Effects of dietary nutrient levels on microbial community composition and diversity in the ileal contents of pregnant Huanjiang mini-pigs

The mammalian gut microbiota influences various metabolic and physiological processes. Substantial metabolic changes occur during a healthy pregnancy that may be related to microbiota composition dynamics. However, the effect of diet on intestinal microbiota composition and diversity during pregnancy remains unclear. We examined the ileal contents of Huanjiang mini-pigs at two pregnancy stages to determine the effects of dietary nutrient levels on such microbial communities. Animals received either a higher-nutrient (HN) diet formulated to meet US National Research Council requirements or a lower-nutrient (LN) diet that met the Chinese National Feeding Standard recommendations. On day 45 or 75 of pregnancy, sows were euthanized and their ileal contents sampled. Experimental diet and pregnancy stage did not affect ileal bacterial richness or diversity, as determined by Chao1 and ACE species richness measures and Shannon and Simpson indices, respectively. The phyla Firmicutes and Proteobacteria, accounting for 69.99–85.44% and 5.82–15.17% of the total reads, respectively, predominated regardless of diet. At the genus level, diet significantly affected the abundance of Lactobacillus species, which was greater in pigs given HN feed (P < 0.05), but had little impact on that of Megasphaera species (P = 0.096). Pregnancy stage had a minimal effect on Proteobacteria numbers (P = 0.053). The number of bacteria of the phylum Firmicutes and genus Lactobacillus decreased, while that of the phylum Proteobacteria, family Enterobacteriaceae, and genus Bacteroides increased between days 45 and 75 of pregnancy. Of the short-chain fatty acids (SCFAs) measured, only propionate levels changed significantly, with higher concentrations observed on day 45 than on day 75. Our findings indicate that Firmicutes and Proteobacteria dominate pregnant sow ileal bacterial profiles. Excepting a tendency for the number of Proteobacteria to increase as pregnancy progressed, pregnancy stage and diet had little effect on ileal microbiotic composition and diversity and luminal SCFA concentrations.


Sample collection
Five and eight pregnant gilts in the HN and LN diet groups, respectively, were examined 45 days after service, and six pregnant gilts in both groups were tested 75 days after service. Initially, eight sows were included in each group, but owing to unsuccessful matings, the final group numbers differed. This was addressed in our statistical analysis.
In a report by Johnston and Trottier (1999), the early, middle, and late stages of pregnancy in pigs are defined as days 1 to 30, 30 to 75, and 75 to delivery, respectively [20]. Considering the size of Huanjiang mini-pigs and the difficulty of collecting conceptus samples to determine fetal development, we chose 45 and 75 days post-service to represent the middle and early-late stages of pregnancy, respectively.
Sows were euthanized for sample collection 12 h after the last feeding on day 45 or 75 postservice [12]. Briefly, general anesthesia was induced by intravenous injection of 4% sodium pentobarbital solution (40 mg/kg BW) and euthanasia carried out by exsanguination following severing of the carotid artery [21]. The ileum was then recovered and its luminal contents collected from a region 10 cm anterior to the ileocecal valve. These were stored at −80˚C for subsequent analysis of gut microbial composition and SCFA concentrations.

Microbial DNA isolation and PCR amplification
Total bacterial DNA was extracted from ileal contents using a QIAamp DNA Stool Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. The DNA concentration of each extract was measured with a NanoDrop ND-1000 instrument (NanoDrop Technologies Inc., Wilmington, DE, USA). The 260/280 nm absorption ratio of all samples was between 1.8 and 2.0.
Bacterial community diversity and composition in each ileal sample was determined by high-throughput sequencing of microbial 16S rDNA genes. Using a previously published protocol [22], DNA was amplified by PCR with primers 515F (5 0 -GTGCCAGCMGCCGCGGTAA-3 0 ) and 806R (5 0 -GGACTACHVGGGTWTCTAAT-3 0 ), which target the V4 region of the 16S rRNA gene. The reverse primer contained a 6-bp error-correcting barcode unique to each sample. DNA samples were sent to a commercial service provider (Novogene, Beijing, China) for pyrosequencing on an Illumina MiSeq platform according to the manufacturer's instructions. Raw data were obtained, before being screened and assembled using the QIIME [23] and FLASH [24] software packages. Sequencing reads were assigned to samples based on the barcodes. Reads flagged as chimeric were removed to form an "effective sequences" collection for each sample. The QIIME software package and UPARSE pipeline were used to analyze these effective sequences and determine operational taxonomic units (OTUs) [25]. Subsequently, the UCLUST algorithm [23] was employed to cluster sequences into OTUs with an identity threshold of 97%. Each OTU was assigned to a taxonomic level with RDP Classifier [26]. The sequences obtained in the present study were deposited in the National Center for Biotechnology Information Sequence Read Archive under accession numbers SRR4156412 to SRR4156415.

