Capturing the Binuclear Copper State of Peptidylglycine Monooxygenase Using a Peptidyl-Homocysteine Lure

Abstract

Peptidylglycine monooxygenase (PHM) is a copper-dependent enzyme that catalyzes C-alpha hydroxylation of glycine extended pro-peptides, a critical post-translational step in peptide hormone processing. The canonical mechanism posits that dioxygen binds at the mononuclear M-center to generate a Cu(II)-superoxo species capable of H-atom abstraction from the peptidyl substrate, followed by long range ET from the CuH center. Recent crystallographic and biochemical data has challenged this mechanism, suggesting instead that an “open-to-closed” transition brings the copper centers closer allowing reactivity within a binuclear intermediate. Here we present the first direct observation of an enzyme-bound binuclear copper species, captured by use of an Ala-Ala-Phe-hCys inhibitor complex. This molecule reacts with the fully reduced enzyme to form a thiolate-bridged binuclear species characterized by EXAFS of the WT and its M314H variant, and with the oxidized enzyme to form a novel mixed valence entity characterized by UV/Vis, and EPR. Mechanistic implications are discussed.

Graphical Abstract

Copper monooxygenases play important roles in human health and climate mitigation.^1–5 Peptidylglycine monooxygenase (PHM) catalyzes the post-translational amidation of neuropeptide hormones,^6–8 using two copper atoms, CuH and CuM, which are ligated to three histidines (H) and two histidines and a methionine (M). Early structures showed the copper atoms separated by ~ 11 Å across a solvent-filled cleft, and a substrate (di-iodo-YG) bound by its carboxy terminus to residues R240 and Y318 (Scheme 1).^9–10 An oxygen adduct bound to CuM was also captured, but with the O₂ in an unproductive conformation pointing away from the substrate.¹¹ These structures, together with spectroscopic^12–16, kinetic^{1, 17–21} and computational^22–26 contributions led to the current canonical mechanism in which the substrate-bound di-Cu(I) enzyme reacts with dioxygen forming a Cu(II)M-superoxo intermediate which subsequently abstracts a H atom at the glycyl C-alpha to form the substrate radical. Long-range electron tunneling (ET) completes the catalytic cycle with the electron being supplied either directly from CuH or via a Cu(II)M-oxyl intermediate.

Scheme 1.

Top Active site structure of PHM (pdb 3PHM) showing positioning of a dipeptide substrate bound via its carboxy terminus to Y318 and R240. Bottom, mixed valence formulation of the PHM-Alo-Alo-Phe-hcys complex.

Recent insights have given cause to reconsider this mechanism.² In particular, new crystal structures for PHM^27–28 and its sister enzyme dopamine β-monooxygenase (DBM)²⁹ show the enzymes in “fully open” (Cu-Cu = 14 Å) and “closed” conformations (Cu-Cu = 4 – 5 Å). The structures suggest flexible dynamics of the two sub-domains allowing the Cu-Cu distance to vary continuously.²⁸ Reformulation of the mechanism to include a binuclear Cu intermediate as proposed from the DBM structure²⁹ and from a recent QM/MM study²⁶ is therefore timely. Here we describe a complex of PHM with a peptide inhibitor complex which establishes the accessibility of a binuclear state.

We reasoned that a substrate analogue bound near CuM (Scheme 1 bottom) might capture the closed state if it carried a Cu(I)-binding functionality capable of bridging the two coppers. We chose the peptidyl inhibitor Ala-Ala-Phe-homocysteine (AAF-hCys) known to form a strong complex with the enzyme³⁰ where binding should occur both via the C-terminal carboxylate and via additional coordination of the thiol to copper (Scheme 1). If binding to both coppers is observed, it would show that a conformation exists with the two Cus in a closed complex, allowing a ligand such as O₂ to bridge and do chemistry. AAF-hCys was prepared by solid phase peptide synthesis. The hCys-containing peptide is largely in the thiol form, and exhibits an m/z of 425.19, which is within 3.29 ppm of expected (Fig. S1).

