The energy metabolism of Cupriavidus necator in different trophic conditions

ABSTRACT

The “knallgas” bacterium Cupriavidus necator is attracting interest due to its extremely versatile metabolism. C. necator can use hydrogen or formic acid as an energy source, fixes CO₂ via the Calvin-Benson-Bassham (CBB) cycle, and grows on organic acids and sugars. Its tripartite genome is notable for its size and duplications of key genes (CBB cycle, hydrogenases, and nitrate reductases). Little is known about which of these isoenzymes and their cofactors are actually utilized for growth on different substrates. Here, we investigated the energy metabolism of C. necator H16 by growing a barcoded transposon knockout library on succinate, fructose, hydrogen (H₂/CO₂), and formic acid. The fitness contribution of each gene was determined from enrichment or depletion of the corresponding mutants. Fitness analysis revealed that (i) some, but not all, molybdenum cofactor biosynthesis genes were essential for growth on formate and nitrate respiration. (ii) Soluble formate dehydrogenase (FDH) was the dominant enzyme for formate oxidation, not membrane-bound FDH. (iii) For hydrogenases, both soluble and membrane-bound enzymes were utilized for lithoautotrophic growth. (iv) Of the six terminal respiratory complexes in C. necator H16, only some are utilized, and utilization depends on the energy source. (v) Deletion of hydrogenase-related genes boosted heterotrophic growth, and we show that the relief from associated protein cost is responsible for this phenomenon. This study evaluates the contribution of each of C. necator’s genes to fitness in biotechnologically relevant growth regimes. Our results illustrate the genomic redundancy of this generalist bacterium and inspire future engineering strategies.

IMPORTANCE

The soil bacterium Cupriavidus necator can grow on gas mixtures of CO₂, H₂, and O₂. It also consumes formic acid as carbon and energy source and various other substrates. This metabolic flexibility comes at a price, for example, a comparatively large genome (6.6 Mb) and a significant background expression of lowly utilized genes. In this study, we mutated every non-essential gene in C. necator using barcoded transposons in order to determine their effect on fitness. We grew the mutant library in various trophic conditions including hydrogen and formate as the sole energy source. Fitness analysis revealed which of the various energy-generating iso-enzymes are actually utilized in which condition. For example, only a few of the six terminal respiratory complexes are used, and utilization depends on the substrate. We also show that the protein cost for the various lowly utilized enzymes represents a significant growth disadvantage in specific conditions, offering a route to rational engineering of the genome. All fitness data are available in an interactive app at https://m-jahn.shinyapps.io/ShinyLib/.

INTRODUCTION

The soil bacterium Cupriavidus necator H16 (formerly Ralstonia eutropha H16) is a widely recognized model organism for chemolithoautotrophic growth (1, 2) and for the production of the bioplastic polymer polyhydroxybutyrate (PHB) (3, 4). Its large inventory of enzymes enables it to grow on a wide range of heterotrophic substrates as well as on mixtures of molecular oxygen (O₂), molecular hydrogen (H₂), and carbon dioxide (CO₂) (5–7). To fuel CO₂ fixation by the Calvin-Benson-Bassham (CBB) cycle, C. necator generates ATP and reduction equivalents from the oxidation of molecular hydrogen (H₂ → 2e⁻ + 2H⁺) or formic acid (HCOOH → CO₂ + 2e⁻+ 2H⁺). Genome redundancy and phylogenetic comparison suggest that the ability for autotrophic growth was only recently acquired (8). The large genome (~6,600 genes) is divided into two chromosomes and one significantly smaller megaplasmid (pHG1) (2). However, pHG1 encodes most of the proteins associated with lithoautotrophic growth and carries operons that extend metabolic versatility, such as nitrate respiration and aromatics degradation functions (9). Both pHG1 and chromosome 2 were likely acquired from other Burkholderia species by horizontal gene transfer, while chromosome 1 represents the evolutionarily most ancient DNA molecule (8). In a recent study on the C. necator proteome, we have shown that about 80% of its protein mass is encoded by chromosome 1 irrespective of the growth condition (10). We have also shown that the expression of autotrophic pathways during heterotrophic growth leads to suboptimal utilization of resources. One example is the costly expression of CBB cycle genes during glycolysis, which leads to reassimilation of emitted CO₂ (11, 12), a process that is unlikely to have a biomass yield or growth rate benefit (10). Another example is the redundancy of CBB cycle enzymes, which are encoded by two largely identical cbb operons on pHG1 and chromosome 2. Knocking out genes from either copy had no negative effect on autotrophic growth. The second cbb copy is therefore completely dispensable (10, 13).

What remains less explored is C. necator’s energy metabolism reviewed by Cramm, 2009 (5). It prefers organic acids over sugars (degraded via the Entner-Doudoroff pathway) to produce ATP during heterotrophic growth under respiratory conditions. The bacterium is also able to obtain energy from hydrogen oxidation via two remarkable O₂-tolerant metalloenzymes, a soluble NAD⁺-reducing and a membrane-bound [NiFe]-hydrogenase (SH and MBH, respectively) (14). The maturation of hydrogenases requires dedicated accessory proteins (encoded by the hyp genes) that synthesize and insert the Ni-Fe-(CN)₂-CO cofactor into the enzymes (15–17). The enzymes that accomplish formate oxidation, two soluble, NAD⁺-reducing [soluble formate dehydrogenase (SFDH)] and two membrane-bound formate dehydrogenases (MFDHs), also require accessory proteins for maturation (5, 18) and biosynthesis of the important molybdenum cofactor (MoCo) (19, 20).

The soluble (de-)hydrogenases transfer electrons directly to NAD+ generating reducing power in the form of NADH. Membrane-bound (de-)hydrogenases couple the oxidation reaction to the reduction of a universal quinone e⁻ carrier which in turn drives the electron transport chain (ETC) to build up a proton motive force. The electron transport chain of C. necator is highly complex (5). Whereas Escherichia coli has a total of three respiratory complexes, C. necator has no fewer than nine (three copies of bo₃ quinol oxidase, two copies of bd quinol oxidase, a bc₁ cytochrome reductase, and three different terminal cytochrome oxidases, bb₃, cbb₃, and aa₃) (5). These complexes are named after the different types of heme cofactors they contain and differ in their biochemical properties such as (i) type of the e⁻ donor and acceptor, (ii) high or low affinity for the terminal e⁻ acceptor, and (iii) the number of protons pumped across the cytoplasmic membrane. Little is known about the utilization of these complexes in different trophic conditions. In addition to using oxygen as the terminal e⁻ acceptor, C. necator also has the ability to respire nitrate, and other nitrogen oxides in anaerobic conditions. The genes encoding the four required enzyme complexes (nitrate, nitrite, nitric oxide, and nitrous oxide reductase) are located on pHG1 and expressed only when oxygen is limited (9, 21).

