Halophilic Anaerobic Cultures Enriched with CO₂:H₂ from Different Saline Environments Reveal Unknown Autotrophic Bacterial Diversity and Modular Carbon Fixation Pathways

Abstract

The subsurface sediments of saline-aquatic systems host diverse microbes, with unclear ecological roles and challenging lab cultivability. Chemolithotrophic anaerobes involved in CO₂-fixation are one of the poorly studied groups. This study focused on understanding these bacteria from subsurface sediments of four representative saline environments, two marine (i.e., Coastal Arabian and Bay of Bengal seas) and two lake (Sambhar and Lonar) systems through enrichment and metagenomics. Enrichment cultures with bicarbonate/CO₂ and hydrogen as the carbon and energy sources, respectively, showed CO₂ fixation, producing acetic and formic acids as the major organic products. Enriched culture with Sambhar Lake sediment produced more formic acid (391 ± 8 mg/L) than acetic acid (92 ± 20 mg/L); however, other enriched cultures produced considerably higher acetic acid (up to 966 ± 24 mg/L) than formic acid (up to 367 ± 30 mg/L). The organics production was accompanied by unique thread-like (up to 500 μm long) aggregates, harbouring chains of rod and oval-shaped microbes in all cultures. Metagenome sequencing revealed dominance of Vibrio spp. (relative sequence abundance of 91% to 97%) across all cultures, while canonical CO₂-fixing taxa were nearly absent (< 0.01%). KEGG analysis revealed partial genes for various CO₂ fixation pathways, including Wood-Ljungdahl, reverse-TCA, dicarboxylate-hydroxybutyrate, hydroxypropionate bicycle, hydroxypropionate-hydroxybutyrate, and the reductive-glycine pathway. The presence of a near-complete serine variant of the reductive glycine pathway, which has been demonstrated in engineered systems, suggests that this pathway may play an operational role in natural systems. The consistent production of organic acids and incomplete pathway representation suggests modular CO₂ fixation within the Vibrio-dominated enriched mixed cultures.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00248-025-02654-6.

Introduction

Saline environments exhibit various physicochemical characteristics, including salinity, temperature, pH, pressure, and a range of redox processes and conditions. The diversity of these ecosystems is reflected in the variety of microbial communities that inhabit them. They are important sites of nutrient regeneration and carbon burial, shaping phylogenetically and metabolically diverse halophilic microbial communities [1, 2]. While diversity surveys have uncovered a wide range of previously undiscovered microorganisms in extreme environments, it is challenging to gain deeper insights into their physiology and ecological roles. This is primarily due to the lack of enrichment and pure cultures, as it is challenging to mimic the same physiological conditions in laboratories [3, 4]. Aquatic sediments comprise clay, decomposing organic material, calcium-based remnants, and various solid materials. The underlying sub-surface sediment layers remain anoxic, and the biogeochemical composition and organic matter content in sediments play a crucial role in determining whether autotrophic or heterotrophic processes dominate such systems [5, 6]. The anoxic environments are characterized by high concentrations of oxidized inorganic compounds such as nitrate, sulfate, manganese oxides, and iron oxyhydroxides [7–9], with microbial fermentation and anaerobic respiration as the predominant metabolic processes [10]. Organic matter oxidation in sediments proceeds through multiple enzymatic reactions involving different microorganisms and oxidative and reductive reactions coupled with electron donors and acceptors. Electron acceptors such as O₂, Mn(VI), NO₃⁻, Fe(III), SO₄²⁻, and CO₂ are commonly used in the order of their decreasing redox potential, followed by methanogenesis and/or fermentation [11–13]. Therefore, sediment microbes exhibit a great deal of metabolic diversity, encompassing phototrophic, lithotrophic, and heterotrophic pathways for energy production [5, 14].

The intensive anaerobic mineralization in sediments results from a series of processes in which products of one metabolic group of microorganisms form the substrate for others. The microbial communities thus formed, therefore, consist of groups that are highly dependent on each other’s metabolic activities. The participants of such processes are of particular interest for studying microbial interactions that occur in these habitats [15]. In these environments, chemolithoautotrophy is the primary process for generating the organic matter to support heterotrophic growth [16, 17]. For instance, carbon isotopic labelling has revealed dark CO₂ fixation in the sediments of three cascade reservoirs in the middle reaches of the Wujiang River basin (Hongfeng Reservoir, Dongfeng, and Canyon Reservoir) [18]. Additionally, CO₂ fixation has been documented in the Cariaco Basin’s redox transition zone, primarily by chemolithoautotrophic sulfur-oxidizing bacteria [19]. Many sulfate-reducing species in marine sediment incubations are autotrophs that can fix inorganic carbon using H₂ as an electron donor, indicating their role as CO₂ assimilators in regulating carbon fluxes in marine sediments, according to a recent study on enrichment with temperate and cold marine sediments [20]. Understanding of the physiology and metabolic capabilities of chemolithoautotrophic microbial groups in saline environments is often limited. This is mainly because of the limited dedicated studies on the enrichment and detailed characterization of these microbes from marine or saline lake sediments. Moreover, the pure cultures of these microbes have barely been isolated and characterized in detail. For instance, a bacterial strain known as YN-1, capable of fixing CO₂ under conditions containing H₂, O₂, and CO₂, has been reported from seawater [21]. An obligate anaerobic halophilic bacterium Acetohalobium arabacticum, which requires 10–25% NaCl (optimum at 15–18%) for growth, has been reported to fix carbon autotrophically under H₂:CO₂ conditions [22]. In addition, the strain Fuchsiella alkaliacetigena Z-7100T, an alkaliphilic, strictly anaerobic, hydrogen-utilizing homoacetogen, was isolated from sediment samples collected from the soda lake Tanatar III in Atlay, Russia. This strain is reported to grow chemolithotrophically with CO₂ and H_2, producing acetate as the sole metabolic product [23]. From the same site, two additional haloalkaliphilic homoacetogenic bacterial strains, namely Fuchsiella ferrireducens Z-7101^T and Z-7102, were reported. These strains are also capable of chemolithotrophic growth with H₂ and CO_2, producing acetate as the only metabolic product [24]. A deep-sea bacterium, Sulfurovum lithotrophicum strain 42BKT^T has also been shown to grow chemolithoautotrophically, utilizing either elemental sulfur or thiosulfate as the sole electron donor and oxygen (at 5% in the gas phase) or nitrate as an electron acceptor [25]. Another haloalkaliphilic CO-utilizing acetogen, Natranaerofaba carboxydovora, isolated from hypersaline soda lake sediments (Kulunda Steppe, Altai region, Russia), is capable of using CO as both a carbon and energy source under 4.5 M salinity and an alkaline pH (9 to 10.5), but is not reported to grow on H₂ and CO₂ [26]. Additionally, a facultative anaerobe, Varunavibrio sulfuroxidans TC8^T isolated from a sulfidic shallow-water marine gas vent in the Tyrrhenian Sea, Italy, has been reported to grow chemolithoautotrophically with CO₂ as a carbon source, sulfur and thiosulfate as electron donors, and nitrate, oxygen (5% v/v), and ferric iron as electron acceptors [27]. The Supplementary Table S1 highlights the reported chemolithotrophic isolates from saline environments. Overall, there are limited pure culture isolates capable of growing on H₂:CO₂; therefore, scarce information on microbial CO₂ fixation in saline environments restricts understanding of the diversity and ecology of this important microbial group. It also limits their possible use for developing CO₂ bioconversion technologies, such as gas fermentation and microbial electrosynthesis approaches. For instance, saline electrolytes in microbial electrosynthesis reactors can lower the ohmic losses, thereby improving the performance and energetic efficiency of the system [28]. The use of CO₂:H₂ fixing microbes from saline habitats thus offers a promising way to overcome the ohmic losses-related limitation of the microbial electrosynthesis process [29]. Moreover, microbes inhabiting saline environments represent a large portion of the biosphere. Gaining insight into their ecological roles is crucial for understanding global carbon cycling processes and other biogeochemical processes such as organic mineralization and biomass production. Understanding their metabolic processes can transform the existing view of these microbial diversities and functionalities [30].

Given the apparent knowledge gap in understanding the chemolithoautotrophic microbes in anoxic saline sediments, this study aimed to understand the diversity of these microbes from the subsurface sediments of four representative saline habitats in India: two marine (Coastal Arabian and Bay of Bengal seas) and two lake systems (Lonar and Sambhar) through the enrichment culture approach. Sub-surface sediments from marine and lake systems were used as an inoculum to enrich CO₂-fixing lithotrophs. CO₂/bicarbonate was provided as the sole carbon source, and H₂ served as the sole electron donor under saline conditions. The CO₂ fixation products were analyzed to confirm their lithoautotrophy. The enriched mixed microbial communities were characterized using microscopy techniques, and metagenome sequencing was employed to elucidate microbial diversity and CO₂ fixation metabolic pathways.

Materials and Methods

Sediment Sampling, Characterization and Inoculum Preparation

Subsurface sediment samples from a depth of about 10–15 cm were collected in air-tight amber containers from the coastal area of four saline aquatic systems: two marine habitats (i.e., Shivrajpur Beach, Arabian Sea, Gujarat, and Puri Beach, Bay of Bengal, Orissa) and two lakes (i.e., Lonar Lake, Maharashtra, and Sambhar Lake, Rajasthan) geographically located in different parts of India. The coordinates for the sampling sites are as follows: Shivrajpur Beach (22.301723° N, 68.975642° E), Arabian Sea, Gujarat; Puri Beach (19.794749° N, 85.825288° E), Bay of Bengal, Orissa; Lonar Lake (19.97884° N, 76.50814° E), Maharashtra; and Sambhar Lake (26.906117° N, 75.077056° E). The sediment samples were used immediately for enrichment purposes and stored at 4 °C when not in use. All samples were analyzed for pH, ammonium ions, phosphate, nitrate, sulfate, and chemical oxygen demand (COD) according to the standard APHA protocols [31] (Supplementary Table S2).

