A haem peroxidase different from other microbial, plant and animal peroxidases is described. The enzyme is secreted as two isoforms by dikaryotic Pleurotus eryngii in peptone-containing liquid medium. The corresponding gene, which presents 15 introns and encodes a 361-amino-acid protein with a 30-amino-acid signal peptide, was isolated as two alleles corresponding to the two isoforms. The alleles differ in three amino acid residues and in a seven nucleotide deletion affecting a single metal response element in the promoter. When compared with Phanerochaete chrysosporium peroxidases, the new enzyme appears closer to lignin peroxidase (LiP) than to Mn-dependent peroxidase (MnP) isoenzymes (58–60% and 55% identity respectively). The molecular model built using crystal structures of three fungal peroxidases as templates, also showed high structural affinity with LiP (Cα-distance 1.2 Å). However, this peroxidase includes a Mn2+ binding site formed by three acidic residues (E36, E40 and D175) near the haem internal propionate, which accounts for the ability to oxidize Mn2+. Its capability to oxidize aromatic substrates could involve interactions with aromatic residues at the edge of the haem channel. Another possibility is long-range electron transfer, e.g. from W164, which occupies the same position of LiP W171 recently reported as involved in the catalytic cycle of LiP.
Lignin degradation by white-rot fungi is a key step for carbon recycling in terrestrial ecosystems (Kirk, 1983). Although many microbial, plant and animal haem peroxidases are able to oxidize phenolic substrates, only lignin peroxidase (LiP), an enzyme secreted by Phanerochaete chrysosporium and other ligninolytic fungi, has been reported to oxidize high redox potential aromatic compounds, including the non-phenolic phenylpropanoid units of lignin. Therefore, LiP has been considered a key enzyme in lignin degradation (Kirk and Farrell, 1987). In addition to LiP, most ligninolytic fungi secrete a second type of peroxidase, named manganese peroxidase (MnP) (Kirk and Farrell, 1987), characterized by the ability to oxidize Mn2+ to Mn3+. This is a strong oxidant that could play an important role at the initial stages of lignin degradation in the plant cell wall, where the small size of ‘molecular pores’ between cellulose microfibrils precludes the action of LiP because of steric hindrances (Flournoy et al., 1993).
MnP and LiP were described in P. chrysosporium almost simultaneously (Glenn et al., 1983; Tien and Kirk, 1983; Kuwahara et al., 1984). Both enzymes are haem proteins secreted as several isoenzymes. As shown in 1Fig. 1A and B, which summarizes their catalytic cycles, P. chrysosporium MnP requires Mn2+ to reduce compound-II, whereas LiP can complete the catalytic cycle with non-phenolic aromatic electron donors, such as veratryl (3,4-dimethoxybenzyl) alcohol. MnP and LiP have been thoroughly investigated in P. chrysosporium (Kirk and Farrell, 1987; Gold and Alic, 1993; Schoemaker et al., 1994), although they are also secreted by other ligninolytic fungi (Hatakka, 1994; Peláez et al., 1995). Predicted amino acid (aa) sequences for LiP and MnP enzymes from P. chrysosporium were reported for the first time by Tien and Tu (1987) and Pease et al. (1989), respectively, and several genes were then cloned from this and other fungi (see Fig. 4 for GenBank references). However, P. chrysosporium is not the most promising organism for biotechnological applications based on selective removal of lignin (Eriksson, 1990), and other fungal species are being investigated with this purpose. They include Poria (synonym Ceriporiopsis) subvermispora for delignification of several types of wood, and Pleurotus eryngii for lignin removal from cereal straw (Blanchette et al., 1992; Martínez et al., 1994). Recent studies revealed that P. eryngii secretes five peroxidases, two in peptone-containing liquid media and three during lignin degradation under solid-state fermentation conditions, that efficiently oxidize Mn2+ to Mn3+ but are different from P. chrysosporium MnP because of their Mn-independent activity on aromatic substrates (Martínez et al., 1996a,b). Cloning and sequencing of the new peroxidases from liquid culture of P. eryngii is described here, and their catalytic properties are considered in the light of a molecular model built using the crystal models of P. chrysosporium LiP (Piontek et al., 1993; Poulos et al., 1993) and MnP isoenzymes (Sundaramoorthy et al., 1994) and Coprinus peroxidase (Petersen et al., 1994) as templates.
