Tyrosyl Radicals in Dehaloperoxidase

Background: The catalytically active species in dehaloperoxidase (DHP) contains both a ferryl heme and a tyrosyl radical. Results: Radicals were shown to form on three tyrosines (Tyr-28, Tyr-34 and Tyr-38) in DHP. Conclusion: Mutants that lacked tyrosines showed increases in the rates of both substrate oxidation and heme bleaching. Significance: Tyrosyl radical formation is an evolutionary adaptation to protect the enzyme from irreversibly oxidizing itself. Dehaloperoxidase (DHP) from Amphitrite ornata, having been shown to catalyze the hydrogen peroxide-dependent oxidation of trihalophenols to dihaloquinones, is the first oxygen binding globin that possesses a biologically relevant peroxidase activity. The catalytically competent species in DHP appears to be Compound ES, a reactive intermediate that contains both a ferryl heme and a tyrosyl radical. By simulating the EPR spectra of DHP activated by H2O2, Thompson et al. (Thompson, M. K., Franzen, S., Ghiladi, R. A., Reeder, B. J., and Svistunenko, D. A. (2010) J. Am. Chem. Soc. 132, 17501–17510) proposed that two different radicals, depending on the pH, are formed, one located on either Tyr-34 or Tyr-28 and the other on Tyr-38. To provide additional support for these simulation-based assignments and to deduce the role(s) that tyrosyl radicals play in DHP, stopped-flow UV-visible and rapid-freeze-quench EPR spectroscopic methods were employed to study radical formation in DHP when three tyrosine residues, Tyr-28, Tyr-34, and Tyr-38, were replaced either individually or in combination with phenylalanines. The results indicate that radicals form on all three tyrosines in DHP. Evidence for the formation of DHP Compound I in several tyrosine mutants was obtained. Variants that formed Compound I showed an increase in the catalytic rate for substrate oxidation but also an increase in heme bleaching, suggesting that the tyrosines are necessary for protecting the enzyme from oxidizing itself. This protective role of tyrosines is likely an evolutionary adaptation allowing DHP to avoid self-inflicted damage in the oxidative environment.

As a globin, DHP has a characteristic fold composed of the canonical 3/3 ␣-helical structure (11,(13)(14)(15) and binds oxygen reversibly when the heme iron is in the ferrous (Fe 2ϩ ) oxidation state (8 -10). Yet DHP is itself distinct from other globins; it has little sequence homology with other hemoglobins, possesses a crystallographically characterized internal substrate binding site for 2,4,6-trihalophenols (16,17), and has been shown to utilize hydrogen peroxide to oxidize a range of substrates such as mono-, di-, and trisubstituted halophenols with bromine, chlorine, or fluorine as substituents (18,19). Moreover, the DHP peroxidase cycle can be initiated from both the ferric and ferrous (or oxyferrous) oxidation states (20,21). Thus, as a bifunctional globin-peroxidase capable of functioning from the reduced form, DHP may represent a new emerging class of peroxidases that shares neither sequence nor structural homology with any of the known peroxidases.
With the help of the Tyrosyl Radical Spectra Simulation Algorithm (TRSSA) (40,41), the free radical EPR signal reported for DHP A Compound ES was shown to be a pH-dependent superposition of two different free radical EPR signals assigned to different tyrosine residues, Tyr-34 and Tyr-38. We hypothesized at that time the following (40); (i) the dominant radical associated with the peroxidase function of Compound ES is located on Tyr-34, (ii) in the absence of substrate and when His-55 is in the open conformation (i.e. rotated away from the heme center), the Tyr-34 radical combines with a porphyrin radical formed upon ferryl heme autoreduction, which leads to formation of Compound RH, (iii) the Tyr-38 radical was interpreted as a result of the ferryl heme reduction when His-55 is in the closed conformation (i.e. rotated toward the heme), and (iv) Compound RH is a state that protects DHP against undesired peroxidase activity in the absence of substrate. Thus we proposed that the conformation of His-55, either open or closed, had an immediate effect on the evolution of Compound ES either to Compound RH or to a biradical state (40).
As the above hypotheses for the free radical chemistry of Compound ES were drawn from the TRSSA based radical site assignments, the focus of the present report was to obtain data on the DHP radicals using a more traditional mutagenesis approach. Of the five tyrosine residues in DHP A (Tyr16, Tyr-28, Tyr-34, Tyr-38, Tyr-107; Fig. 1), only three (Tyr-28, Tyr-34, and Tyr-38) are reasonably close (Ͻ10 Å) to the heme to reduce Compound I (the remaining two, Tyr-16 and Tyr-107, are at distances Ͼ15 Å away from the heme). As the DHP B isoenzyme bears an Asn at position 34 (42), only Tyr-28 and Tyr-38 were hypothesized as the likely site(s) for free radical formation (37). Thus, the experimental strategy employed herein focuses on altering radical formation in DHP via site-directed mutagenesis of three tyrosine residues, Tyr-28, Tyr-34, and Tyr-38, singly or in combinations, in both isoforms. We examined the reaction of the ferric state of the mutants with hydrogen peroxide using stopped-flow UV-visible and rapidfreeze-quench EPR spectroscopic methods. The data suggest a unique interplay between catalytic activity, protection from selfoxidation, and formation of Compound I, Compound ES, Compound RH, and tyrosyl radicals. Overall, the results reported herein help to understand what has been termed the dehaloperoxidase paradox (43), namely the phenomenon of a single heme protein being able to function both as an oxygen binding globin and as a peroxidase.

