A De Novo Designed Artificial Metallopeptide Hydrogenase: Insights into Photochemical Processes and the Role of Protonated Cys

Abstract

Hydrogenase enzymes produce H₂ gas, which can be a potential source of alternative energy. Inspired by the [NiFe] hydrogenases, we report the construction of a de novo designed artificial hydrogenase (ArH). The ArH is a dimeric coiled coil where two cysteine (Cys) residues are introduced at tandem a/d positions of a heptad to create a tetrathiolato Ni binding site. Spectroscopic studies show that Ni binding significantly stabilizes the peptide producing electronic transitions characteristic of Ni-thiolate proteins. The ArH produces H₂ photocatalytically, demonstrating a bell-shaped pH-dependence on activity. Fluorescence lifetimes and transient absorption spectroscopic studies are undertaken to elucidate the nature of pH-dependence, and to monitor the reaction kinetics of the photochemical processes. pH titrations are employed to determine the role of protonated Cys on reactivity. Combined, we find that a fine balance between solution acidity and the electron transfer steps need to be maintained such that the yield of reduced photosensitizer can be maximized to produce the Ni^I-peptide and the Ni^II-H⁻ intermediate (Ni-R) is protonated by a Cys (pK_a~6.4) to produce H₂.

Graphical Abstract

An artificial Ni hydrogenase is designed within a de novo coiled coil peptide, which produces H₂ under light. Mechanistic studies employing time-resolved kinetics, transient absorption spectroscopy, and pH-titration experiments demonstrate that a suitable control of solution acidity need to be maintained to maximize the yield of Ni^I-peptide and to enable a Cys thiol to protonate the Ni^II-H⁻ intermediate for an optimal catalytic activity.

Introduction

A major challenge of the 21^st century is to attain energy security by mitigating the carbon footprint in the environment. This can be achieved by exploiting natural resources such as sunlight as an alternative energy source. In this regard, H₂ gas is a promising fuel that can be combusted to produce energy without causing environmental pollution.^[1] However, H₂ is not available in abundance from any natural resource. Therefore, it must be produced in an environmentally friendly and sustainable manner before using it as a fuel. One approach is to store solar energy in the form of chemical energy of H₂ by performing the reductive half of photosynthetic water splitting.^[2] Typically, a photosensitizer (PS) absorbs sunlight leading to charge separation and the transfer of electrons to a metal catalyst, in the presence of a sacrificial electron donor and a H⁺ source, to produce H₂ (2H⁺+2e⁻ ➔ H₂). The experimental conditions must be optimized such that each of the components can communicate with maximum efficiency.

Hydrogenase enzymes (H₂ases), found in many microorganisms are natural producers of H₂.^[3] The H₂ can also be oxidized reversibly, back to H⁺, in order to drive cellular processes. Pathways for H⁺ transfer, electron transfer (ET), and substrate/product channels facilitate this highly efficient and attractive enzymatic reaction. In [NiFe] H₂ases, the initial binding of substrates is thought to occur at the Ni site, which is coordinated to two terminal cysteine (Cys) and two bridging Cys residues that are shared with the Fe. A fifth bridging ligand is also bound to the Ni site, which is H⁻ in the proposed active states (Ni-R, Ni-C) and OH⁻ in the inactive states (Ni-B, Ni-A). The innate complexity, combined with its tendency towards deactivation and limited production yield have led to the development of biomolecular H₂ evolution catalysts where the metal site is encapsulated within a polypeptide scaffold.^[4] Such examples include assemblies of Ni catalysts with photosystem I,^[5] peptide-bound synthetic catalysts,^[6] replacing native cofactors with non-native cofactors,^[7] and modifying existing protein scaffolds to incorporate Ni.^[8]

Here we describe the de novo design as a complimentary approach to construct an artificial hydrogen evolving catalyst (ArH). The de novo metalloprotein design approach is distinct in that the precise self-assembling rules of peptide chains provide a unique handle to produce oligomers with controllable coordination properties.^[9] The de novo constructs serve as simpler functional analogs of complex metalloenzymes with the potential to provide a working view of the reaction mechanism in aqueous media. With the potential to expand the repertoire of chemical functionality supported in de novo designed constructs, an important area to be explored is to catalyze transformations relevant in alternative energy, e.g. artificial photosynthesis. A novel ArH is reported herein, based on the de novo design of two stranded coiled coils (2SCCs) featuring -Cys-Xxx-Xxx-Cys- motifs on each peptide strand to produce the putative Ni-tetrathiolate site of the [NiFe] H₂ases. The ArH produces H₂ photochemically, demonstrating a bell-shape activity profile with respect to solution pH. Photophysical and transient absorption spectroscopic (TAS) investigation reveals that a fine balance of the solution acidity needs to be maintained for optimum activity to maximize the yield of active metallopeptide catalyst and the availability of H⁺. The protonation state of Cys is deemed a critical determinant to H₂ production. Mechanistic implications derived from our study applicable to the Ni-R and Ni-C intermediates proposed in [NiFe] H₂ases are discussed.

