|Home | About | Journals | Submit | Contact Us | Français|
Tyrosine Hydroxylase (TH) is a pterin-dependent non-heme iron enzyme that catalyzes the hydroxylation of L-tyr to L-DOPA in the rate-limiting step of catecholamine neurotransmitter biosynthesis. We have previously shown that the FeII site in Phenylalanine Hydroxylase (PAH) converts from 6C to 5C only when both substrate + cofactor are bound. However, steady-state kinetics indicate that TH has a different cosubstrate binding sequence (pterin + O2 + L-tyr) than PAH (L-phe + pterin + O2). Using x-ray absorption spectroscopy (XAS), and variable-temperature-variable-field magnetic circular dichroism (VTVH MCD) spectroscopy, we have investigated the geometric and electronic structure of the WT TH and two mutants, S395A and E332A, and their interactions with substrates. All three forms of TH undergo 6C → 5C conversion with tyr + pterin, consistent with the general mechanistic strategy established for O2-activating non-heme iron enzymes. We have also applied single-turnover kinetic experiments with spectroscopic data to evaluate the mechanism of the O2 and pterin reactions in TH. When the FeII site is 6C, the two-electron reduction of O2 to peroxide by FeII and pterin is favored over individual one-electron reactions, demonstrating that both a 5C FeII and a redox-active pterin are required for coupled O2 reaction. When the FeII is 5C, the O2 reaction is accelerated by at least 2 orders of magnitude. Comparison of the kinetics of WT TH, which produces FeIV=O + 4a-OH-pterin, and E332A TH, which does not, shows that the E332 residue plays an important role in directing the protonation of the bridged FeII-OO-pterin intermediate in WT to productively form FeIV=O, which is responsible for hydroxylating L-tyr to L-DOPA.
Mononuclear non-heme iron enzymes catalyze a wide variety of reactions involving O2. Tyrosine Hydroxylase (TH, EC 184.108.40.206) belongs to the family of enzymes that utilize FeII and a pterin cofactor to activate dioxygen for aromatic amino acid hydroxylation (Scheme 1).1–3 In humans, TH is found in the central and peripheral nervous systems and catalyzes the rate-limiting step in catecholamine neurotransmitter biosynthesis by converting L-tyrosine into L-DOPA. L-DOPA is subsequently converted to dopamine, norepinephrine and epinephrine, which play key roles in regulating motor-coordination, behavior, learning and memory, sleep-wake cycle and the stress response. Dysfunction of this enzyme has been linked to several neurological disorders including Parkinson’s disease,4 bipolar disorder,5 and Segawa’s syndrome.6,7 The other two pterin-dependent hydroxylases also play key roles in proper neurofunction: Phenylalanine Hydroxylase (PAH) converts phenylalanine into tyrosine, and its dysfunction results in severe mental retardation (phenylketonuria, PKU).8,9 Tryptophan Hydroxylase makes 5-hydroxytryptophan in the first step of serotonin biosynthesis and plays an important role in mood regulation.10
In bacterial PAH the cosubstrates are reported to bind in a specific order – phenylalanine, then pterin, and finally O2, for productive enzyme turnover,11 although there is some uncertainty in this.12 Using magnetic circular dichroism (MCD) and x-ray absorption spectroscopies (XAS), we have found that the resting ferrous site of rat PAH is 6-coordinate (6C).13 Binding of either phenylalanine or 5-deaza-6-methyltetrahydropterin (5-deaza-6MPH4) only slightly perturbs the 6C FeII site, whereas binding of L-phe + 5-deaza-6MPH4 converts the site to 5-coordinate (5C) through the loss of a water ligand, and opens a coordination position for O2 to react.14,15 This structural change at the PAH active site was subsequently observed by crystallography.16,17 The crystal structures showed that the resting 6C FeII site was coordinated by 2 histidines (His), 1 glutamate (Glu) and 3 waters. When substrate analogs (norleucine or thienylalanine) and cofactor are bound, the FeII site becomes 5C, with the Glu binding in a bidentate fashion.
A different sequence of co-substrate binding was observed with steady-state kinetics in TH, where the order is first pterin, followed by O2 and then tyrosine before product release.18 In fact, the binding of tyrosine to FeIITH results in a dead end complex. These findings suggest that the FeII active site in TH might undergo a different structural mechanism as compared to PAH, even though the enzymes share high sequence homology 3 and perform similar reactions. However, there is limited structural information available on TH. The two crystal structures of TH are of the enzyme in the ferric form, one in resting state (1TOH) and the other with oxidized cofactor, 7,8-dihydrobiopterin, bound (2TOH).19,20 Both of these structures show a five-coordinate FeIII site, with 2 His, 1 Glu and 2 water ligands; the presence of oxidized pterin did not affect the structure of the active site. Mössbauer studies on TH confirmed that the FeII is high spin S = 2, and XAS studies on the resting FeIITH and FeIITH + tetrahydrobiopterin (BH4) indicate that both sites are 6C.21 The electron paramagnetic resonance (EPR) spectrum of FeIIITH is characteristic of a high-spin, predominantly rhombic site.22 At present, there is no structural information on catalytically relevant FeIITH with tyrosine or tyrosine + pterin bound.
When both pterin and tyrosine are present, TH catalyzes the dioxygen dependent hydroxylation of both molecules (Scheme 1). The O2 molecule is partitioned between the hydroxylated substrate and cofactor,23–25 leading to the proposed formation of a putative FeII-peroxy-pterin intermediate (FeII-OO-pterin),24 which undergoes heterolytic O-O bond cleavage to give the hydroxylating intermediate FeIV=O and a 4a-hydroxy-pterin adduct (4a-OH-pterin). The FeIV=O intermediate in TH was recently trapped and studied using Mössbauer spectroscopy.26 The 4a-OH-pterin subsequently dehydrates to produce quinonoid pterin.
