Tài liệu Báo cáo khoa học: Phylogenetic relationships in class I of the superfamily of bacterial, fungal, and plant peroxidases pptx

Phylogenetic relationships in class I of the superfamily of bacterial, fungal, and plant peroxidases Marcel Za ´ mocky ´ Institute of Molecular Biology, Slovak Academy of Sciences, Bratislava, Slovakia Molecular phylogeny among catalase–peroxidases, cyto- chrome c peroxidas es, and ascorbate peroxidases was ana- lysed. Sixty representative sequences covering all known subgroups of class I of the superfamily of bacterial, fungal, and plant heme peroxidases were selected. Each sequence analysed contained the typical p eroxidase motifs evolved to bind effectively the prosthetic heme group, enabling per- oxidatic activity. The N-terminal and C-terminal domains of catalase–peroxidases matching the ancestral tandem g ene duplication event were treated separately in the phylogenetic analysis to reveal their specific evolutionary history. The inferred unrooted phylogenetic tree obtained by three dif- ferent methods revealed the existence of four clearly separated c lades ( C-terminal and N-terminal d omains of catalase–peroxidases, ascorbate peroxidases, and cyto- chrome c peroxidases) which were segregated early in the evolution of this superfamily. F rom the results, it is obvious that the duplication e vent in the g ene f or catalase–peroxid ase occurred in the later phase of evolution, in which the indi- vidual specificities of the peroxidase families distinguished were already formed. Evidence is presented that class I of the heme peroxidase superfamily is spread among prokaryotes and eukaryotes, obeying the birth-and-death p rocess of multigene family evolution. Keywords: ascorbate peroxidase; catalase–peroxidase; cyto- chrome c peroxidase; birth-and-death process; lateral gene transfer. Heme peroxidases are very abundant enzymes present in all living forms. These oxidoreductases are involved in a wide array o f physiological p rocesses, the m ost important of which are involved in the response to various forms of oxidative stress [ 1,2]. Attention was mainly drawn to the family of catalase–peroxidases by the representative enco- ded by KatG in Mycobacterium tuberculosis,whichis capable of oxidative activation of isoniazid (isonicotinic acid hydrazide) [3], still the most widely used antitubercu- losis drug. All heme peroxidases have important features of their catalytic mechanism in common. After their initial oxidation w ith a molecule of hydrogen peroxide, they oxidize from the reactive intermediate known as compound I a wide variety of substrates according to the simplified reaction scheme: H 2 O 2 +2AH fi 2H 2 O+2A. The d etailed reaction mechanism a nd substrate specificity of numerous peroxidases have been investigated for decades, and a large amount of experimental data has accumulated (e.g [4], for r eview). It was suggested that similar heme- containing peroxidases w hich are very abundant in plants, fungi and some bacteria should constitute the plant peroxi- dase superfamily [5]. They were further classified into three subclasses according to their cellular localization and function. All representatives possess the same heme pros- thetic group containing high-spin f erric iron, so the reaction specificity is a pparently determined by the protein surround- ings of the heme. Catalase–peroxidases, which belong to class I , are the only group of this superfamily that possess notable catalase activity (i.e. they can oxidize and reduce hydrogen peroxide; see [6] f or details). All other members of the superfamily can only reduce hydrogen peroxide with subsequent oxidation of a secondary substrate. These ÔnoncatalaseÕ members of class I exhibit strong specificity for electron donors: the preferred substrate is ascorbate in the case of a scorbate peroxidases and cytochrome c for cytochrome c peroxidases. Several crystal structures of heme peroxidases have been solved, now covering all subgroups of this superfamily. In clas s I, the crystal structure of cytochrome c peroxidase (CCP) from Saccharomyce s cerevisiae [7] a nd ascorbate peroxidase (APX) from Pisum sativum [8] is known; the former has served as a benchmark for peroxidase structures for two decades. APX was already crystallized in a complex with its substrate [9]. After many unsuccessful attempts, the crystals of