Binding of exogenous cyanide reveals new active-site states in [FeFe] hydrogenases

[FeFe] hydrogenases are highly efficient metalloenyzmes for hydrogen conversion. Their active site cofactor (the H-cluster) is composed of a canonical [4Fe-4S] cluster ([4Fe-4S]H) linked to a unique organometallic di-iron subcluster ([2Fe]H). In [2Fe]H the two Fe ions are coordinated by a bridging 2-azapropane-1,3-dithiolate (ADT) ligand, three CO and two CN− ligands, leaving an open coordination site on one Fe where substrates (H2 and H+) as well as inhibitors (e.g. O2, CO, H2S) may bind. Here, we investigate two new active site states that accumulate in [FeFe] hydrogenase variants where the cysteine (Cys) in the proton transfer pathway is mutated to alanine (Ala). Our experimental data, including atomic resolution crystal structures and supported by calculations, suggest that in these two states a third CN− ligand is bound to the apical position of [2Fe]H. These states can be generated both by “cannibalization” of CN− from damaged [2Fe]H subclusters as well as by addition of exogenous CN−. This is the first detailed spectroscopic and computational characterisation of the interaction of exogenous CN− with [FeFe] hydrogenases. Similar CN−-bound states can also be generated in wild-type hydrogenases, but do not form as readily as with the Cys to Ala variants. These results highlight how the interaction between the first amino acid in the proton transfer pathway and the active site tunes ligand binding to the open coordination site and affects the electronic structure of the H-cluster.


tion

The DdHydAB C178
The DdHydAB C178A variant was generated by site directed mutagenesis using PCR with non-overlapping primers (forward primer: GCCTGTCCGGGTTGGCAA, reverse primer: AGACGTGAACTGCGGCA) in which the forward primer contained the mutagenic codon at the 5' end.Following PCR, template DNA was digested with DpnI restriction endonuclease, the blunt ended PCR product was gel-purified, circularized using T4 polynucleotide kinase and T4 DNA ligase, and used for transformation of NEB 10β competent E. coli cells, which were plated on LB-agar containing 35 μg mL -1 chloramphenicol.Single colonies were used to inoculate LB containing 35 μg mL -1 chloramphenicol and grown overnight at 37 o C. Cells were then harvested, DNA was extracted, purified and sent for sequencing.

variant was generated by si
e directed mutagenesis using PCR with non-overlapping primers (forward primer: GCCTGTCCGGGTTGGCAA, reverse primer: AGACGTGAACTGCGGCA) in which the forward primer contained the mutagenic codon at the 5' end.Following PCR, template DNA was digested with DpnI restriction endonuclease, the blunt ended PCR product was gel-purified, circularized using T4 polynucleotide kinase and T4 DNA ligase, and used for transformation of NEB 10β competent E. coli cells, which were plated on LB-agar containing 35 μg mL -1 chloramphenicol.Single colonies were used to inoculate LB containing 35 μg mL -1 chloramphenicol and grown overnight at 37 o C. Cells were then harvested, DNA was extracted, purified and sent for sequencing.