EXPERIMENTAL PROCEDURES
Materials and Methods-Buffer salts and acetonitrile (HPLC grade) were purchased from Fisher. All other reagents, unless otherwise specified, were from Sigma. EPR tubes were purchased from Norell (Landisville, NJ). Solutions of 2,4,6-trichlorophenol (TCP) were freshly prepared in 100 mM potassium phosphate (KP i ) buffers and kept at 4°C while protected from light. UV-visible spectra were recorded periodically to ensure that the substrate had not degraded. Hydrogen peroxide solutions were also freshly made before each experiment; initially, a 10 mM stock solution of H 2 O 2 was prepared and maintained at 4°C (typically for less than 15 min), during which time the protein/substrate solutions were loaded into the stopped-flow apparatus. The stock H 2 O 2 solution did not exhibit any measurable degradation over this time as judged by its absorbance at 240 nm (⑀ 240 ϭ 43.6 M Ϫ1 cm Ϫ1 ) (44). The stock H 2 O 2 solution was then diluted to the appropriate premixing concentration and immediately loaded into the stopped-flow apparatus.
The DHP B (Y38F) plasmid was further used as a template for generating the double mutant DHP B (Y28F/Y38F) with the mutagenic primers for pDHPB (Y28F). The plasmids were extracted using the QIAprep spin miniprep kit (Qiagen Sciences, Valencia, CA). DNA sequencing of the resulting mutated genes in their entirety confirmed the success of the site-directed mutagenesis and the absence of secondary mutations. Recombinant DHP proteins were obtained by expression in Escherichia coli as described elsewhere (22,24,37). The protein yield was greater than ϳ9 mg/liter culture in all cases. A twopart purification strategy (immobilized metal affinity chroma-tography followed by ion-exchange chromatography) resulted in a purification level of Ͼ95% homogeneity, with the DHP mutants being indistinguishable by SDS-PAGE gel from their WT counterparts.
Molecular Weight Determination-The molecular weights of the proteins employed in this study were experimentally determined via electrospray ionization mass spectrometry in the positive-ion mode (Agilent Technologies 6210 LC-TOF, Santa Clara, CA). The protein samples were prepared in 50 mM ammonium acetate buffer, pH 7.0. The mobile phase consisted of a linear gradient using the following HPLC grade solvents: water ϩ 0.1% formic acid (v/v) and water:acetonitrile (5:95) ϩ 0.1% formic acid (v/v). The injection volume was 5 l, and the flow rate was 300 l/min. The experimentally determined and calculated monomeric molecular weights of the dehaloperoxidase mutants as determined by electrospray ionization mass spectrometry were as follows: DHP A(Y34F), 16 Preparation of Ferric DHP-DHP as purified was treated with an excess of potassium ferricyanide to obtain a homogeneous solution of the enzyme in the ferric state. The ferricyanide was then removed using a PD-10 desalting column pre-packed with Sephadex G-25 medium. The protein was concentrated using an Amicon Ultra centrifugal filter equipped with a 10,000 kDa cutoff molecular mass membrane, and the purity of DHP was determined as previously published (22,37). The concentrations of the mutant enzymes were determined spectrophotometrically based on the extinction coefficients for WT DHP A (⑀ 406 ϭ 116.4 mM Ϫ1 cm Ϫ1 ) (19) and WT DHP B (⑀ 407 ϭ 117.6 mM Ϫ1 cm Ϫ1 ) (37).
UV-Visible Spectroscopy and DHP Activity Assays-The enzymatic activity of DHP was assayed optically from the rate of TCP conversion to dichloroquinone (2,6-dichloro-1,4-benzoquinone) at variable H 2 O 2 concentrations (10 -200 M, pH 7, 100 mM KP i ). Although the enzyme and the starting TCP concentrations were kept constant in all assays (0.5 and 150 M, respectively), the dependence of the rate on the H 2 O 2 concentration was fit with a classical Michaelis-Menten model using the Grafit 4.0 software package. The kinetics parameters K m and k cat resulted from the optimization of the fitting procedure. Optical spectra were recorded at 25°C using a Cary 50 UVvisible spectrophotometer equipped with thermostatted cell holders. The TCP concentration was measured by using the molar absorption coefficient ⑀ 312 ϭ 3752 M Ϫ1 cm Ϫ1 (22).
