Comparative Physiology and Genomics of Thermincola and Carboxydocella Strains and Description of Two Novel Isolates
Anastasia Galani, Melissa Antony Venancius, Detmer Sipkema, Diana Z. Sousa
TL;DR
Two new thermophilic bacteria were isolated and studied, revealing insights into their CO oxidation capabilities and expanding the known diversity of CO-oxidizing microbes.
Contribution
The isolation of two novel strains and the discovery of numerous candidate CO-oxidizing organisms in the Bacillota_B phylum.
Findings
All described Thermincola and Carboxydocella isolates represent a single species each.
Carboxydocella genomes have six [Ni–Fe] CO dehydrogenase gene copies, indicating high metabolic flexibility.
A genomic survey identified many Bacillota_B organisms with potential for CO oxidation.
Abstract
Thermophilic microorganisms, such as those inhabiting hydrothermal environments, play key roles in carbon monoxide (CO) metabolism, thereby influencing global carbon cycling. Members of the genera Thermincola and Carboxydocella are capable of CO oxidation via the water‐gas shift reaction, generating H2, but comprise only two and three validly described species, respectively. In this study, we report the isolation of two novel bacterial strains: strain AZ34E, affiliated with Thermincola, and strain AZ29I, affiliated with Carboxydocella. Comparative genomic and phylogenetic analyses revealed that all currently described isolates within Thermincola and Carboxydocella each represent a single species within their respective genera. Thermincola genomes contain four gene copies encoding [Ni–Fe] CO dehydrogenases, while Carboxydocella genomes harbour six copies with diverse predicted functional…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
FIGURE 1
FIGURE 2
FIGURE 3
FIGURE 4| Sample name | Environment | Location | Latitude | Longitude | Temperature (°C) | pH |
|---|---|---|---|---|---|---|
| AZ29 | Hot spring | Caldeiras Ribeira Grande | N 37°47.8717′ | W 025°29.2334′ | 49.6 | 3.64 |
| AZ34 | Hot spring | Caldeira Velha | N 37°46.9158′ | W 025°30.0281′ | 51.3 | 4.62 |
| AZ34E |
|
| AZ29I |
|
|
|
| |
|---|---|---|---|---|---|---|---|---|
| Isolation source | Hot spring (Caldeira Velha, Azores) | Hot spring (Baikal Lake area, Russia) | Hot spring (Kunashir Island, Russia) | Hot spring (Caldeiras Ribeira Grande, Azores) | Hot spring (Kamchatka Peninsula, Russia) | Hot spring (Uzon Caldera, Kamchatka Peninsula, Russia) | Hot spring (Karymskoe Lake area, Kamchatka Peninsula, Russia) | Hot spring (Uzon Caldera, Kamchatka Peninsula, Russia) |
| Temperature range (optimum)°C | 40–65 (60) | 37–68 (55) | 45–67 (57–60) | 40–70 (60) | 40–68 (58) | 45–68 (58) | 50–70 (60) | 26–70 (58–60) |
| pH range (optimum) | 6.0–8.0 (7.0) | 6.7–9.5 (8.0) | 5.9–8.0 (7.0–7.1) | 6.5–8.0 (7.0) | 6.5–7.6 (7.0) | 6.5–7.6 (7.0) | 6.2–8.0 (6.8) | 5.5–8.0 (6.5.) |
| Carbon sources/Electron donors | ||||||||
| CO | + | +* | + | + | + | + | + | − |
| H2 | − | − | +^ | − | − | − | − | +^ |
| Formate | − | − | − | − | − | − | − | + |
| Lactate | − | − | − | − | − | − | +^ | + |
| Glucose | − | − | − | + | − | + | − | + |
| Fructose | − | − | − | + | − | n.r. | − | + |
| Maltose | − | − | − | + | − | + | − | + |
| Sucrose | − | − | n.r. | + | − | + | + | + |
| Pyruvate | − | − | + ^ | + | − | + | + | + |
| Electron acceptors | ||||||||
| Sulfate | − | − | − | − | − | − | − | − |
| Nitrate | − | − | − | − | − | − | − | + |
| Fe (III) | − | − | + | + | − | + | − | + |
- —Nederlandse Organisatie voor Wetenschappelijk Onderzoek10.13039/501100003246
- —Koninklijke Nederlandse Akademie van Wetenschappen10.13039/501100001722
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsMicrobial Fuel Cells and Bioremediation · Metal Extraction and Bioleaching · Metalloenzymes and iron-sulfur proteins
Introduction
1
Anaerobic microbes capable of utilising carbon monoxide (CO) as a carbon and energy source, known as carboxydotrophs, are commonly found in hydrothermal environments (Brady et al. 2015; DePoy et al. 2020; Fukuyama et al. 2020; Omae et al. 2019; Slobodkina et al. 2022; Sokolova et al. 2005). Most of the thermophilic carboxydotrophs are hydrogenogenic, that is, they produce H_2_ as the main product via the water‐gas shift reaction (CO + H_2_O → CO_2_ + H_2_) (Diender et al. 2015; Fukuyama et al. 2020). Oxidation of CO is catalysed by a mono‐functional [Ni–Fe] CO dehydrogenase (CODH), and the deriving electrons are transferred to a ferredoxin‐like electron transfer protein (CooF). The oxidation of this protein is coupled to the reduction of protons via an energy‐converting hydrogenase (Ech), resulting in the production of H_2_ and the simultaneous generation of a proton motive force, which drives ATP conservation (Schoelmerich and Müller 2019; Svetlitchnyi et al. 2001). The three key proteins involved in this reaction are typically encoded within a single operon, known as the CODH‐Ech complex.
Among the thermophilic hydrogenogenic carboxydotrophs, only two genera are currently known to harbour obligate CO‐utilising bacteria that grow exclusively through hydrogenogenic CO oxidation: Carboxydocella and Thermincola. The first representative of the Thermincola genus, T. carboxydiphila (DSM 17129^T^ = 2204^T^ = VKM B‐2283^T^ = JCM 13258^T^), was isolated from a hot spring in the Baikal Lake region (Sokolova et al. 2005). This strain was shown to be an obligate hydrogenogenic carboxydotroph, but its genome was never sequenced. Subsequently, a second isolate, T. ferriacetica (DSM 14005^T^ = Z‐0001^T^ = VKM B‐2307^T^), was obtained from a hot spring on Kunashir Island (Zavarzina et al. 2007). In addition to CO oxidation, this strain could also perform Fe(III) reduction with H_2_ and CO_2_ as electron and carbon sources, respectively. The genome of T. ferriacetica was sequenced and published in 2015 (Lusk et al. 2015). Another Thermincola species, strain JR, was isolated from a microbial fuel cell (Wrighton et al. 2008). Although referred to as ‘Thermincola potens’, this isolate was not fully characterised. Nevertheless, its genome is publicly available (Byrne‐Bailey et al. 2010).