WT and C169A CrHydA1 1 , WT and C178A DdHydAB w WT and C169A CrHydA1 1 , WT and C178A DdHydAB were recombinantly expressed in E. coli BL21(DE3) ΔiscR as Strep-tagII fusion proteins in their apo-form (i.e.lacking the [2Fe] H subcluster) and purified by affinity chromatography, as previously described 2,3 (but without co-expression of the maturases).The [2Fe] H precursor (Et 4 N) 2 [Fe 2 (ADT)(CO) 4 (CN) 2 ] was synthesized according to literature procedures. 4Protein samples were artificially maturated with the [2Fe] H precursor as described. 3,5 riefly, for the artificial maturation apo-hydrogenases diluted in 100 mM Tris pH 8, 150 mM NaCl were mixed with (Et 4 N) 2 [Fe 2 (ADT)(CO) 4 (CN) 2 ] (2 eq. for CrHydA1, 5 eq.for DdHydAB) and incubated at room temperature (RT) for 1 h (CrHydA1), or at 35 °C for 50 h (DdHydAB).Samples were then buffer exchanged to 25 mM Tris pH 8, 25 mM KCl to remove excess of [2Fe] H precursor using a Sephadex G-25 desalting column.WT DdHydAB required an additional incubation (>48 h at RT) in an open vessel to allow reactivation of the CO-inhibited state. 3Protein maturation and handling were performed under dim light due to light-sensitivity of these enzymes (DdHydAB in particular). 6Concentrated protein samples were stored in gas-tight vials at -80 °C until use.As maturated CrHydA1 C169A slowly forms the H trans -like state over time, samples of this variant were used within one week of preparation.Sodium dithionite (NaDT) was excluded from all protein preparations to avoid contamination by its oxidation products. 7eparation of samples for spectroscopic measurements Protein samples were prepared in 25 mM Tris pH 8, 25 mM KCl under anaerobic conditions (2% H 2 in 98% N 2 ) or under air where specified.NaCN (1-5 mM), hexaammineruthenium (III) chloride (HAR) (100 μM -10 mM) and NaDT (10 mM) were added where specified in figures and captions.For the preparation of CrHydA1 C169A in the H trans -like state, 200 μM protein was mixed with 100 μM hexammineruthenium (III) (HAR) and 5 mM NaCN (or K 13 CN) and then buffer exchanged to 25 mM Tris pH 8, 25 mM KCl on a Sephadex G-25 desalting column to remove excess CN -and H-cluster degradation products.WT DdHydAB was prepared in the H inact state as previously described. 8 Spectroscopy IR spectra were recorded on a Bruker Vertex 80v FT-IR spectrometer equipped with a mercury cadmium telluride (MCT) detector cooled by liquid nitrogen.Samples were placed between CaF 2 windows separated by a Teflon spacer (50 μm) and sealed in home-built brass holders with rubber rings.Spectra were recorded in the double-sided, forward-backward mode, with a resolution of 2 cm -1 , an aperture setting of 3 mm and scan velocity of 20 kHz.FTIR data were processed using home-written routines in the MATLAB environment.
re recombinantly expressed in E. coli BL21(DE3) ΔiscR as Strep-tagII fusion proteins in their apo-form (i.e.lacking the [2Fe] H subcluster) and purified by affinity chromatography, as previously described 2,3 (but without co-expression of the maturases).The [2Fe] H precursor (Et 4 N) 2 [Fe 2 (ADT)(CO) 4 (CN) 2 ] was synthesized according to literature procedures. 4Protein samples were artificially maturated with the [2Fe] H precursor as described. 3,5 riefly, for the artificial maturation apo-hydrogenases diluted in 100 mM Tris pH 8, 150 mM NaCl were mixed with (Et 4 N) 2 [Fe 2 (ADT)(CO) 4 (CN) 2 ] (2 eq. for CrHydA1, 5 eq.for DdHydAB) and incubated at room temperature (RT) for 1 h (CrHydA1), or at 35 °C for 50 h (DdHydAB).Samples were then buffer exchanged to 25 mM Tris pH 8, 25 mM KCl to remove excess of [2Fe] H precursor using a Sephadex G-25 desalting column.WT DdHydAB required an additional incubation (>48 h at RT) in an open vessel to allow reactivation of the CO-inhibited state. 3Protein maturation and handling were performed under dim light due to light-sensitivity of these enzymes (DdHydAB in particular). 6Concentrated protein samples were stored in gas-tight vials at -80 °C until use.As maturated CrHydA1 C169A slowly forms the H trans -like state over time, samples of this variant were used within one week of preparation.Sodium dithionite (NaDT) was excluded from all protein preparations to avoid contamination by its oxidation products. 7eparation of samples for spectroscopic measurements Protein s

ples were prepared in 25 mM Tris pH 8, 25 mM KCl under
anaerobic conditions (2% H 2 in 98% N 2 ) or under air where specified.NaCN (1-5 mM), hexaammineruthenium (III) chloride (HAR) (100 μM -10 mM) and NaDT (10 mM) were added where specified in figures and captions.For the preparation of CrHydA1 C169A in the H trans -like state, 200 μM protein was mixed with 100 μM hexammineruthenium (III) (HAR) and 5 mM NaCN (or K 13 CN) and then buffer exchanged to 25 mM Tris pH 8, 25 mM KCl on a Sephadex G-25 desalting column to remove excess CN -and H-cluster degradation products.WT DdHydAB was prepared in the H inact state as previously described. 8 Spectroscopy IR spectra were recorded on a Bruker Vertex 80v FT-IR

ectrometer equip
ed with a mercury cadmium telluride (MCT) detector cooled by liquid nitrogen.Samples were placed between CaF 2 windows separated by a Teflon spacer (50 μm) and sealed in home-built brass holders with rubber rings.Spectra were recorded in the double-sided, forward-backward mode, with a resolution of 2 cm -1 , an aperture setting of 3 mm and scan velocity of 20 kHz.FTIR data were processed using home-written routines in the MATLAB environment.

