Spectroscopic, ligand binding, and enzymatic properties of the spleen green hemeprotein. A comparison with myeloperoxidase.

The bovine spleen green hemeprotein, a peroxidase which exhibits spectrophotometric properties similar to those of granulocyte myeloperoxidase, was purified using an improved method. The ligand affinity of the ferric enzyme was spectroscopically determined using chloride and cyanide as exogenous ligands. The pH dependence of the apparent dissociation constant of the enzyme-chloride complex showed the presence of a proton dissociable group with a pKa value of 4 on the enzyme; chloride binds to the enzyme when this group is protonated with a dissociation constant of 60 microM. The cyanide affinity of the enzyme is also regulated by the group with a pKa value of 4, but in this case cyanide binds to the unprotonated enzyme with a dissociation constant of 0.6 microM; only the protonated, uncharged form of cyanide reacts with the enzyme. Cyanide binding was competitively inhibited by chloride, and chloride binding was also competitively inhibited by cyanide. The EPR spectrum of the resting enzyme exhibited a rhombic high spin signal at g = 6.65, 5.28, and 1.97 with a low spin signal at g = 2.55, 2.32, and 1.82. Upon formation of the chloride complex, the spectrum was replaced with a new high spin EPR signal with g-values of 6.81, 5.04, and 1.95. The cyanide complex showed a low spin EPR signal with g-values of 2.83, 2.25, and 1.66. Examination of the enzymatic activity of the spleen green hemeprotein by following the chlorination of monochlorodimedon has indicated that the enzyme has the same chlorinating activity as myeloperoxidase; the spleen green peroxidase can catalyze the formation of hypochlorous acid from hydrogen peroxide and chloride ion. Comparison of the present data with those of myeloperoxidase has led to the conclusion that the structure of the iron center and its vicinity in spleen green hemeprotein is very similar, if not identical, to that of myeloperoxidase. The spleen enzyme can thus be used as a model to study the active center, and its environment, in myeloperoxidase.

The bovine spleen green hemeprotein, a peroxidase which exhibits spectrophotometric properties similar to those of granulocyte myeloperoxidase, was purified using an improved method. The ligand affinity of the ferric enzyme was spectroscopically determined using chloride and cyanide as exogenous ligands. The pH dependence of the apparent dissociation constant of the enzyme-chloride complex showed the presence of a proton dissociable group with a pK, value of 4 on the enzyme; chloride binds to the enzyme when this group is protonated with a dissociation constant of 60 NM. The cyanide affinity of the enzyme is also regulated by the group with a pK,, value of 4, but in this case cyanide binds to the unprotonated enzyme with a dissociation constant of 0.6 p~; only the protonated, uncharged form of cyanide reacts with the enzyme. Cyanide binding was competitively inhibited by chloride, and chloride binding was also competitively inhibited by cyanide. The EPR spectrum of the resting enzyme exhibited a rhombic high spin signal at g = 6.65, 5.28, and 1.97 with a low spin signal at g = 2.55,2.32, and 1.82.
Upon formation of the chloride complex, the spectrum was replaced with a new high spin EPR signal with gvalues of 6.81, 5.04, and 1.95. The cyanide complex showed a low spin EPR signal with g-values of 2.83, 2.25, and 1.66. Examination of the enzymatic activity of the spleen green hemeprotein by following the chlorination of monochlorodimedon has indicated that the enzyme has the same chlorinating activity as myeloperoxidase; the spleen green peroxidase can catalyze the formation of hypochlorous acid from hydrogen peroxide and chloride ion.
Comparison of the present data with those of myeloperoxidase has led to the conclusion that the structure of the iron center and its vicinity in spleen green hemeprotein is very similar, if not identical, to that of myeloperoxidase. The spleen enzyme can thus be used as a model to study the active center, and its environment, in myeloperoxidase. Recently, there have been several reports of unusual green hemeproteins from various sources (Jacob and Orme-Johnson, 1979;DeFilippi and Hultquist, 1978;Davis and Averill, * This investigation was supported by Research Grant AI-20463 from the National Institute of Allergy and Infectious Diseases. 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. 1981). One is a peroxidase' found in hovine spleen (Davis and Averill, 1981) which has optical spectral properties very similar to those of myeloperoxidase (EC 1.11.1.7) of polymorphonuclear leukocytes. The resonance man (Babcock et al., 1985) and magnetic circular dichr O P 'sm (Ikeda-Saito et al., 1985b) results indicate that the structure of the chromophore of the spleen peroxidase is identical to that of myeloperoxidase, probably an iron chlorin (Sibbett and Hurst, 1984;Ikeda-Saito et al., 1985a;Babcock, et al., 1985). However, the two enzymes are clearly distinct in their size; the spleen green hemeprotein consists of a single polypeptide chain (Mr 57,000) with one prosthetic group per enzyme molecule (Davis and Averill, 1981), while myeloperoxidase is tetrameric (M, 140,000), consisting of two heavy chains (Mr 55,000 with a single prosthetic group per chain) and two light chains (M, 15,000) (An+ews and Krinsky, 1981).