The genus Carboxydocella currently comprises three validly described species: Carboxydocella thermautotrophica (DSM 12326^T^ = 41^T^ = VKM B‐2282^T^), C. sporoproducens (DSM 16521^T^ = Kar^T^ = VKM B‐2358^T^) and C. manganica (DSM 23132^T^ = SLM 61^T^ = VKM B‐2609^T^). The first C. thermautotrophica strain (DSM 12356^T^) was isolated from a hot spring in the Geyser Valley (Sokolova et al. 2002). Like T. carboxydiphila , the original strain of C. thermautotrophica (DSM 12326^T^) is also considered an obligate carboxydotroph, growing exclusively by hydrogenogenic CO oxidation. A second strain of C. thermautotrophica (strain 019) was isolated in 2018 from the East Thermal Field at Uzon Caldera (Toshchakov et al. 2018). In contrast to the type‐strain, strain 019 is not an obligate carboxydotroph and can ferment sugars and pyruvate. C. sporoproducens (DSM 16521^T^) was isolated from a hot spring in the Kamchatka Peninsula (Slepova et al. 2006). This strain is also capable of sucrose and pyruvate fermentation. C. manganica (DSM 23132^T^), isolated from a hot spring in the Kamchatka Peninsula in 2012, differs significantly from other Carboxydocella species as it cannot utilise CO. Instead, it grows fermentatively with various sugars or pyruvate and respiratorily with H_2_/CO_2_ and Mn(IV) or Fe(III) (Slobodkina et al. 2012). Lastly, two Carboxydocella strains (JDF658 and ULO1) were obtained from a hot spring and a lake in Izu‐Bonin Trench and Ryukyu Trench in Japan, respectively. Although these strains have not been fully characterised and are not available in strain collections, their genomes have been sequenced and published (Fukuyama et al. 2017).
In this study, we isolated two new bacterial strains, designated strains AZ29I and AZ34E, from two hot springs on the island of São Miguel in the Azores, Portugal. Strains AZ29I and AZ34E are affiliated with the genera Carboxydocella and Thermincola, respectively. Furthermore, we performed comparative genomics between all publicly available Carboxydocella and Thermincola species, as well as other members of the Bacillota_B phylum, to investigate the genetic background and potential for CO utilisation among these groups.
Materials and Methods
2
Source of Samples
2.1
Sediment samples were collected from different high‐temperature environments during a sampling expedition on the island of São Miguel, Azores (Portugal) in September 2021. The locations and metadata of the two samples used in this study can be found in Table 1. The sediment samples were collected into 50 mL Falcon tubes and filled to the top to minimise the presence of air. To maintain anoxic conditions, samples were stored in vacuum‐sealed bags along with Oxoid AnaeroGenTM (ThermoFisher Scientific, Waltham, USA). Samples were kept at room temperature during transport to the lab and subsequent processing.
Composition of the Medium
2.2
For cultivation purposes, an AZ freshwater medium was used. This contained (per L): 0.33 g KCl, 0.14 g KH_2_PO_4_, 0.25 g NH_4_Cl, 0.5 mg NiCl_2_6H_2_O, 0.5 mg resazurin, 0.01 g (NH_4_)2_Fe(SO_4)26H_2_O, 0.1 g MgCl_2_6H_2_O, 0.3 g NaCl, 0.14 g CaCl_2_H_2_O, 1.0 g L‐Cysteine‐HClH_2_O, 2.0 g NaHCO_3_. The medium was supplemented with 10 mL/L of trace element solution containing (per L): 1.5 g N(CH_2_CO_2_H)3, 3 g MgSO_4_7H_2_O, 0.5 g MnSO_4_H_2_O, 1 g NaCl, 0.1 g FeSO_4_7H_2_O, 0.18 g CoSO_4_7 H_2_O, 0.1 g CaCl_2_2H_2_O, 0.18 g ZnSO_4_7H_2_O, 0.01 g CuSO_4_ 5H_2_O, 0.02 g AlK(SO_4_)212H_2_O, 0.01 g H_3_BO_3_, 0.01 g Na_2_MoO_4_2H_2_O, 0.03 g NiCl_2_6H_2_O, 0.3 mg Na_2_SeO_3_5H_2_O, 0.4 mg Na_2_WO_4_*2H_2_O.
All medium components were mixed, and the pH was adjusted to 7.0 using 1 M NaOH (unless stated otherwise). Afterwards, the medium was boiled and cooled to room temperature under a continuous stream of N_2_ gas. Then, 47 mL of medium was dispensed into 117 mL serum bottles under the desired gas atmosphere. The bottles were capped with bromobutyl rubber stoppers (Rubber BV, Hilversum, the Netherlands) and sealed with crimp seals. The headspace of the bottles was filled with the desired gas combination (e.g., H_2_/CO/CO_2_ or N_2_/CO/CO_2_, final pressure ranging from 1.5 to 1.7 atm). The bottles were then autoclaved for 20 min at 121°C.
Before inoculation, the medium was further supplemented with 1 mL/L of a filtered sterile vitamin stock solution containing (per L): 20 mg biotin, 200 mg nicotinamide, 100 mg para‐aminobenzoic acid, 200 mg thiamine, 100 mg pantothenic acid, 500 mg pyridoxamine, 100 mg cyanocobalamin, 100 mg riboflavin (DSM medium 141). Finally, the medium was reduced with 1 g/L of L‐Cysteine‐HCl*H_2_O. When needed, substrates such as sugars, carboxylic acids, and/or electron acceptors were added from sterile stock solutions to final desired concentrations.
Enrichment Cultures
2.3
Two enrichment cultures were initiated from the sediment samples by inoculating 6.00% (w/v) of sediment in 117 mL serum bottles containing 50 mL of AZ freshwater medium. Afterwards, the bottles were gas exchanged to a final composition of 1.7 atm H_2_/CO_2_/CO (48:12:40 v/v). The cultivation of the enrichments was performed under non‐shaking conditions, in the dark, at 50°C.