Cryogenic IR spectra (Figure S10) were recorded on a Bruker Tensor 27 FT-IR Cryogenic IR spectra (Figure S10) were recorded on a Bruker Tensor 27 FT-IR spectrometer linked to a Hyperion 3000 IR microscope equipped with a 20× IR transmission objective and a mercury cadmium telluride (MCT) detector as previously described. 9The temperature was set to 80 K by a liquid-N 2 -cooled cryo-stage (Linkam Scientific instruments).Illumination was performed using the focused beam of a collimated 530 nm LED with a power density of ca.500 mW cm -2 .Data were processed using the OPUS software version 7.5 from Bruker.
spectrometer linked to a Hyperion 3000 IR microscope equipped with a 20× IR transmission objective and a mercury cadmium telluride (MCT) detector as previously described. 9The temperature was set to 80 K by a liquid-N 2 -cooled cryo-stage (Linkam Scientific instruments).Illumination was performed using the focused beam of a collimated 530 nm LED with a power density of ca.500 mW cm -2 .Data were processed using the OPUS software version 7.5 from Bruker.


EPR spectroscopy

For X-band EPR spectroscopy, samples (~200 μL) were transferr

EPR spectroscopy
For X-band EPR spectroscopy, samples (~200 μL) were transferred to X-band quartz EPR tubes and frozen in liquid N 2 .In parallel, one aliquot of each sample was measured by IR spectroscopy at room temperature.Spectra were recorded on a Bruker ELEXSYS E500 CW X-band EPR spectrometer.The temperature was controlled with an Oxford ESR900 helium flow cryostat.Measurement parameters: modulation frequency 100 kHz; modulation amplitude 7.46 Gauss; time constant 81.92 ms; conversion time 81.92 ms.Temperature and microwave power were varied and are specified in figure legends.Spectra were processed using home-written routines in the MATLAB environment.Spectral simulations were performed with EasySpin package 10 in MATLAB.

to X-band quartz
EPR tubes and frozen in liquid N 2 .In parallel, one aliquot of each sample was measured by IR spectroscopy at room temperature.Spectra were recorded on a Bruker ELEXSYS E500 CW X-band EPR spectrometer.The temperature was controlled with an Oxford ESR900 helium flow cryostat.Measurement parameters: modulation frequency 100 kHz; modulation amplitude 7.46 Gauss; time constant 81.92 ms; conversion time 81.92 ms.Temperature and microwave power were varied and are specified in figure legends.Spectra were processed using home-written routines in the MATLAB environment.Spectral simulations were performed with EasySpin package 10 in MATLAB.


X-ray crystallography

The DdHydAB C178A protein was buffer exchanged into 10 mM Tris-

X-ray crystallography
The DdHydAB C178A protein was buffer exchanged into 10 mM Tris-HCl at pH 7.6 using a Sephadex G-25 desalting column prior to crystallization.Crystals were obtained using the sitting drop vapor diffusion method in 1 M lithium chloride, 0.1 M sodium acetate, and 30 % polyethylene glycol 4000 with a protein:precipitant ratio of 1:1.The DdHydAB C178A H inact -like state crystallized at 8 °C under aerobic conditions in the dark.We used a protein concentration of 25 mg/mL and a drop size of 2.0 μL.Crystallization of DdHydAB C178A H trans -like state was performed at 22 °C under anaerobic atmosphere (2% H 2 and 98% N 2 in a vinyl anaerobic chamber) in the dark.The protein concentration of the sample was 35 mg mL -1 and the drop size was 2.2 μL.Crystals were harvested using MicroMounts or MicroLoops after three days, transferred into a cryo-protectant solution consisting of 50% (w/v) aqueous polyethylene glycol 4000 and stored in liquid N 2 .Datasets were collected at an energy of 15 keV and 100 K at PETRA III, beamline P11, Deutsches Elektronensynchrotron (DESY, Hamburg, Germany).Data were processed using XDS 11 and data reduction was performed with AIMLESS 12 within the CCP4i2 13 suite.Molecular replacement was performed using Phaser 14 within the CCP4i2 suite with the coordinates of PDB ID 6SG2 as starting model.Structure refinement, model building, and validation were performed using REFMAC, 15 Coot and CCP4i2.The electron density maps were calculated using FFT 16 in CCP4 and the omit map was calculated with REFMAC using a model lacking the [2Fe] subcluster and additional CN -ligand.PyMOL was used to prepare the figures and to calculate the root-mean-square deviation (RMSD) of the Cα atoms of residues 2-397 using the align command with the number of cycles set to 0, thus, not including outlier rejection.The final models of DdH C178A in both states contained 98 % in the favored region, 2 % in the allowed region, and 0 % in outlier regions of the Ramachandran plot as defined by MolProbity. 17The atomic coordinates have been deposited with the Protein Data Bank, Research Collaboratory for Structural Bioinformatics at Rutgers University (PDB ID: 8BJ7 for DdHydAB C178A in the H inact -like state and 8BJ8 for DdHydAB C178A in the H trans -like state).

