The Catalytic M-center of Copper Monooxygenases probed by Rational Design. Effects of Selenomethionine and Histidine Substitution on Structure and Reactivity.

Abstract

The M-centers of the mononuclear monooxygenases peptidylglycine monooxygenase (PHM) and dopamine β-monooxygenase (DBM) bind and activate dioxygen on route to substrate hydroxylation. Recently we reported the rational design of a protein-based model wherein the CusF metallochaperone was repurposed via a His to Met mutation to act as a structural and spectroscopic biomimic. The PHM M-site exhibits a number of unusual attributes including a His₂Met ligand set, a fluxional Cu(I)-S(Met) bond, tight binding of exogenous ligands CO and N₃⁻, and complete coupling of oxygen reduction to substrate hydroxylation even at extremely low turnover rates. In particular, mutation of the Met ligand to His completely eliminates catalytic activity despite the propensity of Cu^I-His₃ centers to bind and activate dioxygen in other metalloenzyme systems. Here we further develop the CusF-based model to explore methionine variants where Met is replaced by selenomethionine (SeM) and histidine. We examine the effects on coordinate structure and exogenous ligand binding via XAS and EPR and probe the consequences of mutations on redox chemistry via studies on the reduction by ascorbate, and oxidation via molecular oxygen. The M-site model is 3-coordinate in the Cu(I) state and binds CO to form a 4-coordinate carbonyl. In the oxidized forms the coordination changes to tetragonal 5-coordinate with a long axial Met ligand which like the enzymes is undetectable at either the Cu or Se K edges. The EXAFS data at the Se K-edge of the SeM variant provides unique information on the nature of the Cu-methionine bond which is likewise weak and fluxional. Kinetic studies document sluggish reactivity of the Cu(I) complexes with molecular oxygen and rapid rates of reduction of the Cu(II) complexes by ascorbate, indicating a remarkable stability of the Cu(I) state in all three derivatives. The results show little difference between the Met ligand and its SeM and His congeners and suggest that the Met contributes to catalysis in ways that are more complex than simple perturbation of the redox chemistry. Overall the results stimulate critical re-examination of the canonical reaction mechanisms of the mononuclear copper monooxygenases.

Graphical Abstract

Introduction

Methionine histidine and cysteine are common ligands for copper in biological systems. Because these ligands can bind to both oxidation states of copper, they are often found together in proteins whose function is electron transfer such as the blue copper proteins azurin and plastocyanin,^{1, 2} and purple CuA-containing proteins cytochrome oxidase,^{3, 4} N₂O reductase,^{3, 5} and engineered azurins.^{2, 6} Very recently, a CuA site formed at a dimer interface has been identified in pmoD, an accessory protein involved in particulate methane monooxygenase assembly and/or metalation.^{7, 8} Met ligation is also common in periplasmic transporters such as CopC,⁹ PCuaC,¹⁰ and the CusCBAF exporter,^11–13 where the methionines are either found as the sole Cu(I) binding residues,^14–16 or in a mixed complex with histidine.^{17, 18} These latter systems are designed to stabilize the Cu(I) and/or Ag(I) states, and in most cases adopt multi-met motifs where the preponderance of uncharged thioether donors inhibit oxidation of copper to its dicationic state.¹⁹

The family of mononuclear monooxgenases that encompass the enzymes PHM (peptidylglycine monooxygenase) DBM (dopamine β-monooxygenase) and TBM (tyramine β-monooxygenase) are unusual in that they exhibit a His₂Met ligand set at the copper center (CuM) that binds oxygen and activates it towards hydroxylation of the substrate bound nearby.^20–22 These proteins contain a second copper center (CuH) which is coordinated by three histidines, but whose role appears to be electron transfer.^23–27 The role of the Met ligand in the catalytic chemistry has been enigmatic, for a number of reasons. First it is absolutely essential to catalysis, second it only coordinates in the reduced form, and thirdly it appears to be only weakly bound to copper in the state that binds oxygen.^28–31 Mutagenesis of the catalytic Met to Ile³² His,^{29, 30, 33} and Cys^{30, 34} has largely failed to illuminate the electronic role, as these mutants are either inactive due to copper loss, or in the case of Cys, lead to suicide inhibition.^{30, 34} Studies^{35, 36} on the reactivity of thioether containing inorganic models in aprotic solvents at low temperature are also inconclusive as to whether thioether coordination is expected to amplify or leave unchanged the electrophilic properties of the cupric superoxo ³⁷ believed to be the reactive species in these enzymes.^38–42 Understanding the role of this essential yet chemically silent residue has thus been challenging.

