Long-read MinION™ sequencing of 16S and 16S-ITS-23S rRNA genes provides species-level resolution of Lactobacillaceae in mixed communities

Abstract

The <i>Lactobacillaceae</i> are lactic acid bacteria harnessed to deliver important outcomes across numerous industries, and their unambiguous, species-level identification from mixed community environments is an important endeavor. Amplicon-based metataxonomics using short-read sequencing of partial 16S rRNA gene regions is widely used to support this, however, the high genetic similarity among <i>Lactobacillaceae</i> species restricts our ability to confidently describe these communities even at genus level. Long-read sequencing (LRS) of the whole 16S rRNA gene or the near complete rRNA operon (16S-ITS-23S) has the potential to improve this. We explored species ambiguity amongst <i>Lactobacillaceae</i> using <i>in-silico</i> tool RibDif2, which identified allele overlap when various partial and complete 16S rRNA gene and 16S-ITS-23S rRNA regions were amplified. We subsequently implemented LRS by MinION™ to compare the capacity of V3-V4, 16S and 16S-ITS-23S rRNA amplicons to accurately describe the diversity of a 20-species <i>Lactobacillaceae</i> mock community in practice. <i>In-silico</i> analysis identified more instances of allele/species overlap with V3-V4 amplicons (<i>n</i> = 43) compared to the 16S rRNA gene (<i>n</i> = 11) and partial (<i>n</i> = up to 15) or complete (<i>n</i> = 0) 16S-ITS-23S rRNA amplicons. With subsequent LRS of a DNA mock community, 80% of target species were identified using V3-V4 amplicons whilst the 16S rRNA gene and 16S-ITS-23S rRNA region amplicons resulted in 95 and 100% of target species being identified. A considerable reduction in false-positive identifications was also seen with 16S rRNA gene (<i>n</i> = 3) and 16S-ITS-23S rRNA region (<i>n</i> = 9) amplicons compared with V3-V4 amplicons (<i>n</i> = 43). Whilst the target species affected by allele overlap in V3-V4 and 16S rRNA gene sequenced mock communities were predicted by RibDif2, unpredicted species ambiguity was observed in 16S-ITS-23S rRNA sequenced communities. Considering the average nucleotide identity (ANI) between ambiguous species (~97%) and the basecall accuracy of our MinION™ sequencing protocol (96.4%), the misassignment of reads between closely related taxa is to be expected. With basecall accuracy exceeding 99% for recent MinION™ releases, the increased species-level differentiating power promised by longer amplicons like the 16S-ITS-23S rRNA region, may soon be fully realized.