Chapter 1: Self-processing of Peroxidases

The peroxidases, like other families of hemoproteins, include members that undergo post-expression, autocatalytic processing to generate the mature proteins. Hemoproteins have a potential for autocatalytic self-modification reactions beyond that available to most other proteins. This potential stems from the inherent high reactivity of the Compound I and Compound II intermediates generated during the catalytic cycle, the fact that the substrate specificity of peroxidases is broad and often involves outer sphere electron transfer reactions, and the reactivity of the radicals that can be formed by peroxidase reactions. In many instances, these autocatalytic modification reactions are incidental and impair or terminate peroxidase function, but in other instances evolution has optimized such reactions to tailor the catalytic functions of proteins for specific purposes. This maturation process can result in modification of specific protein side chains, remodeling of the prosthetic heme group, or covalent crosslinking of the heme to the protein through one or more bonds.

This chapter focuses on the mechanisms and potential roles of autocatalytic reactions that: (a) modify the protein structure of peroxidases, (b) result in simultaneous modification of both the heme and protein, or (c) only impact the prosthetic heme group. It does not, however, deal with protein or heme modifications that require the intervention of secondary enzymes or the general modifications of the protein or heme that result in loss of peroxidase activity. Specifically, this chapter does not discuss post-translational modifications such as phosphorylation, the mechanisms of heme or protein degradation caused by excess H₂O₂, or, except for one exception, formation of the cysteine–vinyl link characteristic of cytochrome c. Nevertheless, it includes a discussion of site-specific peroxidase mutants that give rise to well-defined alterations in the protein or heme structure, as these help to understand the mechanisms and determinants of these reactions.

Lignin peroxidase (LiP), a fungal enzyme, promotes the oxidative degradation of lignin,¹  a complex polymeric structure that accounts for approximately 30% of the carbon in the biological sphere.²  LiP, like manganese peroxidase,³  catalyzes the one-electron oxidation of a diffusible mediator that can more easily penetrate the dense lignin matrix to promote its degradation. The diffusible mediator in the case of LiP is veratryl alcohol, which is oxidized to the veratryl radical cation. The radical cation can then oxidize lignin or other substrates, or can be further oxidized by removal of a second electron to the aldehyde (Scheme 1.1).

The crystal structure of LiP from Phanerochaete chrysosporium at a resolution of 1.7 Å revealed the unexpected presence of electron density that could be assigned to a hydroxyl group located on the C^β atom of Trp171.⁴  Biochemical and crystallographic evidence showed that this protein modification was the result of an autocatalytic process that occurred when the newly synthesized protein reacted with H₂O₂. Thus, a crystal structure of the naïve protein without prior exposure to H₂O₂ had an intact Trp171 without electron density for the extra hydroxyl group, whereas it was present after the protein was exposed to H₂O₂.^4,5  Tryptic digestion of the mature protein and isolation of the peptide containing Trp171 showed that it was modified in a manner consistent with introduction of a hydroxyl adjacent to the aromatic ring. This modification was only found in the peptide after reaction of LiP with H₂O₂.⁶  As molecular oxygen was not required for introduction of the Trp171 hydroxyl modification, it was postulated that the mechanism for its introduction involves two sequential one-electron oxidations, resulting in formation of an exocyclic conjugated imine. Michael addition of water to this intermediate produces the final modified amino acid (Scheme 1.2).⁶  Involvement of the normal catalytic mechanism in the oxidation of Trp171 is confirmed by the finding that when an alternative substrate is present, increased concentrations of H₂O₂ are required for full modification due to a competition between protein and substrate oxidation.⁶  The involvement of a Trp171 radical in the modification process is supported by spin trapping experiments with 2-methyl-2-nitrosopropane (MNP) in which the spin trap was shown by tryptic mapping to bind to the C6 atom of Trp171 and thus to prevent formation of the hydroxylated tryptophan residue.^4,7 

The Trp171Phe and Trp171Ser mutants in which Trp171 is replaced by redox-inactive residues no longer oxidize veratryl alcohol, a physiological substrate.⁸  However, both mutants form Compound I normally with H₂O₂ and retain most of their ability to oxidize conventional peroxidase substrates, including ABTS (2,2′-azinobis-(3-ethylbenzthiazoline-6-sulfonic acid) and DFAD (4-[(3,5-difluoro-4-hydroxyphenyl)azo]benzenesulfonic acid). Similar results were found with a nonphenolic tetrameric lignin model, in that the wild-type enzyme oxidized this substrate whereas the Trp171Ser mutant did not.⁹  In contrast, a Glu146Ser or Glu146Gly mutation in the channel leading from the surface to the heme edge did not prevent oxidation of veratryl alcohol or the tetrameric lignin model.^8,9  These results show that the oxidation of veratryl alcohol and lignin-like products is mediated at the Trp171 site, whereas more conventional peroxidase substrates appear to be oxidized preferentially at a second site, probably one involving the channel leading into the heme crevice. However, the specific role of Trp171 hydroxylation remains unclear. The finding that Trp171 hydroxylation in pristine LiP competes with veratryl alcohol oxidation suggests that the hydroxyl modification is not absolutely essential for veratryl alcohol oxidation,⁶  whereas the lack of activity of the Trp171Ser mutant towards a lignin model suggests Trp171 is critical for that function.

