Characterization of Nonheme Iron and Reaction Mechanism of Bromoperoxidase in Corallina pilulifera"

The properties of the nonheme iron of bromoperoxi- dase from Corallina pilulifera were studied. The enzyme lost its activity when reduced with formamidine- sulfinic acid and recovered it when oxidized by air. Incubation of the enzyme with ferric or ferrous ion- chelating agents indicated that its nonheme iron was ferric. Analyses of circular dichroism and proton NMR spectra suggested that the ferric ion tightly bound to cysteine, histidine, or tyrosine residues of the enzyme. The enzyme catalyzed Br--dependent catalase reac- tions to yield 1 mol of O2 from 2 mol of H,02. No O2 evolution was observed when bromination reaction of monochlorodimedone occurred. From these results, to-gether with previous knowledge of this enzyme, it was concluded that it activated bromide anion (Br-) to bro- monium cation (Br+) using one molecule of H202, and this Br+OH- formed at the active site then decomposed another H20z to yield O2 in the absence of halogen acceptors (substrate). When substrate was present in the reaction mixture, it and HzOz competitively reacted with the reaction intermediate (Br+OH-) to give brom-inated products. are tyrosine phenolate and histidine imidazole anions.

The detailed properties of the enzyme of C. pilulifera were reported in previous papers (6, 7), but those of the enzyme's nonheme iron were not clear. Elucidation of its valence state and ligand properties is important in understanding the halogenation mechanism of the enzyme.
The halogenation intermediate of the enzyme is a halonium cation (X+), not a radical (4,7,14). It has been found that the bromination reaction of the enzyme occurs at its active * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$. To whom all correspondence should be addressed. Present address: Research Laboratory, Amano Pharmaceutical Co., Ltd., Nishiharu,Aichi 481,Japan. site (14). On the other hand, bromination by the H-type chloroperoxidase of the fungus C. fumago is caused by the release of molecular halogen into the reaction mixture (10,14,15). In addition, no catalase and only slight peroxidase activities exhibited by bromoperoxidase of C. pilulifera (7) imply that its reaction mechanism is quite different from that of H-type haloperoxidase. This work describes the halogenation mechanism of nonheme bromoperoxidase of C. pilulifera and clarifies the differences between the reaction mechanisms of heme and nonheme haloperoxidases.

MATERIALS AND METHODS
Collection of Algae and Enzyme Purification-C. pilulifera (Corallinaceae, Rhodophyta) was collected from shallow water (0.5-1.0 m deep) on the shores of Takahama (Fukui Prefecture), Japan, in April 1985, and frozen at -20 "C until use.
The enzyme was purified from crude extracts of C. pilulifera as described in the previous paper (6).
Characterization of the Purified Enzyme-The properties of the purified enzyme were analyzed according to the methods described in previous papers (7, 16) and compared with those of enzyme purified from the same algae collected at a different site (Shirahama, Wakayama Prefecture, Japan). Enzyme purified from algal samples collected at Takahama was used throughout this study.
Enzyme Assay-Bromoperoxidase activity was assayed by measurement of the change in absorbance at 290 nm caused by the conversion of monochlorodimedone to monobromomonochlorodimedone (4).' Catalase activity was measured by the modified method of Beers and Sizer (17), whose detailed reaction conditions were described in the previous paper (4). Peroxidase activity was measured with o-dianisidine at 25 "C (18). 0 2 concentration in the reaction mixture was monitored with a Beckman Fieldlab oxygen analyzer equipped with a Beckman 39533 polarographic oxygen sensor. Before analysis, the reaction mixture was saturated with nitrogen gas, and the reaction was carried out in a closed system with constant stirring by a magnetic stirrer. Throughout this report, 1 unit of enzyme activity is equal to the amount of enzyme that converted 1 pmol of substrate in 1 min at 25 "C.
('HNMR) spectra were recorded at 300 MHz on a Nicolet NT-300 Instrumental Analysis-Proton nuclear magnetic resonance spectrometer equipped with a 1280 computer system. For recordings of paramagnetically shifted proton NMR spectra, typically 50K transients were accumulated to obtain the Fourier transformed spectrum with 8K data points and 6.0-ps 90" pulse. The enzyme sample was dialyzed against 5 mM potassium phosphate buffer (D20, pH 7.0) before measurement. Electron spin resonance (ESR) spectra were measured at 77K with a JEOL JES-FE3X ESR spectrometer. Before analysis, the ESR samples were dialyzed against 50 mM potassium phosphate buffer (pH 7.0) containing 1 mM EDTA, and then against the buffer, and frozen in liquid nitrogen. Circular dichroism (CD)' spectra were recorded at room temperature in a 10-mm light-path cuvette with a JASCO 5-500 C spectropolarimeter equipped with an electromagnet (1.5 T). N. Itoh, A. K. M. Q. Hasan, Y. Izumi, and H. Yamada, manuscript The abbreviation used is: CD, circular dichroism. in preparation.

