Genomic Diversity of Burkholderia pseudomallei Isolates, Colombia

We report an analysis of the genomic diversity of isolates of Burkholderia pseudomallei, the cause of melioidosis, recovered in Colombia from routine surveillance during 2016–2017. B. pseudomallei appears genetically diverse, suggesting it is well established and has spread across the region.


Materials and Methods
Through laboratory-based surveillance activities, 11 Burkholderia pseudomallei isolates were received by the microbiology group of the Instituto Nacional de Salud in Colombia during 2016-2017. Cultures from blood, sputum, urine, abscesses, and throat swabs generated as part of routine diagnostic procedures were processed according to the protocols of the clinical laboratory of each hospital. We performed preliminary identification of isolates and susceptibility tests using a VITEK 2 (Biomerieux, https://www.biomerieux-usa.com). Isolates that we identified as Burkholderia spp., oxidase positive, gram-negative, and non-Pseudomonas aeruginosa bacteria, were further tested by MALDI-TOF MS (Bruker, https://www.bruker.com) (1).
Six isolates presumptively identified as B. pseudomallei or Burkholderia spp. were sent to the U.S. Centers for Disease Control and Prevention (CDC) for confirmatory testing, whole genome sequencing, and genetic analysis. DNA from an additional 5 B.
pseudomallei isolates were also sent to CDC for sequencing and genetic analysis. Colombia has previously reported 20 cases as sporadic, isolated events in a few geographic areas. The departments with melioidosis cases from this study are noted on the map in the Appendix Qubit v4.0 fluorometer (https://www.thermofisher.com). We eluted samples in PCR-grade water and RNase A, filtered through a 0.1 µm filter, and checked for sterility before whole genome sequencing (11).
We determined isolate sequences from paired-end Illumina reads which were generated on an Illumina MiSeq or iSeq 100 (https://www.illumina.com). We sheared genomic DNA to a mean size of 600 bp using a Covaris LE220 focused ultrasonicator (https://www.covaris.com). We cleaned DNA fragments with a Beckman Coulter Ampure system (https://www.beckmancoulter.com) and used them to prepare dual-indexed sequencing libraries using NEBNext Ultra library prep reagents (New England Biolabs, https://www.neb.com) and barcoding indices synthesized in the CDC Biotechnology Core Facility for the genomes run on the MiSeq. Libraries were analyzed for size and concentration, pooled, and denatured for loading onto the flow cell for cluster generation.
We used 2 × 250 bp cycle paired-end sequencing kits to perform sequencing for the Illumina MiSeq. We used a Nextera Flex kit (Illumina) to produce libraries for the iSeq 100 runs, which we performed using 2 × 150 bp cycle paired-end sequencing kits. On completion, sequence reads were filtered for read quality, base called, and demultiplexed using bcl2fastq, version 2.19 (Illumina). We generated assemblies as previously described and assessed them with QUAST v5.0 (https://github.com ; 12,13). Features of the genome assemblies are noted in Appendix Table 1.
We submitted genomes to the B. pseudomallei MLST website (http://pubmlst.org/bpseudomallei) to identify the sequence types or assign new sequence type identifiers, as needed (14,15). We analyzed core SNPs for the genomes from Colombia using Parsnp in the Harvest 1.3 suite (https://github.com) along with a reference panel previously described, plus genomes associated with the Western Hemisphere that have recently become available (11,(16)(17)(18)(19). The Colombian genomes had an average of 3,822 SNPs in nonprotein-encoding (intergenic) positions compared with K96243; 2.1 × more SNPs were observed in genes that had no predicted amino acid changes (Appendix Table   2). The dendrogram was generated in MEGA 7 (https://www.megasoftware.net) (20). SNP effects of the Colombian isolates compared with the K96243 reference strain were predicted with SnpEff v4.3t (https://github.com; 21).