Results and discussion
Characteristics of Pleurotus eryngii peroxidases
Peroxidases were isolated from liquid cultures of P. eryngii grown in low manganese glucose–peptone medium as described by Martínez et al. (1996b). Two electrophoretically homogeneous peaks, with an approximate ε406 of 150 mM−1 cm−1 and A410/A280 ratio of 450, were obtained in the last purification step (Fig. 2, top left). Both proteins, labelled MnPL1 and MnPL2, presented the same molecular mass, which was estimated as 37 kDa by matrix-assisted laser desorption and ionization time of fly (MALDI-TOF) mass spectrometry, and as 43 kDa by SDS–PAGE. However, they differed in the isoelectric point (3.70 and 3.65), which allowed their separation by ion exchange chromatography. In addition, their N-terminal sequences differed in two of the 17 aa residues sequenced, suggesting that proteins MnPL1 and MnPL2 were isoenzymes encoded by two different genes, as reported for peroxidases of other ligninolytic fungi (Gold and Alic, 1993).
The purified proteins MnPL1 and MnPL2 presented Mn2+ peroxidase activity (pH optimum 5), Mn-mediated activity on phenols (pH optimum 4) and Mn-independent activities on phenolic and non-phenolic aromatic substrates and dyes, even in the presence of EDTA (pH optimum 3). Therefore, the catalytic cycle illustrated in 1Fig. 1C is proposed. This new type of peroxidase is not produced only by P. eryngii. The authors identified similar enzymes in Pleurotus pulmonarius (Camarero et al., 1996) and Pleurotus ostreatus (Sarkar et al., 1997), the latter being different from the peroxidase studied by Asada et al. (1995), as well as in Bjerkandera adusta (Heinfling et al., 1998a,b). A similar enzyme has been recently purified from Bjerkandera sp. (Mester and Field, 1998).
The steady-state kinetic constants for different substrates of the two P. eryngii peroxidase isoforms were very similar. They showed high affinity for H2O2 with an apparent Km of 6–9 μM. Among the enzyme-reducing substrates, the highest affinity is for substituted hydroquinones, Mn2+ and some dyes (apparent Km values below 20 μM), whereas efficient oxidation of 2,6-dimethoxyphenol and the LiP substrate veratryl alcohol require higher substrate concentrations (apparent Km values around 0.3 and 3 mM respectively). A comparison of the different reducing substrates of this P. eryngii peroxidase has been recently reported (Heinfling et al., 1998b). No typical LiP was detected in culture filtrates of Pleurotus species, and no hybridization was obtained with DNA probes carrying the P. chrysosporium gene lpo (Huoponen et al., 1990). This suggests that the polyvalent peroxidase described here, which can oxidize Mn2+ directly, as well as phenolic and non-phenolic aromatic substrates (the former inhibiting LiP activity), could play an important role in preferential degradation of lignin by this fungus.