Differences in Enzyme Properties of Algae
Collected from Different Sites-The enzyme purified from C. pilulifera collected at Takahama (Japan Sea) was identical to that from Shirahama (Pacific Ocean) in the following ways: elution patterns of DEAE-Sepharose and Sepharose 6B chromatographies, K, values for KBr and H202, optimum pH, electrophoretic mobilities on disc gel and sodium dodecyl sulfateslab gels, and immunological properties (16). However, the contents of iron and magnesium, which affected the specific activities, absorption spectra, and dry weight extinction coefficients (E% nm), were found to be different, as shown in Table  I Mini~rint.~ Valency State of Nonheme Iron of the Enzyme-Incubation of the native enzyme with 2 mM formamidinesdfinic acid, a mild reducing agent, for about 2 h under anaerobic conditions caused the enzyme's complete inactivation. When it was reoxidized thereafter by bubbling with air, its activity returned almost to the original level ( Fig. 1, Miniprint). Addition of 2 mM sodium dithionite caused a similar change, but enzyme activity was not restored by oxidation with air.
When the enzyme was incubated with 2 mM Tiron which is a chelating agent specific for ferric ion (19), changes in spectra of the enzyme were observed, which indicated the formation of a chelated complex (Fig. 2a, Miniprint). Absorbance at 450 nm was increased, with a concomitant decrease in enzyme activity (Fig. 2b, Miniprint).
Incubation of the enzyme with a ferrous ion chelatingagent, o-phenanthroline, had no effect on the enzyme activity, but incubation with o-phenanthroline under reducing conditions completely inhibited it. These data showed that nonheme iron of bromoperoxidase was ferric.
Instrumental Analyses of the Enzyme-As described in the previous paper, the enzyme's nonheme iron tightly bound to the polypeptide residues, and no decrease in iron content was observed after prolonged incubation with EDTA or o-phenanthroline (7). It was also found that the enzyme contained no acid-labile sulfide (7). In the CD spectra of native and reduced enzyme samples, there were no characteristic peaks corresponding to [2Fe-2S] and [4Fe-4S] clusters, which are common in nonheme iron proteins (Fig. 3, Miniprint) (20, 21). These spectra were rather similar to that of rubredoxin, whose nonheme iron forms complexes with sulfur donor ligands of 4 cysteine residues (22).
Only a broad peak at 22.3 ppm, caused by protons in paramagnetic ferric ion field was observed in proton NMR spectrum of the enzyme. Further information concerning the ferric ion ligands could not be obtained from the proton NMR spectrum because of the enzyme's high molecular weight (790,000).
No ESR-active ferric ions were observed in native enzyme under the conditions tested (77K). Addition of 2 mM Hz02, Tables I and I1  with or without 20 mM KBr to the enzyme caused no change in the ESR spectrum.

