Comparative Analysis of Gut Microbiota Among the Male, Female and Pregnant Giant Pandas (Ailuropoda Melanoleuca)

Abstract The giant panda (GP) was the most endangered species in China, and gut microbiota plays a vital role in host health. To determine the differences of the gut microbiota among the male, female and pregnant GPs, a comparative analysis of gut microbiota in GPs was carried out by 16S rRNA and ITS high-throughput sequencing. In 16S rRNA sequencing, 435 OTUs, 17 phyla and 182 genera were totally detected. Firmicutes (53.6%) was the predominant phylum followed by Proteobacteria (37.8%) and Fusobacteria (7.1%). Escherichia/Shigella (35.9%) was the most prevalent genus followed by Streptococcus (25.9%) and Clostridium (11.1%). In ITS sequencing, 920 OTUs, 6 phyla and 322 genera were also detected. Ascomycota (71.3%) was the predominant phylum followed by Basidiomycota (28.4%) and Zygomycota (0.15%). Purpureocillium (4.4%) was the most prevalent genus followed by Cladosporium (2.5%) and Pezicula (2.4%). Comparative analysis indicated that the male GPs harbor a higher abundance of phylum Firmicutes than female GPs with the contribution from genus Streptococcus. Meanwhile, the female GPs harbor a higher abundance of phylum Proteobacteria than male GPs with the contribution from genus Escherichia/ Shigella. In addition, the shift in bacteria from female to pregnant GPs indicated that phylum Firmicutes increased significantly with the contribution from Clostridium in the gut, which may provide an opportunity to study possible associations with low reproduction of the GPs.


Introduction
In 2013, nearly 1,860 individual giant pandas (GP, Ailuropoda melanoleuca) were found in Western China (http://www.forestry.gov.cn/main/72/content-742880. html). The GP is a rare wild animal, ranking at the top of the list of endangered species on earth [1]. It is well known that the GP harbors a special dietary preference to bamboo, a high-fiber food. Although GP belongs to the Order Carnivore [2], it consumes a unique herbivorous diet [3]. Previous studies showed that about 8% and 27% of the cellulose and the hemicelluloses, respectively, in bamboo could be digested by GPs [2]. Through wholegenome sequencing, however, no specific genes that are responsible for the digestion of cellulose and hemicellulose were found in GPs [4], suggesting that gut microbiota play a vital role in digesting bamboo fibers [2]. Besides, the gut microbiota also has an impact on the health of the host [5]. The gut microbiota is involved in energy harvesting and storage, as well as in a variety of metabolic functions such as fermenting and absorbing undigested carbohydrates [6]. More importantly, it has become clear that the gut 2 Materials and Methods

Sample collection
This study was carried out with approval of the China Conservation and Research Center for Giant Panda in Sichuan, China. All the giant pandas were fed the same diet. Eighteen fecal samples were collected from the adult male (n=7), female (n=5) and pregnant (n=6) GPs (6 to 10 years old) once in 2016. The GPs had similar husbandry conditions and were housed at Bifengxia Base, China Conservation and Research Center for Giant Panda. Fecal samples were collected immediately after defecation, snap frozen (-80ᵒ C), and shipped to the laboratory in dry ice.