Subsurface sediments from marine and lake environments exhibited notable differences in chemical composition, particularly the level of key nutrients. Marine sediments from the Arabian and Bay of Bengal Seas have salinity levels between 35 and 36 g/L [32, 33]; consequently, enrichment for marine sediments was performed at 35 g/L salinity. For Lonar Lake, a wide range of salinity values has been reported, ranging from 5 to 24 g/L [34, 35]. Based on previously reported studies, we chose 20 g/L salinity for enrichment purposes with Lonar Lake sediments. The sampling of the Sambhar Lake was conducted during the winter season. While Sambhar Lake’s salinity can range from ~ 9 g/L post monsoon to as high as 400 g/L during summer [36–38], we selected an enrichment salinity of 35 g/L as it was close to the analyzed salinity data of the samples (~ 40 g/L), and it corresponds to marine-like salinity conditions. It was chosen to selectively favor the growth of both halotolerant and moderately halophilic microorganisms present in the sample while avoiding excessive osmotic stress that could interfere or inhibit initial enrichment. To maintain consistency and comparability with the marine sediments, enrichment from Sambhar Lake sediments was conducted at 35 g/L. The pH was maintained at 7 in all enrichment media. The microbial inoculum was prepared by mixing sediment with anaerobic saline buffer solution, followed by centrifugation at 3000 rpm. The large sediment particles were then allowed to settle, and the supernatant was used as the microbial inoculum source for the enrichment experiments under anoxic conditions.

Enrichment Medium Composition and Preparation

A modified DSM 135a medium with 3.5% salinity was used for marine and Sambhar Lake sediments. The medium consists of NaCl, 35 g/L; NH₄Cl, 1 g/L; KH₂PO₄, 0.33 g/L; K₂HPO₄, 0.45 g/L; MgSO₄.7H₂O, 0.10 g/L; NaHCO₃, 5 g/L; L-cysteine, 0.50 g/L; vitamins, 10 mL/L; trace elements, 20 mL/L. Vitamin and trace element solutions were prepared as per DSM 135a medium composition. For Lonar Lake sediment, a modified M9 medium with 2% salinity was used. It consists of Na₂HPO₄, 4.33 g/L; NaH₂PO₄, 2.69 g/L; NH₄Cl, 0.30 g/L; KCl, 2.69 g/L; NaHCO₃, 5 g/L; L-cysteine, 0.50 g/L; vitamins, 10 mL/L; trace elements, 20 mL/L. In all cases, 3 g/L sodium 2-bromoethanesulfonate was added to inhibit the growth of methanogens, 0.5 mL/L resazurin from a 0.1% stock solution was used as a redox indicator, 0.15 g/L Na₂S.9H₂O was added to the medium as a reducing agent, and pH was adjusted to 7. For medium preparation, all ingredients except bicarbonate, vitamins, trace metals, L-cysteine, and sulfide were dissolved in distilled water. The mixture was boiled and then cooled to room temperature under an 80% N₂ and 20% CO₂ gas mixture. It was then transferred to serum bottles and subjected to gassing-degassing cycles under N₂:CO₂ (80:20) conditions. Serum bottles were sealed with butyl rubber stoppers and crimp seals and were then autoclaved. Vitamins, trace elements, reducing agents (L-cysteine and Na₂S.9H₂O), and NaHCO₃ from anaerobic stock solutions were added to the medium through sterile filters (0.22 μm, Cole-Parmer) just before microbial inoculation.

Enrichment Experiments

All enrichment experiments were performed in 100 mL serum bottles with a working or liquid medium volume of 40 mL, including an inoculum size of 10% v/v. The specific nutritional and growth conditions were applied as a selective pressure to enrich chemolithoautotrophs from marine and lake sediments. After inoculation, the headspace of the bottles was filled with H₂:CO₂ (80:20) at a maximum pressure of 1.5 bar and incubated at 30 °C, under 30 rpm shaking conditions (New Brunswick™ Innova 42, Eppendorf). A fed-batch enrichment technique was used, where 10% of the spent medium after each batch cycle served as the inoculum for the next cycle with fresh medium. In the inoculated serum bottles, an increase in OD₆₀₀ was observed in initial enrichment experiments (Supplementary Figure S1). The bulk phase samples from these enrichment bottles were used for further sub-culturing and detailed analyses. Two control experiments, one with inoculum but no carbon (CO₂) and energy (H₂) sources (please refer to the next section) and another an abiotic control without inoculum but with CO₂ and H₂, were also performed under the same conditions.

Control Experiment with only L-cysteine as a Sole Carbon and Energy Source

To investigate the ability of enriched cultures to utilize L-cysteine as the sole carbon and energy source, batch culture experiments were carried out in 100 ml serum bottles with a working volume of 40 mL (inoculum size of 10% v/v). The experiment was performed under strictly anaerobic conditions using the same basal medium composition described previously for each respective culture, with the following modification: no bicarbonate/CO₂ and H₂ was added; instead, L-cysteine at a concentration of 0.5 g/L was supplied as the sole carbon and energy source. Serum bottle headspace was filled with N₂ gas at a pressure of 1.5 bar, and the inoculated bottles were incubated at 30 °C.

Analytical Methods

The liquid samples from the serum bottles were monitored daily for pH, OD₆₀₀, and metabolites (i.e., C1-C4 carbon chain length organics and alcohols). Short-chain fatty acids and alcohols were analyzed using High-Pressure Liquid Chromatography (HPLC, Agilent Infinity 1260 equipped with an RI detector and Agilent Hi-Plex H C-18 column). 5 mM H₂SO₄ was used as the mobile phase with a flow rate of 0.5 mL/min at 50 °C. The calibration range for HPLC was from 10 to 4000 mg/L (detection limit 1 mg/L). Formic acid was also analyzed and confirmed using an enzymatic and colourimetric Formate Assay Kit according to the manufacturer’s instructions (MAK059, Sigma Aldrich). The gas samples from the headspace of the serum bottles were analyzed for H₂, CO₂, O₂, CH₄, and H₂S using GC-TCD (Agilent 490 Micro GC). GC was equipped with three channels for different gases (Channel 1: Column-Molecular sieve for H₂ carrier gas-Ar; Channel 2: Column-Molecular sieve for O₂, N₂, CO, and CH₄, carrier gas-He; Channel 3: Column-Pora plot U for CO₂ and H₂S, carrier gas-He). The column oven temperature for each channel was set at 80 °C, and the run time was two minutes for all channels. The calibration range for GC was from 2 to 300 ppm (detection limit 2 ppm). Microbial growth was monitored by measuring the optical density (OD) at 600 nm using a UV-VIS spectrophotometer (Photolab UV-Vis 7600, Xylem) and the biomass protein content using the Bradford Assay with Bovine Serum Albumin (BSA) standard. 1 mL of sample was sonicated at 30% power with 10 s pulse and 10 s gap cycles for 5 min using a PreCiSonic PKS-250 probe sonicator (PCI Anaalytics Pvt. Ltd., India). 160 µL of each sample was pipetted into microtiter plate wells, followed by the addition of 40 µL of Bradford reagent (Bio-Rad Laboratories, USA), following the manufacturer’s protocol. The sample and dye were mixed thoroughly and incubated at room temperature for 5 min. Absorbance was measured at 595 nm, and protein concentration was calculated using a BSA-based standard curve. The pH was measured using a pH meter (Oakton PC2700). The data is reported as averages and standard deviations based on at least triplicate experiments. Statistical significance among the four enrichment cultures (Shivrajpur, Puri, Lonar and Sambhar) was evaluated using one-way Analysis of variance (ANOVA) in Microsoft Excel. Differences with P < 0.05 were considered statistically significant. ANOVA was applied to compare organic acid production and protein estimation data among the enriched cultures.

Microscopy of the Enriched Microbial Cultures

The morphology and growth pattern of the microbes in the enriched cultures were observed using field emission scanning electron microscopy (FESEM) and confocal microscopy. For FESEM, the cell suspension was filtered through a 0.22 μm polycarbonate membrane (Isopore TM, Merck Millipore Ltd.) using a 13 mm Swinny Syringe Filter holder (Merck Millipore). Cells trapped on filters were fixed using a 2% glutaraldehyde and paraformaldehyde solution and incubated at 4 °C for 12 h. The samples were dehydrated stepwise in a graded ethanol series (i.e., 20, 40, 60, 80, 100%) and dried overnight in the desiccator. The fixed samples were gold-coated using a JEOL JEC-1600 Auto-Fine Coater (JEOL Ltd., Japan) at 20 mA for 35 s and observed under the microscope (FESEM Hitachi, SU8010 Japan). For confocal microscopy, bacterial cells were treated with LIVE/DEAD BacLight bacterial viability stains as per the manufacturer’s protocol (Thermo Fischer Scientific) and observed under the upright confocal microscope (Leica SP8, Germany). The excitation wavelength was 488 nm, and the emission wavelength was 512 nm.

