Characterization of a putative sensory [FeFe]-hydrogenase provides new insight into the role of the active site architecture

[FeFe]-hydrogenases are known for their high rates of hydrogen turnover, and are intensively studied in the context of biotechnological applications. Evolution has generated a plethora of different subclasses with widely different characteristics. The M2e subclass is phylogenetically distinct from previously characterized members of this enzyme family and its biological role is unknown. It features significant differences in domain- and active site architecture, and is most closely related to the putative sensory [FeFe]-hydrogenases. Here we report the first comprehensive biochemical and spectroscopical characterization of an M2e enzyme, derived from Thermoanaerobacter mathranii. As compared to other [FeFe]-hydrogenases characterized to-date, this enzyme displays an increased H2 affinity, higher activation enthalpies for H+/H2 interconversion, and unusual reactivity towards known hydrogenase inhibitors. These properties are related to differences in active site architecture between the M2e [FeFe]-hydrogenase and “prototypical” [FeFe]-hydrogenases. Thus, this study provides new insight into the role of this subclass in hydrogen metabolism and the influence of the active site pocket on the chemistry of the H-cluster.


Experimental procedures General
All chemicals were purchased from Sigma-Aldrich or VWR and used as received unless otherwise stated. Protein expression was analyzed by 12% SDS-PAGE minigels in a BIO-RAD Mini PROTEAN® system. The proteins were stained with Page Blue protein staining solution (Thermo Fisher Scientific) according to the supplier instructions. All anaerobic work was performed in an MBRAUN glovebox ([O2] < 10 ppm). The [2Fe] adt and [2Fe] pdt subsite mimics were synthesized in accordance to literature protocols with minor modifications, and verified by FTIR spectroscopy. 1-5

Protein expression and purification
The gene encoding for Tam HydS was synthesized and cloned in pET-11a(+) by Genscript® using restriction sites NdeI and BamHI following codon optimization for expression in E. coli. Transformation of the plasmid into E. coli was performed using heat shock at 42°C. Transformed E. coli were grown overnight in 10 mL LB medium containing 100 µg mL -1 ampicillin at 37°C. These cultures were subsequently used to inoculate 1 L of M9 medium (22 mM Na2HPO4, 22 mM KH2PO4, 85 mM NaCl, 18 mM NH4Cl, 0.2 mM MgSO4, 0.1 mM CaCl2, 0.4% (v/v) glucose) containing 100 µg mL -1 ampicillin. Cultures were grown at 37°C and 150 rpm until an optical density (OD600) of appr. 0.4-0.6 was reached. Protein expression was induced by addition of 0.1 mM FeSO4 and 1 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) and cultures were incubated at 20°C and 150 rpm for appr. 16 h. Cells were thereafter harvested by centrifugation in a Beckman Coulter Avanti J-25 centrifuge (5,000 rpm/4,424 x g, 10 min). Cr HydA1 was prepared as previously described. 6 Dd HydAB was kindly provided by Prof. Juan Fontecilla-Camps.

Aerobic purification
The cell pellet was resuspended in lysis buffer (Tris-HCl (100 mM, pH 8), NaCl (150 mM), MgCl2 (10 mM), lysozyme from chicken egg white (1 mg mL -1 ), DNAse I from bovine pancreas (0.05 mg mL -1 ) and RNAse A from bovine pancreas (0.05 mg mL -1 ) and incubated on ice for 30 min. Cell lysis was performed by three cycles of freezing/thawing in liquid N2 followed by sonication. Cell debris was removed by centrifugation in a Beckman Coulter Optima L-90K Ultracentrifuge (55,000 rpm/222,592 x g, 60 min) and the supernatant was collected and filtered (0.45 µm syringe filter) before being loaded on a StrepTrap™ HP (GE Healthcare) affinity column using an ÄKTA pure FPLC system (GE Healthcare) and purified according to the manufacturer´s instructions.

