Antarctic Bacterium Yields Kineochelins, a Novel Class of Iron-Binding Molecules

Researchers have discovered a previously unknown family of iron-binding molecules, named kineochelins, in a bacterium isolated from Antarctic soil. The compounds were found in Actinokineospora sp. UV203, a strain that appears to represent a species new to science, and their characterization links a distinct biosynthetic gene cluster to a suite of mixed-ligand siderophores whose antimicrobial activity depends strongly on iron availability. The work, drawing on a long-term biodiversity campaign on James Ross Island, adds a structurally novel entry to the catalogue of microbial metallophores and underscores how little-explored polar ecosystems remain as sources of new chemistry.

Siderophores—literally "iron carriers"—are small molecules secreted by bacteria and fungi to scavenge ferric iron from their surroundings. Because iron, despite its abundance in the Earth's crust, exists in near-insoluble ferric form under aerobic conditions and is fiercely contested by co-occurring microbes, the ability to sequester it can determine survival in soil, sediment and host environments. Decades of natural products research have catalogued hundreds of siderophore structures, yet the chemical space of these molecules remains far from exhausted, particularly in environments that have been sampled only sporadically. The kineochelins join this family with an unusual mixed-ligand architecture that combines a salicylate-derived core with peptide extensions, a design not previously reported from an Antarctic isolate.

The discovery emerged from a collection of 972 bacterial strains built up since 2007 from the active layer of soils above permafrost on James Ross Island, near the northern tip of the Antarctic Peninsula. The active layer—the shallow zone of ground that thaws each austral summer and refreezes beneath—is home to most microbial life in permafrost landscapes and is increasingly studied as climate change alters its depth and dynamics. Most isolates in the collection belonged to four phyla: Actinomycetota, Bacillota, Bacteroidota and Pseudomonadota. Actinomycetota, the phylum that includes the streptomycetes responsible for many classical antibiotics, has historically been the most productive source of microbial natural products, making novel members of this group especially attractive for chemical exploration.

From this pool, 145 phylogenetically divergent strains were prioritized for whole-genome sequencing, and most, including UV203, turned out to represent previously undescribed species or higher taxa. That proportion is striking: it suggests that even a modest sampling effort in a remote polar locality yields organisms absent from existing reference databases, consistent with the broader observation that Antarctic microbial diversity is substantially under-characterized. Strain UV203 itself was recovered as a colony on starch-casein agar from a soil sample collected at Abernethy Flats during the austral summer of 2022.

Genomic and phylogenetic analyses placed UV203 firmly outside the boundaries of known Actinokineospora species. Modern prokaryotic taxonomy relies on quantitative thresholds rather than subjective judgment: strains sharing less than roughly 95 to 96 percent average nucleotide identity, or less than about 70 percent digital DNA–DNA hybridization, are conventionally considered distinct species. UV203 fell well below both benchmarks, with an average nucleotide identity of 93.6 percent and a hybridization value of 51.8 percent against its closest relative, Actinokineospora alba DSM 45114T. A phylogeny built from up-to-date bacterial core genes—conserved loci whose combined signal is more reliable than any single marker—supported UV203 as a distinct species-level lineage, and 16S rRNA gene comparisons indicated it does not cluster with three validly described Actinokineospora species whose genomes are not yet available, leaving their relationships unresolved.

With the organism's novelty established, the team turned to its biosynthetic potential. Genome mining for natural products has become a standard first step in modern drug discovery, because the genes encoding the enzymes that assemble specialized metabolites are typically clustered together and can be recognized computationally. The antiSMASH algorithm predicted 21 biosynthetic gene clusters in the UV203 genome, many without close matches in public databases, and the researchers noted this figure is likely an underestimate because closely spaced or overlapping clusters are often not fully resolved by automated tools. Untargeted liquid chromatography–mass spectrometry detected more than 28 specialized metabolites under various culture conditions. Many could not be matched to any known natural product in the GNPS, NPAtlas or CAS databases, prompting an effort to isolate and characterize selected unknowns.

A key clue came from an engineered strain. Initial attempts to characterize low-abundance chlorinated metabolites associated with a halogenase-containing gene cluster failed to yield sufficient material, but one genetically engineered derivative unexpectedly lost the antimicrobial activity seen in the wild type. Comparative metabolomics—profiling the full chemical inventory of wild type and mutant side by side—revealed that a large group of structurally related congeners, spanning monoisotopic masses from 219 to 722 daltons, was absent from the inactive mutant. The three smallest members were tentatively identified as known small siderophores—asteroidic acid, pseudomobactin A and vulnibactin 2—while the larger congeners shared a 2-(2-hydroxyphenyl)-5-methyl-oxazoline or oxazole subunit but differed by increments matching amino acids such as glycine, serine and N-hydroxy-ornithine derivatives. Because the tentative structures matched nothing in the databases, the group was designated kineochelins, with letters A through E reflecting the number of amino acids attached to the vulnibactin 2 core and subscripts denoting the oxidation state of the oxazole ring.