Metagenome Sequencing-based Microbial Community Analysis of the Enriched Cultures

Genomic DNA was extracted from all the enriched cultures using the DNeasy PowerSoil Pro Kit (Qiagen), following the manufacturer’s protocol. The purity and concentration of the DNA samples were determined using NanoDrop (Thermo Scientific™ NanoDrop 2000). The Illumina Novaseq platform was used for sequencing (150 cycles at Genotypic Technology Pvt. Ltd., Bengaluru, India). SqueezeMeta version 1.6.5 [39], a complete automatic pipeline for metagenomics, was used to analyze the reads from the four enriched samples. Briefly, this pipeline uses Trimmomatic [40] for adapter removal, trimming, and filtering of reads; SPAdes [41] for the assembly in co-assembly mode (recommended), Prodigal [42] for gene prediction, barrnap [43] for rRNA prediction, and RDP classifier [44] for classifying the resulting 16 S rRNA sequences. SqueezeMeta performed taxonomic assignment by Diamond search [45] of gene sequences against the GenBank nr database and implemented a fast LCA algorithm for finding the last common ancestor of the hits for each query gene. Further, individual genes were used for consensus taxonomic assignments of the contigs. Diamond was used in the pipeline for comparing genes against the KEGG database [46] for KEGG ID annotation. For estimating the coverage and abundance of genes and contigs, it uses Bowtie2 [47] for mapping reads back onto the contigs, followed by Bedtools [48] for extracting the raw number of reads and mapping bases to each gene and contig. The binning was carried out using CONCOCT [49] and MetaBAT2 [50], and further, the DAS tool [51] was used to merge the multiple binning results into a single set. The completeness of the bins was determined using CheckM software [52]. The downstream analysis of the SqueezeMeta output was performed using the SQMtools R package [53]. The raw metagenomics datasets are deposited and publicly accessible at NCBI Sequence Read Archive (SRA) under Bioproject PRJNA1138324 with accession numbers SRR29933917, SRR29933918, SRR29933919, and SRR29933920.

16 S rRNA Amplicon sequencing-based Microbial Community Analysis of Marine and Lake sub-surface Sediments

Genomic DNA was extracted from all the original sediment samples used as inoculum sources using the DNeasy PowerSoil Pro Kit (Qiagen), following the manufacturer’s protocol. The purity and concentration of the DNA samples were determined using NanoDrop (Thermo Scientific™ NanoDrop 2000). The Illumina MiSeq platform was used for sequencing (300 cycles at Eurofins Genomics Pvt. Ltd., Bengaluru, India). Raw reads were processed using the fastp tool for removing adapters and low-quality bases [54], and high-quality reads were classified with Kraken2 using the Standard Kraken2 database [55]. The 16 S rRNA raw sequence datasets are deposited and publicly accessible at NCBI Sequence Read Archive (SRA) under Bioproject PRJNA1138324 with accession numbers SRR29927254, SRR29927255, SRR29927256, and SRR29927257.

Results

Subsurface sediments from marine and lake environments exhibited notable differences in chemical composition, particularly the level of key nutrients (Supplementary Table S2). A considerable COD level in all cases suggests the presence of organic matter to support the growth of different microbial groups. Similarly, the presence of ammonia, nitrate, sulfate, and phosphate suggests the availability of key nutrients and electron acceptors to support the growth and metabolic activities of microorganisms.

Enrichment of the CO₂-fixing Chemolithotrophic Microorganisms

The increase in OD₆₀₀ and organic acid production in the enriched cultures suggested the growth of halophilic chemolithoautotrophic microorganisms with all four inoculum sources (Fig. 1). The organic analysis of the bulk phase samples revealed that CO₂ was fixed mainly into acetic and formic acids by the microbial communities. For instance, the enriched culture from Shivrajpur sediments (marine) with bicarbonate in a liquid medium and H₂:CO₂ in the headspace produced up to 569 ± 19 mg/L acetic acid and 355 ± 7 mg/L formic acid in the third batch cycle (Fig. 1A). Similarly, the enriched culture from the Puri (marine) sediments produced up to 966 ± 24 mg/L acetic acid in the second batch cycle and 367 ± 30 mg/L formic acid in the third batch cycle (Fig. 1B). The enriched cultures from the Lonar Lake sediments also produced 573 ± 29 mg/L acetic acid and 370 ± 21 mg/L formic acid in the first batch cycle (Fig. 1C). In contrast, the enriched culture from the Sambhar Lake sediment exhibited lower acetic acid production (92 ± 20 mg/L) but higher formic acid (391 ± 8 mg/L) in the second batch cycle (Fig. 1D).

Fig. 1.

Microbial growth (in terms of OD₆₀₀) and organic acid production profiles of the enriched microbial cultures (n = 3) from (A) Shivrajpur (marine), (B) Puri (marine), (C) Lonar lake, and (D) Sambhar lake sub-surface sediments with bicarbonate and H₂:CO₂ (80:20). The cycles indicate the data for three batch cycles. Note: The four enriched cultures from Shivrajpur, Puri, Lonar, and Sambhar are significantly different in terms of acetic acid (P* = 1.26E-11 < 0.05), while the formic acid concentration was statistically insignificant

The formate concentrations detected using the Formate assay kit closely corresponded with the values obtained by HPLC analysis, thereby confirming the production of formate by the enriched cultures (Supplementary Table S3). In all cases, organic acid production is correlated with growth, as confirmed by estimating protein content as a proxy for biomass growth (Supplementary Figure S2). The data highlight the differences in protein concentrations and variability among the four samples across the three cycles. The enriched culture from Puri sediment exhibited a relatively higher average protein concentration of 62 ± 13 mg/L compared to the enriched cultures from Lonar and Shivrajpur, which had average protein concentrations of 49 ± 11 mg/L and 45 ± 24 mg/L, respectively. This corresponds with the organic acid production pattern of these cultures. The lowest average protein concentration of 29 ± 5 mg/L was observed for the enriched culture from Sambhar Lake, which aligns with the lowest organic acid production by this culture. Additionally, no considerable change in pH was observed, suggesting that the used media had sufficient buffering capacity. Neither organic acid production nor any increase in OD was detected in the controls (Supplementary Figure S3), suggesting that organic acid production and an increase in OD were due to the metabolic activity of the enriched microbial communities. No methane was detected in the headspace of enrichment culture bottles, confirming the inhibition of methanogenesis by the addition of sodium 2-bromoethanesulfonate. This suggests that microbes other than methanogens in the enriched cultures utilized H₂ as an electron source and CO₂ as the carbon source in the enrichment cultures. The production of organic acids by enriched microbial cultures from both marine and lake sediments, using CO₂ and H_2, suggests their chemolithoautotrophic nature.

Growth and Organic Acid Production by Enrichment Cultures with L-cysteine as the Sole Carbon and Energy Source

L-cysteine is commonly used as a reducing agent in anaerobic cultivation to scavenge oxygen and lower the redox potential of the media, facilitating the growth of obligate anaerobes [56]. However, L-cysteine can also be actively metabolized by bacteria, as it can also act as both a carbon and an energy source [57]. The metabolism of L-cysteine involves desulfuration of L-cysteine to pyruvate (catalyzed by L-cysteine desulfurases), which subsequently leads to the production of acetate [57].

To assess the impact on microbial growth and organic acid production, a control experiment was conducted where L-cysteine (0.5 g/L) served as the sole carbon and energy source, with N₂ in the headspace to maintain an anaerobic environment. Three enriched cultures from Shivrajpur, Puri, and Lonar showed low acetic acid accumulation over time, reaching peak concentrations between days 2 and 4, followed by a gradual decline (Fig. 2). The Shivrajpur culture produced up to 36 ± 7 mg/L of acetic acid, while the Puri culture reached 42 ± 3 mg/L. The Lonar culture achieved the highest acetic acid production, reaching 55 ± 8 mg/L, while the enriched culture from Sambhar did not produce any detectable organic acids. The production of relatively very low acetic acid and no formic acid compared to CO₂:H₂ condition by the enrichment cultures confirms that CO₂ and H₂ served as primary substrates for chemolithotrophic organic acid production in the experimental conditions tested here. While the enriched microbial communities can metabolize L-cysteine, it does not contribute significantly to the production of organic acid. Further validation was achieved through a protein estimation experiment performed towards the end of the experiment (Day 7), when organic acid production had declined or plateaued (Supplementary Figure S4). These data confirmed biomass accumulation across the cultures, reinforcing that microbial metabolism was responsible for L-cysteine utilization. Interestingly, the Sambhar culture exhibited growth/protein content with L-cysteine, but it did not produce detectable organic acids, suggesting that L-cysteine was utilized for biomass synthesis but not for organic acid production. These observations confirm that CO₂ and H₂ served as primary substrates for chemolithotrophic organic acid production, while L-cysteine can be metabolized, but with limited organic acid accumulation by these enrichment cultures.

Fig. 2.