Anaerobic purification
The cell pellet was brought into an anaerobic glovebox, resuspended in lysis buffer (Tris-HCl (100 mM, pH 8), NaCl (150 mM), MgCl2 (10 mM), lysozyme from chicken egg white (1 mg mL -1 ), DNAse I from bovine pancreas (0.05 mg mL -1 ) and RNAse A from bovine pancreas (0.05 mg mL -1 ) and incubated on ice for 30 min. Cell lysis was performed by three cycles of freezing/thawing in liquid N2. Cell debris was removed by centrifugation in a Beckman Coulter Optima L-90K Ultracentrifuge (55,000 rpm/222,592 x g, 60 min) and the supernatant was collected and filtered (0.45 µm syringe filter) before being loaded on a StrepTrap™ HP (GE Healthcare) affinity column using a BioLogic DuoFlow™ FPLC system (Bio-Rad) and purified according to the manufacturer´s instructions.
The concentration of all purified proteins was quantified using the Bradford assay with bovine serum albumin as a protein standard. 7 Quantification of Fe-content was performed using a previously reported assay. 8

Iron-sulfur cluster reconstitution
Iron-sulfur cluster reconstitution was performed under strictly anaerobic conditions and was preceded by a 10minute incubation with a tenfold molar excess of dithiothreitol at room temperature. Ferrous ammonium sulfate and L-cysteine was then added in a 1.5-fold molar excess to the desired amount of Fe-atoms to be added (dependent on the initial Fe-content in the purified protein). Reconstitution was initiated by the addition of 1% molar equivalent of cysteine desulfarase (CsdA) from E. coli. The reaction was followed by monitoring the absorbance increase around 400 nm by UV/Visible spectroscopy using an AvaSpec-ULS2048-USB2-UA-50: Avantes Fiber Optic UV/VIS/NIR spectrometer. After the reconstitution was finished, it was stopped by running the reaction through a PD-10 column (GE Healthcare).

H2-production assay
Activation of Tam HydS with subsequent measurement of H2-production activity was performed under strictly anaerobic conditions by mixing apo-Tam HydS (1 µM) with sodium dithionite (20 µM, 20x excess), [2Fe] adt (12 µM, 12x excess) and Triton X-100 (1% v/v) in sodium phosphate buffer (100 mM, pH 6.8). The activation mixture was incubated at room temperature for 1 h. The reaction was then initiated by the rapid addition of methyl viologen (10 mM) and sodium dithionite (100 mM). The reaction was incubated at 30°C for 15 min. The amount of produced H2 was then determined by analyzing the reaction headspace on a PerkinElmer Clarus 500 gas chromatograph (GC) equipped with a thermal conductivity detector (TCD) and a stainless-steel column packed with Molecular Sieve (60/80 mesh). The operational temperatures of the injection port, the oven and the detector were 100 °C, 80 °C and 100 °C, respectively. Argon was used as carrier gas at a flow rate of 35 mL min −1 .

Homology modeling
Homology modeling of truncated Tam HydS was performed using YASARA Structure version 18.3.23 as previously described. 9 EPR sample preparation EPR samples were prepared under strictly anaerobic conditions by diluting reconstituted Tam HydS to 50 µM in Tris-HCl buffer (100 mM, pH 8). Activated samples were treated with 600 µM [2Fe] adt or [2Fe] pdt (12x excess) and reduced samples were treated with 1 mM sodium dithionite (20x excess). Samples were then incubated at room temperature for 30 min. All gas treated samples were exposed to the specific gas (N2, H2 or air) for 10 min. Samples were finally transferred to EPR-tubes and flash-frozen in liquid N2.
EPR spectroscopy X-band EPR measurements were performed on a Bruker ELEXYS E500 spectrometer equipped with a SuperX EPR049 microwave bridge and a cylindrical TE011 ER 4122SHQE cavity in connection with an Oxford Instruments continuous flow cryostat. Measuring temperatures were achieved using liquid helium flow through an ITC 503 temperature controller (Oxford Instruments). Modulation frequency at 100 kHz was set in all measurements. Bruker Xepr software package was used for all data processing. The EPR simulation was performed using Bruker-Xsophe/XeprView software package (v.1.1.4). For the spin ½ system, only Zeeman interaction is included in the spin Hamiltonian operator. The simulated spectra were generated by matrix diagonalisation of anisotropic gtensors for a randomly oriented spin system.

Protein film electrochemistry sample preparation
Samples for protein film electrochemistry were prepared under strictly anaerobic conditions by mixing reconstituted Tam HydS (50 µM) with sodium dithionite (1 mM, 20x excess) and [2Fe] adt (600 µM, 12x excess) in sodium phosphate buffer (100 mM, pH 6.8) and incubated in room temperature for 30 min. The activation was stopped by running the reaction through a PD-10 column (GE Healthcare) equilibrated with Tris-HCl buffer (100 mM, pH 8), 150 mM NaCl, 5% v/v glycerol and 2 mM sodium dithionite. The sample was concentrated using Amicon®Ultra 30 kDa cutoff centrifugal filters (Merck Millipore Ltd.) and aliquoted into airtight serum vials that were flash-frozen in liquid N2 and stored at -80°C.