Stopped-flow Kinetics Studies-Rapid changes in the optical spectra of the DHP variants upon reaction with H 2 O 2 were studied by a Bio-Logic SFM-400 Triple Mixing Stopped-Flow instrument equipped with a diode array UV-visible spectrophotometer. The reactions were carried out at pH 7 in 100 mM KP i buffer at 20°C. The temperature was maintained using a circulating water bath. Experiments were performed in singlemixing mode wherein enzyme at a final concentration of 10 M was reacted with 2.5-25 eq of H 2 O 2 . Data were collected at variable intervals over three different time ranges using the Bio-Kinet32 software package (Bio-Logic). All data were evaluated using the Specfit Global Analysis System software package (Spectrum Software Associates). Singular value decomposition (SVD) analysis was applied to the raw spectra sets when different mechanistic hypotheses were explored and fit to exponential functions as either one-step, two species or two-step, three species irreversible mechanisms, where applicable. Kinetics data were baseline-corrected using the Specfit autozero function.
Rapid Freeze Quenching of the Reaction Intermediates-A BioLogic SFM 400 Freeze-Quench apparatus was used to prepare samples for EPR spectroscopic analysis of the short-lived reaction intermediates. A 50 M enzyme solution (final concentration) was reacted with a 10-fold excess of H 2 O 2 in 100 mM KP i (pH 7) at 25°C. Variable reaction times were achieved by means of different aging lines. To make a sample, a standard 4-mm outer diameter quartz EPR tube connected to a Teflon funnel was submersed in, and therefore filled with, isopentane cooled in a liquid nitrogen bath to Ϫ110°C. The reaction mixture was sprayed into the funnel, and the resulting frozen material was then packed at the bottom of the quartz tube using a packing rod fitted with a Teflon tip. The sample was then transferred to a liquid nitrogen storage Dewar for subsequent spectroscopic analysis.
EPR Spectroscopy-EPR spectra were recorded on an X-band (9 GHz) E-9 EPR spectrometer (Varian, El Palo, CA). A quartz finger Dewar insert was used to measure the EPR spectra at 77 K. This required periodic refilling of the Dewar with liquid nitrogen due to its evaporation during longer acquisition runs. The typical spectrometer settings were as follows: field sweep 200 G, scan rate 3.33 Gauss/s, modulation frequency 100 KHz, modulation amplitude 4.0 G, and microwave power 2 milliwatts. The microwave frequency for each EPR experiment, typically close to 9.2977 GHz, was measured by an EIP-578 in-line microwave frequency counter (PhaseMatrix, San Jose, CA).
EPR Spectra Simulation and Assignment to Specific Site-The free radical EPR spectrum simulation was performed by using SIMPOW6 (45). TRSSA (41) was used to generate the EPR spectrum simulation parameters from an input of two variables, the phenoxyl ring rotation angle, and the spin density on atom C1 of the radical. The Phenol Ring Rotation Angle Database (46) was used to identify the tyrosine residues in the DHP structures (as Tyr-28) with the closest rotational conformation of the phenol group (the ring rotation angle) to the conformation determined from the simulation of the tyrosyl radical EPR spectrum.

UV-Visible Spectra of the Mutants in the Resting (Fe 3ϩ Heme)
State-The optical spectra of the two WT DHP isoforms in comparison with their corresponding tyrosine mutants are shown in Fig. 2. The spectra of these heme proteins in the oxidized state are mainly typical of the high spin (with reference to the electron spin heme iron) (S ϭ 5/2) ferric heme forms; their slightly different optical characteristics are detailed in Table 1 as is their optical purity ratio (A Soret /A 280 , also termed the Reinheitzahl value, R z ) and the ratio of absorbances A Soret /A 380 for all proteins studied. The latter ratio has been employed before in estimating the relative proportion of the five-and six-coor-dinated high spin ferric heme populations; the greater the A Soret /A 380 ratio, the greater the six-coordinate to five-coordinate population ratio (47)(48)(49). Note that this ratio is smaller (indicating a greater relative proportion of the five-coordinate form) in all proteins that retain Tyr-38. Fig. 2 shows that whenever Tyr-38 is missing either in DHP A or DHP B, there is a characteristic deviation in the absorption spectrum, and the A Soret /A 380 ratio is greater (Table 1), indicating a greater proportion of the six-coordinate ferric heme form as compared with the five-coordinate. Thus, it appears that the proteins that retain Tyr-38 have a greater proportion of the five-coordinate ferric heme state, which is likely to be associated with the  His-55 "open" conformation. We propose that this indicates that Tyr-38 serves as the primary source of the proton for His-55 in regulating the balance of the open/closed conformations and, therefore, the route of free radical evolution (note the proximity of these two residues in Fig. 1). We previously showed that the pH-dependent pattern of H 2 O 2 -induced free radical formation in DHP A (40) is linked to the balance of open/closed conformations of the distal histidine (His-55) (23,28,50), and the protonation of His-55 promotes its open conformation (associated with the five-coordinate ferric heme form). Thus, at lower pH values, H 2 O 2 has easier access to the heme but a lower chance to be kept there long enough to react with the heme.  Table 2. Each mutant exhibited a H 2 O 2 concentration dependence on the rate of TCP enzymatic oxidation as observed previously for WT DHP A (22). In the absence of DHP (non-enzymatic control), no product was observed under the conditions examined, in agreement with previous reports (19,22). The catalytic efficiency, defined as the ratio k cat /K m , is also provided in Table 2 for the different DHP variants. The data presented in Table 2 allow for the following observations. (a) Removal of Tyr-34 or Tyr-38 in DHP A resulted in an increase of the maximal turnover number k cat (notably more so when both tyrosines were removed). A similar result was obtained for the DHP B mutants for Tyr-28 and Tyr-38. (b) The K m values for H 2 O 2 for the mutants were not Ͼ3-fold greater than those of the WT enzymes with the exception of DHP B (Y28F). The latter 34-fold increase demonstrates that H 2 O 2 binding can be affected significantly by replacing a residue that is 10 Å away from the heme, possibly due to the fact that this mutation occurs at the junction of the B and C helices. (c) The generally higher turnover numbers (k cat ) and higher K m values in the mutants resulted in the catalytic efficiencies (k cat / K m ) being only marginally different in the DHP A and B variants when compared with the WT isoenzymes.
