3.2. Processing and DNA Hybridization

Genomic DNA was extracted from archived stool isolates at each collection site using the following: the boil method (used in Peru [112]), Masterpure DNA and RNA Complete Purification Kit (Epicentre Biotechnologies, Madison, WI, USA; used in Egypt), R.E.A.L. Prep Kit (used in Kenya), DNeasy Blood and Tissue Kit (used in USA), or QIAamp DNA mini (used in Cambodia) (last three kits from QIAGEN, Germantown, MD, USA; USA), according to manufacturers’ instructions.

With the exception of Cambodia, all subsequent steps were performed in NRL facilities (amplification, labeling, hybridization, microarray interrogation) as previously described [113]; all steps for preparation and analysis of Cambodian samples were performed on-site in NAMRU-2 facilities (Phnom Penh, Cambodia). Each sample was amplified using 10 ng of template DNA and Illustra GenomiPhi v.2 DNA Amplification Kits (GE Healthcare, Pittsburgh, PA, USA), as per the manufacturer’s instructions. After quantification using Qubit 2.0 fluorometer (Thermofisher, Rockland, IL, USA), 2 µg of the whole genome amplicons were fragmented for 1 min at 37 °C with 2.7 units DNase I (total volume of 60 µL; enzyme and buffer from GeneChip Resequencing Assay Kit, AffyMetrix, Santa Clara, CA, USA), incubated for 10 min at 95 °C to inactivate the DNase I, then purified on DNA Clean & Concentrator-5 columns (Zymo Research, Irvine, CA, USA). Fragmented, purified DNA was then biotinylated using ULS Platinum Bright Biotin Nucleic Acid Labeling Kits (Kreatech Diagnostics, Durham, NC, USA; 10 µL reaction volume), as per the manufacturer’s instructions. The resulting biotinylated fragments were then applied to pre-hybridized ARDM v.2 microarrays (Customarray, Bothell, WA, USA [39]), hybridized overnight at 60 °C, and labeled with 1000 × diluted multimeric streptavidin-horseradish peroxidase (65R-S104PHRP; Fitzgerald Industries, North Acton, MA, USA), washed, processed, and electrochemically interrogated using the ElectraSense reader (Customarray), as previously described [39]. The content of the ARDM v.2 microarray includes 25- to 35-mer probes directed against 238 gene sequences (between 8 and 10 probes per gene) predicted to confer resistance to 15 categories of antimicrobials: β-lactams (n = 46 genes), aminoglycosides (n = 42), macrolides (n = 27), lincosamides (n = 22), streptogramins (n = 18), quaternary amines (n = 2), ansamycins (n = 1), diaminopyrimidines (n = 28), antimicrobial peptides (n = 1), tetracyclines (n = 38), phenicols (n = 10), glycopeptides (n = 12), platensimycin/platensin (n = 1), fluoroquinolones (n = 4), sulfonamides (n = 3). Many of the macrolide, lincosamide, and streptogramin ARDs overlap in specificity. Samples with > 85–90% gene sequence identity to the reference sequence can successfully hybridize to the microarray, allowing a broader variety of ARDs—i.e., families of genes—to be detected. However, the microarray is unable to distinguish between these similar genes. A gene was deemed present if at least 50% of its representative probes had signals above the 95% probe threshold (mean signal from lowest 2128 probes + 3 SD) or if ≥ 70% of its probes had signals above either of two less stringent thresholds (mean signal from lowest 2016 probes + 3 SD or mean signal from lowest 2128 probes + 2 SD) [39,113]. Using these algorithms, the sensitivity and specificity of the microarray were calculated as 96.3–100% and 97.9–100%, respectively [39,44].