l at pH 7.6 using a Se
hadex G-25 desalting column prior to crystallization.Crystals were obtained using the sitting drop vapor diffusion method in 1 M lithium chloride, 0.1 M sodium acetate, and 30 % polyethylene glycol 4000 with a protein:precipitant ratio of 1:1.The DdHydAB C178A H inact -like state crystallized at 8 °C under aerobic conditions in the dark.We used a protein concentration of 25 mg/mL and a drop size of 2.0 μL.Crystallization of DdHydAB C178A H trans -like state was performed at 22 °C under anaerobic atmosphere (2% H 2 and 98% N 2 in a vinyl anaerobic chamber) in the dark.The protein concentration of the sample was 35 mg mL -1 and the drop size was 2.2 μL.Crystals were harvested using MicroMounts or MicroLoops after three days, transferred into a cryo-protectant solution consisting of 50% (w/v) aqueous polyethylene glycol 4000 and stored in liquid N 2 .Datasets were collected at an energy of 15 keV and 100 K at PETRA III, beamline P11, Deutsches Elektronensynchrotron (DESY, Hamburg, Germany).Data were processed using XDS 11 and data reduction was performed with AIMLESS 12 within the CCP4i2 13 suite.Molecular replacement was performed using Phaser 14 within the CCP4i2 suite with the coordinates of PDB ID 6SG2 as starting model.Structure refinement, model building, and validation were performed using REFMAC, 15 Coot and CCP4i2.The electron density maps were calculated using FFT 16 in CCP4 and the omit map was calculated with REFMAC using a model lacking the [2Fe] subcluster and additional CN -ligand.PyMOL was used to prepare the figures and to calculate the root-mean-square deviation (RMSD) of the Cα atoms of residues 2-397 using the align command with the number of cycles set to 0, thus, not including outlier rejection.The final models of DdH C178A in both states contained 98 % in the favored region, 2 % in the allowed region, and 0 % in outlier regions of the Ramachandran plot as defined by MolProbity. 17The atomic coordinates have been deposited with the Protein Data Bank, Research Collaboratory for Structural Bioinformatics at Rutgers University (PDB ID: 8BJ7 for DdHydAB C178A in the H inact -like state and 8BJ8 for DdHydAB C178A in the H trans -like state).


Single crystal infrared microspectroscopy

Crystals from a batch of crystals of DdHydAB C178A obta

Single crystal infrared microspectroscopy
Crystals from a batch of crystals of DdHydAB C178A obtained under aerobic conditions were collected, transferred to homemade MgF 2 plates and immediately frozen in liquid nitrogen.The same crystals were later analyzed by resonance Raman spectroscopy (see below).IR spectra were recorded on a Bruker Tensor 27 FT-IR spectrometer linked to a Hyperion 3000 IR microscope equipped with a 20× IR transmission objective and a mercury cadmium telluride (MCT) detector as previously described.The temperature was set to 233 K utilizing a liquid-N 2 -cooled cryo-stage (Linkam Scientific instruments).The raw data were processed using the OPUS software version 7.5 from Bruker.

ed under aerobic conditions were collected
transferred to homemade MgF 2 plates and immediately frozen in liquid nitrogen.The same crystals were later analyzed by resonance Raman spectroscopy (see below).IR spectra were recorded on a Bruker Tensor 27 FT-IR spectrometer linked to a Hyperion 3000 IR microscope equipped with a 20× IR transmission objective and a mercury cadmium telluride (MCT) detector as previously described.The temperature was set to 233 K utilizing a liquid-N 2 -cooled cryo-stage (Linkam Scientific instruments).The raw data were processed using the OPUS software version 7.5 from Bruker.


QM/MM calculations

The quantum mechanics/molecular mechanics (QM/MM) model of the DdHydAB C178A varia

QM/MM calculations
The quantum mechanics/molecular mechanics (QM/MM) model of the DdHydAB C178A variant was based on a previous wild-type model 18 with the mutation Cys178 to Ala added.The minimum QM region defined included the [2Fe] H subcluster, the Ala178 amino acid close to the ADT ligand, and Cys382, the cysteine bridging [2Fe] H and [4Fe-4S] H (see Figure S3).The [4Fe-4S] H and the additional coordinating cysteine residues were added to the minimum QM region to obtain a larger model (named minimum + [4Fe-4S] H ). A larger (medium) model included 12 more residues to account for hydrogen bonding around the [2Fe] H : Ala109, Pro108, Ala107, Ser202, Pro203, Ile204, Met232, Lys237, Gly292, Ala293, Thr294, and Sol-2427 (water molecule), shown in Figure S3.