Results and Discussion

Peptide design.

The parallel 2SCCs is chosen as the peptide scaffold to harbor the NiS₄ active site. The M(Cys)₄ binding motif is also prevalent in rubredoxin, DNA polymerase III, alcohol dehydrogenase, and the DNA repair protein Ada.^[10] The peptides are designed according to the repeat pattern of the heptad rule where an Ile at the a site and a Leu at the d site produces the parallel dimeric assembly.^[11] The hydrophobic collapse of these residues into specific knob-into-hole (KIH) packing drives the self-assembly.^[12] Gly residues at each of the termini are introduced, which are acetylated at the N-terminus and amidated at the C-terminus. The b and c positions of the heptad are selected as Ala to favor α-helix formation. The e and g position residues are chosen as oppositely charged Glu and Lys that form ion pairs to control the orientation and produce parallel assemblies. Three of the f sites are filled with polar residues to promote water solubility while in the last heptad, a Trp is introduced as a spectroscopic handle. An Asn is introduced at a site of the 3rd heptad to increase the specificity of the dimeric assembly by Asn-Asn side chain H-bond. Cys residues are placed at the a and d sites of the 2nd heptad to introduce the -Cys_a-Xxx-Xxx-Cys_d- chelate motif such that a Ni(Cys)₄ coordination can be harbored in the peptide, denoted hereafter as [2SCC-(I9CL12C)₂]. Another possibility is to introduce the Cys residues via the -Cys_d-Xxx-Xxx-Xxx-Cys_a- motif. However, the longer distance between the Cys in the latter motif may produce NiS₃-type structures instead of the intended NiS₄ coordination. Indeed, the a/d vs d/a Cys binding has been shown to control Cd(Cys)₄ vs Cd(Cys)₃ coordination in 3SCC systems.^[13] The Cd(Cys)₄ coordination has also been introduced in a 2SCC scaffold bearing the a/d bis-Cys chelate motif.^[14] We hypothesized that, analogously, a Ni(Cys)₄ site can be introduced in our system as well. The final sequence Ac-G-IAALEQKCAACEEKNAALEWKIAALEKK-G-CONH₂ (Table 1) was energy-minimized in CC Builder.^[15] The minimized model shows that the Cys site is nestled between two layers of Ile and Leu residues (Fig. 1). The N-C_α-C_β-S_γ dihedral angle in all 4 Cys residues is −70°. The a and d layer S_γ atoms are rotated by ~70° with respect to each other. The H-bond formed between the Asn residues (Fig. 1B) in the middle of the bundle has been shown to increase the specificity of helix-helix packing.^[16] A total of 5 layers of hydrophobic packing provides the driving force for self-assembly.

Table 1.

Peptide sequences used herein.

Peptide	Sequence
[2SCC(I9CL12C)2]	G-IAALEQK CAACEEK NAALEWK IAALEKK-G
[2SCC(I9C)2]	G-IAALEQK CAALEEK NAALEWK IAALEKK-G

Figure 1.

Overall design of [2SCC-(I9CL12C)₂]. A) The helical wheel diagram of the parallel dimers showing the hydrophobic core (green circles) and the inter-strand ionic interactions (red/blue circles). B) CC Builder model showing the hydrophobic residues at the core (Leu: wheat spheres; Ile: gray spheres); the H-bonding between two Asn residues shown as red dash. C) A view of the Cys site along the helical axis and D) perpendicular to the helical axis showing the relative orientation of the a (green) and d (cyan) site Cys residues.

Ni binding stabilizes the peptide.

The folding behavior of the peptide is probed by circular dichroism (CD) spectroscopy. No apparent concentration dependence on the mean residue ellipticity (MRE) is observed for either the unbound apo or the Ni^II-bound form of the peptide in the range of 25–50 μM (Fig. 2A,B). The data suggests that in this concentration range, no further association of the helices to higher ordered assembly is occurring, a conclusion which validates the specificity of the CC design. This result is corroborated by size exclusion chromatography, which shows that both the apo and [2SCC-Ni(I9CL12C)₂]²⁻ forms exist as dimers (Fig. S1). The apo form is 14% α-helical. In the presence of Ni^II, however, a significant increase in the overall α-helical character to 25% is observed where the peak at ~222 nm became more negative by ~2-fold compared to the apo peptide, with a concomitant red shift of the high energy peak to ~205 nm (Fig. 2C). This result confirms Ni^II binding and that the binding of Ni^II induces the α-helical character to the peptide. A similar behavior was observed for a dimeric peptide reported by Ogawa, which underwent Cd(II)-dependent coiled-coil formation^[14] with ~50% α-helix observed at 10-fold excess Cd(II) concentration.

