Diet, Environments, and Gut Microbiota. A Preliminary Investigation in Children Living in Rural and Urban Burkina Faso and Italy

Diet is one of the main factors that affects the composition of gut microbiota. When people move from a rural environment to urban areas, and experience improved socio-economic conditions, they are often exposed to a “globalized” Western type diet. Here, we present preliminary observations on the metagenomic scale of microbial changes in small groups of African children belonging to the same ethnicity and living in different environments, compared to children living on the urban area of Florence (Italy). We analyzed dietary habits and, by pyrosequencing of the 16S rRNA gene, gut microbiota profiles from fecal samples of children living in a rural village of Burkina Faso (n = 11), of two groups of children living in different urban settings (Nanoro town, n = 8; Ouagadougou, the capital city, n = 5) and of a group of Italian children (n = 13). We observed that when foods of animal origin, those rich in fat and simple sugars are introduced into a traditional African diet, composed of cereals, legumes and vegetables, the gut microbiota profiles changes. Microbiota of rural children retain a geographically unique bacterial reservoir (Prevotella, Treponema, and Succinivibrio), assigned to ferment fiber and polysaccharides from vegetables. Independently of geography and ethnicity, in children living in urban areas these bacterial genera were progressively outcompeted by bacteria more suited to the metabolism of animal protein, fat and sugar rich foods, similarly to Italian children, as resulted by PICRUSt (Phylogenetic Investigation of Communities by Reconstruction of Unobserved States), a predictive functional profiling of microbial communities using 16S rRNA marker gene. Consequently, we observed a progressive reduction of SCFAs measured by gas chromatography–mass spectrometry, in urban populations, especially in Italian children, respect to rural ones. Our results even if in a limited number of individuals point out that dietary habit modifications in the course of urbanization play a role in shaping gut microbiota, and that ancient microorganisms, such as fiber-degrading bacteria, are at risk of being eliminated by the fast paced globalization of foods and by the advent of westernized lifestyle.