One of the important properties of myeloperoxidase is its ability to catalyze the formation of hypochlorous acid from hydrogen peroxide and chloride ion (Harrison and Schultz, 1976). In the catalytic cycle, myeloperoxidase first reacts with hydrogen peroxide to form compound I which has two oxidizing equivalents more than the resting enzyme. Compound I is converted to the resting ferric enzyme in a two-electron reduction by which chloride ion (the electron donor) is oxidized into a chloronium ion, and hypochlorous acid is produced (Harrison and Schultz, 1976). With its ability to cause rapid degradation of various biological compounds, hypochlorous acid is considered to be the pertinent bacteriocidal agent. Like other peroxidases, myeloperoxidase also forms compounds I1 and I11 upon reaction with excess amount of hydrogen peroxide (Odajima and Yamazaki, 1970;Harrison et al., 1980). As one of the essential components of the antimicrobial systems of polymorphonuclear neutrophils (Klebanoff, 1975), myeloperoxidase is an important protein.
But studies of purified myeloperoxidase have remained limited to a small number of laboratories due to the difficulties of preparing the enzyme. With spectrophotometric properties similar to those of myeloperoxidase, the spleen green hemeprotein has the potential to serve as a model of myeloperoxidase. Before this can be addressed, however, a detailed comparison of the spleen green hemeprotein with myeloperoxidase is required. Since the substrate binding and redox reactions are processes which involve the heme moiety, the properties associated with the heme, such as exogenous ligand binding and EPR spectra, will be excellent probes for judging to what extent the active centers of these two enzymes are similar. It is also of interest to test whether the spleen peroxidase is capable of catalyzing the peroxidation of chloride. To this end, I have purified the spleen green hemeprotein and compared its ligand binding, enzymatic, and the EPR spectral properties with those of myeloperoxidase. Since chloride and cyanide have been used to study the ligand binding properties of myeloperoxidase, as high spin and low spin ligands, respectively (Stelmaszynska and Zgliczynski, 1974;Bakkenist et al., 1980;Bolscher and Wever, 1984a), they were also used here as exogenous ligands. Enzymatic properties were studied by steady-state kinetic measurements of the formation of hypochlorous acid, as followed by the chlorination of monocholordimedon, again as reported for myeloperoxidase . Here, I report that the spleen green hemeprotein has ligand binding, enzymatic properties, and EPR spectra, very similar to those of myeloperoxidase.

EXPERIMENTAL PROCEDURES
The green hemeprotein was extracted from bovine spleens obtained from the local abattoir and partially purified by batch absorption to and elution from cellulose phosphate according to the method of Davis and Averill (1981). The partially purified enzyme was then diluted to 0.1 M KCl, pH 5.5, and loaded to a column of CM-Sepharose CL-GB previously equilibrated with 0.1 M phosphate buffer, pH 6.0. The column was eluted with a linear gradient of 0.1 M phosphate buffer, pH 6 (starting buffer), and 0.6 M phosphate buffer, pH 7 (limiting buffer). The green colored fractions with A,mnm/Amnm values larger than 0.4 were pooled, and concentrated by ultrafiltration (Amicon Centriflo CF-25). The enzyme fraction was gel-filtered on a column of Sephacryl S-200, equilibrated with 0.2 M phosphate buffer, pH 7.0. Fractions with Amnm/Amm values larger than 0.83 were pooled and concentrated as above. The enzyme fraction was pure as judged by sodium dodecyl sulfate-polyacrylamide electrophoresis, and the rechromatography on an ionic exchange column, which was used in the original procedure to remove purple acid phosphatase contamination (Davis and Averill, 1981), was unnecessary. The A u o -/ A~~, , , value of the preparations used in this study was 0.84. Myeloperoxidase was prepared from out-dated leukophoretic preparations (supplied from the Penn-Jersey Blood Program of the American Red Cross, Philadelphia, PA) as described previously (Ikeda-Saito and Prince, 1985).