A different protein modification has been found for the LiP from Trametopsis cervina,^10,11  a protein that lacks the highly conserved tryptophan residue equivalent to Trp171 in the Phanerochaete chrysosporium enzyme. Crystallographic, kinetic, and spectroscopic data indicate that the oxidation of veratryl alcohol or polymeric substrates by this enzyme depends critically on Tyr181, a surface-exposed tyrosine. An EPR analysis indicates that Tyr181 is oxidized during the catalytic cycle to a radical that is proposed to mediate substrate oxidation.¹⁰  Subsequent work revealed that pristine LiP showed a lag period in veratryl alcohol oxidation, whereas the enzyme that had already been involved in veratryl alcohol oxidation did not.¹¹  Mass spectrometric analysis revealed that the mature enzyme had a veratryl alcohol moiety covalently attached to Tyr181. Although the exact nature of the covalent link remains unclear, it is likely to involve a crosslink between the tyrosine and veratryl alcohol aromatic rings. Exposure of the naïve enzyme to H₂O₂ in the absence of veratryl alcohol resulted in hydroxylation of Tyr181 to a catechol and loss of the ability to oxidize veratryl alcohol and 1,4-dimethoxybenzene (Scheme 1.3), although both the protein with the hydroxylated Tyr181 and that with the Tyr181-veratryl adduct were able to oxidize low potential substrates such as ferrocytochrome c. No covalent changes were observed with the Y181N mutant of the protein. Kinetic studies demonstrated that covalent binding of veratryl alcohol to Tyr181 stimulated oxidation of veratryl alcohol by Compound II by >100-fold.¹¹  Thus, formation of the covalently modified tyrosine in the case of the Trametopsis cervina enzyme has direct consequences on the catalytic performance of the enzyme.

The KatG catalase-peroxidases that are found in microorganisms have both a high catalase activity in the range of 5000–16 000 s⁻¹ and a much lower, but still significant, peroxidase activity of 10–25 s⁻¹.^12,13  Mechanistic studies have been carried out with several of these enzymes and crystal structures are available for the KatG proteins from Burkholderia pseudomallei,¹⁴  Haloarcula marismortui,¹⁵  Synechocystis PCC 7942,¹⁶  and Mycobacterium tuberculosis.¹⁷  The crystal structure of a KatG excreted from the eukaryotic fungus Magnaporthe grisea has also been reported.¹⁸  The most unusual feature of all the KatG enzymes revealed by these crystal structures is the presence of a highly conserved Met-Tyr-Trp cross-linked tripeptide in the distal (substrate) side of the heme pocket (Figure 1.1). The presence of this cross-linked tripeptide is confirmed by mass spectrometric data on tryptic peptides from the B. pseudomallei,¹⁹  Synechocystis,²⁰  M. tuberculosis,²¹  and Magnaporthe grisea proteins.²²  In the case of M. tuberculosis KatG, the residues in the tripeptide are Met²⁵⁵-Tyr²²⁹-Trp¹⁰⁷ (Figure 1.1). The tripeptide is located in the active site of the protein directly above the prosthetic heme iron atom.^17,21 

Autocatalytic formation of this tripeptide is clearly demonstrated by the fact that heterologous expression of Mycobacterium tuberculosis KatG in Escherichia coli yields a naïve protein in which the three relevant residues, as shown by tryptic digestion, mass spectrometry, and UV-vis spectrometry, are not cross-linked. However, incubation of the protein with six equivalents of H₂O₂ triggers a rapid process in which the cross-linked tripeptide is quantitatively formed.²¹  Expression of the Met255Ile mutant of M. tuberculosis KatG, in which the methionine of the cross-linked peptide is missing, gives rise to a protein with only the Tyr-Trp crosslink, as well as a two-electron oxidized form proposed to be an intermediate in the formation of the cross-linked dipeptide.²³  These results are consistent with earlier findings that mutants of B. pseudomallei KatG that lacked the appropriate tryptophan (Trp111) or tyrosine (Tyr238) did not form the fully cross-linked tripeptide,^19,20  but in addition specifically showed that its formation is due to an autocatalytic process. These results strongly support a crosslinking mechanism in which autocatalytic one-electron oxidation of both the Tyr and Trp aromatic side-chains produces the corresponding radicals that, in turn, undergo radical combination to form a carbon–carbon bond (Scheme 1.4). After proton tautomerizations to regenerate the two aromatic rings, two electrons are removed in a second catalytic turnover, producing an iminoquinone to which the methionine adds as a nucleophile in a Michael reaction. A final proton tautomerization step then yields the mature cross-linked tripeptide. This same mechanism is applicable to the formation of the cross-linked tripeptide in all the KatG enzymes.

