Mechanistic and structural characterization of an iridium-containing cytochrome reveals kinetically relevant cofactor dynamics

Abstract

Artificial metalloenzymes (ArMs), which contain non-native, typically synthetic, metal cofactors, are a flourishing class of biocatalyst for unnatural reactions. Although the number of these reactions is rapidly increasing, multi-faceted mechanistic studies of ArMs comprising structural, kinetic, computational and cofactor binding data to reveal detailed mechanistic information on the effects of the protein scaffold on the structure and reactivity of ArMs are more limited. Here we report the structure of an unnatural P450 analogue using X-ray diffraction. We also report the kinetic analysis of its reaction, catalyst activation during an induction period, and the origins of the stereoselectivity for the cyclopropanation of a terpene catalysed by the iridium-containing P450 variant (Ir(Me)-CYP119). Our data reveal a mechanism initiated by the flip of the cofactor from an inactive to an active conformation. This change in conformation is followed by thousands of turnovers occurring by rate-determining formation of an iridium-carbene intermediate, thereby highlighting the influence of cofactor dynamics within a single active site on an ArM-catalysed reaction.

Fig. 1 |. The structural details of Ir(Me)–CYP119.

a, Ir(Me)–MPIX has a single open coordination site and axial methyl ligand. b, Three major factors that influence the specific binding of haem and haem analogues. L is His/Ser, X is Cys/Ser/Tyr and M is Fe, Co or Mn. c, Comparison of Ir(Me)–CYP119 at 3 Å resolution (PDB 7UOR) and WT CYP119 (PDB 1IO7). The Ir–CHO distance is 2.2 Å in the refined structure. The calculated Ir–CH₃ distance is 2.1 Å. The Cys S–Fe distance in WT CYP119 (PDB 1IO7) is 2.37 Å. Note that the original Ir(Me) moiety has been oxidized, likely by reactive species generated by synchrotron X-ray radiation during data collection. Residues 213, 254 and 317 are highlighted in dark blue. The flexible FG loop is highlighted in yellow. Residue 155 is highlighted in red. The Ir(Me)–MPIX cofactor is highlighted in cyan. d, The FG loop (yellow) is pinned back by intercalation of W155 (red) in Ir(Me)–CYP119, resulting in the opening of the substrate channel. WT CYP119 is shown for reference. e, The axial ligand at Ir points towards the ArM active site (ax-ligand-up orientation), shielding the open coordination site at iridium. Coordination from the protein backbone is not observed.

Fig. 2 |. Potential kinetic mechanisms of cyclopropanation catalysed by Ir(Me)–CYP119.

a,b, Sequential (a) and ping-pong (b) kinetic mechanisms for the cyclopropanation reaction. The major difference between a and b is the binding of both substrates before the reaction proceeds in a versus the binding and reaction of the carbene precursor to form an intermediate that reacts with the olefin in b. Mechanism b is technically a hybrid ping-pong Theorell–Chance mechanism because accumulation of the nitrogen–enzyme complex is highly unlikely. RDS, rate-determining step.

Fig. 3 |. Kinetic data on the cyclopropanation of carvone with EDA catalysed by Ir(Me)–CYP119.

a, Cyclopropanation of carvone catalysed by Ir(Me)–MPIX (free cofactor) and Ir(Me)–CYP119 (holoenzyme-containing Ir(Me)–MPIX). b, Nitrogen evolution time course of 3.3 μM Ir(Me)–MPIX (free cofactor) or 3.3 μM Ir(Me)–CYP119 (holoenzyme) with 25 mM EDA, 5 mM carvone, 2.5 μM Ir(Me)–CYP119. c, Data for 0–10.0 mM carvone (25 mM EDA, 2.5 μM Ir(Me)–MPIX). a.u., arbitrary units. d, Data for 3.13–50.0 mM EDA (5.00 mM carvone, 2.5 μM Ir(Me)–CYP119). e, Data for 0.313–5.00 μM Ir(Me)–CYP119 (5.00 mM carvone, 25 mM EDA). Shaded areas and error bars represent the standard deviations of three consecutive trials with the same batch of purified enzyme. All error bars and bands are centred at the mean value of three consecutive reactions with the same batch of protein. Experiments were conducted with a Man On The Moon X104 gas evolution kinetics kit.