Optical spectra were recorded with a HITACHI 557 spectrophotometer, which was interfaced to an IBM Personal Computer for storing optical data. All the measurements were performed at 20 "C unless specified. Ligand binding measurements were performed by addition of a known amount of ligand solutions (sodium chloride or potassium cyanide) into the enzyme solution followed by optical measurements around the Soret peak. Fractional saturation was calculated from the absorbance difference induced by ligand binding. Ligand affinity was expressed in terms of dissociation constant. All the ligand equilibrium data in the present study followed n = 1 titration curve, and the equilibrium constants, Kam, were obtained by analyzing the experimental data using the equation, Y = Kapp/(Kqp , where Y and [X] are fractional saturation and the ligand concentration, respectively, by least squares fitting. All the calculations were done on an IBM Personal Computer. EPR spectra were recorded by a Varian E-109 spectrometer operating at 9.326 GHz with 100-kHz field modulation. Measurements were carried out at 10 K by using an Air-Products liquid helium flow cryostat. Microwave frequency was measured with an EIP model 545 frequency counter, and magnetic field was calibrated with various standards of known g-values. The rate of hypochlorous acid formation was measured by following the chlorination of monochlorodimedon, using the technique described by Hager et al. (1966) for chloroperoxidase, that has been used for myeloperoxidase . The concentration of hydrogen peroxide was determined by the methods of Cotton and Dunford (1973). Monochlorodimedon was purchased from Sigma. Chlorination of monochlorodimedon was spectrophotometrically monitored at 290 nm at 20 "C. The experimental conditions were chosen to be close to those used for myeloperoxidase by  so that the results on the spleen green hemeprotein can be easily compared with those of myeloperoxidase.
Buffers used were 0.2 M citrate-phosphate, pH 3-6,0.2 M potassium phosphate, pH 6-8. Fig. 1 illustrates the optical spectra of the spleen green hemeprotein and its chloride and cyanide complexes. The resting state exhibits major peaks at 624,569, and 428 (Soret) nm with small peaks around 686 and 500 nm. These did not change between pH 3 and 8. Addition of chloride slightly changes the spectrum; the chloride complex shows the Soret peak at 434 nm and major peaks in the visible region at 623 and 574 nm with small peaks at 500 and 687 nm. Upon binding of cyanide, a new spectrum appears with peaks at 633 and 454 (Soret) nm. These spectral shapes, and the position of the peaks of the resting and chloride and cyanide complexes, agree well with those reported for myeloperoxidase (Agner, 1941;Wever and Plat, 1981;Bolscher and Wever, 1984a). Although the features of the optical spectra of ferric cyanide, ferrous, ferrous cyanide, and ferrous carbon monoxide forms of the present preparation of the spleen green hemeprotein agree very well with those reported by Davis and Averill (1981), there is a discrepancy in the peak positions of the resting ferric enzyme. The Soret and major peaks in the visible region of their resting enzyme were reported as 434 and 574 nm, which correspond to the peaks of the chloride complex in the present study.

RESULTS
The chloride affinity of the spleen green hemeprotein was determined between pH 3 and 9 at 20 "C. An isosbestic point was observed at 432 nm during the titration of the enzyme with chloride, and the chloride equilibrium curves were expressed by n = 1 titration curves. These indicate that there is only one binding process. Fig. 2 (dataset A ) plots the apparent dissociation constant for the chloride complex between pH 3 and 7 at 20 "C. Between pH 7 and 5, the slope of the pH dependence of the dissociation constant was unity, showing an involvement of a proton in the reaction of the enzyme with chloride. Such a pH dependence can be expressed by assuming that either a protonated form of the enzyme or the protonated form of the ligand undergoes the reaction. Since the pK value of HCl does not fall in the pH range of interest, it can be concluded that the pK must be on the enzyme and that only the protonated form of the enzyme binds chloride. The follow- Symbols are experimental data points, and the line for data set A was drawn by Equation 1 with pK. of 3.95 and KC, of 60 PM, and that for data set E was drawn using Equation 2 with pK. of 3.9 and KCN of 0.6 p M .
ing scheme can be used to express such a reaction: SCHEME A where E , K,, and KC, are enzyme, dissociation constant of an ionizable group of the enzyme, and dissociation constant for the enzyme-chloride complex. Then, the apparent dissociation constant, Kapp, for this chloride binding can be written as (1) The experimental data were best fitted with the values of pKa = 3.95, and KC, = 60 p~, as seen in Fig. 2 (datu set A ) . The temperature dependence of chloride affinity was also determined at pH 4.1 between 8 and 30 "C. The plot of the logarithm of association constants against the inverse absolute temperature was linear, and the apparent heat of chloride binding was estimated to be about -7.8 kcal/mol.