In the absence of the intact cross-linked tripeptide, the catalase activity of the KatG enzymes is greatly attenuated (Table 1.1). In contrast, the peroxidase activity is modestly increased, presumably because the Compound I intermediate is not being drained as effectively by the much faster catalase reaction. The table shows that the cross-linked tripeptide is important for the catalase, but not peroxidase, function of the enzyme, although autocatalytic formation of the tripeptide depends on the peroxidase activity.

The exact role of the Met-Tyr-Trp tripeptide in KatG catalytic activity continues to be investigated. It has been proposed that reaction of KatG with H₂O₂ generates Compound II at the heme center together with the tripeptide radical. Reaction of this intermediate with a second molecule of H₂O₂ produces the ferrous dioxygen (or ferric superoxide, Compound III) complex, still with the tripeptide radical. Electron transfer from the dioxygen complex (nominally Fe⁺³O₂^•−) to the tripeptide radical then regenerates the starting enzyme and oxygen, effectively completing the conversion of H₂O₂ to molecular oxygen.^31–35  This mechanism is supported by detection of the ferrous dioxy heme/tripeptide radical intermediate and demonstration that this intermediate cycles to produce molecular oxygen.^32,33,36  The intact tripeptide may also help rescue catalase activity by reducing the catalase-inactive Compound II to the ferric state.²³  The detailed catalytic mechanism of KatG enzymes has been reviewed.³⁷ 

The mammalian peroxidases, as exemplified by lactoperoxidase (LPO), myeloperoxidase (MPO), eosinophil peroxidase (EPO), and thyroid peroxidase (TPO), are distinguished from the plant and fungal peroxidases by the presence of covalent bonds that attach the heme group to the protein. The crystal structures of caprine, buffalo, and bovine LPO^38,39  and of canine and human MPO (Figure 1.2)^40,41  show that the 1- and 5-methyls of the heme are attached through ester bonds to the carboxylic acid side-chains of Asp or Glu residues. The presence of the ester bonds is supported by the observation of appropriate carbonyl vibrations by difference Fourier transform infrared spectroscopy.⁴²  Their presence has also been confirmed in LPO by NMR and mass spectrometric studies of proteolytically generated peptides.⁴³  In caprine LPO, the 1-methyl forms an ester bond with Glu252 and the 5-methyl with Asp108, while in human MPO the corresponding bonds are to Glu242 and Asp94. In accord with these assignments, mutation of either Glu242 to a Gln or Asp94 to a Val in human MPO results in loss of some of the ester carbonyl resonance bands.⁴²  However, a recent crystallographic study of MPO isolated from human leukocytes showed low electron density for Glu242, suggesting that it had high mobility and therefore that the ester bond to this residue might have relatively low occupancy.⁴⁴  Nevertheless, electron density on the 1-methyl group was consistent with covalent attachment of a hydroxyl group to the methyl. It is therefore possible that under some conditions the Glu242 ester bond is only fully formed in a subpopulation of the MPO molecules.

No crystal structure of human eosinophil peroxidase is available, but proteolytic and mass spectrometric evidence has demonstrated the presence of covalent ester bonds between the heme 1- and 5-methyls and the carboxyl groups of Glu241 and Asp93, respectively.⁴⁵  In parallel with the suggestion that not all MPO isolated from human leukocytes had both heme–protein ester bonds, biochemical work with eosinophil peroxidase showed that up to 90% of the protein had only one ester bond when isolated. The missing ester bond was that between the 5-methyl and Asp93. Interestingly, a similar mass spectrometric analysis carried out with bovine LPO indicated that the ester links to both Asp125 and Glu275 were fully present in all the protein molecules.⁴⁵ 

In contrast to the direct crystallographic or biochemical evidence for LPO, MPO, and EPO, the presence of covalent ester links with the heme methyl groups in TPO rests on indirect evidence. When human thyroid peroxidase is expressed in Chinese hamster ovary (CHO) cells, only 2% of the protein reaches the cell surface.⁴⁶  The peroxidase activity at the cell surface due to this protein was greatly decreased in the presence of an inhibitor of heme synthesis, but was increased if heme precursors were added to the cells. Furthermore, incubation of the CHO cells with H₂O₂ resulted in a 65% increase in the surface peroxidase activity. These experiments indicated that heme incorporation and exposure to H₂O₂, as required for autocatalytic processing, increased the level of catalytically active thyroid peroxidase at the cell surface.

