A large-scale bioinformatic survey of the soil and plant-associated bacterium Pseudomonas fluorescens has catalogued thousands of biosynthetic gene clusters scattered across the species, many of them pointing to molecules that chemists have never characterised. The study, published as an open access research article on 9 July 2026, analysed 174 high-quality, publicly available genomes retrieved from the NCBI nucleotide database and identified a total of 2098 biosynthetic gene clusters, or BGCs, with an average of 12 per genome. Individual strains carried anywhere from 6 to 26 of these clusters, and roughly 40 percent of everything detected showed no meaningful similarity to any cluster in the MIBiG reference database of known natural products.
Biosynthetic gene clusters are groups of genes that together encode the enzymatic machinery for making secondary metabolites, the chemically diverse small molecules bacteria use to compete, communicate and survive. In P. fluorescens, a Gram-negative bacterium prized for its plant growth-promoting traits, such clusters encode celebrated compounds including the antimicrobials pyrrolnitrin and 2,4-diacetylphloroglucinol, the iron-scavenging siderophore pyoverdine, and lipopeptide surfactants such as viscosin. The research team, led by Sajid Iqbal and Farida Begum, used antiSMASH version 7 to detect the clusters and BAGEL version 4 to predict ribosomally synthesised and post-translationally modified peptides, known as RiPPs. Only high-quality, complete genomes were considered, because poor sequence assembly can distort predictions of the metabolites a cluster might produce.
Non-ribosomal peptide synthetases, or NRPSs, emerged as the dominant category, accounting for 42.45 percent of all detected BGCs, or 905 of the 2098 clusters. These giant multifunctional enzymes assemble peptide products without the ribosome, and their domain architecture often hints at the structure of the final molecule. Within the NRPS category, the researchers distinguished 678 complete NRPS clusters, 46 NRPS-like clusters and 181 NRPS fragments. Notably, 53.39 percent of the complete NRPS clusters, 362 of 678, were predicted to synthesise pyoverdine, the fluorescent siderophore that helps the bacterium seize iron from its environment. Pyoverdine was detected in 151 of the strains, while 29 strains lacked it entirely, and no single BGC was found in every genome, underscoring how unevenly biosynthetic capabilities are distributed across the species.
Beyond the abundant core pathways, the mining uncovered a set of rare, strain-specific clusters that the authors highlight as promising bioprospecting targets. A cluster encoding the peptide prosekin was found only in strain BBc6R8, where three NRPS regions and nine modules are predicted to assemble a linear peptide with the sequence LETVLLSI before any post-assembly modifications such as cyclisation or methylation. A viscosin cluster was described in strain SBW25, a brabantamide cluster in strain Fw300 25-6, and a pyochelin cluster combining NRPS and phenazine domains in strain AU11518. Kirromycin appeared only in strain PA3G8, and a cluster in strain PfR 37, predicted to incorporate dihydroxybenzoic acid, was flagged as a possible novel obafluorin-like peptide. Several other clusters, including those for anikasin, nunapeptin and pyrrolnitrin, were found in fewer than five strains each.
Polyketide synthase pathways told a similar story of dominance by a few familiar routes and rarity among the rest. Across the 174 strains, 29 PKS-related BGCs representing 13 distinct PKS clusters were identified, while 144 strains carried no PKS clusters at all. The most abundant was the type III PKS pathway for 2,4-diacetylphloroglucinol, detected in 19 strains. Three PKS clusters were unique to single strains: tetarimycin in strain EK007-RG4, and mupirocin and pseudomonic acid A, both in strain 2P24. Two type I PKS clusters, a resorcinol pathway in strain PfR_37 and a cluster in strain Myb115, showed no similarity to any known metabolites, marking them as candidates for entirely new chemistry.
The RiPP category, comprising peptides made on the ribosome and then chemically modified, included bacteriocins such as carocin, microcin, colicin, lanthipeptides and putidacin L1. Many of these clusters were incomplete, lacking core peptides or the immunity, transport and post-translational modification genes needed for a functional system, which the authors attribute to incomplete genome sequencing in the corresponding strains. Only four peptide clusters, including a colicin cluster in strain SBW25 and lanthipeptide class II and PaeM peptide clusters in strain 2P-24, carried immunity or transport-related genes. The lanthipeptide cluster contained a LanM gene encoding the enzyme responsible for post-translational cyclisation. The widespread occurrence of bacteriocins, the authors argue, points to a significant role in competitive exclusion within environmental and plant-associated niches.
