Description of a Versatile Peroxidase Involved in the Natural Degradation of Lignin That Has Both Manganese Peroxidase and Lignin Peroxidase Substrate Interaction Sites*

Two major peroxidases are secreted by the fungusPleurotus eryngii in lignocellulose cultures. One is similar to Phanerochaete chrysosporiummanganese-dependent peroxidase. The second protein (PS1), although catalyzing the oxidation of Mn2+ to Mn3+ by H2O2, differs from the above enzymes by its manganese-independent activity enabling it to oxidize substituted phenols and synthetic dyes, as well as the lignin peroxidase (LiP) substrate veratryl alcohol. This is by a mechanism similar to that reported for LiP, as evidenced byp-dimethoxybenzene oxidation yielding benzoquinone. The apparent kinetic constants showed high activity on Mn2+, but methoxyhydroquinone was the natural substrate with the highest enzyme affinity (this and other phenolic substrates are not efficiently oxidized by the P. chrysosporium peroxidases). A three-dimensional model was built using crystal models from four fungal peroxidase as templates. The model suggests high structural affinity of this versatile peroxidase with LiP but shows a putative Mn2+ binding site near the internal heme propionate, involving Glu36, Glu40, and Asp181. A specific substrate interaction site for Mn2+ is supported by kinetic data showing noncompetitive inhibition with other peroxidase substrates. Moreover, residues reported as involved in LiP interaction with veratryl alcohol and other aromatic substrates are present in peroxidase PS1 such as His82 at the heme-channel opening, which is remarkably similar to that of P. chrysosporiumLiP, and Trp170 at the protein surface. These residues could be involved in two different hypothetical long range electron transfer pathways from substrate (His82-Ala83-Asn84-His47-heme and Trp170-Leu171-heme) similar to those postulated for LiP.

The biodegradation of lignin represents a key step for carbon recycling on earth because forest ecosystems contain about 150,000 million tons of wood (1). Although the main wood constituent, cellulose, can be utilized by a variety of organisms, its hydrolysis in situ by most of them is hampered by the recalcitrant lignin polymer, which is formed by polymerization of plant p-hydroxycinnamyl alcohols (2). However, the so-called white rot fungi, a group of species from class Basidiomycetes, have developed a remarkable capability for oxidative depolymerization and subsequent mineralization of lignin (3) enabling cellulose utilization by other organisms.
During recent years many studies on lignin biodegradation have been carried out in the fungus Phanerochaete chrysosporium (order Stereales). The first evidence of peroxidase involvement came from inhibition of lignin degradation by catalase (4). Then two peroxidases involved in lignin degradation, the socalled lignin peroxidase (LiP) 1 and manganese peroxidase or manganese-dependent peroxidase (MnP), were described (5)(6)(7). LiP is characterized by high redox potential enabling oxidation of hardly biodegradable nonphenolic aromatic compounds, such as veratryl alcohol and methoxylated benzenes. P. chrysosporium MnP strictly requires Mn 2ϩ to complete the catalytic cycle, and the chelates of Mn 3ϩ formed can act as efficient oxidizers of phenols and other compounds. Because most units in lignin are nonphenolic (2), LiP was considered as the main enzyme responsible for lignin depolymerization (3). This is still generally accepted, although some evidence suggests that lignin biodegradation could proceed by endwise attack at the phenolic units (8) and that, in addition to Mn 3ϩ chelates, strong chemical oxidants can be generated by MnP (9).
Since 1983-1984 a large amount of information on peroxidases from P. chrysosporium has been accumulated (10 -13), and crystal models have been described (14 -16). At the time of the LiP model, only the cytochrome c peroxidase crystal structure was available. The information on their molecular structure not only clarified some peculiarities of these enzymes but also contributed to better understanding of some aspects of peroxidase structure and function (17). Despite the above progresses, the mechanism by which lignin is biodegraded is still to be fully understood, and only a limited number of biotechnological applications have been developed. This is related to the fact that the above studies have been focused on a single organism and that little is known about lignin degradation by fungi in their natural environment. Studies on lignocellulose degradation under solid state fermentation (SSF) conditions, which are close to those of natural habitat of white rot fungi, should provide information to contrast that obtained using liquid media (18 -20). * This work was supported by European Contract AIR2-CT93-1219 and Project BIO96-393 of the Spanish Biotechnology Programme. 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.