The cyanide affinity of the spleen green hemeprotein was studied between pH 8.3 and 3 (Fig. 2, data set B ) . An isosbestic point was observed at 442 nm during the titration of the enzyme with cyanide. Above pH 8.5, the cyanide complex of the enzyme was unstable and exhibited a time-dependent change in its absorption spectrum. This change seemed to correspond to the decomposition of the cyanide complex to the resting enzyme, and it was not feasible to determine the dissociation constants above pH 8.5. The equilibrium curves were described as n = 1 titration curves as was the case for chloride binding. In contrast to the chloride complex formation, cyanide affinity is essentially independent of pH between pH 5 and 8.3, but declines below pH 5. The pH independence of the apparent cyanide affinity on pH between pH 8.3 and 5 is indicative of binding of the protonated form of cyanide (HCN) to the enzyme, since the pK, of HCN is known to be around 9.3 at 20 "C. The decrease in the apparent affinity below pH 5 indicates that cyanide binds only to the unprotonated form of the enzyme. The following scheme can express the enzyme cyanide reaction.
where KCN is a dissociation constant for the enzyme-cyanide complex, and other symbols hold the same meaning as in Scheme A. Then, the apparent dissociation constant, Kepp, is expressed as where K, is an ionization constant of proton dissociable group of the enzyme and KCN is a dissociation constant for the enzyme cyanide complex. The pKa value of 3.95 and KCN of 0.6 p~ fit the experimental data, as seen in Fig. 2 (data set  B ) . Although the enzyme exhibits the opposite pH effect on the cyanide and chloride affinities, it seems likely that the same ionizable group with a pK value of 4 may influence both the chloride and cyanide ligand affinities. An addition of a large excess of chloride to the cyanide complex of the enzyme in the low pH region changes the absorption spectrum of the cyanide form to the chloride form, as reported for myeloperoxidase (Zgliczynski and Stelmaszynska, 1979). This indicates that the enzyme-bound cyanide is released from the enzyme when chloride binds to the enzyme. The effect of cyanide concentration on the chloride affinity of the enzyme was determined at pH 4.6 (Fig. 3). The equilibrium curves were expressed by n = 1 Hill plots, and the optical absorption spectra during the chloride titration gave clear isosbestic points. These indicate that cyanide, chloride, and the enzyme are in a rapid equilibrium. Since equilibrium data show that cyanide and chloride bind only to the unprotonated and protonated form of the enzyme, respectively, chloride equilibrium in the presence of cyanide and cyanide equilibrium in the presence of chloride can be expressed by combination of the Schemes A and B (Scheme C). SCHEME C Then, the apparent chloride equilibrium constant is given as where symbols hold the same meaning as in Equations 1 and 2. The curve in Fig. 3 was drawn by using the values of pKa = 3.9, KCN = 0.6 p~, and KC, = 60 p~, as independently determined in chloride and cyanide titrations described above. The set of equilibrium constants fits the experimental data well. Cyanide competitively inhibits the binding of chloride to the enzyme.
I have also measured the chloride concentration dependence of the apparent cyanide equilibrium constants at pH 4.6 and 7.1 (Fig. 4). The Hill plots for cyanide equilibria also showed slopes values of unity, and clear isosbestic points were observed during the cyanide titration. Both at pH 4.6 and 7.1, an increase in chloride concentration decreases the cyanide affinity of the enzyme. From Scheme C the apparent cyanide dissociation constant is written as Fig. 4 also plots the curves calculated by using the same equilibrium constants as used above. Again, as is the case of the effect of cyanide on the chloride affinity, the equilibrium constants derived by the independent cyanide and chloride equilibria and Equation 4 can adequately explain the effect of chloride on the cyanide affinity of the enzyme. Chloride competitively inhibits cyanide binding to the enzyme. I have also studied the effect of cyanide and chloride on the EPR spectral properties of the enzyme. The spectrum A in Fig. 5 is an EPR spectrum of the resting enzyme (enzyme concentration was 150 FM in 0.1 M phosphate buffer, pH 8). This spectrum shows that the enzyme is in a mixture of high spin (g = 6.65, 5.28, and 1.97) and low spin (g = 2.55, 2.32, and 1.81) states at 10 K. Addition of 0.17 M chloride to the resting enzyme increased the rhombicity in the high spin EPR signal ( g = 6.81, 5.04, and 1.95), and the low spin signal observed in the resting state is diminished, as seen in spectrum B. Addition of 130 p~ KCN to this chloride complex reduces the intensity of the high spin signal and a new low spin signal appears at gl = 2.83, g, = 2.25, and g3 = 1.66 (spectrum C). Further addition of cyanide to a concentration of 1 mM eventually converts all the high spin EPR signal to the low spin one. Addition of cyanide to the resting enzyme also gave the low spin EPR spectrum identical to spectrum D. In our previous report on myeloperoxidase (Ikeda-Saito and Prince, 1985), where the EPR measurements were carried out in 0.1 M citrate-phosphate buffer, pH 4.0, the rhombicity of the high spin EPR signal was reduced upon chloride binding. Thus, I have also measured the EPR spectra of the green hemeprotein in 0.1 M citrate-phosphate buffer, pH 4, as well as those of myeloperoxidase of human granulocytes in 0.1 M phosphate buffer, pH 8.0. The general features of the EPR spectrum of the spleen green hemeprotein at pH 4 are the same as those at pH 8: a rhombic high spin signal and a low spin signal at g = 2.55, 2.32, and 1.81, and the low spin signal is absent in the chloride complex. This was also the case for myeloperoxidase at pH 8. In Table I, the low field g-values of the high spin EPR signal are listed together with those of myeloperoxidase. Both enzymes exhibit a change in their high spin EPR signal in the same manner upon chloride binding: a decrease in rhombicity at pH 4 and an increase in rhombicity at pH 7. Davis and Averill (1981) also reported the EPR spectrum of the spleen green hemeprotein. However, there is a discrepancy in the EPR spectrum of the resting enzyme. Their spectrum exhibited high spin EPR signal at g = 6.81, 4.99, and 1.95 without low spin signal. Comparison of their spectrum with those studied here reveals that their spectrum resembles that of the chloride complex rather than that of the resting state. Such a discrepancy is also seen in their optical spectra, as described above.
