Use of a Metallopeptide Based Mimic Provides Evidence for a Proton Coupled Electron Transfer Mechanism For Superoxide Reduction By Nickel Containing Superoxide Dismutase

Superoxide (O₂^•–) is a highly toxic reactive oxygen species that can induce cellular death at subnanomolar concentrations.^[1] Because O₂^•– is an unavoidable fact of aerobic life, aerobic organisms have evolved superoxide detoxification systems, the most prevalent of which are the superoxide dismutases (SODs).^[1,2] All known SODs have redox-active metal cofactors and detoxify superoxide through a ping-pong type mechanism:

$${\mathrm{M}}^{\mathrm{red}}+{{\mathrm{O}}_2}^{\cdot -}+2{\mathrm{H}}^{+}\to {\mathrm{M}}^{\mathrm{ox}}+{\mathrm{H}}_2{\mathrm{O}}_2$$ (1a)

$${\mathrm{M}}^{\mathrm{ox}}+{{\mathrm{O}}_2}^{\cdot -}\to {\mathrm{M}}^{\mathrm{red}}+{\mathrm{O}}_2$$ (1b)

where M^red and M^ox are the reduced and oxidized forms of the metalloenzyme's metal cofactor. The most recently discovered SOD contains nickel at its active-site (NiSOD), is found in several soil and aquatic bacteria, and operates by cycling between the reduced Ni(II) and oxidized Ni(III) oxidation states.^[2f,3] In its reduced state, square planar Ni(II) is ligated by two cis-cysteinate sulfur atoms (Cys2 and Cys6), an amidate nitrogen atom from Cys2 and the N-terminal amine nitrogen atom from His1 (Scheme 1).^[4] Oxidation to Ni(III) yields a square pyramidal structure via the ligation of the His1 imidazole δ-nitrogen to Ni. Thus, all of the ligating residues to Ni are contained within the first six residues of the NiSOD primary sequence.

Scheme 1.

NiSOD (R = H) and Ni^M1 (R = CH₃) active sites.

We,^[3d,5] and others,^[6] have exploited this convenience of nature to generate active NiSOD biomimetics by utilizing peptides containing the first 6-12 residues of the NiSOD sequence. Previously we demonstrated that one of these metallopeptide based NiSOD mimics, {Ni(SOD^M1H(1)H^Me)} (Ni^M1; SOD^M1H(1)H^Me = H₂N-H^MeCDLP CGVYD PA; H^Me = ^εN-methyl-histidine), is isolable and stable at room temperature in both the Ni(II) (Ni(II)^M1) and Ni(III) (Ni(III)^M1) oxidation states. ^[5a] Ni^M1 also performs catalytic superoxide disproportionation at reasonably slow velocities (k = 6(1) × 10⁶ M^–1 s^–1; pH = 8.0) making it an ideal candidate for detailed mechanistic study.

The mechanism by which NiSOD disproportionates superoxide has not been explore in any detail experimentally. Besides issues involving inner-sphere vs. outer-sphere O₂^•– oxidation/reduction, one mechanistic point that has not been explored is the role of proton donation in the superoxide reduction half-reaction (1a). Herein, we present evidence using Ni^M1 that superoxide disproportionation catalyzed by NiSOD can occur through a proton-coupled electron transfer (PCET) type mechanism.

We first turned our attention to the superoxide disproportionation kinetics catalyzed by Ni^M1 using stopped-flow techniques as previously described.^[5a,7] The reaction was performed at pH 8.0 to suppress the O₂^•– self-disproportionation reaction. At 25.0 °C a k_cat = 6.1(2) × 10⁶ M^–1 s^–1 was obtained, in line with what was previously observed (Figure 1). Performing the reaction in D₂O (pD = 8.0)^[8] resulted in a substantial solvent KIE; the D₂O k_cat = 3.1(2) × 10⁵ M^–1 s^–1, which translates into a solvent KIE of 20(4). Such a large room temperature KIE is suggestive of a PCET event with a substantial quantum mechanical tunneling component. We note that we measure a superoxide self-disproportionation solvent KIE of 1.2(1) under these conditions.

Figure 1.

Superoxide decay kinetics in the presence and absence of Ni^M1 at pH/pD 8.0 (25.0 °C). Data is represented by the solid lines and the second order fits are represented by the dashed lines. The gradual increase in absorbance at long time lengths is related to the H₂O₂ initiated decomposition of Ni^M1 taking place after the O₂^•– disproportionation is complete.

To determine if a quantum mechanical tunneling event is a tenable hypothesis we examined the temperature dependence of k_cat on the Ni^M1 catalyzed disproportionation of superoxide. We were able to accurately determine rate constants over the temperature range of 2.0 – 75.0 °C. The Arrhenius plot of superoxide disproportionation is relatively flat, but none-the-less displays a slight concave appearance (Figure 2). A traditional Arrhenius analysis suggests a near barrierless reaction (E_a = 0.60(1) cal mole^–1; A = 6.45(3) × 10⁶ s^–1). These values are wholly inconsistent with estimated and measured SOD energetics.^[10] What the variable temperature kinetics data are consistent with is that SOD catalysis facilitated by Ni^M1 is operating via a tunneling mechanism.^[11]

Figure 2.