After several 10‐fold and 100‐fold dilutions (dilution: 10^45^ for AZ29 and 10^29^ for AZ34), the enriched samples were plated onto 1.50% agar plates supplemented with AZ freshwater medium. The plates were incubated in anaerobic jars under a gas phase of 1 atm N_2_/CO (80:20 v/v) at 50°C. When colonies were formed, they were picked up and transferred to 117 mL serum bottles containing 50 mL of AZ freshwater medium and a gas phase of 1.7 atm N_2_/CO_2_/CO (48:12:40 v/v). The bottles were incubated again under non‐shaking conditions at 50°C. From these samples, two were used for further analysis (AZ29I, AZ34E). Two subsequent 100‐fold serial dilutions were performed in each sample (twice up to 10^10^) to increase the chances of obtaining pure cultures. Afterwards, the purity of the cultures was first checked with phase contrast microscopy using a Leica DM2000 microscope (Leica, Microsystems, Weltzar, Germany) and further confirmed with whole genome sequencing (Novogene, Beijing, China).
Physiological Studies
2.4
The temperature range and optimum for growth of strains AZ34E and AZ29I were determined in 117 mL serum bottles containing 50 mL of AZ freshwater medium, with a gas phase of 1.7 atm N_2_/CO_2_/CO (48:12:40 v/v). For temperature tests, the initial medium pH was set to 7.0. Tests of the pH range were carried out at 60°C, under non‐shaking conditions. Growth of both strains was tested on several soluble (20 mM) and gaseous (1.7 atm) substrates, including D‐glucose, D‐fructose, sucrose, maltose, pyruvate, lactate, formate, H_2_/CO_2_ (80:20 v/v), and CO (100% v/v). The ability to use different electron acceptors was tested using nitrate (10 mM), sulfate (10 mM), and Fe(III) oxide (20 mM), with CO [N_2_/CO/CO_2_ (48:12:40 v/v)] as the electron donor. All substrate and electron acceptor experiments were conducted at 60°C under non‐shaking conditions. All tests were performed in duplicate.
The consumption and production profiles of metabolites of the enrichment cultures and the isolates were investigated using Gas Chromatography (GC) and High‐Pressure Liquid Chromatography (HPLC). Gas samples of 0.2 mL were taken from the headspace of the bottles with a 1.0 mL syringe and measured with a Compact‐GC 4.0 (Global Analyser Solutions, Breda, The Netherlands) equipped with two channels and a thermal conductivity detector. A Carboxen 1010 column was used as a pre‐column separating CO_2_ from other gases. CO and H_2_ were measured with a Molsieve 5A column operating at 140°C while CO_2_ was measured with an RT‐Q‐bond column operated at 60°C. In both channels, argon was used as a carrier gas. Organic acids and alcohols were analysed with an HPLC‐LC‐2030C (Shimadzu, Kyoto, Japan) using a Shodex SH1821 column. The system was equipped with UV and refractive index detectors and was operated at 40°C with a flow rate of 1 mL/min. The eluent used was 0.01 N H_2_SO_4_. Liquid samples of 1 mL were obtained from the liquid phase of the cultures and centrifuged for 15 min at 10,000 g at room temperature. Subsequently, HPLC vials were prepared with the supernatant and arabinose (20 mM) as an internal standard at a 2:3 (sample: internal standard v/v) ratio.
Sulfate and nitrate were measured with Ion Chromatography (IC; ThermoICS2100, Dionex, Sunnyvale, USA) equipped with a suppressed conductivity detector and a Dionex AS19 column (250 × 2 mm) (Dionex, Sunnyvale, USA) set at 30°C. As eluent, KOH was used in a gradient ranging from 1 mM during the first 4 min up to 40 mM after 20 min, at a rate of 0.4 mL/min. Samples were prepared as described for HPLC but without adding internal standards.
The Chromeleon data analysis software v7.2.9 (ThermoFisher Scientific, Waltham, USA) was used for peak integration and analysis in all machines.
DNA Extraction and Whole Genome Sequencing
2.5
For whole genome analysis of the two cultures, genomic DNA was extracted using the FastDNA Spin Kit for Soil (MP Biomedicals, Irvine, USA). To prepare the samples, 15 mL of liquid sample was centrifuged in 15 mL Falcon tubes at 3000 g for 15 min at room temperature. The supernatant was discarded, and 978 μL Sodium Phosphate Buffer was added to the pellet. Homogenisation of the mixture was done using a FastPrep 24 (MP Biomedicals, Irvine, USA) for 3 cycles of 60 s at a speed of 5.5 ms^−1^. For the DNA elution step, the catch tube was incubated at 55°C for 5 min before centrifuging the final eluate for increased elution efficiency. The quantity of extracted DNA was determined with the Qubit dsDNA BR assay (ThermoFisher Scientific, Waltham, USA) and with NanoDrop (ThermoFisher Scientific, Waltham, USA).
In addition, DNA from a growing culture of Thermincola carboxydiphila (DSM 17129^T^) was obtained from the German Collection of Microorganisms and Cell Culture (DSMZ, Braunschweig, Germany). Afterwards, whole genome sequencing was performed for all samples with the Illumina NovaSeq 6000, paired‐end 150 bp (PE150) platform by Novogene (Beijing, China).
Whole Genome Analysis of Cultures AZ29I and AZ34E and
Thermincola carboxydiphila (DSM 17129T )
2.6
Raw reads obtained from sequencing were initially inspected with FastQC v0.11.9 (Andrews 2010) for overall quality assessment. Adapter removal and quality filtering were performed with the BBDuk.sh script from BBTools v38.84 (parameters: ktrim = r, k = 23, mink = 7, hdist = 1, tpe, tbo, qtrim = rl, trimq = 20, ftm = 5, maq = 20, minlen = 50) (Bushnell 2018). BBDuk was also used to remove sequencing artefacts and phiX contamination, with default settings. Quality‐filtered reads were assembled with SPAdes v3.15.2 (Prjibelski et al. 2020) with the isolate parameter enabled.
The assemblies of AZ29I, AZ34E, and T. carboxydiphila (DSM 17129^T^) were further curated by removing contigs shorter than 1000 bp with the reformat.sh script from BBTools v38.84. The overall quality of the resulting assemblies was then assessed with QUAST v5.0.2 (Gurevich et al. 2013). Bowtie2 v2.4.1 (Langmead and Salzberg 2012) was used to map the quality‐filtered reads to the assembled contigs, resulting in sequence alignment map (SAM) files. The SAM files were converted into binary alignment map (BAM) files, sorted and indexed with SAMtools v1.18 (Danecek et al. 2021). Calculation of the genome coverage was performed with the ‘genomecov’ script from BedTools v2.29.1 (Quinlan and Hall 2010). Completeness and contamination values were calculated with CheckM v1.0.18 (Parks et al. 2015).