several catalase– peroxidases we re also obtain ed ( e.g. [10,11]). Recently, the structure of catalase–peroxidase from the halophilic arch- aeon Haloarcula marismortui was solved t o high resolution [12], and this was followed b y the highly resolved structure o f KatG from Burkholderia pseudomallei [13]. The phylogenetic relations of heme peroxidases have only been analysed to a certain extent: the evolutionary analysis of the mammalian peroxidase superfamily has been per- formed, and even a prokaryotic member has been detected [14]. The phylogenetic relations i n t he plant peroxidase Correspondence to M. Za ´ mocky´ , Institute of Molecular B iology, Slovak Academy of Sciences, Du´ bravska ´ cesta 21, SK-845 51 Bratislava, Slovakia. Fax: + 4212 59307416, Tel.: + 4212 59307441, E-m ail: umikm zam@savba.sk Abbreviations: APX, ascorbate pe roxidase; CCP , f ungal c yto- chrome c peroxidase; CP, catalase–peroxidase; CPn , N-t erminal domain of a c atalase–peroxidase; CPc, C-terminal do main o f a catalase–peroxidase; KatG, gene for cat alase–peroxidase; NJ, neighbor-joining. (Received 22 April 2004, revised 1 0 June 2004, accepted 2 1 June 2004) Eur. J. Biochem. 271, 3297–3309 (2004) Ó FEBS 2004 doi:10.1111/j.1432-1033.2004.04262.x superfamily have been analysed only partially: the common phylogeny of catalase–peroxidases and APXs have been outlined [15], and a dendrogram of 29 lignin a nd manganese peroxidases have been presented [16]. The present study should contribute to our understanding of the possible modes of evolution of multigene families. In principle, two possible s chemes have been suggested for this type o f phylogeny: (a) concerted evolution and (b) evolution by a birth-and-death process. In the first case, multigene families arise i n the genomes after gene duplications by the mechanisms of unequal crossing over and gene conversion, followed b y natural selection [17]. In the second case, new genes are created by repeated gene duplication; some are maintained in the genome for a long time, whereas others are d eleted or become nonfunctional [18–20]. It is well documented that almost all KatGs contain two fused copies of the primordial peroxidase gene [5,15]. The copy trans- lated into the N-terminal domain participates in catalysis and possesses t he prosthetic heme group, but the c atalytic function of the C-terminal domain is not apparent [12], and thus the role of the corresponding part of the gene is also unknown. These g ene f usions togeth er with single-copy genes of APXs and CCPs are ideal for investigating the evolution of a widespread multigene family. H ence, the most probable evolutionary route leading to clades of extant heme peroxidases will help to explain the occurrence and function of multigene families present in both prokaryotic and eukaryotic genomes. Experimental Procedures Sequence data All protein sequences used in this study were obtained from the U niProt database and are listed in Table 1 together with their accession numbers and the organisms from which they originate. The protein sequences were used to infer the phylogenetic relationships. Both codon usage bias of analysed sequences ranging from a rchaea to high er plants and the presence of introns in only some of the members analyzed (plant APXs) can cause serious problems with analyzing the DNA sequences directly. Owing to the currently unequal availability of the sequences of the proposed groups of heme peroxidases (known from previ- ous analysis in [15]), only representative sequences from each group and f rom each kingdom were selected for a statistically equilibrated phylogenetic analysis. All 34 cata- lase–peroxidases analysed here were div ided into N -terminal and C-terminal domains because of the apparent tandem gene-duplication event reporte d previously [5,15]. The border between the domains was easily discernible because of conserved residues a nd motifs present i n all known KatGs. All sequenced N-terminal domains are l onger (average length 430 amino acids) because o f several insertions not present in C-terminal domains (309 amino acids on average [6]). Twenty-one APXs, from red algae to higher plants, were selected for the analysis. APX genes expressed both in cytoplasm and chloroplasts were chosen in equal amounts. Two APXs from Euglenozoa were also included. The 25 amino acid-long fragment of