Although a mechanism for the formation of the FeIV=O hydroxylating intermediate has been proposed and generally accepted,27–29 earlier steps in the pterin-dependent reaction, specifically the steps leading to the formation of the putative FeII-OO-pterin intermediate, are still poorly understood (Scheme 2). In a computational study, Siegbahn and colleagues have proposed that O2 first reacts with FeII to form a Fe3+-O2−• complex.30 However, this process is thermodynamically unfavorable by ~ 20 kcal/mol.31 In contrast, spectroscopic and kinetic studies on two PAH mutants suggested a different mechanism in which O2 reacts first with pterin to form a peroxy-pterin species (HOO-pterin) before bridging to FeII.32
In this study, Fe K-edge XAS edge and EXAFS (extended x-ray absroption fine structure) spectroscopy combined with near-IR MCD and variable-temperature-variable-field MCD (VTVH MCD) are applied to investigate the geometric and electronic structure changes of the FeIITH active site and the effects of tyrosine and pterin binding. We also studied two mutants of TH that have different reactivity towards the pterin cofactor to correlate these differences in reactivity to active site structural differences. The homologous mutations in PAH have been shown to result in PKU.33,34 The S395A mutant produces 4a-OH-pterin at the same rate as WT, but does so in predominantly uncoupled reaction (2% of WT tyrosine hydroxylation).35 The S395 residue in TH hydrogen-bonds to the His336 FeII ligand and is in the vicinity of the tyrosine binding pocket. The E332A mutant hydroxylates less than 1% tyrosine compared to WT and does not produce 4a-OH-pterin.36 The E332 residue is hydrogen-bonded to the first sphere ligand His331 as well as to the pterin cofactor. Together, these mutants probe the mechanism of the oxygen/pterin reaction in TH. These results are correlated to the results of new single-turnover kinetic studies on WT TH and the E332A mutant to evaluate the mechanisms of O2 reaction, formation of a bridged species, and O-O cleavage.
Wild-type (WT), E332A and S395A rat TH were overexpressed in E. coli and purified as previously described.35–37 After purification, the enzyme was precipitated with 70% ammonium sulfate and dissolved in deuterated MOPS/KCl buffer, pD=7.3, (containing either 5 mM EDTA for WT and E332A or 1 equiv of ferrous ammonium sulfate, Fe(NH4)2SO4, per monomer for S395A), followed by dialysis against the same buffer. Glycerol-d3 was then added during concentration steps to give final buffer conditions of 100 mM MOPS, pD=7.3, 65% glycerol-d3 and 300 mM KCl. Iron contents were determined using a Perkin-Elmer Analyst 700 atomic absorption spectrophotometer.38
All commercial reagents were of the highest grade and used without further purification. Solutions were made anaerobic either by ten freeze-pump-thaw cycles or by purging with argon on a schlenk line. Solids were purged with argon for 1–2 hours. Anaerobic reagents were transferred into a nitrogen atmosphere maintained in a Vacuum Atmosphere Nexus-1 glovebox (< 5 ppm oxygen). Fe(NH4)2SO4 (J.T. Baker) and sodium dithionite (Sigma-Aldrich) were dissolved in anaerobic D2O (99.9 atom % D). 6-Methyltetrahydropterin (6MPH4, Schircks Laboratories) was dissolved in an anaerobic solution of 10 mM HCl. The concentration of the 6MPH4 stock solution was determined from the absorbance at 266 nm (ε266nm = 17.8 mM−1cm−1). The redox inactive cofactor 5-deaza-6MPH4 was synthesized according to literature procedures.39 Concentrated solutions of L-tyrosine (L-tyr) were prepared in D2O adjusted to pD = 10 using NaOD.
Enzyme samples were made anaerobic by purging with argon. In the glovebox, approximately 0.85 equiv of Fe(NH4)2SO4 was added to the enzyme, followed by five reducing equiv of dithionite solution. The FeII-TH samples were incubated for 10 minutes, then tyrosine and 6MPH4 were added as needed and the samples incubated for another 10 to 20 minutes. Samples for XAS experiments were each syringed into Lucite 23 × 2 × 3 mm pinhole type XAS cells with 37 μm Kapton windows. MCD samples were injected into an MCD cell, which consists of two Infrasil quartz disks (Esco Products) separated by a neoprene spacer (0.3 cm pathlength) and held in place by two fitted copper plates. After removal from the glovebox, the XAS and MCD samples were quickly frozen in liquid nitrogen and stored at 77K.
Samples for spectroscopy included TH-[ ], TH-[L-tyr], TH-[6MPH4], TH-[5-deaza-6MPH4] and TH-[L-tyr, 6MPH4] and typically contained 0.8–2 mM FeII, 2–5 mM tyrosine, 3–5 mM 6MPH4 (or 5-deaza-6MPH4) and approximately 60% v/v glycerol-d3.
X-ray absorption spectra were recorded at the Stanford Synchrotron Radiation Laboratory (SSRL) on beam lines 7–3 40 and 9–3 41 over an energy range of 6785 to 8000 eV (k = 15 Å−1). Hutch slits were set at 2 × 10 mm and 1 × 4 mm on beam lines 7–3 and 9–3 respectively, with the first inflection point of the Fe foil set to 7111.2 eV as an internal calibration.42 Fluorescence data were collected using a solid-state 30-element Ge detector43 with a Mn filter and Soller slits aligned at 90°, with sample orientated at 45° to incident beam. Sample cryogenic temperature of 10K was maintained using an Oxford Instruments CF1208 continuous-flow liquid helium cryostat. Data collection was continuously monitored in order to ensure sample integrity by averaging blocks of two scans and comparing to the other previous averages. No sample degradation was observed in the edge or EXAFS data in any of the samples, thus all scans were used in the final average. Total scan numbers varied based on sample concentration and desired data quality resulting in 22, 31, 39, and 20 scans for TH-[ ], TH-[L-tyr], TH-[6MPH4], and TH-[L-tyr, 6MPH4] respectively. Reduction and normalization of the data was performed according to established methods44–46 using the program PySpline47 for background subtraction and spline fitting. In order to compare multiple data sets, the full data set was truncated to k= 9.5 Å−1 for pre-edge and edge analysis. The pre-edge data (k = 9.5 Å−1) were normalized using a third-order post-edge polynomial background and a two-segment spline. For the EXAFS data (k = 15 Å−1) a second-order post-edge polynomial and three-segment spline was used in order to normalize the data.
Using the fitting program EDG_FIT,48 pre-edge features were successfully modeled using a pseudo-Voigt line shape in a 50:50 ratio of Lorentzian:Gaussian functions. The energy position, full width at half-maximum (FWHM), and peak intensity were all allowed to vary throughout the fitting process. A function modeling the background was empirically chosen to give the best fit. For a fit to be acceptable it had to reasonably match both the pre-edge data as well as that of the second derivative. In all cases three acceptable fits with different FWHM (± 0.5 fixed from float) backgrounds were acquired over the energy ranges of 7108–7116, 7108–7117, and 7108–7118 eV resulting in a total of nine pre-edge fits per data set, which were averaged to get mean values. Standard deviations for the peak energies and intensities from the successful pre-edge fits were used to quantify the error.
