Rapid identification of pathogens, antibiotic resistance genes and plasmids in blood cultures by nanopore sequencing

Abstract

Bloodstream infections (BSI) and sepsis are major causes of morbidity and mortality worldwide. Blood culture-based diagnostics usually requires 1-2 days for identification of bacterial agent and an additional 2-3 days for phenotypic determination of antibiotic susceptibility pattern. With the escalating burden of antimicrobial resistance (AMR) rapid diagnostics becomes increasingly important to secure adequate antibiotic therapy. Real-time whole genome sequencing represents a genotypic diagnostic approach with the ability to rapidly identify pathogens and AMR-encoding genes. Here we have used nanopore sequencing of bacterial DNA extracted from positive blood cultures for identification of pathogens, detection of plasmids and AMR-encoding genes. To our knowledge, this is the first study to gather the above-mentioned information from nanopore sequencing and conduct a comprehensive analysis for diagnostic purposes in real-time. Identification of pathogens was possible after 10 minutes of sequencing and all predefined AMR-encoding genes and plasmids from monoculture experiments were detected within one hour using raw nanopore sequencing data. Furthermore, we demonstrate the correct identification of plasmids and bla_CTX-M subtypes using de novo assembled nanopore contigs. Results from this study hold great promise for future applications in clinical microbiology and for health care surveillance purposes.

Figure 1

Relative distribution of reads in sequence data generated by nanopore sequencing of DNA purified from the three selected blood cultures. The blood cultures were spiked with E. coli A2-39 (A and D), K. pneumoniae A2-37 (B and E) and E. coli A2-39 + K. pneumoniae A2-37 (C and F). Upper panel (A–C) show results obtained using Centrifuge and lower panel (D–F) show results based on BLAST search against the RefProk database that contains prokaryotic sequence data only. The “Others” group represents taxa with relative read counts below 1%. All results are based only on the first output file for each experiment from the MinION sequencing platform, containing 4000 reads (available after approximately 10 minutes of sequencing).

Figure 2

Centrifuge-based relative taxonomic assignments of sequence data from DNA purified from the three selected blood cultures. Samples were spiked with - (A) E. coli A2-39, (B) K. pneumoniae A2-37 and (C) E. coli A2-39 + K. pneumoniae A2-37. The “Others” group shows the average relative amount of all incorrectly assigned (non-target) taxa. Blood cultures spiked with E. coli (A2-39) and with a combination of E. coli (A2-39) and K. pneumoniae (A2-37) were sequenced for 68 and 64 hours respectively.

Figure 3

Overview of the earliest generated reads from each sequencing experiment where bla_CTX-M (yellow) or mecA genes (green) and plasmid-derived sequences (red) were identified based on BLAST search. Blue lines represent length of the nanopore reads, length of hits is illustrated by arrow-length and similarity to the database entry is given on the arrow. The time of read generation is noted on top of each read.

Figure 4

Assembly graphs for E. coli A2-39. (A) Hybrid assembly of nanopore and Illumina data; reads tagged as S. enterica by Centrifuge and bla_CTX-M are highlighted (S. enterica in multicolour, bla_CTX-M in blue). The bla_CTX-M gene was located on a plasmid-derived read which mapped to the IncHI2 plasmid. (B) Unicycler assembled nanopore generated data. (C) SPAdes assembled Illumina generated data.

Figure 5

Genome coverage of target species over time. (A) Blood cultures spiked with monocultures. Reference strain E. coli (NCTC13441) experiment was stopped after 2 hours; at that point it reached 84% of the genome coverage. (B) Blood culture spiked with a combination of E. coli (A2-39) and K. pneumoniae (A2-23) (C) blood culture spiked with a combination of E. coli (NCTC13441) and S. aureus (CCUG35600). S. aureus reached 35% coverage by the end of the sequencing run of 33 hours. Dashed horizontal line denotes 95% genome coverage.

Figure 6

Timeline for the information gathered from nanopore sequencing of positive blood cultures.