Amplicon Quantitation, Pooling and Pyrosequencing. Amplicon DNA concentrations were determined using the Quant-iT PicoGreen dsDNA reagent and kit (Invitrogen) following the manufacturer's instructions. Assays were carried out using 10 µl of cleaned PCR product in a total reaction volume of 200 µl in black, 96-well microtiter plates. Fluorescence was measured on Perkin Elmer Victor Plate reader using the 485/530 nm excitation/emission filter pair with measurement time 0.1 second. Following quantitation, cleaned amplicons were combined in equimolar ratios into a single tube. The final pool of DNA was precipitated on ice for 45 minutes following the addition of 5 M NaCl (0.2 M final concentration) and two volumes of ice-cold 100% ethanol. The precipitated DNA was centrifuged at 7,800 g for 40 minutes at 4°C, and the resulting pellet was washed with an equal volume of ice-cold 70% ethanol and centrifuged again at 7,800 g for 20 minutes at 4°C. The supernatant was removed and the pellet was air dried for 10 minutes at room temperature and then resuspended in 100 µl of nuclease-free water (Ambion). The final concentration of the pooled DNA was determined using a NanoDrop spectrophotometer (Thermo Fisher). Pyrosequencing was carried out using primer A on a 454 Life Sciences Genome Sequencer FLX instrument (Roche) following Titanium chemistry. Data Analysis. Data analysis integrated together the previous data obtained for BR and EU populations (De Filippo, Cavalieri et al. 2010)  Reads of all data sets were pre-processed using the MICCA pipeline (version 1.5, http://compmetagen.github.io/micca/) (Albanese, Fontana et al. 2015). Forward and reverse primer trimming and quality filtering were performed using micca-preproc truncating reads shorter than 280nt (quality threshold=18). Denovo sequence clustering, chimera filtering and taxonomy assignment were performed by micca-otu-denovo (parameters -s 0.97 -c). Operational Taxonomic Units (OTUs) were assigned by clustering the sequences with a threshold of 97% pair-wise identity, and their representative sequences were classified using the RDP software version 2.7 (Wang, Garrity et al. 2007). Template-guided multiple sequence alignment was performed using PyNAST57 (version 0.1) (Caporaso, Bittinger et al. 2010) against the multiple alignment of the Greengenes 16S rRNA gene database (DeSantis, Hugenholtz et al. 2006) filtered at 97% similarity. Finally, a phylogenetic tree was inferred using FastTree (Price, Dehal et al. 2010) and micca-phylogeny (parameters: -a template-template-min-perc 50). Sampling heterogeneity was reduced by rarefaction, obtaining 12`964 sequences per sample.
Bacterial species were assigned, based on Basic Local Alignment Search Tool nucleotide (BLASTn) software in the National Center for Biotechnology Information (NCBI) database, considering the highest percentage of identity (Query cover 100%-99% and Identity 99% or 95%). Expectation value (E-value) was used to select significant BLAST hits, keeping only outcomes with the lowest E-value, given a minimal E-value of 10 −3 (generally, significant match when the E-value is close to zero).Chao1 index and Shannon entropy (indicators of alpha diversity) and UniFrac (Lozupone, Lladser et al. 2011) and Bray-Curtis dissimilarities (indicators of beta diversity) were calculated using the phyloseq package (McMurdie and Holmes 2014) of the R software suite. Exploratory analysis was performed by Principal coordinates analysis (PCoA) using the phyloseq package of the R software suite. Multiple-rarefaction PCoA plots ("jackknifed" PCoA plots) (Lozupone, Lladser et al. 2011) were computed to assess the robustness of the beta-diversity analyses.
The significance of between-groups differentiation on the UniFrac distances and Bray-Curtis dissimilarity was assessed by PERMANOVA using the adonis() function of the R package vegan with 999 permutations.
To compare the relative abundances of OTUs among the four groups, the two-sided, unpaired Wilcoxon test was computed, removing taxa not having a relative abundance of at least 0.1%, in at least 20% of the samples, and using the function mt() in the phyloseq library and the p-values were adjusted for multiple comparison controlling the family-wise Type I error rate (minP procedure).
Based on sequence abundances in each population, heatmap plots of percentage abundances, at different taxa, were obtained by using STAMP (Parks, Tyson et al. 2014), and supported by dendogram, obtained with Average Neighbour and Unweighted Pair Group Method with Arithmetic Mean (UPGMA), useful to cluster fecal samples of the children populations based on taxa abundances.
Based on the relative abundances, the metagenomic biomarker discovery and related statistical significance were assessed using the linear discriminant analysis (LDA) effect size (LEfSe) method (Segata, Izard et al. 2011). LEfSe uses the Kruskal-Wallis rank-sum test to identify features with significantly different abundances between assigned taxa compared to the groups, and LDA to estimate the size effect of each feature. An alpha significance level of 0.05, either for the factorial Kruskal-Wallis test among classes or for the pairwise Wilcoxon test between subclasses, was used.
A size-effect threshold of 2.0 on the logarithmic LDA score was used for discriminative microbial biomarkers.
To infer the functional contribution of microbial communities on 16S rDNA sequencing data set, we applied PICRUSt (Phylogenetic Investigation of Communities by Reconstruction of Unobserved States) (Langille, Zaneveld et al. 2013), that implements an extended ancestral-state reconstruction algorithm to predict which gene families are present, and then combines gene families to estimate the significant differences in the main functional classes (KEGG categories) of the composite metagenome. From a OTUs table with associated Greengenes identifiers, we obtained the final output from metagenome prediction as an annotated table of predicted gene family counts for each sample, where the encoded functions of each gene family are orthologous groups or other identifiers such as KEGG orthologs (KOs). The functional pathways discovery and related statistical significance were assessed by LEfSe.
In general, PICRUSt maps the subset of 16S sequences to their nearest sequenced reference genome.
To evaluate accuracy of PICRUSt, we used the Nearest Sequenced Taxon Index (NSTI), developed to quantify the availability of nearby genome representatives for each microbiome sample (Supplementary Table 6).