Peroxidase cloning and description of mnpl nucleotide and amino acid sequences
The cloning strategy described in Fig. 2 was used. Partial N-terminal sequences of protein MnPL1 and two internal peptides were used to design polymerase chain reaction (PCR) primers, which amplified fragments of ≈1 and 0.4 kb from P. eryngii DNA and cDNA respectively. Nucleotide sequencing of the amplification products revealed partial identity with already known genes of other ligninolytic peroxidases and coincidence with previously obtained amino acid sequences. Therefore, these fragments were used to screen genomic and cDNA libraries. Nucleotide sequencing of positively hybridizing clones showed two different sequences encoding, respectively, proteins MnPL1 and MnPL2, as revealed by coincidence with their N-terminal amino acid sequences. cDNA clones corresponding to the two different DNA sequences were obtained, both revealing a single open reading frame. The fact that the two different DNA clones gave positive hybridization with the MnPL1 probes indicated high identity. This was confirmed by complete sequencing revealing 99% identity over 4000 nucleotides (nt). Moreover, the identical patterns obtained after partial SalI digestion of DNA fragments of around 17 kb from MnPL1 and MnPL2 clones (in λEMBL3) and hybridization with MnPL1 probe, strongly suggested that the two proteins purified from cultures of dikaryotic P. eryngii were encoded by two allelic forms (mnpl1 and mnpl2) of the same gene mnpl. Until now, we had failed to obtain monokaryotic cultures showing allelic segregation, but a single peroxidase with the same N-terminal sequence of P. eryngii allelic variant MnPL1 and similar catalytic properties has been found in cultures of related P. pulmonarius and P. ostreatus (Camarero et al., 1996; Sarkar et al., 1997).
A comparison of mnpl1 and mnpl2 sequences is shown in Fig. 3. The most noteworthy characteristics of the gene mnpl and its flanking regions are also indicated in this figure. The latter include general transcription and putative regulatory elements (related to nitrogen and carbon regulation, response to metals and oxidative stress). The mnpl gene is made up of 1911 nt (same length in both alleles) and has 15 introns, some of them separated by very short exons. Some unusual intron processing sequences, compared with other fungal genes (Johansson and Nyman, 1996), are indicated in Fig. 3. Both mnpl alleles encoded 361 aa proteins, and a 30 aa signal peptide was identified before the known N-terminal sequences of the two mature proteins, after a KR sequence reported to be involved in protease processing. Thirty-nine amino acid residues matched the sequences previously obtained by N-terminal sequencing. A cDNA encoding 273 aa, which seems to correspond to an alternative 5′ splicing of intron XII, was also isolated and sequenced (248–273 predicted sequence: TTNRRFRTVSLLPCRRSLFLAKTRPN) but the protein product was not found. The molecular mass, from sequence 34553 for protein MnPL1 and 34581 for protein MnPL2, agrees with the estimation using MALDI-TOF after deducing 5–7% carbohydrate content. A potential N-glycosylation site (NXS/T) exists in peroxidase MnPL, in agreement with the presence of around 3% N-linked carbohydrate content. Direct evidence for N-glycosylation of P. chrysosporium LiP and MnP at the sites predicted (N257 and N131 respectively) has been obtained by X-ray diffraction. The potential N-glycosylation site in P. eryngii peroxidase MnPL (N96) is at a different position, matching with a potential glycosylation sequence in T. versicolor, P. ostreatus and B. adusta peroxidases (see Fig. 4 for GenBank references).
A detailed description of the two alleles of the mnpl gene (99% identity) and the corresponding proteins was possible after sequencing several clones from the DNA and cDNA libraries. With a single exception, all differences affected a single nucleotide (in the form of 33 one-nucleotide substitutions, a single one-nucleotide deletion, and a single one-nucleotide insertion). The existence of very similar ‘variation levels’ in 5′ (837 nt) and 3′ (1261 nt) flanking regions (1.0%), introns (1.0%) and exons (0.5%) confirmed that the two sequences correspond to alleles. The most significant differences between the two alleles are (see Fig. 3) (i) a 7 nt deletion, affecting 5 nt of a putative metal response element (MRE); and (ii) three one-nucleotide substitutions resulting in three different amino acid residues in the mature protein (two of them in the N-terminus).
Elements of response to metals (MRE), as well as to heat shock (HSE), have been reported in mnp genes (Gold and Alic, 1993). The MRE deletion suggests that only mnpl2 could be regulated by metals in P. eryngii. However, peroxidase production in liquid cultures of P. eryngii is strongly inhibited by Mn2+ concentrations over 10 μM (Martínez et al., 1996b). Differential regulation by Mn2+ of genes encoding MnP isoenzymes has been reported in P. chrysosporium (Gettemy et al., 1998). Studies on regulation of P. eryngii peroxidase MnPL by H2O2 and other active oxygen species, peptone and Mn2+ are currently in course.