It is now clear that covalent heme binding is the result of an autocatalytic maturation process. The first evidence for this was the finding that when LPO was heterologously expressed in a baculovirus system and then purified, only a fraction of the protein molecules had a covalently bound heme group.⁴⁷  However, a high level of covalently bound heme was obtained when the freshly isolated protein was incubated with H₂O₂.^47,48  Hydrolysis and analysis of the covalently bound prosthetic group in this protein confirmed that two of its original heme methyl groups now bore hydroxyl groups. To confirm that crosslinking occurred with the same carboxylic acid groups as in the native protein, the Glu375Asp and Asp225Glu mutants of bovine LPO were expressed and purified.⁴⁸  The heme in the Glu375Asp mutant was only partially covalently bound, but incubation with H₂O₂ resulted in full covalent binding and full catalytic activity. Furthermore, analysis of the heme after digestion of the protein established that the heme 5-methyl formed the bond with Asp225. The Asp225Glu mutant, in contrast, had little covalently bound heme, low catalytic activity, and heme covalent binding did not significantly increase on incubation with H₂O₂. The double Glu375Asp/Asp225Glu mutant had no covalently bound heme and no catalytic activity. Although two covalent heme bonds are normally formed, these results demonstrate that the single Asp225 covalent bond is sufficient for high catalytic activity. An independent study of the Glu375Gln and Asp225Val mutants of recombinant LPO expressed in CHO cells agreed with the finding that the Asp225 ester bond is more critical than the Glu375 bond for catalytic activity.⁴⁹ 

Additional evidence for an autocatalytic process is provided by a study of human EPO, which showed that the heme in the freshly isolated protein has a crosslink between the heme 1-methyl and Glu241, but upon incubation with H₂O₂ forms a second covalent bond between Asp93 and the heme 5-methyl.⁴⁵  The methyl groups involved in these bonds were not specifically identified, but were inferred from a comparison to MPO. The EPO data clearly establishes that the bond to the 1-methyl is formed first, but does not require that formation of the 1-methyl bond must precede formation of the 5-methyl bond. These results are consistent with the suggestion from the crystallographic evidence that the bond between the 1-methyl and the protein in MPO is generally present, whereas that involving the 5-methyl may not always be.⁴⁴ 

Formation of the ester bonds to the heme methyl groups is best formulated as proceeding via the free radical mechanism outlined in Scheme 1.5. Compound I formed by the non-covalently bound heme–protein complex oxidizes a side-chain carboxylic acid to a carboxylate radical, concomitantly reducing the heme center to the Compound II state. In turn, the carboxylate radical abstracts a hydrogen atom from a heme methyl group, yielding a methylene radical and regenerating the carboxylic acid anion. Intramolecular transfer of the unpaired electron from the methylene to the iron produces the methylene cation with concomitant reduction of the iron to the ferric state. In the final step, the carboxylate anion traps the methyl cation to form the ester bond. Of course, this process must occur twice, with different carboxylic acid residues and different heme methyls, to forge the two covalent bonds.

Support for this mechanism is provided by a study in which a carboxyl group was introduced by site-specific mutagenesis near one of the heme methyl groups of horseradish peroxidase (HRP).⁵⁰  Thus, the heterologously expressed Phe41Glu mutant is purified with no heme covalent bonds, but upon incubation with H₂O₂ quantitatively forms a covalent ester bond between Glu41 and the nearby 3-methyl group of the heme. In contrast, incubation of the Ser73Glu mutant with H₂O₂ oxidized the prosthetic group to 8-hydroxymethylheme without actually forming a covalent crosslink, presumably because the carboxylic acid group could not compete effectively with a water molecule in the final step of the reaction (Scheme 1.5) in which the methylene cation is trapped. A demonstration that the hydroxyl group derives from water when ¹⁸O-labeled water is used strongly supports this mechanism. Formation of the covalent heme bond increased the catalytic activity of the Phe41Glu mutant 100-fold.⁵⁰ 

A study of the inhibition of MPO by various hydrazides revealed that benzoic acid hydrazide and 4-(trifluoromethyl)benzoic acid hydrazide inhibit this enzyme by hydrolyzing the ester bond between heme and Glu242 of the MPO heavy chain.⁵¹  This is the first instance in the literature of agents that inhibit peroxidatic activity by promoting hydrolysis of one of the protein–heme covalent ester bonds.

Phylogenetic analysis has led to the discovery of a class of bacterial peroxidases that resemble LPO in many ways, including the presence of at least one heme–protein covalent bond.⁵²  Cloning, heterologous expression in E. coli, and biophysical and catalytic characterization of one of these enzymes, that from Lyngbya sp PCC 8106, yielded a protein in which 20% of the heme was covalently bound. However, incubation with several equivalents of H₂O₂ produced an enzyme in which more than 70% of the heme was covalently attached to the protein and whose spectroscopic properties were nearly identical to those of LPO. Indirect evidence was obtained that the protein residue involved in the covalent bond was Glu299, although the additional participation of Asp100 in a similar bond could not be ruled out. The site of attachment to the heme was not determined, but based on the similarity of the UV-visible spectrum to that of LPO it was presumed to be one of the heme methyl groups. Differential scanning calorimetry demonstrated that the protein after treatment with H₂O₂, i.e., when the heme was mostly covalently bound, was thermally more stable than the protein without the heme–protein crosslink. Finally, the Lyngbya peroxidase after exposure to H₂O₂ was found to have a higher ability to oxidize Br⁻ than LPO, although, like LPO, it could not oxidize Cl⁻ at neutral pH. This bacterial peroxidase may prove to be useful in studies of covalent heme binding, as the protein is not glycosylated and does not have the high number of disulfide bonds that complicate studies of the mammalian proteins.

