Metagenomic Sequencing of Multiple Soil Horizons and Sites in Close Vicinity Revealed Novel Secondary Metabolite Diversity

Abstract

Discovery of novel antibiotics is crucial for combating rapidly spreading antimicrobial resistance and new infectious diseases. Most of the clinically used antibiotics are natural products-secondary metabolites produced by soil microbes that can be cultured in the lab. Rediscovery of these secondary metabolites during discovery expeditions costs both time and resources. Metagenomics approaches can overcome this challenge by capturing both culturable and unculturable hidden microbial diversity. To be effective, such an approach should address questions like the following. Which sequencing method is better at capturing the microbial diversity and biosynthesis potential? What part of the soil should be sampled? Can patterns and correlations from such big-data explorations guide future novel natural product discovery surveys? Here, we address these questions by a paired amplicon and shotgun metagenomic sequencing survey of samples from soil horizons of multiple forest sites very close to each other. Metagenome mining identified numerous novel biosynthetic gene clusters (BGCs) and enzymatic domain sequences. Hybrid assembly of both long reads and short reads improved the metagenomic assembly and resulted in better BGC annotations. A higher percentage of novel domains was recovered from shotgun metagenome data sets than from amplicon data sets. Overall, in addition to revealing the biosynthetic potential of soil microbes, our results suggest the importance of sampling not only different soils but also their horizons to capture microbial and biosynthetic diversity and highlight the merits of metagenome sequencing methods. IMPORTANCE This study helped uncover the biosynthesis potential of forest soils via exploration of shotgun metagenome and amplicon sequencing methods and showed that both methods are needed to expose the full microbial diversity in soil. Based on our metagenome mining results, we suggest revising the historical strategy of sampling soils from far-flung places, as we found a significant number of novel and diverse BGCs and domains even in different soils that are very close to each other. Furthermore, sampling of different soil horizons can reveal the additional diversity that often remains hidden and is mainly caused by differences in environmental key parameters such as soil pH and nutrient content. This paired metagenomic survey identified diversity patterns and correlations, a step toward developing a rational approach for future natural product discovery surveys.

Keywords: Oxford Nanopore; amplicon sequencing; biosynthetic gene clusters; metagenome; natural products; secondary metabolites; soil horizons.

FIG 1

Geographic location, study outline, and analysis workflow. (A) Sampling site geographic location map of Tübingen, Germany (map data from Google ©2021). Multiple soil horizons from three sites were sampled. Photo depicting the 3 horizons of cambisol soil. (B) Sample and sequencing information, See Table S10a at https://doi.org/10.5281/zenodo.5195507 for details about soil names and profile description. (C) Amplicon sequencing (amplicon-seq) and analysis workflow. (D) Shotgun sequencing (shotgun-seq) and analysis workflow.

FIG 2

Microbial composition across 3 sampling sites (podzol, stagnosol, and cambisol) and 3 soil horizons (O, A, and B). (A) Bar plot showing taxonomic profile for 16S rRNA amplicon data set. (B) Bar plot showing taxonomic profile for shotgun-seq data set. Taxonomic profile at the phylogenetic rank of phylum is shown. The top 10 phyla are depicted in different colors, and remaining phyla are grouped as “remainder” and depicted in gray. The same colors for each phylum are used for side-by-side visualization. The SILVA rRNA database was used for classifying amplicons and the maxikraken2 database was used for classifying shotgun-seq reads.

FIG 3

Rarefaction curves for 16S rRNA gene amplicons, adenylation (A) domain amplicons, and Ketosynthase (KS) domain amplicons. The bold curve shows mean value of operational taxonomic units (OTUs)/operational biosynthetic units (OBUs) at a particular sequencing depth for all horizons of a particular site. The faint colored area around each curve shows a confidence interval of 67%.

FIG 4

Intersections and distribution of 16S (A), KS domain (B), and A domain (C) (amplicon sequence variants [ASVs] clustered at 97% similarity). The bar plot (top) in each panel shows the intersection size (the number of ASVs) in the combinatorial sets of relevant samples. The matrix below the bar plot indicates sets of samples that are represented by each bar.

FIG 5

Biosynthetic gene cluster (BGC) abundance distribution. (A) BGC abundance distribution across soil sampling sites (grouped according to BiG-SCAPE class). (B) BGC abundance distribution across soil horizons.