Organic acid production profiles (n = 3) of the enriched microbial cultures from (A) Shivrajpur (marine), (B) Puri (marine), and (C) Lonar lake subsurface sediments with L-cysteine as the sole carbon and energy source

Microscopic Characterization of the Enriched Halophilic Chemolithoautotrophic Cultures

The enriched cultures from both the marine and lake sediments showed an aggregate-type growth pattern in the suspension and at the bottom surface of the serum bottles (Fig. 3A, B, C & D). Even under shaking conditions (30 rpm) during incubation, the enriched cultures formed aggregates in the serum bottles. Various microscopic techniques were employed to investigate the aggregate morphology and cell distribution within aggregates (Fig. 3). Confocal microscopy with the nucleic acid stain STYO™ 9 confirmed the growth of microbes and revealed long filament-like structures along with oval and rod-shaped cells (Fig. 3E, F, G, H). Since SYTO 9 is a DNA-binding dye, confocal microscopy confirmed that the thread-like structures contain microbial cells both inside and along their length. Further analysis of aggregates through FESEM revealed long, thread-like structures ranging from 20 μm to 500 μm in length. At 5 μm resolution, the thread-like structures appear tube-like with various rod and oval-shaped cells inside them (Fig. 3I, J, K, L).

Fig. 3.

Serum bottles showing aggregate formation in the enriched cultures from (A) Shivrajpur (marine), (B) Puri (marine), (C) Lonar Lake, and (D) Sambhar Lake subsurface sediment samples. Confocal Laser Scanning Microscopy (CLSM) images (E, F, G, H) show long, thread-like structures and cells, and scanning electron microscopy (SEM) images (I, J, K, L) show oval and rod-shaped cells inside the thread-like structures (scale bar: 2 μm, 10 μm, 5 μm) of the respective enriched cultures.

Bacterial filamentation can occur in response to starvation [58, 59] salt, pH, and temperature shocks [60–63], ionizing radiations [64, 65] or exposure to the host immune response during infection [66, 67]. In this study, aggregate formation is likely influenced by high salinity levels in the enrichment media. It is well known that, along with salinity, nutritional limiting conditions can also stimulate the formation of cell aggregates as a stress response for the survival mechanism [68–70]. Under predominantly autotrophic conditions, the close spatial proximity of microorganisms within aggregates suggests a physiological interaction that facilitates efficient metabolic exchanges. The microbial communities might have syntrophic interactions, as the establishment of cell-to-cell contact resulting in decreased microbial distance, which, according to Fick’s law, should facilitate increased metabolite exchanges (e.g., Hydrogen fluxes) between species [71], and it can ultimately enhance microbial growth. In natural environments, bacteria exhibit distinct behavioural states, such as living in close proximity to each other within surface-attached biofilms, planktonic cells, or aggregates. The aggregate formation is widely observed in many environmental and pathogenic microorganisms [72]. However, the functionality of such behaviour, which likely represents adaptations to various environmental conditions, remains poorly understood and unclear and thus warrants further detailed investigations.

Microbial Community Composition in the Enriched Halophilic Chemolithoautotrophic Cultures

A stacked bar chart showing the relative abundance of different bacterial species across four enriched cultures from Shivrajpur, Lonar, Puri, and Sambhar is presented in Fig. 4. At the genus level, all four enriched cultures exhibit a clear dominance of genus Vibrio, with relative sequence abundances of 96% in Shivrajpur, 97% in Puri, 94% in Lonar, and 91% in Sambhar cultures. This overwhelming prevalence suggests that the selective conditions of the enrichment process strongly favoured Vibrio, likely due to its metabolic adaptability and resilience in saline environments [73]. The second most dominant group was the unclassified Gammaproteobacteria, with 0.9% abundance in the enriched culture from Shivrajpur and 0.06% in Lonar. The genus Halodesulfovibrio was found to be the second most dominant group, with an abundance of 0.43% in the enriched culture from Puri. In contrast, in the enriched culture from Sambhar Lake sediment, the genus Oceanidesulfovibrio emerged as the second most dominant bacterial group, with an abundance of 5%. Although this genus was also detected in the other three enriched cultures, its presence was significantly lower, with an abundance of less than 0.003%. There were other bacterial taxa also present across all cultures; however, their abundances remained low, typically less than 0.4%. At the species level, unclassified Vibrio was found to be the dominant group, accounting for 96% of the enriched culture from Shivrajpur, 94% from Lonar Lake, and 90% from Sambhar. The only exception was the Puri culture, where V. fluvalis was dominant, comprising 86% of the population. The relative abundance of unclassified Vibrio (only ~ 10%) in Puri sets it apart from the other three samples. V. fluvalis was also found to be present in other samples. For instance, in Shivrajpur, V. fluvalis showed 0.28% abundance; in Lonar, it accounted for 0.25%; and in Sambhar, it was slightly higher at 0.41%. In the Shivrajpur culture, V. parahaemolyticus was the second most abundant species at 0.21%. Oceanidesulfovibrio indonesiensis was the second most abundant species in the Sambhar sample, with 5% abundance. In contrast, its abundance in the other enrichment cultures remained very low, at less than 0.002%. Other bacterial species were also detected, but occurred at low abundance (< 0.4%). In all the enriched cultures, the relative sequence abundance of canonical CO₂-fixing acetogens, such as Acetobacterium spp., Clostridium spp., and Moorella thermoacetica [74] was found to be low or absent. Though unclassified Clostridia were present in all samples, their abundance was very low (< 0.001%).

Fig. 4.

Stacked bar charts showing microbial communities of the enriched cultures with CO₂ and H₂ substrates from sediments of Shivrajpur, Puri, Lonar, and Sambhar samples, categorized at the (A) genus and (B) species levels

16 S rRNA sequencing of the sediment inoculum sources, which were used to enrich the mixed cultures, confirmed the presence of Vibrio in the original sediments, albeit at low abundances (Supplementary Figure S5). In the Shivrajpur sediment, it accounted for 1.62% of the microbial community; in Lonar, it was only 0.09%; in Puri, the abundance was 0.22%; and in Sambhar, it was present at an abundance of 1.52%. Despite these initially low abundances, the chemolithotrophic enrichment conditions with CO₂ as a carbon source and H₂ as an electron source exhibited a significant increase in the Vibrio population above 90% across all cultures. The drastic increase in Vibrio abundance across all enriched cultures indicates that the enrichment conditions strongly favoured its proliferation, resulting in a marked shift in community composition and dominance of Vibrio species.

Carbon Fixation Pathways: Genes Encoding Enzymes for CO₂ Fixation

To explore microbial metabolic strategies employed by the enriched cultures, the distribution of marker genes linked to the following carbon fixation pathways was investigated. The genes related to seven carbon fixation pathways, namely Wood-Ljungdahl pathway (WLP), reductive citric acid cycle (rTCA), 3-hydroxypropionate bicycle (3-HP), hydroxypropionate-hydroxybutyrate cycle (3HP-4HB), dicarboxylate-hydroxybutyrate (DC-4HB), Calvin-Benson-Bassham cycle (CBB), and the reductive glycine pathway (rGly), were checked as elaborated below.

The heatmap for the WLP is shown in Fig. 5. The presence of marker genes was checked for all four enriched samples. Not all genes of the WLP were detected; however, there was consistent detection of several genes involved in the methyl branch of the pathway, particularly those associated with folate-mediated one-carbon transformations. Notably, metF (methylene-tetrahydrofolate reductase), folD (methylene-tetrahydrofolate dehydrogenase), and fhs (formate tetrahydrofolate ligase) were detected in all four samples. These genes are essential for converting formate to methylene-THF and subsequently to methyl-THF, a key intermediate in acetyl-CoA synthesis. In contrast, the carbonyl branch of the WLP was sparsely represented. Key genes such as codh/acs (acetyl-CoA decarbonylase/synthase) were detected in very low abundance in the enriched culture from Sambhar only. The genes for the alpha and beta subunits of fdh (Formate dehydrogenase) showed low and variable abundance across the enriched cultures (Fig. 5). The alpha subunit was detected only in Puri and Sambhar, whereas the beta subunit was present in both Shivrajpur and Sambhar, but absent in Lonar and Puri. This patchy distribution and low-level abundance suggest that while formate dehydrogenase activity may be present in some samples, the canonical WLP is unlikely to be fully functional in the enriched cultures. The cooS, acsA (carbon monoxide dehydrogenase) gene was found only in the enriched cultures from Puri and Sambhar Lake. In contrast, both genes were completely absent in the enriched cultures from Shivrajpur and Lonar. Furthermore, acetyl-CoA synthase, 5-methyltetrahydrofolate corrinoid/iron-sulfur protein methyltransferase, and homocysteine S-methyltransferase were also identified in low abundance in Sambhar only. Additionally, the genes for the phosphate trans-acetylase-acetate kinase pathway, which enables the conversion of acetyl-CoA to acetate which is a critical energy conservation step in WLP, were found to be present in all four cultures.

Fig. 5.

Heatmap showing the presence of the predicted Wood-Ljungdahl pathway genes across four different enrichment cultures labeled as Shivrajpur, Puri, Lonar, and Sambhar in Transcripts per Million (TPM) (Threshold > 0)

The heatmap representing the genes for the rTCA cycle is shown in Fig. 6. The marker genes of the following enzymes were predicted across all the enriched cultures: isocitrate dehydrogenase, succinyl CoA synthetase, succinate dehydrogenase, malate dehydrogenase, phosphoenol pyruvate carboxylase, pyruvate orthophosphate dikinase, pyruvate water dikinase, fumarate hydratase (Class I, Class II), 2-oxoglutarate oxidoreductase (subunits alpha, beta), aconitate hydratase 2/2-methylisocitrate dehydratase, fumarate reductase (subunits flavoprotein, iron-sulfur, C and D) and succinate dehydrogenase/fumarate reductase (subunits iron-sulfur, flavoprotein and membrane anchor). Additionally, other rTCA genes, such as aconitate hydratase and pyruvate carboxylase, were present in the enriched culture from Puri and Sambhar, and 2-oxoglutarate ferredoxin oxidoreductase (subunits gamma and delta) and fumarate hydratase (subunits alpha and beta) were present in the enriched cultures from Puri and Sambhar, whereas, citryl-CoA synthetase large subunit and pyruvate carboxylase subunit B were detected at low levels in the enriched culture from Sambhar. Some rTCA genes showed a patchy distribution, being completely absent in certain samples while present at low abundance in others, suggesting functional variability and potential modularity of the pathway across microbial communities. However, the marker genes of the key enzymes of the rTCA cycle, citrate lyase, which is crucial for the ATP-dependent cleavage of citrate to acetyl-CoA and oxaloacetate, were not detected.

