[FeFe]-Hydrogenase: Defined Lysate-Free Maturation Reveals a Key Role for Lipoyl-H-Protein in DTMA Ligand Biosynthesis

Abstract

Maturation of [FeFe]-hydrogenase (HydA) involves synthesis of a CO, CN⁻, and dithiomethylamine (DTMA)-coordinated 2Fe subcluster that is inserted into HydA to make the active hydrogenase. This process requires three maturation enzymes: the radical S-adenosyl-l-methionine (SAM) enzymes HydE and HydG, and the GTPase HydF. In vitro maturation with purified maturation enzymes has been possible only when clarified cell lysate was added, with the lysate presumably providing essential components for DTMA synthesis and delivery. Here we report maturation of [FeFe]-hydrogenase using a fully defined system that includes components of the glycine cleavage system (GCS), but no cell lysate. Our results reveal for the first time an essential role for the aminomethyl-lipoyl-H-protein of the GCS in hydrogenase maturation and the synthesis of the DTMA ligand of the H-cluster. In addition, we show that ammonia is the source of the bridgehead nitrogen of DTMA.

Graphical Abstract

The aminomethyl-lipoyl-H-protein (H_met) of the glycine cleavage system (GCS) is shown to play a key role in maturation of the [FeFe]-hydrogenase. Including H_met, or H_met plus other components of the GCS, allows maturation by purified HydE, HydF, and HydG in the absence of cell lysate for the first time.

[FeFe]-hydrogenase (HydA) catalyzes the reversible reduction of protons to H₂ at a complex metal cluster, the H-cluster, composed of a [4Fe-4S] cluster bridged via a cysteine thiolate to a [2Fe] subcluster coordinated by CO, CN⁻ and dithiomethylamine (DTMA) ligands (Figure 1).^[1] [FeFe]-hydrogenases can catalyze rates of H₂ production that are approximately 100x higher than other hydrogenases, making HydA an appealing target for biohydrogen production technologies.^[2] Early efforts to heterologously express [FeFe]-hydrogenase, however, led to inactive enzyme, a major obstacle to practical applications of this enzyme.^[3] Subsequent studies showed that three maturation enzymes, HydE, HydF, and HydG, were required to produce an active [FeFe]-hydrogenase.^[4]

Figure 1.

Current model for biological maturation of the H-cluster of the [FeFe]-hydrogenase. Question marks indicate steps that are not fully understood.

HydF is a GTPase that binds redox-active iron-sulfur clusters and serves as a scaffold for assembly of a 2Fe precursor ([2Fe]_F) delivered to HydA to yield active hydrogenase.^[5] HydF can be loaded with [2Fe]_F enzymatically by co-expression with HydE and HydG, yielding a [2Fe]_F subcluster resembling the [2Fe] subcluster of the H-cluster ([2Fe]_H).^[6] In an alternative semi-synthetic approach, pre-formed synthetic [2Fe] clusters coordinated by CO, CN⁻, and DTMA can be used to load HydF, which then can activate HydA.^[7] HydG synthesizes the CO and CN⁻ ligands from tyrosine,^[8] and also provides iron in the form of a [Fe(Cys)(CO)₂(CN)] synthon.^[9] HydE binds a synthetic analog of this synthon, and uses it as a substrate to catalyze an adenosylation reaction,^[10] and may convert two of these adenosylated Fe(I) synthon units to a [2Fe] subcluster precursor.^[11] The origin of the DTMA ligand is not well understood. It was thought to arise from HydE catalysis^[12] but now appears to be formed independent of HydE.^[11] Maturation of HydA^ΔEFG (HydA expressed in the absence of the maturase enzymes) can be achieved in vitro by combining extracts of E. coli cells overexpressing the individual maturases, or by using purified maturase enzymes together with E. coli cell lysate.^{[8d, 13]} In work to date, the requirement for the E. coli lysate (20% to 100% v/v) has been absolute.^{[4c, 4d, 8d, 9a, 11, 13–14]}

We report here development of a fully defined enzymatic system for maturation of HydA using HydE, HydF, and HydG, their substrates and reductants, and components of the glycine cleavage system (GCS, Scheme S1),^[15] but without E. coli lysate. The results reveal a key role for the aminomethyl-lipoyl-H-protein (H_met) as a precursor of the DTMA ligand of the H-cluster.