Portions of this paper (including
Br--dependent Catalase Reaction of the Enzyme-Catalase activity of the enzyme was assayed by spectrophotometric measurement of the decrease in H202 concentration at 240 nm and oxygen electrode measurement of the formation of 0,. The enzyme showed no catalase activity in the absence of bromide. However, in the presence of bromide it decomposed 2 mol of H,Oz to give 1 mol of 02, as shown in Table 11, Miniprint. In the presence of monochlorodimedone, a good acceptor of bromonium cations, no 0, formation was observed. The specific activity of the Br--dependent catalase reaction was just twice that of the bromination of monochlorodimedone. The K, values for KBr and H2O2 of both reactions were identical. Therefore, it was thought that the enzyme's Br-dependent catalase and bromination reactions would proceed by the same mechanism. Catalase Reaction of the Enzyme in the Presence of Monochlorodimedone-The rate of 0 2 formation from H,O, was carefully measured at different monochlorodimedone concentrations (Fig. 4, Miniprint). It was found that O2 evolution occurred only after the complete consumption of monochlorodimedone, indicating that monochlorodimedone was brominated preferably by the bromination intermediate (Br+). After the consumption of monochlorodimedone, the bromination intermediate was then able to act on HzOz.
Comparison of the Enzyme Bromination Reaction with the NaOBr Reaction-NaOBr (HOBr) is known to oxidize H20, to yield 0, as follows: NaOBr + H20, + NaBr + 0, + HzO (23)(24)(25)(26). Supposing that the reaction intermediate of bromoperoxidase were NaOBr (HOBr), and that it was released into the reaction mixture, then the enzymic and chemical reactions of NaOBr should show the same rate of 0, evolution in the presence of several halogen acceptors. Bromoperoxidase of C.
pilulifera catalyzes the bromination of not only monochlorodimedone but also of many other organic compounds, such as substituted phenols (4), heterocycles (14), and substituted alkenes.' We compared the 0, formation rate of the enzyme reaction with that of the NaOBr reaction in the presence of either monochlorodimedone, phenol, or cytosine. The rate of feeding NaOBr solution into the reaction mixture was adjusted until the 0, formation rate was the same as in the enzymic reaction in the absence of halogen acceptor. As shown in Fig. 5, Miniprint, in the NaOBr reaction, 0, formation was observed at high concentrations of monochlorodimedone (1.25 mM), a result completely different from that obtained with the enzymic reaction (Fig. 4, Miniprint). The rates of 0, formation in the enzymic reaction in the presence of phenol or cytosine also differed from those of the NaOBr reaction. Thus, it was obvious that the enzyme possessed some affinity for monochlorodimedone and cytosine. These data suggested that bromination reaction of the enzyme was not due to the release of NaOBr or molecular bromine into the reaction mixture, which was in accord with previous results (14).
Kinetics of 0, Formation by the Enzyme Reaction in the Presence of Phenol-The effects of the presence of the halogen acceptor, phenol, on the kinetics of 0, formation were examined in detail (Fig. 6, Miniprint). The kinetic data indicated that phenol's inhibition of 0, formation was competitive, with a Ki value calculated from the plot data to be 3.8 X

M.
The results revealed that H,O, and phenol competitively reacted with the bromination intermediate formed at the enzyme's active site.

