The Impact of Dietary Grape Seed Meal on Healthy and Aflatoxin B1 Afflicted Microbiota of Pigs after Weaning

The study investigated the effect of grape seed (GS) meal, aflatoxin (AFB1), or their combination on the large intestine microbiota of weanling piglets. Twenty-four piglets were allocated into four groups based on diet composition: (1) Control group; (2) AFB1 (320 g/kg feed) group; (3) GS group (8% inclusion in the diet); (4) AFB1 + GS group. After 30 days of experiment, the colon content was used for microbiota analyses; after isolation of total bacterial genomic DNA, V3/V4 regions of the 16S rRNA amplicons were sequenced using the Illumina MiSeq platform. The raw sequences were analyzed using the v.1.9.1 QIIME pipeline software. 157 numbers of OTUs were identified among all four dietary groups with 26 of them being prevalent above 0.05% in the total relative abundance. GS and AFB1 increase the relative abundance of phylum Bacteroidetes and Proteobacteria, while decreasing the Firmicutes abundance in a synergic manner as compared with the individual treatments. An additive or synergistic action of the two treatments was identified for Lactobacillus, Prevotella and Campylobacter, while rather an antagonistic effect was observed on Lachnospira. The action mechanisms of aflatoxin B1 and grape seed meal that drive the large intestine microbiota to these changes are not known and need further investigations.


Introduction
Aflatoxins are an important group of mycotoxins mainly produced by the Aspergillus flavus and parasiticus species, which contaminates a large quantity of the world's crops, colonizing cereals (especially maize) and other important animal and human food [1,2]. Extensive studies show that aflatoxin B1 is the most harmful of the aflatoxin group, due to its potent toxicity and immune toxicity when is metabolically activated in the liver [2,3]. Chronic exposure to low levels of aflatoxins is a risk factor for hepatocellular carcinoma; based on toxicological data AFB1 has been categorized as a group 1 human carcinogen by the International Agency for Cancer Research [4]. A pig with a diet rich in cereals is very exposed to aflatoxins [5]. Extreme effects can lead to death, but the greatest impact comes from reduced reproductive potential, suppressed immune function, reduced performance, and various pathological effects on organs and tissues [6]. At lower concentrations, AFB1 induced depressed appetite, lower growth rate and lower feed conversion efficiency, decreased proteosynthesis, beneficial effect in the absence of gut microbiota that can catabolize flavonoids in various kinds of catabolites using a large enzymatic equipment [23].
In recent years, next generation sequencing technologies have revealed all kinds of intricate connections among gut-microbiota, dietary composition and host health [24][25][26]. In this 3-way relationships, oral exposure to xenobiotics or dietary composition could lead to the modulation of gut microbiota, and the changes of gut-microbiota may further influence host health in a significant way [27].
Changing the gut microbiota metabolites, either through changing nutritional habits or by changing the microbiome itself, can also have the potential to mitigate a number of metabolic diseases [17,28]. After our knowledge there are no studies concerning the effect of the concomitant administration of aflatoxin and grape waste on microbiota composition.
The aim of this study was to investigate the AFB1 as a contaminant in feed for its effects at the level of microbiota and whether grape seed meal had the potential to modulate by itself the dynamics of microbiota in pig after weaning and to counteract the possible negative effects on microbiota of pigs fed diet contaminated with AFB1.