APX from bovine eye (accession No. PC4445) could not be used in this analysis. This N-terminal stretch is insufficient in length and of rather unclear origin. M oreover, no other homologs of APX are known in the whole kingdom Animalia. Besides the w ell-known S . cerevisiae CCP s equence, two additional ascomycetous CCP sequences (as putative ORFs f rom sequencing projects) were also included in this study. No homologs of y east CCP from other kingdoms are known, and, for example, bacterial CCPs belong to a d ifferent protein family. Multiple sequence alignments Multiple sequence a lignments of catalase–peroxidases, and of APXs with CCPs, were perf ormed using CLUSTALX [21]. In the case of catalase–peroxidases, two partial alignments were performed, for the N-terminal and C-terminal domain. Suitable parameters for all three partial alignments were: gap opening penalty, 10.0; gap extension penalty, 0.2; and gap separation distance, 8. The Blosum 62 series protein– weight matrix was used in all three cases. These parameters were the same as those used for the first alignment of class I of the peroxidase superfamily [15]. Varying the gap opening penalty setting in the range 5.0–20.0 did not change the alignment output significantly. T he seq uence alignments were displayed with GENEDOC [22] and refined manually with respect to known structural homology. Profile alignments The profile alignment mode of CLUSTALX was used stepwise on the partial alignments. Firstly, the N-terminal domains of catalase–peroxidases were aligned with the prealigned group of APXs and CCPs where the known secondary- structure e lements were taken into account. This new profile was final ly used to align the group of C-terminal domains of catalase–peroxidases which share the lowest sequence similarity with other superfamily members in catalytically essential regions. Suitable parameters used for all profile alignments were: gap opening penalty, 10.00; gap extension penalty, 0.1; Blosum 30 protein–weight matrix; helix and strand gap p enalty, 4; a nd loop gap penalty, 1. Finally, a reas of extensive gaps (i.e. longer than 10 amino acid positions and present in more than 90% of sequences) were omitted from the entire alignment to prevent long-branch attraction in the following procedures. Phylogenetic analysis The profile alignment used for the phylogenetic analysis comprised 9 4 sequences (each KatG divided in t he two corresponding domains) and a total length of 398 amino- acid positions. Three different phylogenetic methods were applied. First, the phylogenetic relationships were inferred using the neighbor-joining (NJ) method selected from the package MEGA [23]. T he following parameters were used: the Poisson correction of substitutions; the option o f Ôcomplete deletionÕ for handling gaps; and 100 bootstrap replications as a test of inferred phylogeny. The resulting unrooted tree topology was visualized in the TREE EXPLORER . The s ame profile alignment of 94 sequences was subjected to the bootstrap p rocedure of the PHYLIP package [24]. After 100 bootstrap cycles, the data s et was s ubjected to the 3298 M. Za ´ mocky´ (Eur. J. Biochem. 271) Ó FEBS 2004 Table 1. Sequences of enzymes used in this study. Abbreviations for all peroxidase s included in this evolutionary an alysis, with their accession numbers from the UniProt database and organisms from which they originate. In the case o f catalase–peroxidases, the parts coding for the N- terminal and C-terminal domains of the co rresponding genes (KatG) w ere t rea ted se pa rately. S eq uence data for Candida al bicans was o btained from the Stanford Genome Technology Center website at h ttp://www-sequence.stanford.edu/group/candida. Abbreviation Accession number Enzyme Organism (strain) ArathaAPXc Q05431 Ascorbate peroxidase 1 Arabidopsis thaliana ArathaAPXt Q42593 Ascorbate peroxidase (thylakoid) Arabidopsis thaliana ArchfulCP O28050 Catalase–peroxidase Archaeoglobus fulgidus AspefumCP Q7Z7W6 Catalase–peroxidase Aspergillus fumigatus AspenidCP Q96VT4 Catalase–peroxidase Emericella nidulans BacihalCP Q9KEE6 Catalase–peroxidase Bacillus halodurans BacisteCP P14412 Catalase I Geobacillus stearothermophilus BlumgraCP Q8 · 1 N3 Catalase–peroxidase Blumeria graminis BurkcepCP Q9AP06 Catalase–peroxidase Burkholderia cepacia BurkpseCP Q939D2, pdb: 1MWV Catalase–peroxidase Burkholderia