was based on a pre
ious wild-type model 18 with the mutation Cys178 to Ala added.The minimum QM region defined included the [2Fe] H subcluster, the Ala178 amino acid close to the ADT ligand, and Cys382, the cysteine bridging [2Fe] H and [4Fe-4S] H (see Figure S3).The [4Fe-4S] H and the additional coordinating cysteine residues were added to the minimum QM region to obtain a larger model (named minimum + [4Fe-4S] H ). A larger (medium) model included 12 more residues to account for hydrogen bonding around the [2Fe] H : Ala109, Pro108, Ala107, Ser202, Pr 203, Ile204, Met232, Lys237, Gly292, Ala293, Thr294, and Sol-2427 (water molecule), shown in Figure S3.

Calculations were performed with the QM/MM code, ASH, developed by R. Björnsson. 19ASH has an interface to the OpenMM molecular mechanics library and the ORCA quantum chemistry code.Standard electrostatic embedding with linkatoms and charge-shifting was used in the QM/MM calculations.The MM part were described using a modified CHARMM36 force-field 20 (with a simple nonbonded model for the metal clusters) with periodic boundary conditions enabled.The ORCA code 21 was utilized for the QM calculations.The r 2 SCAN 22 density functional was used for the QM-part which has been shown to work well for both open-shell and closed-shell Fe-S dimers according to a recent benchmarking study. 23The Split-RI-J approximation 24 and with a decontracted auxiliary basis set 25,26 (named SARC/J in ORCA) was used to speed up the Coulomb integral evaluations.The ZORA scalar relativistic Hamiltonian 27,28 was used with relativistically recontracted basis sets: the ZORA-def2-TZVP 25,29 basis set was used on [2Fe] H , the extra ligand on Fe d , and the [4Fe-4S] H cubane (when included in the QM-region), and other atoms were calculated with ZORA-def2-SVP.The D4 dispersion correction 30,31 was used in all calculations.A partial Hessian approach, 32 was used to calculate numerical vibrational frequencies, with two Fe ions, all CO, and all CN -groups included in the definition of the partial Hessian.IR int Calculations were performed with the QM/MM code, ASH, developed by R. Björnsson. 19ASH has an interface to the OpenMM molecular mechanics library and the ORCA quantum chemistry code.Standard electrostatic embedding with linkatoms and charge-shifting was used in the QM/MM calculations.The MM part were described using a modified CHARMM36 force-field 20 (with a simple nonbonded model for the metal clusters) with periodic boundary conditions enabled.The ORCA code 21 was utilized for the QM calculations.The r 2 SCAN 22 density functional was used for the QM-part which has been shown to work well for both open-shell and closed-shell Fe-S dimers according to a recent benchmarking study. 23The Split-RI-J approximation 24 and with a decontracted auxiliary basis set 25,26 (named SARC/J in ORCA) was used to speed up the Coulomb integral evaluations.The ZORA scalar relativistic Hamiltonian 27,28 was used with relativistically recontracted basis sets: the ZORA-def2-TZVP 25,29 basis set was used on [2Fe] H , the extra ligand on Fe d , and the [4Fe-4S] H cubane (when included in the QM-region), and other atoms were calculated with ZORA-def2-SVP.The D4 dispersion correction 30,31 was used in all calculations.A partial Hessian approach, 32 was used to calculate numerical vibrational frequencies, with two Fe ions, all CO, and all CN -groups included in the definition of the partial Hessian.IR intensities and regular Raman activities (polarizability derivatives) were calculated using ORCA.

sities and regular Raman acti
ities (polarizability derivatives) were calculated using ORCA.


Resonance Raman spectroscopy

Small aliquots of CrHydA1 C169A in the H trans -like state (2.5 μL, 4 mM) were transferred to quartz plates and frozen in liquid nitrogen under an anaerobic atmosphere.For isotopic labelling experiments the samples were prepared with natural abundance NaCN and 13 C-labeled KCN (see IR spectra in Figure 6) then exposed to air for 5 min at 10 °C to form the H inact -like state and frozen.Resonance Raman spectra were recorded on a LabRam HR-800 Jobin Yvon confocal Raman spectrometer connected to a liquid-N 2cooled charge-coupled device as previously described. 9,33 he 514 nm emission line of an Ar + -ion laser with 2 mW power was used for excitation.Immediately before measurements, the sample was thawed at 10 °C and exposed to atmospheric oxygen for 5 minutes to form the H inact -like state, then frozen again, setting the temperature to 80 K by using by a liquid-N 2 -cooled cryo-stage (Linkam Scientific instruments).Data were processed using OPUS software version 7.5 from Bruker.Labels refers to the clockwise sequence of the three groups from the exogenous position.The calculations were p