Selenomethionine (SeM) labeling is a useful technique for isolating the contributions to electronics and structure from thioether coordination, and relating these to function. This approach has been used successfully to assess the role of the M121 ligand to the spectroscopy of azurin,⁴³ and the M160 ligand to the structure of the CuA center in T. thermophilus cytochrome oxidase.⁴⁴ The utility of the approach derives in part from the ability to isolate and track the contribution to bonding via measurements at the Se K absorption edge, which report directly on the Se-Cu bond length and its force constant via the magnitude of its Debye Waller factor. In addition, SeM labeling has been leveraged to study the reaction mechanisms of metalation or metal transfer between chaperone target pairs where each has Cu-Met coordination, but where one member of the pair can be selectively labeled with SeM.⁴⁵ In this way we have reported the identity of intermediates in the metallation of the CuA centers in an isolated subunit 2 of cytochrome oxidase⁴⁶ and engineered CuA azurin,⁶ determined the pathway and direction of metal flux in the CusCBAF exporter⁴⁷ and identified a shared ligand intermediate of metal transfer between the CusF and CusB components of this metal export complex.⁴⁸ These studies share the requirement that all Met residues in the proteins be ligands to copper in order to provide maximum sensitivity to the Se label.

Against this background SeM labeling should be an ideal approach to assess the structural and electronic contributions of coordinated methionine to the catalytic chemistry of PHM and its homologues. Unfortunately, two factors have hindered application of the method. First the catalytic core of PHM residues (42 −356) contains 16 Met residues only one of which is a ligand to copper, and second the protein is expressed in a mammalian cell line which precludes simple SeM incorporation via supplementation of the medium due to the transmethylation of homocysteine by methionine synthase. Recently we reported the construction, bacterial expression, and characterization of a protein-based model for the M-center of PHM, built from a small metallochaperone (CusF) scaffold, and carrying (via mutation) the His₂Met ligand set.⁴⁹ The spectroscopy and exogenous ligand binding properties of this model have validated its close resemblance to the M-site of PHM, and furthermore it contains a single Met residue which acts as a ligand to the copper center. This model system therefore satisfies all the conditions for using SeM labeling to interrogate the structural and electronic properties of the His₂Met ligand set. The model system also allows for facile substitution of the Met ligand by histidine, forming a derivative that in the enzyme is inactive.

Here we report spectroscopic studies of the Cu(I) and Cu(II) states of the SeM and His variants of the CusF M-site model, and discuss the results in terms of how the data informs the structure and function of the mononuclear monooxygenase family of metalloenzymes. We examine the effects on coordinate structure and exogenous ligand binding via XAS, FTIR, and EPR and probe the consequences of mutations on redox chemistry via studies on the reduction by ascorbate, and oxidation via molecular oxygen. The EXAFS data at the Se K-edge of the SeM variant provides unique information on the nature of the Cu-methionine bond while kinetic comparisons document a remarkable stability of the Cu(I) state in all three variants. The results show little difference between the Met ligand and its SeM and His congeners and suggest that the Met contributes to catalysis in ways that are more complex than simple perturbation of the redox chemistry.

Materials and Methods

Construction of CusF Mutants.

W44A M49H and W44A M47M49H mutations were introduced into CusF M8M59Ile double mutant background as previously described.⁴⁹

Expression and Purification of CusF W44A M49H and W44A M47M49H variants.

S(Met) CusF variants were expressed and purified as previously described.⁴⁹

Expression and Purification of Selenium (SeM) labeled CusF M49H.

The pETDuet-1-CusF plasmid containing the M49H mutation was transformed into Met auxotrophic E. coli cells. Liquid culture media (LB and 100mg/mL ampicillin) was inoculated from a freezer stock and incubated overnight at 37°C with shaking at 250 rpm. A 100 μl aliquot of the overnight culture was used to inoculate 10 ml of minimal culture medium containing L-methionine. After overnight incubation at 37°C with shaking, this culture was used to inoculate a 1 L flask of minimal medium substituted with selenomethionine. After approximately 12 hours, when the cell culture reached an OD₆₀₀ of approximately 1, the culture was induced with 500 μM isopropyl β-D-1 thiogalactopyranoside. Growth was continued at 17 °C for 16–20 hr after which cells were harvested and purified as previously described.⁴⁹ Se and protein concentrations were determined by ICP-OES and BCA assay from which the Se:protein ratio was determined and used to verify the SeM substitution yield, which in all cases was found to be >95%.

Sample Preparation.

Apo CusF protein was reconstituted with 100 mM CuSO₄ at a ratio of 2.5:1 using a syringe pump operating at 25 μL/hr and allowed to incubate on ice for 1 h with stirring. Excess Cu(II) was removed by overnight dialysis in 50 mM sodium phosphate (NaP), pH 8. Se and Cu concentrations were determined by ICP-OES. Protein was determined by BCA assay and combined with Se and Cu analyses to calculated metal to protein ratios. Holo protein was then exchanged into pH 5.5 combination buffer, comprised of 50 mM each of formate, MES, and HEPES, via a rapid 4-fold dilution of the 4-fold concentrated protein into buffer. A small protein loss was observed, but the copper to protein ratios remained at 1:1 as confirmed by ICP. For Cu(I) samples, Cu(II)-reconstituted protein in pH 5.5 combination buffer was reduced anaerobically by addition of a 2 fold excess of ascorbate buffered pH 5.5. The ascorbate was then removed by two cycles of desalting on spin columns (Zeba) as previously described.⁴⁹ Cu(I) reconstituted protein could also be prepared by addition of a 2.5-fold excess of [Cu(I)(CH₃CN)_4}PF₆ in buffer supplemented with 10% acetonitrile, followed by serial dialysis against the same buffer containing 5% and 0% acetonitrile respectively. This procedure was carried out in a Coy chamber at 30 ppm O₂ and took ~24 hours to complete. Surprisingly we observed 20% oxidation of samples prepared in this way. The ascorbate reduction protocol could be carried out more expeditiously and did not lead to oxidation so was used in all future experiments.

