Introduction

Biological CO2 reduction is an essential reaction for life on Earth responsible for the synthesis of most organic carbon molecules. Until recently, six CO2 fixation metabolic pathways were known [1, 2]. One of these pathways is the reductive tricarboxylic acid cycle (rTCA), which shares multiple enzymatic reactions with the oxidative TCA cycle (oTCA) of anaerobic microorganisms, but operating in reverse for CO2 reduction [3,4,5,6]. The oTCA cycle is a common metabolic pathway performing the oxidation of acetyl-CoA for the generation of energy and redox equivalents. The only key enzyme of the oTCA cycle that is not involved in the rTCA cycle is the citrate synthase (CS), which catalyzes the condensation of acetyl-CoA and oxaloacetate into citrate. In the rTCA cycle, citrate is converted into acetyl-CoA and oxaloacetate by an ATP-dependent citrate lyase (ACL) or by an ATP-dependent two-step pathway catalyzed by a citryl-CoA synthetase and a citryl-CoA lyase (CCS/CCL) [7,8,9,10,11].

Until recently, the CS reaction was regarded as irreversible under physiological conditions [1, 12,13,14]. Two studies have demonstrated that it is not always the case and that the oTCA cycle can operate in reverse with CS instead of ACL or CCS/CCL as key enzyme for CO2 reduction in the hydrogen or acetate-oxidizing and sulfur-reducing deltaproteobacterium Desulfurella acetivorans and hydrogen-oxidizing and sulfur-reducing Aquificae bacterium Thermosulfidibacter takaii [15, 16]. This carbon fixation pathway has been named reversed oxidative TCA cycle (roTCA) since it involves the same series of enzymes as the oTCA cycle but operating the opposite direction. Based on this new discovery and due to the widespread presence of the oTCA cycle in microbial community, it is likely that the capacity of chemolithotrophic inorganic carbon fixation in many microbes cannot be investigated only with genome annotation [17]. Likewise, microbial species with oTCA cycle isolated via laboratory cultivation with organic carbon substrates may have a hidden capability of inorganic carbon fixation encoded in their genomes, which could be revealed if put under the right selection pressure.

Fe(III)-reducing anaerobic Geobacter species are major contributor to the carbon and mineral cycles in iron-rich habitats including soils, aquatic sediment, and subsurface environments where they are often found as the dominant population [18, 19]. Because of their capacity to couple acetate oxidation via the oTCA cycle with Fe(III) reduction, one of the main ecological functions of Geobacter spp. is the degradation of organic acids generated by fermentative microbes [20,21,22]. Besides Fe(III)-based acceptors, Geobacter spp. can also transfer electrons directly to methanogens for methane production in diverse environments such as wetlands and anaerobic digesters [23,24,25]. Another widely investigated attribute of Geobacter spp. is their capacity to either transfer electrons to an electrode for the production of electricity in microbial fuel cell or to use an electrode as electron donor [26,27,28]. The most commonly studied Geobacter is Geobacter sulfurreducens PCA because it is the first one to be sequenced in the genus and gene recombination is easy to be applied [29, 30]. G. sulfurreducens was isolated as an obligately heterotroph with acetate as an electron donor and carbon source [31]. On the other hand, the functional rTCA cycle in G. sulfurreducens was developed with expressing ACL from Chlorobium limicola [26].

Besides genetic engineering, the metabolism and cell functions of G. sulfurreducens have been investigated and modified via adaptive laboratory evolution (ALE). ALE can serve to study cell metabolism and response to stresses as well as to develop more efficient and robust industrial strains [32,33,34,35,36]. During ALE, a microbe is cultivated under a selective pressure over many generations, which can lead to the acquisition of mutations conferring or altering metabolic capacities [37,38,39]. For G. sulfurreducens, ALE has been used to investigate and improve extracellular electron transfer to insoluble electron acceptor as well as the metabolic pathway involved in the oxidation of the organic substrate lactate [40,41,42].

Here, G. sulfurreducens PCA wild-type was serially transferred into a chemolithoautotrophic growth medium with formate as electron donor and carbon source and Fe(III) as electron acceptor (FcF). The capability of chemolithoautotrophic growth of the FcF-evolved strain was further evaluated in a medium, where H2 was the electron donor, CO2 was the carbon source, and Fe(III) was the electron acceptor. Genome of FcF-evolved clones was sequenced to identify the mutations that may be responsible for the observed metabolic changes. Enzymatic assays, quantitative PCR, functional genetic, and isotopologue analysis of proteinogenic amino acids were employed to investigate the carbon fixation pathway responsible for chemolithoautotrophic growth.

