The strain AHT1^T isolated from a mixed anaerobic sediments of northeastern Mongolian soda lakes (Sorokin et al., 2008) was kindly provided for the study by Dr. D. Yu. Sorokin. The ability of strain AHT1^T for dissimilatory iron-reduction was tested in an optimal liquid medium of the following composition (per liter distilled water): NH₄Cl 0.5 g; KH₂PO₄, 0.2 g; MgCl₂ × 6H₂O 0.1 g; CaCl₂ × 2H₂O 0.02 g; KCl 0.2 g; NaCl 3.0 g; Na₂CO₃ 15.0 g; NaHCO₃ 20.0 g; Na₂S × 9H₂O 0.1 g; 1 mL trace element solution (Kevbrin and Zavarzin, 1992); 1 mL vitamin solution (Wolin et al., 1963); yeast extract 0.2 g. Molecular hydrogen (10% v/v in the gas phase) was used as the electron donor, and SF prepared as previously described (Zavarzina et al., 2006) was used as the electron acceptor. SF was added to the culture medium prior to sterilization to achieve a final Fe(III) content of 50 mM. The medium was prepared under pure N₂ flow. Afterwards, NaHCO₃, vitamins, Na₂S × 9H₂O, and SF were added. The medium was dispensed in 20 mL aliquots into 50 mL flasks and the headspace was filled with pure N₂. The medium was autoclaved at 1 atm, 121°C for 20 min. The pH of the sterile medium was 9.5. After three successful transfers to the optimal medium, sodium sulfide and yeast extract were omitted for further cultivation steps in order to test the ability of strain AHT1^T to reduce SF lithoautotrophically in the absence of any reducing agents.

Strain Z-1002 was isolated in 2010 from the sediments of a hypersaline soda lake Magadi (Kenya). Anaerobic sediment samples were collected by Prof. G.A. Zavarzin in 1996 in pre-sterilized flasks, sealed on-site and stored since then at +4°C. The ability of strain Z-1002 to reduce SF during autotrophic growth with molecular hydrogen or formate in the presence of vitamin solution was previously reported (Zavarzina et al., 2018). Here we focused on determining its optimal growth conditions, the ability to reduce sulfur compounds and capacity for chemoorganotrophic growth.

The basal medium (BM) used for physiological tests of both strains contained (per liter of distilled water): NH₄Cl 0.5 g; KH₂PO₄ 0.2 g; MgCl₂ × 6H₂O 0.1 g; CaCl₂ × 2H₂O 0.02 g; KCl 0.2 g; Na₂S·9H₂O 0.1 g, 1 mL trace elements solution (Kevbrin and Zavarzin, 1992), 1 mL vitamins solution (Wolin et al., 1963). Sodium formate (1 g l⁻¹) and SF [final Fe(III) content of 50 mM] were added to test the Fe(III) reducing activity under optimal growth conditions. The pH growth range of strain Z-1002 was determined on BM medium supplemented with 60 g l⁻¹ NaCl and 10 g L⁻¹ NaHCO₃. pH lower than 8.0 was adjusted with 6 M HCl, in the range between 8.0 and 9.5—by titrating with 10% Na₂CO₃ solution, in the range between 9.5 and 10.3—with 12 M NaOH solution. In this experiment, three tubes with culture media were prepared for each pH value: two tubes for cultivation and one for measuring the initial pH value of a medium after autoclaving. The рН stability of the cultures was checked at the end of the incubation.

The carbonate/bicarbonate growth optimum was determined on BM medium supplemented with 20 g L⁻¹ NaCl, 55 g L⁻¹ NaHCO₃, and 95 g L⁻¹ Na₂CO₃. This modified medium was gradually diluted with another one—BM containing 20 g L⁻¹ NaCl but without carbonate/bicarbonate. In the latter medium, the optimal pH value for growth was maintained by the addition of 20 mM N-cyclohexyl-2-aminoethanesulfonic acid (CHES) buffer (Merck). Three subsequent transfers were performed to confirm the ability of strain Z-1002 to grow without carbonates.

The sodium chloride dependence of strain Z-1002 was determined on BM medium supplemented with 20 g L⁻¹ NaHCO₃ and 40 g L⁻¹ Na₂CO₃. NaCl was added to this medium at the concentrations of (g L⁻¹): 0, 3, 5, 10, and up to 140 in increments of 10. Three subsequent transfers were performed to confirm the ability of strain Z-1002 to grow without chloride.

In all the experiments, soluble electron donors and acceptors were added from sterile anaerobic stock solutions prior to inoculation. All the organic substrates (peptides, carbohydrates, alcohols and organic acids) were filter-sterilized using 0.2 μm pore size syringe filters (Millipore) and added to a final concentration of 3 g L⁻¹. The capability for dissimilatory reduction of sulfur compounds (S⁰ (1% w/v); S₂O₃²⁻ (10 mM); SO₄²⁻ (20 mM)) was tested with formate (1 g L⁻¹), H₂ (10% v/v in the gas phase), ethanol or organic acids (10 mM in each case) added as the electron donors. The ability to disproportionate sulfur or thiosulfate was tested without the addition of any of the electron donors.

