Assembly and Analysis of Concurrent XDR and HMV K. pneumo Substrains Using Nanopore Sequencing

The emergence of extensively drug-resistant (XDR) bacteria is a serious medical and public health concern. Both study and treatment of these infections is hampered by lack of information provided by classic testing... [ view full abstract ]

The emergence of extensively drug-resistant (XDR) bacteria is a serious medical and public health concern. Both study and treatment of these infections is hampered by lack of information provided by classic testing methodologies. Here we present our work using nanopore sequencing to characterize a concurrent metastatic infection of a highly virulent hypermucoviscous (HMV) K. pneumoniae strain and OXA-181-producing XDR K. pneumoniae strain from a patient with recent hospitalization in India. Klebsiella pneumoniae, a Gram-negative, lactose-fermenting member of the Enterobacteriaceae family that is responsible for a number of infections including urinary tract infections (UTIs) and pyogenic liver abscesses. During a two month stay at Johns Hopkins, the patient was sampled multiple times and treated with a varying course of antibiotics. We obtained 12 samples in total from patient blood, sputum, kidney abcess, endo/nasal swab and even environmental samples from the patient’s hospital room.

From these samples, we extracted DNA using MoBio PowerBiofilm extraction kits with bead beating to overcome the hypermucoviscous nature of the samples. We subsequently generated Oxford Nanopore libraries for 9 of the isolates and Illumina libraries for all 12 isolates.  We sequenced the nanopore libraries on a range of chemistries (R7.3 and R9.4), granting different yields and error rates, but with a fragment size range of around ~5kb.  Illumina libraries were generating using a tagmentation method (Nextera XT) then sequenced on a MiSeq using a 2x300 v3 kit and multiplexing 3 samples per run.

Using the nanopore sequencing data, we assembled the genomes and plasmids with canu, followed by correction with Illumina MiSeq data using pilon.  The increased yield available from the newest chemistry allowed our most recent run to assemble an isolate genome into 4 contigs.  As an alternative, we assembled using Illumina MiSeq data only with SPAdes. We annotated the assemblies with prokka, then identified plasmids and mutations in the chromosome to parse the separate strains in the patient. To narrow in on antimicrobial resistance genes, we BLASTed our assemblies using the CARD database. From the genomic data, we generated a phylogenic tree using parsnp to cluster the different isolates and trace the mutation profile during the 2 month hospitalization.

Finally, we leveraged the real-time sequencing available from nanopore sequencing to demonstrate the potential speed of detection and analysis for antimicrobial resistance.  As each DNA fragment translocates throught the nanopore (at a current top speed of 450bp/s), the read is sequenced and the data available.  In a retrospective analysis of our data, we found that reads which BLASTed as positive for CTX-M-15 and OXA-181 completed sequencing within 15 minutes after start of the sequencing run.  This rapid turnaround could prove advantageous for diagnosis and treatment.