Spectroscopic Characterization of the Mutants Reacting with H 2 O 2 -The intermediates of the reactions of the variants of DHP A and DHP B with hydrogen peroxide were characterized by UV-visible and EPR spectroscopies. Optical spectra were collected for the reaction of enzyme (10 M) with H 2 O 2 (2.5-25 eq), and the fast time-scale changes were subjected to SVD analyses. The rapid freeze-quench method was used to prepare samples for the EPR analysis of the free radicals rapidly formed and decayed upon reaction of enzyme (50 M) with a 10-fold excess of H 2 O 2 . Typical for the EPR method, a higher concentration of H 2 O 2 , as compared with optical spectroscopy or activity assays, was employed to facilitate spectral acquisition. As the ferryl intermediates were identified in the stopped-flow studies at a significantly lower peroxide concentration (25 M) and the biochemical assays demonstrated that DHP at a physiological concentration exhibits its enzymatic activity for a range of H 2 O 2 concentrations (10 -200 M), we conclude that the chemistry observed at higher peroxide concentrations reflects the chemistry that occurs at more physiologically relevant concentrations. Table 3 summarizes these optical and EPR data related to the intermediates formed under H 2 O 2 treatment for the different DHP A and DHP B variants.
A Note on the "Compound" Terminology-There is an ambiguity in the terminology adopted in the literature for describing the oxidation states of heme proteins and enzymes. We will use the Compound terminology when referring to the oxidation states of the heme as opposed to the case when it is referred to the whole protein molecule. In this paper the term "Compound I" describes a state when the heme iron is in a ferryl state (Fe 4ϩ AO 2Ϫ ) and the porphyrin macrocycle is one electronoxidized, i.e. a cation radical state. The term "Compound II" refers to a ferryl heme (Fe 4ϩ AO 2Ϫ ) irrespective of whether a radical is present or not on the protein moiety of the enzyme.  From this point of view, "Compound ES" is a particular case of Compound II, which becomes confusing, so we will try, whenever possible, to explicitly indicate the ferryl heme oxidation state (Fe 4ϩ AO 2Ϫ ) rather than use the term Compound II. We also feel that the term Compound ES should be only applicable to the WT enzyme to avoid ambiguity in describing Compound ES in different mutants where the protein radical might be located on different amino acids. DHP A (Y34F); a Ferryl Heme ϩ a Protein Radical-An SVD analysis of the rapid changes in the UV-visible spectra of this mutant resulted in a simple two component mechanism, a 3 b, wherein the starting ferric state was converted into an intermediate assignable to a ferryl (Fe 4ϩ AO 2Ϫ ) heme state, i.e. Compound II (Fig. 3A, Table 3). The rate constant of this pseudo first order conversion was found to be k obs ϭ (3.56 Ϯ 0.02) ϫ 10 4 M Ϫ1 s Ϫ1 . In the WT enzymes of both isoforms, this state is associated with a Compound ES (22,37), when the heme is in the ferryl oxidation state and a radical is formed on the protein moiety of the enzyme. The EPR spectra of DHP A (Y34F) react-ing with H 2 O 2 , measured at different times after the initiation of the reaction, showed a time-dependent multicomponent radical signal (Fig. 4A). There was no evidence of formation of Compound RH within the first 3 s of the reaction; however, at a longer time (ϳ1 min) a species with optical characteristics (416, 560, 598 nm) reminiscent of Compound RH was observed (Table 3).