In this study, we investigated which enzymes of energy metabolism contribute most to cell fitness in different growth regimes including heterotrophic, lithoautotrophic, and formatotrophic growth. We focused on two areas of energy metabolism, first, the enzymes that regenerate reduced cofactors (NADH, quinol), and second, the electron transport chain that consumes the reduced cofactors to generate ATP. We wondered which of the different FDH and hydrogenase genes/operons are most utilized in C. necator. Which of the many accessory genes for FDH/hydrogenase maturation are truly required to produce functional enzymes? And what are the fitness costs associated with the expression of these pathways? To answer these questions, we employed a barcoded transposon library (22, 23) that was recently used to study the versatile carbon metabolism of C. necator (10). The mutant library was created by conjugating C. necator H16 with an E. coli donor strain, yielding around 60,000 individually barcoded transposon integration mutants [Fig. 1A (23)]. The barcodes were mapped to their respective genome integration site using the TnSeq workflow (24) (Fig. 1A). The enrichment or depletion of individual mutants was then tracked by next-generation sequencing (NGS) of the 20 nt barcode (Fig. 1B). We calculated a gene-wise fitness score from mutant abundances that represent the relative importance of a gene for a particular growth condition. The use of barcodes significantly increases the experimental throughput, which allowed us to probe a wide range of genes important for the various energy-generating pathways of C. necator.

Fig 1.

Creation of the C. necator transposon library and screening workflow. (A) C. necator H16 was conjugated with an E. coli donor strain carrying a transposon integration construct [sketch of the vector was adopted from reference (23)]. The construct contains two inverted repeat (IR) elements that serve as recognition sites for the transposase, an R6K plasmid replication origin, a kanamycin resistance cassette (kanR), two universal priming sites (U1 and U2), and the 20 nt random barcode (N20). Transposon integration sites were mapped to barcodes using the TnSeq workflow. Genomic DNA was isolated, shear-fragmented, and end-repaired. Transposon-containing fragments were PCR amplified including the barcode and a portion of genomic DNA and then subjected to NGS. (B) Library competition experiments were mostly performed in chemostat bioreactors. Samples were taken after 0, 8, and 16 generations, genomic DNA was extracted, and barcodes were sequenced using NGS. The changes in barcode abundance over time were used to calculate a fitness score for each gene.

RESULTS

Genes involved in energy metabolism dominate fitness screening

We probed the contribution of genes to fitness by growing the transposon mutant library in five different trophic conditions: with fructose, succinate, or formate as carbon and energy source, and O₂ as the terminal e⁻ acceptor, whereby formate is first oxidized to CO₂ which is then fixed via the CBB cycle. In addition, we grew the mutant library in anoxic conditions with fructose as carbon and energy source and nitrate as the terminal electron acceptor (nitrate respiration), and finally, in the presence of a gas mixture of CO₂, H₂, and O₂ (lithoautotrophy). Except for the latter, all cultivations were performed with two different feeding regimes, continuous feed (C) and pulsed feed (P), in order to select either for an advantage of growth rate or for biomass yield, respectively (25). Depletion or enrichment of mutants was tracked by barcode abundance, and a fitness score was calculated for each gene and growth condition (Fig. 1B). Our analysis revealed 354 genes with a fitness score of f ≤ −2 or f ≥ 2 in any condition, which we clustered by similarity (Fig. 2A, Fig. S1). The seven emerging clusters were each dominated by genes related to specific functional groups: genes in cluster 1 were almost exclusively related to the biosynthesis of amino acids and cofactors and depleted in all growth conditions (Fig. 2B and C). These genes became essential when the mutant library was transferred from the complete medium of pre-cultures to the minimal medium in bioreactor cultures. Genes in cluster 2 were depleted in formatotrophic growth and functionally enriched for formate dehydrogenases (classified in KEGG pathways as “methane” related, Fig. 2C) and electron transport complexes. Cluster 3 contained genes related to molybdenum cofactor biogenesis, and mutants were strongly depleted during nitrate respiration. The smallest cluster, number 4, contained almost exclusively genes related to H₂ metabolism, and its mutants were strongly depleted during lithoautotrophic growth but, surprisingly, enriched in all other conditions. The remaining clusters 5–7 were weaker in terms of differential fitness and more heterogeneous in terms of assigned functions (Fig. 2C). Overall, the initial analysis of fitness data showed that energy-related pathways were often conditionally essential, while the central carbon metabolism was less variable regarding fitness (Fig. S2). All trophic conditions have in common that they rely on flux through the lower section of Embden-Meyerhof-Parnas (EMP) glycolysis (both directions) and partially also the tricarboxylic acid (TCA) cycle. Many of the primary enzymes of central carbon metabolism are strictly essential; hence, no fitness score can be obtained for them (7, 10). We therefore focused our analysis on the three energy conversion-related groups formate dehydrogenases, hydrogenases, and the electron transport chain.

Fig 2.

Clustering of genes with significantly changed fitness. (A) Fitness scores for 354 genes exceeding a threshold f ≤ −2 or f ≥ 2. Genes were clustered based on similarity of fitness scores for the various growth conditions (dendrogram colored by cluster number). A cluster number of 7 was used after testing cluster separation by silhouette width, see Fig. S1. P, pulsed feeding; C, continuous feeding. (B) t-SNE dimensionality reduction of genes, color coded and labeled according to the clusters in A. (C) Gene enrichment for KEGG pathways. The top three pathways by P-value were selected. Colors and numbers in brackets correspond to the clusters in A.

Fitness related to formate dehydrogenases and their cofactors

The formate dehydrogenases and nitrate reductases (NAR) of C. necator are metalloenzymes that require MoCo, synthesized by a dedicated pathway (20). The MoCo molecule is a polycyclic aromatic compound, molybdopterin, which binds a single molybdate anion (MoO₄²⁻) via a dithiolene group (26). Four different steps are necessary to synthesize and insert MoCo into the apo-enzyme (26). First, the molybdopterin backbone is synthesized by cyclization of GTP to cGTP catalyzed by MoaA, conversion to pyranopterin by MoaC, and insertion of two sulfur residues by MoaD and MoaE (Fig. 3A). Only the genes moaD, moaE, and the MoCo-unrelated fluoride transporter moaF are organized in an operon structure on chromosome 1 (18). The moaA gene is located further downstream on chromosome 1 together with mobA and moeA2. We found that the moaACDE genes showed extremely low fitness scores and hence high importance for growth on formate and for nitrate respiration (Fig. 3A). Interestingly, moaA (H16_A2581) was only essential for nitrate respiration but not for formatotrophic growth. A second non-quantified gene copy, moaA2 (H16_B1466), probably compensates for the moaA knockout, suggesting that moaA is specific for GTP cyclization during nitrate respiration. The next step in MoCo biosynthesis is the import of the molybdate anion. In C. necator, a dedicated high-affinity membrane transport system encoded by modABC exists, but its knockout had no effect on fitness (Fig. 3A). An explanation for this might be the comparatively high concentration of MoO₄²⁻ in the growth medium, permitting uptake by other, non-specific transporters (27). The molybdate anion is then inserted into molybdopterin by MogA and MoeA (18), whereby MogA increases the affinity for molybdate at low concentrations without being strictly required (26). This was also reflected in the fitness scores: both mogA as well as moeA1 and moeA2 knockouts showed slightly but not dramatically reduced fitness, confirming the non-essentiality of mogA and suggesting that moeA2 can compensate for moeA1 knockout and vice versa. Knockout of a third moeA copy, moeA3, showed no effect on fitness. Our results correspond to a previous report stating that moeA1 becomes only essential when the other two moeA copies are inactivated (18). Finally, the basic MoCo can be further modified by the MobA-catalyzed attachment of GMP residues. The very low fitness score of mobA suggests that the GMP modification is strictly essential for FDH and nitrate reductase activity, a feature known from related E. coli enzymes of the DMSO reductase family to which C. necator’s FDHs belong (26). MobB, whose role in MoCo biosynthesis remains elusive, did not show any fitness effect for its three paralogs mobB, mobB2, and mobB3.