The ability to oxidize a natural siderite-based mineral mixture (Fe₂CO₃ as the major component) under anaerobic conditions was tested using an optimized medium without sodium sulfide with the following composition (per liter distilled water): NH₄Cl 0.5 g; KH₂PO₄ 0.2 g; MgCl₂ × 6H₂O 0.1 g; CaCl₂ × 2H₂O 0.02 g; KCl 0.2 g; NaCl 50.0 g; Na₂CO₃ 40.0 g; NaHCO₃ 20.0 g; 1 mL trace mineral solution (Kevbrin and Zavarzin, 1992); 1 mL vitamin solution (Wolin et al., 1963). This medium was dispensed under N₂ flow in 10 mL aliquots into Hungate tubes containing 0.1 g dry mineral powder each. Uniform grains of the siderite-based mineral mixture of hydrothermal origin (Bakal deposit, the Urals, Russia) were preliminarily selected and ground to powder (<100 μm particle size) in an agate mortar. The presence of the Fe(III)-containing green rust (10.6%) and iron oxides (13%) in the mineral mixture was detected by Mössbauer spectroscopy (Table 1). The siderite-containing medium was autoclaved at 1 atm, 121°C for 20 min. Three subsequent transfers were performed to confirm the ability of strain Z-1002 to grow with siderite by Fe(II) oxidation.

All the cultivation experiments were performed in duplicate using Hungate tubes.

Subsurface water from the Yessentuki Mineral Water Deposit (YMWD, Stavropol Krai, Russia) was sampled in September, 2020, from Well 9 extracting Na-Ca-HCO₃-SO₄-type mineral water from the Lower Cretaceous (K₂s-m) aquifer. This well (E 42°48′10” N 44°2′30″) is 600 m deep and has open boreholes in the interval of 485–556 m. The temperature, pH value, and salinity of the mineral water at the wellhead during the sampling were 21.9°C, 7.9, and 0.4 g L⁻¹, respectively. One hundred liter of water was filtered through a 0.2 μm pore size track membrane filters (JINR, Dubna, Russia) under the natural overpressure of the well, as described previously (Gavrilov et al., 2022). The filter was cut in half and one of its parts was used for DNA extraction and phylogenetic profiling of the natural water microbial community according to a previously described procedure (Gavrilov et al., 2022). Another part of the filter was placed in a pre-sterilized 100 mL flask, filled with mineral water (80 mL) and CO₂ gas headspace (20 mL), and was stored at +4°C for 1.5-years. After storage, the flask was supplemented with SF [final Fe(III) content of 10 mM], sodium sulfate, sodium acetate, and sodium formate (10 mM each), and used as a primary enrichment of sulfate and iron reducing microorganisms inhabiting the mineral water. It was incubated in the dark at 35°C for 3 weeks. Subsamples for DNA extraction and phylogenetic profiling were taken from this culture at the beginning and after the end of the incubation. Further transfers from this primary enrichment were made to the medium with the following composition (per liter distilled water): NH₄Cl 0.015 g; KH₂PO₄ 0.115 g; MgCl₂ × 6H₂O 0.6 g; CaCl₂ × 2H₂O 0.1 g; NaHCO₃ 0.3 g; Na₂S·9H₂O 0.1 g, 1 mL trace element solution (Kevbrin and Zavarzin, 1992); 1 mL vitamin solution (Wolin et al., 1963). Sodium formate (1 g L⁻¹) or sodium acetate (1 g L⁻¹) were used as the electron donors and SF (final Fe(III) content of 50 mM) as the electron acceptor. The medium for the enrichments was prepared by boiling and cooling it under pure N₂ flow, then NaHCO₃, vitamins, Na₂S·9H₂O, and SF were added, and the pH was adjusted to 8.0 with 2.5 M NaOH solution. The medium was dispensed in 10 mL aliquots into 16 mL Hungate tubes and autoclaved at 1 atm, 121°C for 20 min.

Growth of D. alkaliphilus strains was monitored by direct cell counting using an Axio Lab.A1 phase-contrast and fluorescent microscope (Zeiss, Germany). Subsamples of SF-grown cultures were pre-stained with acridine orange dye for DNA.

Molecular hydrogen consumption, the formation of gaseous metabolites, sulfide concentration, and Fe(II) production were monitored as previously described (Khomyakova et al., 2022). Acetate was analyzed using the same chromatograph with FID detector and an Optima FFAPplus 0.25 μm × 0.32 mm × 30 m capillary column (Macherey-Nagel) with argon as the carrier gas. Separation was carried out with temperature programming. Samples for gas–liquid chromatography (0.2 mL each) were pre-treated by centrifugation at 12,600 g for 2 min, followed by stepwise acidification of the clear supernatants with H₃PO₄ and 5 M formic acid to the pH of 2.0. The detection limit of the method was 0.2 mM.

