Metatranscriptomes Reveal That All Three Domains of Life Are Active but Are Dominated by Bacteria in the Fennoscandian Crystalline Granitic Continental Deep Biosphere

Abstract

The continental subsurface is suggested to contain a significant part of the earth's total biomass. However, due to the difficulty of sampling, the deep subsurface is still one of the least understood ecosystems. Therefore, microorganisms inhabiting this environment might profoundly influence the global nutrient and energy cycles. In this study, in situ fixed RNA transcripts from two deep continental groundwaters from the Äspö Hard Rock Laboratory (a Baltic Sea-influenced water with a residence time of <20 years, defined as "modern marine," and an "old saline" groundwater with a residence time of thousands of years) were subjected to metatranscriptome sequencing. Although small subunit (SSU) rRNA gene and mRNA transcripts aligned to all three domains of life, supporting activity within these community subsets, the data also suggested that the groundwaters were dominated by bacteria. Many of the SSU rRNA transcripts grouped within newly described candidate phyla or could not be mapped to known branches on the tree of life, suggesting that a large portion of the active biota in the deep biosphere remains unexplored. Despite the extremely oligotrophic conditions, mRNA transcripts revealed a diverse range of metabolic strategies that were carried out by multiple taxa in the modern marine water that is fed by organic carbon from the surface. In contrast, the carbon dioxide- and hydrogen-fed old saline water with a residence time of thousands of years predominantly showed the potential to carry out translation. This suggested these cells were active, but waiting until an energy source episodically becomes available.IMPORTANCE A newly designed sampling apparatus was used to fix RNA under in situ conditions in the deep continental biosphere and benchmarks a strategy for deep biosphere metatranscriptomic sequencing. This apparatus enabled the identification of active community members and the processes they carry out in this extremely oligotrophic environment. This work presents for the first time evidence of eukaryotic, archaeal, and bacterial activity in two deep subsurface crystalline rock groundwaters from the Äspö Hard Rock Laboratory with different depths and geochemical characteristics. The findings highlight differences between organic carbon-fed shallow communities and carbon dioxide- and hydrogen-fed old saline waters. In addition, the data reveal a large portion of uncharacterized microorganisms, as well as the important role of candidate phyla in the deep biosphere, but also the disparity in microbial diversity when using standard microbial 16S rRNA gene amplification versus the large unknown portion of the community identified with unbiased metatranscriptomes.

Keywords: deep biosphere; groundwaters; mRNA; metatranscriptomes; rRNA.

FIG 1

Active deep biosphere communities inferred from metatranscriptome data. (A) Distribution of SSU rRNA reads based on cmsearch of domain-level covariance models available on Rfam. SSU rRNA reads were mapped to the reconstructed SSU rRNA contigs (≥300 bp and ≥5 average coverage) whose phylogenetic placement was assessed by the RAxML evolutionary placement algorithm (EPA), while the mRNA transcripts were given a taxonomic assignment using Kaiju. (B) Total prokarytic and eukaryotic community based on SSU rRNA gene amplicon sequencing and the active portion according to SSU rRNA gene transcripts from the metatranscriptome. Phylogenetic assignment was carried out at the phylum level, including the most recent candidate phyla (26, 37), except that the Proteobacteria were split into classes. Only phyla identified in the three samples with >0.01% relative abundance were included, and the remaining rare lineages were included in “other.” “Unknown” refers to tree nodes with poor taxonomic information and “Unassigned” to SSU transcripts that could not be reliably placed on the reference tree.

FIG 2

Diversity of active community members from all three domains. The reconstructed SSU rRNA gene contigs were placed on the tree of life using RAxML-EPA. The tree includes the most recent candidate phyla (26, 37) and was inferred by RAxML using the GTRCAT evolutionary model. Where possible, the leaves were collapsed into phyla, except for the Proteobacteria, which are shown in classes. The colored circles denote the origin of the sequences from OS (blue; OS1 in the inner blue ring and OS2 in the outer blue ring) and MM (green; MM1) waters. The circle size relates to the TPM of domain-level SSU rRNA distribution. Only bacterial phyla with a TPM distribution higher than 0.1% in at least one of the waters are named in the tree. Candidate phyla are labeled in bold, and SSU rRNA transcripts that could not be placed in specific phyla are shown by red circles. The scale bar shows 10% sequence divergence.

FIG 3

Gene expression profiles in the deep biosphere. Transcripts annotated to Gene Ontology (GO) “Biological Process” terms are shown for OS (OS1, left-hand bar; OS2, right-hand bar) and MM (MM1) waters. All GO processes for which mRNA transcripts were identified in at least one sample were included, and transcripts not assigned to any GO processes were named “Unassigned.” GO processes with TPM assigned to a phylum (or other taxonomical clasiffication) of less than 1% of the total for that term were amalgamated as Bacteria, Archaea, or Eukarya. Superscripts identify the respective classification included in the main text and metabolic model.

FIG 4

Metabolic model representing the active processes in the OS and the MM waters. Processes were based upon Gene Ontology (GO) terms and include the relevant Pfam and InterProScan identifications. Processes with TPM assigned to a phylum (or other taxonomical classification) are color coded as described in the legend to Fig. 3. Only major processes are shown for the MM water, while all identified processes have been included for the OS water.