pseudomallei CandalbCCP Contig19–10046* Cytochrome c peroxidase Candida albicans CapsannAPX Q84UH3 Ascorbate peroxidase Capsicum annuum CaulcreCP O31066 Peroxidase/catalase Caulobacter crescentus ChlamspAPX Q9SXL5 Ascorbate peroxidase Chlamydomonas sp. W80 ChlareiAPX O49822 Ascorbate peroxidase Chlamydomonas reinhardtii CucusatAPX Q96399 Ascorbate peroxidase (cytosolic) Cucumis sativus CucurcAPXt O04873 Ascorbate peroxidase (thylakoid-bound) Kurokawa Amakuri Cucurbita cv. DesulfiCP ZP_00096951 Catalase–peroxidase Desulfitobacterium hafniense E_coliHPI P13029 Catalase HPI Escherichia coli E_coliPCP P77038 EHEC-strain catalase peroxidase (strain 0157:H7) Escherichia coli EuglgraAPX Q8LP26 Ascorbate peroxidase Euglena gracilis FraganaAPX O48919 Ascorbate peroxidase Fragaria x ananassa GaldparAPX Q8GT26 Hybrid-type ascorbate peroxidase (Rhodophyta) Galdieria partita GeobactCP AAR35476 Catalase–peroxidase Geobacter sulfurreducens GloeobaCP Q7NGW6 Catalase–peroxidase Gloeobacter violaceus GlycmaxAPX Q43758 Ascorbate peroxidase 1 Glycine max GosshirAPX Q39780 Ascorbate peroxidase Gossypium hirsutum HalomarCP O59651, pdb: 1ITK Catalase–peroxidase Haloarcula marismortui HalosalCP Q9HHP5 Catalase–peroxidase Halobacterium salinarum LegipneCP Q9ZGM4 Catalase–peroxidase Legionella pneumophila LycoesAPXt Q8LSK6 Ascorbate peroxidase (thylakoid) Lycopersicon esculentum MesecryAPX Q42909 Ascorbate peroxidase Mesembryanthemum crystallinum MesolotCP Q987S0 Catalase–peroxidase Mesorhizobium loti MethaceCP Q8TS34 Catalase–peroxidase Methanosarcina acetivorans MycoforCP O08404 Catalase–peroxidase Mycobacterium fortuitum MycosmeCP Q59557 Catalase–peroxidase Mycobacterium smegmatis MycospeCP Q9R2E9 Catalase–peroxidase Mycobacterium vanbaalenii MycotubCP Q08129 Catalase–peroxidase Mycobacterium tuberculosis NcrassaCP Q8 · 182 Catalase–peroxidase Neurospora crassa Ncrassahyp Q7SDV9 Hypothetical protein Neurospora crassa NictabAPXc Q42941 Ascorbate peroxidase (cytosolic) Nicotiana tabacum NictabAPXt Q9XPR6 Ascorbate peroxidase (thylakoid-bound) Nicotiana tabacum OryzsatAPX P93404 Ascorbate peroxidase Oryza sativa PenimarCP Q8NJN2 Catalase–peroxidase Penicillium marneffei PisusatAPX P48534, pdb: 1APX Ascorbate peroxidase Pisum sativum PorpyezAPX Q7Y1X0 Ascorbate peroxidase (cytosolic) (Rhodophyta) Porphyra yezoensis PseuputCP Q88GQ0 Catalase–peroxidase HPI Pseudomonas putida KT2440 RhizlegCP Q8RJZ6 Catalase–peroxidase Rhizobium leguminosarum SacchceCCP P00431, pdb: 2CYP Cytochrome c peroxidase Saccharomyces cerevisiae ShewoneCP Q8EIV5 Catalase–peroxidase HPI Shewanella oneidensis SpinolAPXt O46921 Ascorbate peroxidase (thylakoid) Spinacia oleracea Ó FEBS 2004 Molecular evolution of heme peroxidases (Eur. J. Biochem. 271) 3299 pairwise protein distance calculation method in which the JTT protein matrix [24] was formed. This output was put in the F itch–Margoliash Ôleast sq uaresÕ phylogenetic tree estimation method, in which the search for the best trees was allowed. T n addition, global rearrangement of the sequence order after each cycle in the FITCH program was activated. The series of tree s prod uced was analysed by the Consense method to reveal the majority r ule c onsensus tree. This tree was visualized with the program TREEVIEW [25]. The maximum likelihood unrooted phylogenetic tree was also calculated using the program PUZZLE , version 5.0 [26]. The WAG model of amino acid substitution was applied [27]. Slow and accurate parameter estimation and 50 000 puzzling s teps were used on the set of data subjected to the above methods. The c-distribution of rate h eterogeneity with parameter estimation from the actual data set was used (value obtained for parameter Gamma ¼ 0.62). In total, 230 300 quartets were analysed, and an unrooted quartet puzzling tree was produced. This tree was also visualized with the program TREEVIEW [25]. The highest likelihood trees resulting from all three methods described were compared to arrive at the e xpected tree. Structural comparisons Experimental 3D co-ordinates of two catalase–peroxidases, one APX and one CCP, were obtained from the Protein Data Bank, R esearch Collaboratory f or Structural Bio- informatics, Rutgers University, New Brunswick, NJ, USA (http://www.rcsb.org). Their c odes a re mentione d i n Table 1 by the corresponding sequences. The secondary- structure elements of all structures