The The fungus Pleurotus eryngii (order Poriales) has been reported as having the capability to remove lignin selectively when growing on natural substrates and is therefore considered a model organism for studies on biodegradation of lignin in annual plants and related biotechnological applications (21,22). Peroxidases oxidizing Mn 2ϩ have been reported in liquid cultures of P. eryngii (23), together with the AAO responsible for H 2 O 2 generation in Pleurotus species and laccases (21,24). However, no typical LiP has been described in this or other Pleurotus species. The aim of the present study was to identify and characterize the peroxidases involved in natural degradation of lignin by this fungus and to investigate the eventual production of LiP. P. eryngii was therefore grown on a lignocellulosic substrate under SSF conditions, and several peroxidases were isolated. Their catalytic properties were investigated using different substrates, and a three-dimensional model for a P. eryngii peroxidase was built using existing crystal models from other fungal peroxidases as templates.

EXPERIMENTAL PROCEDURES
Organisms and Culture Conditions-P. eryngii ATCC 90787 (IJFM A169), Trametes versicolor IJFM A136, and P. chrysosporium ATCC 24725 (VKM F-1767) were grown on sterile wheat straw under SSF conditions. These are characterized by the presence of enough liquid phase to saturate the solid lignocellulosic substrate, maintaining an air phase between particles. The fungi were grown in an horizontal rotary fermentor, which included six 2-liter bottles containing 125 g of straw (length, 1-2 cm) and 375 ml of water (including inoculum). The inoculation and growth conditions were already described (22). Samples were taken during 2 months to analyze enzymatic activities and substrate degradation. Lignin content was estimated by the Klason method after Saeman's acid hydrolysis and polysaccharide composition by gas chromatography of the acid hydrolyzate, following T-222 and T249 rules (25).  (23). AAO was determined as the veratraldehyde formed from 5 mM veratryl alcohol in 0.1 M phosphate buffer, pH 6. Laccase was measured with 10 mM syringol in 0.1 M sodium tartrate, pH 5. One activity unit was defined as the amount of enzyme transforming 1 mol of substrate/min.

Chemicals-Reactive
Enzyme Purification-Peroxidases were purified from 15-day-old SSF cultures. Water extracts from the straw treated with the fungus were obtained (by adding 3 liters/bottle and shaking for 1 h at 200 rpm), filtered (0.8 m), concentrated by ultrafiltration (5-kDa cut-off), and dialyzed against 10 mM sodium tartrate, pH 4.5. The concentrate (approximately 140 ml) was loaded onto a Bio-Rad Q-cartridge (1 ml/min), and retained fractions were eluted with 1 M NaCl. Fractions with peroxidase activity were concentrated, and 1-ml samples applied onto a Sephacryl S-200 HR column (0.8 ml/min). The peroxidase fractions were pooled, concentrated, and dialyzed against 10 mM sodium tartrate, pH 5, and 1-ml samples were applied to a Mono-Q column. Proteins with peroxidases activity were separated using a 0 -0.25 M NaCl gradient (30 min, 0.8 ml/min).