SCFA and BCFA analyses
Straight-chain fatty acids, namely acetate, propionate, butyrate, and valerate, and BCFAs, namely isobutyrate and isovalerate, were analyzed as described previously [27]. To ensure their homogeneity, intestinal samples were freeze-dried using a vacuum freeze-dryer (ALPHA 2-4/LSC; Martin Christ, Osterode am Harz, Germany) at −80˚C. Our preliminary data indicate that freeze-drying has little effect on the concentration of organic acids in biological samples (S1 Table). Briefly, the freeze-dried samples (0.5-0.6 g) were placed in 10-mL centrifuge tubes, mixed with 8 mL double-distilled H 2 O, homogenized, and centrifuged in sealed tubes at 7,000 × g at 4˚C for 10 min. The resulting supernatant (0.9 mL) was mixed with 0.1 mL 25% metaphosphoric acid solution in a sealed 2-mL tube, and left to stand at 4˚C for over 2 h, before being centrifuged at 20,000 × g at 4˚C for 10 min. The supernatant was then passed through a 0.45-μm polysulfone filter and analyzed on an Agilent 6890 gas chromatograph (Agilent Technologies, Inc., Palo Alto, CA, USA) connected to a flame ionization detector and a 1.82 m × 0.2 mm (length × internal diameter) glass column packed with 10% SP-1200/1% H 3 PO 4 on 80/100 Chromosorb W/AW (HP, Inc., Boise, ID, USA).

Statistical analyses
Clustering and determination of alpha and beta diversity were performed in QIIME [23]. Apparent relative abundance at the phylum and genus levels, alpha diversity indices of bacterial communities, and ileal luminal SCFA concentrations were analyzed using a completely randomized design with a general linear model implemented in SAS (SAS Institute, Inc., Cary, NC, USA). Principal coordinate analysis (PCoA) of overall microbial community diversity based on an unweighted UniFrac metric was performed by the Bray-Curtis distance method to compare all samples. Rarefaction curves were created using Excel 2010 (Microsoft, Redmond, WA, USA). Phyla and genera with relative abundances below 0.5% in sows of both diet groups were excluded from further analysis. Differences were deemed statistically significant when associated with a P-value < 0.05. P < 0.10 was considered to indicate a trend toward significance.

DNA sequence coverage and alpha diversity of bacteria from ileal contents
To assess the impact of diet and pregnancy stage on bacterial communities, sequences of the 16S rRNA gene V4 region were amplified. A total of 1,050,719 sequences (42,028.76 ± 2,520.12 per sample) were obtained, including 43,195, 42,229, 41,653, and 41,356 raw reads acquired from samples in the HN diet group on days 45 and 75 of pregnancy and those in the LN diet group at the same time points, respectively. After trimming, assembly, and quality filtering, 41,167, 40,116, 39,601, and 39,007 sequences from samples in these groups, respectively, were selected for further analysis. Considering all samples, sequence read number ranged from 32,169 to 43,472 per sample, with an average of 39,895. The average sequence read length after primer removal was 253 bp. A total of 31,460 effective sequences were extracted from each sample for comparisons at the same sequencing depth. Overall, 3,381 OTUs were detected according to a nucleotide sequence identity of 97% between reads (S2 Table).
Based on normalized subsamples of 31,460 reads per sample, rarefaction curves showed that the selected sequences were sufficient to determine the majority of bacterial diversity parameters (Fig 1). Indices of community richness (Chao1 and ACE) and diversity (Shannon and Simpson indices), with cut-off values of 0.03, are shown in Table 2. None of these measures were significantly affected by diet or pregnancy stage, although a trend toward fewer OTUs in the later stage of pregnancy (P = 0.083) was observed ( Table 2).