The optical absorption spectra of compounds I, 11, and I11 of myeloperoxidase have been reported (Odajima and Yamazaki, 1970;Harrison et al., 1980, Bolscher andWever, 1984a). Compound I of myeloperoxidase spontaneously decays to compound I1 with a half-time of about 0.1 s (Harrison et al., 1980), and it is not feasible to record its optical spectrum with a conventional spectrophotometer. Compounds I1 and 111, however, are known to be stable enough to record their spectra. Great similarities in the spectroscopic properties of ferrous and ferric derivatives of the spleen green hemeprotein and those of myeloperoxidase have prompted me to compare the absorption spectra of compounds I1 and I11 of the spleen enzyme with those of myeloperoxidase. Fig. 6 shows the light absorption spectra of the spleen green hemeprotein (8 p~) in the absence (spectrum A ) and presence (spectrum B) of a 50fold excess and a 150-fold excess (spectrum C) of hydrogen peroxide at pH 7.0, 5 "C. These ratios are the same as those used for formation of compounds I1 and I11 of myeloperoxidase (Bolscher and Wever, 1984a;Odajima and Yamazaki, 1970).

Spectrum B shows absorption bands at 626 and 454 (Soret)
nm, and spectrum C has absorption bands at 624, 570, and 449 (Soret) nm. As expected from the similarities in the optical properties between these two enzymes, spectra B and C are quite similar to those of compounds I1 and I11 of myeloperoxidase, respectively. Spectrum B was instantaneously changed back to spectrum A by reaction with stoichi-  MPO in calf + 6.78,

5.00
? leukocytes + denotes the presence of the low spin EPR signal at g = 2.55, This paper. Davis and Averill (1981).
2.32, and 1.81, anddenotes the absence of a low spin signal. ometric amount of ascorbic acid, as reported for compound I1 of myeloperoxidase (Bolscher et al., 1984b). When the spleen enzyme (8 p M ) was titrated with hydrogen peroxide to 0.5 mM, spectrum A was converted to spectrum B with a set of isosbestic points at 666, 586, 497, and 442 nm. Further titration with hydrogen peroxide to 2 mM converted spectrum B to spectrum C, with a set of isosbestic points at 636,604,574, 514, and 446 nm. Spectrum C was also formed by oxygenation of the reduced enzyme. Such features are analogous to the observations reported by Odajima and Yamazaki (1970) for the formation of compounds I1 and I11 of myeloperoxidase.
Addition of 8 p~ ferrocyanide to the sample of spectrum B slowly converted the spectrum to that of the resting enzyme (spectrum A ) . When 16 p~ ferrous cytochrome c was added to the sample of spectrum B, the absorbance at 550 nm decreased slowly due to the oxidation of cytochrome c, and spectrum B was converted to spectrum A. The absorbance change at 550 nm after correcting for the absorbance difference between spectra A and B corresponded to the oxidation of about 8 p~ cytochrome c, indicating that 1 mol of ferrous cytochrome c was oxidized, and 1 mol of the enzyme was reduced back to the resting state. The stable compound of spectrum B thus has one oxidizing equivalent more than the resting enzyme. The light absorption spectra B and C of Fig.   6 are thus assigned to those of compounds I1 and I11 of the spleen green hemeprotein.