On the other hand, the substitution of one nucleotide in the 3′-flanking region is probably responsible for the 30 nt 3′ displacement of the RNA polyadenylation site observed in mnpl1. Alleles of genes encoding other ligninolytic peroxidases have been isolated from P. chrysosporium (Gold and Alic, 1993). However, in the present paper we purified and characterized for the first time both the gene and the protein allelic variants.
Comparison with other fungal peroxidases
A noteworthy characteristic of P. eryngii mnpl is the high number of introns, much higher than found in all the other fungal peroxidases. Only the P. ostreatus peroxidase gene isolated by Asada et al. (1995) has similar intron number and position, although some differences exist in introns I and III. Partial coincidence in intron position was also found with the T. versicolor gene pgv (Jönsson et al., 1994). The P. eryngii peroxidase MnPL is the smallest fungal peroxidase described (331 aa mature protein), whereas P. chrysosporium (357–358 aa) and P. subvermispora MnP (364 aa) are the largest ones. The predicted amino acid sequence of the P. eryngii peroxidase MnPL presents the highest identity with the unpublished sequence of protein PS1, a second Mn-oxidizing peroxidase isolated from wheat straw degraded by this fungus (Martínez et al., 1996a). However, the identity degree between both peroxidases (74%) is lower than found between the three MnP isoenzymes of P. chrysosporium (83–84%). It is interesting that the identity between the peroxidase MnPL and the MnP isoenzymes of P. chrysosporium (55%) is lower than found with the LiP of the same fungus (60% and 58% with isoenzymes H2 and H8 respectively). This could justify the reaction of P. eryngii peroxidases with antibodies against P. chrysosporium LiP reported by Orth et al. (1993). However, differences between P. eryngii and P. chrysosporium peroxidase genes were shown by the lack of hybridization of P. chrysosporium DNA with a probe including a P. eryngii peroxidase gene compared with cross-hybridization between P. eryngii and P. ostreatus (Sarkar et al., 1997).
A comparison of the different fungal peroxidases in terms of mature protein sequences is shown in the dendrogram of Fig. 4. During multiple alignment, it was observed that a total of 58 amino acid residues of P. eryngii peroxidase MnPL were invariant in all peroxidases examined. It is interesting that alanine and valine residues at the positions corresponding to A4 and V16 of P. eryngii allelic variant MnPL1 exist in other peroxidases but not aspartate and isoleucine residues as found in the allelic variant MnPL2. The clustering obtained evidenced that C. cinereus peroxidase is unrelated to ligninolytic peroxidases. As the latter shows 99% identity with a crystallized peroxidase from Arthromyces ramosus (ARP) (Kunishima et al., 1994), an invalid species that could correspond to a Coprinus anamorph, the name ARP-CIP is used here. The other peroxidases can be placed in three main groups: (i) P. chrysosporium and P. subvermispora MnP; (ii) P. eryngii, P. ostreatus and T. versicolor peroxidases; and (iii) P. chrysosporium, P. radiata and B. adusta LiP. The recently sequenced MnP from P. subvermispora (Lobos et al., 1998) presents some Mn-independent activity, but the present comparison reveals that its whole protein sequence is related to those of typical MnP from P. chrysosporium (group i), and that this is the peroxidase with the lowest sequence identity (only 49%) with the P. eryngii peroxidase described here. Group iii includes four LiP-H8 proteins clustered together with the product of the P. chrysosporium gene lip-glg6, which probably corresponds to another H8-type isoenzyme. This group clusters together with group ii, revealing higher affinities of Pleurotus and Trametes peroxidases with P. chrysosporium LiP than MnP. Finally, it is interesting to mention that proteins PGV and PGVII are still to be isolated, but pgv was reported as a new gene having both lip and mnp characteristics (Jönsson et al., 1994).
