Re-engineering a NiFe hydrogenase to increase the H2 production bias while maintaining native levels of O2 tolerance

A single site amino acid exchange yields a NiFe hydrogenase with increased bias towards hydrogen production but conserved oxygen tolerance.

In order to make the Escherichia coli strains used in this study, chromosomal mutations were created using a modified version of the streptomycin counterselection protocol detailed by Heermann et al. 5 This is the first time such a method has been used for manipulation of hydrogenases. In the original protocol, 5 electroporation is used to transform DNA into E. coli, but we used chemically competent E. coli cells. 6 E. coli strains were cultured in LB media and on LB agar plates containing ampicillin (50 µg/ml) to select for the presence of the pRed/ET plasmid, kanamycin (50 µg/ml) to select for the presence of the rpsL-neo cassette and streptomycin (50 µg/ml) to select for the absence of the rpsL-neo cassette.
More, precisely, our stepwise protocol proceeded as follows: i.
E. coli W3110 was made streptomycin resistant (LAF-001) by insertion of the rpsL150 allele from E. coli MC1061. The rpsL150 from E. coli MC1061 was amplified by PCR (with GoTaq polymerase, Promega) using primers rpsl_up1 and rpsl_down1 (Table S1), and transformed into E. coli W3110 containing the pRed/ET amp plasmid (obtained from the "Quick and Easy E. coli Gene Deletion Kit" (Cambio)). ii.
Strain LAF-007 (E73Q) was generated directly from LAF-001 via a four-part process: a. The rpsL-neo cassette (template DNA was obtained from the "Counter-Selection BAC Modification Kit" (Cambio)) flanked by hyaB sequences (amplified using the rpsL-neo primers hyaBe73q_neo_sense and hyaBe73q_neo_antisense and Q5 polymerase, NEB), was inserted into the hyaB gene, giving strain LAF-004. b. The rpsL-neo cassette was replaced by a version of hyaB containing DNA encoding for the E73Q mutation (amplified with the "overlap" primers hyaBe73q_olap_sense and hyaBe73q_olap_antisense and Q5 polymerase, NEB). This gave strain LAF-005. c. The rpsL-neo cassette (template DNA was obtained from the "Counter-Selection BAC Modification Kit" (Cambio)) flanked by hyaA sequences (amplified using the rpsL-neo primers hyaA_histag_neo_sense and hyaA_his2_neo_antisense and Q5 polymerase, NEB), was inserted into the hyaA gene to give strain LAF-006. d. The rpsL-neo cassette was replaced by a version of hyaA containing DNA encoding a C-terminal polyhistidine tag (amplified with the "overlap"primers hyaA_his2_olap_sense and hyaA_his2_olap_antisense and Q5 polymerase, NEB), yielding strain LAF-007. iii.
A strain used to express "Native" enzyme, containing polyhistidine tagged E. coli hydrogenase-1 (strain LAF-003), was constructed from strain LAF-001 using the latter two processes from step ii. iv.
The LAF-003 strain was used as the parent strain to generate LAF-008 strain (with the rpsL-neo cassette inserted into hyaB close to the coding region of amino acid 73 using primers hyaBe73a_neo_sense and hyaBe73a_neo_antisense). Strain LAF-008 was then mutated to give LAF-009 (E73A), LAF-010 (E73K) and LAF-011 (E73N) via removal of the cassette using the relevant overlap primers (Table S1).
Successful insertion of the correct single site mutation was verified by amplification of the region of the chromosome and sequencing (GATC Biotech) as shown in Fig. S1. Table S1 lists full details of all the primers, and  Table S2 gives the strain information.