Body Weight and Diarrhea Score
Piglets fed for 30 days with a diet contaminated with AFB1 registered a significantly lower body weight (p < 0.05) when compare either with control or GS group (Figure 1). This effect was counteracted by the inclusion of 8% of GS meal into the diet contaminated with AFB1. Reported to the entire period of the experiment only 2.78% of the control piglets had diarrhea. In change, AFB1 was responsible for a higher percentage of piglets with diarrhea (16.67%), while both GS and GS + AFB1 piglets had almost no diarrhea ( Figure 2).
Toxins 2019, 11, x FOR PEER REVIEW 3 of 16 absorbed in the upper gastrointestinal tract [22]. Polyphenols cannot exert their beneficial effect in the absence of gut microbiota that can catabolize flavonoids in various kinds of catabolites using a large enzymatic equipment [23].
In recent years, next generation sequencing technologies have revealed all kinds of intricate connections among gut-microbiota, dietary composition and host health [24][25][26]. In this 3-way relationships, oral exposure to xenobiotics or dietary composition could lead to the modulation of gut microbiota, and the changes of gut-microbiota may further influence host health in a significant way [27].
Changing the gut microbiota metabolites, either through changing nutritional habits or by changing the microbiome itself, can also have the potential to mitigate a number of metabolic diseases [17,28]. After our knowledge there are no studies concerning the effect of the concomitant administration of aflatoxin and grape waste on microbiota composition.
The aim of this study was to investigate the AFB1 as a contaminant in feed for its effects at the level of microbiota and whether grape seed meal had the potential to modulate by itself the dynamics of microbiota in pig after weaning and to counteract the possible negative effects on microbiota of pigs fed diet contaminated with AFB1.

Body Weight and Diarrhea Score
Piglets fed for 30 days with a diet contaminated with AFB1 registered a significantly lower body weight (p < 0.05) when compare either with control or GS group ( Figure 1). This effect was counteracted by the inclusion of 8% of GS meal into the diet contaminated with AFB1. Reported to the entire period of the experiment only 2.78% of the control piglets had diarrhea. In change, AFB1 was responsible for a higher percentage of piglets with diarrhea (16.67%), while both GS and GS + AFB1 piglets had almost no diarrhea ( Figure 2).

16S rRNA Gene Sequencing and Metagenomic Analysis
The metagenomic analysis of gut microbiota revealed a total of 1,154,427 reads with the lowest sample count of 24,903 and the highest count of 93,622 and an average of 50,192 reads per sample of dietary group, respectively. To remove sampling depth heterogeneity, rarefaction on the OTU table was performed with a cut-off of 24,903, the lowest number of reads recovered in a single sample which was used as the sample depth. More than 90% of sequences from each sample passed quality   The metagenomic analysis of gut microbiota revealed a total of 1,154,427 reads with the lowest  sample count of 24,903 and the highest count of 93,622 and an average of 50,192 reads per sample of  dietary group, respectively. To remove sampling depth heterogeneity, rarefaction on the OTU table  was performed with a cut-off of 24,903, the lowest number of reads recovered in a single sample which was used as the sample depth. More than 90% of sequences from each sample passed quality filtering, showing that the sequencing method is both robust and efficient.

Alpha Diversity
Dietary groups containing Grape Seed Meal, Aflatoxin B1 or both did not result in altered microbial community richness and diversity compared to the control group. The indices inspected included Chao1 (species richness) and PD_whole_tree (phylogenetic diversity units). The lack of influence on richness and diversity was evident at all phylogenetic levels ( Figure 3A,B). There were no major community shifts caused by dietary treatments showed in either unweighted (p = 0.2019) or weighted (p = 0.4817) UniFrac ( Figure 4A,B respectively).

16S rRNA Gene Sequencing and Metagenomic Analysis
The metagenomic analysis of gut microbiota revealed a total of 1,154,427 reads with the lowest sample count of 24,903 and the highest count of 93,622 and an average of 50,192 reads per sample of dietary group, respectively. To remove sampling depth heterogeneity, rarefaction on the OTU table was performed with a cut-off of 24,903, the lowest number of reads recovered in a single sample which was used as the sample depth. More than 90% of sequences from each sample passed quality filtering, showing that the sequencing method is both robust and efficient.

Alpha Diversity
Dietary groups containing Grape Seed Meal, Aflatoxin B1 or both did not result in altered microbial community richness and diversity compared to the control group. The indices inspected included Chao1 (species richness) and PD_whole_tree (phylogenetic diversity units). The lack of influence on richness and diversity was evident at all phylogenetic levels ( Figure 3A,B). There were no major community shifts caused by dietary treatments showed in either unweighted (p = 0.2019) or weighted (p = 0.4817) UniFrac ( Figure 4A and 4B respectively).

