H₂-dependent reduction of rubredoxin by FrhAGB hydrogenase in Thermococcus onnurineus NA1

Abstract

Abstract

The oxidative stress defense mechanism in the order Thermococcales functions through an electron-transfer cascade that transfers electrons from NAD(P)H to various antioxidant enzymes, utilizing rubredoxin (Rd) as the electron carrier. Thermococcus onnurineus NA1 possesses a unique soluble hydrogenase encoded by the frhAGB genes. This hydrogenase has been reported to mediate direct electron transfer from H₂ oxidation through protein–protein interactions. In this study, it was found that the FrhAGB hydrogenase can directly reduce Rd using H₂ as an electron source. Additionally, its subcomplex, FrhAG, also reduced Rd. Binding affinity, as measured by surface plasmon resonance, showed strong interactions between both FrhAGB and FrhAG with Rd, with similar binding constants. In the structural model predicted by AlphaFold 3, the [4Fe-4S] clusters in FrhG and the mononuclear Fe center in Rd are arranged at distances allowing for sequential electron transfer. This research identifies a novel pathway for supplying electrons to Rd and elucidates the role of FrhAGB hydrogenase in the oxidative stress defense system.

Key points

• Identification of direct electron transfer between FrhAGB hydrogenase and rubredoxin.

• Quantitative analysis of protein-protein interactions between FrhAGB and rubredoxin.

• The role of hydrogenase in a H ₂ -driven oxidative stress defense mechanism.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00253-026-13863-5.

Introduction

Hydrogenases are enzymes that catalyze the reversible oxidation of hydrogen gas (H₂ ↔ 2H⁺  + 2e⁻), playing a critical role in various biochemical processes. Hydrogenases are classified into [NiFe]-hydrogenases, [FeFe]-hydrogenases, and [Fe]-hydrogenases based on their catalytic metal center (Shima and Thauer 2007; Vignais et al. 2001). The [NiFe] hydrogenases can be categorized into five groups, membrane-bound uptake hydrogenases (Group 1), regulatory/sensor hydrogenases (Group 2), cytoplasmic/bidirectional hydrogenases (Group 3), energy-converting hydrogenases (Group 4), and Actinobacterial hydrogenases (Group 5), based on phylogenetic analysis of their large subunits, exhibiting distinct cellular locations and functions (Vignais and Billoud 2007; Constant et al. 2011). Hydrogenases employ diverse electron-transfer mechanisms and cofactor-reduction strategies to link hydrogen metabolism with cellular redox pathways. Some hydrogenases, particularly those with multimeric subunits, can reduce soluble cofactors such as NAD, NADP, or F₄₂₀ (8-hydroxy-5-deazaflavin). Additionally, certain hydrogenases are also able to transfer electrons to both NADP⁺ and another electron acceptor, like ferredoxin, through a process known as electron bifurcation (Schuchmann et al. 2018; Vignais and Colbeau 2004). Moreover, several hydrogenases have been characterized for their function in coupling H₂ oxidation to the reduction of cytochromes (type b or c) or methanophenazine, which serves as an electron carrier within the electron transport chain (Tietze et al. 2003).

The hyperthermophilic archaeon Thermococcus onnurineus NA1 has seven [NiFe]-hydrogenases, including three soluble hydrogenases (Frh, SulfI, and SulfII: group 3) and four membrane-bound hydrogenases (Mfh1, Mfh2, Mch, and Mbh: group 4), which have been well characterized in the order Thermococcales (Rittmann et al. 2015; Jenney and Adams 2008; McTernan et al. 2014; Kanai et al. 2011). The FrhAGB hydrogenase, encoded by the operon frhAGB (TON_1559-1561), is homologous to the F₄₂₀-reducing hydrogenases commonly found in methanogens (Cho et al. 2017; Hendrickson and Leigh 2008). However, it is noteworthy that the FrhB subunit of FrhAGB lacks a conserved F₄₂₀-binding site, indicating that it does not utilize the F₄₂₀ cofactor in its enzymatic action (Jeon et al. 2015). In T. onnurineus NA1, the FrhAGB enzyme is capable of transferring electrons derived from H₂ through direct interaction with multiple proteins. For example, while thioredoxin reductase (TrxR) predominantly utilizes NAD(P)H as an electron donor, TrxR can also accept electrons generated by H₂ oxidation, mediated by FrhAGB (Jung et al. 2020). In addition, FrhAGB has been demonstrated to directly transfer electrons from H₂ to three flavin adenine dinucleotide (FAD)-dependent oxidoreductases, TON_0542, TON_0702, and TON_1376, which ultimately leads to the reduction of NADP⁺ (Yang et al. 2023).