Figure 2.

Concentration-dependent CD of apo (A) and Ni-bound peptide (B). C) Overlay of 30 μM apo (black) and [2SCC-Ni(I9CL12C)₂]²⁻ (red) showing that Ni-binding induces α-helical character. D) Thermal melts of the corresponding samples as in C. All data were collected in 5 mM phosphate pH 8.5.

Consistent with the CD data, the apo [2SCC-(I9CL12C)₂]⁴⁻ peptide shows characteristics of an unstable and primarily random coil structure in the thermal melt profiles, where no obvious unfolding transition is observed (Fig. 2D black curve). The Ni^II-bound form, in contrast, undergoes transition from the folded to unfolded state (Fig. 2D red curve) with a T_m of ~55°C. While dimeric CCs have been crystallized and shown to exist as well-folded helical assemblies with intact hydrophobic cores,^[11b] the lack of such behavior in the case presented here is not unexpected, given that substitution of hydrophobic a and d resides disrupts Van der Waals interactions leading to a loss of stability. Ni^II binding counteracts this effect and stabilizes the assembly. This is further substantiated by the single Cys variant [2SCC-I9C)₂]²⁻, which is 31% α-helical even in the apo form and the presence of Ni^II did not change the folding behavior (Fig. S2), indicating Ni^II does not bind to the single-Cys variant.

Electronic and NMR spectroscopy.

Ni^II binding is probed by UV-vis spectroscopy. At pH 8.5, a gradual addition of Ni^II to 30 μM [2SCC-(I9CL12C)₂]⁴⁻ results in the appearance of peaks in the UV to visible regions. Intense LMCT bands appear at 279 nm, 330 nm, 382 nm, and 478 nm (Fig. 3A) with molar absorptivity (Δε, obtained by subtraction of apo peptide) of 26500, 8500, 4400, and 800 M⁻¹cm⁻¹, respectively (Table 2).

Figure 3.

A) UV-vis titration spectra of 30 μM [2SCC-(I9CL12C)₂]⁴⁻ with Ni^II at pH 8.5. B) An energy plot of the metal-bound form with corresponding peak positions in nm. C) Plots of differential absorptivity vs equivalents of Ni^II/dimer. The solid lines represent fits to the data to extract binding constant.

Table 2.

Physical parameters.

Peptide	λmax (nm)	Δε (M−1cm−1)	KD (μM)	Tm (°C)	pK a
	279	26500
[2SCC(Ni(I9CL12C)2)]2−	330	8500	3±1	55	6.4±0.2
	382	4400
	478	800
	598	260
	654	130

Less intense d-d transitions centered at 598 nm and 654 nm with Δε of 260 and 130 M⁻¹cm⁻¹ (Fig. 3B, Table 2), respectively, are also present. These spectral features are characteristic of Ni-thiolate ligation. For example, in Ni^II-rubredoxin, where Ni^II binds to the four Cys from two -Cys-Xxx-Xxx-Cys- motifs in a T_d geometry, LMCT bands at 275 nm, 357 nm, and 448 nm are observed, while d-d bands appear at 670 and 720 nm.^[17] The stoichiometry of Ni^II to peptide is determined from metal titration, which shows 1 equivalent of Ni^II binding (Fig. 3C). From analysis of the binding curves,^[18] the K_d for Ni^II binding to [2SCC-(I9CL12C)₂]⁴⁻ is derived to be ~3 μM (Table 2). These features are absent when 1 equivalent of Ni^II is added to [2SCC-(I9C)₂]²⁻ (Fig. S3), which confirms that the observed electronic transitions of [2SCC-Ni(I9CL12C)₂]²⁻ are attributable to the metal-thiolate interactions and that all four Cys are involved to bind Ni^II via the -Cys-Xxx-Xxx-Cys- motif. Together, these data demonstrate that the designed 2SCC binds 1 equivalent of Ni^II with micromolar affinity at the desired metal site. ¹H NMR spectroscopy is further employed to examine the spin state of Ni in [2SCC-Ni(I9CL12C)₂]²⁻. The NMR spectra of water-suppressed apo and Ni-bound peptide show good chemical shift dispersion (Fig. S4). The metallated peptide has sharper peaks with some shifts in resonances compared to the apo form. This data further attests that metal binding stabilizes the peptide causing slight shifts in the resonances and giving rise to sharper peaks. The absence of hyperfine-shifted peaks suggest a diamagnetic ground state in a non-T_d geometry. The ¹H NMR of diamagnetic aryl-thiolate Ni^II complexes also do not show hyperfine-shifted resonances.^[19] In contrast, the T_d ground state of Ni^II-rubredoxin displays hyperfine-shifted resonances due to the triplet ground state of Ni^II in a T_d environment.^[17]

Photocatalytic hydrogen evolution.