MPO differs from the other mammalian peroxidases in that it has, in addition to the ester bonds to the heme 1- and 5-methyl groups, a third covalent link between the heme 2-vinyl group and the sulfur atom of Met243.⁴¹  The resolution of the initial X-ray data was not sufficient to firmly establish the nature of the Met–vinyl link and two possibilities were proposed (Figure 1.3),^40,42,53  but the vinyl sulfonium structure (Figure 1.3, structure A) is now well established. Fourier transform infrared difference spectroscopy with the protein generally deuterium labeled with CD₃-methionine has directly detected the sulfonium group in the protein.⁵⁴  This structure readily explains why MPO is distinguished from the other mammalian peroxidases by its abnormal Soret maxima in the ferric state at 428 nm and in the reduced state at 472 nm. Both of these are red shifted relative to the maxima at approximately 413 nm in the ferric state and 444 nm in the reduced state for the other mammalian peroxidases. Mutation of Met243 in MPO to a Gln results in a shift of the Soret band of the reduced enzyme to 445 nm, a position similar to that of both LPO and EPO.^54,55  Furthermore, the Met243Gln mutant is no longer able to oxidize chloride to hypochlorite, which shows that the Met–vinyl crosslink is important for both its spectroscopic properties and its physiological activity.

The sequence in which the covalent links in MPO are formed has been clarified by expression and characterization of the Asp94Asn, Asp94Val, and Glu242Gln mutants. The Glu242Gln mutant, which is incapable of forming the covalent link to the heme 1-methyl, yields a protein that, based on its spectroscopic properties, has a normal Met–vinyl link.⁵⁶  Both of the MPO Asp94 mutants yield two proteins, one that appears to have the normal Met–vinyl link, and one that does not. Later work confirmed that Asp94 is not essential for formation of either the Glu242–methyl link or the Met243–vinyl link.^57,58  Thus, formation of the Met–vinyl link is more sensitive to changes associated with the Asp94 than the Glu242 link.

The mechanism for formation of the MPO Met243–vinyl link remains uncertain, but it is generally assumed that it is the result of an autocatalytic maturation process. In one attempt to examine autocatalytic formation of this link, Gln376 of bovine LPO, a residue close to a heme vinyl group, was replaced by a methionine to determine if an MPO-like structure would be formed on exposure to H₂O₂.⁵⁹  The experiment was unsuccessful, as the resulting protein had neither the spectral shift characteristic of MPO nor its ability to oxidize Cl⁻ anions. However, a comparable but successful experiment with a plant peroxidase confirms the feasibility of autocatalytic bond formation.⁶⁰  In this experiment, a methionine was placed close to the heme 2-vinyl group in pea cytosolic ascorbate peroxidase by site-specific mutagenesis of Ser160. The Ser160Met mutant expressed in E. coli was obtained without a heme prosthetic group, but reconstitution with heme gave a red protein. Incubation of the reconstituted mutant with H₂O₂ yielded a green protein shown by chromatographic and spectroscopic methods to have a methionine–vinyl crosslink. The mass spectrometric data was consistent with addition of the methionine sulfur and a hydroxyl group across the vinyl double bond (Figure 1.4). Although this structure differs from that for the native prosthetic group in MPO (Figure 1.3A), the difference only depends on whether addition of a water molecule to the cation intermediate is faster or slower than deprotonation of the carbon adjacent to the positively charged sulfur atom. This model experiment provides convincing evidence that autocatalytic formation of the Met243–vinyl bond in MPO by an autocatalytic mechanism is feasible.

Ascorbate peroxidases undergo autocatalytic heme–protein crosslinking when they react with H₂O₂ in the absence of ascorbic acid, their normal substrate. Thus, in the presence of H₂O₂, a covalent link is formed between the distal Trp41 and the 4-vinyl group of the heme in cytosolic pea ascorbate peroxidase (Scheme 1.6).⁶¹  The sites of the crosslink are confirmed by failure to form the heme adduct with either the Trp41Ala mutant or with a heme reconstituted with deuteroheme, which lacks the two vinyl groups. Although not conclusively demonstrated, the mass spectrometric data is consistent with attachment of the tryptophan to the terminal carbon and a water molecule to the internal carbon of the original 4-vinyl double bond. The formation of this adduct was proposed to result from autocatalytic oxidation of Trp41 to a radical that adds to the vinyl group.

As discussed in Section 1.3.3, replacement of Ser160 of ascorbate peroxidase by a methionine results in the formation of a methionine sulfonium link that serves as an approximate model for the autocatalytic maturation of MPO.⁶⁰  This heme crosslinking reaction is not highly specific, as replacement of Ser160 in pea ascorbate peroxidase by a tyrosine leads to autocatalytic formation of a tyrosine–heme crosslink via addition of an autocatalytically generated tyrosine radical to either the heme vinyl group or a meso-carbon of the porphyrin framework.⁶² 

An analogous crosslinking reaction occurs in tobacco stromal ascorbate peroxidase.⁶³  In this instance, tryptic digestion, mass spectrometry, and site specific mutagenesis established that Trp35 is the protein residue involved in the crosslink, although the site on the heme to which Trp35 is attached was not identified. Formation of this crosslink occurs in concert with inactivation of the protein. The formation of heme vinyl crosslinks with tryptophan, tyrosine, and methionine radicals in ascorbate peroxidases demonstrates the inherent facility of such addition reactions when an amino acid side-chain that can be oxidized by Compound I is located close to a heme vinyl group.