Enzyme Characterization-Protein concentration was determined with the Bradford reagent. Isoelectric focusing was performed in 5% polyacrylamide gels with a thickness of 1 mm and a pH range from 2.5 to 5.5. SDS-polyacrylamide gel electrophoresis was carried out in 12% polyacrylamide gels. Gels were stained with Coomassie Blue R-250. Proteins were deglycosylated using Endo-H from Boehringer. The Nterminal sequences were obtained by automated Edman degradation of 5 g of protein in an Applied Biosystems 494 pulsed liquid protein sequencer. The steady state kinetic constants were obtained from Lineweaver-Burk plots, and the mean values are presented. The kinetic constants for Mn 2ϩ peroxidase activity were calculated by the formation of Mn 3ϩ tartrate at pH 5. Manganese-independent activities at pH 3 on ABTS (cation radical ⑀ 436 29.3 mM Ϫ1 cm Ϫ1 ), Reactive Black 5 (⑀ 558 50 mM Ϫ1 cm Ϫ1 ), guaiacol (o-methoxyphenol) (oxidation product ⑀ 456 12.1 mM Ϫ1 cm Ϫ1 referred to substrate concentration), syringol, methoxyhydroquinone (methoxybenzoquinone ⑀ 360 1.25 mM Ϫ1 cm Ϫ1 ), veratryl alcohol, and p-dimethoxybenzene (benzoquinone ⑀ 254 21 mM Ϫ1 cm Ϫ1 ) were estimated in the presence of EDTA, and kinetic constants were calculated. The K m for the enzyme-oxidizing substrate H 2 O 2 was also obtained at pH 5 (by measuring Mn 2ϩ oxidation) and pH 3 (syringol oxidation). Reciprocal inhibition between Mn 2ϩ and Reactive Black 5 was investigated by following spectrophotometrically the effect of different concentrations of both compounds as inhibitors on the decolorization by peroxidase PS1 of 0. HPLC Analysis of p-Dimethoxybenzene Oxidation-The oxidation of p-dimethoxybenzene (1 mM) was followed spectrophotometrically, using 0.1 M sodium tartrate, pH 3, and 0.2 mM H 2 O 2 . Samples (20 l) from the reaction mixture were analyzed by HPLC using a C18 column (Spherisorb S5ODS2) at 30°C. A methanol/phosphoric acid (10 mM) gradient (consisting of 10% methanol for 10 min, 10 -100% methanol in 6 min, 100% methanol for 4 min, 100 -10% methanol in 0.5 min, and 10% methanol for 6.5 min) was used (1 ml/min). Dual wavelength and diode array detectors were used. Standard calibration curves were used for quantitation.
Protein Modeling-The atomic coordinates of crystal models of LiP-H8 (PDB entries 1LGA and 1LLP), LiP-H2 (1QPA), and MnP1 (1MNP) from P. chrysosporium and ARP-CIP (1ARP) were obtained from the Brookhaven Protein Data Base. The latter peroxidase has been crystallized from both Coprinus cinereus (CIP) (26) and Arthromyces ramosus (ARP) (27). Because the latter is an invalid species (nomen nudum), which could correspond to a Coprinus conidial state, and both proteins share 99% identity, the name ARP-CIP is used here for both of them. The gene coding for protein PS1 of P. eryngii has been cloned, and a preliminary sequence has been obtained (28). From the predicted amino acid sequence a three-dimensional model for the mature protein without Mn 2ϩ was obtained by sequence homology using the program ProMod and refined by CHARMm (the C-terminal region was not modeled) (29). It was based on alignment of sequence of peroxidase PS1 with the above four fungal peroxidases for which crystal models are available. Secondary structure was determined with the DSSP program (30). The above models for P. chrysosporium MnP and LiP were also used for comparison with P. eryngii peroxidase PS1 using programs Swiss-Pdb Viewer and RasMol. Multiple sequence alignment was prepared with PILEUP from GCG package using W2H interface.

RESULTS
P. eryngii degrades wheat lignin maintaining high cellulose content. This resulted in the increased polysaccharide/lignin ratio shown in Table I. Preferential degradation of lignin is a singular characteristic because most ligninolytic basidiomycetes (including the P. chrysosporium and T. versicolor strains shown in Table I) cause stronger degradation of cellulose than lignin. When the presence of ligninolytic enzymes was investigated in the partially degraded straw no LiP was detected in any case, but AAO, laccase, and Mn 2ϩ -oxidizing peroxidase were all produced by P. eryngii and T. versicolor. The high Mn 2ϩ -oxidizing activity produced by P. eryngii suggested that the enzyme responsible for this activity could be involved in lignin degradation.