The activity of [2SCC-Ni(I9CL12C)₂]²⁻ towards photocatalytic hydrogen evolution reaction (HER) is investigated using Ru(bpy)₃²⁺ (Ru^II) as the photosensitizer and ascorbic acid (H₂A) as a sacrificial electron donor. In this 3-component system, the solution acidity is a critical factor in determining the yield of photogenerated H₂.^[20] To test whether such a pH dependence exists in the biomolecular system described here, we monitored the H₂ production using 30 μM of the metallopeptide at varying solution pH. In the studied pH range of 1.9 to 7.4, a maximum H₂ production of 2.6 μmol is observed at pH 5.6 after ~2h (Fig. 4A–B, red) of visible light irradiation. The apo peptide (Fig. 4B, wine), free Ni^II (Fig. 4B, orange), or Ru^II alone (Fig. 4B, navy) produce little to no H₂ under identical conditions. The turn over frequency (TOF) follows a bell-shape pH profile with a maximum TOF of 32/h/mol (Fig. 4C) at pH 5.6, corresponding to a turn over number (TON) of ~44 over 2h. Ni-rubredoxin showed a TON of 32 photocatalytically^[8a] at similar concentrations (25 μM). Our previously reported ArH, NBP had a TON of 115 for light-induced H₂ production.^[8d] In other reported examples where the well-known DuBois catalyst has been encapsulated in photosystem I directly or via flavodoxin, the TON for photogenerated H₂ was 1.87×10³ and 2.8×10³, respectively.^[5a] Kinetic traces depicted in Fig. 4D indicate that increasing the concentration of [2SCC-Ni(I9CL12C)₂]²⁻ increases the yield of H₂. The initial rate of H₂ evolution at each concentration (Fig. 4D, inset) varies linearly, suggesting a first order dependence of catalyst concentration on the rate in the studied range of 7.5–90 μM. The effect of photosensitizer concentration on H₂ evolution is further studied by varying the concentration of Ru^II from 0.5–2 mM (Fig. S5). The initial rate of H₂ evolution increases linearly, suggesting a first order dependence of Ru^II concentration on the rate. The photostability of the sample is studied by drawing aliquots at different time intervals after illumination and measuring the absorption spectra (Fig. S6). No major change in the absorption spectra is observed during the time period of HER, suggesting that photodegradation of the sample is not occurring. Due to spectral overlap, absorption of the ArH in the intense Ru^II region cannot be distinguished.

Figure 4.

A) Representative pH-dependent photocatalytic H₂ production by 30 μM [2SCC-Ni(I9CL12C)₂]²⁻. B) Plots of time-dependent H₂ production at pH 5.6 by 30 μM [2SCC-Ni(I9CL12C)₂]²⁻ (red), apo peptide (wine), 30 μM NiSO₄ (orange), and Ru^II alone (navy). C) Plot of initial TOF (h⁻¹) vs pH. D) H₂ production by the ArH as a function of catalyst concentration. Inset shows a plot of initial rate vs concentration. All experiments are done using 100 mM ascorbic acid, 1 mM Ru^II in 10 mM Tris. Error bars in black are from 3 independent measurements.

pH-dependent steady-state and time-resolved emission spectroscopy are employed to obtain more insight into the photochemical process. In the presence of H₂A, the steady-state emission intensity of *Ru^II at 610 nm decreases with an increase in pH (Fig. S7). Accordingly, in time-resolved emission profiles the lifetime of the excited state species, *Ru^II, decreases with an increase in basicity, showing faster decay profiles at higher pH (Fig. S8). The bimolecular rate constant of electron transfer (k_q) to *Ru^II is determined to be ~5×10⁷ M⁻¹s⁻¹ from this pH-dependent quenching behavior, which reaches saturation at ~pH 5. This can be explained by the fact that the first pK_a of H₂A is 4.1, meaning that at pH 5, H₂A exists as the conjugate base HA⁻ (the 2nd pK_a is ~11), while in acidic conditions (pH <4.1), H₂A predominates. As HA⁻ is a more efficient electron donor to *Ru^II than H₂A,^[21] a steady pool of reduced Ru^I is maintained at pH >5. Therefore, one contributing factor to the observed bell-shaped pH dependence on the HER activity is the fact that *Ru^II can be efficiently reduced to Ru^I, which in-turn reduces Ni^II-peptide to the active Ni^I form. The fluorescence of *Ru^II is quenched by HA⁻ following a linear Stern-Volmer behavior with a quenching rate constant of k_q = 1.1×10⁷ M⁻¹s⁻¹ (Fig. S9). Under similar conditions the metallopeptide does not quench *Ru^II fluorescence. Together, these results lead to the conclusion that the HER in this system proceeds through a reductive quenching pathway where excited *Ru^II is quenched through ET from HA⁻ to produce Ru^I, which in-turn produces the active Ni^I form of the metallopeptide.