used for comparison were ou tlined b y PDBSum (http://www.biochem.ucl.ac.uk/ bsm/pdbsum) [ 28]. The secondary-structure content was quantified from the resulting plots with the program PRO- MOTIF implemented in PDBsum. Results and discussion Conserved regions and typical motifs in the sequences of class I peroxidases Sixty heme peroxidases belonging to class I of the plant peroxidase superfamily were aligned with the option of profile alignment in CLUSTALX . The overall sequence s imi- larity is 28.5%, as calculated from the 398 amino acid-long alignment u sed for the phylogen etic analysis. The three most important sequence areas possibly involved i n the catalytic mechanism are presented in Figs 1–3. Region A is located on the distal side of the prosthetic heme group, and regions B and C are located on proximal side. The unambiguous sequence s imilarities in these regions can also be traced in the known 3D crystal structures of class I peroxidases presented in Fig. 4 for members of each group analysed. The greatest s equence conservation is achieved in the area on the distal side of the heme prosthetic group (known as peroxidase consensus p attern PS00436 in the P rosite database) surrounding the active s ite, where it reaches 76% (Fig. 1). The catalytic triad Arg92, Trp95, and His96 in HalomarCPn (abbreviations of all sequences analysed are listed i n T able 1) located i n t he distal heme cavity is invariantly conserved among all N-domains of catalase– peroxidases, all CCPs and all A PXs. Whereas the essential arginine (Arg92) and histidine (His96) are responsible for compound I formation [4], the latter allowing the heterolytic cleavage of the peroxide bond via acid-base catalysis [29], the coessential tryptophan (Trp95) facilitates the two- electron reduction of compound I by hydrogen peroxide [30]. Site-directed mutagenesis in E_coliHPIn [31], MycotubCPn [32], and SyncyspCPn [29], a s well a s in PisusatAPX [33] and SacchceCCP [34], supported the role of the catalytic triad by affecting the typical reactivity. The level of decrease in the peroxidase activity (in contrast with catalase activity) correlated with the ability o f the re spective mutants t o bind heme. Residues corresponding to the catalytic t riad are not conserved i n the C-domains of catalase–peroxidases which do not bind heme. The position corresponding to Arg92 (the numbering c orresponds to HalomarCP, in which the residues c an also be found; Fig. 4A,B) is variable in C-domains, but there a similar basic residue occurs (e.g. Lys465; Fig. 4B). The position corresponding to His96 is even more variable in all C-domains of the catalase–peroxidases investigated. In contrast, the positions Trp95 and Trp468 were invariantly conserved among all class I representatives except Mesecry- APX. From Fig. 1 it is obvious that the extension of the distal active site exhibits high sequence conservation, although lower than the region d irectly involved in the reaction with the peroxidic substrate. An essential aspara- gine (Asn126 of H alomarCPn) was located here, a nd its role is supported by a mutagenesis study [35]. The hydrogen- bonding network in which this residue is involved has subtle differences from that present in PisusatAPX [36] visible in Table 1. (Continued). Abbreviation Accession number Enzyme Organism (strain) StreretCP O87864 Catalase–peroxidase Streptomyces reticuli SyncyspCP P73911 Catalase HPI Synechocystis sp. (PCC6803) SynecspCP Q55110 Catalase–peroxidase Synechococcus sp. (PCC7942) TrypcruAPX Q8I1 N3 Ascorbate-dependent peroxidase Trypanosoma cruzi VibrchoCP Q9KRS6 Catalase–peroxidase Vibrio cholerae VignungAPX Q41712 Ascorbate peroxidase Vigna unguiculata XantcamCP Q8PBB7 Catalase–peroxidase Xantomonas campestris XylefasCP Q9PBB2 Catalase–peroxidase Xylella fastidiosa YerspesCP Q9X6B0 Catalase–peroxidase Yersinia pestis ZeamaysAPX Q41772 Ascorbate peroxidase Zea mays 3300 M. Za ´ mocky´ (Eur. J. Biochem. 