DHP A (Y38F); a Ferryl Heme ϩ a Protein Radical-Similar to the Y34F variant above, the rapid changes in the optical spectra after mixing DHP A (Y38F) with peroxide were fitted with a two component model a 3 b, wherein the starting ferric state ( Table 1, DHP A (Y38F)) converted (k obs ϭ (2.78 Ϯ 0.01) ϫ 10 4 M Ϫ1 s Ϫ1 ) into a ferryl heme state (Fig. 3B, Table 3). Unlike either WT DHP A or the Y34F variant, formation of Compound RH was not seen for DHP A (Y38F). Rather, the ferryl intermediate converted to a new, stable species whose spectral features strongly resembled those for the oxyferrous form of DHP, Compound III (Table 3). A free radical was also formed in the DHP A (Y38F) mutant upon H 2 O 2 addition (Fig. 4B); however,  NOVEMBER 15, 2013 • VOLUME 288 • NUMBER 46

JOURNAL OF BIOLOGICAL CHEMISTRY 33475
the EPR spectrum of the radical is notably different from that in DHP A (Y34F) (see above).
DHP A (Y34F/Y38F); Formation of Compound I followed by a Species Reminiscent of a Ferryl Heme ϩ a Protein Radical-The chemistry observed with the Y34F/Y38F double mutant of DHP A is markedly different from that observed for the Tyr-34 or Tyr-38 single mutation variants. An SVD analysis of the optical spectra (Fig. 3C) generated upon rapid mixing of DHP A (Y34F/ Y38F) with an excess of peroxide yielded a three-component mechanism, a 3 b 3 c wherein the initial ferric heme enzyme, a, is converted (k obs ϭ (1.13 Ϯ 0.02) ϫ 10 5 M Ϫ1 s Ϫ1 ) to an intermediate, b, whose spectral characteristics are consistent with a Compound I state (51-57) (see the justification of this assignment under "Discussion"). This new intermediate further converted (k obs 5.4 Ϯ 0.1 s Ϫ1 ) to a state, c, that exhibits spectral features (particularly the red shift of the Soret band) that are suggestive of a mixture of Compound I and II (Table 3) (48,49,53,54,58,59). The EPR spectroscopic data also revealed that yet another type of free radical was formed in the Y34F/Y38F double mutant of DHP A upon reaction with H 2 O 2 (Fig. 4C).
Mini Summary for the DHP A Mutants-The two single mutants of DHP A, Y34F and Y38F, when reacting in the ferric heme state with H 2 O 2 , showed a rapid transformation to the ferryl heme species (Compound II). No fast conversion of the latter to Compound RH was detected; rather, a species with spectral characteristics similar to those of Compound III was formed later in the reaction. In contrast to the single mutants, the first observed intermediate upon reaction of H 2 O 2 with the double mutant DHP A (Y34F/Y38F) was Compound I, which rapidly converted to a mixture of Compounds I and II. In all three mutants, free radicals were detected under peroxide treatment (Fig. 5). Remarkably, the radicals were different in the three DHP A mutants. Their EPR spectra are shown for easy comparison on a common magnetic field axis in Fig. 4D.
DHP B (Y28F); a Ferryl Heme ϩ a Protein Radical followed by a Species Reminiscent of Compound RH-UV-visible characteristics of ferric DHP B (Y28F) are given in Table 1. Upon the addition of H 2 O 2 , spectral features consistent with a ferryl state (Compound II) were observed (k obs ϭ (1.58 Ϯ 0.08) ϫ 10 5   Table 3). The EPR signal of the radical thus formed in the Y28F mutant of DHP B is similar to the "pH 5 radical" in DHP A (22,40), previously assigned to a Tyr-38 radical (Fig. 7A and Table 4).
Although the two spectra in Fig. 7, A, are not exactly superimposable (possibly because the coefficient of subtraction for the first spectrum, 0.0616, was estimated), it is important that all eight components in the line shapes of the two signals coincide with a very good accuracy, indicating that the same EPR signal is the main contributor to both spectra.
DHP B (Y38F); Formation of Compound I followed by a Species Reminiscent of a Ferryl Heme ϩ a Protein Radical That Is the Same as in DHP A (Y34F/Y38F)-When the DHP B (Y38F) mutant in the ferric heme state (optical characteristics given in Table 1) reacted with hydrogen peroxide under the same concentration and pH conditions as employed above, a species was formed (k obs ϭ (1.46 Ϯ 0.08) ϫ 10 5 M Ϫ1 s Ϫ1 ; Fig. 6B) whose optical spectral features (Table 3) matched very well those reported for DHP A (Y34F/Y38F) Compound I. This compound in DHP B (Y38F) was rapidly converted (k obs ϭ 6.5 Ϯ 0.1 s Ϫ1 ) in the absence of an exogenously added substrate to a more stable state that had optical characteristics that resembled a mixture of Compounds I and II (note the red shift of the Soret band), again exhibiting similar behavior as DHP A (Y34F/Y38F). Fig. 7B demonstrates that the double mutant of DHP A, in which Tyr-34 and Tyr-38 are replaced with phenylalanines, shows the same EPR spectrum as DHP B in which Tyr-38 is replaced and that lacks Tyr-34 naturally. In WT DHP B, the same EPR signal (37) is seen at a later time after the addition of a 10-fold excess of peroxide (Fig. 7B).