Spectroscopic Measurements.

EPR, FTIR, XAS and UV/vis protocols, including azide titrations to determine binding constants were as described previously.⁴⁹ X-ray absorption measurements were carried out on beam lines 7.3 and 9.3 at the Stanford Synchrotron Radiation Lightsource operating at 3 GeV under continuous top-up mode. Copper spectra were collected and analyzed as described previously.⁴⁹ Se K-edge data was collected on BL 7.3 using 20 percent detuning of the monochromator, or on BL 9.3 using a Rh-coated mirror set to15 keV energy cutoff to reduce harmonic contamination of the beam. Se data were calibrated by means of a Se metal foil placed between the second and third ionization chamber. XAS was measured at 10 K in fluorescence mode using either a 30 element (BL 7.3) or 100 element (BL9.3) Ge detector (Canberra). Elastic scatter was reduced by means of a Soller slit assembly and 6μ As filter placed immediately in front of the detector, and total counts from each channel were adjusted below 50 kHz to avoid saturation. 4–6 scans of each sample were collected and averaged to improve signal to noise and the average of 4–6 scans of a buffer blank were subtracted from each averaged spectrum to correct for pre-edge curvature and As Kβ fluorescence emanating from the Soller slit filter. Data were analyzed and simulated as described previously.^{31, 47–49}

Kinetics of oxidation by O₂.

Rates of oxidation by molecular oxygen were determined using the 400 nm CT band of the azido adduct to determine the rate of formation of the Cu(II) form. Anaerobic ascorbate-free Cu(I) protein was produced as described above and then transferred to a reaction chamber and diluted with buffer previously saturated with 100 percent oxygen gas at 25°C and ambient pressure. During the course of the ensuing oxidation reaction the reaction chamber was gently purged with a stream of buffer-saturated O₂ gas to maintain the O₂ level at a constant (100%) level. Aliquots of the reaction mixture were sampled at appropriate time points and mixed with 100 mM sodium azide resulting in conversion of all oxidized (Cu(II)) species into the chromophoric azido adduct with λ_max = 400 nm. Rates were fitted to a single exponential rise to maximum using non-linear regression in SigmaPlot 14.

Stopped Flow Spectrophotometry of ascorbate reduction.

Reduction rates of Cu(II) azido derivatives by ascorbate were determined from the rate of disappearance of the 400 nm CT band using an Applied Photophysics SX20 stopped-flow instrument contained in a Vacuum Atmospheres anaerobic chamber with oxygen levels below 1 ppm. Anaerobic Cu(II) azido samples were placed in one syringe and shot against anaerobic buffer containing a 5-fold excess of ascorbate. Data were fitted to single exponential decay curves to extract pseudo first order rate constants.

Results and Discussion

Characterization of the copper complexes of the SeM-substituted monooxygenase model.

In a previous study, we reported the properties of the His₂S(Met) Cu(I) complex engineered into the protein scaffold of CusF, a small periplasmic metallochaperone.⁴⁹ In the present work we investigate the effect on the structure, electronics, adduct formation and redox properties of Met substitutions, SeM and His. Both Cu(I) and Cu(II) complexes could be prepared by reaction of the apo-protein with a small excess of either [Cu(I)(CH₃CN)₄]⁺ under anaerobic conditions, or aqueous Cu(II) salts, followed by desalting to remove excess metal ions. Cu(I) complexes could also be prepared by ascorbate reduction of the Cu(II) species. The Cu(I) and Cu(II) complexes formed with simple 1:1 stoichiometry. The structure of the complexes was studied using Cu and Se K edge XAS.

Structural characterization using X-ray absorption spectroscopy at the Cu and Se absorption edges.

The EXAFS data for the Cu(I) and Cu(II) complexes of the His₂SeM model were fit using full curved-wave multiple scattering theory with metrical parameters listed in Table 1. For the Cu(I) complex, the best fit at the Cu edge (Figure 1 and Table 1) is a 3-coordinate complex comprising two Cu-N(His) and 1 Cu-SeM interactions with Cu-N and Cu-Se distances of 1.95 and 2.43 Å respectively. The Se edge fit gave an identical Se-Cu distance. The Se-Cu bond length is typical of Cu(I)-selenoethers in both inorganic^{50, 51} and protein^{45, 47} systems but the Debye Waller (DW) factor for the Se-Cu (2σ²= 0.018 Ų) is high for a single absorber-scatterer interaction, and is further discussed below.