To determine the ET processes, the underlying timescales, as well as the rate limiting step of the photocatalytic process, transient absorption spectroscopy (TAS) is employed. In these studies, the formation of Ru^I is monitored by following the growth and decay kinetics of the characteristic absorption feature exhibited by this species at ~510 nm. At pH 5.6, where the maximum activity is observed, the Ru^I species is found to have already started forming at low ns (~10 ns) timescales. This suggests that the formation of the first excited state, *Ru^II, happens faster than the response function of the instrument (<10ns). With time, the Ru^I species matures with a maximum absorbance at ~400 ns (Fig. 5a). The time constant (τ) for the formation of Ru^I is calculated to be ~105 ns by global fitting of the kinetics data at two wavelengths, 510 nm and 490 nm (Fig. 5b). The next step must involve ET from Ru^I to Ni^II-peptide with concomitant formation of the reduced Ni^I-peptide catalyst along with Ru^II. This process is associated with a decay of the 510 nm peak of Ru^I, found to occur within ~17 μs. The absorbance decays to zero at ~20 μs, which indicates relaxation of the system with the regeneration of Ru^II and the Ni^II-peptide. A Ni bleached signal that was previously observed in a PSI-Ni hybrid system is not present in the photo-excited reaction under study here.^[5b] The kinetic scheme from TAS studies is summarized in Fig. 5C. The pH-dependent TAS data shows that below pH 5, the decay of Ru^I to Ru^II is pH-independent (Fig. 5D) with τ of ~90 μs. At pH 5.6 a much faster rate (~17 μs) of formation of Ru^II from Ru^I is observed, which is consistent with the fact that the activity is the highest at this pH. In the absence of the metallopeptide, a much slower rate (~70 μs) of formation of Ru^II at this pH (Fig. 5D red) is observed. This result suggests that when the metallopeptide is present, it can readily accept the electron from Ru^I to produce the Ni^I-form and Ru^II. At higher pH, the reduction to Ni^I and Ru^II production is still fast (Fig. 5D), but the HER activity is low. To address this aspect, the protonation state of the Cys residues is probed by pH-titrations.

Figure 5.

A) TAS data of the ArH as a function of time. B) Kinetic traces and fits of TAS data at 510 nm and 490 nm for the formation and decay of Ru^I. C) Schematic showing the timescale of photochemical events. D) pH-dependent decay times of reduced Ru^I back to Ru^II. The red data point is in the absence of the metallopeptide at pH 5.6.

pH titrations and the Cys pK_a.

The pH-dependence of Ni^II binding via the characteristic LMCT bands of [2SCC-Ni(I9CL12C)₂]²⁻ is probed by a gradual increase in the pH of solutions containing 30 μM peptide and 1 equivalent of Ni^II. The UV-vis features attributable are absent in the pH range of ~3–5.5 (Fig. S10). Beyond this, the LMCT bands appear and mature to full intensity at pH ~8. This result suggests that the protonation state of Cys determines the LMCT transitions where the fully occupied 3p orbitals of thiolates promotes CT to Ni^II. The absorbance changes with pH following S-shaped transitions with a pK_a of ~6.4 (Fig. 6A). To determine the protonation states of the absorbing species and to associate the pH transition to a chemical event, several models are tested to analyze the pH titration curve. These include modeling the transition to a single H⁺ event as well as 2H⁺ events (see the SI for details) with either sequential deprotonation of 1H⁺ (Fig. S11A) at a time or simultaneous deprotonation of 2H⁺ (Fig. S11B). The model that correlates closely with the experimental data and gives the best statistics for non-linear least square analyses (Table S1) is the one assuming a single H⁺ event (Fig. 6B). This is consistent with the general shape of the transition profile resembling a single deprotonation step. This model assumes that at pH below 5.5, three of the Cys residues are deprotonated, while the fourth Cys is bound as thiol forming the (NiS₃SH)⁻ complex. With an increase in pH the 4th Cys is deprotonated leading to the formation of tetrathiolato (NiS₄)²⁻ species. From this analysis, a pK_a of 6.4 is computed, with Δε similar to that which is obtained from the UV-vis spectra of [2SCC-Ni(I9CL12C)₂]²⁻ at pH 8.5 (Table 2). Due to the local C2 symmetric nature of the Cys sites, it is challenging to assign the pK_a to a specific Cys. It is expected that the local packing effect, differential solvent access, and the relative orientation of the Cys in the Ni^II bound form will alter the intrinsic protonation states.

Figure 6.

pH titration curve of 30 μM peptide in the presence of 1 equivalent Ni^II. A) Plot of differential absorbance vs pH obtained from changes in the UV-vis spectra. The solid lines represent global fits to the experimental data using equation I (see the SI) derived from the equilibria model shown in B.

Mechanistic insights.