EXAFS signals were calculated using FEFF (version 7.0) and a ferric tyrosine hydroxylase crystal structure truncated to an iron ligated by two waters, two His and one Glu. The starting crystal structure (PDB ID 1TOH)49 was then modified with either the addition of a water ligand to make a 6C model, or modification of the Glu to bind in a bidentate fashion in order to achieve a final fit. The models were fit using OPT as part of the program suite EXAFSPAK50 by allowing the bond lengths (R) and bond variance (σ2) to float. The shift in threshold energy (k = 0, E0) was also varied but constrained as a common variable (ΔE0) for all fit paths for a given data set. The amplitude reduction factors (S02) was fixed to a value of 1.0 and coordination numbers (N) were varied systematically to accomplish the best fit to the data based on the model. All samples fit reasonably well using a combination of 4 or 5 paths except for TH[L-tyr, 6MPH4] which required two additional paths. First coordination spheres were fit using a nitrogen backscatter, with a nitrogen/oxygen mix chosen in cases with a split first shell. As oxygen and nitrogen are indistinguishable by EXAFS, best first shell fits were chosen on the basis of normalized error (F) for all the available paths. Second and outer shell fits were again chosen on the basis of best normalized error (F) for all available paths resulting in the use of carbon single scattering and carbon multiple scattering through the His rings. σ2 values were set to initial values of 0.005 Å2 and then allowed to float in all fits, even for those where single scattering and multiple scattering paths were related to one another. The best overall fit was evaluated based on comparison of the normalized error (F) for each proposed fit and by inspection of the fit to the EXAFS to the data, Fourier transform and individual EXAFS wave components. Based on EXAFS studies of known complexes, the uncertainties in final distances are within 0.02 Å.
Low temperature near-IR MCD data were collected on a Jasco J200D spectropolarimeter (600 – 2000 nm) equipped with a liquid nitrogen-cooled InSb photodiode detector and fitted with an Oxford Instruments SM 4000-7T superconducting magnet along the beam path. MCD spectra were corrected for zero-field baseline effects either by subtracting the corresponding 0T scan or by averaging the positive and negative field data at each temperature. Gaussian fitting of the MCD spectra was performed using the program PeakFit (SPSS Science). VTVH MCD data were collected at 8 temperatures between 1.6 K and 25 K and at 12 fields between 0 T and 7 T.
Single turnover reactions of WT TH and E332A TH were monitored using an Applied Photophysics SX20 stopped-flow spectrophotometer in single wavelength absorbance mode. The stopped-flow instrument was made anaerobic by incubation with a solution of sodium dithionite for at least two hours. Anaerobic solutions of FeIITH (with or without substrates) were mixed with oxygen-saturated buffer (200 mM HEPES, pH 7.5, 10% glycerol and 0.1M KCl) at 5 °C. Solutions of apoenzyme (100–300 μM in 200 mM HEPES, pH 7.5, 10% glycerol and 0.1 M KCl) were made anaerobic in a tonometer using at least 20 argon-vacuum cycles, ferrous ammonium sulfate and substrate solutions were then added under argon. Different oxygen concentrations in buffers were achieved by using a modified MaxBlend medical oxygen blender (Maxtec) to mix argon and oxygen in different ratios and then bubbling through buffer-containing syringes on ice for at least 15 min.
Single wavelength absorbance changes consisted of one or more exponential kinetic phases, which were analyzed by first fitting to single or multiple exponentials using the program KaleidaGraph (Synergy Software); results from these fits were than taken as initial estimates for global fits and simulations. Initial values for kon and koff of the reversible oxygen complex formation steps and estimates for unknown extinction coefficients were determined by global fitting of multiple traces (obtained at multiple wavelengths for varying O2 concentrations) using the program DynaFit.51 Initial concentrations were allowed to float within 10% of the experimental values. Global fitting process was performed in an iterative fashion, in which one or two parameters were varied at a time and the rest of the parameters were kept constant. Best fit parameters obtained this way were than employed for the kinetic simulation (DynaFit) of the experimental data to appropriate mechanisms.
The XAS Fe-K pre-edge feature at ~ 7112 eV, while formally a 1s → 3d electric dipole forbidden transition, gains intensity through the weaker electric quadrupole mechanism, resulting in some pre-edge intensity for centrosymmetric complexes. In lower symmetry, the metal 3d and 4p orbitals mix, and the pre-edge feature increases in intensity through the much larger electric dipole allowed 1s → 4p character.52 As a result, XAS pre-edge features are very sensitive to coordination number and geometry.52,53 For a typical ferrous model complex in a 6C environment, pre-edge intensities average about 4 units, whereas 5C structures average 12 to 13 intensity units.52 For a protein environment as in TH, pre-edge intensities for a 6C structure are larger than in symmetric model complexes due to variations in ligation, bond lengths and angles.
The pre-edge and edge spectra of four TH complexes are shown in Figure 1. The pre-edge energies and intensities for the four TH complexes listed in Table 1. The normalized data for TH-[ ] (resting, in red), TH-[L-tyr] (substrate, in blue) and TH-[6MPH4] (cofactor, in orange) all look very similar with respect to pre-edge shape and intensity. When both cofactor and substrate are present, TH-[L-tyr, 6MPH4] (green), the pre-edge shows a notable increase in intensity as well as a shift to lower energy both of which are reflected in the second derivative (Figure 1, inset). A difference is also observed in the edge slope as well as the maximum intensity of the rising edge. Small differences are also observed at higher energies above the edge (7150 eV in Figure 1), suggesting that a structural change has occurred when both substrate and cofactor are bound to TH.
For TH-[ ], the pre-edge region was accurately fit using two features at 7112.0 and 7113.8 eV with intensities of 4.4 and 3.9, respectively, for a total intensity of 8.3 units. TH-[L-tyr] and TH-[6MPH4] were both fit with two features at 7111.9 and 7112.8 eV with total intensities of 8.5 and 8.2 respectively. While the total intensities for TH-[ ], TH-[6MPH4], and TH-[L-tyr] are greater than that which is typical for Oh 6C model complexes with uniform ligation, they are similar to those found for 6C structures in FeII metalloenzymes.13,21,32,54
When both substrate and cofactor are present, the total pre-edge intensity of TH-[L-tyr, 6MPH4] increases to 12.5 units, and is fit using two bands at slightly lower energies, 7111.8 and 7113.5 eV with 6.4 and 6.1 intensity units, respectively. This is similar to the 4 units of intensity increase observed in model complexes in going from 6C to 5C, which distorts the ligand field of the FeII and increases 4p mixing.
EXAFS data and the corresponding Fourier transform (FT) are shown in Figure 2. The intensity of the EXAFS beat pattern of TH-[ ] (red) and TH-[6MPH4] (orange) are very similar with respect to the wave minima and maxima contributing to a first shell FT with the same intensity and position. When only substrate is present, an increase in the EXAFS signal is observed (TH-[L-tyr], Figure 2, blue). This increase in EXAFS amplitude is represented in the FT with an increase of about 20% (from 0.98 to 1.16 units) relative to TH-[ ] and TH-[6MPH4]. Upon binding of both substrate and cofactor, TH-[L-tyr, 6MPH4], shows both a decrease in the EXAFS amplitude as well as a slight shift to higher k. There are also subtle differences in the beat pattern including an additional feature at k = 8.5 Å−1. The EXAFS amplitude reduction, and shift in k are also reflected in the FT as the first shell signal decreases from 0.98 in resting to 0.61, and shifts toward shorter bond distances.