First, we examined the reaction of AAF-hCys with the fully reduced PHM using EXAFS spectroscopy (Figure 1). The intensity of the Cu-S scattering peak in the FT grew significantly relative to the unligated reduced enzyme (Fig. S2 and Table S2 fit 1), and required simulation by 1.5 Cu-S at 2.27 Å with a Debye Waller (DW) of 0.006 Ų (Fig. 1 (a) and Table S2 fit 2). This result suggests that in addition to the CuM-S(Met314) interaction (0.5 Cu-S per Cu), both coppers now form one additional bond each to the thiolate ligand implying a mono-thiolate bridge. The correlation between coordination number and DW factor introduces uncertainty into the Cu-S shell occupancy, and a reasonable fit could be obtained with 1 Cu-S with a DW of 0.002 Ų (Table S2 fit 5). However, when the S shell was split into Cu-S(thiolate) and Cu-S(met) components and the measured DW of the Cu-S(Met) shell in the reduced WT enzyme (0.012 Ų, Table S2, fit 1) fixed in the simulation, the Cu-S(thiolate) refined to 1.0 S per Cu with a DW of 0.004 Ų (Table S2 fit 3). Furthermore, limits can be put on Cu-S(thiolate) DW factors from other studies that have established values >0.010 A² for Cu-S(Met) interactions^31–34, and values of >0.005 Ų for Cu-S(thiolate) in mixed histidine/cysteine containing systems.³⁵ Specifically, the DW factor (2σ²) for the homocysteine analogue of azurin³⁶ (C112hCys) is 0.006 Ų while that of the bridging thiolates of CuA type sites³³ is >0.009 Ų. Thus, a value of 0.006 Ų is close to the lower limit expected for mixed His/Met/Cys or hCys coordination providing validation of the bridged thiolate interpretation. Parameters used to construct these fits are listed in Table S2³⁷.

Figure 1.

Fourier transforms and k³-weighted EXAFS (insets) for the reaction between PHM and AAF-hCys. Experimental (black) versus simulated (red) spectra for the Cu K edge of (a) the AAF-hCys complex of ascorbate-reduced WT PHM (b) the AAF-hCys complex of ascorbate-reduced M314H simulated with 1S per Cu and (c) the AAF-hCys complex of ascorbate-reduced M314H simulated with 0.5 S per Cu. Metrical details extracted from the simulations are listed in Table S2.

To remove ambiguities associated with the presence of the M314 sulfur ligand, we carried out an identical reaction with the M314H derivative which lacks the coordinating Met ligand. The ascorbate-reduced protein gave an EXAFS spectrum (Fig. 1(b)) with the same intense S wave now simulating to 1.0 Cu-S at 2.23 Šwith a DW factor (0.006 Ų) similar to the WT protein (Table S2 fit 9). Simulations that restricted the shell occupancy to 0.5 (expected for non-bridging thiolate coordination) were completely inadequate (Fig. 1(c) and Table S2, fit 10). This result confirms the assignments made for the WT protein and establishes that the thiol must bridge the two Cu atoms in a binuclear configuration.

Next we examined the reaction chemistry of AAF-hCys with the fully oxidized PHM. Binding of ~2.5 equivalents elicited a purple species with broad intense absorption at 925 nm (estimated ε = ~1300 M⁻¹ cm⁻¹) and weaker bands at 580 and 460 nm (Fig. 2). The 925 nm feature is characteristic of intervalence charge transfer transitions (IVCT) of a mixed-valence (MV) center such as the valence localized (class II) halide complexes of hemocyanin ³⁸ or the valence delocalized (class III) centers of CuA of cytochrome oxidase and N₂O reductase ^39–45 These two classes can be distinguished by their copper hyperfine patterns in the EPR spectra, where class II give rise to 4-line patterns and class III exhibit 7-line patterns, the latter due to additional splittings arising from strong coupling of the electron spin to both copper nuclear spins. EPR spectroscopy (Fig. 3) of the PHM MV complex confirmed that ~30% of the copper was undetectable while reaction with ascorbate led to an immediate further reduction in intensity to 35% of the original oxidized enzyme with no loss of color in the first 3 minutes. These data are consistent with the initial presence of a mixture of MV and an oxidized PHM species in a ~65:35 ratio⁴⁶. Serendipitously, while the residual oxidized species is rapidly reduced by ascorbate, reduction of the MV species is slow (Fig. S3), allowing it to be observed in the EPR of the 3-minute ascorbate-reduced sample with no interference from any residual oxidized PHM species (Fig. 3 and Fig. S4). Simulation using EASYSPIN⁴⁷ suggests that the unpaired spin is localized at low temperature since the spectrum shows a 4- rather than 7-line pattern in the parallel region (Fig. 3 top panel and Fig. S4 and Table S1 fit 1). The species remaining after 3-min ascorbate reduction is best described by a 2-component system in a 3:1 ratio, with the major component giving g and A values consistent with a Cu(II)-thiolate species (g₃ = 2.21, A₃ = 389 MHz)³⁵, and the minor component (g₃ = 2.30, A₃ = 393 MHz) more characteristic of mixed O/N coordination. Inclusion of Cu-Cu nuclear interaction did not improve the simulation (Table S1 fit 2, and Fig. S4). The data are consistent with a MV thiolate-bridged binuclear species as shown in Scheme 1 exhibiting localized class II behavior at low temperature. Whereas the UV/vis data suggest that the Cu(II) component of the MV complex is type 1-like with weak S to Cu(II) CT at 460 nm, and stronger S to Cu(II) at 580 nm, the EPR spectra are more typical of a “green” or “red” than a “blue” copper site ³⁵. These preliminary assignments are supported by EXAFS of the MV species which shows a long Cu-S distance of 2.28 Å (Fig. S5, Table S2 fit 11).