DNA extraction and Miseq sequencing
The whole genome DNA from the samples was extracted by using the Mobio Power Fecal TM DNA Kit (Laboratories Inc., America) and EZNA Fungal DNA Mini prep Kit (Omega Inc., America) according to the manufacturer's instruction. The hypervariable region V4 of the 16S rRNA genes was amplified by PCR, using primers 520F (5'-barcode + AYTGGGYDTAAAGNG-3) and 802R (5'-TACNVGGGTATCTAATCC-3'); the first region of the ITS genes were amplified by PCR, using primers ITS1 (5'-barcode + TCCGTAGGTGAACCTGCGG-3') and ITS2 (5'-GCTGCGTTCTTCATCGATGC-3'). The PCR products were then submitted to the Shanghai Personal bio-tech Co. Ltd for sequencing that based on an Illumina MiSeq 2500 platform [21].
The 16S ribosomal RNA (16S rRNA) and Internal Transcribed Spacer (ITS) high-throughput sequencing overcame the limitations of culture-based bacterial and fungal detection [9], and allowed exploration of the gut microbiota in depth, exhibiting its complete bacterial and fungal diversity [10,11]. In recent years, 16S rRNA and ITS high-throughput sequencing were also applied to analyze the gut microbiota community composition of GPs [12]. Using such sequencing, comparisons of the gut microbiota in GPs have been subsequently conducted, including the differences in age [9] and season [13]. However, the composition of the gut microbiota is able to be affected by various factors such as intestinal environment, nutritional and non-nutritional dietary components, antibiotic use [14] and gender difference [15]. In addition, gut microbiota has also been reported to be modified during pregnancy [16]. It has been reported that the GP has a low fecundity, which may be one of the causes behind its population decline [17]. According to previous studies, pregnancy has impact on the diversity of gut microbiota to some extent and the maternal intestinal microbiota is modified over the course of healthy pregnancy. It is possible that maternal gut bacterial profiles may be associated with the known endocrine changes that accompany the female reproductive (estrous) cycle [18]. On the other hand, the microbiota could lead to host maternal gestational weight gain after pregnancy [19]. In addition, some bacteria could cause host adverse pregnancy outcomes [20]. Therefore, there may be a strong relationship between gut microbiota and pregnancy. The variation of gut microbiota should also have an effect on pregnancy. Hence the variation or difference in gut microbiota during pregnancy should also be examined to test this hypothesis.
Until now, the differences of gut microbiota composition among male, female and pregnant GPs have not been examined. Here, the 16S rRNA and ITS highthroughput sequencing were used to characterize the gut microbiota among male, female and pregnant GPs in order to make a comparative analysis of gut bacterial and fungal communities. curves became flat gradually and reached a plateau with more data indicating that the number of OTUs for each sample was sufficient and reasonable (Fig.S1). The rank abundance curves that reflected the evenness and abundance of species in fecal samples horizontally and vertically, respectively, were demonstrated in Fig.S2.

Analysis of discrepancies for male, female and pregnant groups
For bacteria, the principle coordinates analysis (PCoA) plots (Fig.2a) demonstrated that each group tended to assemble together within respective groups. Meanwhile, the PCoA result was verified by ANOSIM (R>0, P<0.01) (Fig.S4). For fungi, however, each group fails to assemble together within respective groups (Fig.2b). Therefore, the fungal samples were not used to make a group comparative analysis.
In Venn figures, 171 OTUs were found in all the groups shared by the male, female and pregnant group (Fig.S5). In addition, the shared OTUs comprised approximately 39.2% of the total OTUs, while 59, 30 and 52 OTUs were each OTU was screened for further annotation. Based on OTUs and species annotation, the dominant species in various samples (groups) and OTU differential abundance testing information were determined.
Six indices including observed species, Chao1, Shannon, Simpson, ACE and Good's coverage were used to analyze the complexity of species diversity for all samples. All indexes (including observed species, Chao1, Shannon, Simpson, ACE and Good's coverage) for bacteria and fungi were analyzed by using software R (https:// www.r-project.org/) with ANOVA (Analysis of Variance) method. Rarefaction curves and Rank abundance curves were delineated to evaluate the reasonableness of all the samples.
A one-way analysis of similarity (ANOSIM) [25] was performed to determine the differences among the male, female and pregnant groups. The principal coordinate analysis (PCoA) that using the OTU-based weighted Unifrac distance matrix was performed basing on "Out table" by using software R (https://www.r-project.org/) with package GUniFrac, ape and ggplot2 [26] to visualize the discrepancy among the male, female and pregnant groups. Meanwhile, a Venn diagram was employed to describe the common and unique OTUs in each group. The top 10 phyla and 35 genera were chosen to generate the percentage-stacked histogram of relative abundance for each sample and group respectively. According to the PCoA, the specific species that had significant difference among the 3 groups at each level were calculated by using T test and LDA effect size (LEfSe) analysis.
All raw sequences obtained in this study were archived at NCBI Sequence Read Archive (SRA) under accession number SRP149033.