Fig 3.

Fitness related to formate dehydrogenases and their cofactors. (A) Fitness score for transposon insertion mutants of the MoCo biosynthesis pathway. (B) Fitness score for transposon insertion mutants of various formate dehydrogenase genes. A detailed description can be found in the Results section. The fitness of the fdoHI genes was not quantified. The scale bar indicates the gene fitness score after eight generations of growth. Inset numbers in genome plots represent the number of transposon insertions per gene. cGTP, cyclic GTP; cPMP, cyclic pyranopterin monophosphate; MPT, molybdopterin; MoCo, molybdopterin cofactor; GMP-MoCo, GMP modified MoCo; SFDH, soluble formate dehydrogenase; MFDH, membrane-bound formate dehydrogenase.

The genome of C. necator possesses four known operons encoding FDH subunits. The fdsGBACD operon encodes the subunits of a soluble, heteromultimeric NAD⁺-reducing FDH [SFDH (5)]. The fdsGBA and D mutants showed greatly reduced fitness in formatotrophic growth (Fig. 3B). FdsA, FdsB, and FdsG contain iron-sulfur clusters and FdsB also the molybdenum cofactor, which makes them likely to participate in electron transfer. FdsD does not bind a cofactor but may assist in maintaining the quaternary structure of the enzyme (28, 29). The role of FdsC remains unclear, but it was suggested that it transfers sulfur to MoCo or protects it from oxidation, prior to insertion into SFDH (30). The fdsC mutant showed moderately reduced fitness, suggesting that it is not strictly essential for growth on formate. However, since the essential fdsD gene is located downstream of fdsC, the reduced fitness for fdsC may also be caused by a negative effect on fdsD expression. Another soluble FDH (fdwAB) and two other MFDHs have been annotated but remain poorly functionally characterized (18). The primary MFDH is encoded by the fdhABCDE operon; a secondary three-subunit enzyme is presumably encoded by the fdoGHI operon (5). While the fds and fdh operons are located on chromosome 1, the fdw and fdo operons are located on chromosome 2, which is thought to have been acquired more recently during evolution (8, 10). None of the fdw, fdh, and fdo knockouts showed a significant effect on fitness (Fig. 3B), suggesting that the SFDH encoded by the fds operon carries by far the main load of formate oxidation.

Fitness related to hydrogenases and their cofactors

The [NiFe]-hydrogenases of C. necator are metallo-enzymes that require a nickel-iron-(CN)₂-CO cofactor for the reversible oxidation of H₂ into 2e⁻ and 2H⁺. All hydrogenase-related genes are located on the megaplasmid pHG1 (9). The metal cofactor is synthesized by a set of accessory Hyp proteins arranged in different operons on pHG1 (Fig. 4A, (31)). The largest operon, hypA1B1F1C1D1E1X, is located within an array of hox genes encoding the MBH (hoxKGZ) and the regulatory hydrogenase (RH, hoxABCJ) (9). Two more partial copies of the hyp gene cluster can be found approximately 40 kb downstream as part of an operon including the soluble NAD⁺-reducing hydrogenase (SH, hoxFUYHI). The function of the hyp gene products is well-understood. HypX generates carbon monoxide (CO) from formyl-THF (17), HypEF synthesizes the cyanide ligands from carbamoyl phosphate, HypCD serves as protein scaffold binding the central iron atom to which the CN and CO ligands are coordinated (16), and HypAB delivers the nickel ion for the active site. Fitness data for all hyp genes suggest that the primary, MBH-associated hyp operon is strictly essential for hydrogen assimilation (Fig. 4A), as knockout of these genes led to a dramatic decrease in lithoautotrophic growth. None of the alternative hyp genes were able to rescue a primary hyp gene knockout nor did they have any effect on fitness. However, the dramatic decrease in fitness could also be explained by the disruption of hoxA transcription, which is located directly downstream of hypX. HoxA is the major transcriptional regulator of hox/hyp gene expression (see detailed discussion below). Moreover, we observed an unexpected increase in fitness for all primary hyp genes in non-lithoautotrophic conditions. The growth benefit was consistently greater with continuous cultivation (primarily growth rate selective) but still evident with pulsed feed cultivation (primarily yield selective).

Fig 4.

Fitness related to hydrogenases and their cofactors. (A) Fitness score for the genes involved in biosynthesis of the Ni-Fe-(CN)₂-CO cofactor (hyp genes). (B) Fitness score for the genes forming MBH, soluble hydrogenase (SH), RH, and accessory genes for maturation of hydrogenases. For details, see the Results section. The scale bar indicates gene fitness scores after eight generations of growth. Inset numbers in genome plots represent the number of transposon insertions per gene. Asterisk: note that HoxJ is inactive in the H16 strain and cannot regulate HoxA activity. CN⁻, cyanide; Ni, nickel; Fe, iron; UQ, ubiquinone; UQH₂, ubiquinol; NDH, NADH dehydrogenase (complex I).

The gene clusters encoding the two metabolic hydrogenases, MBH (hoxKGZ) and SH (hoxFUYHI), are separated by a 75 kb stretch containing accessory hox and hyp genes (Fig. 4B). Inactivation of hoxKGZ and hoxFUYHI genes led to decreased fitness in lithoautotrophic conditions; however, fitness was not as strongly decreased as for the hyp operon. This indicates that the inactivation of either MBH or SH can be partially compensated for by the activity of the respective other hydrogenase. Mutation of the accessory genes hoxMLOQRTVW responsible for hydrogenase maturation also showed reduced fitness during lithoautotrophic growth on a similar scale as that of the structural hydrogenase subunits. The genes for a third, non-enzymatic hydrogenase (hoxBC) termed the RH are located between the MBH and SH operons (Fig. 4B, (32, 33)). The RH is the hydrogen sensor of a two-component transcriptional regulation system. In the absence of H₂, the histidine kinase HoxJ is phosphorylated and transfers its phosphate residue to HoxA, the master transcriptional regulator, turning it off. In the presence of H₂, RH binds HoxJ and leaves the non-phosphorylated HoxA in its active state. The C. necator H16 strain used in this study lost the ability to respond to H₂ due to a mutation in hoxJ, turning hyp/hox gene expression constantly on in energy-limited conditions (14, 32). Accordingly, hoxB, hoxC, and hoxJ genes were dispensable, while the knockout of hoxA led to a dramatic loss of fitness in lithoautotrophic conditions. On the other hand, the fitness of the hoxA mutant increased for all other growth conditions similar to hyp gene knockout, suggesting that unnecessary hydrogenase expression imposes a high metabolic cost. In addition to hoxA, there is a second level of transcriptional regulation for hydrogenases, the σ⁵⁴ sigma factor encoded by rpoN (Fig. S2). It de-represses hox genes depending on substrate quality. Preferred substrates such as succinate lead to low expression and less preferred substrates such as glycerol to high expression (34). The inactivation of rpoN led to a strong decrease in fitness during lithoautotrophic growth and an increase in fitness on fructose similar to the hoxA knockout. In contrast to hoxA, fitness on succinate and formate was slightly reduced, probably because rpoN inactivation led to loss of (unknown) σ⁵⁴-dependent functions important for growth under these conditions.