Protein film electrochemistry
Protein film electrochemistry experiments were carried out under strictly anaerobic conditions in an Ar atmosphere. The three-electrode system used comprised an isolated Ag/AgCl reference electrode (Fisher Scientific) held in a Luggin sidearm containing 0.1 M NaCl and a 0.5 mm platinum wire (Sigma Aldrich) as a counter electrode. All potentials were converted to the Standard Hydrogen Electrode (SHE) scale using the correction ESHE = ESCE + 210 mV at 25 °C. The working electrode was either a home-build pyrolytic graphite edge plane electrode (PGE) or glassy carbon electrode (GCE) rotated at 0−3000 rpm depending on the experiment. The PGEs working electrodes were made of a 2 mm diameter graphite rod embedded into a PEEK body. A tin rod attached to the graphite by a silver conductive epoxy adhesive (MG Chemicals) was used as an electrical lead to the electrodes. The GCEs used were purchased from Pine Research (E2M Fast Speed, 5 mm diameter). The glass cell used featured a water jacket for temperature control.
Four different procedures were tested for protein film preparation: (A) The PGE electrode surface was abraded with P1200 sandpaper before being placed in a sonicator for approximately 1 min and then rinsed with purified water. An enzyme solution (5 μL of 2-5 μM in 10 mM HEPES buffer pH 7) was then pipetted onto the electrode surface and left for 5 min to adsorb before the excess of the solution was removed by pipette. (B) The GCEs were successively polished with alumina powders (1.0, 0.3 and 0.05 μm) followed by ultrasonication in distilled water. The GCEs were functionalized with multi-wall carbon nanotubes as described earlier. 10 The MWCNTs-modified electrodes were used for enzyme immobilization as described in procedure (A). Data analysis was performed using OriginPro 8 software. Buffer composition was a mixture of MES, CHES, HEPES, TAPS and sodium acetate, 5 mM each, with NaCl (0.1 M) as carrying electrolyte titrated with NaOH or HCl to the desired pH. Electrochemical data were acquired using an Eco/Chemie PGSTAT10 and the GPES software (Metrohm/Autolab). Data were analyzed using Origin 8 software.
Temperature study experiments on either proton reduction or hydrogen oxidation were completed with the same film at temperatures varying between 10 and 70 o C. No significant protein film loss was observed during a set of experiments up to 60 o C. The pH study experiments were performed with the same film at pHs between 5-9.

FTIR sample preparation
Samples for FTIR were prepared under strictly anaerobic conditions by mixing reconstituted Tam HydS (100 µM) with sodium dithionite (2 mM, 20x excess) and [2Fe] adt (500 µM, 5x excess) in sodium phosphate buffer (100 mM, pH 6.8) and incubated in room temperature for 1 h. The activation was stopped by running the reaction through a NAP-25 column (GE Healthcare) equilibrated with Tris-HCl buffer (10 mM, pH 8). The sample was concentrated using Amicon®Ultra 30 kDa cutoff centrifugal filters (Merck Millipore Ltd.) and aliquoted into airtight serum vials that were flash-frozen in liquid N2 and stored at -80°C.

FTIR spectroscopy
FTIR spectroscopy was performed in ATR configuration on hydrated films of 1 µl isolated [FeFe]-hydrogenase (about 500 µM). Enzymes included Tam HydS, Dd HydAB, and Cr HydA1. All experiments were performed inside a Coylab glovebox (N2 atmosphere with 1 -2% H2). Absorbance spectra were recorded on a Bruker Tensor 27 with a spectral resolution of 2 cm -1 and a varying number of interferometer scans at 80 MHz (time resolution 1 -10 s). The spectrometer was equipped with a 3-reflections Si ATR unit (Smiths DuraSamplIR II). All gas-, light-, and potential titrations were performed according to established protocols. [12][13][14] For the temperature titrations, a stainless steel heat jacket was modified to fit the ATR crystal plate, which simultaneously served as a gas cell (see Fig. S2 below). A Julabo circulation pump was used to adjust the temperature of the heat jacket and ATR elements (20° -40°C). A digital thermo couple was used to measure the temperature at the interface of reflection element and crystal plate. This conservative approach was chosen to compensate for the difference in heat conductance of the silicon ATR crystal (~1.3 W K -1 m -1 ) and the stainless steel crystal plate (~15 W K -1 m -1 ).    Figure S4.  In agreement with previous reports, it was also observed in this work that the signals arising from the [4Fe4S] + clusters have pronounced spin anisotropy, resulting in much broader EPR spectra than those from the H-cluster, and the latter tends to suffer from microwave power saturation at much lower power than the [4Fe4S] + cluster(s) at equal temperature. This is illustrated in the obtained P1/2 values (listed in the tables inserted in S4A and B) studied at varied temperature.