Reaction of cytochrome c peroxidase with H₂O₂ has been found to generate several protein radicals, one of which is essential for its catalytic mechanism,⁶⁴  but specific intramolecular protein modifications and protein–heme crosslinks have not been observed with the wild-type protein.⁶¹  However, when His52, the distal catalytic histidine of S. cerevisiae cytochrome c peroxidase is mutated to a tyrosine, reaction of the protein with H₂O₂ results in the formation through a diradical combination mechanism of a crosslink between Tyr52 and the indole ring nitrogen of Trp51.⁶⁵  The crystal structure of the protein shows that this crosslink involves the formation of a bond between the tryptophan nitrogen atom and one of the carbons in the tyrosine adjacent to that bearing the hydroxyl group (Figure 1.5A).

In cytochrome c peroxidase, the catalytically relevant radical cation of Compound I is not located on the heme porphyrin ring, as it is in HRP and most peroxidases, but rather on Trp191, a tryptophan on the proximal (histidine iron ligand) side of the heme group.⁶⁴  As replacement of Trp191 by a less readily oxidized amino acid might be expected to stabilize a porphyrin radical cation form of Compound I, Trp191 of cytochrome c peroxidase was mutated to a phenylalanine.⁶⁶  Incubation of this mutant with H₂O₂ results in the formation of a covalent bond between the heme 4-vinyl group and Trp51 (Figure 1.5B). The KatG catalase-peroxidases, ascorbate peroxidases, and cytochrome c peroxidases are all distinguished from other peroxidases by the presence of a tryptophan in the distal heme site. The fact that they are all susceptible to autocatalytic formation of Trp-heme adducts reinforces the conclusion that oxidation of an amino acid side-chain to a free radical species is conducive to the formation of modified protein structures if the oxidized side-chains are appropriately located close to the heme or another oxidizable amino acid.

The reactive free radical and electrophilic products catalytically produced by peroxidases can covalently alter the prosthetic heme group. This aspect of peroxidase function has been most extensively investigated with HRP, the heme of which has been shown to be modified by the free radicals obtained on oxidation of, among others, aryl and alkyl hydrazines,^67,68  azide anion,⁶⁹  nitromethane,⁷⁰  cyclopropanone hydrate,⁷¹  and alkyl acids.⁷²  In peroxidases, the free radicals can potentially add at the δ-meso position of the heme, at the two vinyl groups, or, in some instances, can abstract a hydrogen from a heme methyl group to produce a hydroxylated product (Scheme 1.7). A correlation of the energies of catalytically generated free radical products with the positions at which they react suggests that high energy free radicals can add at either the δ-meso carbon or a vinyl group of the heme in HRP, whereas radicals with a lower intrinsic reactivity (e.g., nitrite radical) only add efficiently to heme vinyl groups.⁷³  A C–H bond dissociation energy of approximately 90 kcal mol⁻¹ to generate a given radical is required for the radical to add to a meso-position, whereas radicals with lower bond C–H dissociation energies can add to the vinyl groups (Table 1.2). For example, the oxidation of phenyldiazene to the phenyl radical by HRP yields products from two of the reaction pathways, addition to the δ-meso carbon of the heme and abstraction of a hydrogen from the adjacent 8-methyl to produce the 8-hydroxymethyl-heme derivative (Scheme 1.8).^67,68  Reactions with the δ-meso carbon and 8-methyl group are favored by the HRP active site architecture, which channels the substrate to the δ-meso edge of the heme.⁷⁴ 

Studies with HRP and related peroxidases have shown that the electrophilic two-electron oxidation products generated from halide ions or NCS⁻ can also modify the prosthetic heme group of the enzyme. The oxidation of Br⁻ by HRP, although not a fast reaction, results in the addition of HOBr across one or both of the vinyl substituents of the heme (Figure 1.6).⁷⁵  The oxidation of NCS⁻ to HOSCN by HRP gives rise to similar vinyl group adducts.⁷⁶  As already noted, Cl⁻ is not oxidized by LPO at pH 7, but at acidic pH a low oxidation of Cl⁻ occurs, as demonstrated by the formation of similar adducts of the heme vinyl groups with catalytically generated HOCl.⁷⁵  In addition to the vinyl adducts, the oxidation of Cl⁻ and NCS⁻ (but not Br⁻) resulted in addition of a chloride or thiocyano moiety, respectively, to the δ-meso carbon of the heme,^75,76  a reaction that probably proceeds via a free radical intermediate, as chloride addition is not suppressed by monochlorodimedone, whereas the vinyl additions are.⁷⁷  Reactions of the heme group with electrophilic and radical intermediates are not limited to HRP, as similar products have been observed in the reactions of the peroxidase from Arthromyces ramosus.⁷⁶  Although not discussed here, catalytically-generated electrophilic metabolites may also react with nucleophilic groups on the protein.⁷⁸ 

