Local and global chemical shaping of bacterial communities by redox potential

Abstract

Thermodynamics predicts a positive correlation between environmental redox potential and oxidation state of molecules; if found for microbial communities it would imply a new kind of deterministic eco-evolutionary process. This study examines evidence for local- and global-scale correlations between oxidation-reduction potential (ORP or Eh) in environmental samples and carbon oxidation state (ZC) of estimated bacterial and archaeal community proteomes. Seventy-nine public datasets for seven environment types (river & seawater, lake & pond, alkaline spring, hot spring, groundwater, sediment, and soil) were analyzed. Taxonomic abundances inferred from high-throughput 16S rRNA gene sequences were combined with NCBI Reference Sequence (RefSeq) proteomes to estimate the amino acid compositions and chemical formulas (CcHhNnOoSs) of community proteomes, which yield ZC. Alkaline hot springs have the lowest ZC for both bacterial and archaeal domains of any environment. Positive global correlations between redox potential and ZC are found for bacterial communities in lake & pond, groundwater, and soil environments, but not archaeal communities, suggesting a broad ecological signal of chemical shaping in Bacteria.

The proteome is a terminal electron acceptor

“…to the chemical character of their habitats (47). Consistent with our hypothesis, we found strong and widespread positive correlations between the ZC values of conserved proteins (Fig.…”

“…Directional fluxes are a hallmark of non-equilibrium systems. In contrast with equilibrium descriptions (47), the magnitudes of these intracellular fluxes alter the optimal biomass redox state (Z * C,B) in our model.…”

“…Based on the correlations documented in Fig. 4F, we propose an additional mode of protein evolution wherein mutations neutral or even weakly deleterious to individual proteins' biochemical or structural function might be selected due to an organismal benefitthey cause ZC,B to better align with the redox chemistry of the host's typical environment (47). The affected proteins can be involved in any cellular structure or biochemical process, but large and highly-expressed proteins will affect ZC,B more.…”

Annotation-free prediction of microbial dioxygen utilization

“…S4). Further, it has been argued that the sequence content of the whole genome adapts to O 2 (21). If this is correct, we do not yet understand the biochemical logic underpinning this adaptation.…”

“…Annotation-free feature sets included the number of predicted open reading frames ("gene count"), counts of genomic DNA K-mers (lengths 1-5), counts of coding sequence amino acid Kmers (lengths 1-3), a manually selected list of simple chemical features of nucleotide and amino acid sequences in each genome ("chemical features"), and genome embeddings. Chemical features included the number of open reading frames, the genomic GC content, the number of carbon, nitrogen, oxygen and sulfur atoms (22) in protein and RNA coding sequences as well as the average redox state of carbon (Z C ) in those same sequences (21). The "gene embedding" approach requires no annotation but adds a computationally intensive step as it involves a forward pass through a large language model (12).…”