On the Molecular Mechanism of Lactoperoxidase-catalyzed H202 Metabolism and Irreversible Enzyme Inactivation*

Lactoperoxidase-catalyzed HzOz metabolism pro- ceeds through one of three different pathways, depending on the nature and the concentration of the second substrate as an e- donor and/or on pH conditions. In the lactoperoxidase (LP0)-Hz02 system, at low HzOz concentrations and/or alkaline conditions the peroxidatic cycle involves ferric LPO -i compound I * com- pound I1 4 ferric LPO conversion, whereas high HzOz concentrations and/or acidic conditions favor the ferric LPO -P compound I -+ compound I1 * compound I11 + ferrous LPO -* ferric LPO pathway. The compound III/ferroperoxidase states are associated with irreversible enzyme inactivation by cleavage of the heme moiety and liberation of iron. It is likely that either singlet oxygen or superoxide and hydroxyl radicals are involved in the attack on heme iron, because inactivation correlates with oxygen production and can be decreased to a certain degree by scavengers such as ethanol, 1-propanol, 2-propanol, or mannitol. In the LPO-H202-I- system, the enzyme may also be inacti- vated by Iz generated in the course of enzymatic I-oxidation (Le. during ferric LPO * compound I + ferric LPO cycles). or of 0 2 and spectral changes of LPO, were performed on an Aminco DW 2a TM UV-visible spectrophotometer equipped with a special cuvette mixer. In addition to spectral recordings, 0, concentrations could be measured simultaneously by a Clark electrode (39, 40) fitted to the cuvette from the side. Furthermore, the reaction vessel could be equipped with a thin glass pH electrode through the cover. Reactions were started by adding Hz02 with a syringe to the Nz-saturated medium through narrow slots in the lids of the spectrophotometer and cuvette. Statistical Calculations-Irreversible LPO inactivation as of were assessed by FX-501 P/FX-502 P program library.

Various suggestions have been made on the molecular mechanism and the structures of the catalytic inteymediate compounds I and I1 (reviewed in Refs. 1-4). Horseradish peroxidase is a well studied example for peroxidases because of its stability, its ease of isolation in large amounts, and its availability from commercial suppliers. Newer models for horseradish peroxidase compounds I and I1 are based either on experimental data (including spectra of electron nuclear * This work was supported by Swiss National Science Foundation Grant 3.987-0. 84. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C.
Section 1734 solely to indicate this fact.
$ To whom reprint requests should be addressed.
double resonance (5, 6), electron paramagnetic resonance, Mossbauer (7-lo), high-field proton nuclear magnetic resonance (11-13), resonance Raman (14-19), and x-ray absorption experiments (20, 21)) or on quantum mechanical calculations (including extended Hiickel calculations (22-25), unrestricted Hartree-Fock studies at the INDO (intermediate neglect of differential overlap) (26) and ab initio levels (27), and X a multiple scattering calculations (28)). Although the results of experiments and those of calculations on various model compounds do not always correlate, they all support a low-spin ferry1 moiety (Fe0") for compounds I and 11. The resting enzyme contains a penta-coordinated high-spin heme at neutral pH and a six-coordinated low-spin heme at alkaline pH (15). Compound I has been assigned almost unanimously with a 'AzU cation radical state (5-12) or a mixture of 2Azu and 'Al, electronic states (13). It has an oxyferryl center (Fe(IV)=O) with an axial symmetry. A significant mixing (spin coupling) between the FeO (S = 1) and the porphyrin spin ( S = 112) systems, however, complicates the pure 2A2u cation radical state (28). The iron-oxygen bonding of compound I1 is still under discussion. It has been suggested that this enzyme intermediate contains a Fe(IV)=O structure like compound I, but has no radical state any more (16, 18).
Alternatively, due to the uptake of a proton and an electron during the transformation from compound I (Reaction 21, compound 11 has also been proposed to contain a Fe(1V)-OH structure (29). The presence of an iron-oxygen single bond is supported by x-ray absorption studies, which show a longer iron-oxygen distance than found in compounds I and I11 (20,21). This data seems to be more reliable than the results from resonance Raman experiments which suffer from potential photoreduction on laser irractation at certain wavelengths (457.9 and 406.7 nm) resulting in a mixure of ferric and ferrous peroxidase (19, 30).