Monochlorodimedon has been used to study the chlorinating activity of myeloperoxidase (Harrison and Schultz, 1976;Bakkenist et al., 1980), and Fig. 7 examines this reaction for the spleen green hemeprotein. Incubation of the spleen enzyme with hydrogen peroxide, chloride, and monochlorodimedon results in a decrease in absorbance at 290 nm, due to the formation of dichlorodimedon (Hager et al., 1966) as observed for myeloperoxidase . Fig. 7A shows the initial rate of this reaction at pH 4.5 with 0.1 M chloride at different hydrogen peroxide concentrations. The rate depends on the sequence of adding substrates and enzyme to the cuvette. If the reaction was started by the addition of chloride to a cuvette that had contained hydrogen peroxide and the enzyme for 30 s (solid circles of Fig. 7 A 1, the rate was slower than that obtained by starting the reaction by either hydrogen peroxide or the enzyme (open circles of Fig. 7A). rate of the chlorination reaction was proportional to the This indicates that hydrogen peroxide in the absence of amount of enzyme present and independent of monochlorochloride inactivates the enzyme as reported for myeloperoxi-dimedon concentration from 5 to 50 p~. Fig. 7B illustrates &se . Reaction ofthe the Lineweaver-Burk plots' of the chlorination reaction at enzyme with excess hydrogen peroxide in the absence of Various NaCl concentrations at pH 4.5. The reactions were chloride rapidly converts the enzyme into compound 11. Compound 'I Of myeloperoxidase does not function in hYPochlorate of the reaction, so that the reaction rates can be directly compared In the plots, "absorbance change at 290 nm/min" was used as the also seems to be the case for the spleen green peroxidase. The axis unit.
acid formation (Harrison and Schultz, 19761, and this with those reported by Bakkenist et al., (1980) who used the same started by addition of enzyme into a cuvette containing monochlorodimedonation, hydrogen peroxide, and chloride. As is the case for myeloperoxidase, it is evident that chloride is not only a substrate for the spleen green peroxidase, but also behaves as a competitive inhibitor with respect to hydrogen peroxide. The chlorination rates are about 80% of those catalyzed by myeloperoxidase. The effects of hydrogen peroxide and chloride concentrations on the chlorination rates for the spleen green hemeprotein are analogous to those for myeloperoxidase , and Fig. 7B and the Lineweaver-Burk plots reported for myeloperoxidase (Fig. 6 of Bakkenist et al., 1980) are similar. The procedure used by Bakkenist et al. (1980) yielded an estimated Michaelis constant for hydrogen peroxide of approximately 0.05 2 0.02 mM at zero chloride concentration for the spleen enzyme, which seems to be in agreement with 0.11 mM reported for myeloperoxidase under the same conditions .
It is concluded that the spleen green hemeprotein catalyzes the formation of hypochlorous acid from hydrogen peroxide and chloride ion as myeloperoxidase.

DISCUSSION
The present study confirms the original report of Davis and Averill (1981) in two aspects: (i) the presence of a green peroxidase in spleen with optical spectral properties similar to those of myeloperoxidase of polymorphonuclear leukocytes, and (ii) the effectiveness of the acid extraction method and the batch absorption to and elution from cellulose phosphate. In this paper, I present an improved method of preparation and a further characterization of the enzyme, together with a detailed comparison of its ligand binding character with that of myeloperoxidase.
The major modification in the preparative methods is the use of CM-Sepharose CL-GB column chromatography with a pH and ionic strength gradient, in place of the ionic strength gradient at pH 5.0 on a column of CM52 at the first column chromatographic step in the original report of Davis and Averill (1981). The effectiveness of the present chromatographic procedure over the previous one can be shown by comparison of a conventional purity index (A430nm/A280 nm value) at this preparative step. In my experiments, the highest A430nm/A280nm value of the enzyme fractions from the CM-Sepharose CL-GB column reached about 0.75, which was about 2-fold better than the value reported previously (calculated from Fig. l of David and Averill, 1981). The complete removal of purple acid phosphatase from the enzyme fraction was readily achieved at this step. This facilitated the preparative method by eliminating the awkward CM52 column sectioning, which had to be used at the final step of purification to remove purple acid phosphatase in the original procedure. The enzyme preparation showed the same high purity on electrophoresis as that reported by Davis and Averill (1981). The optical spectra of the ferric cyanide and ferrous derivatives of the present preparation are identical to those reported by them. The only serious discrepancy exists in the spectroscopic properties of the resting enzyme. The optical and EPR spectra of the resting enzyme reported by them agree well with those of the chloride complex but not with those of the resting state of the present preparation. Since the solution conditions, such as buffer and pH, were not given in their paper (Davis and Averill, 1981), it is not certain whether their buffer contained chloride or not, but it is conceivable that their preparation contained chloride because 2.0 M KC1 was used to get enzyme out of the carboxymethylcellulose in the final step of purification without any state-ment about a further procedure to remove chloride from the enzyme.