The process of peroxidase purification included ultrafiltration of the extract from SSF culture of P. eryngii, followed by Q-cartridge (removing colored compounds from lignin degradation), molecular size exclusion, and ion exchange chromatogra- phy. As shown in Fig. 1A, the last purification step resulted in the isolation of two major protein peaks (labeled PS1 and PS3) and a minor one (protein PS2). Both major peaks were electrophoretically homogeneous (Fig. 1B) and were fully characterized. The three proteins exhibited high absorbance at 410 nm and were able to oxidize Mn 2ϩ in the presence of H 2 O 2 . The optimum for this reaction was at pH 5. Proteins PS1, PS2, and PS3 differed in Mono-Q retention volume, A 410 /A 280 ratio (4.9, 4.9, and 5.5), isoelectric point (3.67, 3.65, and 3.80, respectively), molecular mass (45,45, and 42 kDa, respectively), and N termini (VTCATGQTT for PS1 and PS2 and VTCADGNTV for PS3). In the case of protein PS1 a total of 25 amino acid residues were identified, and no double sequences were observed confirming protein purity. Proteins PS1 and PS3 also differed in the N-glycosylation degree (4 and 2%, respectively). A different Mn 2ϩ -oxidizing peroxidase produced by P. eryngii when grown in liquid media (23) has a molecular mass intermediate between those of the two major peroxidases isolated here (Fig. 1B) and also differs in the N-terminal sequence.
In addition to their activity on Mn 2ϩ , the peroxidases isolated from lignocellulose cultures were able to oxidize substituted phenols, such as guaiacol, syringol, and methoxyhydroquinone. The proteins PS1 and PS2 were able to also oxidize nonphenolic aromatic molecules such as the LiP substrate veratryl alcohol. The fact that peroxidase activity on veratryl alcohol (LiP-type activity) was not detected in the cultures (Table I) is probably due to its inhibition by phenols (or other aromatic compounds in extracts) as described for LiP (19;31). Because veratryl alcohol is a substrate of Pleurotus AAO (and can be oxidized by other fungal enzymes in the presence of mediators), p-dimethoxybenzene was assayed as a more specific LiP substrate. The results of incubation with peroxidases PS1 and PS3, only the first enzyme oxidizing the substrate, are shown in Fig. 2 (A and B). The HPLC analysis revealed benzoquinone as the reaction product by peroxidase PS1 (Fig. 2C). The higher redox potential of peroxidase PS1 was evidenced also by oxidation of some synthetic dyes. Whereas both enzymes oxidized ABTS (PS1 with 3-fold higher specific activity than PS3), only peroxidase PS1 was able to decolorize the Reactive Black 5, a high redox potential azo dye. The optimum for all the manganese-independent reactions was around pH 3.
