A comparative whole-genome approach identifies bacterial traits for marine microbial interactions

Abstract

Microbial interactions shape the structure and function of microbial communities with profound consequences for biogeochemical cycles and ecosystem health. Yet, most interaction mechanisms are studied only in model systems and their prevalence is unknown. To systematically explore the functional and interaction potential of sequenced marine bacteria, we developed a trait-based approach, and applied it to 473 complete genomes (248 genera), representing a substantial fraction of marine microbial communities. We identified genome functional clusters (GFCs) which group bacterial taxa with common ecology and life history. Most GFCs revealed unique combinations of interaction traits, including the production of siderophores (10% of genomes), phytohormones (3–8%) and different B vitamins (57–70%). Specific GFCs, comprising Alpha- and Gammaproteobacteria, displayed more interaction traits than expected by chance, and are thus predicted to preferentially interact synergistically and/or antagonistically with bacteria and phytoplankton. Linked trait clusters (LTCs) identify traits that may have evolved to act together (e.g., secretion systems, nitrogen metabolism regulation and B vitamin transporters), providing testable hypotheses for complex mechanisms of microbial interactions. Our approach translates multidimensional genomic information into an atlas of marine bacteria and their putative functions, relevant for understanding the fundamental rules that govern community assembly and dynamics.

Metabolic Phenotyping of Marine Heterotrophs on Refactored Media Reveals Diverse Metabolic Adaptations and Lifestyle Strategies

“…It also suggests that the ethylmalonyl-CoA pathway could play a dual role in organic acid assimilation (as supported by the enrichment mentioned above) and CO 2 fixation, as documented in other photoheterotrophs ( 81 ). Taken together, these cryptic associations can be interpreted as correlated adaptations with a potential ecological significance, similar to recently identified linked trait clusters ( 49 ).…”

“…The same was true when we compared the growth profile distance to the distance between Kyoto Encyclopedia of Genes and Genomes (KEGG) modules, which may better approximate the metabolic distance between strains ( Fig. 3C ; adjusted R 2 = 0.004, P = 0.0004) ( 49 ). Taken together, global genomic data, at least with the metrics used so far, do not seem to capture the differences in metabolic phenotypes observed between strains.…”

“…In order to generate a collection of strains representative of major marine lineages common to both the global oligotrophic and temperate oceans, we collected 63 heterotrophic isolates from different sources ( Table S1 in supplemental material). Strains were selected via a comprehensive genomic analysis conducted in a recent effort ( 49 ), in which over 400 high-quality reference genomes were clustered into functional groups using a trait-based approach focusing on metabolism and microbial interactions. Representative strains were chosen from each functional cluster, applying the additional criteria that they be culturable in standard laboratory conditions (at 26°C in Marine Broth) and have a biosafety level 1 rating.…”

Prediction of representative phenotypes using Multi-Attribute Subset Selection

“…Defined as the study of arrays of organismal behaviors (phenotypes) under multiple conditions or perturbations, phenomics approaches the challenge of understanding biological systems from a top-down perspective: empirical phenotypes are measured, often in a high-throughput manner, and then analyzed in search for associations (between genotypes and phenotypes, or between different phenotypes) (Baran et al, 2011(Baran et al, , 2013Bowen et al, 2011;Forchielli et al, 2022;Houle et al, 2010;Jim et al, 2004;Ohya et al, 2005;Tamura & D'haeseleer, 2008). These associations can, in turn, be used to understand how the individual subsystem properties of diverse organisms give rise to cell-level or community-level functions, including mutual interdependencies in microbial communities (DiMucci et al, 2018;Zelezniak et al, 2015;Zoccarato et al, 2022). An important goal towards the development and applicability of these approaches is to understand and efficiently organize high-throughput phenotypic data of organisms grown in a variety of environments (Barnett et al, 1990;Price et al, 2018;Segrestin et al, 2021).…”

Interactions between culturable bacteria are predicted by individual species’ growth

“…Metabolic modeling and genome-based models have been commonly used to predict microbial interactions [22][23][24] . These approaches predict interactions by considering the overlap and complementarity between species' metabolic capabilities and/or their resource consumption and secretion 25,26 .…”