Resonance Raman spectroscopy
Small aliquots of CrHydA1 C169A in the H trans -like state (2.5 μL, 4 mM) were transferred to quartz plates and frozen in liquid nitrogen under an anaerobic atmosphere.For isotopic labelling experiments the samples were prepared with natural abundance NaCN and 13 C-labeled KCN (see IR spectra in Figure 6) then exposed to air for 5 min at 10 °C to form the H inact -like state and frozen.Resonance Raman spectra were recorded on a LabRam HR-800 Jobin Yvon confocal Raman spectrometer connected to a liquid-N 2cooled charge-coupled device as previously described. 9,33 he 514 nm emission line of an Ar + -ion laser with 2 mW power was used for excitation.Immediately before measurements, the sample was thawed at 10 °C and exposed to atmospheric oxygen for 5 minutes to form the H inact -like state, then frozen again, setting the temperature to 80 K by using by a liquid-N 2 -cooled cryo-stage (Linkam Scientific instruments).Data were processed using OPUS software version 7.5 from Bruker.Labels refers to the clockwise sequence of the three groups from the exogenous position.The calculations were preformed using the medium QM region (see Figure S3).The model with apical-CN -on Fe d with the other CO/CN -in their regular positions (CN-CN-CO) is the most favorable for both ADT and ADTH models.

QM region (see Figure
S3).The model with apical-CN -on Fe d with the other CO/CN -in their regular positions (CN-CN-CO) is the most favorable for both ADT and ADTH models.


Supplementary Tables

The trends in energies for can be rationalized by more favorable hydrogen-bonding involving apical CN - and ADT (especially when protonated).Also, a Lys237 cation (not shown) donates a hydrogen bond with the ligand pointing away from the viewer, which is more favorable with a CN -ligand in this position (CO-CN-CN and CN-CN-CO models).   CN -in the H inact -like state are displayed in black and gray respectively, with the isotopic shifts from 12 C to 13 C displayed as red numbers.RR spectra were measured at 80 K using the 514 nm line of an Ar + laser for excitation.The small shifts to lower frequency are in line with a slightly weaker metal ligand vibration originating from the heavier 13 C atoms bound terminal at the [2Fe] H subcluster.The spectra were normalized to modes at 622 cm -1 and 644 cm -1 that correspond to the amino acid side-chain vibrations of phenylalanine and tyrosine, respectively. 38Calculated Raman spectra, based on the DdHydAB C178A model, for (B) ADTH and (C) ADT models, using the medium QM region.Calculated frequencies were scaled to match the most easily identifiable peak on the experimental spectrum: the intense 587 cm -1 peak.This was done by dividing the experimental and corresponding calculated frequencies of the exp.587 cm -1 peak.This resulted in scaling factors of 0.945 (ADTH) and 0.938 (ADT) which were then applied to all frequencies.The mass of the carbon in the apical CN -ligand was either 12 or 13 u.Regular Raman activities were calculated from polarizability derivatives.The experimental isotope shift on the spectra is overall well reproduced (some overestimation seen).The experimental peak at 603 cm -1 is reproduced computationally in the ADT model at 605 cm -1 (c) but not with the ADTH model (b).This peak is primarily associated with the bending of the bridging CO, which is likely perturbed to different degrees via a transeffect relating to different degrees of CN -binding in ADTH model vs. ADT model.Each vibrational transition was broadened with a Gaussian with a width of 10 cm -1 (FWHM).The broad experimental peak at 455 cm -1 was calculated as either an intense peak at 405 cm -1 (ADTH) or a split peak at 414-429 cm -1 .

Figure S2 .
S2
Figure S2.CW X-band EPR spectrum of WT DdHydAB in the H trans state.The H trans state was prepared by treating DdHydAB with 1 mM HAR and 10 mM Na 2 S as described previously.8The experimental spectra (black) are overlaid with

