Comparative Genomics of Completely Sequenced Lactobacillus helveticus Genomes Provides Insights into Strain-Specific Genes and Resolves Metagenomics Data Down to the Strain Level

Although complete genome sequences hold particular value for an accurate description of core genomes, the identification of strain-specific genes, and as the optimal basis for functional genomics studies, they are still largely underrepresented in public repositories. Based on an assessment of the genome assembly complexity for all lactobacilli, we used Pacific Biosciences' long read technology to sequence and <i>de novo</i> assemble the genomes of three <i>Lactobacillus helveticus</i> starter strains, raising the number of completely sequenced strains to 12. The first comparative genomics study for <i>L. helveticus</i>-to our knowledge-identified a core genome of 988 genes and sets of unique, strain-specific genes ranging from about 30 to more than 200 genes. Importantly, the comparison of MiSeq- and PacBio-based assemblies uncovered that not only accessory but also core genes can be missed in incomplete genome assemblies based on short reads. Analysis of the three genomes revealed that a large number of pseudogenes were enriched for functional Gene Ontology categories such as amino acid transmembrane transport and carbohydrate metabolism, which is in line with a reductive genome evolution in the rich natural habitat of <i>L. helveticus</i>. Notably, the functional Clusters of Orthologous Groups of proteins categories "cell wall/membrane biogenesis" and "defense mechanisms" were found to be enriched among the strain-specific genes. A genome mining effort uncovered examples where an experimentally observed phenotype could be linked to the underlying genotype, such as for cell envelope proteinase PrtH3 of strain FAM8627. Another possible link identified for peptidoglycan hydrolases will require further experiments. Of note, strain FAM22155 did not harbor a CRISPR/Cas system; its loss was also observed in other <i>L. helveticus</i> strains and lactobacillus species, thus questioning the value of the CRISPR/Cas system for diagnostic purposes. Importantly, the complete genome sequences proved to be very useful for the analysis of natural whey starter cultures with metagenomics, as a larger percentage of the sequenced reads of these complex mixtures could be unambiguously assigned down to the strain level.