Beta Diversity
The Unifrac method, of unweighted and weighted taxon abundances, was used for the construction of the principal coordinate analysis (PCoA) to enable the examination and visualization of the beta diversity within the colon. The Unifrac being phylogeny-based, Weighted-UniFrac takes into account the relative abundance of species/taxa shared between samples. Weighted UniFrac is

Beta Diversity
The Unifrac method, of unweighted and weighted taxon abundances, was used for the construction of the principal coordinate analysis (PCoA) to enable the examination and visualization of the beta diversity within the colon. The Unifrac being phylogeny-based, Weighted-UniFrac takes into account the relative abundance of species/taxa shared between samples. Weighted UniFrac is also useful for examining differences in community structure, which aligned with our study goal. Explaining the low variance of the two axes (22.43% and 10.19%) from the weighted plot can be done considering the multitude of factors that can influence the samples ( Figure 5).

Beta Diversity
The Unifrac method, of unweighted and weighted taxon abundances, was used for the construction of the principal coordinate analysis (PCoA) to enable the examination and visualization of the beta diversity within the colon. The Unifrac being phylogeny-based, Weighted-UniFrac takes into account the relative abundance of species/taxa shared between samples. Weighted UniFrac is also useful for examining differences in community structure, which aligned with our study goal. Explaining the low variance of the two axes (22.43% and 10.19%) from the weighted plot can be done considering the multitude of factors that can influence the samples ( Figure 5).
The phylum Bacteroidetes was statistically and progressively increased in AFB1 and GS group when compared to control. The Firmicutes phylum was also significantly decreased by AFB1 and GS; the concomitant administration of AFB1 + GS have a synergic or additive effect on Firmicutes phylum.

Comparisons between Community Compositions
The most highly represented phyla within the large intestine microbiota of the weaning piglets in the control group were Firmicutes (58.33%), and Bacteroidetes (34.51%), while phyla present in low abundance (<0.01%) included Proteobacteria (1.2%), Spirochaetes, Tenericutes, and Actinobacteria, with traces of Euryachaeota, Fibrobacteres, Cyanobacteria, and Deferribacteres ( Figure 6). The phylum Bacteroidetes was statistically and progressively increased in AFB1 and GS group when compared to control. The Firmicutes phylum was also significantly decreased by AFB1 and GS; the concomitant administration of AFB1 + GS have a synergic or additive effect on Firmicutes phylum.
In comparison, in the AFB1 group, Firmicutes had a 33.2% presence and Bacteroidetes 54.4%, in GS group, Firmicutes accounted for 32.3% and Bacteroidetes 60.3% and in the AFB1 + GS group, Firmicutes had just a 12.5% presence while the Bacteroidetes reached 73.8%. (Figure 7). The relative abundance of the microbiota at the genus level identified Prevotella (21.93%), Lactobacillus (24.8%), Lachnospira (0.89%) and Campylobacter (0.55%) to be the most abundant in colon. In this study, there were also identified Megasphaera, Anaerovibrio, Trembayales, and Clostridiaceae ( Figure 8). As can be seen in the Figure 9, grape seed meal had a stimulating effect on Prevotella relative abundance, and GS seems to act synergistic with AFB1, as the highest relative abundance was observed in the AFB1 + GS1 (48.1%) group (p < 0.05) when compared to the Control (25.3%) and GS   As can be seen in the Figure 9, grape seed meal had a stimulating effect on Prevotella relative abundance, and GS seems to act synergistic with AFB1, as the highest relative abundance was observed in the AFB1 + GS1 (48.1%) group (p < 0.05) when compared to the Control (25.3%) and GS group (39.45%) As shown in Figure 9, both AFB1 and GS significantly impacted (p < 0.05) the genus of Lactobacillus present in the colon (19% and respectively 17.54%), and they synergistically decrease the Lactobacillus abundance from 50.5% in control to 3.8% in the AFB1 + GS dietary group.
The Lachnospira genus growth was stimulated significantly (p < 0.05) by AFB1 (4.53% vs. 1.69% in control). GS administration doesn't affect the relative abundance of Lachnospira (1.93%) as compared with control. However, a synergic effect on Lachnospyra abundance was obtained when AFB1 and GS were administered together. Aflatoxin B1 could also account for a significant growth of 3.03% (p < 0.05) of the Campylobacter genus in AFB1 group and of 3.8% in AFB1 + GS group while the grape seed meal had no noticeable effect compared to control (0.84% and 1% respectively).
Generally speaking the AFB1 impacted the relative abundance of the large intestine microbiota when compared to the control as seen in the heatmap from Figure 10. AFB1 stimulated some of the bacteria species such as Campylobacter, Lachnospira, and Megasphaera, bacteria from the Tremblayales order, the Paraprevotelaceae and Erysipelotrichaceae microbial families while suppressing bacteria like Lactobacillus and Prevotella, and some bacteria from the Lachnospiraceae family. On the other hand, the grape seed meal had an almost inverse effect to the AFB1 stimulating the growth of several bacterial species. It positively impacted the growth of Prevotella, Anaevibrio, and Megasphaera, Lachospira, some bacteria from the order Clostridiales and Tremblayales while suppressing the rest (Figure 10).