Supplementary tables and figures
Guided by Figure S4 only the spectra retaining linearity were used for spin quantification. The linear region in graph B is noteworthy. The broad, seemingly "never saturating" [4Fe4S] + signals have a very narrow linear region, at least below 10 K, although the P1/2 values are large. This clearly indicates that the spin diffusion mechanism largely contributes to the relaxation mechanism for the [4Fe4S] + cluster even at a temperature as low as 5 K. Nevertheless, for an accurate spin quantification, it is essential that the linearity hold. Thus, microwave power with sub-milliwatts were applied in spin quantifications, albeit at the expense of signal/noise quality.  Figure 4, spectrum B main manuscript). Top spectra is the overlay between the simulation (solid line) and the experimental data (dashed line). Apart from the Hox-CO signal, three additional components contribute. Two rhombic spectra R1 and R2 and one axial signal. g-tensors are illustrated above. Based on the simulated spectra, double integration of the signals indicates that R2 accounts for 54.6%, R1, 25.2%, Hox-CO signal, 7.8% and the axial A1 signal for 12.4%. This 2:1 ratio between R2 and R1 is in reasonable agreement with the ratios observed between Hox and HoxH by FTIR at pH 8 ( Figure S7), further supporting the assignment of R2 to Hox and R1 to HoxH.    Figure S8. 1µL Tam HydS solution (~500 µM, pH 8) was injected into a gold mesh that had been placed onto the ATR silicon crystal before. No electron mediators were used. Following established protocols 12 , an electrochemical cell was attached to the ATR unit. Then, the beaker was filled with electrolyte solution (10 mM "mixed buffer" pH 8 with 100 mM KCl) and brought into contact with the protein film (the hydration process was monitored by FTIR spectroscopy). The electrolyte was carefully purged with N2 throughout the experiment to remove traces of H2. The gold mesh was connected to the working electrode, a platinum wire was used as counter electrode, and an Ag/AgCl element was exploited as reference electrode.
At the open circuit potential, the enzyme adopted the reduced state, Hred. (A) In the first step, we increased the potential from -400 mV vs SHE by 200 mV for 30 min each (until equilibrium conditions) and plotted the absorbance at 1947 cm -1 (Hox, black squares) and 1896 cm -1 (Hred, red circles). At +600 mV vs SHE, >90% of the film was oxidized, suggesting a midpoint potential of Em ox ≈ +200 mV (sigmoidal fits, solid black/red traces). The film showed <10% HoxH that appeared to behave similar to Hox. (B) In the second step, we lowered the potential by 100 mV for 30 min each. Surprisingly, no changes were observed until -200 mV vs SHE when the enzyme converted back into Hred. At -500 mV vs SHE, ~100% of the film was reduced, suggesting a midpoint potential of Em red ≈ -300 mV vs SHE (sigmoidal fits, solid black/red traces).
Electrochemical reduction is well compatible with single electron transfer, as expected for the transition of Hox into Hred. At -300 mV, Em red is comparatively anodic, reflecting the nature of Hred as metastable 'resting state' in Tam HydS (see main text). For comparison, the Hox/Hred midpoint potential of the sensory hydrogenase Tm HydS is equally high 15 whereas prototypical hydrogenases were reported with Em = -390 mV (Dd HydAB) 16 or Em = -380 mV (Cr HydA1) 17 . It should be noted that the observed potential of the Hox/Hred interconversion is pH dependent due to the protonation of the [2Fe] subsite, and in the case of Dd HydAB also dependent on the oxidation states of the F-clusters. For a more in-depth discussion see e.g. ref. 18 . In Tam HydS, electrochemical oxidation occurs at ~500 mV more positive potentials and displayed a broad profile incompatible with single electron transfer. 19 Figure S9. Formation of Hair and Hox-CO under oxidizing and reducing conditions.  Tam      In order to compensate for the CO-inhibited fraction of Dd HydAB, we calculated: The diagram illustrates a continous increase of vox with temperature in Tam HydS. Between 20°C and 30°C, vox is approximately ten times higher in Dd HydAB. At temperatures > 30°C, the activity drops significantly, which is in marked difference to Tam HydS.  