With these combined data, a mechanistic model for HER by the ArH can be formulated (Fig. S12A). The first step is proposed to be the reduction of Ni^II to Ni^I, followed by H⁺ binding and formation of the Ni^III-H⁻ intermediate, analogous to the Ni-C intermediate proposed for [NiFe] H₂ases (Fig. S12B). As this intermediate is unstable, an ET step should produce the reduced Ni^II-H⁻ (Ni-R) species primed for protonation of the hydride from the thiol H⁺ to release H₂. At pH 5.6, a Cys remains protonated, given the pK_a is ~6.4. The H⁺ on Cys facilitates H₂ production by combining with the H⁻ of the Ni-R intermediate. In [NiFe] H₂ases, the active site Cys has been proposed to serve an analogous role.^{[3a, 3c]} While the activity tapers off sharply at higher pH, the TAS data suggest that the production of Ni^I from Ru^I is still fast at pH >5.6. Therefore, it is unlikely that a lack of Ni^I limits the activity at high pH. Rather, the absence of H⁺ on the Cys disfavors the protonation of Ni^II-H⁻ for H₂ production.

Conclusions

In conclusion, we have described a de novo ArH with Cys residues placed in the ubiquitous -Cys-Xxx-Xxx-Cys- chelation to form the Ni(Cys)₄ coordination inspired by the Ni site of the [NiFe] H₂ases. The principles of specific KIH packing of hydrophobic residues as originally proposed by Pauling and Crick,^[22] along with ion pair/H-bonding interactions are suitably positioned in the design. In the apo form, the 2SCC is weakly helical. Ni^II binding is necessary to impart a significant α-helical character and to induce the unfolding transition. Rich UV-vis spectral features and metal titrations show that one equivalent of Ni^II is bound producing features characteristic of metal-thiolate interactions. Control experiments indicate that the bis-Cys chelation is necessary for Ni^II binding in the NiS₄ environment. The ArH produces H₂ under photocatalytic conditions with a maximum activity at pH 5.6 and diminished activity below and above this pH. At lower pH, the production of reduced Ru^I is limited by the reducing ability of ascorbic acid, as the conjugate base form is an efficient reducer of *Ru^II as opposed to the acidic form. At pH >5.6, the ability of Ru^I to form reduced Ni^I-peptide is unabated. However, the activity is limited by the protonation of Ni^II-H⁻ by H⁺ on the Cys. As the pK_a of the Cys for Ni^II-binding is 6.4, it exists as thiolate at higher pH, while at pH 5.6 it exists in the protonated form, which facilitates H₂ production. We demonstrate the potential of de novo metalloprotein design for an energy relevant transformation via artificial photosynthesis. Future studies will be necessary to fully elucidate the structural and spectroscopic properties of the intermediates as well as to elucidate how the local environment influences the pK_a of specific Cys residues deemed important for activity.

Experimental Section

General procedures:

All the chemicals and reagents are of analytical grade and used as received without further purification. All the glassware and plasticware used for the experiments are soaked in a 10 mM ethylenediaminetetraacetic acid (EDTA) bath overnight, followed by overnight soaking in 10 and 1% nitric acid baths, respectively. These are then thoroughly washed with deionized water. All buffers are Chelexed (Sigma-Aldrich) overnight followed by pH adjustment and filtration.

Peptide design:

Desired peptide sequences for dimeric coiled coil peptides are energy minimized in CC-Builder,^[15] with a pitch of 421.87, radius 4.79, and an interface angle of 13.51°. The sequence with the lowest BUDE energies is selected for experimental testing.

Peptide synthesis and purification:

Peptides are synthesized in an automated microwave peptide synthesizer (Liberty Blue, CEM) at 0.1 mmol scale using Rink Amide Pro-Tide Resin (CEM). The amino acids (Advanced Chemtech and CEM) are weighed in concentrations of 0.2M and dissolved in DMF (VWR). Oxyma (1M; CEM) and N,N’-diisopropyl carbodiimide (DIC) (0.5M; Advanced Chemtech) in DMF are used as the activator base and activator, respectively. 20% piperidine (Advanced Chemtech) in NMP is used as the deprotection solution. N-terminal acetylation is done by 10% acetic anhydride in DMF. All coupling steps are performed at 90℃ for 4 min. After synthesis, the resin is washed with dichloromethane and dried under N₂. Cleavage and side-chain deprotection is done using 92.5% trifluoroacetic acid (TFA; Advanced Chemtech), 2.5% triisopropylsilane (TIS; Sigma–Aldrich), 2.5% ethane-1,2-dithol (EDT; Sigma–Aldrich), 2.5% water in a 10 mL total volume for 2h at RT with constant stirring. After cleavage, the crude peptides are filtered under vacuum and excess TFA is evaporated under a gentle stream of N₂. The crude peptides are then precipitated, washed with cold diethyl ether and lyophilized. The peptides are again dissolved in 10% acetic acid and purified by using a C18-semiprep column using 1260 Infinity II HPLC (Agilent) system. A linear gradient from 0.1% TFA in H₂O to 0.1% TFA in acetonitrile is used. The chromatogram is monitored at 214 and 280 nm. The purified peptide is lyophilized and stored at −20℃ until further use. Concentrations of purified peptides are calculated using an extinction coefficient ε_280nm of 5500 M⁻¹ cm⁻¹ for one Trp.