Fig. 6.

Heatmap showing the presence of the predicted reductive TCA cycle genes across four different enrichment cultures labeled as Shivrajpur, Puri, Lonar, and Sambhar in Transcripts per Million (TPM) (Threshold > 0)

The heatmap representing the 3-hydroxypropionate (3HP) bicycle is presented in Fig. 7A. The 3 HP bicycle is an autotrophic fixation pathway used primarily by certain bacteria, such as Chloroflexus aurantiacus, a facultative autotroph found in microbial mats. The pathway operates in two interconnected cycles: the first cycle fixes CO₂ into 3-hydroxypropionate via acetyl-CoA carboxylation and reduction, while the second cycle incorporates glyoxylate into central carbon metabolism, ultimately forming pyruvate [75, 76]. In the enriched cultures, genes for the enzyme acetyl-CoA carboxylase (subunits alpha, beta, biotin carboxyl carrier protein, biotin carboxylase), succinate dehydrogenase/fumarate reductase (subunits cytochrome b subunit, membrane anchor, iron-sulfur & flavoprotein), and fumarate hydratase were detected in all samples. Additionally, fumarate hydratase (subunits alpha and beta) and methylmalonyl-CoA mutase were identified in Puri and Sambhar enriched cultures, whereas methylmalonyl-CoA/ethylmalonyl-CoA epimerase was identified exclusively in the enriched culture from Sambhar only. However, the other crucial enzymes required for the complete functioning of the 3 HP bi-cycle, such as malonyl-CoA reductase, propionyl-CoA synthase, and (S)-malyl-CoA lyase, were not detected. The absence of these enzymes suggests either an incomplete pathway or the possibility of alternative carbon assimilation strategies in the studied microbial communities, or limitations in detection due to a limited number of reads [75, 76].

Fig. 7.

Heatmap showing the presence of the predicted (A) 3-hydroxypropionate (3HP) bicycle genes and (B) Dicarboxylate-4-Hydroxybutyrate (DC-4HB) genes across four different enrichment cultures labeled as Shivrajpur, Puri, Lonar, and Sambhar in Transcripts per Million (TPM) (Threshold > 0)

In addition, genes of the Hydroxypropionate-Hydroxybutyrate (3-HP-4HB) pathway (Supplementary Figure S6) were also found. The gene for acetyl-CoA C-acetyltransferase, which facilitates the transfer of coenzyme A (CoA) from one molecule to another, was identified in all the enriched cultures. However, additional key enzymes, including enoyl-CoA hydratase/3-hydroxyacyl-CoA dehydrogenase and methylmalonyl-CoA mutase, were found to be present in the enriched culture from Puri and Sambhar at a very low abundance. Furthermore, two key enzymes, 4-hydroxybutyryl-CoA dehydratase and methylmalonyl-CoA epimerase, were detected at a very low abundance in the enriched culture from Sambhar. Notably, the key enzyme succinyl-CoA reductase, which is involved in the conversion of succinyl-CoA into 4-hydroxybutyrate, was not identified in all the cultures. This incomplete pathway suggests that the 3-HP-4HB pathway may not be fully functional in these microbial communities.

Genes for Dicarboxylate-4-Hydroxybutyrate (DC-4HB) were also identified in all the enriched cultures (Fig. 7B). This pathway shares some similarities with the 3HP-4HB cycle but uses different enzymes for carbon fixation and conversion of intermediates. The data revealed the presence of several genes across all cultures. Genes such as succinate dehydrogenase/fumarate reductase (subunits flavoprotein, iron-sulfur, cytochrome b and membrane anchor), malate dehydrogenase, acetyl-CoA C-acetyltransferase, succinyl-CoA synthetase (subunits alpha and beta), phosphoenolpyruvate carboxylase, pyruvate, water dikinase, fumarate reductase (subunits C, D, flavoprotein and iron-sulfur) and fumarate hydratase (class I, class II) were detected across all cultures. Furthermore, pyruvate-ferredoxin/flavodoxin oxidoreductase was detected in Puri, Lonar, and Sambhar enriched cultures, and fumarate hydratase (subunits alpha and beta) and enoyl-CoA hydratase/3-hydroxyacyl-CoA dehydrogenase were identified in Puri and Sambhar enriched cultures only. Finally, 4-hydroxybutyryl-CoA dehydratase was predicted to be present in low abundance in Sambhar only. However, key enzymes responsible for the conversion of 4-hydroxybutyrate to acetyl-CoA, such as 4-hydroxybutyryl-CoA dehydratase and crotonyl-CoA hydratase, were either absent or detected in low abundance across the samples. This suggests that the cycle may be only partially represented or functionally incomplete in the microbial communities.

Genes associated with the Calvin-Benson-Bassham (CBB) cycle (Supplementary Figure S7) were also found across all four enriched samples. The enzymes included ribulose-phosphate 3-epimerase, triosephosphate isomerase, fructose-1,6-bisphosphatase (I & II), phosphoribulokinase, ribose 5-phosphate isomerase A, phosphoglycerate kinase, fructose-bisphosphate aldolase class II, glyceraldehyde 3-phosphate dehydrogenase and transketolase. The enzyme fructose-bisphosphate aldolase class I was identified in Shivrajpur, Puri and Sambhar enriched cultures, whereas fructose-bisphosphate aldolase class I and ribose 5-phosphate isomerase B were detected in Puri and Sambhar enriched cultures. The fructose-1,6-bisphosphatase III was present only in the Sambhar-enriched culture. But the key enzyme, ribulose-1,5-bisphosphate carboxylase (RuBisCO), which facilitates the fixation of atmospheric CO₂ into organic carbon by catalysing the carboxylation of ribulose-1,5-bisphosphate (RuBP) using CO₂, to produce two molecules of 3-phosphoglycerate that enter the Calvin cycle for glucose biosynthesis [75], was absent. In addition to Rubisco, three other enzymes, sedoheptulose-1,7-bisphosphatase (SBPase), phosphoglycerate mutase (PGM), and sedoheptulose-7-phosphate synthase were also absent. SBPase plays a crucial role in the regeneration phase of the Calvin cycle by catalyzing the dephosphorylation of sedoheptulose-1,7-bisphosphate to sedoheptulose-7-phosphate, an essential step in RuBP regeneration. PGM facilitates the interconversion of 3-phosphoglycerate and 2-phosphoglycerate, linking Calvin cycle intermediates to glycolytic/gluconeogenic pathways. Sedoheptulose-7-phosphate synthase, although less commonly described, is involved in forming seven-carbon sugars necessary for the rearrangement and recycling steps within the Calvin cycle. The absence of these enzymes suggests an incomplete or non-functional Calvin cycle.

The presence of genes encoding enzymes involved in the reductive glycine pathway (rGly) across the four enriched cultures was also examined (Fig. 8; Supplementary Excel File 1). The heatmap indicates that several core genes of the rGly pathway were detected across all samples. The enzymes included formate dehydrogenase subunits (iron-sulfur, alpha, beta, and gamma), suggesting the potential for CO₂ conversion to formate, a key intermediate in the pathway. The other enzymes of the core module, such as formate tetrahydrofolate ligase (catalyzes the conversion of formate to 10-formyl-THF,) and methylene-THF dehydrogenase/methenyl-THF cyclohydrolase (involved in the conversion of 10-formyl-THF to 5,10-methylene-THF) were present. Furthermore, key enzymes of the glycine cleavage system (GCS), which form the second half of the core module of the rGly pathway and catalyze the conversion of methylene-THF, CO₂ and ammonia into glycine, were found to be present across all samples. The detected enzymes included glycine cleavage system H protein (GcvH), aminomethyltransferase (GcvT), dihydrolipoamide dehydrogenase, and ammonium transporter. In addition, the enzyme of the serine variant of this pathway, glycine hydroxymethyltransferase, was detected. This enzyme is involved in the conversion of glycine to serine. Some genes of another variant of this pathway, where there is direct conversion of glycine acetate and pyruvate, were detected. These genes included thioredoxin reductase, phosphate acetyltransferase, acetyl-CoA synthetase, acetate kinase, L-serine dehydratase, and formate transporter. However, the core gene encoding glycine reductase, which is essential for initiating this pathway, was absent. This suggests that although the downstream steps involved in the formation of acetate and acetyl-CoA metabolism may be active, the direct conversion of glycine to acetyl phosphate, which is another variant of rGly pathway is likely inactive and incomplete in these communities. The detection of these marker genes involved in formate assimilation and glycine conversion through the serine route suggests that the reductive glycine pathway might be active through the serine route in these microbial communities.

Fig. 8.

Heatmap showing the presence of the predicted genes reductive glycine pathway (rGly) genes across four different enrichment cultures labeled as Shivrajpur, Puri, Lonar, and Sambhar in Transcripts per Million (TPM) (Threshold > 0).