DHP B (Y28F/Y38F); Formation of Compound I followed by a Species Reminiscent of a Ferric Heme with No Detectable Protein
Radical-Compound I was also formed (k obs ϭ (1.64 Ϯ 0.14) ϫ 10 5 M Ϫ1 s Ϫ1 ) in the double mutant Y28F/Y38F of DHP B when   Table 3). Decay of this Compound I did not yield a detectable Compound II species (note the lack of a red shift in the Soret band) as was observed in the other DHP mutants, such as DHP A (Y34F/Y38F) or DHP B Y38F), but resulted in a final species that exhibited a ferric-like absorption spectrum (k obs ϭ 6.4 Ϯ 0.1 s Ϫ1 ; 404 (Soret), 518, 542 nm (sh)). At longer observation times, heme bleaching was noted (data not shown). No radical was detectable by the EPR spectroscopy in DHP B (Y28F/Y38F) at 77 K.
Mini Summary for the DHP B Mutants-DHP B naturally lacks Tyr-34, so it is not surprising that the single mutant DHP B (Y38F) demonstrates the same phenomenology upon reaction with H 2 O 2 as does the double mutant DHP A (Y34F/Y38F); namely, formation of Compound I followed by its reduction to a state reminiscent of a ferryl heme with a protein radical. Importantly, the EPR spectra of the radicals in the two mutants are identical, indicating that the radical is formed on the same residue in the two proteins (Fig. 5). In the other single mutant of isoenzyme B, DHP B (Y28F), a more conventional ferryl heme state is formed with a protein radical that is likely to be the same as the "pH 5 radical" in WT DHP A, previously attributed to Tyr-38 (40). The ferryl heme ϩ Tyr-38 radical state evolves (within 3 s) to a species that is reminiscent of Compound RH (37). The double mutant DHP B (Y28F/Y38F), which is three tyrosines down when compared with WT DHP A, shows Compound I as the first intermediate of its reaction with peroxide, but it does not show its transformation to a ferryl heme and/or a protein radical and evolves directly to a ferric-like heme form. The latter is likely to be a damaged enzyme as the bleaching of the protein follows. The tyrosyl radicals mostly responsible for the line shapes presented in different panels are indicated in small boxes (see "Discussion" for these assignments). A, the DHP A signal previously obtained as the difference indicated and attributed to the Tyr-38 radical (pH 5 radical) (40) and the EPR spectrum detected in the Y28F mutant of DHP B. The two EPR spectra were obtained on different EPR spectrometers operating at slightly different microwave frequencies causing the same EPR signal to appear at slightly different magnetic field values. Therefore, the absolute values of the magnetic field are not indicated, and the signals were superimposed on the basis of the g-factors.
The positions of the eight components of the EPR signal, common for both spectra, are indicated. B, the EPR signals of the free radicals formed in DHP A (Y34F/Y38F) and DHP B (Y38F). (note: DHP B does not have Tyr-34). The third signal of a similar line shape is from WT DHP B but freeze-quenched at a later time after the addition of peroxide (37). The bottom spectrum is a simulation of the tyrosyl radical EPR spectrum for the TRSSA (41) input parameters C1 ϭ 0.38 and ϭ 46.0°; the simulation parameters yielded by this two-value input are given in Table 4. C, the EPR spectra of the Tyr-34 radical in WT DHP A (as reported in Ref. 40) and in the DHP A Y38F mutant (as in Fig. 4). Similar to panel A, the two spectra were measured on different EPR spectrometers and the magnetic field values are not shown. D, EPR spectra show the presence of the Tyr-38 radical in WT DHP B (as a major species) (37), in WT DHP A (as a minor species but seen well at pH 5) (40) and in the Y34F mutant of DHP A at pH 7 (as in Fig. 4). Note that the free radical spectrum of WT DHP B at a later time point (see panel B) is consistent with the radical being localized on a different residue, Tyr-28. As in A and C, the magnetic field values are not shown.

TABLE 4
The parameters generated by TRSSA for C1 ‫؍‬ 0.38 and 0 ‫؍‬ 46.0°and used to simulate the spectrum in Fig. 7B a The hyperfine interaction constants A are given in MHz, Euler angles are given in degrees.

DISCUSSION
Assignment of Radicals to Residues-The current study of the DHP mutants unequivocally supports our previous TRSSAbased radical assignment in DHP A (40); elimination of Tyr-38 results in the EPR spectrum previously attributed to the Tyr-34 radicals (Fig. 4), and elimination of Tyr-34, when Tyr-38 is kept intact, results in the spectrum of the pH 5 radical previously attributed to Tyr-38 (Fig. 7). As a further confirmation of such radical site assignment, the spectrum of the WT DHP B (37) (note that this isoenzyme has Asn-34 in place of Tyr-34 of DHP A) shows a spectrum of the Tyr-38 radical (Fig. 7D).
