Direct whole-genome sequencing enables strain typing of unculturable Neisseria meningitidis from oropharyngeal carriage specimens

Abstract

Oropharyngeal carriage of Neisseria meningitidis (N.m.) is a prerequisite for invasive meningococcal disease. As such, genomic surveillance of disease-causing carriage strains can inform targeted public health responses. However, whole-genome sequencing (WGS) from isolates is often precluded due to the high rates of culture failure for N.m. samples collected in carriage studies. This study outlines an alternative method to sequence N.m. directly from oropharyngeal specimens that enables high-resolution molecular fine typing.We performed direct probe-capture enrichment WGS (dWGS) of N.m. on oropharyngeal specimens from the 'B part of it' South Australian and 'B part of it NT' Northern Territory meningococcal carriage studies (NCT03089086 and NCT04398849). Sequences were analysed using currently available bioinformatic tools, including the characterization of genogroup, multi-locus sequence typing (MLST), Bexsero Antigen Sequence Typing (BAST), porA and fetA type.Sensitivity of dWGS typing compared to WGS for genogroup, MLST, porA, fetA and BAST schemes was 88.89%, 72.22%, 100%, 94.44% and 88.24%, respectively. Genogroup and porA type were more reliably characterized in unculturable samples compared to the other typing schemes assessed. Factors that influenced accurate fine typing included the amount and proportion of N.m. sequences, and the proportion of other Neisseria species in enriched sequencing libraries. An alternative phylogenetic method (phylotyping) correctly predicted the clonal complex for 93.46% of the samples assessed. These results demonstrate that dWGS enables high-resolution molecular fine typing and can be applied to unculturable samples in N.m. carriage studies.

Keywords: typing; Neisseria meningitidis; metagenomics; whole-genome sequencing.

Fig. 1.. The 20 highest represented genera in pre-capture metagenomic (mNGS) or post-capture enriched (dWGS) sequencing libraries from oropharyngeal samples with detectable N.m.

Fig. 2.. The percentage of reads mapping to the top ten most prevalent Neisseria species present in pre-capture metagenomic (mNGS) and post-capture enriched (dWGS) sequencing libraries.

Fig. 3.. Pre- and post-capture reads were compared for the Neisseria species that were identified in all samples (n=8) sequenced in parallel by mNGS and dWGS, respectively. (a) Log10 transformed read percentages mapping to Neisseria species in post-capture libraries. (b) Read percentage enrichment ratios for Neisseria species were calculated by the read percentages from post-capture libraries (dWGS) divided by read percentages from pre-capture libraries (mNGS) independently for each species in each sample. Statistics were performed by Welch’s ANOVA with Games–Howell multiple comparisons. Statistical significance is summarized as follows for P-value: > 0.05 [non-significant (ns)], ≤ 0.05 (*), ≤ 0.01 (**), ≤ 0.001 (***), ≤ 0.0001 (****).

Fig. 4.. Factors associated with samples that failed or passed quality control assessment. Samples failed QC if they met the following criteria: the top species identified ≠ ‘Neisseria meningitidis’, or the number of N.m reads≤500,000, or the second species identified=other Neisseria species and the proportion of N.m reads≤40 %. (a) The porA PCR Ct of DNA extract, (b) the percentage of N.m. reads in enriched dWGS libraries identified and (c) the percentage of other Neisseria species reads in enriched dWGS libraries. Statistical comparisons were performed by the Wilcoxon rank sum test.

Fig. 5.. Top three species identified in post-capture enriched dWGS sample libraries that passed QC assessment. Samples failed QC if they met the following criteria: the top species identified ≠ ‘Neisseria meningitidis’, or the number of N.m reads≤500,000, or the second species identified=other Neisseria species and the proportion of N.m reads≤40 %. Species identified in the top three are coloured, and ‘Residual’ (grey) represents all other low-prevalence species identified in these samples.

Fig. 6.. Factors associated with correct identification of MLST type for dWGS compared to isolate WGS for all carriage specimens with a corresponding isolate (n=17). (a) The percentage of reads in enriched dWGS libraries matching Neisseria meningitidis, (b) the percentage of reads in enriched dWGS libraries matching other Neisseria species and (c) porA PCR Ct of DNA extract. Statistics were performed by the Wilcoxon rank sum test.

Fig. 7.. Factors associated with typeability of dWGS unculturable carriage specimens (n=89). (a–c) Genogroup was compared to porA Ct, the percentage of N.m. reads or the percentage of other Neisseria species in the sample. This comparison was similarly performed for MLST (d–f) and BAST (g–i) typing results. Statistics were performed by the Wilcoxon rank sum test.

Fig. 8.. Nucleotide proportions of filtered reads aligned to the fumC loci from representative isolates sequenced by WGS (isolate), samples sequenced by dWGS that did not have an exact fumC allele match in the pubMLST database (fumC unassigned) or samples sequenced by dWGS with an exact fumC allele match (fumC assigned).

Fig. 9.. A full core SNP alignment, including all samples passing QC and representative sequences from diverse cc, was used as input for phylotyping. A tree structure was built using FastTree2 -gtr -gamma settings. Samples are coloured according to their cc derived from their previously assigned MLST type. ‘Undetermined’ samples (white) had no identified MLST or cc due to having been previously assigned incomplete or partial MLST allelic profiles. The tree scale indicates evolutionary distance.