Plant roots are inhabited by a wide range of microorganisms, forming diverse communities that could be beneficial or harmful to their hosts. The ability of microbes to successfully colonize the rhizosphere, the soil region closer to the roots, lies in their genomic traits, which enable them to outcompete other microorganisms and interact with the plant hosts.
When comparing the microbial communities in the rhizosphere to those in bulk soil samples a notable change in composition is observed, where some taxa have expanded and are enriched in the rhizosphere, while others stay behind in the soil and are depleted. This phenomenon is commonly referred to as the "rhizosphere effect” and has been observed in many different plants, and experimental settings ranging from various stress conditions such as salinity, acidity, moisture levels, or pathogen presence, to variations in soil types, light cycles, fertilizer input, etc. As soils harbor the most complex and diverse microbial communities on Earth, it has remained challenging to identify what traits render a microbe suitable for colonizing the rhizosphere of plants. Studies typically consider functional or metabolic features, but their results have been difficult to replicate, suggesting that different taxa are enriched in different studies [1-4]. Still, if we want to predict which bacteria are likely to expand in the rhizosphere, it would be interesting to have a more general feature to characterize the successful colonizers.
How could we describe more general factors with a broader perspective? All plant roots release exudates, creating a nutrient gradient in the soil that influences the surrounding microbial communities. Based on this, we hypothesized that the rhizosphere effect would cover a gradient from fast-growing copiotrophs in the rhizosphere to slow-growing oligotrophs in the bulk soil. Although this seems like a straightforward hypothesis, it had never been shown. In fact, until recently the tools were not available to test this in complex microbiomes. Now that thousands of plant- and soil-associated metagenomic samples are available, we could compare the growth rate with other complex traits, if we could estimate bacterial growth rates based on the metagenomic data.
Our previous experience in studying codon usage patterns in prokaryotes  proved to come in handy for this task. Codon usage refers to the frequency at which synonymous codons are utilized in protein-coding genes. Each genome possesses its own distinctive codon usage signature, influenced by factors such as mutational biases and translational selection. However, even within one genome, different genes may have different codon usage. Highly expressed and highly conserved genes exhibit optimized codon usage to enhance translation efficiency. Conversely, lowly expressed and accessory genes tend to display less optimized codon usage frequencies. This disparity becomes particularly prominent in fast-growing microbes, where the genes encoding for ribosomal proteins have undergone substantial optimization of their codon usage to meet the heightened demands of translational machinery. Known as codon usage bias, this genomic signature was shown over a decade ago to reliably predict the growth rate potential of a bacterium by Rocha et al. . Recently, Weissman et al. further developed and refined this concept and published an R package and a useful database containing predicted growth rates for all bacteria .
We started our study comparing the growth rates of bacteria enriched in rhizospheres and their corresponding bulk soils, demonstrating that genera composed of fast-growing bacteria were indeed enriched in rhizospheres across a wide range of plant species, soil types, and environmental conditions (Figure 1).
We then aimed to validate this finding in individual metagenome-assembled genomes (MAGs) derived from diverse plants and soils. From the IMG/M database, we acquired a dataset consisting of thousands of MAGs representing all major branches of the bacterial tree. Using this extensive dataset, we discovered a consistent pattern of preferential enrichment of fast-growing bacteria in the rhizosphere samples across all phyla except Firmicutes (Figure 2). We also confirmed that the fast-growing MAGs were also enriched in rhizosphere-derived shotgun metagenome samples. We were happy now that our first hypothesis was shown in different ways and in unrelated datasets.
We went a bit further then and searched for functional and metabolic traits associated with the rhizosphere and soil bacteria, as well as with the fast-growing and slow-growing bacteria. After obtaining a catalog of all annotatable functional features and complex functional traits for each genome, we assessed their importance for predicting rhizosphere association, and compared the importance score to the growth rate potential by using Random Forest classification machine learning models. The RF models were quite accurate in predicting the niche and growth rate status of bacteria. Our findings further indicated that, when compared to other functional features, the growth rate potential emerged as the most important trait in all RF models. Remarkably, this held true even when comparing it to broader functional modules, including flagella, sugar catabolism, iron acquisition, and diverse metabolisms. This observation underscores the significance of growth rate potential as a highly generalizable characteristic for predicting a bacterium's ability to colonize the rhizosphere. The good performance of these RF models also suggests that it may be feasible to predict bacterial preferences for a given environment or to select biofertilizers or pathogen-suppressing microbes in plants.
To summarize, incorporating growth rate potential as a feature into microbiome models is important for explaining the distribution patterns of bacteria in the rhizosphere, and may also be crucial in predicting other bacterial preferences where nutritional gradients are present.
- Levy A, Salas Gonzalez I, Mittelviefhaus M, Clingenpeel S, Herrera Paredes S, Miao J, et al. Genomic features of bacterial adaptation toplants. Nature Genetics 2017; 50: 138–150.
- Zhou Y, Coventry DR, Gupta VVSR, Fuentes D, Merchant A, Kaiser BN, et al. The preceding root system drives the composition and function of the rhizosphere microbiome. Genome Biology 2020; 21:89.
- Ling N, Wang T, Kuzyakov Y. Rhizosphere bacteriome structure and functions. Nature Communications 2022; 13:836.
- Liu Q, Cheng L, Nian H, Jin J, Lian T. Linking plant functional genes to rhizosphere microbes: a review. Plant Biotechnology Journal 2022; 21: 902–917.
- López JL, Lozano MJ, Fabre ML, Lagares A. Codon Usage Optimization in the Prokaryotic Tree of Life: How Synonymous Codons Are Differentially Selected in Sequence Domains with Different Expression Levels and Degrees of Conservation. mBio 2020; 11:e00766-20.
- Vieira-Silva S, Rocha EPC. The systemic imprint of growth and its uses in ecological (meta)genomics. PLoS Genetics 2010; 6: e1000808.
- Weissman JL, Hou S, Fuhrman JA. Estimating maximal microbial growth rates from cultures, metagenomes, and single cells via codon usage patterns. Proceedings of the National Academy of Sciences 2021; 118:e2016810118.