Strengthening Bordetella pertussis genomic surveillance by direct sequencing of residual positive specimens

Abstract

Whole-genome sequencing (WGS) of microbial pathogens recovered from patients with infectious disease facilitates high-resolution strain characterization and molecular epidemiology. However, increasing reliance on culture-independent methods to diagnose infectious diseases has resulted in few isolates available for WGS. Here, we report a novel culture-independent approach to genome characterization of Bordetella pertussis, the causative agent of pertussis and a paradigm for insufficient genomic surveillance due to limited culture of clinical isolates. Sequencing libraries constructed directly from residual pertussis-positive diagnostic nasopharyngeal specimens were hybridized with biotinylated RNA "baits" targeting B. pertussis fragments within complex mixtures that contained high concentrations of host and microbial background DNA. Recovery of B. pertussis genome sequence data was evaluated with mock and pooled negative clinical specimens spiked with reducing concentrations of either purified DNA or inactivated cells. Targeted enrichment increased the yield of B. pertussis sequencing reads up to 90% while simultaneously decreasing host reads to less than 10%. Filtered sequencing reads provided sufficient genome coverage to perform characterization via whole-genome single nucleotide polymorphisms and whole-genome multilocus sequencing typing. Moreover, these data were concordant with sequenced isolates recovered from the same specimens such that phylogenetic reconstructions from either consistently clustered the same putatively linked cases. The optimized protocol is suitable for nasopharyngeal specimens with diagnostic IS481 Ct < 35 and >10 ng DNA. Routine implementation of these methods could strengthen surveillance and study of pertussis resurgence by capturing additional cases with genomic characterization.

Keywords: Bordetella pertussis; metagenomics; surveillance; whole-genome sequencing.

Fig 1

Schematic overview of the CIWGS workflow, which utilizes a RNA bait library for target enrichment of B. pertussis genome fragments within a sequencing library prepared directly from DNA extracts of residual NP species used for diagnostics. (A) RNA baits are prepared from a diverse mixture of B. pertussis isolates (Table S1), and target enrichment yields significantly more sequencing reads containing B. pertussis genomic data than shotgun metagenomics. (B) Less than 4% of annual reported pertussis cases is culture positive. This approach increases case coverage for genomic surveillance.

Fig 2

B. pertussis genomic data recovery from mock specimens spiked with reducing concentrations of isolate DNA following either 1× or 2× enrichments. (A) The fraction of sequencing reads containing B. pertussis (green), human (blue), or other (black) genomic data. (B) Coverage breadth at 20× depth across a B. pertussis reference genome assembly at 1× (black) or 2× (gray) enrichments.

Fig 3

B. pertussis genomic data recovery from pooled negative clinical specimens spiked with dilutions of isolate cells following either 1× or 2× enrichments. (A) The fraction of sequencing reads containing B. pertussis (green), human (blue), or other (black) genomic data. (B) Coverage breadth at 20× depth across a B. pertussis reference genome assembly following 1× (black) or 2× (gray) enrichments. (C) wgMLST allele calls following 1× (black) or 2× (gray) enrichments.

Fig 4

Phylogenetic reconstruction of matched CIWGS (red squares) and isolate (black circles) sequences from 29 surveillance specimens, reconstructed with 124 core, variable sites using maximum likelihood.