EXAFS data for all samples were collected to k = 15 Å−1; however, depending on data quality, fits were performed to k = 14 Å−1, k = 13 Å−1, k = 14 Å−1, and k = 12 Å−1 for TH-[ ], TH-[L-tyr], TH-[6MPH4] and TH-[L-tyr, 6MPH4], respectively. All spectra were fit to both 5C and 6C models, which were built based on the ferric crystal structure of tyrosine hydroxylase (PDB 1TOH 49). Since a bidentate Glu has been observed in PAH-[L-phe, pterin],16 TH-[L-tyr, 6MPH4] was also fit to a 5C model with a bidentate Glu. Both 6 and 5C EXAFS models and associated fitting paths are shown in Figure 3. Values for the most relevant EXAFS fits are presented in Table 2, with additional fits given in the Supporting Information.
The EXAFS spectrum for TH-[ ] is fit equally well using a 6C model with 4 or 5 paths in a 6 or 5 + 1 first shell distribution (fits 1 and 3 in Table 2). The pre-edge data for TH-[ ] (shown above) also indicate that the iron site is 6C, and this finding is supported by MCD data (vide infra). The outer shell was successfully fit using the remaining carbon single scattering at 3.12 Å and 4.08 Å and multiple scattering at 4.39 Å from the two histidine (His) rings.
Fits to the EXAFS data when only cofactor is bound (TH-[6MPH4]) are very similar to those of TH-[ ]. The best fit to the data is with a 6 coordinate split first shell, in a 1:5 ratio (fit 9) with the shorter bond distance at ~ 2.1 Å and longer bond distances at ~ 2.2 Å. The outer shell is again successfully fit using only the single and multiple scattering from the His rings. When only substrate is bound, the short bond distance which is seen in TH-[ ] and TH-[6MPH4] is no longer observed in the fit of the EXAFS signal of TH-[L-tyr] (fit 6). However, despite the lower error value which results with the addition of a split first shell, a path difference of 0.02 Å is beyond the resolution with data to k =13 Å−1. Thus, the best fit is 4 in which all ligands are incorporated into one 6C shell at 2.16 Å. A symmetric first shell with no resolvable short bond length is consistent with an increase in EXAFS amplitude, as the symmetric shell would provide for a stronger, more uniform signal. This increase is similar to the behavior observed for PAH when the substrate L-phe is present.13 For all three samples, when the data are fit using only 5 backscatters in the first shell rather than 6, a decrease in σ2 value is observed and the F factor increases, indicating a worse fit to the data, wit the smallest effect observed for TH-[ ]. This result is similar to previously reported results for TH-[ ] and TH-[BH4], where both forms of the enzyme were observed to contain a 6C first shell.21 Thus, TH-[ ], TH-[6MPH4] and TH-[L-tyr] forms of tyrosine hydroxylase all contain a distorted 6C site.
For TH-[L-tyr, 6MPH4], a fit using 6 ligands in the first shell results in a σ2 value which is unreasonably large for bond lengths averaging 2.13 Å (Supporting Information Table S2, fit 10) indicating that the data would be better fit with a 5C first shell. Changing the coordination from 6 to 5 also results in a better overall goodness of fit. A split first shell in a 1:4 ratio with bond lengths of 2.00 and 2.15 Å gives a reasonably good fit (Figure 4 and Table 2, fit 12). However, even with the 1:4 split shell, and the outer shell fit in a similar fashion to those for the other tyrosine hydroxylase samples (2 SS and 1 MS carbon paths at ~ 3.1, 4.1, and 4.4 Å respectively; Table 2, fit 12), at least one wave is still present in the residual signal from the EXAFS fit (Figure 4). An additional single scattering wave at 2.53 Å and an accompanying multiple-scattering wave at 2.65 Å were needed to complete the fit, which now has a 2:3 split in the first shell (Figure 4 and Table 2, fit 15). The residual wave is available if the Glu has shifted into a bidentate coordination mode (Figure 4). None of the other TH complexes required an additional signal at ~ 2.5 Å to complete the fit. To test if a His ligand was lost to form the 5C complex, the outer shell signals from the His were halved. This led to unreasonably low σ2 values, indicating that the data do not support His dissociation (Supporting Information Table S2, fits 7 and 8). The data indicate that the ligands which were lost are water, similar to the behavior observed in PAH.15 The k-space data and best fit for all complexes can be found in Supporting Information Figure S3.
The high spin Fe2+ (d6) free ion has a 5D ground state, which splits in an octahedral (Oh) ligand field into a triply degenerate 5T2g ground state and a doubly degenerate 5Eg excited state, which are separated by 10 Dq (~ 10,000 cm−1 for O and N ligands). The transitions are very weak in absorption, but are intense at low temperature in MCD. The Fe sites in proteins have lower symmetry which removes the orbital degeneracy of these states. Hence LF transitions to the two components of the 5Eg excited state are observed. For a distorted 6C site, two transitions centered around 10 000 cm−1 split by ~ 2000 cm−1 are observed. In a 5C distorted square pyramidal site, the excited state splitting (Δ5E) increases, and transitions are observed at approximately 10 000 cm−1 and 5000 cm−1. Transitions in a 5C trigonal bipyramidal site are shifted to lower energy, and are typically observed at < 10 000 cm−1 and < 5000 cm−1. Due to lower values of 10 Dq, LF transitions of 4C distorted tetrahedral ferrous sites are observed at low energy, usually between 4000 and 7000 cm−1. Thus, near-IR MCD spectra probe the 5E excited state and provide information on the geometric and electronic structure of the ferrous active site.55
Two ligand field (d-d) transitions are observed in the MCD spectrum of resting TH-[ ], shown in Figure 5A, that can be resolved into two Gaussian bands centered at 8720 cm−1 and 10 750 cm−1. The pattern of two bands centered around 10 000 cm−1 split by 2030 cm−1 is typical of 6C distorted octahedral ferrous sites. The MCD spectrum of TH-[L-tyr], shown in Figure 5B, appears to be slightly narrower than that of resting TH. The TH-[L-tyr] spectrum is best fit with two Gaussian bandshapes at 9390 cm−1 and 10 950 cm−1 with the higher energy band less intense. Figure 5C shows the MCD spectrum for TH-[6MPH4].56 This spectrum is also narrower than the resting TH spectrum. However, the spectrum cannot be fit with only one band, and is best fit with one band at 9250 cm−1 and another less intense band at approximately 10 800 cm−1 (the position of this band is dependent on the intensity distribution of the two Gaussians used to fit the spectrum). Thus, the MCD spectra of TH-[L-tyr] and TH-[6MPH4] are consistent with both species having distorted 6C geometries with slightly stronger ligand fields (10 Dq ~ 400 cm−1 higher in energy) and less axial distortion (smaller Δ5E) compared to TH-[ ]. These results are consistent with the XAS results (vide supra) which show that resting, substrate, or cofactor bound forms of TH are 6C.