271) Ó FEBS 2004 the sequence alignment (e.g. Asn121 in HalomarCPn and Glu65 in PisusatAPX). T he area arou nd the p roximal heme ligand (His259 in HalomarCPn, His163 in Pisusat APX, His175 in SacchceCCP; Fig. 2) is less conserved. The o verall sequence similarity around this iron ligand is only 48%. Nevertheless, the corresponding peroxidase consensus pat- tern PS00435 (Prosite database) is d iscernible in a ll the peroxidases analysed. The lower sequence similarity com- pared with the distal side can be e xplained by the fact that this rather variable region contributes significantly to the reaction specificity of the respective groups and therefore each family has its own typical feature in this region. The iron of the prosthetic heme group is invariantly co-ordinated by the a bove essential proximal h istidine (Fig. 4A,C,D). However, in all C-domains of catalase– peroxidases, there is a conserved arginine (Arg622 in HalomarCPc) in t he corresponding position, indicating that these domains lost their ability to co-ordinate the heme. Further, a conserved tryptophan (Trp311 in HalomarCPn, Trp179 in PisusatAPX, and Trp191 in SacchceCCP) is thought to participate in an important hydrogen-bond network on the proximal side o f the heme. This residue is not conserved in all C-domains of catalase–peroxidases and some APXs (e.g. position P he171 in MesecryAPX), sup- porting the theory that it is not essential for the reaction mechanism o f A PXs [33]. In contrast, for CCPs, T rp191 has been suggested t o be t he site of the free-radical formation of the c orresponding CCP compound I [ 37]. Site-directed mutagenesis was performed in the proximal heme cavity of SyncyspCPn [38] and SacchceCCP [34] and focused on the function of the two residues. In mutated catalase–peroxid- ases, substitutions i n both residues had a pronounced effect: a decrease in activity and loss of the prosthetic heme group. Similarly to t he distal heme region, close to the essential His and Trp, there are h ighly c onserved positions (Fig. 2) among all the sequences investigated with unknown func- tion. Even though the C-domains of the catalase–peroxid- ases had lost the ability to bind the prosthetic heme group, the structural elements remained conserved. In addition, in the N-domains of the catalase-peroxidases, there is a large, Fig. 1. Multiple sequence alignment of 50 selected representatives of the superfamily of bacterial, f ungal and plant heme peroxidases: r egion on the distal side of the prosthetic heme g roup. Abbreviations o f enzym e sou rces are d efine d in T able 1. Numbers ind icate the position of each presented segment within the corresponding s equence . Sequences are grouped together as discussed in the text (i.e. catalase–peroxidases divided into two separate domains, CCPs, and APXs). Sequence similarity is graded from light grey (low similarity) to b lack (highest similarity). Functionally important residues involved in the catalytic mechanism are marked with an asterisk. This figure was constructed using GENEDOC [22]. The complete alignment of these sequences is available upon re quest. Ó FEBS 2004 Molecular evolution of heme peroxidases (Eur. J. Biochem. 271) 3301 Fig. 2. Multiple sequence alignment of 50 selected representatives of t he superfamily of bacterial, fungal and plant heme peroxidases: region on the proximal side of the prosthetic heme group. Abbreviations of enzyme sources are defined in Table 1. Numbers indicate the position of each presented segment within the corresponding sequence. In the case of cata- lase–peroxidases and some APXs a large insertion is presen t here. Se quenc es are grouped together as discussed in the text (i.e. catalase–peroxidases divided into two separate domains, CCPs, and APXs). Sequence similarity is graded from light grey (low similarity) to black (highest similarity). Functionally im port ant residue s invol ved in the catalytic mechanism are marked with an asterisk. This figure was constructed using GENEDOC [22]. 3302 M. Za ´ mocky´ (Eur. J. Biochem. 271) Ó FEBS 2004 36 amino acid-long insertion (between residues Asp268 and Thr304 in HalomarCPn) which has been suggested to have a function in the strength of Fe–N co-ordination on the proximal side [6]. This unique sequence motif in known KatG structure(s) is in principle a large loop [12] leading from the edge on the proximal side of heme to the molecular surface on the distal side. Part of this loop on the surface, around Glu271 of HalomarCP (Fig. 4A, shown in green), forms an entrance to the substrate access channel, and the remainder interacts with the C-domain of the neighboring subunit [12]. This loop contributes to the typical organiza- tion of the substrate channel to the active site (compare Fig. 4A with Fig. 4C and Fig. 4D), not