In addition to genes involved in CO₂ fixation, several genes related to L-cysteine metabolism were found across all four enriched cultures (Supplementary Figure S8). L-cysteine is a crucial sulfur-containing amino acid essential for maintaining redox balance, regulating energy metabolism, and facilitating biosynthetic pathways. The enzymes detected across four samples were acetate kinase, cysteine synthase, acetyl-CoA synthetase, methylenetetrahydrofolate dehydrogenase (NADP+)/methenyltetrahydrofolate cyclohydrolase, tryptophanase, methylenetetrahydrofolate reductase (NADPH), acylphosphatase, 2-oxoglutarate/2-oxoacid ferredoxin oxidoreductase (alpha and beta subunits), phosphate acetyltransferase, cysteine-S-conjugate beta-lyase, formate-tetrahydrofolate ligase, L-cysteine desulfidase and acetyl-CoA ligase (ADP-forming) (alpha and beta subunits). Additionally, pyruvate-ferredoxin/flavodoxin oxidoreductase was present in Puri, Lonar and Sambhar, and cysteine-S-conjugate beta-lyase, anaerobic carbon-monoxide dehydrogenase catalytic subunit, formate dehydrogenase (NADP+) alpha subunit and D-cysteine desulfhydrase were identified in Puri and Sambhar. The putative phosphotransacetylase was predicted in Puri, and acetyl-CoA decarbonylase/synthase complex (gamma and delta subunits), phosphate acetyltransferase, acetyl-CoA synthase, 5-methyltetrahydrofolate corrinoid/iron sulfur protein methyltransferase and homocysteine S-methyltransferase were found in Sambhar only. These findings suggest that L-cysteine may feed into central metabolism intermediates such as pyruvate and acetyl-CoA, linking sulfur amino acid metabolism with energy metabolism and biomass production. However, several other enzymes related to downstream sulfur metabolism or auxiliary reactions, such as thiosulfate sulfurtransferase and some components of glutathione biosynthesis, were either absent or detected at low abundance, particularly in the Sambhar culture. This suggests that the L-cysteine metabolic pathway may be only partially active, pointing to a fragmented or auxiliary role for L-cysteine metabolism in the enriched cultures. Across all seven carbon fixation pathways analyzed, several marker genes exhibited uneven distribution, being completely absent in some samples while present at low abundance in others. Only the serine variant of the reductive glycine pathway was found to be nearly complete across the four samples.

The binning analysis identified a total of eight MAGs. A summary of the recovered MAGs taxonomy, genome statistics, and quality metrics is shown in Supplementary Table S4. None of the MAGs contained a complete natural CO₂ fixation pathway; the genes associated with different CO₂-fixing routes were found to be fragmented across multiple MAGs. However, we identified the serine variant of the rGly pathway (Supplementary Excel File 2), which has been demonstrated in engineered microbial systems so far [76–78]. This is consistent with the analysis of contigs from the coassembly data.

4. Discussion

The enrichment and growth of microbial cultures from different sub-surface sediment inoculum sources under strictly anaerobic conditions with CO₂ and H₂ as the sole carbon and energy sources strongly support the presence of halophilic chemolithotrophic microorganisms in these habitats. The growth was evidenced by the increase in biomass and organic acid production (Fig. 1, Supplementary Table S3) across three independent batch cycles, collectively suggesting CO₂ fixation by all cultures. In anoxic sediments, methane-producing archaea (methanogens) and acetate-synthesizing microbes (acetogens) are reported to be involved in competitive metabolic interactions [79]. Acetate is a principal precursor for methane production for acetoclastic methanogens. Also, hydrogenotrophic methanogens are specialized in growth with H₂/CO₂ [80]. To remove methanogens, a methanogenesis inhibitor, sodium 2-bromoethanesulfonate, was added to the serum bottles. Consequently, no methane was detected in the headspace of enrichment culture bottles. Microbial community analysis also revealed the absence of methanogens, reinforcing that organic production and biomass growth could be attributed solely to CO₂-fixing chemolithotrophic bacteria. Furthermore, all four enriched cultures produced considerable levels of formic acid. However, in WLP, formate is transiently produced and later converted to acetic acid [81]. This pattern was observed in the enriched cultures from both marine sediments and Lonar Lake, where acetic acid production was higher than formic acid. In contrast, the enriched culture from Sambhar Lake exhibited the opposite trend, with more formic acid production than acetic acid. This variation in organic acid production can be attributed to the differences in the microbial communities present in the enriched cultures.

To further explore the metabolic flexibility of the enriched cultures, L-cysteine was examined as an alternative carbon and energy source. Although L-cysteine is traditionally used as a reducing agent in anaerobic media, the findings of this study suggests that it can be partially utilized as a substrate by the enriched communities. The production of acetic acid was observed in the cultures from Lonar, Shivrajpur, and Puri, with Lonar exhibiting the highest level (55 ± 8 mg/L). Notably, the Sambhar culture showed no detectable organic acid formation despite measurable protein accumulation (Supplementary Figure S4). These results suggest that L-cysteine metabolism is community-specific and may contribute to biomass growth rather than organic production in some cases. Despite biomass growth, no organic acid production in Sambhar culture suggests the possibility that L-cysteine played biosynthetic or redox-balancing roles rather than acting as a primary energy-yielding substrate. These results align with a recent study showing that L-cysteine can be desulfurated to pyruvate and subsequently converted to acetate through central carbon metabolism [57]. Together, these findings underline the metabolic versatility and adaptability of enriched halophilic microbial communities. Their ability to utilize CO₂ and H₂ chemolithotrophically, coupled with L-cysteine metabolism, reflects a high degree of flexibility that likely enables survival and productivity in nutrient-limited or fluctuating anoxic environments.

CO₂ fixation in these enriched cultures cannot be attributed to a single culture; instead, it might occur syntrophically under nutritionally challenging conditions. All enriched cultures from marine and lake sediments, irrespective of the difference in the microbial communities, showed the filamentous aggregate-type growth pattern. Such structural organization may indicate possible syntrophic or cooperative interactions within the enriched microbial cultures mediating biogeochemical cycles in anoxic environments. Close proximity between the microbial partners is advantageous, as intermicrobial distances influence the metabolism rates and specific growth rates, which results in the microbes forming compact aggregates [82, 83]. The aggregate-type growth pattern shown by the enriched mixed cultures from the marine and lake sediments is common for obligate syntrophic interactions often found in acetogenic and methanogenic communities. For example, in high-rate methanogenic bioreactors, anaerobic bacteria and methanogenic archaea form compact microbial granules that operate like a consortium rather than a set of microorganisms working independently [84, 85]. Bacterial aggregates and filaments are also reported in pelagic marine waters and sediments associated with hydrothermal environments that are rich in iron [86–89]. For instance, the aggregate formation is reported to be the result of chemical signalling between the filament-forming Acidithrix strain C25 and heterotrophic iron-reducing Acidiphilumstrain C61 [90]. Loosely aggregated Fe-mineralized filaments, such as twisted stalks and tubular sheaths, along with other morphologies, are reported to be inside the microbial mats found in such environments [91–93]. Microbial aggregates appear in sprout-like formations in subterranean environments, including karst caves. Detailed microscopy showed that these formations are highly organized, with a core that contains various cytoplasmic inclusions surrounded by a thick extracellular matrix, which helps maintain the stability and shape of aggregates [94]. Filament formation has also been reported in Vibrio spp. in conditions mimicking the marine habitat [95]. They can produce biofilms and filamentous architecture, which is an entangled mesh of cells that provides strength to the biofilm and makes it independent from the secretory compounds. Autoinducers mediating quorum sensing in marine Vibrio, which allow bacterial communities to sense small auto-secreting molecules in the environment and have also been reported to activate aggregation in V. cholerae [96]. Bacterial aggregates are ubiquitous; however, relatively few studies on coaggregation between aquatic microbes have been reported [97]. To the best of our knowledge, this type of growth pattern of chemolithoautotrophic microbes from saline habitats, observed in all enriched cultures in this study, have not been reported yet. Even though microscopic observations revealed the presence of bacterial aggregates, which may facilitate close physical associations between the different taxa. However, the existence of syntrophic interactions within these aggregates remains speculative. Understanding this behaviour, characteristics, and functional regulation of such a growth pattern will help improve the knowledge of unique microbial growth patterns in different environments and nutritional conditions. Investigating bacterial aggregates from natural environment sites can provide new perspectives and questions based on fundamental similarities or differences. However, the functionality of such behaviour, which in all likelihood represents adaptations to different environmental conditions, remains understudied and unclear. Further investigations are warranted to understand this unique microbial growth pattern.

Interestingly, in the enriched cultures from marine and Lonar Lake sediments, canonical acetogens known to produce organic acids through the WLP were not found. Although they were present in low abundance in the inoculum sources, under enrichment conditions with CO₂ and H₂ as the major carbon and energy sources, Clostridium and other known acetogens were not detected. The enrichment conditions favoured the overwhelming dominance of the genus Vibrio, which constituted over 90% of the bacterial population in all four cultures. This suggests that the selective conditions with CO₂ as the carbon source and H₂ as the electron donor strongly favoured the growth of Vibrio, likely due to its metabolic flexibility and resilience in saline and alkaline environments [98].