In contrast, TH-[L-tyr,6MPH4] shows only one band at 8960 cm−1 (Figure 5D),57 suggesting that the second band of the 5E has shifted to below the cutoff of the MCD instrument, and the FeII site has gone 5C when both substrate and cofactor are present. This is supported by the variable-temperature-variable-field (VTVH) MCD results in the next section and is consistent with the XAS results (vide supra). The < 10 000 cm−1 energy of the band would support a trigonal bipyramidal site; however, analysis of VTVH MCD data gives a +ZFS system (vide infra) which indicates that the site is closer to square pyramidal.55,58,59 The shift of the transition to less than 10 000 cm−1 suggests a strong axial perturbation to the iron site which would decrease the equatorial ligand field.
While MCD spectra probe the FeII 5E exited state, the splittings of the 5T2 ground state can be obtained by VTVH MCD. The field and temperature dependence of the normalized MCD intensity of the TH complexes are plotted against βH/2kT (where β is the Bohr magnetron, H is magnetic field in Tesla, k is Boltzmann’s constant, and T is temperature in Kelvin) in Figure 5E-H. The resulting isotherms are nested (non-superimposed) and this nesting behavior can be used to gain insight into the ground state electronic structure of the FeII sites.28,58,60 The three-fold orbitally degenerate 5T2g ground state will split in energy in the low symmetry protein environment. This splitting can be defined in terms of an axial splitting of the dxy from the dxz,yz orbitals by an amount Δ, and rhombic splitting of the dxz and dyz orbitals by an amount |V|. The resultant orbitally non-degenerate ground state is still five-fold degenerate in spin (S = 2), and undergoes a zero-field splitting (ZFS), due to spin-orbit coupling of the low symmetry split components of the 5T2 state, to yield MS = ± 2, ± 1 and 0 sublevels split by 3D and D, where D is the axial ZFS parameter. Systems with weak axial ligand fields have negative ZFS (D < 0) and the MS ± 2 doublet is the lowest energy (Figure 6, top right). This will further split by an amount δ due to the rhombic ZFS splitting of the 5T2. When a magnetic field is applied, the splitting of the MS = ± 2 non-Kramers doublet increases due to the Zeeman effect (g||βH) and shows non-linear dependence on magnetic field due to a change of its wavefunction (Figure 6, bottom right). Simulations show that magnitude of δ is directly proportional to the degree of nesting of MCD isotherms.61 For positive ZFS systems (D > 0, Figure 6, top left) which corresponds to strong a axial LF perturbation of the 5T2, the MS = 0 sublevel is lowest in energy. When a perpendicular magnetic field is applied, it interacts with one component of the MS = ± 1, which is D cm−1 above it in energy. The resulting pseudo-doublet will Zeeman split and behave similarly to the MS ± 2 doublet in the −ZFS case to yield similar non-linear field dependence (Figure 6, bottom left). In general, + ZFS systems, which are split by the large axial ZFS (D), have much larger nesting behavior than −ZFS systems.55,61
The spin-Hamiltonian parameters can be obtained from VTVH MCD data by computationally fitting the field and temperature dependence of the normalized MCD intensity, taking into consideration linear B-terms, z-polarization, and thermal population of states (Equation 1 in Ref. 55). Once δ and g// (and D and g// for +ZFS) are known, they can be used 58 to obtain and |V|, the axial and rhombic splitting of the t2 set of d orbitals. All the TH VTVH MCD data were fit with both − ZFS and + ZFS models, which can be distinguished from each other by comparing the spin-Hamiltonian values obtained. In particular, +ZFS systems will give non-physical, low g// values (vs g// = 8.8 – 10 cm−1 for − ZFS 6C), and a low-lying excited state singlet when fit to a − ZFS model.58
From Figure 5E, the saturation data of the TH-[ ] bands exhibit a very small nesting and can be fit with a − ZFS model giving spin-Hamiltonian parameters of δ = 3.1 cm−1 and g// = 9.1 cm−1. These yield 5T2 ground state parameters of Δ = −200 cm−1 and |V| = 60 cm−1. The doublet ground state and small splitting of the t2g d orbitals are typical of a 6C FeII site with moderate rhombic distortion (|V/2Δ| ~ 0.15). The splitting of the 5T2 ground state is combined with the excited state splitting (Δ5E = 2030 cm−1, from the MCD spectrum) to give the experimental d-orbital energy level diagram for TH-[ ] (Figure 7, left).
The VTVH MCD data for TH-[L-tyr] (Figure 5F) show increased nesting compared to resting TH, but are still described as a − ZFS system with δ = 5.4 cm−1, g// = 9.0 cm−1 to yield Δ= −250 cm−1 and |V| = 130 cm−1. The large δ and relatively small Δ are consistent with an FeII site with a higher degree of rhombicity (|V/2Δ| = 0.26), and thus a larger splitting of the t2 orbitals is observed. The excited state data presented earlier shows that tyrosine binding shifts the e orbitals to higher energy (10 Dq = 10,170 cm−1), and decreases their splitting to 1560 cm−1. Overall, the experimental orbital energy levels for TH-[L-tyr] (Figure 7, second panel) are consistent with a distorted 6C iron site.
The VTVH data of TH-[6MPH4] also show more nesting than that of resting TH and fit equally well to both − ZFS and + ZFS models (Information Figure S5). A larger total splitting of the orbitals of the ground state is observed when pterin binds to the enzyme regardless of the sign of the ZFS (Figure 7, third panel). However, from the XAS data (vide supra) that TH-[6MPH4] is 6C, which is consistent with a −ZFS. The parameters for the doublet ground state (− ZFS) in TH-[6MPH4] are δ = 3.7 cm−1, g// = 9.2 cm−1, Δ= −500 and |V| = 220 cm−1. The orbitals of the 5E excited state of TH-[6MPH4] are also shifted to higher energy, suggesting an overall stronger 6C LF at the FeII center relative to resting.