Mass spectrometry:

The identities and purities of the purified peptides are confirmed by MALDI-MS (Bruker Voyager) with sinapinic acid as the matrix (Sigma–Aldrich:10 mg/mL in 50:50 acetonitrile: water with 0.1% TFA). One microliter of 50:50 peptide: sinapinic acid matrix mixture is spotted onto a 100-SS MALDI plate and analyzed. Alternatively, peptides are also characterized by ESI-MS using a Waters Synapt G2 mass spectrometer with peptides dissolved in 50 mM NH₄OAC. Fig. S13 shows the ESI data.

DTNB assay:

Samples for thiol quantification contains peptide and 5,5’-dithio-bis-(2-nitrobenzoic acid) (DTNB, Acros Organics) in 100 mM TRIS pH 8.5. The samples are incubated for 15 min prior to quantification using ε_{412 nm} = 14,150 M⁻¹cm⁻¹ on an Agilent 8454 UV-vis spectrometer.

Preparation of peptide samples:

The solid peptides are dissolved in suitable degassed buffer solutions. Tris(2-carboxyethyl) phosphine hydrochloride (TCEP) is added at 10-fold excess to peptide solutions and stirred for 10 min. TCEP is removed by PD10 or spin filtration. Samples are prepared anaerobically using 1 eq. of NiSO₄ employing Schlenk techniques. The binding of nickel is confirmed by UV–vis spectroscopy.

Size exclusion chromatography:

Size exclusion chromatography is performed using a bio SEC-5 column (Agilent) at 0.34 ml/min flow rate with 150 mM phosphate buffer pH 7 as the mobile phase. The concentration range of 5–85 μM dimer is used for both apo and Ni bound peptides. At all concentrations the peptides elute with same retention time of 11 min. Molecular weights are estimated from the calibration plot utilizing protein standards from Agilent. Molecular weight of both apo and Ni^II-peptides are calculated to be 6039 Da from the calibration plot (theoretical = 6458 Da for apo and 6517 Da for Ni-bound).

¹H NMR spectroscopy:

NMR data are collected at RT in a Bruker 400 MHz spectrometer equipped with a 5 mm broadband cryo probe employing a Watergate pulse program for water suppression. Samples contained 500 μM apo or Ni^II-peptides prepared under inert conditions in 100 mM phosphate buffer pH 8.2 and 10% D₂O. For each sample, 1500 scans with a sweep width of 100 are collected. To suppress the water peak at 4.8 ppm, a pulse power of 30 DB, 1500 number of scans and sweep width of 100 are used. Data processing is performed in MestReNova.

CD spectroscopy:

CD spectra are collected using an AVIV 202 SF spectrometer under constant flush of nitrogen gas supplied from a liquid nitrogen tank. The apo and Ni^II-bound peptides are prepared under inert conditions in 5 mM phosphate buffer at pH 8.5. Spectra are collected at different concentrations using a quartz cuvette with 1 mm path length. Raw data are converted to mean residue ellipticity (MRE) using the equation, Ө_mre (deg. cm². dmol⁻¹ resi⁻¹) = millidegrees / [(AAs − 1) *concentration (M) * cell path length (cm) * 10], where AA is the number of amino acids. Thermal melting experiments are performed by monitoring the CD signal at 222 nm in the temperature range of 18–95℃. 15 μM of dimers with 10-fold TCEP are prepared anaerobically in a 1 cm path length quartz cuvette. A constant flow of N₂ is maintained in the sample chamber during the thermal melts.

UV-vis Ni^II titration:

Ni^II titration is performed by anaerobic addition of NiSO₄ to TCEP-reduced 30 μM [2SCC-(I9CL12C)₂]⁴⁻ in 100 mM Tris pH 8.5 using a 1 cm path length septa-capped cuvette (Starna Cells) and spectral changes are monitored using a Cary 5000 UV–vis NIR spectrophotometer (Agilent). Ni^II is added using a gastight syringe (Hamilton) followed by 10 min incubation to reach equilibrium after each Ni^II addition prior to measurements. The absorbance of apo peptide is subtracted from Ni^II-bound spectra to obtain the differential absorbance. Appropriate dilution correction is done as necessary.

pH titrations:

UV-vis pH titrations are carried out by adding small aliquots of concentrated KOH to unbuffered solutions containing 30 uM peptide dimers in the presence of 1 eq. Ni^II, and the change in absorbance is monitored as a function of pH. Equilibration time of 10 min is allowed before reading the final pH and recording UV-vis. The titration curves of the peptides are fit using various models shown in the SI. The best fit is obtained for a single-H⁺ equilibrium model.