Table 1.

Fits obtained to the Se and Cu K EXAFS of the Cu(I)-SeM-M49H derivative of the CusF M-site model.

Sample	Shell	Fa	Nob	R (Å)c	DW (Å2)d	−ΔE0
	Selenium
SeM-M49H	Se-C	1.33	2	1.958(4)	0.004	5.2
	Se-Cu		1	2.424(9)	0.018
SeM-M49H-CO	Se-C	1.50	2	1.957(5)	0.005	5.4
	Se-Cu		1	2.506(9)	0.017
	Copper
SeM-M49H	Cu-N(His)e	0.42	2	1.945(4)	0.008	3.6
	Cu-Se		1	2.428(6)	0.017
SeM-M49H-CO	Cu-N (His)e	0.41	2	1.981(4)	0.006	1.6
	Cu-C (CO)		1	1.830(6)	0.002
	Cu-O (CO)		1	2.93(1)	0.020
	Cu-Se		1	2.468(4)	0.014
	∠Cu-C-O=172(3)°

^aF is a least-squares fitting parameter defined as $F^{\mathit{2}}=\frac{\mathit{1}}{N}\sum _{i=l}^N{k}^{\mathit{6}}{(Data-Model)}^{\mathit{2}}$ where N is the number of data points and k is the photoelectron wave vector defined as k= 2π/h√(2m_e(E-E₀))

^bCoordination numbers are generally considered accurate to ± 25%.

^cErrors in bond lengths are reported as 95% confidence limits as determined from the least squares analysis. This underestimates the true error in the distances due to experimental factors such as finite data range, errors in the phase shifts, and choice of ΔE₀ which are strongly correlated with R. True errors are probably closer to 0.02 Å for first-shell (coordinated) ligands and 0.05 Å for outer-shell (non-coordinated) ligands.

^dDebye Waller terms (DW) are calculated as exp(−2σ²k²) and reported as values of 2σ² (Ų).

^eFits included both single and multiple scattering contributions from the imidazole ring.

Figure 1.

Fourier transforms and EXAFS (insets) of the reduced (Cu(I)) CusF SeM-M49H derivative. (a) Cu edge data (b) Se edge data. Black traces are experimental data, red traces are simulations using EXCURVE 9.2. Parameters used in fitting the data are listed in Table 1.

Selenomethionine coordination to Cu(I) is expected to give rise to two peaks in the Fourier transform of the Se K EXAFS, Se-C from the ligand itself and Se-Cu from its interaction with the metal. In native CusF with the (SeM)₂ His ligand set the DW term (2σ²) was found to be 0.005(2) Ų indicating a strong Cu-SeM interaction.⁴⁸ Remarkably, the intensity of the Se-Cu interaction is significantly attenuated on transitioning from the (SeM)₂His -Cu(I) of native CusF to (SeM)His₂-Cu(I) in the PHM M-site model. This behavior has been observed previously in the S(Met)-containing model and in PHM and DBM enzymes themselves,⁴⁹ and has been interpreted in terms of fluxionality of the Cu-thioether bond. This hypothesis was based on assessment of the increasing Debye Waller factors for the Cu-S(Met) interaction in the series Cu(I)-S(Met)₃ (CusB), Cu(I)-S(Met)₂His (CusF) and Cu(I)-S(Met)His₂ (monooxygenase model). We suggested that the Cu-Met bond is inherently fluxional leading to the reversible dissociation of any one Cu-(Met), but leaving the other two ligands coordinated with a low average DW. In the (Met)His₂ case the Met ligand would always be labile and would exhibit a large Debye-Waller factor. This argument also explains why Met residues are often found lining importers and export channels of transporters such as CTR1^52–54, CusA¹² and CopA^{55, 56} since a single Cu-Met interaction is in rapid exchange between on- and –off conformers. The Se edge data provide a means of validating this hypothesis with increased confidence since the Se K EXAFS data allows the Se-metal interaction to be isolated from contributions from the other ligands. The Debye Waller terms for the series Cu(I)-Se(Met)₃ (CusB)⁴⁸, Se(Met)₂His (CusF) ⁴⁸ and the (SeMet)His₂ (PHM model) are 0.005, 0.005, and 0.018 Ų respectively. Here the high DW factor observed in the Se(Met)His₂ ligand set of the monooxygenase model, amply demonstrates the lability of a single Cu(I)-methionine bond.

The Cu K EXAFS of the Cu(I) form of the M47HM49H double mutant (3His) ligand set is unremarkable and simulates to 3 Cu-N(His) at 1.95 Å. Experimental and simulated data are shown in Fig. S1 with parameters listed in Table S1.

Reaction of the reduced derivatives with carbon monoxide.