Bacterial community composition in ileal contents
In total, genetic material from 37 bacterial phyla was identified across all ileal samples. There were six phyla with a relative abundance greater than 0.5% in at least one experimental group: Actinobacteria, Bacteroidetes, Firmicutes, Proteobacteria, Spirochaetes, and Tenericutes. Of these six phyla, Firmicutes predominated in all samples, with a relative abundance of 69.99-82.22%, followed by Proteobacteria, at 5.82-15.17%. From day 45 to 75 of pregnancy, Firmicutes abundance exhibited a decreasing trend (P = 0.069), whereas levels of Proteobacteria tended to increase (P = 0.053). Diet marginally affected the presence of Tenericutes  (P = 0.079). However, as for the effect of pregnancy stage on Firmicutes and Proteobacteria, this change was not statistically significant (Table 3).
Of the 17 genera with a relative abundance greater than 0.5% in at least one of the experimental groups, only Lactobacillus was significantly affected by diet (P < 0.05). Sows fed the HN diet displayed higher numbers of sequences assigned to this genus than those given LN feed. A similar, though not statistically significant (P = 0.096), effect was observed in relation to Megasphaera. Both of these genera belong to the phylum Firmicutes. No statistically significant effect of pregnancy stage on the relative abundance of these bacteria was evident. The effect of the interaction between diet and pregnancy stage trended towards statistical significance for Pseudomonas (P = 0.080) and Sutterella (P = 0.084), members of the phylum Proteobacteria (Table 4). OTUs present at <0.5% in any samples were excluded from the analysis. Of those in the phylum Firmicutes, levels of OTU-2 and OTU-906 (genus: Lactobacillus), OTU-45 (species: Lactobacillus delbrueckii), and OTU-460 (genus Mitsuokella) were significantly altered by diet (P < 0.05). All of these organisms as well as OTU-29 (species Megasphaera elsdenii, P = 0.077) were more abundant in sows fed the HN diet than in those fed the LN diet. In contrast, the presence of OTU-4 (genus Clostridium, P = 0.078), OTU-38 (order Clostridiales, P = 0.053), and OTU-2219 (family Clostridiaceae, P = 0.091) in sows given the HN diet was lower than in those fed the LN diet. Concerning pregnancy stage, levels of OTU-45 (species Lactobacillus delbrueckii, P < 0.05), OTU-906 (genus Lactobacillus, P < 0.05), and OTU-28 (family Streptococcaceae, P = 0.051) tended to be higher on day 45 than on day 75. In addition, OTU-10 (family Enterobacteriaceae, P = 0.080) and OTU-184 (genus Bacteroides, P = 0.067) were less abundant on day 45 than on day 75 of pregnancy (Table 5). A significant effect of the interaction between diet and pregnancy stage was noted for OTU-45 (species Lactobacillus delbrueckii, P < 0.05) and OTU-906 (genus Lactobacillus, P < 0.05). This effect was also evident to a certain degree, close to reaching statistical significance, for OTU-460 (genus Mitsuokella, P = 0.089).
Principal component analysis at the genus level (Fig 3) revealed a tendency for each experimental group to form a distinct cluster, although some overlap was apparent. Ten dominant OTUs contributing to the variation between these groups were determined. OTU-1 and OTU-4, identified as Turicibacter spp. and Clostridium spp., respectively, were partly responsible for separating the HN diet group on day 75 of pregnancy from the other treatment groups. OTU-5 (family Peptostreptococcaceae), OTU-8 (genus Clostridium), OTU-10 (family Enterobacteriaceae), OTU-12 (family Peptostreptococcaceae), and OTU-2219 (family Clostridiaceae) partially distinguished the LN diet group on day 75 of pregnancy and the HN diet group on day 45 of pregnancy from the three other experimental groups. OTU-1376, OTU-11, and OTU-2 were identified as Lactobacillus spp., Bifidobacterium spp., and Lactobacillus spp., respectively. In addition, samples from the LN diet group on day 45 of pregnancy were more widely dispersed compared with those of other groups.
Partial least squares-discriminant analysis of bacterial OTUs in ileal contents, using diet and pregnancy stage as factors. The 10 OTUs principally responsible for the separation of samples are shown as follows: OTU-1, genus: Turicibacter, phylum: Firmicutes; OTU-2, genus: Lactobacillus,

Concentrations of SCFA and BCFA in ileal contents
The total SCFA concentration was not significantly affected by diet. Of the individual straightchain fatty acids and BCFAs tested, only propionate was influenced by any of the experimental factors, its concentration being significantly higher (P < 0.05) on day 45 of pregnancy than on day 75 (Table 6).