The Fe²⁺/Fe³⁺ ratio of minerals was determined by ⁵⁷Fe Mössbauer spectroscopy. This method allows for the determination and quantification of different atomic environments, magnetic states, chemical states and transformations of iron-containing compounds (Kamnev and Tugarova, 2021). All Mössbauer measurements were performed at room temperature using the MS-1101 Em spectrometer, operating in the constant acceleration mode, with a ⁵⁷Co source in the Rh matrix. The calibration was carried out relating to α-Fe, and the spectra were fitted using SpectrRelax software. To process the spectra, a model fitting was carried out simultaneously with the extraction of the hyperfine magnetic field distribution.

The composition of the microbial community of Yessentuki mineral water and of the enrichments was determined by amplification of the hypervariable V4 region of the 16S rRNA genes, followed by sequencing and bioinformatic data processing. Total DNA was isolated with the FastDNA^™ Spin Kit for Soil DNA Extraction (MP Biomedicals, United States) according to the manufacturer’s instructions using the FastPrep-24™ 5G Bead Beating System (MP Bio, United States). Amplicon libraries were prepared as described by Gohl et al. (2016) using the following primers: 515F (5′-GTG BCA GCM GCC GCG GTA A-3′) and 806R (5′-GAC TAC NVG GGT MTC TAA TCC-3′) (Hugerth et al., 2014; Merkel et al., 2019) including Illumina technical sequences (Fadrosh et al., 2014). High-throughput sequencing of the libraries was performed using MiSeq Reagent Micro Kit v2 (300-cycles) MS-103-1,002 (Illumina, United States) on a MiSeq sequencer (Illumina, United States) according to the manufacturer’s instructions. Libraries were prepared and sequenced in two replicates for each sample. Amplicon sequence variants (ASV) were obtained using the Dada2 script (Callahan et al., 2016) and the SILVA 138.1 database (Quast et al., 2013). Analysis of the ASV tables was performed using Rhea (Lagkouvardos et al., 2017). All sequencing data were deposited in SRA (NCBI) under BioProject ID PRJNA945437.

A WGS library preparation and sequencing of Dethiobacter alkaliphilus Z-1002 was performed in BioSpark Ltd., Moscow, Russia, using KAPA HyperPlus Library Preparation Kit (KAPA Biosystems, United Kingdom) according to the manufacturer’s protocol and NovaSeq 6,000 system (Illumina, San Diego, CA, United States) with a reagent kit capable of read 100 nucleotides from each end. This Whole Genome Shotgun project has been deposited at DDBJ/ENA/GenBank under the accession JAPDNO000000000. Gene search, annotation, and genome-based phylogenetic reconstructions were performed as previously described (Khomyakova et al., 2022) with an additional use of PGAP service (Li et al., 2021).

Screening of the genomes of both D. alkaliphilus strains for multiheme cytochromes and their sequence analysis was performed as previously described (Toshchakov et al., 2018) using reported cytochromes, involved in extracellular electron transfer (EET) in Geobacter sulfurreducens, Shewanella oneidensis, “Thermincola potens,” Carboxydothermus ferrireducens (Gavrilov et al., 2021 and references therein), the dataset of cytochrome query sequences was supplemented with multiheme proteins reported to be involved in Fe(III) respiration or other EET processes in the thermophilic Fe(III) reducers Carboxydocella thermautotrophica (Toshchakov et al., 2018) and Melioribacter roseus (Gavrilov et al., 2017), anaerobic Fe(II) oxidizing bacteria Sideroxydans lithotrophicus, Gallionella capsiferriformans, and Dechloromonas aromatica (Chakraborty et al., 2005; Emerson et al., 2007; Liu et al., 2012), Fe(III) reducing and syntrophic archaea (Mardanov et al., 2015; Smith et al., 2015; Krukenberg et al., 2018; Kashyap and Holden, 2021). Heme-binding motifs in D. alkaliphilus multihemes were predicted as previously described (Mardanov et al., 2015). Conservative domains, transmembrane helices and signal peptides were predicted using the hmmscan¹ web-service (Potter et al., 2018) with default parameters and all databases included.

The cytochrome c protein sequences of SirA octaheme sulfite reductase from S. oneidensis and the MtoA Fe(II)-oxidizing decaheme of D. aromatica were retrieved from the non-redundant NCBI protein database on November 2022. Homologs for each of these sequences were separately screened for using the BLASTp algorithm and the obtained sets of sequences were manually curated and processed as described previously (Gavrilov et al., 2021). The resulting sets of 48 amino acid sequences for the SirA query and 342 sequences for MtoA query were amended with the sequences of SirA, or the MtoA homologs from the genomes of D. alkaliphilus strains (Supplementary Table S1). The immunoglobulin-like domains were omitted from the sequences of D. alkaliphilus to decrease non-specific alignments within the protein sets. The sets were then aligned using MAFFT 7.490 with default parameters, with 1,000 iterations of the FFT-NS-i refinement method (Katoh et al., 2019). Two final alignments (for SirA and MtoA hits) were subjected to a Bayesian inference and used to construct unrooted phylogenetic consensus trees as described previously (Gavrilov et al., 2021).

