli can cause certain diseases, most of them are harmless bacteria, which can produce vitamin K2 [47] and prevent the colonization of pathogenic bacteria in the intestine, to benefit the host and to have a reciprocal relationship [48,49]. Vitamin K2 is an essential  factor for blood coagulation [50]. Disorder of coagulation is considered as one of the causes of recurrent abortion. Therefore, the abundance reduction of the Escherichia_coli may affect the synthesis of vitamin K2 to interfere with blood coagulation function, thus inducing the MA.
Bacteroides species usually constitute the most important part of the gut microbiota of mammals. Studies have shown that long-term diet is closely related to intestinal microbiome composition. Those who eat a lot of protein and animal fat are mainly Bacteroides, while those who eat more carbohydrates predominate with Prevotella [51]. In the MA cases of this study, the abundance of the Bacteroides (Bacteroides_uniformis) is higher than that of the EP group, while the abundance of the Prevotella (Prevotella_copri) is lower than that of the control group (Fig. 2B, C). This result suggests that the population of the MA group prefers a diet of large amounts of protein and animal fat. Besten et al.have shown that the Prevotella_copri improves insulin-sensitivity [52]. A meta-analysis indicates that insulin resistance is correlated to the susceptibility to MA, it may cause the recurrent miscarriages [53]. Thus, the decline of the Prevotella (Prevotella_copri) may participate the pathogenic process of the MA.
Klebsiella usually exists in human gastrointestinal tract as normal flora. However, it is also a human pathogen. Klebsiella can cause many diseases including pneumonia, sepsis, urinary tract infection, meningitis, peritonitis, diarrhea, and soft tissue infection. Most human Klebsiella infections are caused by Klebsiella_pneumoniae [54,55]. The distribution of the top ten species and genera showed that the abundance of the Klebsiella (genus) and Klebsiella_pneumoniae (species) are significantly greater in the MA patients (Fig. 2B,  C), suggesting that the Klebsiella (Klebsiella_pneumoniae) may play a role in the pathological mechanism of the MA.
In the difference analysis between the MA and EP groups, it also showed that the abundance of the Synechococcus (genus) and Synechococcus_rubescens (species) were significantly higher in the MA patients than those in the healthy controls at the genus and species level ( Table 2). Up until now, there is no report on the pathology of the Synechococcus (Synechococcus_rubescens). Whether they play a role in the pathological mechanism of the MA remains to be seen. It is interesting to note that Acinetobacter and Acinetobac-ter_lwoffii are the pathogen of hospital acquired pneumonia, wound infection, bacteremia and meningitis [56]. However, they are significantly reduced in the MA patients (Table 2). Streptococcus_ Salivarius was found significantly lower in the MA patients compared to that in the EP group (Table 2). It belongs to a probiotic and can produce antibacterial peptide (bacteroid inhibitor), which can inhibit the growth of Streptococcus pyogenes [57]. It is worth noting that rheumatic heart disease is an autoimmune disease caused by Streptococcus pyogenes. Autoimmune disorder is also considered as a risk factor for the MA [22]. Whether the reduction of the Streptococcus_salivarius abundance can induce an autoimmune disorder, resulting in the MA is a subject worthy of further study.
Metastats analysis showed that the Lactobacillus in the MA patients were significantly lower at genus level (Table 3) and absent at species level (Lactobacillus_iners; Table 4). The Lactobacillus has been reported to regulate L-22 to maintain the host-microorganism homeostasis on the surface of intestinal mucosa and intestinal barrier function [58], prevent intestinal inflammation [59], and elevate intestinal IgA production [60]. Inflammation is considered to be one of the causes of the MA. Therefore, the reduction or deletion of the lactobacillus may also be one of the causes of the MA.
Our data shows that there was no significant difference between MA group and EP group in either alpha-or betadiversity analysis. The results were inconsistent with those of Liu et al. [41]. They detected the gut flora in 41 MA patients and 19 healthy controls and found that the alphadiversity of the MA group is significantly lower than those of the control group, indicating that the abundance and evenness of the gut flora in the MA patients are low. Whether this difference is caused by different geographical regions (Shanghai vs Putian in China), or different ages (35y vs 29y) remains to be studied.
We note that the highest correlation analysis results at the genus level showed that the Ruminococcaceae_UCG-008 exhibited a significant positive correlation with the coprococcus (Table 5(A) and Fig. 3). While the Ruminococcaceae was not detectable in the gut flora of the normal controls (EP), however, it was found in the specimens of the MA patients (Table 3). Omura et al. [61] found a significant increase in the abundance of intestinal coprococcus in 35-days mice after infection with Theiler's murine encephalomyelitis virus, accompanied with increased expression of T cell receptor (TCR), IgG, IgA, various complements and major histocompatibility complex The limitation of this study is not able to verify the speculative pathogenic mechanism of the MA in vivo or in vitro experiments. In fact, these works need to be done in the future by multiple institutes in biochemistry, microbiology, animal model, and even clinical studies.

Conclusion
Ethics Approval This study was approved by the Ethics Committee of the First Hospital of Putian City (No. 2022-025).
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.