Primers, full plasmid maps and sequences are provided in supplementary materials. Plasmid pIY003 was constructed by PCR-amplification of the hydrogenase operon from A. macleodii genomic DNA. Two amplicons spanning the operon were produced where the sequences had overlapping ends with each other and to the plasmid pTRC-NS1 (the backbone plasmid from ), which was linearized by double digest with the restriction enzymes BamHI and EcoRI. These three DNA fragments were assembled using Gibson isothermal assembly , and transformed into DH10B (Invitrogen) cells by electroporation. After screening clones by PCR, candidate constructs were sequence-verified by Sanger DNA sequencing. Plasmids pIY004, bearing the H230C substitution and pIY006, bearing the P285C substitution were constructed similarly, except using PCR products generated from primers containing the appropriate DNA nucleotide substitutions; plasmid pIY007, bearing the doubly-substituted enzyme, was constructed by repeating the procedure used to generate pIY006 except using pIY004 plasmid as a template.
Plasmid pIY033 was constructed by moving the hydrogenase operon into a pDEST-23 vector backbone using Gibson isothermal assembly followed by re-amplification of this operon (now including T7 promoter and terminator) and recloning into the pTRC-NS1 plasmid by Gibson isothermal assembly. Successful construction was verified by Sanger sequencing. pIY038 was constructed by PCR amplification of a segment of pIY007 around the hynS ORF followed by Gibson isothermal assembly into plasmid pIY033 which was linearized by double digest with the BamHI and AgeI restriction enzymes. Construction of plasmid pRC41-4 will be discussed elsewhere but the sequence is provided in the Supplemental Information; pCM012 was constructed by PCR amplification of the second half of the hynS ORF from pIY038, followed by Gibson isothermal assembly into pRC41-4 which was linearized by double digest with BamHI and AgeI restriction enzymes. Successful construction of pIY038 and pCM012 was verified by Sanger sequencing initiated with PCR primers used to amplify the modified region. Plasmids pIY085 and pIY086 were constructed similarly to pCM012, with pIY003 and pIY007 as templates, respectively, except using two sequential rounds of amplification with two reverse primers together generating sequence encoding the strep-tag epitope –RSAWSHPQFEK and the sequence bridging the C-terminal stop codon of hynS and the beginning of hynL up to the AgeI restriction site. Successful construction of pIY085 and pIY086 was verified by Sanger sequencing initiated with the PCR primers used to amplify the modified region.
A. macleodii Hydrogenase Expression in E. coli
All four plasmids pIY003, pIY004, pIY006, pIY007 were transformed into E. coli strain FTD147 cells, an MC4100(DE3) strain with the three active E coli hydrogenases removed . Plasmids pIY033 and pIY038 were transformed into BL21(DE3)ΔH4 cells . To grow E. coli cultures used to prepare extracts, 25 mL of autoinducer media  supplemented with 0.1 mM NiCl2 and 40 μg mL-1 spectinomycin was inoculated with 100 μL of a starter culture of the appropriate strain. Following inoculation, the expression culture was grown in an unbaffled 250 mL Erlenmeyer flask, rotating at 200 rpm, under ambient room atmosphere at 30°C overnight (~10-14 hrs). Under these conditions, the culture grew to a density in the approximate range of OD600 6–7. Piloting experiments showed that growth in baffled flasks resulted in higher specific activity but was avoided to maintain comparability with previous experiments . This was then harvested by centrifugation; the pellet was separated from the supernatant media and resuspended after addition of 0.8 mL lysis buffer (10 mM Tris, pH 7; 1 mM DTT; 0.5 mM EDTA). Lysis was performed by sonication on wet ice for one minute using a microprobe sonicator (Branson, Sonifier Model 250) at setting “4”, 40% duty cycle. The lysate was cleared by centrifugation for 10 minutes at 16,800 x g, at 4°C, and the supernatant was reserved as the “cleared lysate”. Protein content in the cleared lysate was measured with Bradford reagent (Bio-Rad) and comparison against a BSA (NEB) standard curve.