Arrhenius plot of {Ni(SOD^M1H(1)H^Me)} catalyzed decay of superoxide at pH/pD 8.0. The main plot depicts the data collected in H₂O, the inset depicts the data collected in D₂O. The error bars are the size of the dots or smaller. The trend lines were generated using a modified Bell Tunneling expression^[11a] (see supporting information).

Based on the above experiments it seems reasonable to propose that Ni^M1 can facilitate the reduction of O₂^•– through a PCET reaction. We should therefore be able to facilitate the reduction of Ni(III)^M1 with H-atom donors possessing moderate X-H bond dissociation free energies (BDFEs). Ni(III)^M1 was thus subjected to a series of reactions at pH 8. We could effect the reduction of Ni(III)^M1 to Ni(II)^M1 using ascorbate (BDFE = 73.6 kcal mole^–1)^[12] and TEMPO-H (BDFE = 71.0 kcal mole^–1).^[12] Furthermore, Ni(III)^M1 could be regenerated by adding TEMPO• to Ni(II)^M1. In contrast, we could not facilitate the reduction of Ni(III)^M1 utilizing reagents with slightly stronger X-H bonds, such as 1,4-dihydroquinone (first BDFE = 81.5 kcal mole^–1)^[12] and hydrazine (first BDFE = 83.4 kcal mole^–1).^[12] This would suggest that 1) there is a hydrogen atom acceptor site within Ni(III)^M1 that allows for the generation of Ni(II)^M1–H (i.e. protonated Ni(II)^M1), and 2) that the Ni(II)^M1–H BDFE is between 70 to 80 kcal mole^–1.

To better estimate the Ni(II)^M1–H BDFE an equilibrium reaction between TEMPO•/TEMPO-H and Ni(II)^M1–H/Ni(III)^M1 was established (Scheme 2). By varying the concentration of TEMPO• and Ni(II)^M1–H and allowing the reaction to come to equilibrium we could calculate the equilibrium constant, K, for the reaction. At 25.0 °C we obtain a K = 26.5(8) for the Ni(III)^M1 + TEMPO-H ⇌ Ni(II)^M1–H + TEMPO• equilibrium process, which translates into a ΔG = –1.9(7) kcal mole^–1. We can thus estimate a BDFE for the Ni(II)^M1–H bond of 73(1) kcal mole^–1. The measured BDFE is in line with the calculated BDFE based on the Ni(II)^M1–H pK_a and the previously determined redox potential of Ni^M1 (440 mV vs. NHE).^[5a] Measuring differences in the ligand-field bands of Ni(II)^M1–H/ Ni(II)^M1 as a function of pH yields a Ni(II)^M1–H pK_a of ~8.2. Application of a modified Bordwell equation^[12] yields a Ni(II)^M1–H BDFE of ~79 kcal mole^–1, which is in good agreement with the above measurement considering the associated errors in the Ni^M1 redox potential measurement.^[5a] Both of these measurements are in line with the superoxide reduction process; the deprotonated product, HO₂^–, has an O-H BDFE of 81.6 kcal mole^–1.^[12] For Ni(II)^M1–H to effectively facilitate the reduction of O₂^•– through a PCET reaction it must have a BDFE less than the HO₂^– O-H BDFE. Otherwise the reaction would be overall uphill and not thermodynamically viable.

Scheme 2.

Equilibrium reaction between Ni(III)^M1 + TEMPO-H and Ni(II)^M1-H + TEMPO•

If there is an abstractable H-atom within Ni(II)^M1–H, then there must be a well defined protonation site near the Ni(II) active site. An initial clue to the location of the well-defined protonation site within Ni(II)^M1–H was provided by Ni K-edge X-ray absorption spectroscopy (XAS) recorded at elevated pH. Previously we demonstrated that at pH 7.4 Ni(II)^M1–H has an average Ni-S bond length of 2.18 Å.^[5a] At pH 9.5 Ni(II)^M1 has an average Ni-S bond length of 2.20 Å. This increase in bond length is consistent with a thiolate deprotonation event.^[13a] Although counterintuitive, the origin of the contraction of the Ni-S bond in square planar Ni(II)N₂S₂ complexes upon protonation or hydrogen-bond formation is well know.^[13] Prontonation of the coordinated S via a filled S(p)-type lone-pair reduces unfavorable filled-filled Ni(π)/S(π) interactions leading to a bond contraction. Thus, the elongation of the Ni-S bond at high pH is consistent with cysteinate deprotonation.