Strain name Genotype Reference
This work

Protein purification protocol
The purification protocol is modified from Lukey et al. 9 10 mL starter cultures were grown overnight in LB plus streptomycin (50 µg mL -1 ) at 37 °C with shaking. Of this, 3 mL was used to inoculate 6 L prewarmed bottles of LB, each containing 1 % glycerol, 0.4 % sodium fumarate, and 50 µg mL -1 streptomycin. These were grown overnight at 37 °C until an OD of ~1 was reached.
The cells were harvested by spinning at 5500 g for 15 min on a Beckman Avanti centrifuge (JA10 rotor) at 4 °C. Pellets (~50 g total) were resuspended in 125 mL chilled buffer (100 mM Tris pH 7, 300 mM NaCl) before the addition of sucrose (20% w/v) followed by stirring for >45 min at 4 °C. This mixture was then centrifuged at 6500 g for 20 min on a Beckman Avanti centrifuge (JA25.50 rotor) at 4 °C. The supernatant was once again discarded and the pellet was resuspended in 300 mL ice-cold pure water (Purite). This mixture was stirred in the fridge for 30 min before making the solution up to 100 mM Tris pH 7, 300 mM NaCl and 9% Triton X-100 and adding EDTA-free protease inhibitors, DNAse and lysozyme. This solution was stirred overnight in the fridge to facilitate solubilisation. The mixture was then divided between 150 mL sample pots and each aliquot was sonicated on ice for 10 x 30 s. The mixture was once again combined and centrifuged at 20000 g for 30 min on a Beckman Avanti centrifuge (JA25.50 rotor) at 4 °C. The supernatant was collected and diluted to 600 mL with 100 mM Tris pH 7, and the mixture was made up to 50 mM imidazole in preparation for loading onto the column.
The solubilised protein was loaded onto a 5 mL HisTrap Ni affinity column (GE Healthcare) using an Akta Start, the column was washed with 8 column volumes of buffer A (100 mM Tris, 150 mM NaCl, 50 mM imidazole, pH 7) and protein was eluted using a gradient elution (0-100% buffer B: 100 mM Tris, 150 mM NaCl, 1 M imidazole, pH 7). Presence of hydrogenase enzyme was confirmed using 10% SDS-PAGE (see Fig. S2). Fractions were dialysed overnight in the fridge (100 mM Tris pH7, 150 mM NaCl) in 0.2 micron dialysis tubing. An Amicon Ultra centrifugal filter with 30 kDa cutoff (Merk Millipore) was used to concentrate the protein.
A full spectrum of the purified enzymes (250-700 nm) was obtained on a UV-1061 (Shimadzu) UV-Vis spectrophotometer using a Semi-Micro Cell 108B-QS cuvette (Hellma Analytics). The concentration of the protein was calculated from the A 280 using the molar extinction coefficient 171,335 M -1 cm -1 as determined using the ExPASy ProtParam tool. 10

Assay method
All assays were performed in an anaerobic glovebox (Faircrest) filled with N 2 . A H 2 gas flow of 100 scc min -1 (controlled by a Sierra SmartTrack 50 mass flow controller) was bubbled through a solution of 0.1 mM methylene blue in pH 4.5 "mixed hydrogenase" buffer (see below) for a minimum of 1 hour. Enzyme solution was diluted 100-fold in the same buffer. 2 mL of the methylene blue solution was then placed inside a 3 mL cuvette containing a 1 mm magnetic bead, and the cuvette was placed in an LED spectrophotometer (built in-house, Department of Chemistry Electronic Workshops, University of York) on top of magnetic stirrer plate, which was turned on. A light blocking cuvette lid with a small injection hole was placed over the cuvette, and the solution was illuminated at a wavelength of 626 nm for 150 sec, before injecting 50 µL of diluted enzyme solution. The absorbance was then measured until the methylene blue had been completely decolourised. An inhouse determined methylene blue molar extinction coefficient of 28,000 µM -1 cm -1 was used to convert the fastest rate of absorbance change into H 2 oxidation rates, with protein concentrations determined as described above.
The quoted assay rates are calculated averaging three repeat experiments.