To place this biosynthetic diversity in an evolutionary context, the team performed a core pan-genome analysis on 166 strains using the BPGA pipeline, clustering predicted proteins at a 95 percent identity threshold. The analysis revealed an open pan-genome: as genomes are added, the total gene pool keeps expanding rather than plateauing, and the core genome, the set of gene families shared by all strains, steadily shrinks. In this dataset, 1287 genes, or 23.76 percent, were shared by all strains, while the pan-genome continued to grow with each additional genome. A total of 2.3 percent of genes were strain-specific, with the largest private gene repertoires found in strains Ps_22, with 662 strain-specific genes, FW300-N1B4 with 623, ZL22 with 620 and PS652 with 583. Novel genes continued to appear even at high genome counts, reinforcing the open character of the pan-genome.
A phylogenetic tree built from 1287 concatenated core genes resolved the strains into 17 clades and revealed substantial genetic diversity. The largest clade, clade VI, contained 29 strains isolated from varied sources including nematodes, the kiwifruit plant Actinidia deliciosa and the environment, while the smallest clades each held just two environmental strains. The strains carrying the largest numbers of BGCs clustered in adjacent clades VIII, IX and XI. Mapping BGC types onto the tree showed that a small subset of clusters, including a conserved NRPS found in over 95 percent of strains and redox cofactor pathways, are broadly distributed, whereas many hybrid clusters and polyketide pathways are lineage-specific. The authors note that the repeated emergence of a particular hybrid BGC architecture could reflect either descent from a common ancestor or convergent assembly through horizontal gene transfer, and that their data cannot yet distinguish between the two scenarios.
Among the strains with exceptional biosynthetic potential, NZ17 stood out with 26 BGCs, including 17 NRPS, RiPP and polyketide clusters, while strains AH4 and ATCC 13525 also harboured unusually rich repertoires. The authors caution that in silico tools have limits: some orphan BGCs may be misidentified, and secondary screening with tools such as DeepBGC and PRISM would increase confidence. Every cluster identified in the study remains a bioinformatic prediction, and the authors frame their catalogue as a foundation for natural product discovery rather than a set of confirmed compounds. The core genome reported here, 1287 genes, is smaller than those reported in earlier pan-genome studies that used 14 or 4 genomes, and comparable to a previous analysis of 93 genomes from the broader P. fluorescens complex, a pattern the authors attribute to the larger and more diverse genome sample.
The broader significance lies in what the orphan clusters represent. With 880 of the 2098 BGCs showing no similarity to characterised clusters and a further 87 showing less than 5 percent similarity, the species appears to hold a deep reservoir of uncharacterised biosynthetic potential, particularly in phylogenetically divergent strains and those isolated from underexplored habitats. The authors suggest that horizontal gene transfer and niche-driven diversification, especially in rhizosphere, soil and marine environments, have shaped this accessory repertoire. Future work, they argue, should focus on functional validation and metabolite characterisation of the orphan clusters, with strains such as NZ17, AH4 and ATCC 13525 offering the richest starting points for the search for new antibiotics, biocontrol agents and other bioactive natural products.
Subject of Research: Comparative genome mining of biosynthetic gene clusters across Pseudomonas fluorescens genomes
Article Title: Genome mining and comparative genome analysis of Pseudomonas fluorescens revealed diverse biosynthetic and metabolic potential
Article References: Iqbal, S., & Begum, F. (2026). Genome mining and comparative genome analysis of Pseudomonas fluorescens revealed diverse biosynthetic and metabolic potential. Discover Genetics and Evolution, 72(1), Article 3. https://doi.org/10.1007/s00294-026-01336-1
Image Credits: AI Generated
DOI: 10.1007/s00294-026-01336-1
Keywords: Pseudomonas fluorescens, genome mining, biosynthetic gene clusters, NRPS, polyketides, RiPPs, pan-genome, secondary metabolites, siderophores, natural product discovery, bioinformatics, orphan clusters
News Source: Juliet Wilcox. (October 28, 2026). Genome Mining of 174 Pseudomonas fluorescens Strains Reveals Thousands of Hidden Biosynthetic Gene Clusters. Scienmag.