Fitness related to electron transport chain complexes

C. necator H16 is strictly dependent on respiration in order to generate sufficient ATP to support growth. Electrons stripped from a wealth of substrates are either transferred directly (e.g., from MBH and MFDH) to universal electron carriers such as ubiquinone (UQ), or indirectly via NADH (e.g. from SH and SFDH) and the activity of NADH dehydrogenase (respiratory complex I) to UQ (5). Electrons from UQ can then take different paths through the ETC, generating a proton-motive force across the cytoplasmic membrane that is used to generate ATP through the rotational movement of ATP synthase. Judging by the number of different complexes potentially involved in electron transport, the ETC of C. necator is both highly flexible and highly complex (Fig. 5). The structural and functional differences of ETC complexes are described in more detail elsewhere (5, 35).

Fig 5.

Fitness related to electron transport chain complexes. Fitness score for the genes encoding subunits of different respiratory complexes in the electron transport chain. For details, see the Results section. The scale bar indicates gene fitness scores after eight generations of growth. Inset numbers in genome plots represent the number of transposon insertions per gene. UQ, ubiquinone; UQH₂, ubiquinol; e⁻, electrons.

Here, we focus on basic functional properties such as electron acceptor affinity and contribution to the proton gradient. The direct route that electrons can take from the e⁻ donor ubiquinol (UQH₂) to the terminal e⁻ acceptor oxygen is through the quinol oxidase complexes bo₃ and bd. In C. necator, the bo₃ complex is encoded by three different cyoABC operons and the bd complex by two cydAB operons (Fig. 5). The bd family complexes do not pump protons but nevertheless contribute to the membrane potential by consuming two protons through the reduction of O₂ to H₂O at the cytoplasmic side of the membrane (35). Of the bo₃ and bd-related genes, none showed any significant effect on growth after inactivation. Although it is possible that duplicate genes compensate for the inactivation of the respective complexes, we assume that an overwhelming majority of ETC flux takes the alternative route via cytochrome reductase (bc₁ complex). C. necator has one copy of the bc₁ complex encoded by the qcrABC operon and inactivation of any of its genes led to significantly reduced fitness in all conditions except with nitrate as the e⁻ acceptor (Fig. 5). The fitness penalty was highest for formatotrophy and lowest for growth on fructose. The bc₁ complex, which is the most abundant ETC complex in C. necator (Fig. S3), transfers two e⁻ from UQH₂ to a cytochrome c carrier protein and pumps a total of four protons to the periplasmic side.

Cytochrome c is then oxidized by one of three complexes, bb₃, aa₃, or cbb₃, all of which transfer the e⁻ to O₂ to reduce it to water. Inactivation of the genes encoding the bb₃ complex (coxMNOPQ) did not show any effect on fitness. Inactivation of the genes encoding the aa₃ complex (ctaABCDEG) on the other hand led to moderately reduced fitness in fructose and succinate-driven growth, predominantly with pulsed feeding. The strongest effect on fitness was observed for the genes of the cbb₃ complex (ccoGNOP) during growth on formate, to a lesser extent for succinate, and not at all for fructose. This suggests that C. necator prefers the longer but more energy-efficient route via cytochrome reductase and oxidase (4 + 2 protons pumped per UQH₂), instead of the direct reduction of O₂ through quinol oxidases (2/4 protons pumped per UQH₂). The low O₂ affinity complex aa₃ was preferred in heterotrophic growth and the high O₂ affinity complex cbb₃ in formatotrophic growth. This result is supported by mass spectrometry data, where the most abundant aa₃ subunit (CtaC) was more abundant in fructose, while the most abundant cbb₃ subunit (CcoO) was more abundant in the formate condition (Fig. S3). We did not observe ETC complexes that showed a fitness loss specific for lithoautotrophy.

C. necator also has the ability to respire anaerobically using nitrate as the terminal electron acceptor. Cells were cultivated under anaerobic conditions (Fig. S4) with fructose or formate as carbon and energy sources (Fig. S5). We found that C. necator was able to grow on fructose with nitrate supplementation, but remarkably, growth on formate and nitrate was impossible (Fig. S5B). Next, we analyzed the effect on fitness from knockout of the four denitrification pathways using our transposon mutant library. The C. necator genome contains operons for enzymes catalyzing all four steps of denitrification from nitrate (NO₃⁻) via nitrite (NO₂⁻), nitric oxide (NO), and nitrous oxide (N₂O) to dinitrogen (N₂), all located on chromosome 2 or pHG1 (5). Surprisingly, we found no change in fitness for denitrification-related genes (Fig. S6), regardless of whether nitrate was respired or not. The only two exceptions were the regulatory genes narX2 and narL2, whose mutants had slightly reduced fitness in one nitrate respiration condition. This result is puzzling as it is known that the nar, nir, nor, and nos operons are strongly upregulated during nitrate respiration, and the intermediary products of each denitrification step can be detected in C. necator (21). We assume that gene knockouts were compensated for by iso-enzymes of the denitrification pathway, thus rescuing the mutants.

Protein cost explains the growth advantage of hydrogenase mutants

A surprising result was the positive fitness score of a range of hydrogenase mutants (Fig. 4). We estimated the growth advantage of hox and hyp mutants based on the change of NGS read counts over time (Fig. 6A; Materials and methods). We found that inactivation of the accessory hypABCDE and hypX genes as well as inactivation of the master regulator hoxA led to an estimated increase in growth rate from 0.1 h⁻¹ for the average population to 0.12–0.14 h⁻¹ for the mutants. Such a pronounced increase in growth rate implies the removal of a considerable burden associated with hydrogenase-related genes. We therefore investigated two scenarios, the protein cost hypothesis and the metabolic cost hypothesis, which could explain this observation.

Fig 6.