Figure 2.

Titration of fully oxidized PHM with Ala-Ala-Phe-hCys.

Figure 3.

EPR spectra of WT PHM titrated with AAF-hCys. Bottom panel: (a) WT oxidized PHM (blue); (b) WT + 3 equivalents AAF-hCys (pink); (c) 3 minute reduction by 5 mM ascorbate (green), and (d) 20 minute reduction by 5 mM ascorbate (black). Top panel: Simulation of the 3 minute reduced sample using Easyspin. Best fit (Fit 1 Table S1) was a 2-component system in a 1 : 0.35 ratio.

The NIR band observed in the PHM-AAF-hCys complex is typical of an IVCT were one copper is in the Cu(II) state and the other in the Cu(I) state. These IVCTs arise from transitions from a doubly occupied HOMO centered on the Cu(I) center to a singly occupied LUMO centered on the Cu(II) center. Under the Robin-Day classification, class II MV complexes arise from weakly coupled binuclear sites with a thermal barrier to intersite ET, and class III from strongly coupling sites with the coupling energy H_AB > the reorganizational energy λ, and a low to non-existent thermal barrier for intersite ET⁴⁸. Both class II and class III systems require a ligand bridge to couple the two metal centers. Examples of class II includes the halide-bridged complexes of half-met hemocyanin³⁸ while class III includes the CuA site of cytochrome oxidase ^{40, 43–44, 49}, N₂O reductase^{42, 50}, purple copper azurin^51–52, pMMOD⁵³ and a bis-thiolato dicopper model.^{43, 45} Class III complexes exhibit intersite ET rates that are fast on the timescale of electronic transitions, resulting in full delocalization of the unpaired electron over both metal centers. Class II systems on the other hand have slower rates of ET and often show a temperature dependence due to the thermal barrier⁵⁴.

The class II behavior observed for PHM suggests that the bridging thiol provides weak coupling between the H and M centers, insufficient to overcome the reorganizational barrier to bringing the cupric center into a Cu(I)-favorable state, likely arising from the chemical inequivalence of the H and M centers. We note that class III to class II interconversion has been observed due to ligand set perturbation of the CuA center of purple copper azurin ⁵⁵. Notwithstanding, the IVCT provides strong confirmatory evidence that the binding of the AAF-hCys peptide induces the closed conformation and, when considered alongside the EXAFS data showing thiolate bridging in the fully reduced state, provides compelling evidence for a binuclear state.

The stability of the MV state as evidenced by its slow reduction is notable. Wang and coworkers have proposed a reaction mechanism for DBM (and by inference PHM²⁰) from QM/MM calculations involving a MV binuclear intermediate with lower activation energy for H-atom abstraction than its mononuclear analogue²⁶ (Scheme 2). Our current work suggests that reaction chemistry leading to such a species is accessible.

Scheme 2.

Proposed mixed valence intermediate adapted from reference 26

A binuclear mechanism could explain many experimental findings that are inconsistent with the canonical mononuclear mechanism. In addition to structural evidence for sub-domain dynamics^{2, 27–29}, the following observations can be cited. First, mononuclear protein models of the CuM site in aqueous solution show sluggish reactivity towards O₂³¹ despite demonstrated stability of inorganic Cu(II)-superoxo complexes in organic solvents at low temperature;^56–67 second, ν(CO) for the CuM-CO complex is red shifted by 30 cm⁻¹ in the presence of substrate reminiscent of the semi-bridged carbonyls of bi- or hetero-bimetallic sites^{2, 15, 68–69}; third, substrate binding is always coupled to product hydroxylation establishing a requirement for substrate activation²; and fourth, substrate hydroxylation from oxidized enzyme and hydrogen peroxide is viable and catalytic, but leads to approximately 60 percent ambient ¹⁶O exchange when H₂¹⁸O₂ is used as the oxygen source⁷⁰. This requires an intermediate where Cu(II)-peroxide can equilibrate with Cu(I)-dioxygen, also suggestive of a binuclear state. Our future efforts will be directed towards generating binucleating substrate-analogues that bring us closer to illuminating these important questions.

ABBREVIATIONS

ET

electron transfer

PHM

peptidylglycine monooxygenaase

DBM

dopamine β-monooxygenase

AAF-hCys

Ala-Ala-Phe-homocysteine

MS

mass spectrometry

MV

mixed-valence

DW

Debye-Waller factor

IVCT

intervalence charge transfer