For proteomic analysis, biomass of D. alkaliphilus strain AHT1^T grown with H₂ and thiosulfate or SF as the electron acceptors was harvested from 100 mL cultures by centrifugation at 16,000 g for 15 min. Both cultures were grown in triplicate. Before harvesting, the SF-grown culture was separated into liquid and mineral phases using low speed centrifugation (5 min, 1,000 g). The mineral phase was ultrasonicated inside the centrifuge bags in an ultrasonic bath (Sapphire, Russia) at 99% power, 5 min, to separate adherent cells from the minerals. The ultrasonicated mineral phase was then combined with the liquid phase, mixed thoroughly, the bulk of the minerals were held in the bags with a hand magnet and the remaining suspension was transferred to another centrifuge bag to harvest the biomass at 13,000 g for 20 min.

Cell lysis, reduction, alkylation and digestion of the proteins, as well as the processing and LC-MS/MS analysis of the obtained peptide sets were performed as previously described (Gavrilov et al., 2021). Label-free protein quantification was made by MaxQuant software version 1.5.6.5 using D. alkaliphilus strain AHT¹ amino acid FASTA dataset and a common contaminant database through the Andromeda search engine according to the previously described protocol (Gavrilov et al., 2021). The iBAQ algorithm (Schwanhäusser et al., 2011) implemented in MaxQuant software was used to quantify proteins in each sample. Normalization of each protein’s iBAQ value to the sum of all iBAQ values generated a relative iBAQ (riBAQ) values corresponding to the molar percentage of each protein in the sample, with the entire set of proteins in the sample taken as 100% (Shin et al., 2013).

The mass spectrometry proteomics data were deposited at the ProteomeXchange Consortium via the PRIDE (Perez-Riverol et al., 2022) partner repository with the dataset identifier PXD040929.

We used 16S rRNA gene sequences and related metadata to analyze the relative abundance of Dethiobacters in different environments. For this purpose, we took all sequences assigned to the Dethiobacter genus in the SILVA database 138.1 (Quast et al., 2013). Next, we analyzed the similarity of these sequences to the 16S rRNA gene sequences of D. alkaliphilus strains AHT1^T and Z-1002 using BLASTN (Altschul et al., 1990), discarding all sequences with less than 94.5% similarity to these cultured strains of the genus. Finally, to analyze the distribution of the genus Dethiobacter in different types of ecotopes, we obtained 108 sequences of 16S rRNA genes. Next, we analyzed the metadata related to the detection sources of these sequences.

To analyze the distribution of the representatives of the Dethiobacteraceae family (Sorokin and Merkel, 2019), we used the metadata on MAGs from the GTDB 207 database (Parks et al., 2022), i.e., the information on isolation sources, description of sampling sites, geological or geochemical information on them if publicly available.

The Fe(III)-reducing strain of D. alkaliphilus, Z-1002, was previously isolated in a pure culture with SF and formate under highly alkaline conditions (Zavarzina et al., 2018). Here, we describe its metabolic characteristics in detail. Strain Z-1002 appeared to be an obligate alkaliphile with the pH growth range from 7.8 to 10.1 and an optimum at pH 9.2. It is an obligate natronophile growing in the range of carbonate/bicarbonate concentration ratios from 13.5/7.5 to 82.0/45.0 g L⁻¹ with an optimum at 40.0/20.0 g L⁻¹. It could not grow without carbonates in the presence of CHES buffer after the second transfer from a carbonaceous medium. It was halotolerant and grew in the range of NaCl concentrations from 0 to 120 g L⁻¹ with an optimum at 50–60 g L⁻¹. Strain Z-1002 grew chemoorganotrophically with lactate, succinate, butyrate, pyruvate, propionate, or ethanol as the electron donors and SF or thiosulfate as the electron acceptors. Slow growth with a low cell yield of 5 × 10⁵ cells mL⁻¹, accompanied by the production of 2.5 mM sulfide, was observed on thiosulfate with the addition of formate or H₂ as electron donor. The most intense thiosulfate reduction accompanied by the formation of 15 mM sulfide was observed with ethanol as the electron donor. Strain Z-1002 was unable to use elemental sulfur and sulfate as the electron acceptors, as well as it was unable to disproportionate sulfur or thiosulfate. Strain Z-1002 was unable to grow by fermentation of carbohydrates, peptides, and amino acids.

Considering the metabolic features of D. alkaliphilus isolate from the lake Magadi, we tested the ability of the type strain AHT1^T for dissimilatory iron reduction. After three sequential transfers on the medium supplemented with SF and H₂ in the absence of yeast extract and sodium sulfide, strain AHT1^T produced 4.1 ± 0.5 mM Fe(II) from ferrihydrite with concomitant oxidation of 1.6 ± 0.2 mM molecular hydrogen (Figure 1). The mineral phase changed from brownish to dark brown color during the growth, indicating the formation of Fe(II)-containing minerals (Supplementary Figures S1). The cell yield under these conditions was 3.5 ± 2 × 10⁷ cell mL⁻¹. Ferrihydrite-grown cells of both D. alkaliphilus strains were strongly associated with iron minerals (Figure 2).