A. macleodii Hydrogenase Expression in S. elongatus
Plasmids pRC41-4 and pCM012 were mobilized into S. elongatus PCC 7942 strain PW416 by conjugative transfer from E. coli as previously described [16, 44]. Genomic integration into neutral site I (NSI) was verified by PCR. Cyanobacterial cultures were grown in 50 mL Bg11 medium in baffled 250 mL flasks to an OD730 of approximately 0.5. The cultures were induced by addition of IPTG to a final concentration of 0.25 mM, and additionally 0.5 μM NiCl2 was supplemented. Cells were grown for an additional 24 h, centrifuged, resuspended in 800 μL sonication buffer and sonicated as for E. coli. The lysates were cleared as described above for E. coli and the supernatant was reserved as the “cleared lysate”. In vitro hydrogenase evolution assays were conducted as described below.
Evolution activity assay
Cleared lysate (0.2 mL) was added to 1.7 mL of methyl viologen assay buffer containing 1.5 mL deionized water, 0.1 mL 40 mg mL-1 methyl viologen (Aldrich), and 0.1 mL of 500 mM potassium phosphate, pH 7.0 (Gibco). The combined samples in 13 mL gastight vials was sealed using rubber septa (Aldrich), and sparged under argon (Westair) for 20 minutes to remove oxygen. After sparging, 0.1 mL of 2 M sodium dithionite (Aldrich) was added by syringe anoxically. The resulting assay solution was a 1:10 dilution of lysate containing 25 mM potassium phosphate, 8 mM methyl viologen, and 100 mM sodium dithionite, pH ~7.0. E. coli samples were incubated for 2 hours at 30°C, followed by gas chromatography (CP-3800, Varian) using a Fused Silica Molsieve 5A column (CP7537, Varian) of 100 μL samples taken from the vial headspace. Previous experiments [13, 14] and piloting experiments for this experiment demonstrated time-dependent linearity of hydrogen evolution over the course of ~20 hours, suggesting that enzyme quality is unlikely to have been affected by the reaction conditions, including damage by oxidized dithionite products. S. elongatus samples were prepared similarly but were incubated at 30°C overnight before GC analysis. Hydrogen peaks were identified and integrated to quantify hydrogenase yield by comparison to a standard curve prepared with pure hydrogen from which a specific activity was calculated based on total protein in the lysate. For plasmids pIY003, pIY004, pIY006, and pIY007, the evolution activity was measured for four biological replicates with three technical replicates for each expression. For plasmids pIY033, pIY038, the evolution activity was measured for three technical replicates, and for pRC41-4, and pCM012, the evolution activity was measured for three biological replicates.
Uptake activity assay
Cleared lysate (0.1 mL) was added to 0.9 mL of benzyl viologen assay buffer containing 0.05 mL of benzyl viologen (65 mg mL-1) and 0.05 mL of 0.5 M potassium phosphate buffer, pH 7.0, and 0.8 mL deionized water. The combined assay sample was prepared in a 2 mL round-rim disposable UV cuvette (Brandtech) and sealed with a rubber septum (Aldrich), coated with a small amount of vacuum grease (Fisher). The cuvette was sparged for 20 minutes under 10% hydrogen balanced by nitrogen (Westair). Cuvettes were then transferred to a multichannel spectrophotometer and monitored at 555 nm at 30°C. For all four variants, four instances of expression were conducted with four replicates for each instance, such that three of these replicate experiments appeared to have self-consistent progress curves without prejudice to the data processing. Reaction rates were identified by fitting a linear regression to the data between OD555 = 1 to OD555 = 3.
Thermal exposure assay
The assay was conducted as the evolution assay, except cleared lysate was further incubated at 85°C for one hour and centrifuged a second time for 10 minutes at 16,800 x g and 4°C to clear additional insoluble debris prior to addition to MV buffer. Equivalent volumes of heat-treated and control-treated cleared lysate were added to assay buffer, and total hydrogen evolution activities without normalizing to protein concentration were used for subsequent calculations. Tolerance to thermal exposure was expressed by dividing the mean of three technical replicates of the denatured sample by the mean of three technical replicates of the undenatured sample. This process was subjected to three biological replicates; the geometric mean and geometric standard error are reported.
