What Is the Best Lens? Comparing the Resolution Power of Genome-Derived Markers and Standard Barcodes

MinION Sequencing of Yeast Mock Communities To Assess the Effect of Databases and ITS-LSU Markers on the Reliability of Metabarcoding Analysis

Microbial communities play key roles both for humans and the environment. They are involved in ecosystem functions, maintaining their stability, and provide important services, such as carbon cycle and nitrogen cycle. Acting both as symbionts and as pathogens, description of the structure and composition of these communities is important. Metabarcoding uses ribosomal DNA (rDNA) (eukaryotic) or rRNA gene (prokaryotic) sequences for identification of species present in a site and measuring their abundance. This procedure requires several technical steps that could be source of bias producing a distorted view of the real community composition. In this work, we took advantage of an innovative "long-read" next-generation sequencing (NGS) technology (MinION) amplifying the DNA spanning from the internal transcribed spacer (ITS) to large subunit (LSU) that can be read simultaneously in this platform, providing more information than "short-read" systems. The experimental system consisted of six fungal mock communities composed of species present at various relative amounts to mimic natural situations characterized by predominant and low-frequency species. The influence of the sequencing platform (MinION and Illumina MiSeq) and the effect of different reference databases and marker sequences on metagenomic identification of species were evaluated. The results showed that the ITS-based database provided more accurate species identification than LSU. Furthermore, a procedure based on a preliminary identification with standard reference databases followed by the production of custom databases, including only the best outputs of the first step, is proposed. This additional step improved the estimate of species proportion of the mock communities and reduced the number of ghost species not really present in the simulated communities. IMPORTANCE Metagenomic analyses are fundamental in many research areas; therefore, improvement of methods and protocols for the description of microbial communities becomes more and more necessary. Long-read sequencing could be used for reducing biases due to the multicopy nature of rDNA sequences and short-read limitations. However, these novel technologies need to be assessed and standardized with controlled experiments, such as mock communities. The interest behind this work was to evaluate how long reads performed identification and quantification of species mixed in precise proportions and how the choice of database affects such analyses. Development of a pipeline that mitigates the effect of the barcoding sequences and the impact of the reference database on metagenomic analyses can help microbiome studies go one step further.

Near Chromosome-Level Genome Assembly and Annotation of Rhodotorula babjevae Strains Reveals High Intraspecific Divergence

The genus Rhodotorula includes basidiomycetous oleaginous yeast species. Rhodotorula babjevae can produce compounds of biotechnological interest such as lipids, carotenoids, and biosurfactants from low value substrates such as lignocellulose hydrolysate. High-quality genome assemblies are needed to develop genetic tools and to understand fungal evolution and genetics. Here, we combined short- and long-read sequencing to resolve the genomes of two R. babjevae strains, CBS 7808 (type strain) and DBVPG 8058, at chromosomal level. Both genomes are 21 Mbp in size and have a GC content of 68.2%. Allele frequency analysis indicates that both strains are tetraploid. The genomes consist of a maximum of 21 chromosomes with a size of 0.4 to 2.4 Mbp. In both assemblies, the mitochondrial genome was recovered in a single contig, that shared 97% pairwise identity. Pairwise identity between most chromosomes ranges from 82 to 87%. We also found indications for strain-specific extrachromosomal endogenous DNA. A total of 7591 and 7481 protein-coding genes were annotated in CBS 7808 and DBVPG 8058, respectively. CBS 7808 accumulated a higher number of tandem duplications than DBVPG 8058. We identified large translocation events between putative chromosomes. Genome divergence values between the two strains indicate that they may belong to different species.

When barcoding fails: genome chimerisation (admixing) and reticulation obscure phylogenetic and taxonomic relationships

DNA barcoding is based on the premise that the barcode sequences can distinguish individuals (strains) of different species because their sequence variation between species exceeds that within species. The primary barcodes used in fungal and yeast taxonomy are the ITS segments and the LSU (large subunit) D1/D2 domain of the homogenized multicopy rDNA repeats. The secondary barcodes are conserved segments of protein‐encoding genes, which usually have single copies in haploid genomes. This study shows that the analysis of barcode sequences fails to reconstruct accurate species trees and differentiate species when the organisms have chimeric genomes composed of admixed mosaics of different origins. It is shown that the type strains of 10 species of the pulcherrima clade of the ascomycetous yeast genus Metschnikowia cannot be differentiated with standard barcodes because their intragenomic diversity is comparable to or even higher than the interstrain diversity. The analysis of a large group of genes of the sequenced genomes of the clade and the viability and segregation of the hybrids of ex‐type strains indicate that the high intragenomic barcode differences can be attributed to admixed genome structures. Because of the mosaic structures of the genomes, the rDNA repeats do not form continuous arrays and thus cannot be homogenized. Since the highly diverse ITS and D1/D2 sequences of the type strains form a continuous pool including pseudogenes, the evolution of their rDNA appears to involve reticulation. The secondary barcode sequences and the nonbarcode genes included in the analysis show incongruent phylogenetic relationships among the type strains, which can also be attributed to differences in the phylogenetic histories of the genes.

The mutL Gene as a Genome-Wide Taxonomic Marker for High Resolution Discrimination of Lactiplantibacillus plantarum and Its Closely Related Taxa

The current taxonomy of the Lactiplantibacillus plantarum group comprises of 17 closely related species that are indistinguishable from each other by using commonly used 16S rRNA gene sequencing. In this study, a whole-genome-based analysis was carried out for exploring the highly distinguished target genes whose interspecific sequence identity is significantly less than those of 16S rRNA or conventional housekeeping genes. In silico analyses of 774 core genes by the cano-wgMLST_BacCompare analytics platform indicated that csbB, morA, murI, mutL, ntpJ, rutB, trmK, ydaF, and yhhX genes were the most promising candidates. Subsequently, the mutL gene was selected, and the discrimination power was further evaluated using Sanger sequencing. Among the type strains, mutL exhibited a clearly superior sequence identity (61.6–85.6%; average: 66.6%) to the 16S rRNA gene (96.7–100%; average: 98.4%) and the conventional phylogenetic marker genes (e.g., dnaJ, dnaK, pheS, recA, and rpoA), respectively, which could be used to separat tested strains into various species clusters. Consequently, species-specific primers were developed for fast and accurate identification of L. pentosus, L. argentoratensis, L. plantarum, and L. paraplantarum. During this study, one strain (BCRC 06B0048, L. pentosus) exhibited not only relatively low mutL sequence identities (97.0%) but also a low digital DNA–DNA hybridization value (78.1%) with the type strain DSM 20314^T, signifying that it exhibits potential for reclassification as a novel subspecies. Our data demonstrate that mutL can be a genome-wide target for identifying and classifying the L. plantarum group species and for differentiating novel taxa from known species.