Cultures of the strain Z-1002 appeared to oxidize Fe(II) from a natural mixture of hydrothermal siderite (76%), green rust (11%), and iron oxides (13%) under anaerobic conditions in the absence of any organic compounds. The maximum cell yield under these growth conditions was 10⁷ cells mL⁻¹ and was observed by the 34th hour of incubation (Supplementary Figure S2). Growth was accompanied by the changes in the mineral phase as no soluble Fe(III) or Fe(II) species were detected in the culture during the incubation. Mössbauer analysis of the minerals from the grown cultures (stationary growth phase) and uninoculated controls revealed a decrease in the relative total intensity of the spectra corresponding to ferrous iron atoms from 87.7 ± 2.9% to 78.2 ± 0.9% in the grown cultures (Table 1). No such changes were observed in the controls. The relative intensity of siderite subspectrum of the microbially impacted mineral sample remained virtually unchanged, but the intensity of the doublet corresponding to Fe²⁺ atoms in the green rust structure decreased from 8.4 ± 2.1% to 2.1 ± 0.6%, while the intensity of the doublet corresponding to Fe³⁺ atoms in this phase increased from 2.2 ± 0.5% to 4.6 ± 0.4% (Table 1 and Supplementary Figure S3). This means that 75% of the bivalent iron atoms contained in the green rust were oxidized. At the end of the incubation, we also observed an increase of 4.2% in the magnetically ordered phase content of the mineral mixture. Thus, on the one hand, we have observed a rearrangement of the bivalent and trivalent iron atoms in the green rust, and on the other hand, an increase in the amount of iron oxides in the mixture.

To assess the relevance of the identified multihemes for EET and to detect any accessory proteins involved in electron transfer to Fe(III) or sulfur compounds, as well as from Fe(II) minerals, the results of a shotgun proteomic analysis of cells, harvested at late logarithmic growth phase, were compared across two different cultivation conditions for each strain. The strain AHT1^T was cultured with ferrihydrite or thiosulfate, and the strain Z-1002 was cultured with ferrihydrite or the mixture of Fe(II) minerals (siderite, green rust, and iron oxides). In each case, the triplicate samples were taken from three different cultures (in three biological replicas).

In the case of the AHT1^T strain, tandem MS revealed 1910 different proteins identified with a least of two unique peptides. The number of proteins identified for each sample is provided in Supplementary Table S3. Profiling of individual multiheme proteins revealed 17 EET-related multihemes that were differentially expressed under different growth conditions. A clear difference in multiheme cytochrome profiles was observed between ferrihydrite- and thiosulfate-grown cells. However, several proteins, such as DealDRAFT_1449 homologous to a CymA-like quinol oxidase (Supplementary Table S1), were expressed at the same level at both culture conditions (Figure 6A). In contrast, the protein DealDRAFT_1,457 appeared to be the only one exclusively expressed in thiosulfate reducing cells (Figure 6A). In ferrihydrite-grown cells of strain AHT1^T, the highest relative abundance was observed for DealDRAFT-1,087 which is homologous to the SmhC (23% identity) Fe(III)-specific secreted multiheme of C. ferrireducens (Supplementary Table S1). In thiosulfate-grown AHT1^T cells, the highest expression level was detected for the multiheme DealDRAFT_2,539 (Figure 6A) but in this case, the difference in the expression level was not statistically supported by the imputation procedure due to the absence of DealDRAFT_2,539 protein in one of the three preparations of the thiosulfate-grown biomass. Statistically relevant upregulation was observed for the proteins DealDRAFT_0324, 1087, 1093, 1435, 1439, and 1441 under ferrihydrite reduction and for the proteins DealDRAFT_1443, 1451, 1670, and 1917 (Figure 6B) under thiosulfate reduction. In addition, the cytochrome DealDRAFT_2033 was overexpressed in thiosulfate-grown cells (Figure 6A) but the absence of this protein in two of three preparations of ferrihydrite-grown biomass did not statistically support the differences in its expression level. The majority of the proteins with positive response to ferrihydrite provided as an electron acceptor are encoded in the clusters 1-Fe-T, 2-Fe-T, 3_Fe-T (Figure 4) and share homology with putative terminal Fe(III) reductases, secreted multihemes, and MtoA putative Fe(II) oxidase which is phylogenetically related to MtrA/MtrD Fe(III) oxidases (Shi et al., 2012). Interestingly, four of the six multihemes upregulated in thiosulfate-grown cells (DealDRAFT_1443, 1451, 1670, and 2539) were also homologous to various cytochromes related to Fe(III) reduction (Supplementary Table S1). Only a single protein homologous to SirA sulfite reductase (22% identity), DealDRAFT_1917, was reliably upregulated during D. alkaliphilus growth on thiosulfate.