Fig. 4 |. Putative mechanism for cyclopropanation involving a flip of the cofactor to generate the active catalyst.

A mechanism that would involve a cofactor flip would begin with Ir(Me)–CYP119 almost entirely in the axial-ligand-up conformation. Reversible dissociation to form a non-specific apoenzyme–Ir(Me)–MPIX complex (apoE–Ir) followed by association of the cofactor in the axial-ligand-down conformation would yield the active enzyme species that would be trapped rapidly by EDA.

Fig. 5 |. Kinetics of cofactor and ligand binding to apo-CYP119 and Ir(Me)–CYP119.

a, Left: fluorometric titration plotted as a modified Stern–Volmer plot for the binding titration of apo-CYP119 with Ir(Me)–MPIX. Each point represents the average of two consecutive trials from the same batch of protein. Right: binding time course of 0.5 μM Ir(Me)–MPIX and 9.6 μM apo-CYP119. The shaded area represents the standard deviation of three consecutive trials with the same batch of purified apo-enzyme. The fit $\left(y(t)=A{\text{e}}^{-k_{\text{obs}}t}+a;R^2=0.97\right)\text{ylelded}{k}_{\text{obs}}=0.634{\text{s}}^{-1}$, which gives $k_{\text{on}}=6.6\times {10}^4{\text{M}}^{-1}{\text{s}}^{-1}$. cps, counts per second. b, In path (i) the cofactor flips to the ax-ligand-down conformation before binding of a fluorescent imidazole ligand L. In path (ii) L binds directly to the holoenzyme with non-specific binding of the cofactor. This complex would be formed by dissociation of the cofactor from the active site; for path (ii) $k_{\text{obs}}=k_{\text{off}}$ at saturating [L]. c, BODIPY-Im synthesis. d, Left: representative emission spectra of the ligand free in solution as well as complexed with apo and holoenzyme. We attribute the minor emission change with the apo-BODIPY-Im complex to a local change in pH within the protein. Right: ligand binding time course of 0.25 μM Ir(Me)–CYP119 and 2.5 μM BODIPY-Im. The shaded area represents the standard deviation of two consecutive trials with the same batch of purified holoenzyme. The fit $\left(y(t)=A{\text{e}}^{-k_{\text{obs}}t}+a;R^2=0.96\right)\text{ylelded}{k}_{\text{obs}}=0.34{\text{s}}^{-1}$. All error bars and bands are centred at the mean value of three consecutive reactions with the same batch of protein. a.u., arbitrary units.

Fig. 6 |. Proposed mechanism of cyclopropanation catalysed by Ir(Me)–CYP119.

Potential binding of EDA to the dissociated Ir(Me)–MPIX to initiate the first turnover of the catalytic cycle was excluded for clarity. RDS, rate-determining step.

Fig. 7 |. DFT calculations of cyclopropanation and MD simulations of Ir(Me)–CYP119.

a, DFT-computed Gibbs free energy profile (ΔG) of the intrinsic reaction mechanism for the cyclopropanation of carvone catalysed by the carbene complex of Ir(Me)–MPIX, using a truncated model (Ir–porphine). Energies are given in kcal mol⁻¹ and key distances in the optimized transition state (TS) structures are given in Å. b, Representative structure describing the preferred orientation of the iridium–carbene intermediate when formed in the Ir(Me)–CYP119 (C317G, L155W, T213G, V254L) active site, as determined from extensive MD simulations (see Supplementary Fig. 22 for details). Iridium–carbene and key active site residues are shown as sticks. Mutated residues are highlighted in orange. c, Representative structure describing the preferential catalytically relevant binding mode of carvone in the Ir(Me)–CYP119 (C317G, L155W, T213G, V254L) active site, as characterized from restrained-MD simulations (see Supplementary Fig. 23 for details). Simulations characterized a major pro-cis near attack conformation (NAC) of the carvone with respect to the iridium–carbene intermediate that leads to the (R,S,S) transition state for cyclopropanation. Iridium–carbene, carvone and key active site residues are shown as sticks. Mutated residues are highlighted in orange.