Vibrio is a ubiquitous group of metabolically flexible marine bacteria that play a major role in the biogeochemical cycling of carbon in oceans. Their ability to utilize many carbon, nitrogen, and phosphate substrates along with the ability to produce external enzymes such as chitinase and laminarinase, allows them access to abundant nutrients that are unavailable to other organisms [99]. Vibrios are facultative anaerobes and have not been reported for CO₂ fixation yet. They can survive in anaerobic zones and are highly adaptable to oxygen-depleted environments. A novel strain Vibrio sp. JC009, which was cultured anaerobically, has been reported, showing that the strict anaerobic properties of this new strain may have resulted from long-term adaptation to a permanently anoxic environment [100]. Dedicated isolation experiments under these selective growth conditions from these enriched cultures could provide more clarity on the microbes responsible for CO₂ fixation in the enriched cultures. The ability of Vibrio spp. to persist in both oxic and anoxic conditions further supports its ecological significance in marine carbon cycling.

Even though the enrichment conditions strongly favored Vibrio-dominated consortia under H₂:CO₂ conditions, the data do not demonstrate direct CO₂ fixation by Vibrio itself; rather, it indicates its potential syntrophic or cross-feeding interactions in this autotrophy-based metabolism. A previous study has shown that V. alginolyticus, isolated from deep-sea environments, can tolerate and recover from elevated CO₂ exposure, while exhibiting distinct morphological and membrane adaptations under high CO₂ and low-pH conditions [101]. In another study, V. natriegenshas been reported to exhibit natural tolerance and metabolic capacity towards formate. Its genome contains six fdh (formate dehydrogenase) genes, two THF cycle genes- ftl (formate-tetrahydrofolate ligase) and folD (bifunctional methylenetetrahydrofolate dehydrogenase/methenyltetrahydrofolate cyclohydrolase and one pfl (pyruvate formate lyase) gene, enabling both formate dissimilation and partial assimilation through tetrahydrofolate-linked one-carbon metabolism [102]. These findings collectively suggest that certain Vibrio lineages may possess intrinsic genetic and physiological traits that support persistence and potential metabolic interactions under CO₂-rich conditions. Although the current dataset indicates the potential for CO₂ fixation within the enriched mixed community, direct validation for autotrophic metabolism by any particular microbial group will require further experimentation with targeted approaches such as ¹³CO₂ stable isotope probing, transcriptomic profiling, and physiological characterization of isolates.

The enriched culture from Sambhar lake sediments also exhibited a dominance of Vibrio, while the second most abundant bacterium was Oceanidesulfovibrio indonesiensis. Oceanidesulfovibrio is a strictly anaerobic bacterium capable of both respiratory and fermentative metabolism. Notably, it has the ability to grow as a chemolithoheterotroph, utilizing acetate both as an organic carbon source and an electron donor, along with H₂ as an additional electron donor [103]. Given these metabolic traits, it is highly likely that Oceanidesulfovibrio actively consumed acetate present in the medium, leading to a decrease in acetate concentration, resulting in lower acetate levels and higher formate concentrations compared to the other enriched cultures.

The detection of genes associated with multiple CO₂ fixation pathways highlights the metabolic versatility of the anaerobic microbial consortia enriched from the lake and marine sediments. The gene annotation data revealed diverse carbon fixation pathways across all enriched cultures. These included the Wood-Ljungdahl pathway (WLP), reductive tricarboxylic acid (rTCA) cycle, 3-hydroxypropionate (3-HP) bicycle, hydroxypropionate-hydroxybutyrate (3-HP-4HB) cycle, dicarboxylate-4-hydroxybutyrate (DC-4HB) cycle, Calvin-Benson-Bassham (CBB) cycle, and the reductive glycine (rGly) pathway. The differential representation and completeness of these pathways suggest the co-existence of multiple autotrophic strategies under nutrient limitation and saline conditions. While several marker genes associated with major autotrophic CO₂-fixing pathways were detected, key enzymatic components were missing in many cases, suggesting incomplete pathways or metabolic reliance on unknown alternative routes. However, the identification of pathway fragments in metagenomic data does not by itself demonstrate functional activity. Incomplete genome recovery, annotation gaps, or metabolic cross-feeding among taxa can yield partial pathway signals.

The WLP, also known as the reductive acetyl-CoA pathway, is among the most energy-efficient CO₂ fixation pathways. While partial gene sets were observed across all samples, several critical enzymes were either undetectable or present at low abundance. The presence of incomplete WLP genes has also been reported in an organohalide-respiring bacterium, Dehalococcoides mccartyi, where acetyl-CoA cleavage, rather than complete CO₂ reduction, supports one-carbon metabolism and methionine biosynthesis [104]. The main product of CO₂ reduction in the WLP pathway is acetate, with formate as an intermediate [105]. This pathway is most commonly utilized by the acetogens that can grow by the oxidation of organic or inorganic substrates coupled with the reduction of CO₂ to produce acetate. Notably, formate can also serve as a substrate for some acetogens [81, 105]. However, it is important to note that in the WLP pathway, formate primarily serves as an intermediate and is subsequently converted to acetate [106]. In this study, the enriched culture produced a higher formic acid concentration than reported in previous studies with mixed and pure CO₂-fixing cultures. However, the patchy distribution and variable low abundance of formate dehydrogenase alpha and beta subunits across the cultures suggest that the canonical enzyme complex may likely not be fully functional. Instead, putative fdh genes that may utilise alternate routes for formate production by using analogous subunits from other formate dehydrogenases were identified in all four enriched cultures (Supplementary Figure S9). A recent study on understanding the metabolic strategies of autotrophic communities in the anoxic region of Blue Hole also reported the presence of putative alternate genes for formate dehydrogenase [107]. Another study on chemolithoautotrophic phosphite-oxidising consortia detected an incomplete WLP pathway, lacking genes for formate dehydrogenase and several methyl-branch enzymes. Metagenome-guided analyses revealed a non-canonical variant of the pathway, supporting one-carbon reactions rather than complete CO₂ fixation [108]. These findings highlight that partial WLP pathway genes may support one-carbon transformations or intermediate recycling, rather than full autotrophic acetogenesis. In the present study, the partial representation of WLP pathway genes suggests that the enriched cultures may harbor noncanonical variants or incomplete segments of the WLP pathway rather than a fully functional route, highlighting the possibility of potential diversity within this ancient metabolic route.

Although gene sets related to the rTCA cycle were partially present across all samples, their distribution suggests that this pathway likely operates in a modular or truncated form. This aligns with the growing view that the rTCA cycle, while ancient in parts, is not universally conserved nor necessarily complete in all microorganisms. [109]. Notably, key citrate-cleaving enzymes such as citryl-CoA synthetase and citryl-CoA lyase are thought to have been acquired via horizontal gene transfer, rather than inherited from a common ancestor [109]. This suggests that the complete rTCA cycle is a derived metabolic adaptation rather than an ancestral trait. In this context, the partial detection of an incomplete gene set for the rTCA cycle across all cultures may reflect its ecological utility in anaerobic niches, where cross-feeding, energy constraints, or metabolic streamlining select for retention of only those segments that confer a functional advantage.

These results suggest the possibility that a modified CO₂ fixation pathway may be at work, utilizing noncanonical enzymes shared between the reductive TCA and WLP pathways. A recent study supports this possibility: high CO₂ concentration can thermodynamically drive parts of the TCA cycle in reverse, promoting reductive fluxes toward succinate and 2-oxoglutarate formation even in facultative heterotrophs [110]. In another study, kinetic simulations of deep-branching chemoautotrophs have shown that rTCA flux becomes thermodynamically favorable only under high CO₂ and strongly reducing conditions, whereas the WLP pathway predominates under lower CO₂ availability [111]. This environmental dependence of CO₂ concentration and electron donor availability provides a mechanistic rationale for the incomplete rTCA representation in metagenome data, supporting the view that CO₂ fixation may proceed via condition-dependent or distributed pathway segments across the enriched mix cultures.

The gene distribution for the 3-hydroxypropionate (3-HP) bicycle in the enriched cultures suggests a fragmented presence, with only some core metabolic enzymes consistently detected. The 3-HP bicycle occurrence is largely restricted to Chloroflexeaceae, particularly Chloroflexus aurantiacus [112]. This pathway’s dual-cycle architecture allows flexible routing of intermediates such as glyoxylate and pyruvate, which may benefit mixed trophic microbial communities. The partial detection of 3-HP genes in the present study likely reflects phylogenetic constraints or adaptation to anoxic conditions where isolated pathway modules may be integrated into other metabolic networks. This observation is consistent with a recent metagenomic dataset from a marine blue hole, where 3-HP and DC/4HB cycle genes were partially detected, suggesting that hybrid or non canonical CO₂ fixation route may operate under stratified anoxic environments [107].

The gene profile associated with the Dicarboxylate-4-Hydroxybutyrate (DC-4HB) cycle indicates a relatively broader distribution across the enriched cultures, particularly in relation to CO₂ fixation intermediates and energy metabolism. However, the absence or low abundance of key enzymes responsible for the conversion of 4-hydroxybutyrate to acetyl-CoA suggests that the complete cycle is likely not fully operational in most samples. This observation supports the notion that the DC-4HB cycle may function in a modular manner, and some metabolic flexibility or pathway variation may exist in these microbes. Originally described in anaerobic, thermophilic archaea like Ignicoccus hospitalis, this pathway is considered one of the most ATP-efficient CO₂ fixation routes, making its partial representation ecologically relevant in low-energy, oxygen-limited environments. Many archaea and bacteria are reported to harbour parts of the DC-4HB cycle, using them not for CO₂ fixation but to generate precursors like acetyl-CoA and succinate, or 4-hydroxybutyrate [113]. Its fragmented presence in the present study reflects a specialized role in providing biosynthetic precursors or supporting syntrophic interactions. The 3-HP-4HB pathway also showed fragmented representation across the enriched microbial communities. While one core metabolic enzyme, acetyl-CoAC-acetyltransferase, was detected across all cultures, the remaining pathway components were either absent or detected in low abundance. This suggests that the pathway is unlikely to be fully functional.