The steady state kinetic constants of the two major peroxidases for oxidation of Mn 2ϩ to Mn 3ϩ and for manganese-independent oxidation of two phenolic and two nonphenolic aromatic substrates, and the dye Reactive Black 5 are shown in Table II. As already mentioned, the three latter substrates were oxidized only by peroxidase PS1, but both peroxidases also differed in the K m values for the other substrates. These revealed higher affinities for Mn 2ϩ and phenols of PS1 than PS3. Moreover, peroxidase PS1 could oxidize methoxyhydroquinone concentrations below 1 M, whereas peroxidase PS3 did not exhibit activity below 100 M substrate. The K m obtained for peroxidase PS1 (17 M) is lower than found for other enzyme-reducing aromatic substrates (with the only exception of synthetic dyes), revealing substituted B, peroxidase PS3. The UV spectrum did not change under the same conditions described in A revealing that p-dimethoxybenzene is not oxidized by this enzyme. C, HPLC analysis (PS1). Benzoquinone formation after p-dimethoxybenzene incubation (20 min) with peroxidase PS1 was confirmed by HPLC using dual wavelength (254 and 286 nm) and diode array detectors, which enabled identification of the quinone formed (the benzoquinone UV spectrum is characterized by maximum at 254 nm) and quantitation of reaction substrate and product. hydroquinones as the best natural substrates of P. eryngii peroxidase PS1 from the point of view of enzyme affinity. However, because of the higher turnover number for Mn 2ϩ oxidation, both peroxidases showed the highest efficiency (t/K m ) for Mn 2ϩ oxidation (even considering that oxidation of aromatic compounds should involve two one-electron oxidations). The apparent K m values for H 2 O 2 , obtained during both Mn 2ϩ oxidation at pH 5 and oxidation of aromatic substrates at pH 3, were low. This corresponds to high affinity of peroxidases for H 2 O 2 , but the value obtained at pH 3 could be considered as extremely low compared with other fungal peroxidases. Mutual inhibition between peroxidase PS1 oxidation of the substrates Reactive Black 5 (an azo dye that is not oxidized by Mn 3ϩ ) and Mn 2ϩ was found. However, the azo dye appeared as a stronger inhibitor because the inhibitor/substrate molar ratio, to cause 40% inhibition, was 125 in the case of Mn 2ϩ inhibition of dye oxidation, and only 0.13 in the case of Reactive Black 5 inhibition of Mn 2ϩ oxidation. The type of inhibition caused by Reactive Black 5 was investigated using different dye concentrations, and although a decrease of the velocity of Mn 2ϩ oxidation was found (the turnover number decreasing from 71 to 43 s Ϫ1 in the presence of 32 M dye), the K m value was maintained (around 50 M). This suggests two different substrate interaction sites with different affinities that are not affected by the presence of the alternative substrate acting as inhibitor. The inhibitory effect would be caused by the reduction of the oxidized enzyme forms (compounds I and II) during oxidation of the substrate acting as inhibitor. Finally, it is possible to mention that the stronger inhibitory effect caused by Reactive Black 5 agrees with the higher peroxidase affinity for this substrate, compared with the affinity for Mn 2ϩ .
Because the above results showed that peroxidase PS1 shared LiP and MnP-type catalytic properties, as well as some characteristics of other peroxidases oxidizing phenolic substrates, a three-dimensional model was built for the mature protein (this model includes the two helical domains characteristic of all peroxidases). The four fungal peroxidases for which crystal structures are available were used as templates for homology modeling, and a multiple alignment of their corresponding sequences together with the PS1 sequence that was modeled is shown in Fig. 3. These peroxidases are P. chrysosporium MnP1 (58% sequence identity with PS1) and isoenzymes H8 (60%) and H2 (62%) of LiP, and Coprinus ARP-CIP (52%). The model obtained showed good geometry with root mean square deviations in bond angles and distances of 2.31°a nd 0.013 Å, respectively. It includes 12 predominantly ␣ helices, the position and size of which is indicated in Fig. 3. The helix BЉcorresponds to helix BЈ of LiP (14), with both being longer than helix BЈ of MnP1. A short helix at the position of helix BЈ of P. eryngii peroxidase PS1 was described in ARP-CIP (27), and some helical conformation at this position can be observed also in LiP. Therefore, the three latter peroxidases could include 12 helices, compared with only 11 helices of P. chrysosporium MnP1. The peroxidase PS1 conserves some structurally important elements including disulphide bridges (the last one is not included in the model because it links the C-terminal region to helix I) and putative Ca 2ϩ sites (distal Ca 2ϩ linked by oxygens of Asp 48 , Gly 66 , Asp 68 , and Ser 70 ; and proximal one by those of Thr 176 , Asp 193 , Thr 195 , Thr 198 , and Asp 200 ), characteristic of all peroxidases in classes II and III (17). The peroxidase PS1 model shows the highest coincidence in protein folding and helical topology and the lowest root mean  Figs. 4 and 6), together with sequences from P. chrysosporium LiP (isoenzymes H8 and H2) and MnP1, and Coprinus ARP-CIP used for homology modeling. The atomic coordinates of the four latter proteins were used as templates to obtain the three-dimensional model (Protein Data Bank entry 1BQW) for the P. eryngii peroxidase PS1 including two helical domains, which contain all the catalytically important residues (the C-terminal region, which has variable length in different peroxidases, was not modeled). The alignment was built by PILEUP from GCG package, and the amino acid residues identical to those found in the PS1 sequence are depicted in white on black (the distal and proximal histidine residues are marked with an asterisk). The position of the 12 helices in the PS1 peroxidase predicted by the DSSP program is indicated on the corresponding sequence. The mature ARP-CIP includes an extra Nterminal sequence of 8 amino acids (QG-PGGGGS), which in other fungal peroxidases corresponds to the signal peptide. a pH 5. b pH 3. c A two-step activity curve was obtained when high substrate concentrations are used resulting in a second apparent K m , the enzyme being more efficient at the lowest one. square distance between backbone C ␣ with the LiP-H8 and ARP-CIP models. However, when superimposed with P. chrysosporium MnP1 differences were found in two superficial loops (PS1 Thr 57 -Ala 59 and MnP1 Leu 228 -Thr 234 ).