Molecular modelling of Pleurotus eryngii Mn2-oxidizing peroxidase
To explain their polyvalent catalytic properties, molecular models for proteins MnPL1 and MnPL2 were built, using as templates the four crystal models of fungal peroxidases in class II currently available (49–60% amino acid identity with the P. eryngii peroxidase). They include LiP-H8, LiP-H2 and MnP1 from P. chrysosporium, and Coprinus peroxidase (ARP-CIP). The refined model for peroxidase MnPL (the two variants present only minor differences) has good geometry with a root mean square (rms) deviation from ideality in bond distances and angles of 0.016 Å and 1.60°, respectively. A schematic representation of the model for the allelic variant MnPL1 is shown in Fig. 5. Homology models for other peroxidases, including P. chrysosporium LiP (Du et al., 1992) and MnP (Johnson et al., 1994), had been reported, and they provided useful information on enzyme architecture and spatial arrangement of amino acid residues.
The molecular model of P. eryngii peroxidase includes (i) 12 helices named, according to cytochrome c peroxidase nomenclature (Finzel et al., 1984), as helices A (A12-N27), B (E36-G51), B′a (S64-A67), B′b (D69-E72), C (I81-H95), D (A99-N113), E (V145-A155), F (P159-I171), G (S195-E200), H (E229-R236), I (A241-M247) and J (E251-A266) (and four 2 aa sheets); (ii) four disulphide bonds (C3:C15, C14:C278, C34:C114 and C242:C307), one less than P. chrysosporium MnP1; (iii) the haem pocket, which also includes characteristic distal- (R43, F46 and H47, involved in H2O2 reaction) and proximal-side residues (H169, F186 and D231; Nε of H169 acting as a fifth Fe3+ ligand) conserved in all fungal peroxidases with the only exception of ARP-CIP, which has a leucine at the position of F186 of P. eryngii peroxidase; (iv) two Ca2+ binding sites involving oxygen from D48, G60, D62 and S64 (distal Ca2+ linking helices B and B′a), and S170, D187, T189, V192 and D194 (proximal Ca2+ linking helix F to loop between helices F and G), which are the same residues of ARP-CIP Ca2+ sites but differ from those of P. chrysosporium LiP and MnP in the V192 position; and (v) a putative Mn2+ binding site, which involves E36, E40 and D175 and the propionate of haem ring III (the ‘internal’ propionate), and enables Mn2+ fixation and electron transfer to compound I porphyrin ring or to compound II Fe4+ (see Fig. 1 for catalytic cycle).
All peroxidases from classes I and II (Welinder, 1992) share a general tertiary fold and helical topography (Banci, 1997). The molecular model built for the P. eryngii peroxidase (allelic variant MnPL1) was superimposed on the crystal models of P. chrysosporium MnP and LiP. The rms distances between Cα were 1.24 Å with LiP-H8 (329 residues computed) and 1.49 Å with MnP1 (324 residues). This indicates high structural affinity with LiP-H8, taking into account that the rms distance between LiP isoenzymes H2 and H8 is 1.1 Å. The most significant differences in global architecture are shown in Fig. 6: (i) the long C-terminal tail of P. chrysosporium MnP1 (right, top arrow), being 14 residues shorter in LiP-H8 and 17 residues shorter in peroxidase MnPL; (ii) the loop of seven residues (right-bottom arrow) in P. chrysosporium MnP1 (L228-T234); (iii) the internal loop around F301 of P. chrysosporium LiP-H8 (bottom), being shorter in P. chrysosporium MnP1 and lacking in P. eryngii peroxidase MnPL; (iv) the loop of six residues (left, top arrow) in P. chrysosporium LiP-H8 (E57-F62), which is more reduced in P. chrysosporium MnP1 and absent in P. eryngii peroxidase MnPL. This region includes some helical conformation in LiP-H8 and was described as an additional helix in ARP-CIP. Therefore, the last two peroxidases and that from P. eryngii could include 12 major helices compared with 11 helices in P. chrysosporium MnP1. However, the position of the additional helices is different (see Fig. 5): B′a and B′b of peroxidase MnPL correspond to helix B′ of LiP-H8, interrupted by a turn at F68, whereas the additional helix in ARP-CIP is between helices B and B′ of LiP-H8.