Initial annotations of the genomes were performed using the Prokaryotic Genome Annotation Pipeline (PGAP v6.6) (Tatusova et al. 2016) during the genome submission process to the NCBI submission portal. For comparative genomics, publicly available genomes of Thermincola and Carboxydocella strains were obtained from NCBI. An overview of the genomes used can be found in Table S1. Genomes were re‐annotated with EnrichM v0.2.1 (‘annotate’ function) (https://github.com/geronimp/enrichM) against the Kyoto Encyclopedia of Genes and Genome (KEGG) Orthologies (KOs) database. Afterwards, the EnrichM's ‘classify’ function was employed to calculate the completeness of KEGG modules. The results were plotted in R v4.3.2 (packages used: ggplot2 v3.5.1, tidyr v1.3.1, ggsci v3.1.0).
Taxonomic Placement of Cultures AZ29I and AZ34E
2.7
Initial taxonomic annotation of cultures AZ29I and AZ34E was performed with the Genome Taxonomy Database (GTDB) toolkit, GTDB‐Tk v2.3.2 (Chaumeil et al. 2020). Afterwards, a phylogenetic tree was constructed, which incorporated the genomes of AZ29I, AZ34E, Thermincola carboxydiphila (DSM 17129^T^), as well as all genomes included in the classes: c__GCA‐003054495 (12 genomes), c__Thermincolia (14 genomes), c__Moorellia (75 genomes) in GTDB r220. For the construction of the tree, 120 single‐copy bacterial gene markers were identified in the genomes and aligned with GTDB‐Tk v2.3.2. The obtained alignment was trimmed with trimAl v1.4.1 (−gappyout) (Capella‐Gutiérrez et al. 2009) (trimmed alignment: 5029 bp). The best‐fit‐substitution model was determined with IQ‐TREE v2.0.6 (Minh et al. 2020) and the ModelFinder (−m MFP) parameter. The final maximum‐likelihood phylogenetic tree was created with IQ‐TREE (model: LG + F + R6) and 1000 standard non‐parametric bootstrap replicates. The tree was visualised in iTOL v6.8.1 (Letunic and Bork 2021).
In addition, digital DNA–DNA hybridisation (dDDH) values (and confidence intervals) between the genomes of AZ29I, AZ34E, Thermincola carboxydiphila (DSM 17129^T^) and their closest relatives were calculated using the Type (Strain) Genome Server (https://tygs.dsmz.de) and the recommended settings of GGDC v4.0 (Meier‐Kolthoff et al. 2013, 2022; Yoon et al. 2017). Average nucleotide identity (ANI) values were calculated using OrthoANIu in the web service by CJ Bioscience (Yoon et al. 2017).
Phylogenetic Analysis of CODHs in Thermincola and Carboxydocella Strains
2.8
To investigate the diversity and evolutionary relationships of [Ni‐Fe] CODH enzymes, the CODH sequences were retrieved from all currently available Thermincola strains (four genomes) and Carboxydocella strains (six genomes). The CODHs of Carboxydothermus hydrogenoformans were also retrieved and used as reference points for functional predictions. A phylogenetic tree was constructed from all CODHs as described above, using the same workflow for alignment, trimming, and tree inference. The final tree was inferred with IQ‐TREE v2.0.6, using the LG + G4 substitution model and 1000 standard non‐parametric bootstrap replicates and visualised in iTOL v6.8.1. The visualisation of the CODH‐encoding gene clusters was performed with R v4.3.2 in RStudio v2023.12.0 + 369 using the R package gggenes v0.5.1 as implemented in ggplot2 v3.4.4.
Results and Discussion
3
Taxonomic Placement of Isolates and Re‐Classification of Thermincola and Carboxydocella Species
3.1
From the hot spring sediments, we obtained two pure cultures, designated as strains AZ29I and AZ34E, capable of hydrogenogenic CO oxidation. Initial taxonomic classification of the isolates, performed with GTDB‐Tk, revealed that strain AZ34E is affiliated with the Thermincola genus, while strain AZ29I is affiliated with the Carboxydocella genus. At the time of analysis, the genomes of five Carboxydocella strains and two Thermincola strains were publicly available and were incorporated in the analysis, along with the genomes of strains AZ29I, AZ34E and Thermincola carboxydiphila (DSM 17129^T^), which were sequenced in this work (Table S1). Phylogenetic analysis based on a tree created by the concatenation of 120 single‐copy bacterial genes placed isolate AZ34E and all Thermincola strains in the same well‐supported clade, indicating that they are closely related. A similar pattern was observed for isolate AZ29I and Carboxydocella strains (Figure 1a).
(a) Maximum likelihood phylogenetic tree based on the concatenation of 120 single‐copy genes. Different colours represent different families (GTDB r220). The tree scale bar represents 0.1 substitutions per site. The numbers in nodes represent bootstrap values, and accession numbers are given in parentheses. Thermoanaerobacter kivui was used as an outgroup and root of the tree. (b) Average nucleotide identity comparisons between publicly available Thermincola strains and strain AZ34E. (c) Average nucleotide identity comparisons between publicly available Carboxydocella strains and strain AZ29I.
To validate the inferred relatedness observed in the phylogenetic tree and ensure accurate assignment at the species level, we performed pairwise whole genome comparisons at the nucleotide level (ANI and dDDH). The established prokaryotic species delineation thresholds are proposed to be 95.00%–96.00% for ANI and 70.00% for dDDH (Chun et al. 2018; Jain et al. 2018; Konstantinidis et al. 2017).
The ANI (Figure 1b) and dDDH (Table S2) values for Thermincola strains were all above these threshold values (ANI 97.95%–99.00% and dDDH 80.70%–91.00%), indicating that all currently available Thermincola strains are the same species, while three Thermincola species have been described in literature: * T. carboxydiphila, T. ferriacetica
- and T. potens (with T. potens not validly characterised nor deposited in culture collections). T. carboxydiphila and T. ferriacetica share a 98.00% 16S rRNA gene similarity and 27.00 (±1.00%) DNA–DNA hybridisation level (Zavarzina et al. 2007). Based on these results, along with physiological differences between the strains and the lack of whole‐genome information for T. carboxydiphila , they were assigned to different species (Zavarzina et al. 2007). However, our analyses suggest that T. carboxydiphila and T. ferriacetica should not be classified as different species, challenging the previous classification based on limited genomic data and emphasising the need for a reassessment of their taxonomic status. While the study by Zavarzina et al. (2007) reported a DDH value of 27.00 (±1.00%), between T. carboxydiphila and T. ferriacetica , the dDDH value calculated in this study was significantly higher at 80.70 (±5.50%). This discrepancy could likely be attributed to experimental errors inherent with traditional wet‐lab DDH methods. Several studies have highlighted the limitations of DDH methods, including their low robustness, as different experimental conditions and methodologies can yield inconsistent results (Auch et al. 2010; Goris et al. 2007; Richter and Rosselló‐Móra 2009). Furthermore, their efficiency is much lower compared to modern genomic methods such as dDDH and ANI, particularly for distinguishing closely related taxa at lower taxonomic levels (Goris et al. 2007). As a result, whole‐genome approaches are now preferred over the traditional DDH methods for species delineation as they can offer greater accuracy, reproducibility and resolution.