Both S(met) and Se(Met) derivatives of the M49H PHM model react with carbon monoxide to form CO adducts. For the S(Met) derivative reported previously, the close correspondence between model and enzyme M-site carbonyl, suggested strong structural and electronic similarities. The CO stretching frequency of a metal-carbonyl is strongly perturbed via an electronic mechanism whereby metal d-electron density is back-donated into the empty π* orbitals of the triply-bonded CO ligand, reducing the bond-order and thus red-shifting ν(CO).⁵⁷ In turn, the magnitude of this effect is influenced by the donor strength of the other ligands in the complex, making ν(CO) an excellent reporter of the electronics of ligand-metal interaction. We are now able to compare the donor strength of thioether with selenoether ligands, where the softer, more polarizable selenomethionine might be expected to form stronger covalent bonds with Cu(I), and thus lead to a larger red-shift. Indeed such behavior has been documented in a selenium substituted H-cluster of the CpI hydrogenase matured using the [Fe₂(μ(SeCH₂)₂NH)(CO)₄(CN)₂]²⁻ synthon, where two seleno-azodithiolate ligands bridge the two Fe atoms of the H cluster.⁵⁸ FTIR data comparing ν(CO) for S(met) and Se(Met) ligand complexes of the M49H model are shown in Figure 2. The Se(met) derivative has a ν(CO) of 2087 cm⁻¹, just 2 wavenumbers less than the S(Met) system. For comparison, FTIR data for the CO complex of the M47HM49H double His mutant show a much larger red-shift to 2072 cm⁻¹ similar to that reported for the M314H mutant of PHM (where the single M314 residue at the M-center is replaced by His).³³ These data indicate (i) that Se substitution has minimal effect on the electronic structure of the complex and (ii) both S(Met) and Se(Met) ligands are much weaker donors than histidine.

Figure 2.

Fourier transform infrared spectra of the CO adducts of the Cu(I) CusF model system.

We also examined the structure of the Se(Met) Cu(I)-carbonyl at both Cu and Se K edges via simulation of the EXAFS spectra. The results are listed in Table 1 with simulations in Figure 3. Se K edge EXAFS provides unique information on the Se-Cu bond-length and Debye Waller factor, and shows the expected lengthening of the Se-Cu bond from 2.43 to 2.51 Å as the result of increase in coordination number to form the 4-coordinate carbonyl. Like the parent complex, the DW factor is large, indicating a weak, kinetically labile bond. Cu edge data confirm details of the Cu-Se bond, together with coordination by the two His residues and a linear Cu-CO adduct (Figure 3 and Table 1).

Figure 3.

Fourier transforms and EXAFS (insets) of the reduced (Cu(I)) CO complex of the CusF SeM-M49H derivative. (a) Cu edge data (b) Se edge data. Black traces are experimental data, red traces are simulations using EXCURVE 9.2. Parameters used in fitting the data are listed in Table 1.

The x-ray absorption near edge structure (XANES) was compared for unligated Cu(I) complexes and their CO analogues and is shown in Fig. S3. Notably the Cu(I) data are emblematic of 3-coordination with partially resolved 1s → 4p transitions at 8983 eV on the rising absorption edges. CO complex exhibit XANES features typical of 4-coordinate Cu(I) carbonyls as reported previously.⁴⁹

Oxidized Samples.

The structure of the oxidized SeM derivative was interrogated by similar methods. Cu edge EXAFS data were best fit by a 4 – 5 coordinate model involving 4 Cu-O/N scatterers at 1.96 Å with multiple scattering contributions that were consistent with two histidine and two non-His O/N donors as equatorial ligands. In one sample, a small contribution at ~2.4 A could be simulated by a weak interaction with Se (0.4 ±0.2), but was absent in other samples and is likely the result of some photoreduction in the beam. Taken together the oxidized Cu and Se edge data are most consistent with the absence of an observable Cu(II)-SeM interaction, and agree with previous conclusions from analysis of both PHM EXAFS and the S(Met) M49H model that the Met ligand is not observable in the oxidized XAS data. This is consistent with the Met residue occupying an axial position in a tetragonally distorted 5-coordinate complex, with two His and two solvent ligands bound in the equatorial plane, and confirms that the system must undergo a significant conformational change on reduction, as previously suggested.^{31, 59} Figure 4 shows experimental and simulated EXAFS data for Cu and Se edges of the Cu(II) SeM complex. EXAFS-derived metrical parameters are listed in Table 2. EXAFS of the oxidized 3His double mutant was unremarkable, and analyzed to three imidazole and one O/N scatterer as expected (Table S1, Fig. S2).

Figure 4.

Fourier transforms and EXAFS (insets) of the oxidized (Cu(II)) CusF SeM-M49H derivative. (a) Cu edge data (b) Se edge data. Black traces are experimental data, red traces are simulations using EXCURVE 9.2. Parameters used in fitting the data are listed in Table 2.

Table 2.

Fits obtained to the Se and Cu K EXAFS of the Cu(II)-SeM-M49H derivatives of the CusF M-site model.