Protein cost explains the growth advantage of hydrogenase mutants. (A) Estimated growth rate µ of hydrogenase-related transposon mutants. The growth rate was calculated from the log₂ fold change in mutant abundance over time compared to the population average. The dashed line marks the nominal growth rate for bioreactor experiments. (B) Corresponding protein cost for the knockout mutants in A in percent of the total protein mass. Open symbols, individual cost. Closed symbols, protein cost when disruption of downstream gene expression is taken into account. (C) Organization of the various hydrogenase-related genes. Membrane-bound hydrogenase MBH (purple), accessory hyp genes (orange), and regulatory hox genes (blue) are located sequentially on megaplasmid pHG1. Genes for soluble hydrogenase (SH) are located around 60 kb downstream of the MBH operon. Expression of the promoters P_MBH and P_SH is controlled by the HoxA master regulator. Inset table: effect on transcription of the selected gene knockouts. It is assumed that knockout of any of the primary hyp genes (orange) also disrupts hoxA expression. (D) Biomass yield of selected, in-frame knockout mutants for batch cultures in fructose-supplemented minimal medium. Stars indicate the significance level of a two-sided t-test of each mutant against wild type. *, P ≤ 0.05. **, P ≤ 0.01. ***, P ≤ 0.001. (E) Growth rate determined from optical density measurement for the same cultures as in D. (F) Gas chromatography measurement in the headspace of batch cultures of wild-type C. necator H16 (green), and a mutant lacking all known hydrogenases (∆hoxG, ∆hoxH, ∆hoxC, and ∆hofG, in red). Open symbols, O₂ measurement. Closed symbols, H₂ measurement. All subfigures: points and bars represent the mean of at least four biological replicates except (F) where three replicates were used. Error bars represent SD.

The protein cost hypothesis states that a bacterium is fundamentally limited in its capacity to synthesize proteins (10, 36–38). As a consequence, the utilization of proteins becomes important to optimize growth rate, and every idle enzyme constitutes an additional cost. The knockout of an unutilized enzyme leads to a reduction of the protein cost. In order to estimate the individual protein cost of hydrogenase-related genes, we determined the protein mass fraction for each Hox/Hyp protein from mass spectrometry data acquired in a previous study (Fig. 6B, (10)). We then summed up the protein cost for the inactivated gene and all downstream affected genes according to their genomic arrangement and known transcripts, yielding the total cost (Fig. 6B and C). For example, the inactivation of hoxK leads to a low reduction of protein cost because it constitutes only the sum of the protein mass for the lowly expressed hoxK to hoxV genes. The inactivation of hoxF disrupts the expression of all SH genes, yielding an intermediate reduction of protein cost. Inactivation of hoxA on the other hand disrupts expression of all hox and hyp operons, resulting in a strong reduction of protein cost. Knockout of the primary hyp genes had the same effect as a hoxA knockout, from which we conclude that hoxA expression is dependent on the secondary P_HypA or P_HypB promoters (32). The pattern of total protein cost reduction correlated well with the estimated increase in growth rate (Fig. 6A and B; Fig. S7).

To validate the observed fitness advantage, we determined the biomass yield and growth rate of selected in-frame knockout mutants. The preparation of additional knockout mutants was necessary because transposon mutants cannot be easily extracted from our library and to exclude that the observed phenotype of our mutants originates from disruption of downstream gene expression (false-positive) or incomplete disruption of a gene by the transposon (false-negative). We generated mutants representative for MBH (hoxG), SH (hoxH), accessory hyp genes (hypB and hypX), and the master transcriptional regulator (hoxA). The mutants were grown in separate bioreactor batch cultures with fructose as the carbon source, such that hydrogenase expression constitutes a growth burden. Unlike transposon insertions, these knockouts are not expected to affect the expression of downstream genes. We therefore expected the highest increase in biomass yield or growth rate for the hoxA mutant, a moderate increase for hoxH, and little to no increase for the lowly expressed hoxG, hypB, and hypX mutants. Indeed, the biomass yield was highest for the hoxA deletion, followed by hoxH and hypX, and lower than or equal to wild type for hoxG and hypB (Fig. 6D). The maximum growth rate determined from OD₇₂₀ readings showed an overall similar result (Fig. 6E), with the exception of hoxA, which showed a lower growth rate than expected. However, we observed a shorter lag phase for hoxA, hypB, and hypX mutants, which could also represent a growth advantage (Fig. S8).

Finally, we tested the metabolic cost hypothesis. In heterotrophic conditions, hydrogenases could catalyze the reverse reaction of H₂ oxidation—the formation of H₂ from protons and an electron donor such as NADH (2H⁺ + 2e⁻ → H₂), which could drain the NAD(P)H pool. The dissipation of excess reducing power in the form of NADH is, for example, known from other microbial phyla such as cyanobacteria (39, 40). It has been reported that C. necator produces H₂ when the terminal electron acceptor, oxygen, or one of the nitrogen oxides is limiting and NADH accumulates (41). In order to determine if C. necator wastes reducing power through the action of hydrogenases, the wild-type H16 strain and a deletion mutant defective in all hydrogenase activity (∆hoxG, ∆hoxH, ∆hoxC, and ∆hofG) were cultured in batch with fructose and glycerol to maximize hydrogenase expression (Fig. 6F). No hydrogen evolution was detectable unless oxygen was severely limited. Hydrogen was only detected for the wild-type strain at the last probed time point (20.5 h) at a concentration of around 1,300 ppm [0.13% (vol/vol)]. Such oxygen-depleted conditions were not present in our bioreactor experiments, ruling out the metabolic cost hypothesis and strengthening the protein cost hypothesis.

DISCUSSION

C. necator has attracted interest not only for its ability to produce the storage polymer PHB in high concentrations but also for its extremely versatile carbon and energy metabolism. Many of the initial groundbreaking studies identifying the key enzymes for C. necator's remarkable metabolic capabilities date back to the 1980s and 1990s (28, 42–44). Another important milestone was the publication of the full genome sequence (2), which made it possible to compile a detailed list of all theoretically present energy acquisition pathways (5). However, since then most studies on C. necator have focused on enhancing PHB production or engineering pathways for biosynthesis of other compounds replacing PHB (45–48). Our study, on the other hand, attempts to gain a holistic view of C. necator’s energy metabolism.

We have recently studied the carbon metabolism of C. necator from the perspective of resource allocation in different trophic conditions (10). To this end, a barcoded transposon library was created to quantify gene essentiality in the central carbon metabolism, particularly the Calvin cycle which is used for CO₂ fixation. Here, we exploited our transposon library to probe the energy metabolism of C. necator. Depletion or enrichment of individual strains thereby revealed the effect of inactivation of a gene on growth rate and biomass yield, which we summarize with the term “fitness contribution.” Since our data were particularly enriched for specific clusters of genes that affect fitness for formate and hydrogen assimilation as well as the molybdenum cofactor assembly (Fig. 2B), our study focused on these pathways. Formatotrophic and lithoautotrophic growth can be regarded as secondary trophic modes that extend the metabolic capabilities of C. necator beyond heterotrophy. The genes encoding these pathways were most likely acquired by horizontal gene transfer (8), and they fully overlap regarding carbon fixation. This leads to the interesting situation that C. necator possesses up to 3 iso-enzymes for central reactions of EMP glycolysis and the CBB cycle, one “original” and two additional copies as part of the cbb operons (Fig. S2). The former are often strictly essential and hence impossible to study with our transposon library, while the latter two showed no fitness effect due to the high level of redundancy (compensation). From a metabolic perspective, the only differences we found between formatotrophic and lithoautotrophic growth were the enzymes for energy metabolism, which are discussed in the following sections.