Chloride binding to myeloperoxidase has been studied by several groups. There has been a conflict on the spectroscopically determined chloride affinity of myeloperoxidase. For example, the apparent dissociation constants at pH 4.5 were reported as 25 mM (calculated from Fig. 6 of Stelmaszynska and Zgliczynski, 1974), 0.3 mM by , 0.4 mM by Bakkenist et al. (1980), and 0.6 mM by Bolscher and Wever (1984a). Bakkenist et al. (1980) measured the pH dependence of the halogen affinity of myeloperoxidase, and found a linear relationship between the logarithm of equilibrium constant and pH over the pH range of 7 to 3 for chloride binding to myeloperoxidase. Recently Bolscher and Wever (1984a) have extended these experiments and reported that the protonated form of myeloperoxidase binds chloride and that the pK, of this protonation is about 4. The present data on the spleen green hemeprotein seem to agree with this: Kc, is 60 PM and pK, is 4 for the green hemeprotein, and they were 170 PM and 4.1 for myeloperoxidase (Bolscher and Wever, 1984a). The cyanide dissociation constant for the spleen green hemeprotein in the neutral pH region was 0.6 PM, which agrees with the equilibrium constants of 0.3 and 0.43 PM reported for myeloperoxidase by Odajima (1980) and Bolscher and Wever (1984a). The lack of an effect of pH on cyanide dissociation constants around neutral pH has been reported for horseradish, intestinal, myelo-, and chloroperoxidases (Makino and Yamazaki, 1973;Kimura and Yamazaki, 1978;Bolscher and Wever, 1984a;Dawson et al., 1984), and the neutral (protonated) form of cyanide has been considered to bind to these peroxidases. This is also the case for the spleen green hemeprotein because the affinity of cyanide of the spleen green hemeprotein did not vary in the range of pH from 8.3 to 5. The present results also show that cyanide and chloride competitively bind to the spleen green hemeprotein. Analogous observations were reported for myeloperoxidase from the chloride concentration dependence of the reaction velocity of cyanide (Bolscher and Wever, 1984a). It can be concluded that the spleen green hemeprotein possesses ligand binding properties very similar to those of myeloperoxidase. Davis and Averill (1981) showed that the optical spectra of the spleen green hemeprotein and its various complexes were very similar to those of myeloperoxidase complexes. By including the optical spectra of the chloride complex, and compounds I1 and 111, the present results further confirm this. I have also extended the comparison to include the EPR spectral properties of the chloride and cyanide complexes of the enzyme. The EPR spectra of the enzyme and its chloride complexes are similar to those of myeloperoxidase (Ikeda-Saito and Prince, 1985). The cyanide complex shows an EPR spectrum essentially the same as that of myeloperoxidase cyanide (Bolscher et al., 1984). The great resemblance in the EPR properties of myeloperoxidase and the spleen green hemeprotein shows that the electronic structure of these iron centers are similar.
One might claim that the present cyanide binding data on the spleen green hemeprotein differ from those in the earlier report for myeloperoxidase (Eglinton et al., 1982). The latter authors claimed that there are at least two cyanide binding processes exhibiting different ligand affinity and spectral changes, resulting in no true isosbestic points during the cyanide titration of myeloperoxidase as monitored by optical and magnetic circular dichroism spectroscopy. Bolscher et al. (1984a) pointed out that such a complication was due to contamination of eosinophil peroxidase in the preparation of myeloperoxidase used by Eglinton et al. (1982). The EPR spectrum of the cyanide complex of the spleen green hemeprotein exhibits a symmetrical g, signal, as seen in the spectrum of myeloperoxidase-cyanide reported by Bolscher et al. (1984a), but differs from that reported by Eglinton et al. (1982), which exhibits a shoulder at the lower fields side of the g, signal. Bolscher and Wever (1984a) reported the presence of a single set of isosbestic points in the spectra of myeloperoxidase and its cyanide complex. The present data on the cyanide complex of the spleen green hemeprotein agree well with these recent data on the cyanide complex of myeloperoxidase.