Sample	Shell	Fa	Nob	R (Å)c	DW (Å2)d	−ΔE0
	Selenium
SeM-M49H	Se-C	3.47	2	1.963(8)	0.004	5.2
	Se-Cu		0.4(2)	2.43(4)	0.018
SeM-M49H-N3−	Se-C	2.3	2	1.973(7)	0.005	5.8
	Se-Cu		0.4(2)	2.45(3)	0.018
	Copper
SeM-M49H	Cu-N(His)e	0.54	2	1.960(4)	0.011	2.9
	Cu-N/Of		2	1.960(4)	0.011
	Cu-Se		0.4(1)	2.41(2)	0.014
SeM-M49H-N3−	Cu-N (His)e	0.54	2	1.977(4)	0.011	3.6
	Cu-N/Of		2	1.977(4)	0.001
	Cu-Se		0.4(1)	2.93(1)	0.014

^aF is a least-squares fitting parameter defined as $F^{\mathit{2}}=\frac{\mathit{1}}{N}\sum _{i=l}^N{k}^{\mathit{6}}{(Data-Model)}^{\mathit{2}}$ where N is the number of data points and k is the photoelectron wave vector defined as k= 2π/h√(2m_e(E-E₀))

^bCoordination numbers are generally considered accurate to ± 25%.

^cErrors in bond lengths are reported as 95% confidence limits as determined from the least squares analysis. This underestimates the true error in the distances due to experimental factors such as finite data range, errors in the phase shifts, and choice of ΔE₀ which are strongly correlated with R. True errors are probably closer to 0.02 Å for first-shell (coordinated) ligands and 0.05 Å for outer-shell (non-coordinated) ligands.

^dDebye Waller terms (DW) are calculated as exp(−2σ²k²) and reported as values of 2σ² (Ų).

^eFits included both single and multiple scattering contributions from the imidazole ring.

^fIn cases where the resolution ΔR for split histidine and non-histidine shells is less than the theoretical resolution of the data (π/2k), histidine and non-histidine scatterers are simulated as a single shell.

XANES data for the Cu(II) complexes and their azido adducts are shown in Fig. S3 and indicate a slightly lower energy for the SeM derivative than for the 3His homologue. The XANES data for the oxidized samples were extracted from single scans, and were collected under fast scanning conditions to eliminate photoreduction. Notwithstanding, the small red-shift observed for the SeM derivative is consistent with some photoreduction as discussed above for the EXAFS analysis. Surprisingly, the SeM derivative shows a lower tendency towards chemical reduction (vide infra), suggesting a photoprocess that implicates an excited state.

EPR data for oxidized SeM-M49H and the M47H49H 3His variant are shown in Fig. S4. Like the M49H counterpart, the spectra were complex and indicative of multiple species. Double integration of the spectra gave spin concentrations less than the expected total copper concentrations but greater than reported previously for the parent S(Met) complex (78% for SeM and 71% for 3His respectively). Due to the complexity of the system, studies on these unligated Cu(II) derivatives were not pursued, rather we focused on characterization the azido adducts, which as previously reported, generate EPR spectra with closer to stoichiometric EPR detectable Cu(II) levels.

Reaction of the oxidized derivatives with azide.

Both Cu(II)-SeM and 3His derivatives react with azide to form azido adducts with absorption maxima at 400 nm (Table 3). Titration data (Fig. S5) reveal formation of 1:1 complexes with K_D= 5.2 and 3.5 mM respectively, but for SeM at higher azide concentrations, the absorbance at 400 nm (corrected for dilution effects) falls. The most plausible interpretation of this behavior is photobleaching of the signal due to photoreductive processes, which can be simulated by formation of a second species with lower extinction coefficient. (Although the “second” species is artfactual, inclusion in the non-linear regression analysis improves the accuracy of K_D). This behavior is not observed for either the parent S(Met) complex or for the 3His derivative where the titration data fit well to a single mono azido adduct. Table 3 compares binding parameters for S(Met), Se(Met) and 3His derivatives. EPR spectra of the azido species are shown with simulations in Fig. 5 with relevant spectral parameters in Table 3. Notably the spin concentrations determined by double integration are increased relative to the unligated derivatives and approach 85 percent.

Table 3.

Spectroscopic and kinetic data for ligand complexes of M49H and M47HM49H model complexes.

Parameter	M49H-S(Met)	M49H-Se(Met)	M47HM49H
ν(CO) (cm−1)	2089	2087	2072
λmax azido (nm)	395	400	400
ε (M−1cm−1)	1805 ± 1	2269 ± 3	1645 ± 2
KD	3.42 ± 0.02	5.2 ± 0.2	3.50 ± 0.04
% Detectable Cu(II)	89 ± 4	79 ± 4	83 ± 2
gz comp 1	2.253	2.254	2.271
Az comp 1 (MHz)	476	486	483
gz comp 2	2.267	2.271	2.281
Az comp 2 (MHz)	390	398	395
Ratio	0.3 ± 0.1	0.3 ± 0.1	0.5 ± 0.1
Rate of reduction (s−1)	57 ± 0.6	28 ± 0.5	277 ± 4.3
Rate of oxidation (s−1)	6 ± 1 × 10−5	14 ± 3 × 10−5	13 ± 1 × 10−5

Figure 5.