Grosu et
In the GS and AFB1 group the highest value of relative abundance was noted for Prevotella and Campylobacter, a modulation of the Paraprevotelaceae family and Bacteroidales and Tremblayales order, while diminishing Lachnospira and Lactobacillus to their lowest extent.

Discussions
The present study has demonstrated that the dynamics of the microbiota was changed by grape seed meal, aflatoxin B1 or their combinations and we have identified a complex microbiota in the large intestine of the piglets. This was consistent with the scientific reports which show that large intestine of pig has a high richness and diversity of microbiota [29]. In a comprehensive study published by Kim and collaborators [30] on the pig gastrointestinal microbiome, from a phylum level perspective, most bacteria were classified in five phyla: Firmicutes, Bacteroidetes, Proteobacteria, Actinobacteria, and Spirochaetes. In this study, bacteria from Firmicutes phylum represented the largest proportion of the total population followed by Bacteroidetes and these results were similar with our data concerning the distribution of the colon microbiota in piglets. Other studies have shown that these two phyla account for approximately 86-95% of all bacteria present in the large intestine [31].
The most abundant genus identified in the colon of control and experimental piglets from our study were represented by Prevotella, Lactobacillus, Lachnospira and Campylobacter. Also, the bioinformatic analysis allowed the identification of Megasphaera, Anaerovibrio, Trembayales, and Clostridiaceae.
Most of the bacteria identified in the colon content samples are known to play important roles in the metabolisms of carbohydrates, nucleotides, and amino acids [32]. Our results are similar with other studies that focused on the identification of the core fecal microbiota genus in the colon of the pigs, represented by: Clostridiales, Ruminococcaceae, Bacteroidetes, Escherichia, Firmicutes, Lactobacillus, Erysipelotrichaceae, Lachnospiraceae [33].
Grape seed meal is rich in polyphenols [34], but only 5-10% of the total polyphenol intake is absorbed in the small intestine, the remaining polyphenols (90-95% of total polyphenol intake) being subject to the enzymatic activities of the gut microbial community [35]. There is evidence from in vitro animal and human studies that phenolic compounds alter gut microbiota and, consequently, alter the Bacteroides/Firmicutes balance [36,37]. The Firmicutes to Bacteroidetes ratio was linked to the general health of vertebrates [38].
In our study, the phylum Bacteroidetes was statistically and progressively increased as compared with control group after the addition of GS in the pig diet, while the Firmicutes phylum was significantly decreased by GS. Many studies have shown that dietary polyphenols and their metabolites contribute to the maintenance of gut health by the modulation of the gut microbial balance through the stimulation of the growth of beneficial bacteria and the inhibition of pathogen bacteria, exerting prebiotic-like effects [35]. Thus, it was shown that grape seed extracts may modulate the intestinal microbiota, producing changes in beneficial bacteria such as Lactobacillus spp. but inhibiting other groups such Clostridium spp. in both in vivo and in vitro studies [9,37]. However, in our study, GS has induced a significant decrease of Lactobacillus, Lachnospiraceae, Bacteroidales, and Campylobacter, while positively impacted other species like Megasphera, Clostridiales, Anaerovibrio, or Prevotella. The higher relative abundance of Megasphaera, Prevotella, and bacteria belonging to Clostridiales order that were found in pigs fed GS diet might have beneficial effects on the host due to the implications of these species in the carbohydrate metabolic chain and the synthesis of short chain fatty acids [39]. Prevotella, Megasphaera, Anaerovibrio represents strictly anaerobic bacteria that have been identified as a dominant species in the large intestine of pigs, and are also abundant in the ileum [29,40,41]. These bacteria are known to be a dominant complex carbohydrate degrading bacterial group in the lower gastrointestinal tract [42] and their increase in the colon of piglet fed a GS diet, might result in response to the higher fiber content of the diet enriched with grape seed meal.
A decrease of Lactobacillus abundance as well as an increase in relative abundance of Clostridium was already described in a recent study in rats fed with grape pomace extract rich in phenolic compounds [43]. Lactobacillus spp. is a beneficial bacterium that enhances gut barrier function, stimulates the host immune system, prevents diarrhea or allergies, contributes to activation of provitamins, and modulates lipid metabolism [44]. An increase or a reduction of growth of Lactobacillus was reported in several in vitro and in vivo studies depending on the precursor or the metabolite used. While some phenolic compounds (gallic acid, isoflavones, proanthocyanidins, and resveratrol) increase the relative abundance of Lactobacillus, other compounds, (catechin, epicatechin, quercetin) had no significant effect or even decrease the Lactobacillus abundance [45]. Studies on pigs fed grape pomace (GP) diets showed that GP decreases the relative abundance of Lactobacillus and Ruminococcus [46]. Also, the grape seed proanthocyanidins decreased the abundance of Lactobacillaceae and increased the abundance of Clostridiaceae in both ileal and colonic lumen [47].
Similar with our results, other in vivo studies using grape pomaces have shown an increase of Clostridiales abundance. For example, proanthocyanidins induce an increase of Clostridiales in pigs fed a diet containing 1% grape seed extract [48] and Clostridium coccoides was increased by a wide variety of polyphenolic compounds both in vivo and in vitro [49][50][51].