Determination of Short-Chain Fatty Acids (SCFAs) in Fecal Samples. For determination of
SCFASs we used an aliquot of frozen fecal samples (about 250 mg). Briefly, fecal samples were homogenized after addition of 1 ml of 10% perchloric acid and centrifuged at 15,000 g for 5 minutes at 4°C. Concentrations of SCFASs were determined in a 1:25 dilution of 500 µl supernatant. We used 5 µl of a mixture of deuterated acids containing 50 ng D3-propionic, 50 ng D7-butyric and 500 ng D4-acetic acids as internal standard. A calibration curve was prepared, adding the mixture of internal standards (5 µl) to scalar amounts of the acids. SPME-GC-MS determinations were performed using a Varian Saturn 2000 GC-MS instrument with 8200 CX SPME autosampler. The SPME fiber was a Carboxen/Divinylbenzene 75 µm. The capillary column was an Agilent HP-Innowax 30 m × 0.25 mm, 0.5 µm film thickness. The injector and transfer line temperatures were 290°C and 260°C, respectively; the ion trap temperature was 180°C. Absorption of analytes was performed in the headspace of the sample solution for 3 min at 70°C; the analytes were desorbed in the GC injector port at 290°C for 20 min. The GC oven temperature program was as follows: initial temperature 45°C for 0.15 min, then to 123°C at 2°C/min, to 159°C at 6°C/min and to 200°C at 20°C/min. The retention times for individual short-chain fatty acids were determined by injecting each standard into the column. The Varian MS workstation software (version 6.6) was used for data acquisition and processing. The SCFAs concentration in fecal sample was expressed in µmol/g of feces. To determine statistical significance of differences observed among the four populations we used unpaired Student's t test (one tailed).

Microbiota characterization of African children populations compared to Europeans
Among the minor components of gut microbiota (relative abundance <0.05), we observed that

Bacterial species assignment
By BLAST alignment of 16S sequences, we found that the majority of sequences belonging to Bacteroides genus and consistently found in BC and EU metagenome, was attributable to B. uniformis, and in minor abundance to B. acidifaciens, B. caccae, B. coprophilus, B. ovatus, and B. plebeius. In BR and BT populations, although Bacteroides was poorly represented, we found Bacteroides sequences attributable with 95% of identity to B. vulgatus. Interestingly, a study on gnotobiotic interleukin-2-deficient mice showed that B. vulgatus has a protective role against E. coli induced-colitis (Waidmann, Bechtold et al. 2003), suggesting that the probable unique Bacteroides species found in BR microbiota could have a possible protective role against potential pathogenic bacteria.
Regarding Bifidobacterium genus, we observed that BC and EU populations were mainly enriched in B. longum, and in minor part in B. adolescentis, B. bifidum and B. breve.

Functional metabolic profiles of gut microbiota by PICRUSt
In order to evaluate how the observed taxonomic differences between the gut microbiota of African We decided to use this approach based on 16S rRNA inference, instead of whole genome sequencing,

Carbohydrate Metabolism
In the BR metagenome, we found enrichment of carbon fixation pathways and oxidative phosphorylation, metabolic functions that are both related to carbohydrate metabolism and involved in releasing energy, associated with the TCA cycle pathway ( Figure 7A-B).
In the BC metagenome, we observed enrichment of starch and sucrose metabolism, as well as pentose phosphate metabolism, a metabolic pathway parallel to glycolysis that generates NADPH and ribose 5-phosphate, a precursor for the synthesis of nucleotides, and erythrose 4-phosphate (E4P), used in the synthesis of aromatic amino acids. Interestingly, the BC metagenome was also enriched in methane metabolism ( Figure 7B), related to the fermentation of polysaccharides (Danielsson, Werner-Omazic et al. 2014).
In the EU metagenome, over the general pathway related to carbohydrate metabolism, pentose, glucuronate interconversion pathways and C5-branched dibasic acid metabolism, deriving from a simple sugar-rich diet, we found enrichment of galactose metabolism, involved in conversion of galactose into glucose ( Figure 7A). The acquisition of these functions in the EU metagenome could arise from consumption of dairy products. Other functions enriched in the EU gut microbiota were glyoxylate and dicarboxylate metabolism, involved in the biosynthesis of carbohydrates from fatty acids. Sulfur and nitrogen metabolism were also enriched in the EU metagenome ( Figure 7B).
Regarding nitrogen metabolism, several enteric bacteria produce reduced sulfur and nitrogen by dietary amino acids and animal protein.