Previous studies revealed significant growth differences between the frhA-deletion mutant (ΔFrhA) and the frhAGB-overexpression mutant (FO) under exposure to oxygen (O₂) levels up to 20% (Lee et al. 2017, 2019). The wild-type strain showed a gradual decrease in cell density, reaching 65% at 10% O₂, compared with the value observed in the absence of O₂. When O₂ levels exceeded 10%, growth declined significantly, leading to a dramatic reduction in cell density to 28% at 16% O₂. The FO strain displayed enhanced growth compared to the wild-type strain, while the ΔFrhA strain showed a substantial decline in growth. Transcriptomic analysis indicated that the FO strain exhibited upregulation of expression for two antioxidant-related genes, peroxiredoxin (TON_0786, 2.5-fold increase), and alkyl hydroperoxide reductase subunit C (TON_0847, 4.1-fold increase). In contrast, the expression levels of these genes remained unchanged in the ΔFrhA strain. In the oxidative stress defense system within the order Thermococcales, the electron transfer cascade occurs sequentially as follows: NADP rubredoxin oxidoreductase (NROR) transfers electrons from NADPH to the electron carrier rubredoxin (Rd), which then donates them to rubrerythrin (Rbr), superoxide reductase (SOR), and flavodiiron proteins (FDPs) that detoxify O₂ and reactive oxygen species (ROS) (Thorgersen et al. 2012). Notably, there was no significant differential expression of the genes encoding these enzymes in either the ΔFrhA or FO strains compared to the wild-type strain (Lee et al. 2017, 2019).

To elucidate the mechanisms by which FrhAGB contributes to O₂ tolerance, our study aimed to explore the potential for direct electron transfer from H₂ to the oxidative stress defense system via FrhAGB. Our findings indicate that Rd reduction can occur through direct interaction with FrhAGB, independent of NROR and NAD(P)H. This discovery suggests a novel electron transfer pathway that may be crucial for effective oxidative stress defense.

Materials and methods

Strains, media, and culture conditions

T. onnurineus NA1 (KCTC 10859) was originally isolated from a deep-sea hydrothermal vent area (Bae et al. 2006). E. coli Rosetta (DE3) pLysS (Stratagene, La Jolla, CA, USA) served as the host for the recombinant plasmids. The basic medium for culturing T. onnurineus NA1 was MM1, which contains 10 g L⁻¹ of sodium formate and has an initial pH of 6.5 (Kim et al. 2010). When culturing T. onnurineus NA1 cells that overexpress the frhAGB or frhAG genes, MM1 medium was supplemented with an increased yeast extract concentration of 5 g L⁻¹. For E. coli, Luria-Bertani medium was used. The culture conditions for both T. onnurineus NA1 and E. coli are detailed in previous studies (Choi et al. 2016; Jung et al. 2020).

Cloning of NROR and Rd

The genes TON_0865 and TON_0867, which encode the proteins NROR and Rd, respectively, were amplified from the genomic DNA of T. onnurineus NA1 using specific primers. For TON_0865, the following primers were used: 5′-GGA CAG CAA ATG GGT CGC GGA TCC GTG AGG GTT GTA ATA GTC GGC G-3′ and 5′-GTG GTG GTG GTG GTG GTG CTC GAG GTC AAT AAG CAG TCC TTC TCT G-3′. For TON_0867, the following primers were used: 5′-GGA CAG CAA ATG GGT CGC GGA TCC ATG GCA AAG TGG AAA TGT ATA GTT TGT GG-3′ and 5′-GTG GTG GTG GTG GTG GTG CTC GAG CTC TAT CTT TTC AAA CAT ATC CTT GGG G-3′. These amplified genes were then inserted into the pET-21a(+) vector (Merck KGaA, Darmstadt, Germany). Each recombinant plasmid was subsequently transformed into E. coli Rosetta (DE3) pLysS, and the appropriate clones containing the plasmid were then identified and selected.