The Clostridia belong to the class of Firmicutes and represent obligate anaerobes bacteria, including Clostridium and other similar genera [41]. Clostridium represent normal constituents of the intestinal flora, with different implication for animal health as some members of the Clostridiales family, as Clostridium difficile are associated with negative implications, such as inflammatory bowel disease [52], while others as C. leptum and C. coccoides are important members of the gut flora by promoting healthy aging [53].
Intoxication of pigs with AFB1 was associated with reduced feed intake and body weight gain, impaired liver and immune functions and altered serum biochemical parameters [54]. Indeed, our results have shown a significant decrease in the body weight of AFB1 intoxicated piglets. Even that diarrhea was less associated with aflatoxin B1 intoxication when compared with other mycotoxin as deoxynivalenol [55], our study has shown an increase of the number of pigs with diarrhea after the AFB1 administration as compared with the control. Also, in our experiment, intoxication with AFB1 induced compositional changes in microbial community, consisting in an increase of the phylum Bacteroidetes, while decreasing the Firmicutes phylum abundance, but the changes were opposites when compared with the changes induced by GS diet. Similar results were obtained also in rats exposed to different concentration of AFB1, species whose compositions significantly decreased among the top 20 OTUs belong mainly to Firmicutes, with an important decrease of Lactobacillus, mainly of Streptococcus sp. and Lactococcus sp. [1]. When looking to the species affected by the AFB1 exposure, the toxin induced a decrease of Prevotella and Lactobacillus in the colon of intoxicated piglets, while increasing Camplyobacter, Lachnospira and some members of Erysipelotrichaceae, Bacteroidales or Tremblayales families. Earlier studies have shown that the antimicrobial spectrum of aflatoxin B1 is narrow and limited and AFB1 was found to be inactive against common Gram-positive and Gram-negative bacteria at a concentration of 100 µg/mL [56]. Thus, a general antimicrobial effect cannot explain the alterations of microbiota found in the colon of intoxicated animals. It rather looks that other more complex mechanisms can be involved and it was suggested that the gut microbiota of a host could be intentionally modified to accumulate toxin-tolerant species, a potential natural biological intervention strategy for mitigation of AFB1-induced toxic effect [1]. For example, it has been reported that some lactic acid bacteria can remove AFB1 or have protective effects against AFB1 [57,58]. Indeed, an increase of AFB1 concentration in feces was observed in intoxicated mice receiving L. plantarum as compared with mice receiving only AFB1 [59].
Less piglets with diarrhea were observed in the group fed AFB1 + GS diet as compared with AFB1 group, suggesting that GS was responsible for a shift in the microbiota that results in a reduction of diarrhea. Concomitant administration of GS and AFB1 caused significant shift in the microbial community by significantly increasing the relative abundance of phylum Bacteroidetes and Proteobacteria, while decreasing the Firmicutes abundance in a synergic or additive manner as compared with the individual treatments. Again, an additive or synergistic action of the two treatments was identified for Lactobacillus, Prevotella, and Campylobacter, while rather an antagonistic effect was observed on Lachnospira. Lachnospira decrease observed in the colon of piglets fed both AFB1 and GS can be correlated with the reduction of diarrhea incidence as compared with AFB1 group, as Lachnospira is positively correlated with diarrhea in piglets [60]. Also, it was shown that the abundance of Prevotella was increased in healthy pigs compared with pigs with diarrhea and the higher abundance of Prevotellaceae family in healthy pigs may provide an adequate prevention strategy for pathogen infection [61]. However, our microbiota results cannot entirely explain the beneficial effect of grape seed meal, and other microbiota-independent mechanisms might be responsible for the positive effects of the grapeseed meal in weaned piglets. Indeed, our preliminary data have shown that the inclusion of grape seed in the diet of AFB1 intoxicated pigs restored toward the control group the level of the phase-II antioxidant enzymes activity and total antioxidant capacity, decreased TBARS level, and had the potential to ameliorate the pro-inflammatory cytokines concentration and the performance of AFB1-treated animals (unpublished data).
In conclusion, piglets fed for 30 days with a diet contaminated with AFB1 registered a significantly lower body weight and a high frequency of diarrhea. These effects were counteracted at least partially by the inclusion of 8% GS meal into the diet contaminated with AFB1. In the analyses concerning large intestine microbiota 157 numbers of OTUs were identified among all four dietary groups with 26 of them being prevalent above 0.05% in the total relative abundance. Most of the OTUs were identified as bacteria involved in polysaccharides (hemicellulose and pectin, host glycan, and α-glucan) and carboxylic and amino acid metabolism. Concomitant administration of GS and AFB1 caused significant shift in the microbial community by significantly increasing the relative abundance of phylum Bacteroidetes and Proteobacteria, while decreasing the Firmicutes abundance in a synergic or additive manner as compared with the individual treatments. Again, an additive or synergistic action of the two treatments was identified for Lactobacillus, Prevotella and Campylobacter, while rather an antagonistic effect was observed on and Lachnospira. The action mechanisms of aflatoxin B1 and grape seed meal that drive the large intestine microbiota to these changes are not known and need further investigations. Twenty-four healthy weaned crossbred TOPIGS-40 hybrid piglets, with an average body weight of 9.13 ± 0.03 kg were allocated into four groups of 6 piglets based on diet composition as follows:

Animals and Diets
(1) Control group fed a standard diet; (2) AFB1 group fed a diet contaminated with 320 µg/kg AFB1; (3) GS group fed a diet supplemented with a 8% grape seed meal; (4) AFB1 + GS group fed a diet containing both 320 µg/kg AFB1 and 8% grape seed meal. The grape seed meal provided by a local commercial (S.C. OLEOMET-SA S.R.L., Bucuresti, Romania) was included in the diet by corn replacement. The total polyphenol content and identification of different classes of polyphenols of grape seed meal was measured by the Folin-Ciocalteu reaction and HPLC-DAD-MS and gas chromatography as described by [62]. The polyphenols composition of the grape seed meal is presented in Table 1. To prepare the toxin contaminated diet, 50 mg AFB1 (FERMENTEC, Jerusalem, Israel) was dissolved in DMSO (dymethil sulfoxide) and mixed into basal diet to provide a feed diet containing 320 µg/kg; final AFB1 concentration in feed was verified by ELISA using a Veratox kit (Neogen, Lansing, MI, USA). The content of other mycotoxins that frequently contaminate cereals (DON, ZEA, OTA) was under the detection limit as resulted from the ELISA assay.
The animals were individually identified by ear tag and housed in pens (two replicates of 3 pigs per pen per treatment) and fed the experimental diets for 30 days. They had free access to feed and water every day of the experimental period. The piglets with diarrhea were recorded every day of the experiment. All diets were formulated to meet specific requirements for weaning feed as indicated by NRC (2012)- Table 2. At the end of the experiment pigs were weighted and then slaughtered. The content of the colon was immediately frozen and stored at −80 • C until use.