In the strain Z-1002, tandem MS identified 1,390 different proteins identified with at least two unique peptides. The number of proteins identified per each sample is provided in Supplementary Table S4. Profiling of individual multiheme proteins uncovered nine EET-related multihemes that were differentially expressed under ferrihydrite reduction or Fe(II) oxidation. The cytochromes OMD50_RS14355, 14365, 14415, and 14880, homologous to MtoA Fe(II) oxidases, CymA-like quinol oxidase and OmhA terminal Fe(III) reductase, respectively (Supplementary Table S1), had almost similar expression levels under ferrihydrite reduction or Fe(II) oxidation (Figure 6C). The highest expression levels under both Fe(III)-reducing and Fe(II)-oxidizing conditions were observed for the OMD50_RS14385 and OMD50_RS14880 proteins (Supplementary Table S4) which are homologous to cell surface-associated cytochromes of C. ferrireducens (30 and 26% identity, respectively, Supplementary Table S1). These proteins are likely to comprise the core part of the EET pathway driving the redox transformation of Fe minerals in D. alkaliphilus. Interestingly, the cytochromes OMD50_RS02905 and 7,060, homologous to OmhA Fe(III) reductase and SirA sulfite reductase (25 and 23% identity, respectively) were exclusively expressed under Fe(II) oxidation, whereas a putative quinol oxidase OMD50_RS09530 showed the strongest upregulation under the same growth conditions (Figure 6C).

We have detected a phylotype belonging to the genus Dethiobacter in a subsurface mineral water sample collected from the Lower Cretaceous aquifer of YMWD through well 9. The phylotype comprised less than 0.1% of the microbial diversity of this sample (Figure 7, 1st column). The sample was then stored at +4°C for 1.5 years and then used for the enrichment of prokaryotes inhabiting the deep subsurface aquifer using a mixture of electron donors and acceptors available in this environment. This primary enrichment containing both sulfate and ferrihydrite (as SF) and incubated at +30°C for 3 weeks was dominated by typical sulfate reducing taxa but carried out the reduction of SF to a black precipitate. No Dethiobacter representatives were detected in the water sample after its long-term storage (1.5 years) or in the primary enrichment at the end of its 3-weeks incubation (Figure 7, 2nd and 3rd columns). To check the presence of iron reducers in the obtained enrichment, it was transferred to selective media containing formate (10 mM) or acetate (10 mM) with SF (50 mM final Fe(III) content) as the only electron acceptor. After 2 weeks of incubation, the color of the Fe mineral changed from reddish to dark-brown, which is characteristic of ferrihydrite reduction. Fluorescence microscopy of mineral samples taken from these cultures revealed that mineral particles were densely settled with rod-shaped cells of different length. 16S rRNA gene profiling of the Fe(III) reducing enrichments revealed the predominance of Dethiobacter phylotypes (96% identity with D. alkaliphilus) in the culture with SF and formate (33.2% relative abundance) and their complete absence in the culture with SF and acetate (Figure 7, 4th and 5th columns, respectively).

The survey of the environmental distribution of 16S rRNA gene sequences of the genus Dethiobacter available in public databases shows that about 25% of the sequences are retrieved from sediments of soda lakes and alkaline soils, 35% of the sequences are retrieved from Fe-enriched serpentinites, about 30%—from anaerobic bioreactors and digesters (Supplementary Figure S5). The remaining small fraction of sequences was found in microbial mats, ground and urban waters. Thus, Dethiobacter is the most abundant in sulfur- or iron-enriched alkaline natural habitats. However, the representatives of the genus have also been detected in anthropogenic organics-rich neutrophilic environments. Unfortunately, the lack of publicly available data on the physico-chemical conditions sustained in these anthropogenic environments, and on the sources of wastewaters treated there do not allow us to discuss the phylogenetic group of Dethiobacteraceae detected in bioreactors in the context of the evolution of their metabolic capacities.

Both D. alkaliphilus strains appeared to lack canonical genomic determinants of sulfur, sulfate or thiosulfate reduction, such as DsrAB or molybdopterin-containing complexes. Instead, each of the strains possessed two octaheme c-type cytochromes homologous to the SirA sulfite reductase or Otr tetrathionate reductase of S. oneidensis (Supplementary Table S1). The sulfur-reducing type strain possesses an additional weak homolog of Otr proteins, DealDRAFT_1454 (20% identity). Phylogenetic reconstruction of these putative determinants of sulfur compounds reduction revealed deep branching of three of them from the common ancestor of SirA and Otr proteins, and association of two D. alkaliphilus proteins with OcwA/OmhA Fe(III) reductases (Figure 5), which are structurally similar to, but phylogenetically distant from octaheme sulfite or tetrathionate reductases (Soares et al., 2022).

Multiheme cytochrome profiling revealed upregulation of the protein DealDRAFT_1917, homologous to SirA, in the cells of the strain AHT1^T grown with thiosulfate (Figure 6). Interestingly, the protein DealDRAFT_1457 homologous to MtoA Fe(II) oxidase was only detected in thiosulfate-grown cells. Furthermore, the most pronounced upregulation in thiosulfate-reducing cells was observed for DealDRAFT_1451 and 1670 (Figure 6) homologous to the putative Fe(III) reductases SmhB and SmhC (21 and 32% identity, respectively, Supplementary Table S1). The highest expression level in these cells was detected for the protein DealDRAFT_2539 homologous (24% identity) to OmcN Fe(III)-reducing outer surface cytochrome of G. sulfurreducens (Aklujkar et al., 2013). All these facts emphasize that the reduction of sulfur compounds in D. alkaliphilus is most likely driven by multiheme cytochromes homologous to Fe-cycling proteins.