The saturation behavior of TH-[L-tyr, 6MPH4] is dramatically different from TH-[ ]. The saturation curves are significantly more nested, and must be fit with a + ZFS system. The ground state of TH-[L-tyr, 6MPH4] is now MS = 0 and gives D = 11.2 cm−1, |E| = 2.6 cm−1, Δ = 700 cm−1 and |V| = 420 cm−1. The large t2 orbital splittings and + ZFS for TH-[L-tyr, 6MPH4] are consistent with a 5C site that is closer to square pyramidal in geometry. The ground state analysis yields |V/2Δ| = 0.3, indicating a high degree of rhombic distortion. In addition, the excited state transitions for TH-[L-tyr, 6MPH4] occur at 8960 cm−1 and < 5000 cm−1, a range consistent with a 5C site with a strong axial distortion. Although there is no crystal structure of this ternary complex of TH, the MCD data of PAH are very similar and examination of the structure of PAH with cofactor and substrate analog bound (PDB ID 1MMT 16) shows a 5C FeII site that has one short axial bond and rhombically distorted equatorial ligand field (Figure 8). Excited and ground state parameters obtained from MCD spectra and VTVH analyses of the four TH complexes are summarized in Table 3.
The S395A-[ ] spectrum, shown in Figure 9A (brown) has a shape very similar to wild-type TH-[ ], but is almost two-fold more intense. The spectrum is fit well with two Gaussian bands at 8850 cm−1 and 10,870 cm−1, and the saturation behavior of the LF bands shows small nesting (Figure 9B) and fits to a −ZFS system. Together, these indicate that the iron site in S395A-[ ] is distorted 6C. As in wild-type TH, the substrate + cofactor bound form of S395A has only one resolvable transition (9330 cm−1) in its MCD spectrum (Figure 9C tan). Also, as with the WT enzyme, the VTVH data of S395A-[L-tyr, 6MPH4] (Figure 9D) show a large increase in nesting and yield a + ZFS system, showing that the active site has become 5C square pyramidal.
The MCD spectrum of E332A-[ ] (Figure 10A, purple) shows two bands centered around 10 000 cm−1, but looks qualitatively quite different from that of wild-type TH. The intensity ratios of the 2 bands are reversed — the higher energy band now has more intensity than the lower energy band. The Gaussian-resolved band positions are 8880 cm−1 and 10 610 cm−1 and give Δ5E of 1830 cm−1. The smaller excited state splitting suggests less distortion of this 6C site than in resting wild-type TH. The saturation data (Figure 10B) of this mutant show small nesting and are fit to a − ZFS system, confirming that the ferrous site is 6C. When both substrate and cofactor are bound to E332A, the MCD spectrum changes to show only one band (9150 cm−1 Figure 10C, light blue). The VTVH data of this band (Figure 10D) again exhibit very large nesting and yield a + ZFS system, confirming that the iron site in E332A-[L-tyr, 6MPH4] is 5C distorted square pyramidal.
The ligand field, spin-Hamiltonian and ground state parameters for TH S395A and E332A are listed in Tables S3 and S4 in the Supporting Information.
Stopped-flow single-turnover kinetic experiments were performed on WT and E332A TH to further probe the mechanism of the reaction of TH with pterin and O2. The oxidation of FeIITH to FeIIITH results in a broad increase in the absorbance spectrum between 200 and 450 nm (Figure 11A), corresponding to a ligand to FeIII charge transfer transitions.37 In addition, the reduced and oxidized pterin species have distinctive absorbance spectra over the same wavelength region (Figure 11B); these can be used to identify pterin-based intermediates formed during the reaction. In the rapid kinetic experiments, the oxidation of FeII was followed as absorbance increases at 246 and 350 nm (Figure 11A, black to orange). The oxidation of 6MPH4 to quinonoid 6MPH2 (q-6MPH2) is uniquely characterized by increased absorbance at 350 – 450 nm (solid red to dashed green in Figure 11B). The formation of 4a-HO-6MPH3 from 6MPH4 was detected by absorbance changes at two diagnostic wavelengths: a maximal initial absorbance increase at 246 nm and an accompanying decrease at 300 – 340 nm (solid red to dotted blue in Figure 11B). The slow dehydration of 4a-HO-6MPH3 to q-6MPH2 is accompanied by a decrease at 246 nm and an increase at 300 – 450 nm (blue to green in Figure 11B).
Representative stopped-flow absorbance traces from the reactions of different TH complexes (~100 μM final concentration) with varied concentrations of O2 (95 μM to 950 μM) at 5 °C are shown in Figures 12 and and13,13, and S7–S10. All the reactions could be successfully simulated using one of the mechanisms in Scheme 3. The rate constants used in the simulations are summarized in Table 4. Addition of O2 to resting FeIITH-[ ] results in monophasic absorbance increases at 246 nm and 350 nm (Figure 12, dashed red trace) with a single step second-order mechanism with a rate constant of 0.037 mM−1s−1 (Scheme 3A, Table 4 and Figure S7), in agreement with previous studies at 20 °C.37 Similarly, the oxidation of 6MPH4 to q-6MPH2 in the presence of apo-TH is well-described as a single second-order reaction with a rate constant of 1.7 × 10−4 mM−1s−1 (Table 4), comparable to the rate of tetrahydropterin auto-oxidation in solution under the same conditions (1.3 × 10−4 mM−1s−1, results not shown).
The absorbance changes accompanying the reaction with oxygen of TH complexes containing substrates or analogs are more complex. All are better described by an initial reversible oxygen binding step to form Enz:O2 complexes followed by chemical transformations (Schemes 3B & C and Figures S8 – S10). The presence of tyrosine or 6MPH2 has little effect on the kinetics of FeII oxidation (Figure 12). The FeIITH-[5-deaza-6MPH4] complex is an exception; in this case, the rate constants are the same as for FeIITH-[6MPH2], but the magnitudes of the absorbance changes are much smaller and are consistent with the amount of the free enzyme in these samples (Figure 12, grey dotted and black solid lines, respectively), suggesting that binding of 5-deaza-6MPH4 to TH significantly decreases the reactivity with oxygen.
The rate constant for FeII oxidation in the FeIITH-[6MPH4] complex is about 5-fold faster than that for the other binary complexes containing the redox-inactive pterins. This suggests that a change in the oxidation state of the pterin accompanies oxidation of the FeII. Scheme 3B, in which the reaction of O2 with FeIITH-[6MPH4] results in concomitant oxidation of FeII to FeIII and 6MPH4 to 6MPH3• in the first step, followed by disproportionation of 6MPH3• to give q-6MPH2 and 6MPH4•, accounts well for the absorbance traces at 246 nm and 350 nm (Figures 12 and S9). The rate constant for 6MPH4 oxidation in the FeIITH-[6MPH4] complex shows a three order of magnitude acceleration over the auto-oxidation reaction in apo-TH-[6MPH4] (0.16 s−1 vs 1.7 × 10−4 s−1, respectively). The mechanism of Scheme 3B is consistent with the redox potentials of the individual species. Using an E0 of + 210 mV for the FeII in TH,62 the 1-electron oxidation of Fe2+ + O2 → Fe3+ + O2− • (O2 + e− → O2− •, E0 = −330 mV)63 has E0 = −537 mV and is uphill by 12.5 kcal/mol.. Similarly, the reaction of 6MPH4 to 6MPH3•H+ (Eo = −270 mV)64 with O2 has E0 = −600 mV and is energetically uphill by 14 kcal/mol. However, the two electron reaction between Fe2+, O2 and 6MPH4, (O2 + 2e− → H2O2, E0 = + 280 mV)63 has E0 = + 83 mV, and is downhill at − 1.9 kcal/mol. Taken together, all these results indicate that the concerted 2-electron transfer to O2 from 6MPH4 and FeII (equation 1) is more favorable than the individual 1-electron oxidation processes.