To better understand the chemistry of H-cluster maturation, we pursued fractionation of E. coli cell lysate to identify component(s) responsible for its requirement during maturation. These efforts revealed one or more components of a low-molecular weight (LMW) fraction (10 – 15 kDa, Figure S1) of lysate could also support maturation. Essential component(s) from E. coli lysate were inactivated by heat/acid treatment, suggesting they may be proteins (Figure S2). Shotgun proteomics revealed the presence of carrier proteins including the acyl-carrier protein (ACP) in the LMW fraction, suggesting the involvement of a small carrier protein in hydrogenase maturation, perhaps in the still poorly understood synthesis of the DTMA ligand, as suggested by Juan Fontecilla-Camps (personal communication). However, no detectable HydA activity was observed following in vitro HydA maturation with holo- or lipoyl-ACP in place of E. coli cell lysate, and with purified HydE, HydF, HydG, SAM, and other small molecule components (see SI).

We next considered the H-protein of the GCS (Scheme S1), as a potential key component of HydA maturation. Lipoylated H-protein acts as a carrier between protein components of the GCS (Scheme S1), and can be in the disulfide form (H_ox), the reduced form (H_red), or the aminomethyl form (H_met). Maturation reactions carried out with H_red in the absence of cell lysate provided no detectable hydrogenase activity (Figure 2). H_met was synthesized enzymatically from H_ox using the P-protein, PLP, and glycine (Scheme S1), purified, and verified via mass spectrometry (Figure S3). HydA maturation with H_met added to our defined system resulted in active HydA, as evidenced by H₂ gas production (Figure 2). Increasing the amount of H_met provided a significant increase in H₂ (Figure 2), demonstrating that HydA activation is dependent on H_met. To our knowledge, this is the first report of biological HydA^ΔEFG conversion to active hydrogenase in the absence of cell lysate, and points to H_met as a previously unrecognized essential component.

Figure 2.

Maturation of HydA without lysate, using the GCS H-protein. Without H-protein (Ctrl) or with lipoyl-H-protein (H_red) no active HydA is formed. Addition of aminomethyl-lipoyl-H-protein (H_met) provides maturation of HydA in a concentration-dependent manner. Assays for H₂ were run for 24 h to allow H₂ to accumulate. Conditions as reported in the SI.

In these experiments, the level of activation of HydA^ΔEFG in the presence of H_met is nonetheless low, requiring 24 h incubation to detect small amounts of H₂. The low hydrogenase activity may be due to instability of H_met, which is known to readily hydrolyze, especially in the presence of partner proteins.^[16] We reasoned that if this is so, regeneration of H_met in situ would improve the hydrogenase maturation. Regeneration of H_met from H_red in situ under the reducing conditions required for maturation was accomplished by including aminomethyltransferase (T-protein), serine hydroxymethyltransferase (SHMT), serine, and NH₄⁺ (Scheme 1) in the maturation mixture. HydA maturation under these conditions yielded dramatically improved hydrogenase activity, with further refinement as described below providing specific activities as high as 390 μmol/min/mg HydA (Figure 3); in comparison, holo-HydA is reported to have a specific activity of ~700 μmol/min/mg.^[17]

Scheme 1.

In situ generation of H_met from H_red via SHMT and the T-protein of the GCS (see Scheme S1 for the full GCS). All reactions are reversible.

Figure 3.

Maturation of HydA in the presence of the T-protein and SHMT. Concentrations of key components were varied to test the hypothesis of the involvement of the GCS and H_met during hydrogenase maturation. Conditions as reported in the SI. Reactions contained no lysate.

To further probe the involvement of H_met of the GCS during HydA maturation, we examined the dependence of maturation on concentrations of the H_met regeneration components (Figure 3). Maturation is dependent on both serine and NH₄⁺, with high concentrations required for optimal activity, consistent with the equilibria outlined in Scheme 1. Glycine has an inverse effect on maturation (Figure 3), consistent with the role of SHMT in producing methylene-tetrahydrofolate (CH₂-THF, Scheme 1). The maturation observed without added NH₄⁺ is attributed to co-purification of NH₄⁺ with T-protein, consistent with the well-defined NH₄⁺ binding site in this protein.^[18] THF also impacts maturation, increasing activity at low concentrations followed by a rapid decrease with increasing THF concentration (Figures 3, S5), consistent with THF involvement in both SHMT and T-protein equilibria; the dominant effect of THF is on the T-protein equilibrium, driving it towards H_red and away from H_met (Scheme 1). Maturation works without added THF because THF co-purifies with the T-protein. The dependence of HydA maturation on these components required for generation of H_met in situ further support a key role for H_met in HydA maturation; these components were provided by lysate in prior lysate-dependent maturations.^{[4c, 4d, 8d, 9a, 13b, 13c, 14]}