X-band CW EPR spectra of the Cu(II) azido adducts of the CusF M-site model. Black traces are experimental data while red traces are simulations using EASYSPIN. (a) SeM-M49H (b) 3His. Spectra were collected at a temperature of 100 K, microwave frequency 9.678 GHz, 100 Khz modulation, 10 G modulation amplitude, 20 mW microwave power and 1000 G sweep width with the field centered at 3100 G. Both spectra were simulated using 2-components with ratios (component 2 : component 1) and EPR parameters listed in Table 3.

Oxygen reactivity of reduced complexes.

The formation of simple 1:1 chromophoric azide adducts provided us with a tool to assess the relative reactivity of the three M-site model derivatives. First we explored the reactivity with molecular oxygen. Previous work⁴⁹ established that the fully reduced parent S(met) complex was unreactive to molecular oxygen when a stoichiometric amount of ascorbate-reduced sample was added to an air-saturated buffer in an oxygen electrode, or when a catalytic amount of reduced or oxidized sample was reacted with air-saturated buffer in the presence of excess reductant. We repeated these experiments and again found neither stoichiometric oxidation, nor catalytic oxidation of ascorbate on the time-scale of the experimental protocol (0–5 minutes). Therefore we designed an additional experiment wherein fully reduced, ascorbate-free complex was allowed to react with oxygen-saturated buffer for several hours during which aliquots were removed and treated with 100 mM sodium azide. This formed an azido adduct with any oxidized product that had formed which was immediately analyzed at 400 nm. In a separate experiment we added 100 mM sodium azide to the reaction mixture at time zero and monitored the formation of the 400 nm Cu(II)-azido peak at defined time points during the course of the oxidation reaction. The results are shown in Figure 6 (parent M49H complex) and Figs S6 and S7 (SeM and 3His) with rate constants listed in Table 3. The analysis indicates extremely slow rates of oxidation which are either comparable or reduced in the presence of azide, suggesting that azido adduct formation does not accelerate the rate via favoring the Cu(II) state.

Figure 6.

(a) Oxidation of CusF M49H PHM M-site model with oxygenated buffer at 400 nm over 24hrs. The reaction was sampled at 0min, 1min, 5min, 15min, 40min, 60min, 90min, 180min, 360min and 1440min. (b) Absorption at 400 nm plotted versus time and fit to a single exponential rise to maximum with a rate of constant of 6 × 10⁻⁵ s⁻¹

Rates of reduction of the azido adducts by ascorbate (stopped flow).

We also determined the rates of reduction of the oxidized azido derivatives by ascorbate. These rates were much faster than the corresponding rates of oxidation, and required stopped flow. Because the speciation of the azido adducts is largely homogeneous, the rate constants should follow the relative value of the redox potentials of each derivative, and lead to an assessment of the role of the S/Se(Met) ligand in setting this potential. Figure 7 shows data for the parent S(met) derivative fit to a single exponential decay while similar data for SeM and 3His derivatives are shown in Fig. S8. Rate constants are listed in Table 3. Whereas all three derivatives undergo rapid reduction, thio- or selenoether ligation appears to provide less stabilization of the Cu(I) state than histidine, with the selenomet derivative exhibiting the slowest rate of reduction, and the 3His being almost ten times faster. This result is notable since it excludes redox stabilization of the Cu(I) state in PHM/DBM by methionine as a rationale for the essential role of Met over His in catalysis.

Figure 7.

Stopped-flow measurements of the reduction of the CusF M-site model M49H with a 5X excess of buffered ascorbate at 400 nm over 50 ms reaction time. Data were base-line corrected with respect to pure buffer and simulated by a single exponential decay with rate constants as listed in Table 3.

Conclusions

Using a rationally designed protein model complex we have investigated the effect of methionine mutation on the reaction chemistry of the His₂Met ligand set that is central to the catalytic activity of mononuclear copper monooxygenases PHM and DBM. Selenomethionine substitution coupled to XAS at the Se edge has provided a unique description of thioether/selenoether coordination which when correlated with Cu EXAFS, has led to an accurate description of the His₂Met reaction center. The approach has determined that the M-site model is 3-coordinate in the Cu(I) state with a highly fluxional Cu-S(Met) bond, and increases to tetragonal 5-coordinate with a long axial Met ligand in the Cu(II) state. Histidine substitution is also informative particularly since the Met to His mutation eliminates activity in the enzymes. Comparison of His with Met has shown that thioether coordination has little effect on redox chemistry, nor does it alleviate the sluggish reactivity of the His₂Met ligand set towards oxygen. These observations provide insights into the role of the Met ligand that have been difficult to determine from studies of the enzymes themselves on account of the overlapping spectroscopic signals and reactivity of the CuH and CuM centers.