We have experimentally demonstrated the ability of the sulfur-reducing strain AHT1^T to reduce Fe(III) from SF and the ability of the Fe(III)-reducing strain Z-1002 to oxidize Fe(II) from green rust supplied as a mixture with siderite. The result of Fe(II) oxidation by strain Z-1002 was indicated by a 6.4% increase in the relative intensity of the Mössbauer spectra of Fe³⁺ atoms compared to that of uninoculated controls (Table 1).

Genomic analysis revealed that each of the D. alkaliphilus strains contained a large set of genes encoding multiheme c-type cytochromes, including those homologous to quinol-oxidases, membrane-associated Fe(III) reductases, such as OcwA or MtrA/D, and soluble electron shuttling cytochromes, previously identified in the model Fe(III)-respiring bacteria of the genera Shewanella and Geobacter or in Gram-positive thermophilic Fe(III) reducers (Figure 4 and Supplementary Table S1). Taken together, these cytochromes could be combined in an electron transfer chain linking the membrane-bound respiratory complexes with extracellular electron acceptors, such as Fe(III)-containing minerals, or electron donors, such as Fe(II)-bearing mineral mixtures. Proteomic profiling of strain AHT1^T for multihemes revealed strong upregulation of six different cytochromes, including the cytochrome DealDRAFT_1439 (Figure 6), encoded in three gene clusters of putative Fe(III) reductases (clusters 1-Fe-T, 2-Fe-T, 3-Fe-T, Figure 4). These cytochromes contain the homologs of all the necessary components of an extracellular electron transfer chain to Fe(III) oxides. Interestingly, the Fe(II)-oxidizing strain Z-1002 possesses four different homologs of the protein DealDRAFT_1,439 which are encoded in the 2-Fe and 3-Fe clusters. Of these, the proteins OMD50_RS14355 and OMD50_RS14365 share similarity with MtoA-type Fe(II)-oxidases (Shi et al., 2012). Phylogeny reconstruction of these and other MtoA homologs of D. alkaliphilus (DealDRAFT_1454, 1457, OMD50_RS14520) revealed clear separation of these proteins from DmsE-family decahemes of other organisms that perform Fe(III) reduction or Fe(II) oxidation (Supplementary Figure S4). Branching most deeply from the common ancestor of all the D. alkaliphilus decahemes is the phylogenetic cluster of MtoA-like cytochromes DealDRAFT_1439, OMD50_RS14355, and OMD50_RS14365 (Supplementary Figure S4), as well as the protein OMD50_RS14375 with closer homology (24% identity) to the MtrD Fe(III)-reducing cytochrome of S. oneidensis (Shi et al., 2012). We propose that three of the four proteins in this phylogenetic cluster might appear in strain Z-1002 by several duplication events with further minor changes occurred under the evolutionary pressure of the conditions favoring Fe-cycling, such as increased abundance of mixed Fe(III)/Fe(II)-containing minerals. Such an assumption correlates with the similarity of expression levels of two MtoA-like cytochromes in strain Z-1002 cells grown under ferrihydrite reducing or Fe(II) oxidizing conditions (Supplementary Table S4 and Figure 6C). In general, multi-omics analysis revealed that the biochemical machinery of Fe redox cycling in D. alkaliphilus strains is more complex and flexible than that of sulfur respiration.

The most likely quinol oxidizing/quinone reducing proteins of the respiratory chains in both D. alkaliphilus strains are the CymA-like homologs DealDRAFT_1449 and OMD50_RS14415 with their stable expression level almost independent of the electron acceptors or donors provided (Figure 6). Importantly, the majority of the cytochromes overexpressed under ferrihydrite- or thiosulfate-reducing conditions in the strain AHT1^T are encoded in three largest gene clusters (Figure 4), which also encode auxiliary flavin-containing electron transfer proteins together with NHL- and TPR-repeat-containing proteins that could be involved in protein–protein interactions mediating the assembly of multiprotein complexes (D’Andrea and Regan, 2003), such as MtrCAB Fe(III)-reducing ones (Shi et al., 2012). Interestingly, the putative Fe-cycling multiheme proteins of D. alkaliphilus showed greater divergence from each other than that observed among the EET-related proteins of the DmsE family from fairly different bacteria. Another important point is that the homologs of OmhA Fe(III) reductase and SirA sulfite reductase are expressed only under Fe(II) oxidation in strain Z-1002. These facts highlight the intricate weaving of EET pathways, that determine the redox transformations of sulfur and iron compounds in the organism, which could be a manifestation of the active evolution of multiheme proteins within D. alkaliphilus species, characteristic of the ongoing adaptation of an evolving population to the changes in its environment (Zheng et al., 2019).

