Hidden Antarctic microbe turns iron into stone beneath the ice shelf

Beneath the floating ice of Antarctica’s Larsen C Ice Shelf, in water that has never seen sunlight and barely knows oxygen, scientists have discovered a previously unknown bacterium that can pull electrons from dissolved iron. The microbe, provisionally named Candidatus Mariimomonas ferrooxydans, belongs to an entirely new order within the phylum Nitrospirota and appears to have been quietly precipitating iron minerals in the dark sediments of the Weddell Sea for thousands of years. The finding, published in the journal Microbiome, matters far beyond the polar south: it offers a living model for how some of Earth’s most famous iron deposits, the banded iron formations of the Precambrian, may have formed without any help from sunlight or photosynthesis.

The story begins with a sediment core called GC16B, recovered from the Larsen C embayment at a water depth of 324 meters. The core stretches 236 centimeters into the seafloor and preserves an unbroken Holocene archive, roughly the last 11,700 years of environmental history. Radiocarbon dating of acid-insoluble organic material allowed the team to assign ages to each layer, revealing a sequence that runs from glacial till deposited beneath a grounded glacier during the Last Glacial Maximum, through parallel-laminated and cross-laminated muds laid down under a floating ice shelf, to sandy, ice-rafted debris near the top under consolidated sea ice. Because the calving front of the ice shelf retreated only recently, these sediments had remained sealed and undisturbed since the last glacial period, making them a rare and pristine window into sub-ice-shelf life.

To read that archive, the researchers extracted environmental DNA from 40 layers and sequenced the V5 to V8 region of the 16S ribosomal RNA gene from 37 of them. The resulting community profiles were strikingly coherent. Using weighted UniFrac distances, the layers clustered into three distinct phases that aligned almost perfectly with the geological facies boundaries. Phase A, from the surface down to 41 centimeters, corresponded to occasionally open marine conditions and showed significantly higher taxonomic richness, evenness, and Shannon diversity than the deeper layers. Phases B and C, spanning roughly 45 to 181 centimeters, reflected anoxic, aphotic conditions beneath the floating ice shelf, dominated by families such as Anaerolineaceae, Burkholderiaceae, and an uncultured lineage within the class Thermodesulfovibrionia. In the deepest phase C layers, the methane-cycling archaeal family Candidatus Methanoperedenaceae made up nearly half of the community on average.

The transition between these microbial phases coincides with known retreat events of the Larsen C Ice Shelf, suggesting that the microbiome recorded the oceanographic and glacial regime shifts of the Holocene in real time. Linear discriminant analysis effect size, or LEfSe, pinpointed the families most responsible for the differences between the open marine and sub-ice-shelf settings, and most of them were unassigned or uncultured, underscoring how little is known about life in these environments. A co-occurrence network built with SparCC correlations split the community into two sub-networks, one characteristic of the open ocean and one of the ice-covered seafloor, with almost no overlap between them. Within the sub-ice-shelf network, the uncultured Thermodesulfovibrionia stood out with the highest betweenness, degree, and closeness centrality values, marking it as a keystone taxon that appeared to shape the structure, and possibly the function, of the entire community.

To understand what that keystone was actually doing, the team turned to shotgun metagenomics. Because DNA concentrations in the ancient layers were extremely low, below one nanogram per layer, they used whole-genome amplification before sequencing on the Ion Torrent and Illumina platforms, and interpreted the results primarily in terms of gene presence rather than abundance. Assembling and binning the reads yielded 20 medium-quality metagenome-assembled genomes, or MAGs. Several of these carried genes for the Wood–Ljungdahl pathway, a carbon fixation route favored under anoxic conditions, along with genes for nitrogen fixation and dissimilatory sulfate reduction, painting a picture of a metabolically versatile chemolithoautotrophic ecosystem running on inorganic chemistry rather than photosynthesis.

Three of the MAGs, all matching the uncultured Thermodesulfovibrionia, contained something especially intriguing: a gene encoding Cyc2, a fused porin-cytochrome protein that sits in the outer membrane and serves as the primary electron acceptor in known iron-oxidizing bacteria. One MAG, LCMS-H39C-Mi from a layer 160 to 161 centimeters deep, carried the full-length gene, complete with a signal peptide, a cytochrome c domain bearing the characteristic CXXCH heme-binding motif, and a transmembrane beta-barrel with 16 antiparallel beta strands. The genomes also encoded a suite of c-type cytochromes, iron-sulfur proteins, and nitrate reductase complexes, hinting at an electron transfer chain that could couple iron oxidation to nitrate reduction in the absence of oxygen.