Our results suggest that the GCS, and specifically H_met, provides a precursor for DTMA biosynthesis. We propose that the aminomethyl group of H_met is the source of the C and N of the C-N-C framework of DTMA. Given that the aminomethyl of H_met is derived from the β-C of serine and NH₄⁺ (Scheme 1), we set out to probe incorporation of isotopic labels from these sources into the DTMA of matured HydA. Maturation reactions were carried out in the absence of lysate (see SI), with [¹³C₃,¹⁵N]-serine or ¹⁵NH₄Cl included in place of the unlabeled component. HydA was purified from the maturation mixture and oxidized with thionin to give an electron paramagnetic resonance (EPR) spectrum typical of the H_ox state of holo-HydA (Figure S6). Electron nuclear double resonance (ENDOR) spectroscopy was used to look for incorporation of isotopic labels into the DTMA of the H-cluster. For the [¹³C₃,¹⁵N]-serine samples, the spectrum collected at ~ g₁ shows two distinct ¹³C hyperfine signals (Figure 4, red) associated with the two carbons of the DTMA ligand, as reported previously for HydA matured in the presence of lysate.^[14b] The presence of signals with different hyperfine splittings for the two DTMA carbons is consistent with the prior work demonstrating their origins from serine and asymmetry in the DTMA ligand environment in holo-HydA.^{[14b, 19]} However, this sample prepared without lysate shows no ¹⁵N ENDOR response, indicating that ¹⁵N from [¹³C₃,¹⁵N]-serine is not present in the DTMA of HydA matured lysate-free (Figure 4, black).

Figure 4.

35 GHz pulsed ¹³C and ¹⁵N ENDOR of HydA matured in the absence of lysate. Top, ¹³C Mims ENDOR spectrum at g=2.093 of HydA matured with [¹³C₃,¹⁵N]-serine. Bottom, ¹⁵N Mims ENDOR spectra at g=2.004 of HydA matured with ¹⁵NH₄Cl (blue) and with [¹³C₃,¹⁵N]-serine (black). Mims holes are indicated with arrows. Microwave frequency, 34.605 GHz; rep time, 50 ms; T = 2 K. For a complete characterization of these signals, see ref 15b.

In our defined maturation, the T-protein uses NH₄⁺ to provide the N for the aminomethyl group of H_met (Scheme 1), and thus NH₄⁺ should be the source of the N of DTMA. ¹⁵N ENDOR spectra of HydA matured in the presence of ¹⁵NH₄Cl reveals this to be true, as the corresponding ¹⁵N coupling of 1.5 MHz is observed (Figure 4, blue). Prior work showed that both N and C of DTMA could be derived from serine, but this was in the presence of E. coli cell lysate (72% v/v).^[14b] Given that cell lysate contains enzymes such as serine dehydratase that can liberate NH₄⁺ from serine, it is likely that the observed incorporation of ¹⁵N from serine in the prior work was a result of ¹⁵NH₄⁺ liberation due to the presence of lysate.

We demonstrate the first maturation of HydA using a defined enzymatic system, thereby revealing an essential role for H_met and the GCS in building the DTMA ligand of the H-cluster. H_met alone, when added to lysate-free maturation components, is sufficient to provide active HydA (Figure 2), revealing the key role for H_met during H-cluster biosynthesis. Inclusion of an H_met regeneration system (T-protein, SHMT, serine, and NH₄⁺) affords high levels of HydA activation, with the roles of T-protein and SHMT highlighted by the dependence of activation on the concentrations of key species (Figure 3). ¹³C/¹⁵N-ENDOR spectroscopy of HydA matured using this defined system confirms that the carbon atoms of DTMA originate from serine, while the nitrogen comes from NH₄⁺ (Figure 4), consistent with our previous prediction^[20] and with the known chemistry of the GCS during H_met formation from H_red (Scheme 1).

The recent report that HydA can be matured in the presence of HydF, a synthetic [Fe₂(SH)₂(CO)₄(CN)₂]²⁻ complex, and cell lysate,^[11] together with the results presented here, lead us to propose that H_met interacts with the HydF-[Fe₂(SH)₂(CO)₄(CN)₂]²⁻ complex to synthesize the dimethylamine backbone of DTMA. In this model, the aminomethyl group of H_met is functionally equivalent to the formaldehyde and ammonia used to synthesize the C-N-C DTMA backbone in the [2Fe]_H model reported by Li and Rauchfuss.^[21] We propose that during biological maturation, each sulfur of the [2Fe]_F precursor cluster [Fe₂(SH)₂(CO)₄(CN)₂]²⁻ nucleophilically attacks the C of an aminomethyl group of H_met, resulting in transfer of two aminomethyl groups from two H_met to the bridging sulfides of the HydF-bound [2Fe]_F precursor cluster. Condensation of the aminomethyl moieties with loss of NH₄⁺ would then form DTMA (Scheme 2). In this model, the sulfurs of DTMA are derived from the precursor diiron cluster, not from the lipoyl group of Hmet; use of a seleno-lipoyl-H-protein during maturation could test this hypothesis.

Scheme 2.

Proposed pathway for formation of the DTMA ligand of the [2Fe]_F cluster on HydF.