So far, compound I11 has been of less interest due to the fact that it does not take part in the peroxidatic cycle. Recently, however, it has been identified as an oxyheme adduct from its diamagnetism and its Mossbauer parameters (9), confirming its formation from native (ferric) peroxidase with superoxide anion O,, from reduced (ferrous) peroxidase with 02, or from compound 11 with excess H20, (reviewed in Ref. 1). X-ray absorption studies (21) have demonstrated an increase in charge density on iron of both compound I11 and the reduced enzyme, compared to a decrease in compounds I and 11, relative to the resting enzyme. This suggests a ferrous state for compound I11 and, therefore, favors the ferroperoxidase-oxygen assignment for compound I11 with the dioxygen not bound to the iron.
Work on lactoperoxidase (LPO,' a bovine milk peroxidase) performed in our laboratory so far (31-33) has revealed com- parable catalytic behaviors for LPO and horseradish peroxidase (reviewed for horseradish peroxidase in Ref. 1). In addition, similarities between LPO and thyroid peroxidase have been reported and ferriprotoporphyrin IX (protoheme IX) has been identified as a common prosthetic group to all three peroxidases (34-38). Consequently, we have used LPO as a model enzyme in all experiments. The link between the peroxidatic cycles in the LPO-H20z system (Reactions 1-3), peroxidase-catalyzed I-oxidation in the LPO-H,O,-I-system (Reactions 4 and 5), and irreversible enzyme inactivation in both systems (31-33) still remains to be demonstrated.
Iz + Hz02 ___* 21-+ 2H+ + 0 2 (5) The aim of this communication is to investigate the relationship between these pathways and to determine their consequences on LPO activity. The effects of H202, I-, enzyme concentration, and pH on intermediate compound formation, Ioxidation, and 0, generation are reported.
UV-uisible Spectroscopy of LPO-Spectral properties of LPO upon addition of HzOz or Na2S20r in 0.05 M phosphate buffer at a range of pH were monitored by recording UV-visible absorption in the Soret region (350-450 nm) on a UVIKON model 810 spectrophotometer combined with a UVIKON 21 recorder using serial overlay in 0.5-or 1-min intervals, or on a Hewlett-Packard 8451 A diode array spectrophotometer in 0.2-or 0.5-s cycles. Due to its molar absorption coefficient (cf. "Materials") LPO was used at a concentration of 3 pM.  In addition to spectral recordings, 0, concentrations could be measured simultaneously by a Clark electrode (39,40) fitted to the cuvette from the side. Furthermore, the reaction vessel could be equipped with a thin glass pH electrode through the cover. Reactions were started by adding Hz02 with a syringe to the Nz-saturated medium through narrow slots in the lids of the spectrophotometer and cuvette.
Statistical Calculations-Irreversible LPO inactivation as a function of excess Hz02 and the effects of Iaddition were assessed by regression analysis. Correlation coefficient r, intercept a, and slope b were calculated for linear (y = a + bx), exponential (y = aebr or In y = In a + bx), logarithmic (y = a + b In x), and potential (y = axb or In y = In a + b In x ) regression using a Casio FX-502 P programming calculator and the Casio FX-501 P/FX-502 P program library.

LPO-H,O,
System-Whereas photometric recordings of compound I require special techniques (e.g. stopped-flow etc.), compounds I1 and I11 can easily be detected by addition of a small excess of H,O, to LPO (3 phi) in a buffered medium.
Up to a concentration of 50 PM H20, compound I1 is the main detectable intermediate at pH 7.33 (Fig. lA). Beyond this limit of Hz02, compound 111 is generated, which has a lifetime ranging from a few minutes up to several hours depending on the excess of H,Oz (Fig. 1B). The effect of changing pH on compound 111 formation is discussed elsewhere (42). Absorption maxima (Xmax) of the different LPO states are listed in Table I.