Protein purification

NROR and Rd proteins were purified from the recombinant E. coli Rosetta (DE3) pLysS strain, as previously described (Choi et al. 2016). The cell pellets obtained were resuspended in an anaerobically prepared buffer containing 100 mM Tris-HCl (pH 8.0), 150 mM NaCl, 19 mM KCl, 10% glycerol, and an EDTA-free protease inhibitor cocktail (Roche Diagnostics, Mannheim, Germany). The target proteins were purified by TALON (Clontech, Mountain View, USA) immobilized metal affinity chromatography with a buffer containing 100 mM Tris-HCl (pH 8.0), 150 mM NaCl, 19 mM KCl, and 10% glycerol, which was pre-equilibrated in an anaerobic chamber. FrhAGB and FrhAG proteins were purified as previously described (Jung et al. 2020).

Rd reduction assay

The reduction of Rd was performed in a reaction buffer containing 50 mM Tris-HCl (pH 8.0), 500 mM KCl, and 75 mM NaCl, which was pre-equilibrated with either N₂ or H₂ gas. To assess the reducing activity of NROR and FrhAGB, 0.2 mM NADPH or H₂ gas was provided as the electron donor, respectively. For the stopped-reaction assays, NROR (36 nM) or FrhAGB (8.6 nM) was added to the gas-charged buffer. After a 2-min preincubation, the reaction was initiated by the addition of 32 µM Rd and incubated at 70 °C for 30 s. The reduction of Rd was monitored by measuring the decrease in absorbance at 494 nm (ε = 9.22 mM⁻¹ cm⁻¹) using a UV-VIS spectrophotometer (UV-2600, Shimadzu, Kyoto, Japan) as previously described (Blake et al. 1991). For time-course experiments, 10 nM FrhAGB or FrhAG was preincubated at 70 °C for 2 min before the addition of 40 µM Rd, and the decrease in absorbance was continuously monitored. For kinetic analysis, 4.3 nM FrhAGB or FrhAG was used with varying concentrations of Rd (0–11 µM). Kinetic parameters were determined based on the rates of Rd reduction under these conditions. One unit (U) of Rd reduction activity was defined as the amount of enzyme catalyzing the reduction of 1 μmol Rd per minute. To account for variations in molecular mass across different enzymes, specific activities were normalized to units per nanomole of enzyme (U nmol⁻¹) rather than to units per milligram of enzyme (U mg⁻¹).

Surface plasmon resonance (SPR) analysis

The analysis was performed using the Biacore 1K + system (Cytiva, Upsala, Sweden). FrhAGB and FrhAG were covalently immobilized on a CM5 sensor chip (Cytiva, Upsala, Sweden) at a surface density of 300 response units (RU). The analyte, Rd, was subjected to a fourfold serial dilution over a concentration range of 0 to 16.384 µM. The protein-protein interaction was measured at 25 ℃ in HBS-EP + buffer (Cytiva, Upsala, Sweden). After each measurement cycle, the chip surface was regenerated using a 10 mM Glycine-HCl solution (pH 1.5) for 40 s. During the experiment, binding signals were recorded as response units (RU) measured at steady state, and the binding affinities, represented as K_D values, were calculated by plotting the RU values against the concentrations of Rd using the Biacore Insight Evaluation software (Cytiva, Upsala, Sweden).

Structural model building

The structural models of FrhAGB–Rd and FrhAG–Rd complexes were generated using AlphaFold 3 (Abramson et al. 2024) via the AlphaFold Server (https://alphafoldserver.com/). The combined model confidence score was calculated using the predicted template modeling (pTM) and interface predicted template modeling (ipTM) scores according to the equation: 0.8 ipTM + 0.2 pTM. Iron–sulfur [4Fe-4S] clusters and a mononuclear iron center were manually introduced and adjusted using the COOT program (Emsley et al. 2010), based on conserved coordination geometry. The final structural model was visualized and rendered in PyMOL (DeLano 2002), where inter-cofactor distances were analyzed by measuring the shortest Fe–Fe distances.