The MoCo is essential for the function of the two structurally and functionally related enzymes formate dehydrogenase and nitrate reductase (5). Our results confirmed the essentiality of various enzymes responsible for MoCo biosynthesis. While most enzymes were essential for both formate assimilation and nitrate respiration, moaA was only essential for the latter. A second copy, moaA2, can probably compensate for this knockout specifically during formatotrophic growth. A set of genes responsible for molybdate insertion into molybdopterin (mogA, moeA1, and moeA2) led to slightly reduced but not completely abolished growth. This confirms that mogA is beneficial but not essential (26), and both moeA copies are functional and can partially compensate for the knockout of the respective other copy. A third copy, moeA3, was hypothesized to play the same role (5), but our fitness data indicate that this copy is fully dispensable in the conditions tested.

The molybdenum cofactor is an essential component of C. necator’s two types of formate dehydrogenases, SFDH and MFDH. The molecular structure and enzymatic mechanism of the main SFDH (fds operon) have been studied in great detail (19, 42, 49). Much less is known about the two MFDH enzymes and the second copy of the SFDH, which share only limited similarity on gene level. It is unknown to which extent these formate dehydrogenases are active during formatotrophic growth and whether all or some of them carry metabolic flux. Fitness data from our library experiments clearly suggest that the SFDH encoded by the fds operon is the enzyme carrying the main metabolic flux of formate oxidation. All fds mutants had strongly reduced fitness, while all other FDH mutants did not. In library screenings, we often see a partial decrease in fitness when multiple iso-enzymes with similar activity share the metabolic load (50), but the FDH genes did not show such a “division of labor.” This also means that the overwhelming majority of electron flux from formate oxidation is channeled toward NADH and only indirectly toward ubiquinone/ubiquinol in the electron transport chain.

The fitness screening for hydrogenase genes—unlike formate dehydrogenase—suggests that both soluble and membrane-bound hydrogenase are equally utilized for hydrogen oxidation. Mutants for the SH genes hoxFUYHI showed moderately reduced fitness scores, similar to the fitness from hoxKGZ, while fitness scores for the hoxA knockout, which disrupts all hox gene expression, were around twice as low. Knockout of the secondary regulator σ⁵⁴ led to a similar outcome, but fitness changes were more nuanced in line with the known substrate-specific regulation (34). Reminiscent of hoxA was the strongly reduced fitness during lithoautotrophic growth and the increased fitness on fructose, where the missing activation of hydrogenase expression presumably confers a growth advantage. Fitness was slightly reduced on succinate and formate, probably due to the loss of other σ⁵⁴-related functions.

The C. necator hydrogenases have been of interest for their oxygen tolerance and have been structurally and functionally well characterized (31, 51). However, less is known about their physiological relevance in different growth conditions. It has been shown that hydrogenase expression is upregulated when the preferred carbon source (e.g., fructose and succinate) is depleted in the culture medium, and expression levels of both hydrogenases are comparable [MBH around 60% of SH (32, 34, 52)]. However, the physiological importance in terms of metabolic flux has not yet been quantified. Our results indicate that SH and MBH most likely have a similar contribution in carrying the metabolic flux from H₂ to reduction equivalents. The SH and MBH enzymes require accessory proteins for the biosynthesis of cofactors (hyp genes) and maturation (hoxMLOQRTVW). Of the three hyp operons on pHG1, only mutants of the primary hyp1 operon showed a reduction in fitness. The strong phenotype was remarkably similar to that of the hoxA mutant, suggesting that the knockout of these genes disrupts the downstream hoxA expression and leads to a complete loss of H₂ oxidation capacity. The hydrogenase maturation genes were all essential for lithoautotrophic growth with the exception of hoxN/N2, which encodes nickel transporters (53) that are most likely not important when the metal concentration in the environment is sufficient.

In formatotrophic or lithoautotrophic lifestyle, redox power generated from formate or H₂ oxidation can take two different routes to fuel the energy requirements of C. necator. The NADH from soluble FDH/SH can reduce ubiquinone (UQ) via the NADH dehydrogenase (NDH, also known as respiratory complex I). All subunits of NDH are fully essential, as the library contained no transposon mutants of the nuo operon encoding them. The second route is the direct e⁻ transfer to UQ. The reduced UQH₂ can then be used as a substrate to pump protons by at least three ETC complexes. We found that C. necator prefers the route via cytochrome c reductase (qcrABC), which is 50% more energy efficient than quinol oxidases in terms of protons pumped (6 vs 2/4), but more complex and therefore likely associated with a higher protein cost. We also found that, of the three annotated terminal cytochrome c oxidases, the cbb₃ complex is preferred during formatotrophic growth and the aa₃ complex for growth on fructose. We note that ETC complexes have different affinity for oxygen, but we only cultivated the library with unlimited O₂ supply (except for nitrate respiration). None of the complexes was clearly essential for lithoautotrophic growth.

When oxygen is replaced by nitrogen oxides, a completely different set of terminal oxidases is used for anaerobic respiration—the dedicated nitrate, nitrite, nitric, and nitrous oxide reductase complexes (5, 21). We grew the transposon library in denitrifying conditions and verified that nitrate supplementation was strictly required for growth. Yet, we found none of the annotated terminal reductases to be essential. A combination of the following two hypotheses could explain this surprising result: (i) inactivation of individual denitrification genes can be compensated by iso-enzymes. For example, a knockout of the primary NAR could be compensated for by the function of either of the two secondary reductases NAP and NAS, although the latter is most likely not expressed in the presence of the nitrogen source we used, ammonium chloride (54). (ii) Nitrate was present at non-limiting concentration, such that C. necator was not dependent on the entire denitrification pathway from nitrate to N₂, making downstream denitrification genes non-essential. However, the genes for biosynthesis of the molybdenum cofactor were strictly essential for denitrification (Fig. 2A). This cofactor is an essential part not only of FDH, but also of all three nitrate reductases (NAR, NAP, and NAS), implying that one or more of these enzymes were actively used. We noted that nitrate respiration was impaired when cells were grown on formate instead of fructose (Fig. S5). It has previously been reported that C. necator cannot respire nitrate when grown lithoautotrophically (55), and this seems to apply to formate as well. Our results did not provide any hints as to why litho- and formato-autotrophic growth is incompatible with nitrate respiration.