The thermodynamic properties of chloride binding to the spleen green hemeprotein could be compared with those of ligand binding to other hemeproteins with a similar coordination structure. Resonance Raman data (Babcock et al., 1985) indicated that the resting enzyme is a six-coordinated high spin state with a water molecule as the sixth axial ligand. The presence of a proximal histidine residue in myeloperoxidase has been suggested (Ikeda-Saito et al., 1984;Bolscher and Wever, 1984b). The resonance Raman, EPR, and magnetic circular dichroism spectra of myeloperoxidase and the spleen green hemeprotein indicate that the coordination structure of the iron centers in these two enzymes are similar, and the presence of a proximal histidine residue in the spleen enzyme, as in myeloperoxidase, is reasonably assumed. The spleen enzyme seems to have a water-iron-histidine axial coordination structure, as methemoglobin. The heat of chloride binding to the enzyme falls in the range of values reported for ligand binding to methemoglobin. For example, the values of heat of fluoride and thiocyanate ligation to methemoglobin were -2.9 and -7.5 kcal/mol, respectively (Anusium et al., 1968). This might suggest that chloride binds to the iron in the enzyme as fluoride and thiocyanate bind to the heme iron in methemoglobin, by replacing a bound water molecule.
Cyanide is a common low spin exogenous ligand for a variety of hemeproteins, and there is no doubt that its binding site is the sixth coordination position of the heme iron. Although chloride is known to bind at the axial coordination position of the iron in hemin-chloride, chloride is not a usual heme ligand in hemeproteins. Myeloperoxidase and chloroperoxidase belong to a small clan which exhibits chloride induced changes in their optical absorption spectra. Changes in the optical, EPR, and resonance Raman spectroscopic properties have been considered as evidence supporting a proposal of coordination of chloride at the sixth coordination position of the iron in myeloperoxidase. In chloroperoxidase, NMR (Krejcarek et al., 1976) and Mossbauer (Champion et al., 1973) studies concluded that chloride does not coordinate to the heme iron, although recent ligand equilibrium measurements have suggested that it does (Dawson et d., 1984).
The data presented in this paper clearly show that cyanide and chloride compete in binding to the spleen green hemeprotein. The steady-state enzyme kinetics study has also shown that chloride behaves as a competitive inhibitor with respect to hydrogen peroxide. The straightforward interpretation of the competition between chloride and typical heme ligands, such as cyanide and hydrogen peroxide, might be the direct competition at the same binding site. Such interpretations have been made for myeloperoxidase and chloroperoxidase (Zgliczynski and Stelmaszynska, 1979;Wever and Bakkenist, 1980;Bolscher and Wever, 1984a;Dawson et al., 1984). It should be borne in mind, however, that the competition in binding by chloride and cyanide does not necessarily imply competitive binding to the same binding site on the enzyme. It is still possible that the inhibitory action of chloride against cyanide or hydrogen peroxide binding to the enzyme may reflect the binding of chloride at a different site which secondarily interacts with the heme ligand binding site in an anticooperative manner. Such a possibility might seem remote when considering the large body of data favoring direct binding of chloride to the heme iron, but more structural and spectroscopic characterization of the chloride complex are necessary to definitely determine the chloride-binding site in myeloperoxidase and the spleen green hemeprotein.
The present enzyme kinetics results suggest similarities in the catalytic properties of the spleen green hemeprotein and myeloperoxidase. However, Davis and Averill (1981) reported that the spleen enzyme could not catalyze the peroxidation of ascorbic acid, which is a good substrate of myeloperoxidase (Agner, 1941), and described the possible difference in the structure of the substrate-binding sites between the two peroxidases. Ascorbic acid is known as a good reducing agent for compound I1 of various peroxidases including myeloperoxidase (Bolscher et al., 1984b): compound I1 is reduced to the native enzyme upon reaction with ascorbic acid. In my experiments, rapid conversion of compound I1 to the resting state by ascorbic acid was also observed for the spleen green hemeprotein, indicating that the spleen enzyme can utilize ascorbic acid as a substrate. A preliminary test of the steadystate kinetics shows that peroxidation of ascorbic acid is catalyzed by the spleen enzyme, in contrast to the original report of Davis and Averill (1981). Although the origin of this discrepancy is not clear, the present results are indicative of the similarities in catalytic properties between the spleen green hemeprotein and myeloperoxidase.
Together with the results obtained by resonance Raman and magnetic circular dichroism, the present results strongly indicate that the structure of the prosthetic group, and its vicinity, in the spleen green hemeprotein is identical to that of myeloperoxidase. It is concluded that the spleen enzyme can be used as a model for the study of the active center of myeloperoxidase of granulocytes, although the relationship between the spleen green hemeprotein and myeloperoxidase is not clear as pointed out by Davis and Averill (1981). There are several practical advantages to the use of the spleen green hemeprotein over myeloperoxidase. The source is easily obtained at the local abattoir in large quantity, preparation is simple, and the smaller molecular size may be of great advantage in the eventual determination of the molecular structure.