Results and discussion

Reduction of rubredoxin by FrhAGB complex

In the effort to identify a potential interaction partner for FrhAGB within the oxidative stress defense system, NROR was selected as the initial target. This decision was based on the fact that NROR is classified as an NADPH-dependent oxidoreductase. This characteristic of NADPH dependency is shared by all previously identified proteins that interact with FrhAGB in T. onnurineus NA1 (Jung et al. 2020; Yang et al. 2023). In addition, they exhibit structural similarities as noted in prior research (Yang et al. 2023) based on the AlphaFold Protein Structure Database (Tunyasuvunakool et al. 2021; Fleming et al. 2025).

To evaluate the reduction of NROR, it was necessary to measure the subsequent electron transfer to Rd within the context of a redox cascade. For this assay, recombinant NROR and Rd proteins were purified from E. coli (Supplementary Fig. S1). The electron transfer from FrhAGB to the NROR-Rd redox cascade using H₂ was assessed by measuring the decrease in absorbance at 494 nm. This decrease indicates the transition of the iron center in Rd from the oxidized state (Fe³⁺) to the reduced state (Fe²⁺). The activity of the recombinant NROR and Rd proteins was confirmed, demonstrating that NROR reduces Rd in the presence of NADPH (Fig. 1). The addition of FrhAGB with H₂ into the reaction mixture facilitated the reduction of Rd, effectively substituting for NADPH. This implies that NROR may acquire electrons from FrhAGB to reduce Rd. Under the condition tested, the molar activity of the FrhAGB with NROR in the presence of H₂ (3.4 U nmol⁻¹) was 3.7-fold higher than that of NROR using NADPH (0.9 U nmol⁻¹). Intriguingly, Rd reduction was observed even in the absence of NROR, with FrhAGB alone exhibiting 53% of the activity measured in the presence of NROR. This finding suggests that FrhAGB can directly reduce Rd using H₂. Consequently, these results indicate that Rd serves as a direct functional interaction partner for FrhAGB, potentially playing a key role in the oxidative stress defense system of T. onnurineus NA1.

Fig. 1.

H₂-driven reduction of Rd by FrhAGB. The reduction of Rd catalyzed by NROR or FrhAGB was quantified by measuring absorbance changes at 494 nm using an extinction coefficient of 9.22 mM⁻¹ cm⁻¹. Symbols “ + ” and “–” denote the addition or omission of each reagent, respectively

It has been reported that in addition to NAD(P)H:Rd oxidoreductase (commonly known as Rd reductase), ferredoxin-NADP⁺ oxidoreductase, purified from spinach chloroplasts, has been shown to effectively reduce Rd in vitro using NADPH as an electron donor (García-Cerdán et al. 2019; Peterson and Coon 1968). Thus, this research has identified hydrogenase as a newly recognized enzyme capable of reducing Rd.

The mechanism by which Rd is reduced by the FrhAGB complex is of interest. The structures of the subunits of FrhAGB have been identified: the FrhA subunit contains a binuclear [NiFe] center, while the FrhG subunit contains three [4Fe-4S] clusters. Additionally, the FrhB subunit is an iron-sulfur flavoprotein that contains one [4Fe-4S] cluster and one FAD prosthetic group (Jeon et al. 2015). It has been reported that the FrhB subunit is not essential for electron transfer to TrxR (Jung et al. 2020) or for formate production through its interaction with Fdh3 (Yang et al. 2023). However, it may be required to reduce Rd. To determine whether the FrhB subunit is essential for this reduction process, an examination was conducted. The result showed that the FrhAG complex performed comparably to the FrhAGB complex in reducing Rd (Supplementary Fig. S2). This suggests that Rd can still be effectively reduced by FrhAG through electron transfer and that this process does not rely on FrhB.