When O2 is mixed with FeIITH-[L-tyr,6MPH4] to form a catalytically competent complex, a reversible oxygen complex is initially formed with no observable absorbance change. This is followed by biphasic absorbance changes at 246 nm, 318 nm and 330 nm over approximately 4 s (Figure 13, dashed green trace). This rapid initial first phase corresponds to the formation of 4a-OH-6MPH3 with a rate constant of 24 s−1 (Scheme 3C and Figure S10). The 4a-OH-6MPH3 is formed in the same step as the FeIV=O intermediate (Schemes 2 and and3C).3C). Rapid-quench Mössbauer spectroscopy has previously shown that the FeIV=O intermediate is also formed with a rate constant of 24 s−1 under the same conditions used here.26 The slower second phase was successfully modeled as the release of 4a-OH-6MPH3 from the enzyme active site into the solution with a rate constant of 0.42 s−1 (Figure S10), a value which matches kcat under these conditions. The 4a-OH-6MPH3 subsequently dehydrates in solution to produce q-6MPH2, resulting in a gradual decrease at 246 nm and increases at 318 nm and 330 nm, with a rate constant of ~0.02 s−1. The reactions of O2 with FeIITH-[L-tyr,5-deaza-6MPH4] and FeIITH-[L-tyr,6MPH2], which presumably have similar 5C sites,65 are similar to FeII oxidation in FeIITH with either 6MPH2 or tyrosine bound (Figure 13A and Table 4). This indicates that both a 5C FeII and a redox active pterin are needed for the two (and greater) orders of magnitude rate enhancement in the reactivity with O2 seen with FeIITH-[L-tyr,6MPH4] compared to FeIITH-[ ], FeIITH-[L-tyr], and FeIITH-[6MPH4].
The absorbance changes upon addition of O2 to E332A-FeIITH-[L-tyr,6MPH4] are shown in Figure 13 (solid blue traces). The enzyme clearly catalyzes a reaction between FeII, O2 and pterin, but the absorbance changes are distinct from those seen with WT FeIITH-[L-tyr,6MPH4]. In the case of the E332A mutant, the observed absorbance changes can be accounted for by initial reversible oxygen addition followed by formation of an intermediate in a single slow step with a rate constant of 0.3 s−1; this value matches kcat for this mutant enzyme. The intermediate decays nonenzymatically with almost the same rate constant (~0.02 s−1) as the dehydration of 4a-HO-6MPH3 to q-6MPH2. The reaction catalyzed by the E332A mutant is clearly slower than that of the wild-type enzyme. The rate constant for the second step is similar to the rate constant for release of the 4a-HO-6MPH3 from the wild-type enzyme. This raises the possibility that the observed rate constant is a net rate constant which includes formation of the intermediate and its release from the enzyme; however, these two rate constants are too similar to resolve. Compared to WT TH, the magnitudes of the absorbance changes at 246 nm and 330 nm and the sign of the absorbance change at 318 nm after O2 binding are different for E332A TH (Figure 13, solid blue traces). The spectrum of the intermediate could be obtained by global analysis of the absorbance changes between 240 and 450 nm, fitting the data to a three step kinetic model (Figure S11). The starting spectrum and those of the final two species match those for 6MPH4, q-6MPH2, and 7,8-6MPH2. The spectrum of the unknown intermediate exhibits a number of differences from that of 4a-HO-6MPH3, including a much lower absorbance at 246 nm and a red-shifted absorbance maximum. These differences establish that, in contrast to WT TH, a 4a-HO-pterin is not formed in the reaction of E332A-FeIITH [L-tyr, 6MPH4] with O2. However, the decay of the intermediate in the E332A TH reaction to q-6MPH2 with a rate constant nearly identical to that for 4a-HO-6MPH3 suggests that this intermediate is similar to it. Spectral differences similar to those between this intermediate and that of 4a-HO-6MPH3 have been observed previously for 4a-HO- and 4a-HOO-flavins, intermediates in the reaction of the flavoprotein phenol hydroxylases.66,67 Thus, a possible candidate for this intermediate species in the E332A TH reaction is a hydroperoxy-pterin, HOO-6MPH3.67
The rapid single-turnover kinetics experiments presented above allow the dissection of individual steps of TH reaction, and show that the presence of a 5C FeII site and a redox-active pterin greatly accelerates the enzyme reaction with O2. The data on the WT vs E332A reactions also provide insights into the nature of the intermediates formed in the coupled vs uncoupled reaction.
In this study, we present new spectroscopic insight and information on the catalytically relevant ferrous active site of TH and its interaction with tyrosine and 6-methyltetrahydropterin (6MPH4). XAS and MCD studies show that the FeII active site is 6C in resting TH. The geometric and electronic structure of the FeII site is slightly perturbed when either tyrosine or pterin is bound to TH, but the site remains 6C in both cases. The active site goes 5C when both tyrosine and pterin are bound, opening up a coordination position for the O2 reaction. The EXAFS data for TH-[L-tyr,6MPH4] are also consistent with a 5C FeII site with 2 His, 1 bidentate Glu and 1 water as ligands, comparable to the crystal structure determined for FeII-PAH with thienylalanine and BH4 bound (PDB 1KWO17). The 6C → 5C change in the TH active site is consistent with the general mechanistic strategy observed in other non-heme iron enzymes that use redox-active cofactors to activate O2 for reaction with substrates.28,60,68
Earlier steady-state kinetic studies on WT TH were consistent with an ordered binding sequence of pterin followed by oxygen in rapid equilibrium and then tyrosine before the DOPA product is released.18 This result suggested that the FeII site in TH could undergo different structural changes upon pterin binding which allow the FeII, pterin and O2 to react before tyrosine binds. The protein itself does undergo conformational changes upon pterin binding.69 However, our spectroscopic data show only a slight perturbation to the 6C FeII site upon pterin binding. While steady-state experiments give information on the overall reaction, they do not provide information on individual mechanistic steps.70 In addition, these experiments were complicated by the use of a large excess of 6MPH4, required to ensure steady-state conditions, because the excess 6MPH4 also reduces the Fe. Here, we have used single-turnover kinetics experiments with tyrosine and 6MPH4 (added in stoichiometric amounts to minimize side reactions) to evaluate the individual steps and identify short-lived intermediates in the TH reaction.