A protonated Ni-S(H)-Cys bond is also consistent with room temperature S K-edge XAS performed at pH 7.4 vs. 9.5. At pH 9.5 the S K-edge X-ray absorption spectrum of Ni(II)^M1 is consistent with a thiolate ligated Ni(II) complex (Figure 3).^[14] The main S(1s) → S(C-Sσ*) transition occurs at 2472.6(1) eV. Prior to the S(1s) → S(C-Sσ*) transition there is a prominent pre-edge feature corresponding to the nominal S(1s) → Ni(3d)/S(σ)* transition at 2470.4(1) eV. Similarly, the pH 7.4 S K-edge spectrum of oxidized Ni(III)^M1 is consistent with a thiolate ligated Ni(III) complex; the main S(1s) → S(C-Sσ*) transition occurs at 2473.0(1) eV while the S(1s) → Ni(3d)/S transitions at ~2470 eV is broad, displays asymmetry and is more intense. The change in intensity and peak shape is consistent with a) the increase in number of holes in the Ni(3d) manifold from 2 to 3 upon oxidation and b) the fact that a lower-energy Ni(3d)/S(π)* type orbital is now only partially filled and will accept an electron upon promotion from the S1(s) orbital. These two spectra can be contrasted with the pH 7.4 S K-edge X-ray absorption spectrum of Ni(II)^M1–H. First, there is a broadening of the main S(1s) → S(C-Sσ*) transition. Second, and more obvious, the pre-edge feature virtually disappears into the baseline. There are three conclusions that can be drawn from these results. One is that the pH 7.4 form of Ni(II)^M1–H is protonated at one (or both) of the coordinated cysteinates.^[15] The second conclusion is that at pH 9.5 the coordinated cysteinate(s) become deprotonated. The last conclusion is that Ni(III)^M1 does not contain a protonated coordinated cysteinate at low pH.

Figure 3.

Sulfur K-edge XAS spectra of Ni(II)^M1–H recorded at pH 7.4 (solid), Ni(II)^M1 recorded at pH 9.5 (dotted), and Ni(III)^M1 recorded at pH 7.4 (dashed).

It seems reasonable to propose that the source of the formal H-atom in reduced Ni(II)^M1–H is a protonated Ni(II)-S(H)Cys moiety. One may expect that the pK_a of a coordinated thiolate would be rather low and inconsistent with this formulation. However, we suspect that the cis-Cys moiety, or possibly the coordinated amine cis to the Cys residue, may substantially raise the pK_a of the Ni(II)-S(H)Cys moiety higher than expected through a cooperative effect as seen in 1,8-bis(dimethylamino)naphthalene (proton sponge), for example. Other structural factors in metallopeptide microenvironment about Ni(II) center may also be contributing to the elevated Ni(II)-S(H)Cys pK_a.

In summary, we have provided evidence that superoxide disproportionation catalyzed by Ni^M1 occurs through a rate-limiting PCET mechanism with a significant tunneling component. It appears that the PCET step occurs during the reduction of O₂^•– by protonated Ni(II)^M1–H, and the formal H-atom is in the form of a coordinated Ni(II)-S(H)^Cys moiety (Scheme 3). This proposal of a PCET mechanism during superoxide reduction appears to be unique amongst the SODs, save a recent study of Cu/Zn-SOD (vide infra).

Scheme 3.

Proposed mechanism of O₂^•– reduction by {Ni^II(SOD^M1H(1)H^Me)-H} and NiSOD.

Computationally, there has been one hybrid DFT study that has explicitly invoked a PCET event during the NiSOD catalyzed reduction of O₂^•–.^[16] In that study the transferred “H-atom” is in the form of a Ni-S(H)-Cys moiety. Our findings are also consistent with the limited experimental data available concerning NiSOD itself, as well as some more recent studies on Cu/Zn SOD. First, there is strong evidence from S K-edge X-ray absorption spectroscopy that reduced NiSOD itself contains a protonated Ni(II)-S(H)Cys moiety while the cysteinate is deprotonated upon NiSOD oxidation.^[15] Second, the crystal structure of NiSOD reveals an unusual structural feature; the Ni-S bond trans to the amidate nitrogen is shorter than the Ni-S bond trans to the amine nitrogen.^[4] This is the opposite of what one would expect based on the synthetic Ni(II)N₂S₂ models with amine vs. amidate coordination.^[13a,b,17] It therefore seems reasonable to suggest that the Ni-S^Cys moiety trans to the amidate is indeed protonated in NiSOD. Analysis of the pH dependence of SOD catalysis by NiSOD itself shows that k_cat is pH independent over the range of 6-8 and then falls off dramatically above 8.0, which is suggestive of the strong coupling of the proton-transfer event with the redox chemistry of O₂^•–.^[3c] Lastly, we note that at least one other SOD appears to operate through a rate-limiting PT event. A recent natural abundance ¹⁸O/¹⁶O KIE study by Roth and Smirnov provided evidence that an initial rate limiting proton transfer event is occurring in the reduction of O₂^•– by Cu/Zn SOD under high pH conditions.^[10a] We are currently in the process of determining the extent to which H-atom transfer from a M^red-S(H)^Cys moiety is relevent in NiSOD itself as well as other metalloenzymes, such as [Ni,Fe]-hydrogenase.^[18]