Photo-induced catalysis:

Photocatalysis experiments are performed with 2 mL solutions of anaerobically prepared 30 μM catalyst, 1 mM of the photosensitizer Tris(2,2′-bipyridine)dichlororuthenium(II)hexahydrate, (Ru(bpy)₃Cl₂·6H₂O, Alfa Aesar), and 100 mM of ascorbic acid (VWR) in 10 mM Tris buffer inside septa-capped Pyrex tubes. The white light source (λ > 400 nm; Thor Laboratories) is adjusted to 180 mW power. Illuminations are performed generally for 4–6 hours with constant stirring. The pH-dependent photocatalysis experiments are performed by making a series of individual samples with varying pH, which is adjusted by addition of small aliquots of NaOH after addition of all reagents. The low buffer concentration is used to provide sufficient buffering capacity to the solutions while minimizing the effect of ionic strength on photocatalytic H₂ production, as the latter factor has been a contributing factor in such assays with molecular catalysts.^[23] After photocatalysis experiments, the pH is checked again. In general, a small variability in pH of ±0.4 is observed before and after photocatalysis experiments. Peptide concentration dependent experiments are performed keeping all other reagents constant. 250 μL of head space gas is syringed every ~30 min using gas-tight syringes (VICI) and injected into a 7890B GC instrument (Agilent) operating with He as the carrier gas. H₂ is detected using a thermal conductivity detector (TCD) and the area under the peak is converted to ppm using calibration standards. TONs are calculated from the ratio of the moles of H₂ produced per mole of the catalyst.

Fluorescence assays:

A solution of 1 mM Ru^II in 10 mM Tris is prepared in a quartz cuvette fitted to a total volume of 2.5 mL and pH is adjusted to 5.6. Aliquots of ascorbic acid are added followed by 10 min stirring. A concentration range from 35 – 120 μM of ascorbic acid are recorded for the Stern-Volmer analysis. Similar experiments are done with the metallopeptide (5–100 μM). The intensity of the fluorescence is monitored using PTI fluorimeter (excitation λ= 510 nm, emission λ = 610 nm) and the data are processed using Origin and fit to the Stern-Volmer equation:

$${\text{F}}_0/\text{F}=1+{\text{k}}_{\text{q}}{\tau }_0\left[\text{Q}\right]$$

F₀ and F are the initial and final intensity, respectively; [Q] is quencher concentration, τ₀ is the lifetime of fluorophore in the absence of quencher (0.63 μs in water), k_q is the quenching rate constant.

Time-resolved fluorescence:

Data is collected using an Edinburgh LP980 optical system. For fluorescence lifetime measurements, 425 nm excitation light is generated from an Nd:YAG pumped dye laser (Continuum, ND6000). Time-resolved fluorescence measurements are monitored at 600 nm and averaged over 50 laser shots. Emission kinetics are modelled by a single exponential function. Samples consisting of 100 mM of ascorbic acid and 1 mM Ru^II dissolved in 10 mM Tris are prepared anaerobically in 1 cm path length. pH of the samples is adjusted by adding NaOH solution.

Transient absorption spectroscopy:

Nanosecond transient absorption spectroscopy (ns TAS) data are collected using an Edinburgh LP980 optical system (Edinburgh Instruments, U.K.). Measurements are performed with 532 nm excitation light (<12 mJ/pulse) provided by pulsed Nd:YAG laser (Continuum, Surelite I) equipped with a frequency doubling crystal. The probe source is a 150 W pulsed xenon arc lamp and absorption is detected with a photomultiplier tube (R928, Edinburgh Instruments, U.K.). Detector outputs are processed using a Tektronix MDO3022 Mixed Domain Oscilloscope (200 MHz, 2.5 GS/s) interfaced to a PC and Edinburgh’s L900 (version 8.2.3) software package. Time-resolved absorption data is monitored at 510 nm and averaged over 200 laser shots. Kinetic data is fit with the exponential function $\Delta \text{OD}\left(\text{t}\right)=\Delta {\text{OD}}_0\cdot e^{-k_1\tau }$, where ΔOD₀ is the initial ΔOD, k₁ is the first order rate constant and τ is the lifetime. Sample solution consisting of 300 μL anaerobically prepared 500 μM metallopeptide in 10 mM Tris buffer pH 5.6 is prepared with 0.02 μM of the Ru^II and 100 mM of ascorbic acid. The PS concentration was optimized to minimize fluorescence. Solutions are stirred and purged with N₂ for 5 min and kept in the dark to avoid degradation prior to the measurements. Data is recorded in a 1 mm path length cuvette positioned at a 45° angle relative to the pump and probe sources in order to maximize overlap and direct scattered light away from the entrance slit to the monochromator.