A superimposition of side-chains of the haem pocket and Mn2+ binding residues, and the C-terminal backbone of the P. eryngii peroxidase and P. chrysosporium LiP-H8 and MnP1 is shown in Fig. 7. Small differences in proximal-side residues were observed, including displacement of aspartate and rotation of the histidine ring. Moreover, the orientation of the benzenic ring of proximal phenylalanine, which has been reported to affect accessibility of small molecules to the haem ring (Kishi et al., 1997), of the P. eryngii peroxidase is similar to that found in P. chrysosporium MnP1. In the model for P. eryngii peroxidase MnPL, Mn2+ could be co-ordinated by carboxylate oxygens of E36, E40 and D175 and the internal propionate of haem. A similar binding site, including two water molecules to complete Mn2+ hexa-co-ordination, was postulated in both theoretical (Johnson et al., 1994) and crystal models (Sundaramoorthy et al., 1994) of P. chrysosporium MnP1, and was then confirmed by site-directed mutagenesis (Sundaramoorthy et al., 1997). LiP from P. chrysosporium presents different residues at these positions (see Fig. 7), resulting in no Mn2+ binding. Moreover, in the model for the P. eryngii peroxidase, the carbonyl oxygen of A174 appears to be rotated 90° with respect to A178 of P. chrysosporium MnP1 (not shown) and in a spatial position compatible with octahedrical co-ordination of Mn2+. It has been reported that R177 also contributes to Mn2+ oxidation by P. chrysosporium MnP1 (Gold et al., 1998) and suggested that this could be due to (i) displacement of the C-terminal tail away from the Mn2+ binding site; or (ii) formation of a H bond, which orients E35 towards Mn2+. The efficient oxidation of Mn2+ by the P. eryngii peroxidase, despite the presence of a small residue (A173) at this position, could be related to its short C-terminal tail (see Fig. 7) or to a favourable position of E36 in the absence of the above H-bond.
The molecular models suggest that Mn2+ access is produced via a small channel situated directly on the internal propionate of haem. This is different from access through the main haem channel and interaction at haem C20 (δ-meso) as initially suggested (Harris et al., 1991). A comparison of amino acid residues at the opening of the main haem channel of four fungal peroxidases is presented in Fig. 8. In this region the existence of a specific veratryl alcohol binding site has been suggested in P. chrysosporium LiP (Poulos et al., 1993), but an aromatic binding site in P. eryngii peroxidases should present analogies with that of horseradish peroxidase (Gajhede et al., 1997), because of its wider substrate specificity. Therefore, the existence of an apolar environment at the haem channel could be a requirement for binding aromatic substrates. This could involve aromatic residues such as F148 in P. chrysosporium LiP-H8 and F142 in the P. eryngii peroxidase, occupying the same position of horseradish peroxidase F142 reported to affect substrate binding (Veitch et al., 1995). It is interesting that this phenylalanine residue is absent in P. chrysosporium MnP1, and two polar residues (Q145 and N81) occupy the positions of MnPL F142 and A79.