Supplementary Tables
The trends in energies for can be rationalized by more favorable hydrogen-bonding involving apical CN - and ADT (especially when protonated).Also, a Lys237 cation (not shown) donates a hydrogen bond with the ligand pointing away from the viewer, which is more favorable with a CN -ligand in this position (CO-CN-CN and CN-CN-CO models).   CN -in the H inact -like state are displayed in black and gray respectively, with the isotopic shifts from 12 C to 13 C displayed as red numbers.RR spectra were measured at 80 K using the 514 nm line of an Ar + laser for excitation.The small shifts to lower frequency are in line with a slightly weaker metal ligand vibration originating from the heavier 13 C atoms bound terminal at the [2Fe] H subcluster.The spectra were normalized to modes at 622 cm -1 and 644 cm -1 that correspond to the amino acid side-chain vibrations of phenylalanine and tyrosine, respectively. 38Calculated Raman spectra, based on the DdHydAB C178A model, for (B) ADTH and (C) ADT models, using the medium QM region.Calculated frequencies were scaled to match the most easily identifiable peak on the experimental spectrum: the intense 587 cm -1 peak.This was done by dividing the experimental and corresponding calculated frequencies of the exp.587 cm -1 peak.This resulted in scaling factors of 0.945 (ADTH) and 0.938 (ADT) which were then applied to all frequencies.The mass of the carbon in the apical CN -ligand was either 12 or 13 u.Regular Raman activities were calculated from polarizability derivatives.The experimental isotope shift on the spectra is overall well reproduced (some overestimation seen).The experimental peak at 603 cm -1 is reproduced computationally in the ADT model at 605 cm -1 (c) but not with the ADTH model (b).This peak is primarily associated with the bending of the bridging CO, which is likely perturbed to different degrees via a transeffect relating to different degrees of CN -binding in ADTH model vs. ADT model.Each vibrational transition was broadened with a Gaussian with a width of 10 cm -1 (FWHM).The broad experimental peak at 455 cm -1 was calculated as either an intense peak at 405 cm -1 (ADTH) or a split peak at 414-429 cm -1 .

Figure S2 .
Figure S2.CW X-band EPR spectrum of WT DdHydAB in the H trans state.The H trans state was prepared by treating DdHydAB with 1 mM HAR and 10 mM Na 2 S as described previously.8The experimental spectra (black) are overlaid with spectral simulations (dashed red lines) and the component spectra are shown underneath.The orange trace (Component 1) corresponds to the H trans state, while the dark cyan trace (Component 2) corresponds to the H ox -CO state.The asterisk indicates a small contribution from the H ox state.The g values for the H trans state (g = 2.064, 1.964, 1.891) are in excellent agreement with those reported previously by Albracht et al (g = 2.06, 1.96, 1.89).37EPR experimental conditions: microwave frequency = 9.63 GHz; microwave power = 100 mW; temperature = 30 K.

Figure S3 .
Figure S3.The definition of different quantum mechanical (QM) regions and the effect on calculated IR spectra in the H inact state.The scaling factor for vibrational frequencies is 0.964.Inclusion of the [4Fe-4S] H cubane in the QM region (minimum + [4Fe-4S] H region) results in very small shifts of the CN -modes and only slightly larger shifts for the CO modes.The larger QM region ("medium") includes protein residues around the [2Fe] H to better account for hydrogen bonding: Ala109, Pro108, Ala107, Ser202, Pro203, Ile204, Met232, Lys237, Gly292, Ala293, Thr294, and Sol-2427 (water molecule); calculations with this region results in larger shifts, particularly for the CO modes with the CN modes less affected.

Figure S4 .
Figure S4.The relative QM/MM energies for different rotameric structures of CN --bound to Fe d in the Hcluster models in both ADT (top) and ADTH (bottom) protonation states in the [Fe(II)Fe(II)] H redox state.Labels refers to the clockwise sequence of the three groups from the exogenous position.The calculations were preformed using the medium QM region (see FigureS3).The model with apical-CN -on Fe d with the other CO/CN -in their regular positions (CN-CN-CO) is the most favorable for both ADT and ADTH models.The trends in energies for can be rationalized by more favorable hydrogen-bonding involving apical CN - and ADT (especially when protonated).Also, a Lys237 cation (not shown) donates a hydrogen bond with the ligand pointing away from the viewer, which is more favorable with a CN -ligand in this position (CO-CN-CN and CN-CN-CO models).

Figure S5 .
Figure S5.IR and RR spectra of DdHydAB C178A crystals in the H inact -like state.A) IR spectra of DdHydAB C178A crystals prepared under aerobic conditions recorded at 233 K in comparison to the corresponding spectrum in solution measured at 293 K.These data clearly confirm the presence of the H inact -like state in the protein crystals.Small phase-dependent (crystal vs. solution) shifts in the band positions are likely related to temperature-dependent changes or crystal packing effects, which may cause e.g.small rearrangements in the hydrogen-bonding networks.B) Corresponding RR spectra of the same DdHydAB C178A crystals recorded at 80 K via excitation by the 514 nm line of an Ar + laser in comparison to the respective spectrum in solution.The assignment of the H inact -like state to the crystal is additionally verified by an almost identical spectral signature in the region of the Fe-CO/CN metal ligand modes of the protein crystals and solution.The spectra were normalized to modes at 622 cm -1 and 644 cm -1 that correspond to the amino acid side-chain vibrations of phenylalanine and tyrosine, respectively.38

Figure S6 .
Figure S6.Overall view of DdHydAB C178A H inact -like with schematic representation of the [4Fe-4S]clusters and H-cluster.