The 3-coordinate description of the M center is based on the following three pieces of evidence: (i) the absorption edges are characteristic of 3-coordination (ii) EXAFS analysis does not support the presence of more than 3 ligands, one of which is the thio- or selenoether and (iii) adduct formation with CO increases the coordination number to 4, supported by both the overall EXAFS analysis and the observed increase in Cu(I)-S or Cu(I)-Se distance in the carbonyl models. Indeed the availability of data from both Cu and Se K EXAFS analysis of the Se(Met) derivative provides compelling evidence for this increase, which is expected to accompany coordination expansion. Additionally, solution data on the oxidized states of the model and its S/Se derivatives are only consistent with 5-coordinate (His)₂ (O/N)₂X coordination with X=S(Met) or Se(Met) occupying an axial position. The latter result reproduces the conformational change from a 3-coordinate distorted trigonal Cu(I) to a 5-coordinate tetragonal Cu(II) state that has been documented by us and others for the M-center in the enzymes.^{27, 31, 59, 60}

Importantly, the copper coordination of the His₂Met ligand set determined from the model chemistry contrasts sharply with that determined from PHM crystal structures. The crystallographic description of the Cu(I)M site in PHM as visualized in the ascorbate-reduced enzyme is 4-coordinate tetrahedral with 2 His, 1 Met and a water ligand while in the oxy form, the water ligand is replaced by a diatomic.^{25, 42} Computational studies have used the crystallographic structures as the starting point for calculations that support end-on cupric superoxo reactive intermediates that progress along a reaction coordinate with attainable activation barriers and overall exergonic thermodynamics ⁶¹. These approaches have been remarkably insightful in delineating a canonical reaction mechanism in which a CuM superoxo abstracts a H atom from bound substrate to generate a substrate-based radical followed by radical rebound and a second electron transfer from CuH. However, their validity rests on whether the M-site structures depicted in the crystal structures are representative of the active forms of the solution structures. Our data reported here as well as a substantial body of spectroscopic data ^{29, 31, 59, 60, 62} on the ascorbate-reduced enzymes that fail to observe the coordinated water molecule, suggest significant differences between crystal and solution coordination at the M-site. We also note that to our knowledge computational studies do not provide insight into why replacement of Met with His kills the reactivity.

Next we turn to the reactivity of the Cu(I)-His₂Met center with oxygen. First we note that all three (His)₂X Cu(I) model complexes are remarkably resistant to dioxygen oxidation in contrast to expectations based on computational and biomimetic investigations,^{63, 64} but consistent with reactivity of the enzymes themselves where the Cu(I) state has low reactivity towards oxygen in the absence of substrate.^{33, 65, 66}. Sluggish oxygen reactivity in the absence of substrate underscores the need for substrate triggering of catalysis, and understanding how the enzymes achieve substrate dependent oxygen activation remains a key unanswered question. For example, substrate activation may perturb second sphere contacts such as H-bonding that are important for HAA reactivity as highlighted recently in model complexes ⁶⁷ and suggested for the unusual thiolate-rich active site of formylglycine generating enzyme.⁶⁸ One possibility supposes that the 5-coordinate tetragonal oxidized M-site induces a “facial triad” (comprised of two equatorial His and one axial Met) similar to alpha-keto-glutarate-dependent Fe enzymes (which also show substrate triggering).⁶⁹ This would leave two open equatorial coordination positions allowing the end-on superoxo to adopt a cis configuration with a coordinated solvent which can then H-bond to the proximal O of the superoxide. Computational studies have also hinted that the peroxo intermediate formed after HAA with the bound substrate is bound to Cu(II) in a side-on configuration necessitating cis binding. ⁶¹

Alternatively, it may be necessary to undertake a more extensive reexamination of established tenets of the canonical monooxygenase mechanism. An important finding relevant to substrate triggering is the observation that substrate binding (in PHM) red-shifts the M-site Cu(I)-CO stretching frequency by 30 cm⁻¹ from 2093 to 2063 cm^-1.³³ In the M314H mutant substrate binding shifts the frequency to 2052 cm⁻¹ a value well below any mononuclear CuN₃CO species reported for copper proteins⁶² or models,^{70, 71} and in the range for binuclear Cu(I)-CO species such as hemocyanin and tyrosinase.^{72, 73} Red shifted CO frequencies are also observed for the bridging CO ligands of the H cluster of hydrogenase.⁷⁴ These observations are relevant to recent crystal structures of both PHM⁷⁵ and DBM²² which have documented the existence of “closed” conformations of the active site where the separation between CuH and CuM centers shrinks from 11 to 4 Å. Therefore it is possible that substrate binding to the ascorbate-reduced enzyme could trigger an open to closed conformational change which would result in the red-shifted (semi-bridging) ν(C≡O) and activate oxygen via binding to the resulting dinuclear Cu(I) center. Our laboratory is engaged in exploring this chemistry as an alternative mechanistic pathway.