Discussion
Owing to the potentially important roles of the intestinal microbiota in swine health and growth performance, its composition and metabolic activities in various physiological and nutritional contexts deserve close attention [28]. Changes in the composition of the intestinal (including bacterial profiles of the digesta and mucosa) and fecal microbiota have been demonstrated in pigs from birth to finishing phases [29-30] and in pregnant animals [31]. To the best of our knowledge, the present study represents the first analysis of changes in the ileal luminal bacterial profiles of Huanjiang mini-pigs. In this study, on average, 39,895 effective reads were obtained for each sample, with high coverage (>99.08%). In general, alpha diversity indices were not influenced by diet or pregnancy stage. Coverage was marginally lower on day 45 of pregnancy compared to day 75. However, this difference was not statistically significant. The PCoA of overall diversity indicated that differences among individual pigs became greater as gestational age increased, as did indices of alpha diversity. The ileal microbial communities of pregnant Huanjiang minipigs were dominated by Firmicutes (69.99-82.22% of the total microbial content) and Proteobacteria (5.82-15.17%). This observation was consistent with the findings of Isaacson and Kim [32]. In the present study, Clostridium (11.00-21.75%), Lactobacillus (4.82-29.83%), and Turicibacter (12.47-19.16%) were the dominant bacterial genera in the ileal contents of Huanjiang mini-pigs, in accordance with previous surveys of the porcine ileal digesta-and mucosa-associated microbiota [33]. Collectively, these data support the assertion that Firmicutes constitutes the dominant phylum in the gut microbiota of mammals, including mice and humans [34].
Lactobacillus species are known for their potentially beneficial effects on gut function and health [35]. In our experiments, species of this genus were relatively more common in sows fed the HN diet (containing 11% soybean meal, a highly digestible plant protein source), which is consistent with a previous report that pigs fed normal levels of protein exhibit a greater abundance of Lactobacillus in the cecum compared with those given lower-protein feed [36]. Our previous in vitro studies indicated that soybean oligosaccharides (SBOS), major bioactive components of soybean meal, can be selectively fermented by commensal bacteria present in the colon, thus improving gut microbiota balance and modulating metabolism [37]. Dietary SBOS supplementation increases SCFAs, but decreases protein-derived catabolites in the intestinal luminal contents of weaned Huanjiang mini-piglets, which may have beneficial effects on the gut [27]. In addition, our results agree with a previously published report describing elevated numbers of lactobacilli in the ilea of pigs given a barley-based diet compared to those fed primarily on corn [28]. At the OTU level, the presence of OTU-45 (species: Lactobacillus delbrueckii) and OTU-906 (genus: Lactobacillus) decreased from day 45 to day 75 of pregnancy. Lactobacillus species have been associated with weight change in humans and animals [38]. However, the mechanism by which these microbes induce body weight loss or gain remains unclear.
Regarding other Firmicutes taxa, Megasphaera abundance was slightly lower in pigs fed the HN diet than in those given LN feed. Overall, our findings are similar to those of Pedersen et al. [39], who found bacteria of this phylum to be more abundant in the terminal ilea of obese pigs. Moreover, excess energy intake, obesity, and glucose intolerance are associated with increased presence of Firmicutes in humans [40,41].
In the present study, OTU-1, OTU-2, OTU-4, OTU-5, OTU-8, OTU-12, and OTU-1376 were among the 10 OTUs whose relative abundances distinguished, to a certain extent, species composition under different experimental conditions, i.e., diet group and pregnancy stage. All of these belonged to the phylum Firmicutes. In addition, considering it as a single indicator, this phylum was more affected by pregnancy stage than by diet.
Representatives of the phylum Proteobacteria were significantly less abundant than those of Firmicutes. Proteobacteria presence tended to be increased in the later stage of pregnancy (day 75). Koren et al. [15] showed that the relative abundance of Proteobacteria in fecal samples from pregnant women is higher in the third trimester of pregnancy than in the first. Moreover, a significant increase in the abundance of Proteobacteria has been associated with gastrointestinal inflammation in response to environmental and genetic factors [42], as observed in inflammation-associated dysbioses [43]. Shin et al. reported that members of Proteobacteria constitute a microbial signature of gut microbiota dysbiosis [44]. In the present study, this phylum was prominent and primarily represented by the genus Sutterella (0.20-1.48%). The abundance of OTU-10, a family (Enterobacteriaceae) within Proteobacteria, was also higher on day 75 than on day 45 of pregnancy. Several studies reported that active inflammatory bowel disease is associated with significantly elevated levels of Proteobacteria (members of Enterobacteriaceae in particular) [45]. Therefore, our results are compatible with suggestions that the structure and composition of bacterial communities in pregnant hosts are reminiscent of disease-associated dysbiosis.