Similarly, the ANI (Figure 1c) and dDDH (Table S3) for Carboxydocella spp. were all above the species delineation threshold values (ANI 98.38%–99.98% and dDDH 85.20%–99.90%), indicating that all tested Carboxydocella strains belong to a single species. Until recently, the genus comprised three distinct species (* C. thermautotrophica, C. sporoproducens *, and C. manganica ) that were distinguished based on DNA–DNA hybridisation and morphological/physiological differences (Slepova et al. 2006; Slobodkina et al. 2012; Sokolova et al. 2002). While we cannot draw any conclusions for C. manganica since the genome of this bacterium is not available, our analyses reveal that C. thermautotrophica and C. sporoproducens are the same species.
Physiological and Genomic Differences Among Thermincola and Carboxydocella Strains
3.2
Several physiological differences between Thermincola and Carboxydocella strains concerning the compounds they utilise as carbon and energy sources have been described in the literature (Table 2). None of the currently isolated Thermincola strains can grow organotrophically, and the same is true for our isolate AZ34E. On the other hand, all of them can grow with CO, both as an energy source and a carbon source. Nevertheless, although T. carboxydiphila is considered an obligate carboxydotroph, in order to grow on CO, it needs supplementation of either yeast extract or acetate (Sokolova et al. 2005). In contrast, strain AZ34E can grow on CO without the requirement for extra substrates. T. ferriacetica can also grow by reducing Fe(III) or Mn(IV) with CO or H_2_/CO_2_ (Zavarzina et al. 2007). Strain AZ34E was not capable of coupling CO oxidation to Fe(III) reduction. Thermincola potens (strain JR) remains uncharacterised, with limited information available regarding its physiology.
Carboxydocella isolates show greater versatility, with significant variation in carbon and energy metabolism among different strains (Table 2). While the original strain of C. thermautotrophica (DSM 12356^T^) is considered an obligate carboxydotroph relying exclusively on hydrogenogenic CO oxidation (Sokolova et al. 2002; Toshchakov et al. 2018), C. thermautotrophica (strain 019) shows additional metabolic capabilities. Strain 019 can grow fermentatively on sugars, such as glucose and sucrose, as well as on pyruvate. Furthermore, it is capable of Fe(III) reduction with CO as the electron donor (Toshchakov et al. 2018). This diverse metabolic profile is also observed in C. sporoproducens (Slepova et al. 2006) and in our isolate, strain AZ29I. In contrast, C. manganica is the only Carboxydocella isolate incapable of CO oxidation. Instead, it grows by fermenting sugars or pyruvate, as well as by reducing nitrate, Fe(III), and Mn(IV) with H_2_ as the electron donor (Slobodkina et al. 2012). The observed physiological differences in carbon and energy source utilisation among Thermincola and Carboxydocella strains cannot be traced back into the genomes. Comparative genomics between publicly available Thermincola and Carboxydocella genomes (Table S1 and Figure 2) showed more similarities than could be predicted from physiological data. The GC content ranges from 45.00% in Thermincola strains to 48.00%–49.00% in Carboxydocella strains. At the same time, Thermincola strains have slightly larger genome sizes (> 3 Mb) compared to Carboxydocella strains (2.5–2.7 Mb) but generally a lower coding density (86.00%–87.00% in Thermincola strains and 91.00%–92.00% in Carboxydocella strains) (Lusk et al. 2015; Toshchakov et al. 2018; Fukuyama et al. 2017).
Bubble plot depicting the completeness of selected KEGG modules in Thermincola and Carboxydocella genomes. Different colours represent different genomes. The size of the bubbles corresponds to different KEGG modules' completeness values.
All currently known Thermincola and Carboxydocella strains feature all genes involved in glycolysis. Nevertheless, under laboratory conditions, none of the studied Thermincola strains or the original strain of C. thermautotrophica (DSM 12356^T^) can utilise glucose or other sugars. While glycolysis genes may be present in the genome, their expression and functionality can be influenced by regulatory mechanisms, missing essential genes, preferential use of alternative energy sources, evolutionary changes, and laboratory conditions that may not accurately reflect natural environments. In some microorganisms, the glycolytic pathway can be primarily involved in gluconeogenesis rather than glycolysis. This is evident in hyperthermophilic archaea such as Thermococcus kodakarensis and Pyrococcus furiosus , where key enzymes of the classical Embden‐Meyerhof (EM) pathway enzymes—particularly glyceraldehyde‐3‐phosphate dehydrogenase (GAPDH) and phosphoglycerate kinase (PGK)—play a major role in gluconeogenesis rather than glycolysis (Matsubara et al. 2011). This is supported by genetic studies where the disruption of GAPDH and PGK genes in T. kodakarensis affected gluconeogenic growth but not glycolytic growth. The direction of flux through the glycolytic/gluconeogenic pathway can be influenced by environmental factors and metabolic needs. In Methanosarcina acetivorans , for example, the pathway operates in the gluconeogenic direction during exponential growth when an external carbon source, such as methanol, is available. However, when the carbon source is depleted, the flux shifts towards glycolysis, utilising stored glycogen (Santiago‐Martínez et al. 2016).
Regarding inorganic carbon assimilation, all Thermincola and Carboxydocella genomes possess the complete set of genes of the Wood Ljungdahl pathway (WLP), while the rest of the carbon fixation pathways (e.g., CBB cycle, rTCA cycle, 3‐HP bicycle, 3‐HP/4‐HB cycle) are incomplete. Furthermore, they all possess the genes responsible for acetate production (phosphotransacetylase and acetate kinase) from acetyl‐CoA, along with close homologues of the subunits of enzymatic complexes involved in co‐factor regeneration and energy conservation during acetogenesis, including energy‐converting hydrogenases. However, none of the strains, including strains AZ34E and AZ29I, seems capable of acetogenic growth on H_2_/CO_2_ under laboratory conditions (Toshchakov et al. 2018).