surprisingly as it is a very flexible region which could not be located i n the electron density map. The role of this unique insertion has been examined in MyctubCPn by mutating Ser315 to a threonine [39]. The mutated protein did not activate isoniazid because o f th e introduction of a steric hindrance in the access c hannel. Hence, it is very likely that this extension from the proximal side to the substrate channel guarantees efficient catalytic reaction of catalase–peroxid- ases via rapid diffusion through a channel to the heme in the active site, similarly to monofunctional catalases [40]. A third conserved sequence pattern is located nearer the C-termini of the investigated sequences. With a sequence similarity of 52%, i t is a bove the average for a ll the sequences. Asp372 and Asp686 in HalomarCP (marked in Fig. 3 with an a sterisk) are 100% conserved i n a ll t he peroxidases a nalysed. This invariant aspartate forms an important hydrogen bond with the proximal heme ligand (His259), facilitating the reactivity of the heme iron [ 8]. Site- directed mutagenesis was performed in this region in SyncyspCPn [38] and SacchceCCP [34], with a large effect on the reactivity of the engine ered peroxidases. In contrast with other c atalytically important regions, this essential aspartate remained conserved in all known C-domains of Fig. 3. Multiple se quence alignment of 50 s elected representatives of the superfamily of bacterial, f ungal and plant heme p eroxidases: conserved region around the essential aspartate on the proximal heme side. Abbreviation s of e nzyme sources are defi ned in T able 1. Numbers ind icate the p osition of each presented segment within the corresponding sequence. Sequences are grouped t ogether as d iscussed in t he text (i.e. catalase–peroxidases divided into two se parate dom ains, CCPs, and APXs). Sequence similarity is graded from light grey (low similarity) to black (highest similarity). Functionally important residues i nvolved in the catalytic mecha nism are marked with an asterisk. This figure was constructed using GENEDOC [22]. Ó FEBS 2004 Molecular evolution of heme peroxidases (Eur. J. Biochem. 271) 3303 catalase–peroxidases, although its function in these domains is not apparent. In this third analysed region, residues in positions for w hich the function h as not yet b een determined are highly conserved (Fig. 3). Phylogenetic relationships The c onsensus phylogenetic trees produced by NJ and Fitch distance methods as well as the reconstructed t ree produced by the PUZZLE method revealed four main clades of heme peroxidases belonging t o class I of the superfamily of bacterial, fungal, and plant heme peroxidases. In Fig. 5 the simplified inferred FITCH tree, is presented, and in Fig. 6 the same tree in a simplified form with an outgroup is shown. The NJ-reconstructed tree exhibited identical topo- logy with slightly different branch lengths, and a very similar maximum likelihood tree was revealed b y PUZZLE .Thelatter method w i th the Ô+GÕ option f or rates of heterogeneity Fig. 4. Structural comparison of four representatives of class I of the superfamily of bacterial, fungal, and plant peroxidases. (A) N-terminal domain and (B) C-terminal domain of catalase–peroxidase from Haloarcula marismortui (PDB code 1ITK); (C) S. cerevisiae CCP (PDB code 2CYP); (D) cytosolic APX from Pisum sativum (PDB code 1APX). All figures are in solid ribbon presentation. In (A), (C) and (D) the prosthetic heme group is presented in ball and stick presentation. Functionally important conserved r esidues discussed in the te xt and marked also in Fig. 1 are shown with their corresponding number in the amino acid sequence. Those on the distal s ide of the prosthetic heme grou p are coloured yellow and those o n the proxim al side blue. The large loop in catalase–peroxidases with an essential residue in the entrance of a substrate chann el is coloured greenin(A). 3304 M. Za ´ mocky´ (Eur. J. Biochem. 