Materials and methods

Strain and growth condition

Strains used in this study are listed in Table S1. G. sulfurreducens PCA was routinely cultured in 25 ml anaerobic pressure tubes (15 ml headspace) with a N2:CO2 (80:20; 1 atm) atmosphere at 30 °C in either acetate-fumarate (NBAF) or acetate-Fe(III) citrate medium (FcA) at pH 6.8 as described previously [29, 31]. Formate-adapted strains were routinely maintained in FcF medium, which has the same composition as FcA but with formate replacing acetate. Heterotrophic growth experiments were conducted with NBAF (acetate 10 mM, fumarate 40 mM) and FcA (acetate 10 mM, Fe(III) citrate 50 mM) media. Chemolithoautotrophic growth experiments were carried out with FcF (formate 40 mM, Fe(III) citrate 50 mM) and FnF (formate 20 mM, Fe(III)-NTA 10 mM) media. FnF medium has the same composition as FcF but with Fe(III)-NTA instead of Fe(III) citrate as electron acceptor. Chemolithoautotrophic growth was also assessed in FcHC medium, which is similar to FcF but without formate, and the N2:CO2 atmosphere replaced by H2:CO2 (80:20; 1.3 atm). Additional growth experiments were conducted with an inorganic carbon-free FcH medium (H2–Fe(III) citrate 50 mM), where the CO2/HCO3− buffer system was replaced by 20 mM MOPS pH 7.0 buffer in the presence or not of 1 mM acetate. In FHC medium, the atmosphere is pure H2 (1.3 atm). The concentration of Fe(II) in Fe(III)-reducing cultures was measured with the ferrozine assay [43].

Adaptive laboratory evolution with formate

The ALE experiment with formate as electron donor and carbon source was started by transferring 15% of an FcA-grown G. sulfurreducens PCA culture into 10 ml sterile FcF medium. Subsequent 15% transfers were done each time when FcF cultures grown anaerobically at 30 °C reached 15 mM Fe(II). The ALE experiment was stopped after 23 transfers as Fe(III) reduction rate stabilized and did not accelerate further. Culture from transfer #17, which was the fastest to reach 15 mM Fe(II), was streaked on plates and several colonies were picked for cultivation into FcF liquid medium. PCR was used to confirm that the isolated clones were G. sulfurreducens.

Whole-genome DNA sequencing

Genomic DNA of formate-adapted clones T17-3 and T17-4 was extracted with the M5 Bacteria Genomic DNA Kit (Mei5 Biotechnology, China). The NEB Next Ultra DNA Library Prep Kit for Illumina (NEB, MA, USA) was used to generate genomic DNA libraries. Adapters were ligated to DNA fragments ranging from 300 to 400 bp, which were then recovered by beads purification. A PCR of 12 cycles was used to enrich the adapter-modified DNA before high-throughput sequencing with an Illumina HiSeq 4000 (Illumina, CA, USA) with a paired-end protocol and read lengths of 150 nt. The sequencing reads were trimmed with PRINSEQ and aligned for variant calling with the Picard DNA-seq analysis pipeline and BBTools [44,45,46]. The reference genome for the analysis was G. sulfurreducens PCA (NCBI reference sequence NC_002939.5). All the samples have an average coverage of at least 30×.

Quantitative reverse transcription PCR (RT-qPCR)

Sequences of all primers used for RT-qPCR are listed in Table S2. Total RNA was extracted with the TRIzol Max Bacterial RNA Isolation Kit (Thermo Fisher Scientific, MA, USA) from mid-log FcF or FcA cultures. RNA samples were then treated with DNase I (RNase-free) (Thermo Fisher Scientific) to remove DNA contamination. cDNA was generated from treated RNA samples with the RevertAid First Strand cDNA Synthesis Kit (Thermo Fisher Scientific). The SYBR green Real-Time PCR supermix (Mei5 Biotechnology) and the CFX Connect Real-Time PCR Detection System (Bio-Rad, CA, USA) were used to amplify and quantify PCR products for GSU0514, sucC, and gltA. The expression of these three genes was normalized with proC expression, a gene expressed constitutively in G. sulfurreducens [47]. Relative levels of expression of GSU0514, sucC, and gltA were calculated by the 2–ΔΔCT method [48].