The mammalian peroxidases, like the plant, fungal, and microbial enzymes, catalyze the one-electron oxidation of diverse organic substrates. However, the mammalian enzymes have an additional physiologically important function: the two-electron oxidation of Cl⁻, Br⁻, I⁻, and NCS⁻, respectively, to HOCl, HOBr, HOI, and HO-SCN.⁷⁹  Thus, TPO catalyzes the oxidative iodination of thyroxine in the synthesis of thyroid hormone,⁸⁰  the HOCl and HOBr generated by MPO, EPO, and LPO function as antimicrobial agents,⁷⁹  and the oxidation of NCS⁻ to HOSCN may serve various purposes, including as an antimicrobial agent, as an antioxidant with respect to the more reactive halohydrins, and in regulating some protein functions.⁸¹  The two-electron oxidation of halides and pseudohalides is thus an important, physiologically relevant, function of the mammalian peroxidases. In contrast, although plant and fungal enzymes can perform some of these oxidations, they are not a part of their normal physiological function. This suggests that the covalent modifications found in mammalian peroxidases may relate, at least in part, to the oxidation of halides and pseudohalides rather than to the one-electron oxidations that are also central to the function of plant and microbial peroxidases.

The MPO and LPO crystal structures show that the three crosslinks to the heme in MPO cause a bowing of the heme planar structure (Figure 1.2) more severe than the more modest distortions of the heme caused by the two ester crosslinks in LPO.^38,41  Resonance Raman confirms that although LPO has some heme distortion,⁸²  it is much smaller than that in MPO.^56,83  As a result of this distortion and the electronic effects of the vinyl sulfonium substituent in MPO, the effects of the crosslinks on the spectroscopic and biophysical properties of LPO and MPO differ markedly. Whereas LPO has spectroscopic properties and a redox potential similar to that of plant peroxidases, such as HRP, which have no heme covalent bonds, those of MPO are quite different.⁸⁴  In addition to the red shift of the Soret band noted earlier, the Fe(iii)/Fe(ii) redox potential of LPO is −190 mV, whereas it is approximately +5 mV for MPO.^85,86  The MPO Compound I/Compound II redox couple is 1.35 V and the Compound II/ferric couple 0.97 V.⁸⁷  In contrast, both redox couples for HRP, which has no heme–protein covalent links, are 0.9 V. Despite this shift, the rate of formation of Compound I is not significantly altered, but the rate of reduction of Compound I by bromide and chloride was much slower in the absence of the methionine link.^57,58  These differences, which primarily reflect the effect of the electron-withdrawing vinyl sulfonium substituent, contribute to the unique ability of MPO to efficiently oxidize Cl⁻ ions. In contrast, EPO readily oxidizes Br⁻, I⁻, and NCS⁻, but only poorly oxidizes Cl⁻, whereas LPO can oxidize I⁻ and NCS⁻, has a low ability to oxidize Br⁻, and is essentially inactive towards Cl⁻ at neutral pH (Table 1.3). The Met–vinyl link thus enhances the oxidizing power of MPO Compound I, enabling the efficient oxidation of Cl⁻ to HOCl, a key constituent of the innate immunity system.

If a primary role of the vinyl–Met link in MPO is to enhance the ability of the enzyme to oxidize chloride ions, what is the role of the two ester links that are common to all the mammalian peroxidases? Resonance Raman shows that the bowed conformation of the heme group in MPO is relaxed in the Asp94 and Glu242 mutants in which one or the other of the two ester bonds is missing.⁵⁶  This change in the heme may contribute to the finding that the overall chlorination and bromination activities of the Glu242Gln MPO mutant are 2% and 24% of those of the wild-type recombinant enzyme, a finding which suggests that the Glu242 crosslink contributes to the ability of the enzyme to oxidize halide ions.⁹¹  The ester bonds linking the heme to the protein may therefore directly influence the oxidative properties of the enzyme.