Overview of the sequencing data
In 16S rRNA sequencing, 1,148,722 high quality reads were obtained, and classified into 435 OTUs with the 97% similarity from the 18 fecal samples of the GPs. In ITS sequencing, 1,052,318 high quality reads were obtained, and classified into 920 OTUs with the 97% similarity.
The Alpha diversity indices for bacteria and fungi (including observed species, Shannon, Chao1, ACE and Good coverage) were shown in Table S1 and Table S2, respectively. However, there is no significance in all alpha diversity indexes (including observed species, Shannon, Chao1, ACE and Good coverage) among the male, female and pregnant groups (p>0.05). The rarefaction   No significance in all alpha diversity indexes (including observed species, Shannon, Chao1, ACE and Good coverage) was observed among the male, female and pregnant groups (p>0.05). The boxplot of Chao1 and Shannon index showed that diversity was not significantly different among the male, female and pregnant groups (Fig.S6).
The LDA effect size (LEfSe) analysis exhibits the specific taxa that had significant difference among male, female and pregnant groups (Fig.4a). A total of 16, 26 and 15 taxa that had discrepancies in relative abundance were presented in the male, female and pregnant groups, respectively (e.g. Firmicutes, Proteobacteria, Streptococcus, and Escherichia/Shigella). It's obvious that, at the phylum level, the relative abundance of Firmicutes showed remarkable difference (p<0.001) in the male group, and Proteobacteria was significantly higher in the female group than other groups (p<0.001). The cladogram in Fig.4b showed the core bacterial species were in remarkable difference at all levels. According to the T-test of bacterial species difference, Firmicutes in the male group had a significantly higher abundance than female at the phylum level (p<0.001). The abundance of Proteobacteria in female group was significantly higher than that in the males (p<0.001). At the genus level, Streptococcus (p<0.001) and Leuconostoc (p<0.05) in male group showed a significantly higher abundance than uniquely identified among the male, female and pregnant groups, respectively. Meanwhile, 9 phyla were found in all the groups shared by the male, female and pregnant groups (Fig.S5). Three phyla were uniquely identified in the female group. 76 genera were found in all the groups shared by the male, female and pregnant groups (Fig.S5). Moreover, 29, 12 and 19 genera were uniquely identified in the male, female and pregnant groups, respectively.
Our study demonstrated the discrepancy between male and female GPs. We found that the male GPs harbor a higher abundance of phylum Firmicutes than female GPs with the contribution from genera Streptococcus. Meanwhile, the female GPs harbor a higher abundance of phylum Proteobacteria than male GPs with the contribution from genera Escherichia/Shigella. Some Streptococcus species are responsible for cases of endocarditis [29], erysipelas [30], and necrotizing fasciitis [31]. However, most of them are not pathogenic, and form part of the commensal microbiota of the intestine tract [32,33]. The strains of Escherichia/shigella are also part of the normal flora of the gut, and can benefit their hosts by producing vitamin K2, and preventing colonization of the intestine with pathogenic bacteria, having a symbiotic relationship [34]. However, some virulent strains can cause gastrointestinal tract infections [35]. In recent years, the differences in gut microbiota between males and females had been successively reported [36]. The difference between males and females in the composition of mouse fecal flora has been examined by denaturing gradient gel electrophoresis (DGGE) [37]. However, the significant taxon cannot be detected by DGGE.. The gender effects female. The abundance of Escherichia/Shigella (p<0.001), Serratia (p<0.001) and Enterobacteria (p<0.05) were significantly higher in the female group than the male. According to the T-test of bacterial species difference, Firmicutes (p<0.05) in the pregnant group has a significant higher abundance than females at the phylum level. Meanwhile, the abundance of Clostridium (p<0.001) and Turicibacteria (p<0.05) in the pregnant group was also significantly higher than females at the genus level. More details on species with significant discrepancy at the phylum and genus level are presented in Fig.5.