Some aspects of the model directly related to catalytic activity were considered more in depth. The amino acid residues at the heme pocket of P. eryngii peroxidase PS1 were the same found in other fungal peroxidases (with the only exception of ARP-CIP): Arg 43 , Phe 46 , His 47 , Glu 78 , and Asn 84 in the distal side and His 175 , Phe 192 (Leu 201 in ARP-CIP), and Asp 237 in the proximal side. Fig. 4 (left side) shows the opening of the main heme access channel in P. eryngii peroxidase PS1 (top), P. chrysosporium LiP-H8 (center), and MnP1 (bottom). This channel enables the access of hydroperoxides to the distal side of heme. Moreover, it has been postulated that some of the edge residues are involved in oxidation of aromatic substrates by LiP and horseradish peroxidase. As shown in Fig. 4, the P. eryngii peroxidase PS1 conserves six of the eight residues delimiting the channel edge in LiP-H8 (and one of the two differences is the substitution of Ile 85 by Leu 85 ). However, the heme channel of P. chrysosporium MnP1 differs from that of LiP-H8 in five residues. Fig. 4 (right side) also shows the narrow Mn 2ϩ channel of P. eryngii peroxidase PS1 and P. chrysosporium MnP1 directly on the internal heme propionate, where the cation is fixed by the three acidic amino acid residues (one aspartate and two glutamates) delimiting the channel edge. Finally, it is interesting that Trp 170 of peroxidase PS1 occupies a position at the protein surface similar to that of LiP Trp 171 (that could transfer electrons to heme because of topological proximity), whereas a similar residue is absent from the ARP-CIP and MnP1 models.

DISCUSSION
LiP and MnP of P. chrysosporium have been considered as two models for all ligninolytic peroxidases. However, this study shows that one of the two major peroxidases produced by P. eryngii during lignin degradation under natural growth conditions can be considered as representative for a third type of ligninolytic peroxidase, different from other microbial, plant, or animal peroxidases (17). As shown in Fig. 5, this versatile peroxidase is able to perform both the oxidative reactions characteristic of P. chrysosporium LiP, i.e. the oxidation of nonphenolic aromatic substrates via aromatic radicals, and MnP, i.e. the oxidation of Mn 2ϩ to Mn 3ϩ . Its affinity for H 2 O 2 and Mn 2ϩ was higher than reported for P. chrysosporium peroxidases, but the turnover number for Mn 2ϩ and the affinity and turnover number for veratryl alcohol were lower (32)(33)(34). Moreover, the P. eryngii peroxidase efficiently oxidizes substituted phenols (the affinity for methoxyhydroquinones being even higher than for Mn 2ϩ ), which cannot be oxidized by the P. chrysosporium peroxidases. This is because these compounds inactivate LiP in the absence of veratryl alcohol (31), whereas compound II of P. chrysosporium MnP needs to be reduced by Mn 2ϩ to close the catalytic cycle (phenols are indirectly oxidized by Mn 3ϩ ) (35). In addition to the above substrates, the P. eryngii peroxidase PS1 can oxidize two compounds, ␣-keto-␥-methylthiobutyric acid and p-dimethoxybenzene, which were those used for LiP detection (7) and for demonstration of oxidation mechanism by this enzyme (36). The oxidation of ␣-keto-␥-methylthiobutyric acid, a substrate characteristic of strong one-electron abstracting agents (such as LiP and hydroxyl radical), to ethylene by peroxidase PS1 has been reported very recently (37). The results of p-dimethoxybenzene oxidation to benzoquinone confirmed that the P. eryngii peroxidase has high redox potential and acts on aromatic substrates by the same mechanism described for LiP, i.e. via an aromatic cation radical formed by one electron oxidation of the benzenic ring (36). Finally, Reactive