The H₂-dependent Rd reduction activity of FrhAGB or FrhAG was compared to the NADH- or NADPH-dependent Rd reduction activity of the NROR enzymes from Pyrococcus furiosus (Ma and Adams 1999), Clostridium acetobutylicum (Petitdemange et al. 1979), and Desulfovibrio gigas (Chen et al. 1993). The specific activities of these three enzymes were reported to be 799, 1.89, and 0.71 U nmol⁻¹, respectively. While these values may appear lower, it is important to note that FrhAGB and FrhAG are operating in different biochemical contexts. Given the solubility of H₂ under ambient conditions, their activity is significant. This potential for FrhAGB and FrhAG to function effectively in various biological systems where NADH/NADPH is limited indicates that they could be especially valuable in specific metabolic pathways.

Kinetic analysis of H₂-dependent Rd reduction by FrhAGB and FrhAG

To gain insight into the catalytic mechanisms underlying Rd reduction by FrhAGB, a kinetic analysis was performed (Table 1). Under the conditions tested, FrhAGB exhibited an apparent Michaelis constant (K_{m, app}) of 0.046 mM and a turnover number (k_{cat, app}) of 289 s⁻¹. In comparison, FrhAG showed a similar affinity for Rd (K_{m, app} = 0.043 mM), but a lower turnover number (k_{cat, app} = 216 s⁻¹). Consequently, the catalytic efficiency (k_cat/K_m) of FrhAGB (6,283 mM⁻¹ s⁻¹) was approximately 1.25-fold higher than that of FrhAG (5,023 mM⁻¹ s⁻¹). These results suggest that although the FrhB subunit may not be essential for binding to the substrate, its presence may contribute to the structural stability or optimal alignment of the electron transfer pathway.

Table 1.

Kinetics of FrhAGB and FrhAG interaction with Rd

	Substrate (mM)a	Km, app (mM)b	kcat, app (s−1)b	kcat/Km (mM−1 s−1)
FrhAGB	Rd (0.0033–0.0109)	0.046	289	6,283
FrhAG	Rd (0.0038–0.0081)	0.043	216	5,023

^aConcentration range used for the determination of kinetic parameters

^bapp denotes apparent values

Protein–protein interaction between FrhAG(B) and Rd

To investigate whether electron transfer from FrhAG(B) to Rd depends on direct protein-protein interactions between FrhAG and Rd, and to evaluate the binding strength, SPR analysis was performed. FrhAGB or FrhAG was used as the immobilized ligand, while Rd was flowed over the surface of the sensor chip as the analyte, at concentrations ranging from 0 to 16.384 μM. The binding affinity was then measured, yielding binding constants, K_D, of 1.27 × 10^–6 M for FrhAGB and 1.26 × 10^–6 M for FrhAG (Fig. 2). These results indicate that hydrogenase and Rd interact strongly, and the absence of the FrhB subunit did not affect this interaction.

Fig. 2.

Surface plasmon resonance (SPR) analysis of the interactions of FrhAGB or FrhAG and Rd. FrhAGB (A) and FrhAG (B) were immobilized on a CM5 sensor chip as ligands. Sensorgrams (left) were obtained using a series of Rd concentrations ranging from 0 to 16.384 μM. The binding curves to Rd were fitted to a kinetic model to determine K_D values (right)