Finally, all Thermincola and Carboxydocella genomes encode complete pathways involved in vitamins (e.g., thiamin, riboflavin and cobalamin) or amino acid biosynthesis. They also contain several transporters for iron assimilation, and several members, including our isolate AZ29I, have been reported to reduce iron. The pathways for nitrate and sulfate reduction were not complete in any of the genomes analysed here, consistent with physiological data.
Carboxydotrophy in Thermincola and Carboxydocella Strains
3.3
One metabolic process common among Thermincola and Carboxydocella strains is hydrogenogenic carboxydotrophy, with C. manganica as an exception, as it is the only strain incapable of CO utilisation (Slobodkina et al. 2012). Our strains AZ34E and AZ29I were also capable of growing with CO as the sole energy and carbon source, producing equimolar quantities of H_2_ (results not shown). No other products were detected during incubation with CO.
The key enzyme involved in anaerobic carboxydotrophy is the [Ni‐Fe] CODH. All Thermincola genomes harbour four gene copies encoding CODHs (cooS genes), while Carboxydocella genomes harbour six cooS gene copies (with the exception of Carboxydocella sp. strains JDF658 and ULO1, which will be discussed later). These cooS genes are found in different gene clusters that are very conserved among the genomes of the same genus. The functional roles of some [Ni–Fe]‐CODHs have been predicted in the genome of Carboxydothermus hydrogenoformans (DSM 6008^T^), which features five cooS gene copies and have been further confirmed by phylogenetic comparisons and gene context analysis (Dobbek et al. 2001; Hedderich and Forzi 2006; Inoue et al. 2019; Svetlitchnyi et al. 2001; Techtmann et al. 2012; Wu et al. 2005).
To assign potential functional roles to the CODHs of Thermincola and Carboxydocella genomes, we conducted a phylogenetic comparison of these CODHs with well‐characterised CODHs from C. hydrogenoformans (Figure 3) and analysed their genomic context relative to other carboxydotrophs. All Thermincola and Carboxydocella genomes have one cooS gene that is part of an acetyl‐CoA synthase gene cluster. This gene cluster codes for the bifunctional CODH‐ACS (CODH‐III in C. hydrogenoformans ), which is crucial for carbon fixation via the WLP (Svetlitchnyi et al. 2004). Furthermore, they all possess a cooS gene copy that is in a cluster with genes encoding an energy‐converting hydrogenase (CODH‐Ech, CODH‐I in C. hydrogenoformans ), which is involved in energy conversion via coupling CO oxidation to H_2_ production (Hedderich and Forzi 2006; Svetlitchnyi et al. 2001).
Phylogenetic comparison of CODHs detected in the Themincola and Carboxydocella genomes. The tree was constructed with the LG + G4 model in IQ‐TREE v2.0.6 and visualised in iTOL v6.8.1. CODH sequences of Thermincola sp. strain AZ34E and Carboxydocella sp. strain AZ29I are depicted in bold. The CODHs of Carboxydothermus hydrogenoformans were used as reference points for functional predictions. The tree scale bar represents 1 substitution per site. The circles in nodes represent bootstrap values > 80. The genomic structure of cooS gene clusters encoding the different CODHs in Thermincola sp. strain AZ34E and Carboxydocella sp. strain AZ291 are shown in the boxes on the top right corner. Gene abbreviations: acsB, Acetyl‐CoA synthase; acsC, corrinoid iron–sulfur protein large subunit; acsD, corrinoid iron–sulfur protein small subunit; acsE, methyltransferase A; ATPase, AAA family ATPase; cooA, CO‐dependent transcriptional activator; cooC, CODH chaperone; cooF, ferredoxin‐like electron transfer Fe‐S protein; cooMKLXUH, sub‐units of the energy‐converting hydrogenase; cooS, CO‐dehydrogenase; FNOR, NAD/FAD oxidoreductase; hp, hypothetical protein; hypA, hydrogenase maturation protein; nqrF, Na(+)‐translocating NADH‐quinone reductase subunit F; TR, transcription regulator; YlmC, YlmC/YmxH family sporulation protein; YpiB, YpiB family protein.
The third cooS gene, found in all analysed genomes, is in a gene cluster with a gene encoding a FAD/NAD(P) oxidoreductase (CODH‐FNOR, CODH‐IV in C. hydrogenoformans ). Several functional roles have been predicted for this CODH. In C. hydrogenoformans , this gene cluster also includes a gene coding for a rubrerythrin‐like protein, leading to the hypothesis that this CODH is involved in oxidative stress responses (Wu et al. 2005). In Geobacter spp., this gene cluster is predicted to be involved in energy conservation via CO oxidation coupled to NAD(P)^+^ reduction (Geelhoed et al. 2016). In the Thermincola and Carboxydocella genomes, the gene coding for the rubrerythrin‐like protein is missing from the CODH‐FNOR gene cluster, so the latter hypothesis is more likely. Finally, the fourth cooS gene, found in all genomes, encodes a CODH that clusters with the CODH‐V from C. hydrogenoformans. No functional role has been assigned to this CODH (Toshchakov et al. 2018; Wu et al. 2005).
In Carboxydocella strains, two additional cooS genes were found that were absent from the Thermincola genomes. The CODHs encoded by these genes clustered separately in the phylogenetic tree and do not have close homologues in the genome of C. hydrogenoformans . The fifth cooS gene, found in all Carboxydocella genomes, seems to be a ‘lone’ cooS gene, not found in the vicinity of genes involved in CO metabolism, with no existing hypotheses regarding its function. In contrast, the sixth cooS gene in all Carboxydocella genomes has a more complex configuration. It is located near the CODH‐Ech gene cluster, positioned just upstream from this gene cluster in the genomes of C. thermoautotrophica (strains DSM 12356^T^ and 019) and Carboxydocella sp. strain AZ29I, while positioned downstream from this cluster in the genome of C. sporoproducens (DSM 16521^T^). This cooS gene is flanked by two genes involved in CO oxidation: the CO‐sensing transcription regulator cooA and the CODH accessory nickel‐insertion protein cooC. Due to these associations, it has been speculated that the CODH encoded by this gene might be involved in CO oxidation coupled to H_2_ production. However, its precise role remains elusive (Toshchakov et al. 2018). In the genomes of Carboxydocella strains JDF658 and ULO1, this cooS gene is found upstream from the CODH‐Ech but is annotated as a pseudogene due to an internal stop codon in strain JDF658 and a split within the gene in strain ULO1. While these errors often stem from the poor quality of an assembly, we decided to remove these two sequences from our analysis.