Figure S7 .
Figure S7.Superposition of the DdHydAB C178A H inact -like structure (gray) with the DdHydAB WT H inact structure (blue) showing a close-up view of the active site.

Figure S8 .
Figure S8.A) Superposition of the DdHydAB C178A H inact -like structure (gray) with the DdHydAB WT H inact structure (blue) and with the DdHydAB C178A H trans -like structure (magenta).

Figure S9 .
Figure S9.Crystal structure of the DdHydAB C178A variant in the H inact -like state with an omit map calculated for the [2Fe] H subsite. (A) An omit map (blue mesh, contoured at 1.0 σ) is shown for the [2Fe] H subsite including the CN -ligand in the open coordination site.(B) An omit map (blue mesh, contoured at 1.0 σ) is displayed for all CO and CN -ligands.(C) The same omit map (blue mesh, contoured at 2.8 σ) as displayed in (B) with a different contour level supports the assignment of the nitrogen and oxygen atoms.The protein backbone is represented as cartoon, amino acid side chains and the [2Fe] H subsite are shown as sticks, [4Fe-4S] H clusters are shown as ball-and-stick model.

Figure S10 .
Figure S10.Cryogenic IR spectra of CrHydA1 C169A in the H inact -like and H trans -like state illuminated with blue light.IR spectra were recorded at 80 K in the dark (top), after 20 min illumination by the focused beam of a collimated power LED at 530 nm (middle) and the corresponding light-minus-dark difference spectrum.This illumination experiment revealed no obvious photosensitivity of the H trans -like and H inactlike states, rather excluding the binding of an additional terminal CO.

Figure S11 .
Figure S11.Comparison of experimental and calculated Raman spectra of CrHydA1 C169A, where the CN - bound apical at Fe d was isotopically labelled.A) Experimental RR spectra from the natural abundance CN - and13 CN -in the H inact -like state are displayed in black and gray respectively, with the isotopic shifts from12 C to13 C displayed as red numbers.RR spectra were measured at 80 K using the 514 nm line of an Ar + laser for excitation.The small shifts to lower frequency are in line with a slightly weaker metal ligand vibration originating from the heavier13 C atoms bound terminal at the [2Fe] H subcluster.The spectra were normalized to modes at 622 cm -1 and 644 cm -1 that correspond to the amino acid side-chain vibrations of phenylalanine and tyrosine, respectively.38Calculated Raman spectra, based on the DdHydAB C178A model, for (B) ADTH and (C) ADT models, using the medium QM region.Calculated frequencies were scaled to match the most easily identifiable peak on the experimental spectrum: the intense 587 cm -1 peak.This was done by dividing the experimental and corresponding calculated frequencies of the exp.587 cm -1 peak.This resulted in scaling factors of 0.945 (ADTH) and 0.938 (ADT) which were then applied to all frequencies.The mass of the carbon in the apical CN -ligand was either 12 or 13 u.Regular Raman activities were calculated from polarizability derivatives.The experimental isotope shift on the spectra is overall well reproduced (some overestimation seen).The experimental peak at 603 cm -1 is reproduced computationally in the ADT model at 605 cm -1 (c) but not with the ADTH model (b).This peak is primarily associated with the bending of the bridging CO, which is likely perturbed to different degrees via a transeffect relating to different degrees of CN -binding in ADTH model vs. ADT model.Each vibrational transition was broadened with a Gaussian with a width of 10 cm -1 (FWHM).The broad experimental peak at 455 cm -1 was calculated as either an intense peak at 405 cm -1 (ADTH) or a split peak at 414-429 cm -1 .

Figure S12 .
Figure S12.A comparison of calculated IR spectra when the exogenous ligand is modelled as CN - or O 2 using the QM/MM model with a singly protonated ADT ligand.Frequencies are scaled by 0.964.Optimization of an O 2 model results in spontaneous superoxide formation according to geometry and spin density distribution.

Table S1 -IR bands of relevant [FeFe] hydrogenase states observed in this and previous studies
This state was previously assigned to be H ox .We suggest that this state to actually have an electronic structural similar to H red (i.e. with ([4Fe-4S] H + -[Fe p (II)Fe d (I)] H ). In the H ox -CO state the two CO ligands bound to Fe d are strongly vibrationally coupled, giving rise to a higher energy symmetric stretching band and a lower energy antisymmetric stretching band. *