Microbial DNA Extraction and 16S rRNA Gene Sequencing
Microbial DNA was extracted using the QIAamp DNA stool minikit (QIAGEN, Dusseldorf, Germany) following the manufacturer protocol additionally increasing the temperature to 95 • C to improve the DNA yield. Sample DNA concentrations were verified on agarose gel electrophoresis and on the Nanodrop Spectrophotometer and adjusted to 10 ng/µL with a volume of 100 µL for each sample for a total of 24 samples.
Microbial profiling was performed by BMR Genomics (Padova, Italy) using the Illumina Miseq platform using 300PE approach on 16S amplicons derived from the bacterial DNA. The library formation and sequencing of the 16S rRNA genes were carried out according to the method above. In the first step the V3 and V4 regions of the amplicons were amplified with universal primers: forward primer: 5 -CCTACGGGAGGCAGCAGT-3 and reverse primer 5 -GACTACCAGGGTATCTAATCCTGTT-3 [63] and forward primer: 5 -CCTACGGGNBGCASCAG-3 and reverse primer: 5 -GACTACNVGGGTATCTAATCC-3 [64]. In the second amplification step, the adaptors and indexes were bound to the sequences. The pooling of the samples was followed by a cluster generation and sequencing with the Paired End 2 × 300 pb format.

Microbiota Bioinformatics
For the microbial raw data, the sequences obtained were investigated using a subsampled open reference OTU (operational taxonomic unit) picking method with default settings performed in QIIME (v1.9.1). The 16S rRNA gene reads were then paired-end and demultiplexed. Subsampled open-reference analyzable samples were clustered into Operational Taxonomic Units. The OTU-picking strategy was carried out by using UCLUST algorithm using a de-novo strategy with a 97% sequence similarity threshold. Representative sequences for each OTU were set in a matchup with the Greengenes database V13_8 and taxonomy was designated through the application of the UCLUST method with a 90% confidence threshold. Data was chimera checked using the Blast fragments approach in QIIME. The raw reads in QIIME were filtered for a length ≥70 and an average quality ≥20. In order to eliminate sampling depth heterogeneity, the OTU table was rarefied to the lowest number of reads obtained from a single sample. Furthermore, the OTU table obtained in the previous step was abundance filtered by removing the OTUs with a relative abundance ≤0.005% across all samples in order to keep only the most abundant species for further data processing.
A phylogenetic tree was constructed by firstly aligning the representative sequences of each OTU using PyNAST algorithm against Greengene reference sequences. Above mentioned procedures were implemented through QIIME with customized commands. The OTU table and phylogenetic tree were the major elements for data analysis.
The diversity was estimated with Chao 1 and Faith's phylogenetic diversity (PD_whole_tree) observed-species indices (alpha or within-sample diversity) and double principal component analysis (DPCoA; beta or between-sample diversity) using the phylogeny-based weighted Unifrac distance matrix.
A heatmap was also built around the OTU table of the species that were found above a 0.005% relative abundance.

Statistical Analysis
Analyses were performed using XLstat software package (http://www.xlstat.com). Comparison between groups was performed using one-way analysis of variance (ANOVA) with the post hoc Tukey test for multiple comparisons at the p < 0.05 level. ANOSIM is an internal statistical method (script) of the QIIME which was used to test the differences between two or more samples. The ANOSIM test is nonparametric and the statistical significance is derived through permutations.