Lipid Metabolism
Studies on animal models have shown that commensal bacteria can regulate and maintain lipid homeostasis (Brestoff and Artis 2013). Our results reveal differential functional lipid metabolism characterized the metagenomes of the four studied populations ( Figure 7C). In BR children, we found enrichment of arachidonic acid, fatty acid biosynthesis and lipid biosynthesis proteins ( Figure 7C).
In BC children, we observed enrichment of linoleic metabolism, an essential polyunsaturated fatty acid involved in the biosynthesis of arachidonic acid, a key inflammatory intermediate. We also found enrichment of primary and secondary bile acid biosynthesis ( Figure 7C). The synthesis of bile acids is one of the predominant mechanisms for the excretion of excess cholesterol, and is involved in intestinal absorption of fat-soluble vitamins.
In EU children we found an enrichment of glycerophospholipid and fatty acid metabolism and synthesis and degradation of ketone bodies ( Figure 7C), clearly derived from a lipid-rich diet. The synthesis and degradation of ketone bodies could be derived from glucose production from noncarbohydrate sources, and could explain the observed enrichment in butanoate metabolism.

Amino acid Metabolism
In the BR metagenome, PICRUSt analysis showed several enriched amino acid metabolism ( Figure   7D). Commensal bacteria can provide amino acids to the host from both dietary and endogenous proteins. Peptides and amino acids are used as carbon, nitrogen and energy sources by both saccharolytic and non-saccharolytic bacteria. Some saccharolytic species, such as Prevotella, the predominant genus in BR children, are able to derive energy from the carbon skeletons of peptides and amino acids.
Glutamate is the principal source of nitrogen for the host. Bacteria can make the carbon skeletons of all amino acids and transaminate those carbon skeletons with nitrogen from glutamine or glutamate to complete the amino acid structures (Berg, Tymoczko et al., 2002). Interestingly, the observed enrichment in beta and D-alanine metabolism represents a way to produce glucose, especially in fasting conditions. Alanine can then be converted to pyruvate in the liver by the glucose-alanine cycle, as a source of carbon for gluconeogenesis, in order to form glucose that can be used as energy for the muscle.
Another metabolic function found in the BR population is seleno-compound metabolism, deriving from consumption of cereal grains, legumes and soybeans (Whanger 2002) (Figure 7D). Seleno compounds are organometallic molecules comprised of selenium, an essential element involved in reproduction, thyroid hormone metabolism, DNA synthesis, and protection from oxidative damage and infection (Sunde and Thompson 2009).
Surprisingly, metabolism of taurine and hypotaurine was enriched in the BR metagenome ( Figure   7D). Generally, taurine is a major constituent of bile acid, important in emulsifying fat. However, taurine is also derived from cysteine, thus enrichment of cysteine metabolism found in BR children could explain this functional acquisition.
In the BC metagenome, we found enrichment of aromatic amino acid (phenylalanine, tyrosine and tryptophan; Figure 7D). In BC children, we also found enrichment of phosphonate and phosphinate metabolism ( Figure 7D), related to carbon-phosphorous bonds (C-P compounds) biosynthesized by Actinobacteria that were abundant in BC microbiota.
In the EU metagenome, we observed enrichment of metabolism of amino acids such as arginine, proline, valine, leucine, isoleucine, histidine and tryptophan, and lysine biosynthesis and degradation ( Figure 7D). These amino acids can originate from the animal protein-rich food, typical of the Western diet. Concerning essential branched-chain amino acid (BCAAs, such as valine, leucine, and isoleucine) our results are in agreement with a recently analyzed metagenome of an Italian population (Rampelli, Schnorr et al. 2015). It is noteworthy that metabolic function related to BCAAs and aromatic amino acids are enriched in obese compared to lean individuals (Newgard, An et al. 2009), and increased BCAA levels are associated with the risk of developing type 2 diabetes (Wang, Larson et al. 2011).