sucCD − mutant construction

Sequences of all primers used for construction of ΔsucCD::Kmr mutant allele are listed in Table S2. The mutant construction was done as previously described [49]. The coding sequence of sucCD was replaced by a kanamycin resistance cassette. Primer pair sucCDUP/sucCDfusUP was used to amplify 508 bp upstream from sucC (GSU1058) and sucCDfusDN/sucCDDN 582 bp downstream from sucD (GSU1059) with G. sulfurreducens T17-3 genomic DNA as template. The KmRXhoIUp/KmRXbaIDn primer pair was used to amplify the kanamycin resistance cassette from pBBR1MCS-2 [50]. UP and DN PCR products were combined by overlapping extension PCR resulting into the ΔsucCD allele, and then cloned with a Zero Blunt TOPO PCR Cloning Kit (Thermo Fisher Scientific). The kanamycin resistance cassette was digested by XhoI–XbaI and ligated with the T4 DNA ligase into the ΔsucCD allele resulting into a ΔsucCD::Kmr allele. The vector carrying ΔsucCD::Kmr was linearized by digesting with ClaI and concentrated by ethanol precipitation [40]. Electroporation into the adapted strain T17-3 of the linearized vector carrying ΔsucCD::Kmr and mutant selection was performed as previously described [29, 51]. The sucCD− mutant was verified by PCR and Sanger sequencing.

Cell counting, high performance liquid chromatography, isotopologue analysis, and enzymatic assays

The methods for cell counting, high performance liquid chromatography, isotopologue analysis, succinyl-CoA synthetase, and CS assays have been included in the Supplementary Information.

Results

Adaptation of G. sulfurreducens with formate as electron donor and carbon source

It has been assumed that G. sulfurreducens PCA has a heterotrophic metabolism and cannot grow with CO2 or other C1 compounds such as formate as sole source of carbon. In previous studies investigating the metabolism of G. sulfurreducens with H2, CO, or formate as electron donor, acetate and/or fumarate were always provided as carbon sources [52, 53]. The genome sequence of G. sulfurreducens also supported the claim that this bacterium is possibly an obligate heterotroph since genes coding for the essential enzymes of the six previously known carbon fixation pathways are absent (Table S3) [30].

With the recent discovery of the CS-dependent carbon fixation pathway roTCA cycle in anaerobic bacteria [15, 16], it became interesting to reassess more carefully the chemolithoautotrophic potential of G. sulfurreducens PCA. Like multiple other bacteria, G. sulfurreducens PCA’s genome codes for the full complement of enzymes necessary for TCA cycles in anaerobes (Fig. 1a). Furthermore, the activity of most of these enzymes has already been demonstrated (Table S3) [21, 54]. The complete components of TCA cycle and the capability of hydrogen-oxidizing metabolism in G. sulfurreducens PCA suggest the genomic capability of roTCA cycle while chemolithoautotrophic growth was not observed in this strain.

Fig. 1: The TCA cycle of G. sulfurreducens in both directions.
figure 1

a roTCA cycle for carbon fixation. All the enzymes required for roTCA cycle are present in G. sulfurreducens. Genes coding for ACL or CCS/CCL required for carbon fixation rTCA cycle are not found in G. sulfurreducens. b oTCA cycle for organic substrate oxidation. When growing with acetate as substrate, G. sulfurreducens has a modified oTCA cycle where succinyl-CoA synthetase has minimal activity and is mostly replaced by the succinyl-CoA:acetate CoA-transferase [21, 54].

Full size image

To assess if G. sulfurreducens PCA has a chemolithoautotrophic capability harbored in its genome, an ALE experiment was initiated where the bacterial culture was transferred serially into FcF medium. The medium contains the C1 compound formate as the electron donor and carbon source and Fe(III) citrate as the electron acceptor. Fe(III), an electron acceptor of Geobacter spp., is commonly supplemented as Fe(III) citrate, and growth and isotope tracing experiments have established that citrate is not used as a carbon source of G. sulfurreducens [26, 55, 56]. These results are to be expected since no bacterial transporter demonstrated experimentally to be capable of citrate uptake are encoded on the genome of G. sulfurreducens [30]. Furthermore, G. sulfurreducens reduces Fe(III) outside of the cell at the outer surface and does not have homologs for the Fe(III) citrate transporter complex FecABCDE found in other Gram-negative bacteria such as E. coli [57, 58].