The CBB cycle was only partially represented across the enriched cultures, with key regulatory enzymes such as phosphoribulokinase either absent or detected at low abundances. The absence of RuBisco and other critical enzymes like Sedoheptulose-1,7-bisphosphatase (SBPase) and Phosphoglycerate mutase suggests that the pathway may not operate as a fully functional CO₂ fixation route in these communities. In the absence of light, the Calvin cycle is powered by the oxidation of H₂ or other reduced compounds, offering an alternative to light-driven autotrophy, particularly in oligotrophic and anoxic subsurface environments. Studies in chemolithotrophs such as Cupriavidus necator have shown that Calvin cycle segments can support anabolic functions and operate under non-canonical forms, particularly fueled by H₂ or formate [114]. Its partial presence in these cultures reflects the pathway’s metabolic plasticity, where segments may be retained to support biosynthesis or energy generation, especially in mixed-trophic or resource-limited environments [115]. Such modular operation has also been observed in sulfur-oxidizing symbionts of marine invertebrates, where the classical enzyme SBPase is replaced with PPi-dependent phosphofructokinase, forming an energy-conserving variant of the cycle [116]. These adaptations highlight the pathway’s flexibility in diverse non-phototrophic environments.

A nearly complete set of genes encoding the rGly pathway through the serine route was identified in all enriched cultures (Fig. 8). This suggests the functional activity of this pathway through the serine route. The key genes of the core module of the rGly pathway, including formate dehydrogenase subunits (iron-sulfur, alpha, beta, and gamma), formate-THF ligase, methylene-THF dehydrogenase/methenyl-THF cyclohydrolase and enzymes of the glycine cleavage system, were detected in all samples. Additionally, a key enzyme in the serine variant of the rGly pathway, serine/glycine hydroxymethyltransferase, also known as serine dehydratase/serine deaminase, was identified. These enzymes catalyze the conversion of glycine to serine, which can subsequently be deaminated to pyruvate. This suggests that the rGly pathway might be active through the serine route. The serine variant route of the rGly pathway has been reported in synthetic biology studies using engineered strains of Escherichia coli [78, 117, 118], Saccharomyces cerevisiae [77] and Cupriavidus necator [76]. In natural systems, a serine variant of this pathway has not been reported to date. However, in the present study with enriched mixed cultures under saline conditions, the serine variant of the pathway might have been active. The natural variant of the rGly pathway has been demonstrated in the chemolithotrophic sulfate-reducing bacterium Desulfovibrio desulfuricans, where the pathway operates efficiently using CO₂ as a carbon source, hydrogen as an electron donor, and sulfate as an electron acceptor under anaerobic conditions [119]. However, to date, rGly pathway has not been shown to be active under fermentative conditions where CO₂ acts as both a carbon source and an electron acceptor [120]. In the present study under H₂:CO₂ conditions, the rGly pathway emerged as the most complete and consistently detected pathway across the samples. Its prevalence reinforces the ecological relevance of the serine variant of rGly pathway in anoxic, hydrogenotrophic environments. Combined with experimental evidence of acetate and formate production, these findings suggest that rGly pathway may play a role as an additional carbon-fixing strategy in microbial communities under anoxic environments, alongside WLP. The identification of rGly pathway as a possible route in the enrichments is consistent with the recent findings highlighting its role in syntrophy-mediated CO₂ fixation. A recent study demonstrated that the rGly pathway is among the most transcriptionally active CO₂ fixation routes in low-complexity anaerobic microbiota, where syntrophic cooperation between bacteria and archaea drives overall carbon conversion. Their work showed that the rGly pathway, which shares partial overlap with the WLP, can mediate formate-based cross-feeding and operate bidirectionally under energy-limited conditions. This has led to growing recognition that the rGly pathway can functionally complement the WLP in syntrophic associations, particularly in systems where the canonical WLP appears incomplete or underrepresented in metagenomic datasets [121].

Another study on a well-known acetogen, Clostridium drakei, a hybrid model, has been suggested, where the rGly pathway has been reported to function in concert with the WLP. In this hybrid model, the format produced from CO₂ is first assimilated through the WLP and then is channelled to rGly pathway to generate glycine and eventually pyruvate. This coupling between the two pathways provides energetic flexibility, particularly under autotrophic conditions. Simultaneously, the WLP contributes to reducing equivalents and intermediate carriers such as methylene-THF, which can integrate into rGly pathway modules [122]. In the present study, although there is partial representation of the genes of the WLP, several core enzymes of the WLP, including formate dehydrogenase, formate-THF ligase and methylene-THF cyclohydrolase, were found to be copresent with the genes of the serine variant route of the rGly pathway. This suggests that similar metabolic cooperation might be at work in the studied enriched cultures.

In addition to CO₂ fixation, the detection of genes related to L-cysteine metabolism across all four enrichment cultures suggests that sulfur amino acid processing plays a supportive role in microbial adaptation across anoxic nutrient limitation conditions. Although core pathway components were detected, the absence or low abundance of certain auxiliary enzymes, particularly those involved in downstream sulfur recycling and redox buffering, indicates that this pathway is likely incomplete and variably expressed across the microbial communities. Experimental results confirmed that L-cysteine can be metabolized by enriched microbial communities, supporting both growth and modest acetic acid production. However, the comparatively low acetate yields, especially when contrasted with CO₂ and H₂ conditions, indicate that L-cysteine was likely utilized for biosynthetic processes and maintenance metabolism rather than as a primary carbon and energy source. Interestingly, even in the absence of detectable organic acids, as observed in Sambhar culture, L-cysteine supported measurable biomass accumulation, reinforcing its metabolic assimilation. These findings align with a recent study showing that L-cysteine can be enzymatically converted to pyruvate via desulfuration, providing both carbon and electrons in certain CO₂-fixing anaerobic bacteria [57]. Collectively, these findings underscore the functional diversity and metabolic flexibility of halophilic chemolithotrophic communities and their ability to incorporate sulfur amino acid metabolism into broader carbon assimilation strategies, especially under nutritionally challenging environments.

5. Conclusions

The selective enrichment conditions with CO₂ as the carbon source and H₂ as the electron donor favored the dominance of Vibrio species across all four cultures, regardless of their initial low sequence abundance in the inoculum sources. This suggests that Vibrio might play an important role in chemolithoautotrophic microbial ecosystems. Even though in the initial sediment inoculum sources few CO₂-fixing acetogens were present, they were not enriched under selective enrichment conditions. This suggests that other microbial groups were more competitive under the provided conditions, or there is a possibility that potentially novel or uncharacterized microbes were likely involved in CO₂ fixation. The production of formate and acetate in varying concentrations across the different cultures suggests that microbes might have employed diverse carbon fixation strategies through metabolic crossfeeding and potential involvement of syntrophic interactions. Metagenomics-based functional profiling revealed partial representations of the multiple CO₂ fixation pathways. Among these, the serine variant of the reductive glycine pathway was most consistently represented. To date, the serine variant of the rGly pathway has been reported in only engineered systems, and the presence of a near-complete rGly pathway in the enriched cultures from the subsurface sediments of saline environments suggests that this pathway may also be operational in natural systems. This warrants further confirmation through transcriptomics and stable isotope labeling experiments. Futhermore, the presence of genes associated with L-cysteine metabolism and its experimental validation confirmed that L-cysteine can be used as a supplementary carbon and energy source. However, its role appeared more auxiliary than primary as growth was observed in the Sambhar culture without organic acid accumulation, suggesting possible biosynthetic or redox functions. The study also highlights the unique aggregate-type growth pattern of the enriched cultures, which is most likely crucial for syntrophic interactions that facilitate metabolic exchanges under nutritionally and environmentally challenging conditions. Overall, this study broadens the understanding of microbial chemolithoautotrophy in saline and anaerobic ecosystems, highlighting the unexpected dominance of Vibrio, the metabolic potential of non-canonical CO₂ fixation pathways, and the importance of microbial interactions in enabling life under nutrient-limiting conditions. The findings also point toward the existence of unknown novel CO₂-fixing organisms, unconventional enzyme systems and the probable presence of a serine variant of the reductive glycine pathway in natural systems, warranting further genomic, transcriptomics, and cultivation-based studies to unravel their roles in global carbon cycling and biotechnological applications.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contributions

Rashmi Kiran: Investigation, Methodology, Data curation, Formal analysis and interpretation, Writing – original draft, Writing review & editing; Mohit Sharma: Whole genome metagenome data analysis and interpretation, Writing – review & editing; Srikrishna Subramanian: Methodology, Software, Writing review & editing; Sunil A. Patil: Conceptualization, Methodology, Funding acquisition, Project administration, Writing – review & editing.

Funding

This work was funded by IISER Mohali.

Data Availability

The metagenome sequencing and 16 S rRNA amplicon raw sequencing data that support the findings reported in this article have been deposited in the NCBI Sequence Read Archive (SRA) repository under the BioProject ID PRJNA1138324.

Declarations

Competing interests

The authors declare no competing interests.