Cofactors and Vitamin metabolism
Commensal colonic bacteria are a significant source of a range of vitamins to the host (Hill 1997).
Unlike dietary vitamins, the uptake of vitamins derived by microbial metabolism predominantly occurs in the colon (Said and Mohammed 2006). Among this, the B vitamins, including thiamin, riboflavin, niacin, folate, vitamin B6, vitamin B12, biotin and pantothenic acid, cooperate to perform many different processes, such as the release of energy deriving from carbohydrates, proteins and fats, or regulation of immune cells (Brestoff and Artis 2013). The infant microbiota appears to be specialized for the acquisition of nutrients, especially the vitamin B (Yatsunenko, Rey et al. 2012).
In BR metagenome, we found several functional acquisitions related to cofactors and coenzyme involved in oxidative reactions, energy, and carbohydrate and amino acids metabolism, and for DNA and RNA building (Supplementary Figure 6B), such as folate biosynthesis, riboflavin, vitamin B6, retinol metabolism, nicotinate and nicotinamide metabolism. In BC metagenome, lipoic acid and biotin metabolism were enriched (Supplementary Figure 6B). The former is an important cofactor for mitochondrial enzyme complexes in antioxidant reactions. The latter is a coenzyme for carboxylase enzymes, involved in the synthesis of fatty acids, isoleucine, and valine, and in gluconeogenesis.
In EU, we found enrichment in functions related to porphyrin and chlorophyll metabolism (porphyrins are essential cofactors of many proteins including cytochrome, haemoglobin and myoglobin), panthotenate and CoA metabolism, and thiamine, a coenzyme involved in the catabolism of sugars and aminoacids (Supplementary Figure 6B).

Secondary metabolisms
Human-associated bacteria produce also a wide range of natural compounds, such as terpenoids and polyketides (Donia and Fischbach 2015), that include well-characterized mediators of microbe-host and microbe-microbe interactions. In the BR microbiome, we observed a great enrichment of functions related to metabolism of terpenoids, probably due to a high consumption of plant-derived foods, and polyketides (Supplementary Figure 6D). Among the metabolic functions related to biosynthesis of polyketides, many of the commonly used antibiotics, such as tetracycline and macrolides, are produced by polyketide synthases. In the four groups of children, we observed differentially and progressively increased antibiotic biosynthesis functions, passing from rural to urban populations. In particular, we found acquired streptomycin biosynthesis function in the BR metagenome; novobiocin biosynthesis in the BT group; butirosin, neomycin and ansamycin biosynthesis in the BC; and beta-lactamase resistance, penicillin and cephalosporin biosynthesis, and tetracycline biosynthesis in the EU children (Supplementary Figure 6C-D). In a rural environment, antibiotic production by bacteria is a competition mechanism that could provide selective benefit for the producing microorganism. Most antibiotics are derived from biomolecules and secondary metabolites produced by soil-dwelling microorganisms (Davies and Davies 2010); therefore microbiota of rural populations may be acquiring this potential for the unique scope of survival and competition among bacteria. The observed beta-lactamase resistance and cephalosporin biosynthesis acquisitions in the EU metagenome confirm that the use of antibiotics in agriculture, in livestock and medical practice are inducing functional acquisitions related to antibiotic resistance, an emerging and dramatic problem in industrialized and globalized populations.
Regarding the xenobiotics metabolism, in the BR metagenome, we found enrichment of aromatic organic compounds, such as toluene, naphthalene and ethylbenzene degradation (Supplementary Figure 6E). This pathway referred to several compounds containing the benzene ring, including phenolic molecules. As observed in our recent study on functional acquisitions related to xenobiotic degradation by microbiota of red colobus monkeys eating several plant species in the forest of Tanzania (Barelli, Albanese et al. 2015), we may hypothesize that the widespread presence of plants and vegetables rich in tannins and phenolic compounds in rural villages of Burkina Faso, require metabolic acquisition of functions for digestion of xenobiotics by commensal bacteria, rather than the acquired functions related to environmental contaminant degradation.
In urban BC and EU metagenomes, we found enrichment of chloralkene, bisphenol, dioxin and xylene degradation, and benzoate and 1,1,1-trichloro-2,2-bis(4-chlorophenyl)-ethane-(DDT) degradation, respectively (Supplementary Figure 6E). The degradation of these xenobiotics by the microbiota of urban populations can be considered a functional response of bacteria to exposure to toxic compounds derived from industrial and urban environmental pollution (especially chloroalkene and dioxin).
Among benzoate compounds, sodium benzoate, a food preservative known as E211, is widely used in acidic foods, carbonated drinks, jams and fruit juices. Furthermore, several chlorinated organic insecticides are persistent contaminants in the urban environment and potentially in cultivated foods.