The first transfer into FcF was from an acetate-grown PCA culture in an Fc medium with acetate as electron donor and organic carbon source and Fe(III) as electron acceptor. This 15% transfer required 8.2 days to reach 15 mM Fe(II) in FcF medium (Fig. 2a). Subsequently, a 15% inoculum from the first FcF culture was transferred into fresh FcF medium for the second transfer, which took 5.9 days to reach 15 mM Fe(II). After five transfers in FcF, acetate concentration was measured at the beginning of transfer 6 and no residual acetate from the initial FcA culture was detected. After ten transfers in FcF fresh medium, the PCA culture was clearly adapting and required only 2.8 days to reach 15 mM Fe(II). The Fe(III) reduction rate of the adapted PCA culture became stabilized after transfer 15 and did not improve significantly afterward. The ALE experiment was stopped at transfer 23. Culture from transfer 17, which reached 15 mM Fe(II) after 1.3 days, was used for further experiments.

Fig. 2: Adaptation, growth, and metabolism of G. sulfurreducens PCA with formate as carbon source and electron donor as well as Fe(III) citrate as electron acceptor.
figure 2

a Number of days to reach 15 mM Fe(II) for each transfer during the ALE experiment with FcF medium. b Growth of PCA, the FcF-adapted culture T17, the clones T17-3, and T17-4 on FcF medium and c FcA medium. d Fe(III) reduction, formate oxidation, citrate concentration, and cell count increase when clone T17-3 grew in FcF medium. e Total formate consumed, oxidized formate equivalent required for observed Fe(III) reduction and f percentage of oxidized formate used for Fe(III) reduction at 48, 96, and 144 h by T17-3 in FcF medium. Formate is oxidized in G. sulfurreducens according to the reaction HCOO− → CO2 + H+ + 2e−. Fe(III) is reduced in G. sulfurreducens according to the reaction Fe3+ + e− → Fe2+. Overall, 1 mol formate reduces 2 mol Fe(III). Each curve and bar for b–f is the mean of three replicates with standard deviation.

Full size image

To obtain individual clonal population, colonies were isolated from the T17 culture and tested for Fe(III) reduction in FcF medium. Culture T17, clones T17-3 and T17-4 reduced Fe(III) at a rate of 490.3 ± 23.0, 417.5 ± 54.0, and 414.3 ± 21.5 μM h−1, respectively (Fig. 2b). In comparison, unadapted PCA after one transfer from FcA to FcF reduced Fe(III) at a rate of 94.3 ± 23.0 μM h−1, which was 5.2 times slower than culture T17. When grown on FcA, culture T17 and derived clones had significantly longer lag phase than PCA, which showed that gain of fitness when growing chemolithoautotrophically with formate was accompanied with a loss of fitness for heterotrophic growth with acetate as substrate (Fig. 2c).

Cell count, formate oxidation, citrate concentration, and Fe(III) reduction

To confirm that adapted bacteria grow in FcF medium and not only reduce Fe(III), T17-3 cells were counted at different time points over a period of 143 h after inoculation (Fig. 2d). From 45 to 93 h, the cell number in the FcF medium increased by 7.45 × 107 ± 1.05 × 107 cells ml−1 (n = 3). In comparison, increase in T17-3 cell number was negligible when no formate was provided (Fig. S1). The results demonstrated that Fe(III) reduction with formate as energy and carbon sources was accompanied with cell biosynthesis, division, and growth. When formate consumption ceased, T17-3 cell number stopped increasing and eventually decreased, which indicates the end of biosynthesis and the beginning of cell death.

Besides Fe(III) reduction and cell number increase, formate concentration was also monitored during growth of the T17-3 clone (Fig. 2d). During exponential growth phase, more formate than required for the observed Fe(III) reduction was consumed by the T17-3 clone (Fig. 2e, f). Only 60.7 ± 21.0% and 73.1 ± 15.5% of all the electrons derived from formate consumed by cells after 48 and 96 h of growth, respectively, were used for energy conservation via Fe(III) reduction. During the stationary growth phase from 96 to 144 h, formate consumption ceased while Fe(III) reduction continued at a slower rate compared with 0 to 96 h. After 144 h, all the electrons derived from oxidized formate have been employed for Fe(III) reduction (Fig. 2e, f). In the absence of formate oxidation in the stationary growth phase, it is highly possible that T17-3 cell starts reducing Fe(III) with electrons coming from substrates excreted by cells, from metabolites released by dead cells or from compounds stored beforehand in cell reserve.