Compound I1 is directly and completely reconverted to the resting ferric enzyme, as indicated by an isosbestic point at 421 nm on the descending limb of the original recording of native LPO (Fig. lA). In contrast, compound I11 has been shown to be metabolically involved in the reaction leading to irreversible inactivation of LPO (33). The loss of enzyme activity is indicated by the difference between original and final peak heights at 411 nm and by the position of the isosbestic point which no longer coincides with the descending limb of the original spectral recording of native LPO (Fig.  1B). Regression analysis has revealed that the recovery of LPO decreases with excess H202 (Fig. 2). Experimentally determined values closely fit to Equation 6 (correlation coefficient r = 0.9934), y = 100.87 e"0.272x or In y = In 100.87 -0.272~ (6) where x corresponds to the concentration of H202 (mM) and y to the percentage LPO recovery.
Compound 111 metabolism is characterized by a red shift (higher wavelengths) prior to reconstitution of the resting enzyme, suggesting the formation of an intermediate compound such as compound I1 (Xmax = 430 nm) or the reduced ferrous state (Xmax = 435 nm) ( Fig. 1, B and C). At the beginning of the red shift, when most of the H2O2 excess has been consumed, enzyme destruction ceases as suggested by the transient constant absorbance value. The remaining fraction of compound 111 is then reconverted to the resting ferric enzyme without any further loss, as indicated by the isosbestic point at 420 nm.
If only selected cycles during compound 111 conversion are plotted, the formation of an intermediate compound with Xmax between two isosbestic points at 430 and at approximately 447 nm can be demonstrated, suggesting the conversion of compound I11 to ferrous LPO rather than to compound I1 (Fig. IC). In addition, the isosbestic point at 430 nm is clearly different from the one formed at 426 nm during the conversion of compound I1 to compound 111. In the visible range ( Fig.  1D) the same assay demonstrates conversion of compound I11 to a state which absorbs at shorter wavelengths than the isosbestic point at 538 nm and the wavelength minimum around 560-570 nm. This intermediate compound is converted back to native LPO with an isosbestic point at 517 nm. Reduction of LPO from the ferric to the ferrous form ( Fig.  3 (1, 41)) by addition of a few grains (approximately 3 mg) of Na2S204 results in a labile compound with X  is very short lived and is rapidly converted into ferrous enzyme and resting LPO (Fig. 3B, cycle 1). Thereafter, ferric LPO steadily decreases in favor of the reduced enzyme and compound I11 which reach a maximum at cycle 17. Reconversion of the two compounds to the resting ferric LPO is terminated prisingly compound I1 is produced first (Fig. 3B, cycle 0.5). at cycle 90 (Fig. 3C). Recovery (89%) is, unexpectedly, lower This is due to a catalytic interaction of LPO with H202 than that obtained by substitution of 260 @ I H202 for dithioriginating from a reaction between O2 contained in the air-onite (94%, Fig. 2 Assuming a catalatic degradation of Hz02 (33,43), 1 mol of O2 should be produced from 2 mol of H202 consumed. This is confirmed by measurements of H202 consumption and O2 generation (Fig. 4). The results correspond very closely to the theoretical stoichiometric ratio up to a concentration of about 500 ~L M H202. Beyond this limit some O2 is lost probably due to oversaturation of the medium and generation of 0, bubbles

(39, 40).
H,Oz consumption (Fig. 4, top) and O2 production (Fig. 4, bottom) consist of two phases. The initial fast a-phase occurs before compound I11 formation, i.e. duringcompound I1 cycles, while the slow P-phase corresponds to the compound I11 state (first arrows on each curve of Fig. 4 indicate the maximal amount of compound I11 formed). O2 production is terminated soon after the beginning of the red shift and during the reconversion of ferroperoxidase to the resting ferric enzyme (second and third arrows, respectively, on each curve of Fig.  4).

LPO-recovery (%)
'O01 is not subsequently consumed in the reactions discussed in Fig. 6.