The oxidation of veratryl alcohol and other aromatic substrates by LiP has been suggested to be produced in close contact with the haem or via long-range electron transfer (Schoemaker et al., 1994), as reported for cytochrome c oxidation by cytochrome c peroxidase (Pelletier and Kraut, 1992). Two histidine residues, H82 and H239, have been suggested as the origin of two electron transfer pathways in LiP, via distal and proximal histidines respectively (Schoemaker et al., 1994; Jojima et al., 1998), and similar pathways have been identified in the P. eryngii peroxidases PS1 (from H82) (S. Camarero, S. Sarkar, F. J. Ruiz-Dueñas, M. J. Rartínez and A. T. Martínez, unpublished) and MnPL (from H232). Recently, the involvement of W171 in LiP catalytic cycle, via another long-range electron transfer to haem, has been proposed (Blodig et al., 1998). This residue, which is absent from P. chrysosporium MnP and ARP-CIP, had been already suggested as involved in a hypothetical aromatic substrate binding site after homology modelling of LiP (Du et al., 1992). The presence of W164 occupying the same position in the model of the P. eryngii enzyme described here constitutes another ‘LiP-type’ characteristic of this Mn2+-oxidizing peroxidase. Additional studies, including site-directed mutagenesis, are in progress to understand better the catalytic properties of this new peroxidase and its role in lignin degradation.
Organisms and growth conditions
Pleurotus eryngii ATCC 90787 (= IJFM A169), obtained from dikaryotic mycelium of a fruit body cap, was grown in glucose–peptone or glucose–ammonium media (Martínez et al., 1996b) at 28°C and 180 r.p.m. Escherichia coli DH5α (Life Technologies) was used for cloning and plasmid propagation, XL1-Blue MRA and MRA-P2 (Stratagene) for screening of the genomic library and L87 and NM514 (Amersham) for the cDNA library. They were grown in LB or NZY media (Sambrook et al., 1989). DNA probes and clones from libraries were subcloned in pBluescript SK (+/−) (Stratagene).
MnP activity was estimated by formation of Mn3+–tartrate complex (ε238 6500 M−1 cm−1) from 0.1 mM MnSO4 in 0.1 M sodium tartrate (pH 5) using 0.1 mM H2O2; or by Mn2+-dependent oxidation of 0.1 mM 2,6-dimethoxyphenol under the same conditions (ε469 27 500 M−1 cm−1) (Martínez et al., 1996b). Mn2+-independent activities on 5 mM 2,6-dimethoxyphenol and veratryl alcohol (the latter measured as veratraldehyde, ε310 9300 M−1 cm−1) were estimated in 0.1 M sodium tartrate buffer (pH 3). The effect of 1 mM disodium salt of EDTA on these activities was investigated. One activity unit was defined as the amount of enzyme oxidizing 1 μmol of substrate per min.
Peroxidase purification and characterization
Manganese-oxidizing peroxidases were purified from cultures in glucose–peptone medium as described elsewhere (Martínez et al., 1996b). After concentration and molecular-size exclusion chromatography, two proteins with peroxidase activity on Mn2+ were separated in a Mono-Q column using 10 mM sodium tartrate buffer (pH 5) and a 0–0.25 M NaCl gradient (20 min at 0.8 ml min−1). Proteins were N-deglycosylated with 125 mU ml−1 of Endo-H from Boehringer. Protein concentration was estimated with Bradford reagent, and carbohydrate with anthrone reagent. SDS–PAGE of native and deglycosylated protein was performed in 12% polyacrylamide gels, and isoelectric focusing in 5% polyacrylamide gels with a pH range of 2.5–5.5 (protein bands were stained with AgNO3 or Coomassie blue R-250). In addition to SDS–PAGE, the protein mass was estimated by MALDI-TOF mass spectrometry (Bruker) in sinapic acid matrix, and by Superose-12 chromatography. The amino acid composition was determined using a Biochrom-20 autoanalyser (Pharmacia) after hydrolysis of 20 μg of protein with 6 M HCl at 110°C. Samples (200 μg) of deglycosylated protein were hydrolysed with trypsin (10 μg), and peptides were separated in a C18 column using 0–70% acetonitrile gradient in 0.1% trifluoroacetic acid (1 ml min−1). The N-terminal sequences of whole peroxidase and tryptic peptides were obtained by automated Edman degradation of 5 μg of protein in an Applied Biosystems 494 pulsed-liquid protein sequencer.