Structural elucidation focused on two representative congeners. Kineochelin E1, purified from fraction 17 of a flash-chromatography separation, was assigned the molecular formula C13H14N2O5, and one- and two-dimensional NMR spectroscopy confirmed the expected oxazoline substructure with a glycine attached via an amide bond. Marfey's analysis—a technique that derivatizes hydrolysed amino acids to determine their stereochemistry by chromatographic comparison—established that the compound derives from L-threonine and ruled out the presence of a second stereoisomer. Kineochelin A1, the largest and most abundant congener at roughly 80 percent purity in fraction 11, was shown by NMR and MS/MS interpretation to extend the E1 scaffold with glycine, N5-hydroxy-N5-formyl-ornithine, two serine residues and a cyclized N5-hydroxy-ornithine forming a cyclic amide at the C-terminus. Marfey's analysis of the hydrolysed compound detected only L-thr, L-orn and D-ser, fully establishing its absolute configuration. Two known 2,5-diketopiperazines, cyclo(Hyp-Leu) and cyclo(Hyp-Phe), were also isolated but not studied further.

Genome mining identified the kin cluster, BGC 2.11, as the most plausible source of the kineochelins. It encodes three core nonribosomal peptide synthetase genes—kinA, kinB and kinC—comprising six modules, two with epimerization domains, plus a salicylate synthase, kinL. Nonribosomal peptide synthetases are giant, modular enzymes that assemble peptide products without ribosomes, with each module typically incorporating one amino acid; epimerization domains flip L-amino acids into their D-forms, explaining the D-serine found in kineochelin A1. The cluster shows about 38 percent homology to the gobichelin pathway and shares biosynthetic logic with the amychelin and cahuitamycin systems, but BiG-SCAPE network analysis placed it in a separate gene cluster family alongside two uncharacterized clusters from A. alba strains. Full-length alignments of core proteins did not exceed 40 percent amino acid identity to related clusters, and the kin cluster uniquely contains the kinC module, absent from the A. alba homologues, consistent with differences in peptide length and composition between the kineochelins and their known relatives.

Direct genetic validation proved difficult. Knockout constructs for kinA and kinB repeatedly failed to yield exconjugants, although a control plasmid integrated successfully, confirming the strain is genetically tractable—a result suggesting the core synthetase genes themselves may be refractory to disruption, perhaps because of their size or genomic context. The researchers instead used comparative transcriptomics under iron-modulated conditions. In a 10-day time course in SM17 medium, siderophore activity rose sharply between days 4 and 6, peaking at day 6, with antimicrobial activity first detectable at day 6 and peaking at day 8. Iron supplementation with 200 micromolar FeCl3 suppressed siderophore production almost completely, while the iron chelator 2,2'-bipyridine did not increase it, suggesting the pathway is already derepressed under baseline conditions—plausibly an adaptation to the chronically iron-limited soils from which the strain was isolated. Transcriptome sequencing at days 3 and 7 showed that kin genes were co-transcribed and significantly upregulated under siderophore-producing conditions and repressed under iron-replete conditions at both time points. An additional co-regulated gene outside the cluster, encoding a lysine/ornithine N-oxygenase and named kinO, plausibly supplies the N-hydroxy-ornithine incorporated into most kineochelins; hydroxamate groups of this kind are classic iron-binding motifs in siderophore chemistry.

Bioactivity testing revealed a pattern consistent with iron sequestration rather than classical antibiotic action. Crude extracts inhibited Antarctic environmental strains most strongly, along with the indicator bacterium Micrococcus luteus and the yeasts Nakaseomyces glabratus and Saccharomyces cerevisiae. A pre-purified kineochelin-enriched fraction confirmed these effects, but potency was modest: minimum inhibitory concentrations were 0.5 milligrams per millilitre against Nakaseomyces strains and 2.5 milligrams per millilitre against S. cerevisiae, with corresponding fungicidal concentrations of 2.5 and 5 milligrams per millilitre. Critically, adding excess iron markedly reduced inhibition, while iron depletion in the presence of the fraction caused complete growth arrest of sensitive yeasts. A modified Chrome Azurol S assay—a colorimetric test in which a dye-labeled iron complex releases its color upon ligand exchange—showed the strongest metal-binding responses toward ferric and ferrous iron, with intermediate activity for copper, gallium and vanadium and weak responses to other metals, in both culture supernatants and the purified fraction.

The selective sensitivity of certain yeasts fits their differing iron acquisition strategies, the authors argue. S. cerevisiae relies mainly on reductive iron uptake, a strategy in which cell-surface reductases strip ferric iron from carriers before import, and N. glabratus has an even more restricted repertoire, lacking several ferric reductases.