Previous investigations indicated an association between raised Firmicutes/Bacteroidetes ratios and obesity [46]. Clostridium bacteria are also suspected to play a role in energy harvesting because they are found at higher levels in obese individuals than in people with low body weights [35]. Our prior study showed that the average back-fat thickness of pregnant Huanjiang mini-pigs in both HN and LN diet groups increases from day 45 (27.20 and 26.90 mm, respectively) to day 75 (36.60 and 28.10 mm, respectively) post-service [18]. The live body weights of sows in these treatment groups also increase over this period, from 73.82 and 67.52 kg to 86.14 and 75.28 kg, respectively (Kong et al., unpublished data). These data suggest that the sows became obese during pregnancy, especially those in the HN group. In the present work, the level of Firmicutes tended to be lower on day 75 than on day 45 of pregnancy. Bacteroidetes levels were stable between the two measured time points in the HN diet group, but were higher during the later stage of pregnancy in sows fed the LN diet. Therefore, the Firmicutes/Bacteroidetes ratio was lower on day 75 than on day 45 of pregnancy in both diet groups.
The presence of Prevotella species positively correlates with the proportion of carbohydrates in the diet. In a previously published study in which samples were clustered according to the prevalence of dietary components, representatives of this genus were found to be more abundant in a "carbohydrate" than in a "fat-protein" cluster [47]. In the current investigation, ileal levels of OTU-184 (genus: Bacteroides) tended to be higher on day 75 than on day 45 of pregnancy in Huanjiang mini-pigs. Bacteria of the genus Prevotella were more abundant in pigs fed the LN diet, whereas pregnancy stage did not significantly affect their numbers. This is consistent with observations of elevated Prevotella levels in goats fed a diet with reduced grain content [48]. In summary, the ratio of Bacteroides to Prevotella was less affected by the LN diet than by the HN diet. The abundance of Proteobacteria and Bacteroidetes in the ileal contents samples was higher on day 75 than on day 45 of pregnancy, suggesting an increase in these bacteria between the first and second trimesters.
Together, the results of the present study showed that bacteria of the phylum Firmicutes and genus Lactobacillus decreased in number, while those of the phylum Proteobacteria, family Enterobacteriaceae, and genus Bacteroides increased from day 45 to 75 of pregnancy. These changes in the gut microbiota are similar to those observed in inflammatory bowel disease, during which, the numbers of several species within Firmicutes are reduced, including those of Lactobacillus. Moreover, Enterobacteriaceae is among the Proteobacteria families whose levels appear to be consistently increased in this condition [49,50].
SCFAs are produced by the microbiota of the large intestine from both indigestible carbohydrates [51] and certain amino acids originating from partially digested dietary and endogenous proteins [52]. SCFAs regulate colonic physiology, metabolism, and gene expression [53]. They are also produced in the small intestine, but at concentrations lower than those in the colon, with the exception of acetate [5]. These molecules are produced via fermentation of indigestible polysaccharides by saccharolytic bacteria [54]. For example, species of Ruminococcus, Oscillospira, Clostridium, and Pseudobutyrivibrio metabolize fiber, while those of Prevotella metabolize hemicellulose, producing acetate and propionate [55]. In our study, the concentration of propionate in the ilea of sows was higher on day 45 than day 75 of pregnancy, which may be related to significant changes in Bacteroides abundance during this period. Notably, propionate is known to inhibit the synthesis of lipids from acetate [56].
In the large intestine, concentrations of BCFAs, bacterial metabolites produced exclusively from amino acids, are generally lower than those of SCFAs [57]. BCFAs such as isobutyrate and isovalerate are breakdown products of fermentation by proteolytic bacteria, including members of Bacteroides and Clostridium [58]. Various Clostridium, Bacillus, Lactobacillus, and Streptococcus species as well as many Proteobacteria species play major roles in the utilization of amino acids by their hosts [59].

Conclusions
The ileal bacterial profiles of Huanjiang mini-pigs were dominated primarily by Firmicutes and Proteobacteria, and, in particular, representatives of the genera Lactobacillus, Clostridium, and Turicibacter. The effects of varying animal feed on the composition and diversity of the large intestinal microbiota also need to be elucidated, considering their potential importance. The HN diet was associated with a higher Lactobacillus abundance in pregnant Huanjiang mini-pigs. Since this diet differs from the LN diet in various characteristics (more digestible energy, higher relative protein content, and lower relative crude fiber content), it was not feasible in the current study to identify one particular dietary component associated with modifications in bacterial communities. In addition, the rice bran fiber and alfalfa used in both diets, but in inverse proportions, are characterized by different fiber type compositions [60,61]. Members of the phylum Firmicutes and genus Lactobacillus decreased, while those of the phylum Proteobacteria, family Enterobacteriaceae, and genus Bacteroides increased in number from day 45 to 75 of pregnancy. Notably, changes in bacterial community structure (e.g., increased number of Proteobacteria) as pregnancy progressed were similar to those observed in disease-associated dysbiosis (e.g., morbid obesity), indicating the need for further studies on a possible causal link between these parameters.