In G. sulfurreducens, formate is oxidized (formate → CO2 + H+ + 2e−, E0′ = −420 mV) by the periplasmically oriented membrane-bound formate dehydrogenase FdnGHI [59]. For chemolithoautotrophic growth on formate as substrate to be feasible, part of the electrons generated by the oxidation reaction catalyzed by the formate dehydrogenase must be carried by the electron transport chain to the final electron acceptor Fe(III), while protons released in the periplasm form the proton gradient necessary for energy conservation. Concomitantly, another fraction of electrons coming from formate oxidation must be used for the metabolic reduction of CO2 required for the synthesis of cell components. Both formate consumption and Fe(III) reduction results during the exponential growth phase of strain T17-3 support this dual metabolism where formate is used simultaneously as a substrate for biosynthesis and as an electron donor for energy conservation.

Besides formate, the C6 organic carbon compound citrate is also present in the FcF medium under the form of Fe(III) citrate. Although a large number of evidences demonstrated that Fe(III) citrate is not used as a carbon source by G. sulfurreducens [26, 55, 56], the citrate concentration was monitored during growth (Fig. 2d). From time 0 to 144 h, citrate concentration did not change, which suggested that it was not used by strain T17-3.

Chemolithoautotrophic growth of G. sulfurreducens T17-3 with alternative substrates

To provide additional evidences of chemolithoautotrophic metabolism, adapted strain T17-3 was also cultivated with H2 as electron donor, CO2 as carbon source and Fe(III) citrate as electron acceptor (FcHC), as well as with formate as electron donor and carbon source and Fe(III)-NTA as electron acceptor (FnF) (Fig. 3a, b). Before collecting data for Fe(III) reduction curves with T17-3, four prior transfers on either FcHC or FnF were carried out from an original FcF inoculum to ensure complete depletion of carbon sources, electron donors, and acceptors found in FcF medium. Fe(III) reduction curves with PCA were done with the second transfer on FcHC or FnF from an FcA culture. T17-3 grew in FcHC at a Fe(III) reduction rate of 346.9 ± 57.3 μM h−1 between 50 and 100 h and significant increase of cell number while PCA did not reduce Fe(III) over an incubation period of 144 h (Fig. S1). In addition, citrate concentration was also assessed during chemolithoautotrophic growth of T17-3 in FcHC medium (Fig. S2). Similar to the results from FcF, citrate concentration did not vary in FcHC medium. For FnF growth, T17-3 had a Fe(III) reduction rate of 95.7 ± 29.7 μM h−1, while G. sulfurreducens PCA did not reduce Fe(III)-NTA over 230 h (Fig. 3b). These results demonstrated that chemolithoautotrophic growth by G. sulfurreducens T17-3 is not specific to formate and Fe(III) citrate, but is also observed with CO2 as carbon source, H2 as electron donor, and Fe(III)-NTA as electron acceptor.

Fig. 3: G. sulfurreducens T17-3 growth with alternative carbon sources, electron donors, and acceptors.
figure 3

a Fe(III) reduction by PCA and T17-3 in an FcHC medium with H2 as electron donor, CO2 as carbon source, and Fe(III) citrate as electron acceptor. b Fe(III) reduction by PCA and T17-3 in an FnF medium with formate as electron donor and carbon source and Fe(III)-NTA as electron acceptor. c Fe(III) reduction by PCA and T17-3 in an FcH medium without inorganic carbon CO2/HCO3−, with H2 as electron donor, Fe(III) citrate as electron acceptor, and MOPS as buffer in the presence or not of 1 mM acetate. Each curve is the mean of three replicates with standard deviation.

Full size image

Removal of CO2 and HCO3 − prevents H2-driven growth of T17-3

Chemolithoautotrophic capability of G. sulfurreducens T17-3 was further investigated by growing cells in an FcH medium with H2 as the electron donor and Fe(III) citrate as the electron acceptor, but where CO2 and HCO3− had been removed from the gas phase and the liquid phase (Fig. 3c). An MOPS buffer was added in this growth medium to replace the bicarbonate buffer system and maintain pH to 7.0. Neither PCA nor T17-3 could grow in the absence of CO2/HCO3− providing further evidence that the formate adaptation process activated a chemolithoautotrophic metabolism in T17-3 dependent on the presence of inorganic carbon sources. When 1 mM acetate was provided as carbon source with the MOPS-based FcH medium, both PCA and T17-3 reduced Fe(III) at a rate of 577.3 ± 73.8 and 724.3 ± 56.8 μM Fe(II) h−1, respectively.

