Effect of the antibiotics on the intestinal microbiota.

Stool samples collected at baseline and on days 13, 25, 45, and 60 after the first antibiotic dose were used to determine the effect of the antibiotics on the microbiota. Quantitative cultures for facultative and aerobic Gram-negative bacilli were performed for five mice per group at each time point by plating serially diluted specimens onto MacConkey agar (Difco Laboratories, Detroit, MI). Organisms recovered on MacConkey agar were characterized as lactose-fermenting or non-lactose-fermenting Gram-negative bacilli, and for a subset of 10 plates a colony growing at the highest dilution was subjected to identification using standard methods.

Real-time PCR was used to monitor the effect of antibiotic treatment on the concentration of Bacteroides spp. and Prevotella spp. for five mice per group at each time point as previously described (8). Purified template DNA from Bacteroides fragilis (American Type Culture Collection [ATCC] 25825) and Prevotella oris (ATCC 33573) was used for melting-curve analysis and to generate standard curves for each primer set using 10-fold serial dilutions of DNA ranging from 10 to 10⁻⁶ ng. The PCR results were used to calculate CFU per gram of stool as described by Louie et al. (5).

Microbiome analysis was completed for three mice per group at each time point by deep-sequencing bacterial 16S rRNA gene amplicons (25). Briefly, DNA was extracted from ∼200 mg of feces using the QIAamp DNA stool minikit (Qiagen, Frederick, MD) according to the manufacturer's instructions with lysis conditions optimized to increase the ratio of nonhuman to human DNA. The V3-V4 region of the bacterial 16S rRNA gene, defined as the most promising bacterial primer pair for deep-sequencing-based diversity (25), was PCR amplified, and single amplicons of ∼460 bp were visualized by electrophoresis in 1% agarose gels and cleaned up using AMPure XP beads (Agencourt AMPure XP). Dual indices (barcodes) and Illumina sequencing adapters were added to the amplicons using the Nextera XT Index kit (Illumina, Inc., San Diego, CA), followed by DNA purification (Agencourt AMPure XP). Individual barcoded DNA samples were then quantified (Qubit 2.0; Thermo Fisher Scientific), normalized, and pooled. Multiplexed libraries, including 5% PhiX as an internal control, were diluted to 20 pM and denatured with NaOH prior to sequencing on the MiSeq system (Illumina) using the MiSeq reagent Kit v3 600 cycle (2 × 300 bp; Illumina). All 66 samples were multiplexed into a single Miseq sequencing run, generating over 52.9 million quality reads, distributed homogenously among all samples (mean, 441,790 reads; interquartile range, 369,237 to 520,492 reads).

Indexed reads were demultiplexed to generate sample-specific fastq files, which were mapped and aligned against a set of 16S rRNA reference sequences, allowing taxonomic classification based on an Illumina-curated version of the Greengenes database (http://greengenes.lbl.gov/) using the 16S Metagenomics app v.1.0.1 (Illumina, Inc.). The algorithm is a high-performance implementation of the Ribosomal Database Project as previously described (26). The total number of reads that passed quality filtering were clustered into operational taxonomic units (OTU) using a 97% similarity threshold. OTU with a number of sequences lower than 0.005% of the total number of sequences generated were discarded, and those remaining were summarized to get taxonomies.