Experiments with HRP suggest that the ester crosslinks between the heme and the protein may also protect the catalytic site from oxidative damage by the antimicrobial products HOCl, HOBr, and HOSCN that are generated by the mammalian peroxidases. The mammalian peroxidases, in contrast to HRP and other plant, fungal, and bacterial peroxidases (Section 1.3.4), are resistant (but not impervious) to heme modification by both catalytically generated free radical and electrophilic metabolites.⁹²  A specific role for the ester crosslinks in mediating this resistance is suggested by experiments with the Phe41Glu mutant of HRP,⁹³  in which one covalent ester bond is formed between the 3-methyl and Glu41 on reaction with low concentrations of H₂O₂.⁵⁰  Incubation of this cross-linked mutant with Br⁻ under conditions that result in virtually complete modification of the heme group in wild-type HRP cause negligible modification of the heme in the mutant protein. Similar protection is afforded against HOSCN generated from NCS⁻. Parallel experiments with the Glu375 mutant of LPO, which retains only one of the two ester links, indicate that a single ester bond is sufficient to protect the heme, as the LPO mutant with a single ester link was similarly resistant to heme modification. The exact nature of this protective effect is not clear, as it could be entirely due to steric effects of the ester link that limit access to the vinyl groups, but could also include more subtle electronic and structural effects. The protection is not absolute, however, as the heme of MPO is partially irreversibly bleached on reaction with HOCl in a reaction that is partially prevented by chloride ions.⁹⁴  Furthermore, even MPO with three heme–protein crosslinks has been shown to form heme adducts in which relatively large 2-thioxanthine substrates become covalently attached to the heme 8-methyl group (Figure 1.7).^95,96  Kinetic studies have shown that this heme alkylation reaction is a mechanism-based inactivation that nevertheless releases a reactive intermediate that can also alkylate other proteins.⁹⁷  A radical hydrogen abstraction, methylene cation formation, and trapping by the thiol analogous to that for formation of the 8-hydroxyheme with phenylhydrazine (Scheme 1.8) have been proposed for this reaction.

The intermolecular and intramolecular oxidation of tyrosine, tryptophan, and thioethers such as methionine by hemoprotein peroxidases is consistent with the relative ease with which these residues can be oxidized. An informative example is provided by early efforts to pin down the catalytically relevant tryptophan in cytochrome c peroxidase by individually mutating the tryptophans in the protein. The result was that mutation of even the critical catalytic tryptophan simply led to observation of other tryptophan and tyrosine radicals formed by the oxidation of alternative amino acid residues to radical cations or radicals.^64,98–101  The peroxidative generation of various tyrosine and tryptophan radicals in myoglobins and hemoglobins upon reaction with H₂O₂ provides a second relevant example.¹⁰²  More immediately relevant is the finding that reaction of KatG from Burkholderia pseudomallei with peroxyacetic acid in the absence of one-electron donors results in the addition of up to three oxygen atoms to Trp and Met residues of the protein.¹⁰³ 

Autocatalytic maturation reactions usually require two one-electron oxidations, or, alternatively, a two-electron oxidation. The coupling of amino acids, as seen in the formation of the KatG cross-linked tripeptide, is best explained by the in situ formation of two aromatic radicals, presumably through the sequential action of Compound I and Compound II of the peroxidase (Scheme 1.4). However, crosslinking of the third residue, a methionine, to the dipeptide that is generated by radical coupling requires formation of an extended iminoquinone by either two one-electron or, less likely, a concerted two-electron oxidation. The actual methionine crosslink is formed by nucleophilic attack of the methionine on the iminoquinone intermediate.

In contrast, the formation of crosslinks between heme vinyl or methyl substituents and protein residues requires the formation of a single radical species, as the Fe(iv) of Compound II can serve as an electron sink for the second electron. The free radical formed by hydrogen abstraction from a heme methyl (Scheme 1.5), or by addition of a protein radical to a heme vinyl, is converted to a cation by electron transfer to the iron. The reactions are then terminated either by trapping the cation with a nucleophile such as a carboxylic acid, forming an ester link, or by loss of a proton, as occurs in regeneration of the double bond in the MPO Met–vinyl link.

Based on the examples that are available, the principal determinants for autocatalytic maturation of hemoproteins are: (a) the presence of one or more amino acids positioned within the protein so that they are susceptible to one-electron peroxidative oxidation, (b) for amino acid crosslinking, location of a second amino acid that can also undergo one-electron oxidation close enough to the first that they can react with each other; and (c) for heme crosslinking, location of the appropriately oriented oxidizable amino acid side-chain close to a heme methyl or vinyl substituent.

These general determinants do not preclude the possibility of crosslink formation via other mechanisms, as in the apparently spontaneous attachment of His117 to the heme 2-vinyl group in Synechocystis hemoglobin.^104,105 

The peroxidases are not exceptional among hemoproteins in their ability to promote intramolecular amino acid or amino acid–heme crosslinks. For example: (a) most CYP4A and CHP4F cytochrome P450 enzymes autocatalytically form a single covalent ester bond that attaches the heme 5-methyl to a protein carboxylic acid residue;^106,107  (b) several catalases have cross-linked dipeptides, including a Cys–Tyr link in a Neurospora crassa catalase,¹⁰⁸  as well as a Tyr–His link in catalase HPII from Escherichia coli;¹⁰⁹  (c) cytochrome c oxidase has a His–Tyr cross-linked dipeptide,¹¹⁰  and (d) the heme in E. coli HPII is autocatalytically converted to a heme d moiety.¹¹¹ 

The autocatalytic formation of intramolecular links between amino acid residues, between amino acid side-chains and the heme group, and modifications of the heme group are being found in a growing number of hemoproteins. Much progress has been made in understanding the mechanisms by which these links are formed in the peroxidases, but the structural, mechanistic, and functional reasons for formation of the covalent links requires further clarification.