Phylogenomic analysis placed these bacteria firmly outside every known group. Their 16S rRNA genes shared less than 87 percent identity with any cultivated member of Nitrospirota, and their closest available reference genome, recovered from deep-sea marine sediments, shared only about 93 to 94 percent average nucleotide identity. Together with related MAGs from subsurface fluids, they formed a novel candidate order, which the authors named Candidatus Mariimomonadales. To confirm that the organisms were really present in the sediment rather than an artifact of sequencing, the team used fluorescence in situ hybridization with a custom oligonucleotide probe, Mari_866F, and visually captured the cells in a layer 70 to 71 centimeters deep, glowing orange against the red-stained background of general bacteria.

The decisive experiment came next. The researchers chemically synthesized the cyc2 gene with codon optimization for Escherichia coli, swapped in a bacterial signal peptide, and expressed the modified protein in a strain engineered to mature c-type cytochromes. Cells expressing the protein turned pink, a sign of successful heme incorporation, and in-gel assays confirmed the cytochrome’s peroxidase activity. Most importantly, whole-cell assays showed that E. coli carrying the full-length Cyc2 significantly accelerated the oxidation of ferrous iron, measured with the ferrozine colorimetric assay, compared with controls expressing only the porin domain or an empty vector. This functional validation demonstrated that Cyc2 from Candidatus Mariimomonas ferrooxydans is a genuine iron oxidase, the first such enzyme functionally confirmed from a Nitrospirota lineage.

The geological implications are considerable. Elevated iron-to-titanium ratios in the anoxic interval of the core broadly coincide with the depth distribution of the Thermodesulfovibrionia, suggesting that microbial iron oxidation contributed to iron mineralization in situ. The Holocene sediments beneath Larsen C, with their laminated, dropstone-free facies, bear a resemblance to the synglacial iron formations deposited during the Neoproterozoic Snowball Earth, when glaciers reached the tropics and iron-rich layers accumulated beneath or adjacent to ice. Popular models of banded iron formation genesis invoke cyanobacteria or anoxygenic photoferrotrophs, organisms that need light. The Larsen C discovery documents a plausible alternative: iron biomineralization by chemolithotrophs in complete darkness, under oligotrophic, anoxic conditions, fueled by dissolved ferrous iron likely delivered from subglacial meltwater or local hydrothermal sources.

The authors are careful to note the limitations of their approach. Whole-genome amplification can introduce bias, so quantitative gene comparisons across layers should be treated cautiously, and the absence of negative extraction controls precluded direct contamination assessment. Yet the continuous, coherent stratigraphic variation of the 16S profiles, without isolated taxonomic spikes, and the close correspondence of community transitions to facies boundaries argue that the signal is genuine. Whether the coupling between iron oxidation and nitrate reduction in the new lineage is direct or indirect, with nitrite from denitrification abiotically oxidizing ferrous iron, remains to be established. Either way, the study expands the known repertoire of iron-cycling life, offers a non-phototrophic complement to models of ancient iron deposition, and even suggests a framework for interpreting ferric oxide accumulations on other planets, such as Mars, where dark, iron-rich, oxygen-poor environments may once have hosted similar chemistry.

Subject of Research: Microbial iron biomineralization by a novel chemolithotrophic bacterium in anoxic Holocene sediments beneath the Larsen C Ice Shelf

Article Title: Novel Antarctic chemolithotroph drives iron biomineralization

Article References: Yoon, J., Lee, B., Yoo, K.-C., Kwak, M.-J., Song, H. J., Hwang, C. Y., Chung, Y., Kim, K., Kwon, S.-K., Song, J. Y., Yoon, H. S., & Kim, J. F. (2026). Novel Antarctic chemolithotroph drives iron biomineralization. Microbiome, 14(1), Article 225. https://doi.org/10.1186/s40168-026-02536-0

Image Credits: AI Generated

DOI: 10.1186/s40168-026-02536-0

Keywords: Antarctica, Larsen C Ice Shelf, chemolithotroph, iron biomineralization, Cyc2, Nitrospirota, banded iron formations, metagenomics, Snowball Earth, biogeochemistry, Holocene sediments, microbiology