Fitness data for lithoautotrophic growth surprisingly revealed that inactivation of several hydrogenase-related genes boosted growth in heterotrophic conditions. We tested two hypotheses, hydrogenases causing metabolic cost by wasteful conversion of NADH to H₂ or protein cost by production of enzymes (and cofactors). We found that the increase in growth rate correlated well with the estimated protein mass of hydrogenases and accessory proteins. Moreover, no emission of molecular hydrogen was detected in heterotrophic growth that would support the metabolic cost hypothesis. The expression of unutilized hydrogenases therefore constitutes a severe growth burden for C. necator, at least for strain H16, which was reported to be defective in H₂-dependent regulation of hox/hyp gene expression (56). A recent study identified targets for rational engineering of C. necator toward faster growth on formate and reached similar conclusions (57). Almost all of the spontaneous mutations inactivated either SH, MBH, or the CBB cycle enzymes, whose under-utilization we have described previously (10). The complete removal of the pHG1 megaplasmid encoding all hydrogenases led to one of the biggest improvements in growth rate (57). The increase in the maximum growth rate of around 20% for ∆pHG1 corresponds well to the increase of yield or growth rate we observed for selected hox/hyp mutants but even more so for the growth rate estimated from library experiments (Fig. 6). Altogether, our results are in line with the findings from Calvey et al. and mount more evidence for the importance of protein cost in C. necator. The tight regulation of hydrogenase expression in strains with functional hoxJ signifies the high protein cost associated with litho-autotrophic lifestyle. The detrimental effect of a protein burden on bacterial growth is well-studied in model bacteria (58 ,36, 59, 60). Microbes nevertheless employ the strategy of keeping enzyme reserves in order to cope with fluctuating environments and stress (59), and C. necator appears to be a prime example of this generalistic lifestyle (10). In conclusion, the genome-wide fitness screening in this study revealed the substrate-specific utilization of energy-generating pathways. Our results show in an unbiased manner, which of the various iso-enzymes for hydrogenases, formate dehydrogenases, and ETC complexes are most important. For other targets like the denitrification pathway, it was difficult to obtain results with our single knockout library, most likely due to functional redundancy. This can be a valuable target for future studies with combinatorial knockout libraries.

MATERIALS AND METHODS

Strains

C. necator H16 was obtained from the German Collection of Microorganisms and Cell Cultures, strain number DSM-428. Strains carrying individual in-frame deletions for hoxG (61), hypB (62), hypX (63), hoxA, and hoxH (61) have been described previously. The negative control strain lacking all four hydrogenases (AH, MBH, RH, and SH) was created by serial in-frame deletions of the respective hydrogenase subunit genes. C. necator HF364, which already carried deletions in the large subunit genes of the MBH, RH, and SH, served as a starting point (64). In this strain, the native AH promoter was already replaced by the MBH promoter. A PCR fragment was amplified with primers 1 and 2 (Table S2), using the megaplasmid pHG1 as a template. The fragment contained the AH large subunit gene hofG as well as flanking regions. The fragment was digested with XbaI and subsequently ligated into XbaI-cut pLO1 (Lenz et al., 1994), resulting in the plasmid pGE815. Next, a 1,131 bp fragment was removed from the hofG gene by restriction using SexAI and subsequent religation, resulting in plasmid pGE816. Double homologous recombination using the suicide vector pGE816 facilitated the incorporation of the hofG deletion into C. necator HF364, yielding the final ∆hoxG, ∆hoxH, ∆hoxC, and ∆hofG deletion.

Creation of barcoded C. necator transposon library

The creation of the barcoded transposon knockout library has been described in detail in references (10) and (23). Briefly, C. necator H16 wild-type was conjugated with an E. coli APA766 donor strain containing a barcoded transposon library. The strain is auxotrophic for DAP, the L-Lysin precursor 2,6-diamino-pimelate, to allow for counter selection. Overnight cultures of E. coli APA766 and C. necator H16 were prepared. The APA766 culture was supplemented with 0.4 mM DAP and 50 µg/mL kanamycin. Fresh cultures were prepared from precultures and then harvested during the exponential growth phase by centrifugation for 10 min, 5,000 × g, RT. Supernatant was discarded, and cell pellets were resuspended in residual liquid, transferred to 2 mL tubes, washed twice with 2 mL phosphate buffered saline (PBS), and finally resuspended in a total amount of 500 µL PBS. Cell suspensions from both strains were combined and plated on 25 cm × 25 cm large trays with Luria-Bertani (LB) agar supplemented with 0.4 mM DAP. For conjugation, plates were incubated overnight at 30°C. All cells were then harvested from mating plates by rinsing with PBS and plating on selection plates with LB agar supplemented with 100 µg/mL kanamycin, without DAP. After colonies of sufficient size appeared, transformants were harvested by scraping all cell mass from the plate and collecting the pooled scrapings in 1.5 mL tubes. The mutant library was diluted 10-fold and then immediately frozen at −80°C. For competition experiments, a 1 mL 10-fold diluted aliquot (pool of all conjugations, ~1 M CFU) was used to inoculate pre-cultures. The identification of mutant insertion sites (“transposon mapping”) is described in detail in reference (10).

Growth medium

Strains were cultivated on complete (LB) medium or minimal medium depending on experimental setup. Minimal medium was composed of 0.78 g/L NaH₂PO₄, 4.18 g/L Na₂HPO₄ × 2H₂O, 1 g/L NH₄Cl, 0.1 g/L K₂SO₄, 0.1 g/L MgCl₂ × 6H₂O, 1.6 mg/L FeCl₃ × 6H₂O, 0.4 mg/L CaCl₂, 0.05 mg/L CoCl₂ × 6H₂O, 1.8 mg/L Na₂MoO₄ × 2H₂O, 0.13 g/L Ni₂SO₄ × 6H₂O, and 0.07 mg/L CuCl₂ × 2H₂O. All components were added to autoclaved sodium phosphate buffer from filter-sterilized stock solutions. Different compounds were added as carbon and energy sources depending on the experimental setup (Table S1). Standard batch cultures were grown in a volume of 10–20 mL medium in 100 mL shake flasks at 30°C and 180 rpm. Precultures of the barcoded C. necator transposon library were supplemented with 200 µg/mL kanamycin and 50 µg/mL gentamicin to suppress the growth of untransformed C. necator recipient or E. coli donor cells.

Pressure bottle cultivation

Pressure bottles were used for batch cultivation in lithoautotrophic or nitrate respiration conditions. For lithoautotrophic growth, precultures of the C. necator transposon library were grown overnight in 10 mL LB medium and then transferred to 20 mL of minimal medium containing 2 g/L fructose. The precultures were used to inoculate 100 mL Duran Pressure Plus bottles containing 30 mL of minimal medium. Samples of the precultures were immediately frozen at −20°C for later use as T0 samples. A GB100 Plus gas mixer (MCQ instruments, IT) paired with a GW-6400–2 anaerobe gas exchange system (G.R. instruments, NL) was used to flush and then fill the bottles to a final pressure of 1 bar above standard pressure, with a mixture of 70% H₂, 15% O₂, and 15% CO₂. These culture bottles were shaken at 30°C, 150 rpm. When the cultures approached an OD₆₀₀ of 1.0, they were diluted with minimal medium to OD₆₀₀ of 0.1, and the gas was replenished to ensure continuous growth occurred until the end point of 12 or 16 generations of growth was reached. Samples were taken at 4, 8, 12, and 16 generations. For nitrate respiration, the same protocol was used except that the minimal medium was supplemented with 2 g/L fructose and 1 g/L NaNO₃. The headspace of bottles was flushed and then pressurized to 1 bar above standard pressure with 100% N₂.