Black 5 is oxidized by the P. eryngii peroxidase, the reaction being inhibited by high Mn 2ϩ concentrations. This azo dye cannot be oxidized by P. chrysosporium MnP because its redox potential is higher than that of Mn 3ϩ tartrate nor by LiP in the absence of veratryl alcohol because of rapid inactivation (38). However, no inactivation is produced in the case of peroxidase PS1. The Reactive Black 5 also acts as a very efficient noncompetitive inhibitor of Mn 2ϩ oxidation by the P. eryngii peroxidase. This result agrees with the noncompetitive inhibition by Mn 2ϩ of the oxidation of dyes and veratryl alcohol by two similar peroxidases recently isolated from liquid cultures of Pleurotus and Bjerkandera species (38 -40) and supports the existence in these proteins of a specific substrate interaction site for Mn 2ϩ , different from that involved in oxidation of other peroxidase substrates.
No crystal structures for P. chrysosporium LiP have been obtained containing veratryl alcohol, which would provide invaluable information for identification of substrate interaction site. However, when the enzyme was crystallized (14,15),it was suggested that oxidation of veratryl alcohol could be produced at the heme access channel, and some amino acid residues were postulated to be involved in substrate binding and electron transfer. Poulos et al. (14) modeled the veratryl alcohol molecule at the heme edge of LiP-H8, H-bonded to His 82 and Gln 222 . These two residues are conserved in the P. eryngii peroxidase PS1, which has a remarkably similar heme access channel as shown in the three-dimensional model obtained (Fig. 4, top and  center). The presence of different amino acid residues at the above two positions (Fig. 4, bottom) probably prevents direct oxidation of aromatic substrates by MnP, together with other differences resulting in a more polar channel environment. It is interesting that Phe 142 of horseradish peroxidase, which occupies a position similar to that of Phe 148 of P. eryngii peroxidase and LiP-H8, has been reported to be involved in aromatic substrate interaction (41,42). Despite the possibility of direct electron transfer from veratryl alcohol at the above binding site to heme as was postulated in P. chrysosporium LiP, substrate oxidation via long range electron transfer (LRET) should also be considered taking into account the distance between channel opening and heme edge. This was suggested by Schoemaker et al. (43), who proposed one of the LRET pathway shown in Fig.  6 (bottom). This initiates in LiP His 82 , mentioned as involved in veratryl alcohol binding, and proceeds via Pro 83 and Asn 84 , the latter being H-bonded to the distal histidine (His 47 ). Such a LRET pathway does not exist in P. chrysosporium MnP, but the three-dimensional model built showed that the His 82 of P. eryngii peroxidase was also exposed at the heme channel opening (Fig. 4, top) and connected by a similar pathway with the distal histidine (PS1 His 47 ) (Fig. 6, top). The difference is that PS1 Ala 83 occupies the position of LiP Pro 83 , resulting in a different position of His 82 side chain at the protein surface. Recently Trp 171 of P. chrysosporium LiP has been described as being involved in the catalytic cycle of the enzyme via a different LRET pathway (Fig. 6, bottom), which does not exist in P. chrysosporium MnP (44). However, Trp 170 of P. eryngii peroxidase PS1 occupies a similar position at the protein surface and it is linked to heme by a LRET pathway as described in LiP (Fig. 6, top). Therefore, this aromatic residue could be involved in oxidation of some substrates by peroxidase PS1, whereas other substrates would be oxidized at the heme channel, as recently shown for P. chrysosporium LiP (45).