DISCUSSION
Bromoperoxidase of C. pilulifera is an interesting haloperoxidase which possesses a nonheme iron as a prosthetic group.
As described in the previous paper (7), the enzyme isolated from C. piluliferu collected at Shirahama (Pacific Ocean, temperate sea) contained 2.3 f 0.2 iron atoms and 1.6 f 0.1 magnesium atoms/molecule of dodecameric enzyme. In this study, enzyme was again purified to homogeneity from algae at a different collection site, Takahama (Japan Sea, temperate sea). It was found that the newly purified enzyme differed from the previous one in its iron and magnesium contents (14 +: 0.5 iron and 0.7 f 0.1 magnesium atoms/molecule of enzyme), but the other physicochemical and immunological properties (16) were identical. It was speculated that the algal growing conditions, e.g. growing season, day light, tide movement, mineral and nutrient compositions, and temperature of the sea water, probably affected the metal content of the enzyme, resulting in different enzyme activities. Seasonal changes in bromoperoxidase activity has been reported in Rhodomela lurix (1). Ferric ions were found to be essential for the enzyme activity of C. piluliferu. However, the reason for the low specific activity of the Takahama enzyme in spite of the high ferric ion content is obscure (Table I). Compared with the Shirahama enzyme, the magnesium content of that from Takahama was low, so, magnesium ions may play a very important role in conformational maintenance of the enzyme, which in turn would affect its specific activity. Other factors resulting in low enzyme specific activity such as protease digestion of the enzyme toward the end of the growing season are possibilities. We have not yet checked the seasonal variations of algal bromoperoxidase activity at any one collection site.
Bromoperoxidase of C. piluliferu rapidly lost its activity during reduction by formamidinesulfinic acid but recovered it following subsequent oxidation by air (Fig. 1). This suggested that the valency state of nonheme iron of the enzyme was a) En -Fa'' trivalent (Fe (111)). Tests on the enzyme's complex formation, using Tiron and o-phenanthroline supported this suggestion. However, the reason why the loss of activity which was plotted in Fig. 2 did not parallel the change in the absorption spectrum is obscure. We speculated that the interaction between the enzyme's ferric ion and Tiron is weak, so the Tiron-treated enzyme recovers its activity under the assay conditions, in the presence of HzOZ and KBr. Inability to detect ESR active ferric ions was probably due to the rapid relaxation time at 77K of the enzyme's nonheme iron electrons. Tiron is known to form chelated complexes with free ferric ions in a short time (19); Kojima et ul. (27) reported that pyrocatechase, an enzyme which contains Fe (III), lost its activity in about 60 min when incubated with Tiron. Thus, the slow rate of complex formation of bromoperoxidase with Tiron (Fig. 2) suggested that the enzyme's ferric ions were strongly bound to the polypeptide residues at the active site, which prohibited easy complex formation with metal-chelating agents. That the enzyme did not lose its activity following prolonged incubation with EDTA or the treatment with 6 M urea for several hours is in accord with this proposal. The enzyme has been found to contain no acid-labile sulfide (7), a finding which was supported by our analysis data of CD spectrum of the enzyme. The similarity of the CD spectra of the enzyme to ferric-rubredoxin implied that ferric ion directly bound to 4 cysteine residues of the enzyme. The results of amino acid analysis of the enzyme, which found 4 cysteine residues/ enzyme subunit (7), are not in conflict with this speculation. We failed to measure the enzyme's resonance Raman spectrum because the purified enzyme was contaminated with a covalently bound red substance. This red dye interfered with resonance Raman measurements because of its fluorescence. Iodoperoxidase and chloroperoxidase are known to show the halide-dependent catalase activity yielding singlet oxygen, '0, (24-26), but they show catalytic activities even in the absence of halide. Thus, this is the first report of an absolutely Br--dependent catalase reaction. The chemical equivalence of H202 consumption and 0, formation in the Br--dependent catalase reaction suggested that one molecule of H202 was used in the oxidation of Br-to Br+, the latter then reacting with another molecule of H20, to yield one molecule of 02.
This proposal is based on the observations that the halogenation intermediate of C. pilulifera bromoperoxidase was a halonium cation (X+), and that hypobromous acid (Br'OH-) catalyzed the oxidation of H20, to 02. However, in the presence of the halogen acceptor, monochlorodimedone, the active intermediate (Br'OH-) was used only in the bromination of that, and no detectable O2 was produced (Table I1 and Fig.   4). In addition, the bromination rate of monochlorodimedone was the same as that of 0, formation in Br--dependent catalase reaction. Therefore, it was concluded that the enzymic bromination proceeded as shown in the formula in Fig.  7a.
On the other hand, the differences in the 0, formation rates between the HOBr chemical and enzymatic reactions in the presence of halogen acceptors showed that the bromoperoxidase-catalyzed bromination occurred at the enzyme's active site and was not dependent on release of molecular bromine or HOBr into the reaction mixture. The result was in fair agreement with the previous study's findings that no formation of tribromide ions was observed in the reaction mixture (14). On this point, C. pilulifera nonheme bromoperoxidase halogenation is quite different from the usual heme haloperoxidases. The results in Fig. 5 also showed that the enzyme had some affinity toward the halogen acceptors, resulting in different rates of bromination from those shown by the NaOBr reaction.
As shown in Fig. 6, 0, formation by bromoperoxidase catalatic reaction was competitively inhibited by phenol, indicating that both phenol and H202 reacted on the same bromination intermediate. Lack of formation of 0, in the presence of monochlorodimedone (Fig. 4) was probably due to a high enzyme affinity toward it. In the previous paper, we showed that inhibition of monochlorodimedone bromination by fluoride ions was uncompetitive over a concentration range of fluoride ions from 1.25 to 5.0 mM (7). This phenomenon can easily be understood from the enzyme's reaction mechanism. Fluoride ions probably bind to the Br'OH-intermediate to give BrF3. The reactivity of BrF3 as an electrophile is lower than that of Br'OH-because of its high binding energy.
The reaction mechanism of nonheme (NH type) bromoperoxidase of C. pilulifera is summarized in Fig. 7a and compared with that of heme (H type) chloroperoxidase of C. furnago (Fig. 76). The characteristic of the bromoperoxidase reaction in terms of no catalase reaction and minimal peroxidase activity are clearly explained by its reaction mechanism.