0-
Increasing LPO from 1 to 2 nM at fixed Hz02 (500 p M ) and 1-(10 mM) concentrations (Fig. 6A) accelerates the reaction rates of Ioxidation and 0 2 generation. Simultaneously, net production of I,/& increases, but no change in the final O2 concentration can be observed.
When LPO (2 nM) and 1-(10 mM) are kept constant and H20, is raised from 100 I.LM up to 500 ~L M (Fig. 6B), 12  This in turn leads to a higher nonproductive H202 consumption and Oz generation due to the now prevailing 12 reduction, whereas the amount of is decreased as a result of this 12 withdrawal.
If LPO (2 nM) and HzO, (500 p~) are kept constant and the concentration of Iis raised from 1 mM up to 100 m M ( Fig. 6C), both the rate and the amount of production increase. 1, reduction is suppressed at high Iconcentrations due to a shift in the equilibrium between I, and I, ((see "Experimental Procedures") (31)). Therefore, rate and amount of 0, generation are significantly diminished. Ioxidation and 0, production at fixed amounts of LPO (2 nM), H,02 (300 p~) , and I-(10 mM) are strongly influenced by changes of pH (Fig. 6D). Whereas net I& generation is drastically reduced upon an increase of pH from 6.6 to 7.6, amount and rate of 0, generation are favored at alkaline pH conditions according to Reactions 4 and 5 . No enzymatic Ioxidation occurs upon addition of 50 pM H202 to a nonbuffered solution containing LPO (1 nM) and I-(10 mM) below pH 4.0 and above pH 10.2. Between these limits of pH, catalysis begins at rates which are comparable to those in buffered media, but the reaction breaks down independent of the H,02 concentration after having produced a certain amount of OH-. The pH at which the reaction breaks down depends on the concentration of I-. As Iincreases, so does the amount of OHformed, resulting in a shift of pH as predicted by Reactions 4 and 5 and by the shift in the equilibrium between I, and 1 : (see "Experimental Proce-dures"). It is possible to restart the reaction by a new portion of HZ02 after titration of the incubation mixture back to pH 7.0. Depending on the concentration of Hz02 this reversible and the irreversible inactivation may overlap as suggested by a lower pH value at the second breakdown (results not shown).

DISCUSSION
Peroxidation of Hz02 by LPO can occur through one of three cycles depending on the nature of the second substrate as an e-donor (Reactions 1-3, Fig. 7 Fig. 1, B-D; Fig. 2). Under alkaline conditions compound I11 can barely be detected, whereas at an acidic pH this state is more stable and detectable.
Compound I formation, i.e. the initial reaction of LPO with HzOz, includes an acid-base catalysis of amino acid residues close to the active site on the enzyme (e.g. Asp-43 and Arg-38 in the case of horseradish peroxidase (44)), with the release of OH-and an e-transfer from the porphyrin ring to iron (44)(45)(46)(47). Restitution of the resting ferric enzyme can be achieved by either a single 2e-transfer (e.g. from I-(48)) or by two le-transfers via compound I1 (Reactions 1-3). In the absence of I-and at low H202 concentrations compound I1 is the main detectable intermediate, suggesting a le-transfer from compound I to compound I1 (Fig. IA ). The e-could either be donated from the apoprotein or from H,O,. In the latter case superoxide radicals (HO; ) are generated as intermediates of the first le-transfer and 0, as final product (Fig.  7). Apart from 02, H30+ is also released on the reconversion of compound I1 to LPO (29), originating from an oxygen atom and a H+ from the first step of the cycle plus two H+ from the pathway compound I + compound I1 + LPO. In this way, OH-from the initial phase is neutralized, and thus the same reaction performed in water instead of buffer solution does not break down as occurs in the presence of I-. These reactions are in line with mechanisms already proposed in the literature (21,29,46,49). ferrous state in the spectrum before it is reconverted to the resting ferric enzyme (Fig. 1, C and D ) . These pathways are in line with the observed stoichiometry of one O2 generated per two H202 consumed (Fig. 4, taking into account the loss of O2 bubbles when the solution is saturated).