The presence of six cooS gene copies in Carboxydocella genomes represents the highest number reported among carboxydotrophic microbes, along with Calderihabitans maritimus, which also has six cooS copies (Omae et al. 2017). This abundance of CODHs in the Carboxydocella and Thermincola strains suggests that CO utilisation plays a crucial role in their metabolism, which could be related to the importance of CO as a substrate in their environments. Although present in minor amounts, CO is commonly detected in several hydrothermal environments around the world (Kochetkova et al. 2011; Toshchakov et al. 2018). Whether the consistently low CO concentrations result from rapid consumption by CO‐utilising hydrogenogenic microbial species remains to be determined. Nonetheless, it has been suggested that these microbes play a pivotal ecological role in hydrothermal environments by maintaining low CO levels, thereby supporting the growth of CO‐sensitive microbes and providing energy for the H_2_‐dependent microbial communities (Techtmann et al. 2012; Yoneda et al. 2015).
Genomic Potential for CO Utilisation in the Bacillota_B Phylum
3.4
According to the GTDB r220, Thermincola and Carboxydocella genera, as well as genera including other well‐known thermophilic carboxydotrophic bacteria, such as Carboxydothermus spp. and Moorella spp., have been re‐classified in the phylum Bacillota_B. This phylum at the time of the analysis, includes 1.092 genomes in GTDB distributed across 59 families. To identify potential CO‐utilising bacteria in this phylum, we checked all genomes in this phylum for the presence of genes encoding anaerobic [Ni‐Fe]‐CODHs (bifunctional: *acsA‐*part of the CODH‐ACS gene cluster and monofunctional: cooS) (Figure 4). Our analysis showed that the potential for CO utilisation is widely distributed in this phylum. Out of 1.092 genomes analysed, 364 genomes featured acsA genes, and 408 genomes featured cooS genes. The largest families within this phylum are Peptococcaceae (173 genomes), the uncharacterised UBA4997 (170 genomes, primarily represented by the Avidehalobacter genus) and Syntrophomonadaceae (116 genomes). The vast majority of genomes from these families lack all genes encoding CODHs.
(a) Maximum likelihood phylogenetic tree, based on the concatenation of 120 singly‐copy bacterial genes (model: LG + F + R5) of the Bacillota_B phylum (1.092 genomes, GTDB r220). The tree was created with IQ‐TREE v2.0.6 and visualised in iTOL v6.8.1. Clades represent distinct families (according to GTDB taxonomy). Numbers in parentheses indicate the number of genomes comprising each clade. Clades are blue when more than 80.00% of their genomes encode for the acsA or cooS genes. The tree scale bar represents 1 substitution per site. The circles in nodes represent bootstrap values. Thermoanaerobacter kivui was used as an outgroup and root of the tree. Purple circles indicate families that include validly described isolates that can grow with CO. (b) Heatmap depicting the proportion of either the acsA gene (red) or the cooS gene (blue) in the genomes belonging to different families of the Bacillota_B phylum. The proportion was calculated by identifying the presence or absence of these genes per genome. If a genome contained at least one acsA/cooS gene copy, it was assigned a positive value (1 for presence, 0 for absence). Then, the number of genomes with a positive value was divided by the total number of genomes constituting this specific family.
The families with the highest number of CODH‐encoding genomes were the Desulfitobacteriaceae (99 genomes; 93 featured the acsA gene, 84 featured cooS genes), Moorellaceae (44 genomes; 42 with acsA, 33 with cooS), Thermacetogeniaceae (42 genomes; 31 with acsA, 36 with cooS), Syntrophobotulaceae (35 genomes; 18 with acsA, 29 with cooS) and Desulforudaceae (24 genomes, 23 with acsA, 21 with cooS). Among these families, only Desulfitobacteriaceae and Moorellaceae include isolates with confirmed CO metabolism (Alves et al. 2013; Slobodkina et al. 2022; Yoneda et al. 2013). It is important to note that the presence of the bifunctional CODH‐ACS (encoded by the acsA gene) is no definitive evidence of CO utilisation. This enzyme also catalyses the reverse reaction, enabling syntrophic acetate‐oxidising bacteria (SAOB) to oxidise acetate to H_2_/CO_2_ through the WLP (Pan et al. 2021). SAOB strains, such as Thermacetogenium phaeum and Syntrophaceticus schinkii , are found within the Thermacetogeniaceae family. Of these, only T. phaeum is capable of acetogenic growth on H_2_/CO_2_. Although growth of this strain on CO has also been observed, it has proven difficult to reproduce experimentally (Oehler et al. 2012).
Our findings show that genes potentially involved in CO metabolism within Bacillota_B are traits present in approximately 50% of the families. Among these families, several are not characterised and lack isolated representatives, such as f__JAJZTU01, f__JAJZSN01, and f__JAJXUE01. This highlights the unexplored diversity of potential CO‐utilising microbes in nature, emphasising the need for further research to isolate and characterise these unknown species. Isolating these novel microbes could provide valuable insights into their metabolic pathways and ecological roles while also unlocking their potential for biotechnological applications focused on converting C1 compounds, like CO, into valuable products. Exploring this unknown microbial diversity could be an important step towards a future where microbes are used to drive a circular, carbon‐neutral economy.
Conclusions
4
In this work, we isolated two novel strains (AZ34E and AZ29I), affiliated with the Thermincola and Carboxydocella genera, respectively. We also employed whole‐genome phylogenetic analyses (phylogenetic tree, ANI and dDDH) to determine the taxonomic relationship between all currently available Thermincola and Carboxydocella species. Our analyses revealed that all known Thermincola strains belong to a single species. The same is true for all Carboxydocella strains, with the exception of C. manganica for which the genome is not publicly available.
All Thermincola and Carboxydocella strains, with the exception of C. manganica, can grow by coupling CO oxidation to the production of H_2_. The carboxydotrophic potential can be traced back to the genomes of these strains, which feature multiple copies of the cooS genes encoding CODHs, the enzymes responsible for CO oxidation. With four cooS copies in Thermincola genomes and six copies in Carboxydocella genomes, it is clear that CO is an important substrate for these microorganisms. This is further supported by the observation that most of them cannot grow organotrophically under laboratory conditions, while they all carry the necessary genes for glycolysis.
Finally, by employing whole genome analyses we revealed the presence of cooS genes in genomes in approximately half of the families within the Bacillota_B phylum, to which the Carboxydocella and Thermincola genera belong. Interestingly, several of the families with carboxydotrophic potential in this phylum remain uncharacterised and lack isolated representatives. This finding highlights the great diversity of potential CO‐utilising microbes that remain unexplored. Given the promising biotechnological applications of these microbes, future research should focus on bridging the evident gap between genomic insights and microbial isolation.