Bioreactor cultivations

For chemostat experiments, the C. necator H16 transposon mutant library was grown as described previously (10). Briefly, an 8-tube MC-1000-OD bioreactor (Photon System Instruments, CZ) was customized to perform chemostat cultivation (36, 65). Bioreactors (65 mL) were filled with minimal medium supplemented with the respective carbon and nitrogen source and inoculated with an overnight preculture to a target OD_720nm of 0.05. Bioreactors were bubbled with air at a rate of 12.5 mL/min and a temperature of 30°C. The OD_720nm and OD_680nm were measured every 15 min. For chemostat mode, fresh medium was continuously added using Reglo ICC precision peristaltic pumps (Ismatec, DE). For turbidostat mode, the target OD₇₂₀ was set to 0.5 for fructose or 0.2 for formate cultures, and 4 mL medium (6.2% of total volume) was added as soon as the set point was reached. For library competition experiments, 15 mL samples were taken after 0, 8, and 16 generations of growth (population average). Cells were harvested by centrifugation for 10 min at 5,000 × g, 4°C, washed with 1 mL ice-cold PBS, transferred to a 1.5 mL tube, and centrifuged again for 2 min at 8,000 × g, 4°C. The supernatant was discarded, and the pellet was frozen at −20°C. For batch cultures in the bioreactor, WT or mutant strains were grown using the same setup but with 50 mL volume instead of 65 mL. The entire culture was harvested after the stationary phase was reached to determine the biomass yield.

Biomass yield and growth rate

To determine biomass yield, 50 mL cell suspension was harvested by centrifugation for 10 min, 5,000 × g, 4°C. The pellet was washed twice with 1 mL mqH₂O, transferred to preweighed 1.5 mL tubes, and dried for 4 h at 70°C or 55°C overnight. Dried cell mass was measured on a precision scale, and yield was calculated according to the equation: yield (gDCW/g substrate) = DCW (mg)/[substrate (g/L) × volume (L) × 1,000]. Growth rate was calculated by applying a sliding window of 5 h length to the log OD₇₂₀ over time and fitting a linear model to each window. The slope of the fitted model was the growth rate for the respective window. From all slopes over time, the maximum was selected as the maximum growth rate.

Hydrogen evolution experiment

Precultures of the H16 wild-type and the hydrogenase negative strain (∆hoxG, ∆hoxH, ∆hoxC, and ∆hofG) were grown in FN medium (14) containing 0.4% fructose for 48 h at 37°C and 120 rpm. Main cultures were grown overnight in FN medium supplemented with 2 g/L fructose and 2 g/L glycerol in baffled Erlenmeyer flasks (filled to 20% of the nominal volume) at 30°C and 120 rpm until an optical density at 436 nm (OD₄₃₆) of approximately 6–8 was reached. The OD₄₃₆ of the culture was adjusted to 6.5 using H16 buffer (C-free minimal medium). The cultures were transferred into 120 mL serum bottles (total bottle volume approx. 147 mL). The bottles were filled with 74 mL of the main culture, and thus, the gas space was 73 mL, and the bottles were subsequently closed airtight with a rubber septum. The bottles were incubated at 30°C and 120 rpm, and the gas composition was regularly analyzed by drawing 1 mL of gas from the bottles per measurement using a gas-tight 1 mL Hamilton syringe. The gas chromatography measurements followed published protocols (64).

Gene fitness analysis (BarSeq)

Frozen cell pellets from the pulsed and continuous competition experiments were resuspended in 100 µL of 10 mM Tris, and genomic DNA was extracted from 10 µL of the resuspension using a GeneJet Genomic DNA Purification Kit (ThermoScientific, US). Amplification of barcodes from genomic DNA was conducted using one of the custom forward indexing primers (BarSeq_F_i7, Table S2) and the reverse phasing primer (BarSeq_R_P2_UMI). For each sample, 9 µL of genomic DNA extract (≥10 ng/µL) was combined with 3 µL of a forward indexing primer (100 nM), 3 µL of the reverse phasing primer pool (100 nM), and 15 µL of Q5 Mastermix (New England Biolabs, US). Cycle conditions were 4 min at 98°C followed by 20× (30 seconds at 98°C, 30 seconds at 68°C, and 30 seconds at 72°C) with a final extension of 5 min at 72°C. Concentration of each sample was quantified using a Qubit dsDNA HS Assay Kit (Invitrogen, US). Samples were then pooled with 40 ng from up to 36 different samples being combined and run on a 1% agarose gel. Gel extraction was performed on the thick band centered around 200 bp, and the concentration of the purified pooled library was quantified again via Qubit assay and diluted to 2 nM. The 2 nM library was then diluted, denatured, and sequenced using a NextSeq 500/550 High Output Kit v2.5 (75 Cycles, Illumina, US) run on an Illumina NextSeq 550 instrument according to the manufacturer’s instructions. Library loading concentration was 1.8 pM with a 1% phiX spike. Calculation of gene fitness was performed using the rebar pipeline (https://github.com/m-jahn/rebar (10), which trims and filters reads, extracts barcodes, and summarizes read counts per barcode. Fitness score calculation based on the log₂ fold change of read count per barcode over time was implemented as an R script as part of the pipeline.

Statistical analysis

Bioreactor cultivations, LC-MS/MS measurement for proteomics, and library competition experiments (“BarSeq”) were performed with four independent biological replicates. Here, biological replicate means that samples were obtained from independently replicated bottle or bioreactor cultures inoculated with the same preculture. If not otherwise indicated in figure legends, points and error bars represent the mean and SD. No removal of outliers was performed. All analyses of fitness data, cultivations, and proteomics results are documented in R markdown notebooks available at https://github.com/m-jahn/R-notebook-ralstonia-energy. A significance threshold of |f| ≥ 2 after at least eight generations was chosen based on the bulk fitness distribution of mutants. Clustering of genes was performed based on the similarity of fitness scores using the R function hclust() from package stats with method ward.D2. The optimal cluster number was determined using silhouette width analysis with the function silhouette() from R package cluster. T-SNE dimensionality reduction of genes was performed using R package tsne. Gene enrichment for KEGG pathways was performed by using the function kegga() from R package limma. The R statistical language was used with version 4.3.2.

DATA AVAILABILITY

Mass spectrometry proteomics data were obtained from the PRIDE repository with the data set identifier PXD024819. Protein quantification results can be browsed and interactively analyzed using the web application available at https://m-jahn.shinyapps.io/ShinyProt. Sequencing data for BarSeq experiments are available at the European Nucleotide Archive with accession number PRJEB43757. The data for competition experiments performed with the transposon mutant library can be browsed and interactively analyzed using the web application available at https://m-jahn.shinyapps.io/ShinyLib/. All analyses of fitness data, cultivations, and proteomics results are documented in R notebooks available at https://github.com/m-jahn/R-notebook-ralstonia-energy.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/aem.00748-24.

Supplemental material. aem.00748-24-s0001.pdf.

Figures S1 to S8; Tables S1 and S2.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.