Additionally, the super-oxidized state Hsox1 increased in the presence of air, and a blue-shifted state Hsox2 was identified. We did not observe Hair (i.e., as in Tam HydS) or Hox-O2 (i.e., as in Cr HydA1). Hsox1 and Hsox2 were partly re-activated under 1% H2. In the last step, the enzyme was subjected to 1% CO, which had no effect on Hsox1 and Hsox2 but converted both Hox and Hred into Hox-CO. In summary, these data demonstrate why a direct comparison between Tam (Table S1). These states are likely related to Hinact and Htrans but show an overall red-shifted IR signature. 20 Table S1. IR signature of H-cluster states observed in Dd HydAB (CO-only) Table S1. IR signature of H-cluster states observed in Dd HydAB (COonly). All data given in cm -1 . Figure S13. EPR-spectra in the g ≈ 2 region of as prepared and reduced Hair   Figure S14. 13 CO isotope editing and spectro-electrochemical characterization of Hair-ox and Hair-red. Figure S14. 13 CO isotope editing and spectro-electrochemical characterization of Hair-ox and Hair-red. (A) A film of Tam HydS was partly oxidized and reacted with 3 atm 13 CO following established protocols. 14 The dark yellow difference spectrum shows the conversion of Hred (as indicated by red arrows) and Hox (*1948 cm -1 ) into Hox-13 CO. White light irradiation in the presence of 13 CO facilitates isotope scrambling and produces an inhibited H-cluster with three terminal 13 CO ligands (Hox-( 13 CO)3), black spectrum). Surprisingly the µCO ligand was barely exchanged and only a very small µ 13 CO band at 1744 cm -1 was observed (α). Reactivating the H-cluster in the presence of 1 atm H2 (red spectrum) demonstrated a 40% editing efficiency which reflects the instability of Hox-CO in Tam HydS (compare Figure S9). Bands at 1871 and 1855 cm -1 are assigned to isotopically labelled 13 13 The conversion was found to be reversible (magenta trace, reduction; blue trace, oxidation). Whether or not Hair still carries the adt ligand is matter of speculation; unfortunately, Hair was not observed in Tam HydS pdt (compare Figure S6). The small but discernable difference between adt and pdt are well established in the case of the full H-cluster, 22 and could have been useful to conclude on the presence of absence of a dithiolate ligand in Hair. Given the octahedral nature of the iron ions in the intact H-cluster, the mono Fe species in Hair may form six bonds -one with the bridging cysteine, one with cyanide and two with the carbonyls. Due to the fact that Hair shows highly limited reactivity towards both H2 and CO (compare Figure S9), the remaining two coordination sites are arguably occupied. Therefore, we suggest the presence of azadithiolate in Hair ( Figure S13).  Figure S15. Suggested geometry of Hair. Figure S15. Suggested geometry of Hair. Model based on pdb coordinates 4XDC. 23 The hydrogen bonding between protein fold and Fep-CN is likely to stabilize the depicted ligand geometry. 24 DD Figure S16. Optimization of the immobilization protocol in PFE.     Figure S18. Influence of pH on currents observed in PFE. CVs obtained at a rotating disk PGE modified with holo-Tam HydS under 1 atm H2 at various pHs from 5 to 9 at 30°C. The potential is normalized to the reversible hydrogen electrode (RHE). The scan rate is 2 mV s -1 , the rotation rate is 3000 rpm. The onset potential for H + reduction decreases with lowering pH, with overpotential required for H + reduction being close to zero at pH 5. The onset potential for H2 oxidation also decreases with pH with observable switching to smaller overpotentials when going from pHs 9-7 to pH 6-5. Figure S19. Stability of protein film in PFE. Figure S19. Stability of a protein film in PFE. CVs obtained at a rotating disk PGE modified with holo-Tam HydS under 1 atm H2 in pH 7 solution at 30°C before the temperature dependence experiment (black) and after the experiment stopped at 60°C (red). The time of the experiment is ~4 hours, during which the solution with the electrode was heated from 10°C to 60°C and CVs were recorded for each 10°C increase in temperature at 3000 rpm. The scan rate is 2 mV s -1 , the rotation rate is 3000 rpm. Background CV is shown in black. The scan rate is 2 mV s -1 , the rotation rate is 3000 rpm, pH 7, 30°C. Once exposed to air to form Hair, the enzyme lost capacity for H2 oxidation, but retained H + reduction capacity, albeit with an increase in overpotential.