Oxygen exposure assay
The H-D exchange assay to test sensitivity to oxygen exposure was performed as follows: cleared lysate (0.1 mL) was added to 0.8 mL of mass spectrometry buffer containing 0.05 mL of 0.5 M potassium phosphate buffer, pH 7.0 (Gibco), and 0.75 mL deionized water. The samples were placed in 10 mL scintillation vials, and capped with new rubber septa. Vials were sparged for 10 minutes under pressurized argon. A polyimide-coated fused silica capillary, ID 50 μm OD 220 μm (Scientific Instrument Services) adapted for mass spectrometry was then inserted into the vial headspace followed by mass spectrometric monitoring (Omnistar, Pfeiffer). The microcapillary substituted for the standard metal probe supplied with this instrument samples very low quantities of total gas making up an inconsequential amount of total gas volume for the duration of the experiment. Hydrogen (100%, Westair) was injected to a final concentration of 10% hydrogen headspace and verified by mass spectrometry. For microaerobic samples, air was injected to create a final 1% oxygen mixture. Anaerobic samples were incubated for 2h at 30°C to ensure quantitative enzyme recovery, followed by injection of 0.1 mL D2O, followed by monitoring of the developing m/z = 3, until χHD > 1000 ppm. Microaerobic samples, prepared in parallel, were then injected with 0.1 mL D2O followed by overnight mass spectrometric monitoring at m/z = 3. Oxygen tolerance was measured by filtering spikes from the progress curve, calculating enzyme rate and normalizing the slope of the microaerobic activity against the slope of the anaerobic activity. This process was subjected to three biological replicates; the geometric mean and geometric standard errors are reported.
Cleared lysate from E. coli strains containing plasmid pIY085 and pIY086 was further purified using Streptactin magnetic bead resin (Qiagen). Cleared lysate was prepared as above but in NP buffer (50 mM sodium phosphate, pH 8.0, 300 mM NaCl) with 1 mM dithiothreitol and NeutrAvidin (Thermo) at 10 μg mL-1 final concentration. Streptactin resin (100 μl) was combined with cleared lysate (400 μl) and incubated at 4°C for 1 h with end-over-end rotation. The supernatant was removed and the resin was washed three times for five minutes each at room temperature in NP buffer containing 0.05% sodium dodecyl sulfate (SDS). This was followed by two washes in NP buffer supplemented with 0.01% Tween 20 as above. Final elution was performed at room temperature in 100 μl NP buffer supplemented with 0.01% Tween 20 and 10 mM biotin.
Lysates were diluted to 5 mg mL-1 total protein. One lysate from an overexpression of WT hydrogenase was further treated by denaturation at 65°C for 10 minutes and clearing by centrifugation, at 20,000 x g for 10 minutes. 25 μL of 4x NuPAGE SDS-PAGE loading buffer (Invitrogen) was added to 75 μL aliquots, followed by boiling for 10 minutes. These samples were frozen at −80°C and on a subsequent day thawed and 10 μL was separated onto a 10% NuPAGE Bis-Tris gel with the NuPAGE MOPS-SDS running buffer system (Invitrogen). Gel electrophoresis was performed at 150 V for 5 hours at 4°C, and the gel was cut separating the gel into two halves at approximately 75 kDa. The top gel was stained using the Sypro Ruby kit (Invitrogen) as a total protein loading control; the bottom gel was transferred onto nitrocellulose using NuPage transfer buffer (Invitrogen) with 30% MeOH for 10 hours at 10 mA constant current. These samples were then hybridized using rabbit-anti-HynL primary antibody  and goat-anti-rabbit-DyLight secondary antibody (Thermo), and imaged using a Typhoon fluorescence scanning imager (GE).
For gels associated with protein purification, samples were prepared as described above, but for cleared lysate, approximately 5 μg was loaded per lane and for purified proteins, approximately 0.5 μg protein was loaded per lane. Electrophoresis of the gels for western blot analysis of Strep-tagged HynS was only performed for 1 h at 150 V. Monoclonal antibody to the strep-tag antigen (Qiagen) was used with goat-anti-mouse-DyLight secondary antibody (Thermo) and imaged as above.