The ability to directly oxidize Mn 2ϩ to Mn 3ϩ is a unique characteristic of Mn 2ϩ -oxidizing peroxidases produced by P. chrysosporium and many other white rot fungi (46). Other enzymes could oxidize Mn 2ϩ via superoxide anion radical, such as that generated by redox cycling (47). The existence of a Mn 2ϩ binding site in P. chrysosporium MnP was predicted in theoretical models of the enzyme built by sequence homology using LiP as a template (48), as well as in the first crystal structure obtained (16). Then it was indirectly confirmed by site-directed mutagenesis of MnP Glu 35 , Glu 39 , and Asp 179 (32), and direct evidence obtained by x-ray diffraction of the enzyme crystal- lized in the presence of Mn 2ϩ (49). As shown in Fig. 4, the P. eryngii peroxidase PS1 has the same three acidic amino acid residues (Glu 36 , Glu 40 , and Asp 181 ) forming a small channel directly on the internal heme propionate that acts as first electron acceptor for Mn 2ϩ oxidation by these peroxidases. The absence of the three above residues in P. chrysosporium LiP (Ala 36 , Glu 40 , and Asn 182 at the corresponding positions) results in inability to bind and oxidize Mn 2ϩ (Fig. 4, center). The pH optimum for oxidation of Mn 2ϩ (pH 5) is higher than for direct oxidation of aromatic substrates (pH 3) by the P. eryngii peroxidase. This is because the three acidic residues and the internal heme propionate at the Mn 2ϩ interaction site should be dissociated to bind the cation. The low pH for manganese-independent oxidations is probably due to increased redox potential of heme at low pH values, but the involvement of an acidic amino acid residue in its protonated form has also been suggested (14).
No typical LiP is produced by Pleurotus species, but several Mn 2ϩ -oxidizing peroxidases have been reported. Those from P. pulmonarius cultures on wheat straw (18) and P. ostreatus on sawdust (MnP2) (50) probably correspond to the same peroxidase described here, because they have the same N-terminal sequence and catalytic properties. Peroxidases similar to that produced by P. eryngii in liquid culture (GenBank accession number AF007223 and its allelic variant AF007224) (23,51) are produced by P. pulmonarius in liquid culture (18) and by P. ostreatus in both liquid (52) and sawdust cultures (MnP1) (50) (N terminus ATCADGRTT). In addition to these peroxidases oxidizing both Mn 2ϩ and aromatic substrates and dyes, the peroxidase PS3 from P. eryngii and a peroxidase from liquid culture of P. ostreatus (53) (differing in N terminus) are closer to P. chrysosporium MnP. Production of specific isoenzymes on natural substrates has also been reported in other fungi (18 -20), but the physiological/ecological significance of the different peroxidase isoenzymes remains to be established. Recently a MnP with some manganese-independent activity has been described in Poria (synonym: Ceriporiopsis) subvermispora (54), but it has low sequence homology with P. eryngii peroxidases and is not able to oxidize veratryl alcohol. However, enzymes with catalytic properties similar to Pleurotus versatile peroxidases (23) have been found in Bjerkandera adusta (55) and Bjerkandera sp. (40), the latter being described as a MnP-LiP "hybrid" enzyme. Some aspects of the three-dimensional model shown here suggests that the catalytic properties of this third type of ligninolytic peroxidase are related to a hybrid molecular structure including MnP and LiP-type features. Site-directed mutagenesis studies are necessary to confirm substrate interaction sites and LRET pathways (as proposed from His 82 and Trp 170 ) in P. eryngii peroxidases. This will be facilitated by the recent expression in Emericella nidulans of the gene encoding a P. eryngii peroxidase.