Taxonomic Conclusions
5
Based on the present study, the genomes of Thermincola carboxydiphila (DSM 17129^T^) and T. ferriacetica (DSM 14005^T^), share a 97.95% ANI value and 80.70% dDDH value. Therefore, they cannot be distinguished at the genome level and should be considered a single species. According to Rule 24b of the International Code of Nomenclature of Prokaryotes (Oren et al. 2023), Thermincola. ferriacetica (DSM 14005^T^) (Zavarzina et al. 2007) is proposed as a later heterotypic synonym of Thermincola carboxydiphila (DSM 17129^T^) (Sokolova et al. 2005). This is supported by similarities in the morphology and physiology between the two taxa (Sokolova et al. 2005; Zavarzina et al. 2007), as well as from whole‐genome analyses conducted in this study. The type strain of Thermincola carboxydiphila is 2204^T^ (=DSM 17129^T^ = VKM B‐2283^T^ = JCM 13258^T^), isolated from a hot spring of the Baikal Lake region in Russia (Sokolova et al. 2005). ‘Thermincola potens’ strain JR (Byrne‐Bailey et al. 2010; Wrighton et al. 2008) and Thermincola sp. strain AZ34E (isolated in this study) also belong to the Thermincola carboxydiphila taxon.
Additionally, in this study, we determined that the genomes of Carboxydocella thermautotrophica (DSM 12326^T^) and C. sporoproducens (DSM 16521^T^) share a 99.57% ANI value and 96.70% dDDH value. Hence, they should also be considered a single species, and Carboxydocella sporoproducens (DSM 16521^T^) (Slepova et al. 2006) is considered a later heterotypic synonym of Carboxydocella thermautotrophica (DSM 12326^T^) (Sokolova et al. 2002). The type strain of C. thermautotrophica is strain 41T (=DSM 12356^T^ = VKM B‐2282^T^), isolated from a terrestrial hot vent of the Geyzer Valley, Kamchatka Peninsula, Russia (Sokolova et al. 2002).
Author Contributions
Anastasia Galani: writing – original draft, visualisation, methodology, investigation, formal analysis, conceptualization. Melissa Antony Venancius: writing – review and editing, methodology, investigation. Detmer Sipkema: writing – review and editing, supervision, conceptualization. Diana Z. Sousa: writing – review and editing, supervision, conceptualization, funding acquisition.
Funding
This work was supported by the Nederlandse Organisatie voor Wetenschappelijk Onderzoek (P16‐10/P1) and the Koninklijke Nederlandse Akademie van Wetenschappen.
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information
Table S1: Genome features of strains AZ34E, AZ29I and publicly available Thermincola and Carboxydocella strains. The features of the publicly available strains were obtained from GTDB r220. Table S2: Pairwise dDDH comparisons among Thermincola strains, 1:
Thermincola carboxydiphila (DSM 17129), 2: Thermincola sp. strain AZ34E, 3:
Thermincola ferriacetica DSM 14005 (GCA 001263415.1), 4: Thermincola potens JR (GCA 000092945.1). C.I values refer to confidence intervals. Table S3: Pairwise dDDH comparisons among Carboxydocella strains, 1: Carboxydocella sp. strain AZ29I, 2:
Carboxydocella thermautotrophica DSM 12326 (GCA_003054495.1), 3:
Carboxydocella thermautotrophica 019 (GCA 003047205.1), 4:
Carboxydocella sporoproducens DSM 16521 (GCA 900167165.1), 5: Carboxydocella sp. ULO1 (GCA 002049255.1), 6: Carboxydocella sp. JDF658 (GCA 002049395.1). C.I values refer to confidence intervals.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Alves, J. I. , A. H. van Gelder , M. M. Alves , D. Z. Sousa , and C. M. Plugge . 2013. “ Moorella stamsii sp. nov., a New Anaerobic Thermophilic Hydrogenogenic Carboxydotroph Isolated From Digester Sludge.” International Journal of Systematic and Evolutionary Microbiology 63, no. 11: 4072–4076. 10.1099/ijs.0.050369-0.23749275 · doi ↗ · pubmed ↗
- 2Andrews, S. 2010. Fast QC: A Quality Control Tool for High Throughput Sequence Data. https://www.bioinformatics.babraham.ac.uk/projects/fastqc/.
- 3Auch, A. F. , M. von Jan , H. P. Klenk , and M. Göker . 2010. “Digital DNA‐DNA Hybridization for Microbial Species Delineation by Means of Genome‐to‐Genome Sequence Comparison.” Standards in Genomic Sciences 2, no. 1: 117–134. 10.4056/SIGS.531120.21304684 PMC 3035253 · doi ↗ · pubmed ↗
- 4Brady, A. L. , C. E. Sharp , S. E. Grasby , and P. F. Dunfield . 2015. “Anaerobic Carboxydotrophic Bacteria in Geothermal Springs Identified Using Stable Isotope Probing.” Frontiers in Microbiology 6: 897. 10.3389/fmicb.2015.00897.26388850 PMC 4555085 · doi ↗ · pubmed ↗
- 5Bushnell, B. 2018. BB Map Short Read Aligner and Other Bioinformatics Tools. https://sourceforge.net/projects/bbmap/.
- 6Byrne‐Bailey, K. G. , K. C. Wrighton , R. A. Melnyk , P. Agbo , T. C. Hazen , and J. D. Coates . 2010. “Complete Genome Sequence of the Electricity‐Producing “Thermincola potens” Strain JR.” Journal of Bacteriology 192, no. 15: 4078–4079. 10.1128/JB.00044-10.20525829 PMC 2916387 · doi ↗ · pubmed ↗
- 7Capella‐Gutiérrez, S. , J. M. Silla‐Martínez , and T. Gabaldón . 2009. “trim Al: A Tool for Automated Alignment Trimming in Large‐Scale Phylogenetic Analyses.” Bioinformatics 25, no. 15: 1972–1973. 10.1093/bioinformatics/btp 348.19505945 PMC 2712344 · doi ↗ · pubmed ↗
- 8Chaumeil, P. A. , A. J. Mussig , P. Hugenholtz , and D. H. Parks . 2020. “GTDB‐Tk: A Toolkit to Classify Genomes With the Genome Taxonomy Database.” Bioinformatics 36, no. 6: 1925–1927. 10.1093/bioinformatics/btz 848.PMC 770375931730192 · doi ↗ · pubmed ↗