The feasibility of electron transfer between FrhAGB and Rd was evaluated using the structural model predicted by AlphaFold 3. Structural modeling indicates that Rd interacts with both FrhAGB and FrhAG through distinct interfaces (Fig. 3A). According to the model of FrhAG-Rd complex, the Rd-binding site overlaps with the region occupied by FrhB in the FrhAGB complex. This FrhB binding site in the model is structurally consistent with the previously reported X-ray structure of F₄₂₀-reducing hydrogenase from Methanothermobacter marburgensis (Vitt et al. 2014) (Fig. 3A). It is plausible that the FrhB subunit undergoes dissociation under the assay conditions, thereby alleviating steric hindrance and allowing Rd access to the FrhAG core. Given the proposed oxygen-resistance mechanism in the frhAGB-overexpression mutant (FO), FrhB may similarly dissociate in vivo to position Rd for sequential electron transfer by FrhAG. Following canonical coordination geometry, the [4Fe-4S] clusters in FrhG and the mononuclear Fe center in Rd were positioned and distances between them were measured (Fig. 3B). The inter-cofactor distances ranged from 8.8 to 12.4 Å, suggesting a rapid and efficient electron transfer between the cofactors, based on the report on the distance between redox centers to allow rapid electron tunneling within 14 Å (Page et al. 1999). In contrast, the distances from the [4Fe-4S] clusters in FrhG or the FAD in FrhB to the Fe center of Rd in the predicted FrhAGB-Rd model both exceeded the 14 Å threshold required for physiological tunneling (Supplementary Fig. S3). This suggests that direct electron transfer between these specific sites is kinetically unfavorable, potentially necessitating a mediator or a specific conformational change for productive interaction. Although the model relies on a predicted structure rather than one determined experimentally, the measured inter-cofactor distances strongly suggest that sequential electron transfer from FrhAG to Rd is feasible. These findings suggest that FrhB is not an essential mediator for electron transfer to Rd. Accordingly, the FAD cofactor within FrhB appears to be non-essential in this pathway, supporting a mechanism of direct electron transfer from the FrhAG catalytic core to Rd. Further studies are needed to elucidate the specific dynamics of electron flow between these metal centers.

Fig. 3.

Predicted interaction models of FrhAGB and FrhAG complexes with Rd. A AlphaFold-predicted structures of FrhAGB and FrhAG with Rd, compared with the X-ray crystal structure of F₄₂₀-reducing hydrogenase from Methanothermobacter marburgensis. The combined model confidence scores are 0.74 for the FrhAGB-Rd complex (ipTM = 0.73, pTM = 0.78) and 0.65 for the FrhAG-Rd complex (ipTM = 0.62, pTM = 0.75). B A schematic illustration of the proposed spatial arrangement of the metallocofactors involved in electron transfer. FrhA is depicted in magenta, FrhG in green, and Rd in cyan. The iron-sulfur [4Fe-4S] clusters within FrhG are shown in stick representation, while the mononuclear iron center of Rd is illustrated as an orange sphere. Inter-cofactor distances are indicated by red dashed lines

This study suggests that although FrhAGB may not be the primary enzyme responsible for reducing Rd, it can contribute to detoxifying the cell from oxidative stress by utilizing abundant H₂ as an alternative electron donor rather than NADPH. Although it has not yet been tested, it is also possible that FrhAGB may interact with terminal reductases, such as rubrerythrin or flavodiiron proteins, in the oxidative stress response system. Notably, a [FeFe] hydrogenase–rubrerythrin chimeric enzyme, identified in Clostridium perfringens, has been shown to be capable of reducing H₂O₂ under H₂ atmosphere (Taylor et al. 2025).

Conclusion

This study demonstrated that FrhAGB hydrogenase can transfer electrons to Rd through direct interaction. The FrhB subunit is not essential for this interaction or for the Rd reduction activity, as FrhAG exhibits similar binding strength and reducing activity on its own. The link between H₂ oxidation by the hydrogenase and Rd reduction indicates a newly characterized pathway that protects cells against oxidative stress. This pathway serves as an alternative to the standard NADPH-dependent system, which is essential but may become depleted during severe stress. Using readily available H₂ provides a supplementary defense mechanism. These findings highlight the need for further investigation into the potential interactions between the FrhAGB hydrogenase and other components of the oxidative stress defense system.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

JY, HP, BGC, and MJ performed the experimental work and analyzed the data. YJA and HO provided bioinformatic support. SGK conceived and designed the research and drafted the manuscript. HSL edited the manuscript. All authors read and approved the final manuscript.

Funding

This work was supported by the Korea Institute of Ocean Science and Technology (KIOST) in-house programs (EA0411 and EA0413).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethical approval

This research on microorganisms complies with relevant institutional, national, and international guidelines and legislation.

Conflict of interest

The authors declare no competing interests.