Accurate prediction of nitrogen fixation in cyanobacteria reveals the dynamic evolution driving high retention rate with mosaic distribution

Abstract

Cyanobacteria are the only prokaryotic group capable of simultaneously performing oxygenic photosynthesis and nitrogen fixation by nitrogenase extremely sensitive to oxygen. However, systematic understanding of the phylogenetic distribution of nitrogen-fixing ability remains incomplete because of difficulties in accurately identifying genes based on genome information due to occasional confusion with nitrogenase-like enzymes and the high degree of fragmentation in nif genes in heterocystous species. In this study, we comprehensively evaluated the distribution of the minimal nitrogen fixation gene set (nifHDKENB) across 586 major cyanobacterial strains using a 2-dimensional similarity plot with the negative logarithm of the e-values and protein length allowing accurate prediction of nitrogen fixation. The evaluation that almost all heterocystous strains are nitrogen fixer supports the accuracy of this method. Overall, 48% of the strains were predicted to be nitrogen-fixing with a mosaic distribution in non-heterocystous strains. Using the results as training data, nitrogen fixation was evaluated for a broader cyanobacterial lineage, including uncultured strains and metagenomes. Approximately 20% of 2,718 strains were evaluated to be nitrogen fixers, revealing a higher retention rate in Cyanobacteria phylum than other bacterial phyla. We also found a novel nif genes belonging to Group II and discussed the dynamic evolution process in cyanobacteria.