*2.2. Sample Collection*

The study was conducted during the fall–winter seasons of 2011–2012. The volunteers provided samples of feces before treatment (basal, T0) and after one month (T1) of isoflavone supplementation. Fresh stools were collected in sterile plastic containers and kept under anaerobic conditions in jars containing Anaerocult A (Merck, Darmstadt,

Germany) for transporting to the laboratory within 2 h. Fecal samples were kept frozen at −80 ◦C until use.

#### *2.3. Total Bacterial DNA Extraction*

Fecal samples (0.2 g) were suspended in 1.8 mL of phosphate buffered saline (PBS) (pH 7.4). These suspensions were homogenized and centrifuged at 800 rpm for 5 min at 4 ◦C to eliminate insoluble material, and the supernatants were transferred to new tubes. These were then centrifuged again at 14,000 rpm for 5 min at 4 ◦C. Pelleted cells were suspended in 1 mL of PBS and lysed in an enzyme solution containing 20 mM TRIS-HCl pH 8.0, 2 mM EDTA, 1.20% Triton X-100, 20 mg/mL lysozyme (Merck), and 20 U mutanolysin (Sigma-Aldrich, Saint Louis, MO, USA). Total bacterial DNA was extracted following the protocol described by Zoetendal [22], and purified using the QIAamp DNA Stool Minikit (Qiagen, Hilden, Germany). Finally, the DNA was eluted in 100 μL of sterile molecular grade water (Sigma-Aldrich), and its concentration and quality were determined using an Epoch microvolume spectrophotometer (BioTek Instruments, Winooski, VT, USA).

#### *2.4. Library Construction and Pyrosequencing*

A segmen<sup>t</sup> of the 16S rRNA genes from the purified bacterial DNA were PCRamplified using the universal primers Y1 (5-TGGCTCAGGACGAACGCTGGCGGC-3 ) (position 20–43 on the 16S rRNA gene of Escherichia coli) and Y2 (5-CCTACTGCTGCCTCC CGTAGGAGT-3) (positions 361–338). These primers amplify a 348 bp stretch of the prokaryotic rDNA embracing the V1 and V2 hypervariable regions. Further, 454-adaptors were included in both the forward (5-CGTATCGCCTCCCTCGCGCCATCAG-3) and reverse (5-CTATGCGCCTTGCCAGCCCGCTCAG-3) primers, followed by a 10 bp samplespecific barcode. Amplifications were performed using the NEBNext High-Fidelity 2x PCR Master Mix Kit (New England Biolabs., Ipswich, MA, USA) as follows: 95 ◦C for 5 min, 25 cycles of 94 ◦C for 30 s, 60 ◦C for 45 s, 72 ◦C for 30 s, and a final extension step at 72 ◦C for 5 min. The amplicons produced were purified using the GenElute PCR Clean-Up Kit (Sigma-Aldrich), and their concentration was measured in a Qubit fluorometer with dsDNA assay kits (Thermo Fisher Scientific Inc., Waltham, MA, USA).

An amplicon library was prepared for pyrosequencing by mixing equal amounts of amplicons from the different samples. Pooled amplicons were then sequenced using a 1/8 picotitre plate in a 454 Titanium Genome Sequencer (Roche, Indianapolis, IN, USA) in the UNC Microbiome Core (University of North Carolina, USA).

#### *2.5. Sequence and Data Analysis*

Raw sequences were denoised and filtered out of the original dataset. Filtering and trimming were performed using the Galaxy Web Server [23], employing the sliding window method. Only reads longer than 150 bp were used in further analysis. Chimeras were eliminated using the USEARCH v.6.0.307 clustering algorithm routine in de novo mode [24]. After demultiplexing, high quality rDNA sequences were classified taxonomically using the Ribosomal Database Project (RDP) Bayesian Classifier [25] with an 80% confidence threshold to obtain the taxonomic assignment and relative abundance of the different bacterial groups. "Genus" was the lowest taxonomic level contemplated. Sequences with at least 97% similarity were clustered into operational taxonomic units using the CD-Hit clustering method [26] and employed in the generation of rarefaction curves using a RarefactWin freeware (produced by S. Holland; http://strata.uga.edu/software/index. html). Different diversity indexes (Sobs, Chao, ACE, Jackknife, Shannon, Simpson) were calculated for each sample and compared between groups of women [27]. As diversity index values increase with sample size, normalization of sequencing effort in all samples was necessary to avoid biases in the results [28]. Thus, diversity indexes were normalized using the median number of sequences obtained in all samples as a scaling factor [29]. Weighted UniFrac analysis [30] was performed to assess the similarity of the microbial

communities between samples and principal coordinates analysis (PCoA) was applied to the distance matrix for visualization.

#### *2.6. Fatty Acids (FAs) Determination*

One hundred microliters of a 1:10 dilution of feces (*w*/*v*) in PBS was supplemented with 100 μL of 2-ethyl butyric acid (Sigma-Aldrich, St. Louis, MO, USA) as an internal standard (1 mg/mL in methanol) and acidified with 100 μL of 20% formic acid (*v/v*). The acidic solution was then extracted with 1 mL of methanol and centrifuged for 10 min at 15,700× *g*. Supernatants were kept at −20 ◦C until analysis in a 6890 N gas chromatography (GC) apparatus (Agilent Technologies, Santa Clara, CA, USA) connected to a flame ionization detector (FID). All samples were analyzed in duplicate and FAs were quantified as previously described [31].