Interestingly, although WT DHP B and Y34F DHP A (having identical set of tyrosines) show the same Tyr-38 radical as the principal species, WT DHP A at pH 5, where the Tyr-34 radical is not seen, exhibits the same EPR line shape but at a lower resolution of the spectral components (Fig. 7D), which indicates that there might be another radical present in WT DHP A at pH 5. It is reasonable to suggest that this type of radical in DHP A is the one we see in the double mutant when both Tyr-34 and Tyr-38 are replaced (Figs. 4C and 7B). As expected, the Y38F DHP B mutant (in which Tyr-34 is missing naturally) shows the same EPR spectrum (Fig. 7B). This suggests that there is a third site of radical localization in addition to the already established Tyr-34 (in DHP A) and Tyr-38 (in both DHP A and B), which is common for both isoenzymes.
Which residue can this be? We suggest that the third site of radical stabilization is another tyrosine. It was possible to simulate the EPR line shape of the radical by using TRSSA (Fig. 7B), which means that a tyrosyl radical can indeed be responsible for the EPR signal. The simulation shown has been obtained for the TRSSA input parameter ϭ 46.0°(the rotation angle of the tyrosyl ring). Now the question should be addressed, Which tyrosines in the DHP A and B have their ring rotation angles close to this value or to the complementing symmetrical angle of 74.0°? (note that complementing angles 46°and 74°, totaling 120°, will yield identical sets of the EPR spectrum simulation parameters.) Table 5 shows the results of the analysis performed on the website of the Tyrosine Residues in Different Proteins database (46). Overall, the evidence unequivocally suggests that Tyr-28 is the most likely site of the radical, with the EPR spectral features presented in Fig. 7B in both DHP A (Y34F/Y38F) and DHP B (Y38F) ( Table 3).
Our hypothesis that the third radical site is Tyr-28 is further supported by the study of the DHP B (Y28F) mutant; the EPR spectrum of this protein when activated by H 2 O 2 is consistent with the Tyr-38 radical (Fig. 7A), an obvious scenario for the protein that is naturally lacking Tyr-34. We also note that although the initial (0.3 s) radical site observed in WT DHP B is Tyr-38 (Fig. 7D), the radical character moves at a later time (2 s) to the nearby Tyr-28 (Fig. 7B), a significantly more solventexposed tyrosine residue.
Compound I Is Detected When Both Tyr-34 and Tyr-38 Are Missing-The optical spectral features of DHP A (Y34F/Y38F) (Fig. 3C), DHP B (Y38F) (Fig. 6B), and DHP B (Y28F/Y38F) (Fig.  6C) that are observed upon their reaction with peroxide, particularly the distinct hypochromicity of the Soret band as well as the appearance of a visible absorbance at ϳ640 -650 nm, are highly reminiscent of Compound I in other hemoproteins, including WT sperm whale myoglobin (51) and its His-64 mutants (52), horseradish peroxidase (HRP) (53), Arthromyces ramosus peroxidase (54), peanut peroxidase (55), soybean peroxidase (56), and Rhodnius prolixus nitrophorin 2 (57). Thus, we assign the first detectable peroxide-induced intermediates in DHP A (Y34F/Y38F), DHP B (Y38F), and DHP B (Y28F/ Y38F) to Compound I (Table 3). In the absence of an exogenously added substrate, this species for the variants that retain Tyr-28, i.e. not in DHP B (Y28F/Y38F), rapidly decays to a more stable state with spectroscopic features that are similar to, but do not quite match those of, a typical Compound II/Compound ES. It is likely that this state, with a red shift of the Soret band, represents an equilibrium mixture of Compounds I and II that has been noted before for other peroxidases, including HRP (53,58), A. ramosus peroxidase (54), and catalase-peroxidase (48,49,59). At longer reaction times (up to 85 s), this state undergoes bleaching.
Thus, Compound I was observable only when the two primary sites of radical localization, Tyr-34 and Tyr-38, were both absent, as in DHP A (Y34F/Y38F), DHP B (Y38F), and DHP B (Y28F/Y38F). For the first two of these three mutants, Compound I was found to partially convert to Compound II, and a tyrosyl radical was formed on the residue identified as Tyr-28. In DHP B (Y28F/Y38F), neither a free radical nor Compound II was observed. This might be explained by a very fast two-electron reduction of Compound I via concerted or two rapid sequential one-electron transfer events, which brings the heme to a ferric state but results in some oxidative damage to the protein.
Peroxidase Activity of the Mutants-Removal of both Tyr-34 and Tyr-38 not only makes Compound I detectable but also  (Fig. 7B).
increases the turnover number in both isoforms (  Table 3). The observation of these differences parallels our earlier conclusion about the formation of a covalent cross-link between the heme and Tyr-34 in WT DHP A Compound RH (40). Obviously, such a cross-link cannot be formed in the other three aforementioned variants that lack Tyr-34.