DNA and RNA extraction
Six-day-old mycelium from glucose-ammonium culture was homogenized in liquid N2, suspended in lysis solution, and DNA extracted with phenol–chloroform–isoamylic alcohol and treated with ribonuclease (González et al., 1992). RNA was obtained from 5-day-old cultures in glucose–peptone medium, using the Ultraspec RNA isolation system (Biotech).
Primer design and preparation of DNA probes
Oligonucleotides were synthesized (Beckman-oligo 1000 M synthesizer), corresponding to N-terminal sequences of the mature protein and two tryptic peptides. PCR was used to prepare the DNA probe and reverse transcriptase (RT)-PCR to prepare the cDNA probe used for library screening. PCR reactions were carried out in 100 μl, using 0.1 μg of DNA, 400 pmol of each primer and 2.5 U of Amplitaq (Perkin-Elmer Cetus), and 55°C as annealing temperature. For RT-PCR, 0.5 μg of total RNA, 1 U of RT and oligo-14 as primer (in 20 μl), were used to obtain the first DNA strand for PCR amplification. The PCR products were separated on 0.8% agarose gels in TAE buffer and purified (Geneclean BIO101 kit) before cloning.
Preparation and screening of genomic and cDNA libraries
Genomic DNA was partially cleaved with Sau3A (Boehringer), 10–23 kb fragments were separated by sucrose gradient centrifugation (Sambrook et al., 1989), 0.175 μg were ligated to 1 μg of BamH1 predigested λEMBL3 arms using vector kit, and the ligation mixture packaged using Gigapack II Gold (Stratagene). mRNA was purified from total RNA using oligo(dT)-cellulose (Pharmacia-Biotech kit). For preparation of first and second cDNA strands the module RPN 1256 (Amersham) was used. Adaptors with EcoRI cohesive ends were added to the cDNA blunt ends using an adaptor-ligation module (Amersham), the ‘adapted’ cDNA was ligated with λgt10 arms using module-λgt10 (Amersham), and products were packaged as described for DNA. The genomic and cDNA libraries were titred, and screened by plaque hybridization using the 32P-labelled DNA probes previously described.
DNA sequencing and sequence analysis
Both DNA chains were sequenced using an automatic sequencer ABI 377. Synthetic oligonucleotides were used as primers in automatic sequencing. The GCG package, and BLAST and EXPASY programs (in NCBI and SWISSPROT http servers) were used for analysis, alignment and comparison of sequences. A dendrogram of affinities between sequences of mature peroxidases was obtained using the ‘unweighted pair group method with arithmetic mean’ clustering from the Kimura distance matrix from progressive pairwise multiple alignment of sequences.
Molecular modelling of P. eryngii proteins, MnPL1 and MnPL2 (without signal peptides), by sequence homology was performed using the PROMOD program, and refined by CHARMM (Peitsch, 1996). It was based on alignment of the four fungal peroxidases for which crystal models are available, i.e. LiP-H8 (Broohaven PDB entries 1LGA and 1LLP), LiP-H2 (1QPA) and MnP1 (1MNP) from P. chrysosporium, and Coprinus ARP-CIP (1ARP). The haem pocket and Mn2+ binding site modelling was completed by additional molecular dynamics refining using X-PLOR ver. 3.1. Helices were determined using the DSSP program (Kabsch and Sander, 1983). The FRODO, RASMOL and SWISS-PDBVIEWER programs were used for model comparison and presentation.
The authors thank S. Sarkar for information on the unpublished gene ps1 and, together with E. Varela and L. Botella, for contributions to obtain the mnpl probe; A. Romero for molecular dynamics of models for peroxidase MnPL; M. A. Peñalva for critical reading of the manuscript; and A. Guijarro and T. Raposo for skilful technical assistance. J. Varela contributed to amino acid analysis and protein sequencing, A. Díaz to DNA sequencing and A. Prieto to MALDI-TOF analysis. This work was supported by the EU-contract AIR2-CT93-1219 and the Spanish Biotechnology Programme.