Isopotomer analysis of chemolithoautotrophic growth with H13CO3 −

To investigate if the roTCA cycle was active in the adapted strain T17-3, 6 mM H13CO3− was added to mid-log phase cultures growing either chemolithoautotrophically with FcHC or heterotrophically with FcA. Sixteen hours after the addition of 13C, cells were harvested, hydrolyzed, and proteinogenic amino acids were derivatized for isotopologue profiling by GC-MS (Fig. S3). 13C enrichments of entire amino acids (m-57)+ and amino acids lacking the α-carboxyl group (m-159)+ were analyzed to determine 13C position [60]. Alanine, aspartate, and glutamate were selected for the isopotologue analysis since they are synthesized from precursors generated by the TCA cycles [56].

13C enrichment of all three amino acids in H2:CO2-grown cells compared with acetate-grown cells indicated that the roTCA cycle was functional in T17-3 (Figs. 4 and S4). For alanine, 13C labeling of C1 indicated carboxylation of acetyl-CoA by PFOR for both acetate-grown and H2:CO2-grown cells. Carboxylation of acetyl-CoA by PFOR is a contributing gluconeogenetic reaction for the biosynthetic metabolism of G. sulfurreducens during growth with acetate as carbon source and Fe(III) as electron acceptor [56]. However, it was only with H2:CO2-grown cells that a significant fraction of alanine was labeled with two to three 13C (Fig. 4a). This result shows that pyruvate molecules, the precursor of alanine, were synthesized with at least two inorganic carbon molecules probably via the roTCA cycle.

Fig. 4: Isopotomer analysis of G. sulfurreducens T17-3.
figure 4

13C enrichments of a alanine, b aspartate, and c glutamate mass fragments from G. sulfurreducens T17-3 grown either chemolithoautrophically or heterotrophically with H13CO3−. (M-57)+ is the 13C enrichment of the entire amino acids while (m-159)+ is the amino acid without the α-carboxyl group. n × 13C or n–y × 13C indicates the presence of n 13C or n to y 13C in the amino acid, respectively.

Full size image

Aspartate and glutamate molecules with 13C were also significantly more frequent when T17-3 was grown chemolithoautotrophically (Fig. 4b). Precursors for aspartate and glutamate are oxaloacetate and 2-oxoglutarate, respectively. The isotopologue profile for aspartate showed that both PFOR and the pyruvate carboxylase carboxylated acetyl-CoA into oxaloacetate. Furthermore, the detection of aspartate molecules labeled with three or four 13C indicated that oxaloacetate underwent multiple turns in the roTCA cycle. 13C enrichment and position in glutamate molecules from T17-3 grown with H2:CO2 versus acetate provides additional evidences of the activity of the roTCA cycle in the adapted strain (Fig. 4c).

Mutations in G. sulfurreducens T17-3 and T17-4

The genomes of formate-adapted clones T17-3 and T17-4 were sequenced to gain insights on how G. sulfurreducens evolved to grow chemolithoautotrophically in FcF (Table S4). T17-3 and T17-4 had a total of 15 single-nucleotide polymorphisms (SNP), 1 insertion, and 2 deletions in common. One SNP was exclusive to T17-4. Mutations were located in or near genes coding for hypothetical proteins, membrane proteins, c-type cytochromes, and enzymes involved in RNA methylation, oxidoreduction, glycerophospholipid metabolism, nucleotide salvage, and branched-chain amino acid catabolism. dcuB, the gene coding for the anaerobic C4-dicarboxylate transporter essential for the growth of Geobacter spp. with fumarate as electron acceptor [61], was mutated with an SNP resulting in the substitution of tyrosine-385 for a histidine. However, this mutation had no major impact since the adapted strains T17-3 and T17-4 could still grow with fumarate as electron acceptor (Fig. S5).

rpoB, the gene coding for the β-subunit of the RNA polymerase, had an SNP resulting in the substitution of glutamine-1089 by an arginine in the C-terminal part of the protein (Table S4) [62]. Mutations in rpoB or in other global transcription-related genes have been observed in multiple ALE experiments where they have major beneficial impacts on fitness by affecting simultaneously the expression of hundreds to thousands of genes [63,64,65]. These important changes in the expression profile usually lead to pleiotropic effects on cell’s function [66, 67]. Here, genome sequencing results suggest that rpoB mutation is involved in the capacity of G. sulfurreducens to adapt and grow chemolithoautotrophically on formate.