Although the exact nature of the Compound RH species will require further elaboration, the present study provides additional evidence that Compound RH is a protected state of DHP, the formation of which is mediated/regulated by tyrosyl radicals. As pointed out above, in an environmental situation when the aggressive oxidant H 2 O 2 is present but the substrates meant to be oxidized are absent, most globins would start oxidizing themselves (and other cell components). Formation of Compound RH in DHP is a two-way solution to this problem. First, the enzyme, although remaining in the ferric heme state, is significantly deactivated in terms of reactivity toward H 2 O 2 , possibly via a mechanism in which the distal histidine is arrested in a conformation that prevents the heme from further interacting with H 2 O 2 (40). Second, the oxygen binding function of DHP can be rescued from Compound RH as the oxyferrous form can be generated from Compound RH under mild reducing conditions (i.e. sodium dithionite in vitro) (22,37).
Why Nature Employs Tyrosyl Radicals in DHP-The data presented in Tables 2 and 3 show that the peroxidase reaction is generally faster when catalyzed by the DHP mutants that lack the ability to form a stable and long-lived radical on Tyr-34 or Tyr-38. An alternative way to look at it is that those mutants that give rise to Compound I is more active toward substrate oxidation. The question thus arises as to why nature utilizes tyrosyl radical formation in DHP, as it appears that the presence of a tyrosyl radical diminishes the ability of the enzyme to catalyze substrate oxidation. We propose that this is likely a protective mechanism that is unique to DHP as a dual function globin-peroxidase. Specifically, in the absence of reducing sub-strate, Compound I rapidly (Ͻ1 s) undergoes heme bleaching (DHP A (Y34F/Y38F) or DHP B (Y38F)) or forms an inactive ferric-like state (DHP B (Y28F/Y38F)) that also exhibits some bleaching (Table 3). This represents the loss of both peroxidase and globin functions of DHP. By invoking a tyrosyl radical, the peroxidase activity of DHP is attenuated, but the oxidized (activated) form of the enzyme (as Compound ES) is nearly 2 orders of magnitude longer lived than Compound I, increasing the likelihood that the enzyme will react with substrate to regenerate the ferric resting form without heme bleaching.
Why is Compound I so much more prone to bleaching than Compound ES? As a globin, DHP possesses a relatively high redox potential (ϳϩ200 mV (37 and 61) for the Fe 3ϩ /Fe 2ϩ couple in comparison to those for peroxidases (HRP, Ϫ266 mV (62), cytochrome c peroxidase, Ϫ182 mV (63), myeloperoxidase, ϩ5 mV (64)) and even other globins (sperm whale Mb, ϩ43 mV (65), horse heart Mb, ϩ46 mV (66), horse heart Hb, ϩ152 mV (67), human Hb, ϩ158 mV (68)). The high redox potential is to ensure that the oxygen binding function is maintained, as only a heme in the ferrous oxidation state can reversibly bind molecular oxygen. In Mb and Hb, which both possess similarly high redox potentials as DHP, Compound I is rapidly reduced to form ferryl species with protein-centered radicals, likely to avoid heme degradation. For DHP, it appears that a similar protective mechanism is in place; as the redox potential for the ferryl/Fe 3ϩ couple correlates and trends with the Fe 3ϩ / Fe 2ϩ couple (69), the unusually high redox potential for the Fe 3ϩ /Fe 2ϩ couple in DHP leads to an estimated redox potential of ϳ1.6 V for the ferryl/Fe 3ϩ couple, which would rank among the highest for any peroxidase. It is likely that it is for this reason that the Compound I intermediate in the DHP tyrosine mutants rapidly undergoes heme bleaching. By introducing tyrosine-mediated electron transfer pathways (70) in DHP, nature has provided an endogenous reducing substrate that effectively quenches Compound I, leveling the redox potential of one of the two oxidizing equivalents to that of tyrosine (ϳ900 mV) (71,72) and protecting the enzyme from irreversible heme bleaching.
Conclusions-The present study provides support for our explanation of the structure-function relationship that enables the dual activity in the dehaloperoxidase hemoglobins. To sustain its oxygen binding function, DHP possesses a high redox potential that is atypical of monofunctional peroxidases because it would imply a fast autoreduction of Compound I that could cause the enzyme's self-oxidation and heme degradation. To function as a peroxidase without self-damaging, DHP employs tyrosines as endogenous reducing co-factors that allow reduction of Compound I to Compound II, which also leads to formation of a radical on the tyrosine. The employment of tyrosines in the catalytic chemistry of DHP allows for Compound ES to live long enough to have a chance to encounter a substrate to oxidize. Such a purpose for an active site tyrosine does not need to be invoked in the classical monofunctional peroxidases, such as HRP, given that their ferryl/Fe 3ϩ redox potentials are generally around 1 V (69), which is similar to that of tyrosine itself. As a hallmark of the globin-peroxidases, tyrosine radical formation in DHP is, therefore, nature's adaptation that enables peroxidase activity within a globin framework.