The single turnover kinetic results indicate that the isolated one-electron oxidation reactions of FeII to FeIII in FeIITH-[ ] and tetrahydropterin to quinonoid dihydropterin (q-6MPH2) in apoTH-[6MPH4] are slow due to unfavorable thermodynamics. However, the combination of FeII and redox-active pterin in FeIITH-[6MPH4] accelerates both the FeII and uncoupled pterin oxidation reactions, even though the FeII site remains 6C. The faster reaction rates in FeIITH-[6MPH4] can be ascribed to the favorable thermodynamics of the two-electron reduction of O2 to peroxide by the combined reaction of FeII → FeIII and 6MPH4 → 6MPH3•. When both pterin and tyrosine are bound to give a 5C FeIITH site, the first observable step in the O2 reaction, corresponding to the breakdown of the putative FeII-OO-pterin intermediate, is 2 – 3 orders of magnitude faster than the FeII, 6MPH4 and FeII + 6MPH4 oxidation reactions in the 6C and non-catalytic 5C forms of TH (Table 4). The measured rate constant for this step necessarily sets a lower limit for the rate constant for the initial reaction with oxygen to form this intermediate. Thus the two electron transfer from the 5C FeII and reduced pterin to O2 is far more favorable than in the 6C FeIITH-[6MPH4] site and than in the 5C FeIITH-[L-tyr,5-deaza-6MPH4] and FeIITH-[L-tyr,6MPH2] sites, confirming that the open coordination position on the FeII and the presence of a redox active pterin are both critical for catalytic TH reactivity.
In an earlier study on PAH mutants related to phenylketonuria (PKU), structural differences in the active site were observed that suggested that the pterin will react with O2 to make peroxypterin which decays in the uncoupled reaction to give off H2O2.32 Here, we are able to observe individual reaction steps in the coupled WT and uncoupled E332A reactions when both tyrosine and 6MPH4 are bound. The first step with observable kinetics in WT TH is the decay of this putative intermediate, which occurs at a rate of 24 s−1. The 4a-OH-pterin that forms in WT TH (and in the S395A mutant) is not observed in the E332A mutant, which also does not hydroxylate tyrosine, indicating that the FeIV=O intermediate also does not form. The first observable step of the E332A TH reaction is the formation of an oxidized pterin species, possibly the HOO-pterin, with a rate constant of 0.3 s−1. This intermediate may form either by direct reaction of O2 with 6MPH4 (however this reaction is much slower at 1.7 × 10−4 mM−1s−1), or, more likely, through a FeII-OO-pterin intermediate as in WT TH. In the latter case, the E332A mutation may change the reactivity of the FeII-OO-pterin intermediate such that it decomposes with a slower rate and through a different mechanism than WT. In both, decay of the intermediate likely involves protonation of the FeII-OO-pterin. In WT TH, the nature of the cleavage products, FeIV=O and 4a-OH-pterin indicate that the O distal (Odis) to the FeII is protonated to heterolytically cleave (Scheme 4, top). In the E332A mutant, an oxidized pterin species, possibly the HOO-pterin, is formed. This would involve the protonation of the O proximal to Fe (Oprox) in the FeII-OO-pterin, followed by the cleavage of the Fe-Oprox bond (Scheme 4, bottom). These results suggest that the E332 residue participates in the reactivity of TH by directing the distal protonation of the FeII-OO-pterin intermediate.
The crystal structure of FeII-PAH with thienylalanine and tetrahydrobiopterin bound (PDB 1KW0, Figure 14)17 indicates that the E332 carboxylate group is oriented towards the pterin, and could hydrogen bond to it, either through direct interaction with the proton on the N3, or through solvent molecules. The E332 residue could either directly transfer the proton from N3 to Odis (N3 is deprotonated in the final quinonoid pterin product), or direct a solvent water via an H-bond to the Odis. The water ligand is not likely to be the proton source for heterolytic O-O cleavage, because the MCD data of the TH-[L-tyr, 6MPH4] complexes in WT and E332A indicate that the geometric and electronic structures of the 5C sites are very similar (Figure 10C). Apparently, when E332 is mutated to an alanine, the enzyme can no longer direct a proton to Odis, and the Oprox is protonated instead, leading to the cleavage of the Fe-O bond to regenerate FeII and HOO-pterin.
In summary, we have shown through XAS and MCD spectroscopy that the FeII active site in TH is 6C until both tyrosine and pterin are bound. This 6C → 5C conversion is consistent with the general mechanistic strategy observed in other mononuclear non-heme iron enzymes. Single turnover kinetic data on WT TH indicate that at a 6C active site, the two electron reduction of O2 to peroxide by FeII and pterin is favored over individual one electron reactions. However, this reaction is still slow. Importantly, this reaction is greatly accelerated in the FeIITH-[L-tyr, 6MPH4] complex, indicating that both 5C FeII and pterin are required for the O2 reaction. Comparison of the kinetics of WT and E332A TH-[L-tyr, 6MPH4] reactions with O2 indicate that the E332 residue plays a role in directing the protonation of the bridged FeII-OO-pterin complex to produce the FeIV=O hydroxylating intermediate and hence is important for tuning the TH active site for productive coupled turnover. Application of the spectroscopic and mechanistic analyses described in this study to variants of TH related to neurological disorders would provide molecular-level insight into enzyme dysfunction and define potential therapeutic approaches.
Financial support was provided by the National Institutes of Health GM 40392 to E.I.S., GM 47291 to P.F.F and RR-01209 to K.O.H, and The Welch Foundation (A-1245) to PFF. SSRL operations are funded by the U.S. Department of Energy (Office of Basic Energy Science). The Structural Molecular Biology program at SSRL is funded by the National Institutes of Health, National Center for Research Resources, Biomedical Technology Program and the Department of Energy, Office of Biological and Environmental Research. This publication was made possible in part by Grant Number 5 P41 RR001209 from the National Center for Research Resources (NCRR), a component of the National Institutes of Health (NIH). Its contents are solely the responsibility of the authors and do not necessarily represent the official view of NCRR or NIH.
6 Supporting Information Available Details of EXAFS fits, the MCD spectrum and VTVH data of WT FeIITH-[5-deaza-6MPH4], ligand field and spin Hamiltonian parameters for E332A and S395A, kinetic traces at different wavelengths for single-turnover experiments and proposed mechanisms based on the kinetics. This material is available free of charge via the internet at http://pubs.acs.org.