Role of the succinyl-CoA synthetase in the adapted strain

Among the 19 mutations found by whole-genome sequencing, one SNP in T17-3 and T17-4 was located upstream of GSU0514, which is a gene that has been shown to have a direct regulatory impact on the oTCA cycle of G. sulfurreducens. GSU0514 is a transcriptional regulator of the IclR family involved in the repression of the transcription of sucCD, the two genes coding for the succinyl-coA synthetase, which is a central enzyme of the oTCA cycle [42] as well as of the roTCA and rTCA cycles [1, 15]. In G. sulfurreducens PCA grown with acetate as substrate, the conversion of acetate into acetyl-CoA prior to oxidation via oTCA is achieved by the succinyl-CoA:acetate CoA-transferase, which does not require ATP hydrolysis contrary to the alternative pathway catalyzed by an acetate kinase and a phosphotransacetylase (Fig. 1b) [21, 54]. When the succinyl-CoA:acetate CoA-transferase is active, the succinyl-CoA synthetase is not required for the completion of the oTCA cycle. This has been highlighted by the absence of succinyl-coA synthetase activity in G. sulfurreducens growing heterotrophically [21, 54]. Since G. sulfurreducens is usually maintained with acetate as substrate in laboratory conditions, the bacterium has adapted and optimized its acetate metabolism by repressing the expression of succinyl-CoA synthetase via a GSU0514-dependent regulatory mechanism.

However, when acetate is not available in sufficient quantity for the succinyl-CoA:acetate CoA-transferase, G. sulfurreducens must adapt and reoptimize its metabolism to ensure efficient growth. This has been demonstrated by adapting G. sulfurreducens to grow on lactate as sole substrate [42]. After serial transfer of G. sulfurreducens PCA on lactate, mutations inactivating GSU0514 appeared and the expression of the genes coding for the succinyl-coA synthetase had a multifold increase. For chemolithoautotrophic growth of G. sulfurreducens with formate or other C1 compounds, the succinyl-CoA synthetase would be required for the completion of a functional roTCA cycle, which may explain the mutation found upstream of GSU0514 in T17-3 and T17-4.

The SNP A → G was found in position −161 in relation to the start of the coding sequence of GSU0514 and was localized in the −35 box of the promoter of GSU0514 as predicted by BPROM (Fig. S6) [68]. Quantitative PCR showed that GSU0514 was downregulated 3.6-fold (3.4–3.8) in T17-3 grown with formate compared with PCA grown with acetate (Fig. 5a). Concomitantly, sucC, the gene coding for the β-subunit of the succinyl-CoA synthetase, was upregulated fivefold (4.2–6.0) in T17-3 grown in FcF compared with PCA grown in FcA. The enzymatic activity of succinyl-CoA synthetase was also measured by monitoring the formation of succinyl-CoA from succinate, ATP, and CoA in whole-cell lysates of T17-3 grown in FcF and PCA grown in FcA (Fig. 5b) [69]. A barely detectable succinyl-CoA synthetase activity of 4.81 ± 2.69 nmol succinyl-CoA min−1 mg per protein was measured in PCA grown in FcA. In comparison, succinyl-CoA synthetase activity in T17-3 grown in FcF was 3.4-fold higher at 16.41 ± 2.86 nmol succinyl-CoA min−1 mg per protein.

Fig. 5: Transcript abundance fold change of succinyl-CoA synthetase and citrate synthase genes, enzymatic activities, and growth of the T17-3 sucCD− mutant.
figure 5

a Transcript abundance fold change of GSU0514, sucC, and gltA in strain T17-3 grown in FcF medium compared with PCA grown in FcA medium. b Enzymatic activity of the succinyl-CoA synthetase (SucCD). Succinyl-CoA synthetase activity was measured via the formation of succinyl-CoA from succinate according to the reaction succinate + CoA + ATP → succinyl-CoA + ADP. c Growth of the T17-3 sucCD− mutant on FcF and FcA medium. d Enzymatic activity of CS (GltA) in whole-cell lysates of PCA grown in FcA and of T17-3 grown in FcF. CS activity was measured via oxaloacetate formation coupled with porcine malate dehydrogenase for NADH oxidation according to the following reactions: (1) citrate + CoA → acetyl-CoA + oxaloacetate and (2) oxaloacetate + NADH + H+ → malate + NAD+. The results shown are from at least three replicates for each experiment with standard deviation.

Full size image

Furthermore, deletion in the adapted strain T17-3 of both sucC and sucD, the gene coding for the α-subunit of the succinyl-CoA synthetase, prevented growth in FcF but not in FcA medium (Fig. 5c). The absence of a functional succinyl-CoA synthetase appears to disrupt the roTCA cycle, which inhibits chemolithoautotrophic growth by G. sulfurreducens. This result confirms the importance of the SNP upstream of GSU0514 for the adaptation of G. sulfurreducens to grow with formate as substrate.