Discussion
In this study, we characterized the gut microbiota among the male, female and pregnant GPs. Consistent with previous studies [3,13], all fecal samples had low diversity and were dominated by bacteria in the phyla Firmicutes (53.6%) and Proteobacteria (37.8%), with contributions from the genera Escherichia/Shigella (35.9%), Streptococcus (25.9%), and Clostridium (11.1%). Our study showed that most fungi in all fecal samples were affiliated to the phyla Ascomycota (71.3%) and Basidiomycota (28.4%), with contributions from classes Saccharomycetes (30.8%), Tremellomycetes (18.5%), Dothideomycetes (21.6%) and Sordariomycetes (14.6%). However, there is no information about the dominant genus of fungi in previous studies [27,28]. Purpureocillium  . T-test bar plot, the species with significant discrepancy at phylum and genus level among the male, female and pregnant group, as well as the relative abundance and p value.
the contribution from genera Escherichia/Shigella. In addition, the shift in bacteria from female to pregnant GPs indicated that phylum Firmicutes increased significantly with the contribution from Clostridium in the gut, which may provide an opportunity to study possible associations with low reproduction of the GPs. on human gut bacteria were firstly observed for the genus Prevotella, which was affiliated to phylum Bacteroides, with higher levels in males than in females. Mueller et al suggested that the existence of a postpubescent gender bias in microbial diversity and representation of individual species became evident [38]. On the contrary to Mueller's result, Haro et al found that the abundance of the Bacteroides was lower in men than women [39]. Gut bacteria has been reported to shift during gestation or pregnancy in human [19] and other mammals [40]. Koren et al demonstrated that pregnancy was associated with alterations to the gut microbiota based on animal (mice, mammal) model [41]. Ji et al discovered a tendency for the abundance of Proteobacteria to increase as pregnancy progressed in sow, even if all the sows share the same dominant phyla Firmicutes, Proteobacteria and Bacteroidetes of gut bacteria [16]. Meanwhile, Jost et al observed a significant decrease (p < 0.05) in Lactobacillus, suggesting that an underestimation of Bacteroidetes occurs during pregnancy [42]. In our study, we also found the shift from female to pregnant GPs that phylum Firmicutes underwent a significant increase (p <0.001) with the contribution of Clostridium and Turicibacteria. Clostridium, affiliated to Firmicutes, was known as a function of cellulose degradation [43]. However, some species of Clostridium like Clostridium difficile could be a threat to the health of pregnant host individuals, in particular, by causing diarrhea [44]. Turicibacter bacteria is commonly detected in the gastrointestinal tracts and feces of humans and animals, but their phylogeny, ecological role, and pathogenic potential remain unclear [45]. Han et al [46] reported that Turicibacter belong to class Erysipelotrichia which have been isolated from swine manure and increase in composition of the mouse gut microbiome for mice switched to diets high in fat [47]. There is evidence that pregnancy is a physiological state that is associated with shifts in gut microbiota [19]. Hence, the physiological and biochemical indexes of the pregnant GPs should be detected to be associated with the shifts in microbiota. If possible, a fecal transplant could be performed to improve the intestinal flora of non-pregnant giant pandas. Besides, the certain functions of the shift microbiota should be confirmed by using metagenomics.
In conclusion, we characterized the gut microbiota among the male, female and pregnant GPs. Through comparative analysis, we determined the discrepancy between male and female GPs which indicated that the male GPs harbor a higher abundance of phylum Firmicutes than female GPs with the contribution from genera Streptococcus. Meanwhile, the female GPs harbor a higher abundance of phylum Proteobacteria than male GPs with