Our analysis of the distribution of Dethiobacters in natural environments revealed that the representatives of this genus show strong addiction to two peculiar alkaline, sulfur- or iron-enriched ecotopes, soda lakes and serpentinite-associated sediments, respectively (Supplementary Figure S5). Serpentinizing environments have been stable throughout the Earth’s geological evolution, whereas soda lakes are subjected to continuous biogeochemical development, being surfaceous ecosystems (Stüeken et al., 2015; Furnes and Dilek, 2022). Their most pronounced restructuring occurred after the Great Oxidation Event followed by the accumulation of sulfates in the sediments and waters of the lakes (Montinaro and Strauss, 2016). Since then, the biogeochemical cycling of sulfur has become one of the major factors determining the physicochemical conditions in soda lakes and influencing the evolution of their inhabitants.

The phylogenetic reconstruction of the determinants of key catabolic processes, sustaining the growth of two D. alkaliphilus strains (Figure 5 and Supplementary Figure S4), offers insights into the evolutionary traits which led to the occupation of two different ecological niches by this bacterial species. The presence of multiheme cytochromes, sharing their phylogenetic root with OmhA/OcwA Fe(III)-reductases, as the only possible determinants of the reduction of sulfur compounds in D. alkaliphilus allows us to propose that soda lakes are secondary habitats for these organisms comparing to Fe-rich subsurface environments associated with serpentinites. This hypothesis correlates with the geological history and current geochemical characteristics of Magadi soda lake from which the Fe-reducing strain Z-1002 was isolated. This lake is located in a geologically young East African Rift Valley characterized by high pH, silica and carbonate concentration. The ecotope was formed ca. 1 million years ago due to faulting of the Rift Valley composed of alkali Pleistocene basalt, trachyte lava flows and phonolite (Jones et al., 1977; Eugster, 1980; Schagerl and Renaut, 2016). Thus, the bedrock of the modern lake appeared to be enriched with iron minerals. In this case, the Fe-reducing D. alkaliphilus could be an alien species for lacustrine sulfur-rich sediments, which was introduced there from the underlying Fe-rich volcanic rocks. The strain was managed to modify its multiheme cytochrome machinery under the evolutionary pressure of renewed geochemical settings with predominant sulfur cycle. A large number of different multiheme cytochrome genes and complete absence of typical genes determining the redox transformations of sulfur compounds could favor the adaptation of Dethiobacter to an ecotope where Fe-rich rocks are combined with sulfur-enriched sediments and waters.

Better fitness of the metabolic features of Dethiobacters to the life in Fe-rich alkaline environments was highlighted by the enrichment of their phylotypes from the Lower Cretaceous aquifer of the Yessentukskoye subsurface mineral water deposit (YMWD). The crystalline basement of the deposit is represented by Proterozoic-Paleozoic metamorphic and magmatic shales and granites, i.e., the rocks depleted with iron-bearing minerals. The sedimentary cover of the YMWD is represented by iron-depleted limestones, mudstones and siltstones of Meso-Cenozoic age. The Lower Cretaceous aquifer of the YMWD, penetrated by the well 9 which we have sampled, is directly connected with the recharge area and contains carbon-free sulfaceous alkaline waters (Elena et al., 2020; Filimonova et al., 2022). This environment is completely different from those formed by serpentinization processes. Not surprisingly, Dethiobacter-related phylotypes comprised a minor part (0.08%) of the microbial community in this aquifer and rather represented the so-called “rare biosphere,” which is considered a phenotypic repository of microbial communities getting advantages upon significant changes of environmental conditions (Jousset et al., 2017). In our experiments, Dethiobacter phylotypes appeared to pass undetected through a long-term storage of a mineral water sample and its further cultivation under sulfate-reducing conditions. The phylotypes restored when the enrichment conditions became favorable for alkaliphilic lithotrophic Fe(III) reducers. In this case, Dethiobacter representatives became the dominant group of the enrichment culture (Figure 7). Thus, Fe-rich alkaline conditions seem to be optimal geochemical settings for yet uncultured Dethiobacter species. This fact together with the unity of origin of the sulfur, thiosulfate and Fe(III) reduction pathways in D. alkaliphilus, allows us to propose an adaptation strategy of the organism to the change of its environment from serpentinizing Fe-rich ecotopes to soda lakes. This strategy is based on the changes in the multiheme cytochrome repertoire aimed to get energy from the reduction of electron acceptors with lower redox potential (sulfur compounds vs. Fe(III) minerals). Such an example of intraspecific microevolution within D. alkaliphilus shows a possible way of a global adaptive response of prokaryotes to the activation of the sulfur cycle after the appearance of sulfates in the oceanic water and free oxygen in the atmosphere during GOE (Catling and Zahnle, 2020).

In addition to the characteristics given for the type strain of the species, AHT1^T (Sorokin et al., 2008; Poser et al., 2013, 2016; Sorokin and Merkel, 2019), and strain Z-1002 (Zavarzina et al., 2018), the following characteristics should be added to the formal description: both strains of the type species can grow by iron-reduction in the presence of molecular hydrogen, formate, acetate, lactate, succinate, pyruvate, butyrate, propionate or ethanol as the electron donors. Strain Z-1002 is unable to reduce elemental sulfur or disproportionate thiosulfate and elemental sulfur. Strain Z-1002 is able to reduce thiosulfate or anaerobically oxidize Fe(II)-containing minerals.