Activity of the citrate synthase in the reductive direction

The CS is a key enzyme catalyzing an essential step of the oTCA cycle in the oxidative direction and of the roTCA cycle in the reductive direction (Fig. 1). In G. sulfurreducens PCA, activity of the CS GltA in the oxidative direction where free CoA and citrate are formed by the condensation of acetyl-CoA and oxaloacetate has already been observed [54]. However, GltA must catalyze the reverse ATP-independent conversion of citrate and CoA into acetyl-CoA and oxaloacetate for the roTCA cycle to be functional. This activity has never been demonstrated before in G. sulfurreducens. Here, GltA activity in the reductive direction was coupled with porcine malate dehydrogenase and NADH oxidation was monitored [15]. The reductive CS activity in the whole-cell lysate of PCA grown in FcA was 102.87 ± 1.97 nmol NAD+ min−1 mg per protein (Fig. 5d). In the adapted strain T17-3 grown in FcF, CS activity was still detected but lower at 40.19 ± 1.97 nmol NAD+ min−1 mg per protein. Quantitative PCR results are in line with enzymatic activity. gltA was expressed 5.1 (4.0–6.4) times less in T17-3 grown in FcF compared with PCA grown in FcA (Fig. 4a).

The lower activity and transcription of CS in T17-3 grown in formate is to be expected if we consider that CS participates into the roTCA cycle during chemolithoautotrophic growth with formate, and its main function is the biosynthesis of cell components from CO2. While during heterotrophic growth on acetate, CS participates into the oTCA cycle, and its principal function is to release energy stored in organic carbon substrate for respiration. Thus, metabolic flux is likely to be more important through CS in an oxidative context for respiration than in a reductive context for biosynthesis and, under these circumstances, the cell may limit energy expense by decreasing CS expression.

Discussion

Genome annotation have led to the suggestion that Geobacter bemidjiensis has a chemolithoautotrophic metabolism via the rTCA cycle [70]. Growth experiment showed that another Geobacter species, Geobacter metallireducens, can grow chemolithoautotrophically, and based on genome annotation, it has been suggested that either the rTCA or the dicarboxylate–4-hydroxybutyrate cycles are involved [71]. The discovery of the roTCA cycle [15, 16] as well as experimental evidences presented here demonstrating chemolithoautotrophic growth via roTCA cycle by G. sulfurreducens raise the possibility that carbon fixation capability is more widespread than previously thought in the Geobacteraceae family. Genes coding for CS and other enzymes of the o/roTCA cycles are present in the vast majority of Geobacter species genomes sequenced until now and this includes G. bemidjiensis and G. metallireducens. When compared with the rTCA and the dicarboxylate–4-hydroxybutyrate cycles, the roTCA cycle appears to be more efficient bioenergetically [15]. Thus, G. bemidjiensis and G. metallireducens thought to reduce CO2 via alternative carbon fixation pathways may also use the roTCA cycle concomitantly or preferentially.

Several recent studies show that Geobacter spp. have a chemolithoautotrophic metabolism [70, 71]. Based on these developments, the ecological niche of Geobacter species as degraders of organic acid substrates released by fermentative microbes may have to be reconsidered. At the very least, the chemolithoautotrophic capacity of Geobacter spp. confers a competitive advantage over other species and may explain why this family can maintain itself in habitats where organic carbon substrates are likely to fluctuate [18, 19].

In addition, the chemolithoautotrophic metabolism of G. sulfurreducens is intriguing for potential industrial applications since this bacterium can synthesize bioelectronic compounds such as electrically conductive pili and can be employed as microbial catalyst in bioelectrochemical systems [26, 72]. Thus, bioproduction processes catalyzed by G. sulfurreducens could be driven by microbial electrosynthesis or could be optimized for the synthesis of bioelectronic parts from the greenhouse gas CO2.

Furthermore, another interesting observation reported here is that an anaerobic bacterial species previously thought to be a strict heterotroph has a chemolithoautotrophic metabolic capacity that was revealed by ALE. G. sulfurreducens PCA was isolated in the laboratory with acetate as electron donor and carbon source [31]. This isolation method might have introduced metabolic biases and adaptive mutations hiding physiological functions with major role in the natural habitat of the isolated microbe. Based on the observation here, it is possible that other laboratory-isolated anaerobic microbes with genes for the roTCA cycle have been characterized prematurely as strict heterotroph.