Electrochemistry method
All protein film electrochemistry was performed in an anaerobic glovebox (Faircrest), filled with nitrogen using previously described methods. 11 A threeelectrode configuration was used in a gas-tight glass cell. The graphite working electrode and platinum wire counter electrode were housed in the water-jacketed main body of the cell. The same "mixed hydrogenase" buffer solution described in the assay section (above) was used in all the electrochemical measurements at sufficient volume to cover the electrode connections. A reference side arm at ambient temperature, filled with 100 mM NaCl, connected to the main cell by a Luggin capillary, housed the saturated calomel reference electrode. Gases (BOC) were flowed through the experimental setup at the stated composition, with constant total rate of 100 scc min -1 (where scc is standard cubic cm) under the control of mass flow controllers (Smart-Trak; Sierra Installations) connected to the electrochemical cell.
The graphite electrode surface (electrodes manufactured in-house) was prepared for enzyme application by sanding with Norton P1200 abrasive sheets before application of 2 µl of enzyme which was adsorbed onto the surface for 30 s. Excess film was removed by abrasion with cotton wool before insertion of the electrode into the electrochemical cell. The working electrode was rotated using an Origatrod rotator (Origalys) at 4000 rev min -1 to allow an adequate supply of substrate and removal of product. A CompactStat potentiostat (Ivium Technologies) and the IviumSoft program were used to control the experiment.
Other experimental conditions are as reported for each data set.
A reference electrode correction factor of E(V vs SHE) = E(V vs Ref) + 0.265 V was determined from 1 mM methylene blue cyclic voltammetry calibration measurements at pH 7, 25  C, using a platinum working electrode and a value of E m, 7 = +19 mV vs SHE, calculated from published 12 reference data.

EPR spectroscopy Method
Small-volume EPR samples were prepared under anaerobic conditions in a Braun UniLab-plus glovebox (O 2 < 0.5 ppm, N 2 atmosphere), as described previously. 13 250 µL of the Hyd-1 E73Q (approx. 12 μM, in 50 mM HEPES, 50 mM sodium phosphate, 150 mM sodium chloride, 30% (v/v) glycerol) was added to a water-jacketed glass electrochemical cell equipped with a 2 mm glass-encased Pt working electrode (both from Scientific Glassblowing Service, University of Southampton) and a Ag/AgCl mini-reference electrode (DRIREF-2, WPI). The enzyme solution, maintained at approximately 4 o C, was stirred continuously and the potential was monitored throughout the titration (EmSTAT3+, PalmSens). Redox mediators (1,2 naphthoquinone, phenazine methosulfate, methylene blue, indigotrisulfonate, 2-hydroxy-1,4naphthoquinone, benzyl viologen, methyl viologen) were added at a concentration of 30 µM each. The solution was adjusted to the desired potential by successive additions (0.1 -0.5 μL) of sodium dithionite or potassium ferricyanide using a microlitre syringe (SGE Analytical Science) and 9 µL aliquots of the resultant enzyme-containing solution were successively transferred to a 1.6 O.D. quartz EPR tube (Wilmad). EPR samples were flash-frozen in the glove box cold finger containing EtOH, cooled externally using a dry-ice/acetone bath. Frozen samples were transferred to liquid nitrogen for storage.
EPR measurements were performed using an X/Q-band Bruker Elexsys E580 Spectrometer (Bruker BioSpin GmbH, Germany) equipped with a closed-cycle cryostat (Cryogenic Ltd, UK). The magnetic field was calibrated at room temperature with a Bruker strong pitch sample (g = 2.0028). All measurements were carried out at 20 K in an X-band split-ring resonator module with 2 mm sample access (ER 4118X-MS-2W). The Q value, as reported by the built-in Q indicator in the Xepr programme (typically 700) as well as the microwave frequency was used as a guide to position each sample in the same position in the resonator. EPR measurement conditions were 2 mW power, 100 kHz modulation frequency and 1.0 mT modulation amplitude for all samples. The Nernst plot (Fig. 4) was obtained by plotting the height of the EPR signal at g = 1.97 (maximal intensity normalised to 1) as a function of potential. The reduction potential (+211 ± 10 mV) varies slightly depending on which peak of the superoxidised proximal cluster EPR signal is monitored as a function of potential. The individual EPR spectra are shown in Fig. S9.