Irreversible LPO inactivation depends on the presence of excess H202 (Figs. lB, 2, and 6B). In the absence of I-, this occurs during compound 111 formation and before establishment of isdsbestic points, which indicate conversion to another state without any further loss of enzyme ( Fig. 1, B-D (33)). LPO is not able to escape from inactivation by suppIy of superoxide dismutase, but reconversion of compound III/ ferroperoxidase to native LPO is gradually slowed down indicating that HO; is not the inactivating species (42). This correlates with the absence of inactivation during LPO * compound I -+ compound I1 + LPO cycles (see above and It is reasonable to compare irreversible inactivation of LPO, which includes cleavage of the heme moiety and liberation of iron (33), with the oxygenation of hemin at one of its amethene bridges and subsequent ring opening to form verdoglobin and biliverdin as primary steps in heme destruction. HO; itself does not attack the autoxidation-sensible porphyrin plane of hemin (50). However, HO; is involved in the Haber-Weiss reaction (Reaction 8, P K , ,~~~ = 4.8 (50)), which in fact consists of a chain mechanism with Fez+, OH', and HO; as chain carriers and in which OH', and not HO;, is capable of attacking H202 (Reactions 9-14 (44); Reaction 10 is the Fenton reaction).
The fact that LPO inactivation can either be decreased by alcohols or I- (Fig. 5 ) in a concentration-dependent manner or proceeds until the enzyme is fully destroyed as soon as of Lactoperoxidase by Excess H202 EDTA is added to the incubation medium (results not shown) supports the view that OH ' may be the most damaging species acting on LPO. On the other hand, I O 2 is able to oxidize many diamagnetic bio-organic molecules (33, [50][51][52]. Whereas O2 production in the LPO-H2OZ system can be attributed to the catalatic activity of the enzyme (33,43), 0 2 generation in the ',PO-H202-I-system occurs due to chemical 1, reduction ( Figs. 4 and 6). Although there is a positive correlation between O2 production and irreversible LPO inactivation (Figs.  4 and 6), the involvement of '02 in the destruction of heme cannot be proven. 0, measured in our systems (Fig. 6) may be generated as a result of enzyme inactivation but is not necessarily the cause of it. 1present in large amounts, or at least additional ' 0 2 quenchers such as 1-or 3-methylhistidine, is supposed to quench 'Ag 0, readily (52), mainly when it is produced in a purely chemical reaction remote from the active site of the enzyme ( i e . in the LPO-H202-I-system, Fig. 6, A-D ) . However, irreversible inactivation of LPO occurs nevertheless, except for cases where 0 2 production is low (e.g. in the "compound I1 way," Fig. 7).
Despite the positive correlation between O2 production and irreversible enzyme destruction, we have also to consider the possibility that LPO could be inactivated during enzymatic 1oxidation. Incubation of LPO (3 pM) in 1 mM aqueous 1, and removal of excessive I2 by lyophilization indeed reduces enzyme activity.' Excess H202 could well inactivate LPO via iodination of the apoprotein, although other residues may be iodinated in the case of enzymatic Ioxidation close to the active site and when incubated in an Iz-containing medium. H202 not used in the enzymatic reaction would be degraded by I& produced up till the breakdown of Ioxidation (Fig.   6). Protection against I2 can be achieved by either raising the concentration of I-, therefore withdrawing I2 due to a shift in the equilibrium between IZ and 1 ; (31), or by adding iodine acceptors such as tyrosine or thyroglobulin (32, 53). Depending on the conditions, irreversible enzyme inactivation caused by excess H202 alone or by iodination may even overlap.
The fact that irreversible LPO inactivation is also caused, even to a higher degree, by dithionite (Fig. 3, A-C) can possibly be assigned to the large amount of monomeric free radicals of SOT, which are generated by dissociation from S202used in large excesses (41). Thus, more work is required to test fully the proposed models. Experiments including spin trapping are in progress which are intended to reveal either I, or oxygen free radicals as damaging species on the peroxidase. The outcome of this work will provide some hints which may help to explain hitherto unsolved problems with special regard to the pathophysiology of the thyroid gland.