Escherichia coli hydrogenase-1 E73A variant electrochemistry
Experiments on the E73A variant confirm that making this amino acid exchange in Escherichia coli hydrogenase-1 has the same effect as was previously reported 11 in a study on Salmonella hydrogenase-5. As summarised in Fig. S3, the catalytic activity of E73A is comparable to Native enzyme (top panels), while the O 2 tolerance is seriously impaired (lower panels). Note that the experimental conditions for the O 2 inhibition measurements were chosen to match those in the Salmonella study.

The impact of p(H 2 ) on enzyme activity
In order to determine the Michaelis constant for H 2 and the product inhibition constant, enzyme was adsorbed onto the electrode and then the cyclic voltammograms shown in Fig. S4 were measured at 5 mV s -1 under different concentrations of H 2 .  Fig. S5A shows the Hanes-Woolf plot used to determine the Michaelis constant for H 2 for both Native and E73Q variant Escherichia coli hydrogenase-1, following the previously published procedure. 9 Fig. S5B shows analysis of the product inhibition constant for H 2 production, again using previously published methodology. 9 The data in Fig. S5 has been extracted from the voltammograms in Fig. S4. Tables S3 and S4 summarise the results of analysing repeat voltammogram experiments.

The impact of pH on enzyme activity
Two different types of experiment were conducted to compare the impact of pH on Native Escherichia coli hydrogenase-1 and the E73Q variant. Fig. S6 shows 5 mVs -1 voltammograms measured by adsorbing enzyme onto the electrode and measuring activity at pH 4.5 under both 3% and 0% H 2 , the pH was then changed to 6.0 and the same measurements were repeated, before a final buffer exchange to pH 7.6, and experiments in both 3% and 0% H 2 . To overcome complications due to enzyme desorption, the voltammograms are normalised relative to the maximum current recorded under 3% H 2 at each pH. Accurate measurements of Eswitch, the potential at which the maximum rate of reductive reactivation of the Ni-B state occurs, 14 are not possible from the experiments shown in Fig. S6 and therefore the measurements shown in Fig. S7 were made. Fig. S8 summarises data extracted from both Fig. S6 and Fig. S7; the "Eonset" provides a measure of the impact of the E73Q amino acid exchange on the catalytic overpotential of the enzyme, 15 while the "Eswitch" data shows that the Native and E73Q enzymes do not differ in their reactivation profile.

Fig. S7
The potential was held at +0.465 V vs SHE in pH 4.5, or +0.375 V vs SHE in pH 6.0, or +0.285 V vs SHE in pH 7.6 for 5 hours before the potential was scanned to -0.535 V vs SHE at a scan rate of 0.1 mVs -1 . Other conditions: 3% H 2 in N 2 with total gas flow rate 100 scc min -1 ; electrode rotation rate 3000 rpm; 37 C.

Fig. S8
Filled shapes show how the potential at which the current crosses the zeroaxis (Eonset) in the 3% H 2 cyclic voltammograms in Fig. S5 changes relative to the reduction potential for the 2H + /H 2 couple calculated from the Nernst equation (E(2H + , H 2 ), grey line). Empty shapes show how Eswitch (determined from the maxima in the first derivative plot of experimental data shown in Fig. S6) changes as a function of pH.

EPR spectra
The EPR spectra which were analysed in order to extract the proximal cluster high-potential redox-transition midpoint potential are shown in Fig. S9. These experiments were conducted as detailed in the relevant methods section above.

Fig. S9
Potentiometric titration of Hyd-1 E73Q. Small-volume (9 µL) EPR samples were prepared at fixed potentials under anaerobic conditions, as described in section 5. The order in which samples were taken is also indicated by the numbers on the left. All EPR measurements were carried out at 20 K with 2 mW power, 100 kHz modulation frequency and 1.0 mT modulation amplitude. All spectra were baseline subtracted.