271) Ó FEBS 2004 produced the parameter a ¼ 1.96 estimated from t he actual data set of 94 a nalysed peroxidase sequences. The bootstrap support in all main nodes is strong; only in some minor nodes is refining of the particular species within groups moderate. In the case of PUZZLE , the likelihood mapping analysis also revealed strong support of all main nodes. Hence, the four distinct clades can be understood as four diverse peroxidase families: ascomycetous CCPs; APXs; C-domains of catalase–peroxidases; and N-domains of catalase–peroxidases. This e volutionary branching a lso Fig. 5. Unrooted ph ylogenetic tree of 60 peroxidase g enes. TheinferredtreeobtainedwiththeFitchmethod [24] is pres ented. This tree is essentially identical w it h t he majority rule consensus t ree o b tained by the NJ method [23]. A very similar maximum likelihood tree was also obtained by PUZZLE [26]. Numbers represent the b ootstrap valu es on the branche s calculated fo r NJ/Fitch, respectively. The third value gives likelihood output from PUZZLE . The scale bar represents 10% of the estimated sequence divergence. Abbreviations of the species are identical with those used in Table 1. In the case o f catalase–peroxidases, the N-terminal and C-terminal domains are analysed separately (giving rise to a total of 94 analysed sequences) due to the evident tandem gene-duplication event discussed in the text. Colour scheme for catalase–peroxidases: brown, Archaeons; cyan, Cyano- bacteria; orange, Proteobacteria; magenta, Firmicut es; black, Actino bacteria; dark blue, Ascom ycota. Ó FEBS 2004 Molecular evolution of heme peroxidases (Eur. J. Biochem. 271) 3305 matches t he reaction specificity of the corresponding enzymes, and, in the case of all known c atalase–peroxidases, the two domains are fused together in one KatG .Itis obvious that CCPs are closely related t o A PXs, and, although the active centre of catalase–peroxidases located exclusively in the N-domains of KatGs resembles the active centres of APX and CCP (Fig. 4), the N-domains are phylogenetically more closely r elated to catalytically inactive C-domains of catalase–peroxidases. The complete sequences are known for catalase–peroxi- dases from various prokaryotes, both e ubacteria and archaea. The systematic a nalysis reveals that KatGsare distributed unequally among closely related genomes. Whereas in some complete genomes, no KatG is present (as discussed below), some bacteria even contain t wo different ones (e.g. Mycobacterium fortuitum). This unequal distribution of KatGs can be attributed to a lateral gene transfer [41] between otherwise phylogenetically unrelated micro-organisms. In the case of KatGs, it was first proposed to occur between archaea and eubacteria based on the analysis of three archaeal and 16 bacterial KatGs [42]. Later it was postulated that this phenomenon often occurs in all lineages of hydroperoxidases capable of catalytic reaction [43]. From the phylogenetic tree presen- ted here (with 34 KatGs divided in the separate domains), it is obvious that archaeal and eubacterial KatGsare phylogenetically more closely related than the rest of the genomes. Moreover, several lateral gene transfer events are discernible in t he phylogenetic tree in both branches of catalase–peroxidases (Fig. 5). Interestingly, the sequence of ArchfulCP segregated very early on from the remaining known KatGs. In th is paper, I focus on the analysis of lateral gene transfer between KatGs of Firmicutes, Cyano- bacteria, and Proteobacteria, which is also supported by high bootstrap values for both domains. Their positions on the branches indicate that KatGs from pathogenic proteo- bacteria are d escendants o f gen es from Firmicute s and Cyanobacteria. T here is also an obvious discrepancy between the rather high GC c ontent o f p roteobacterial KatGs a nd the GC content of the whole o rganism (Table 2), s upporting the hypothesis on the direction of the lateral gene transfer. Pathogenic and soil proteobacteria could profit from such a mode of lateral gene transfer by causing new genes to resist more efficiently the harmful effects of oxidative stress often caused by the host immune response or t he environment. However , no KatG was foundinthecompletedgenomeofBacillus subtilis, indicating gene loss in some Firmicutes. Fig. 6. Simplified presentation of the inferred t ree with the use of an outgroup (manganese peroxidase from Phanerochaete chrysosporium belonging to class II of this superfamily) presented to demonstrate the order of evolutionary events in class I of the peroxidase superfamily. Table 2. Analysis of GC c ontent in KatGs thought to be involved in lateral gene transfer between organisms. Values were obtained from the codon usage database at kazusa.or.jp. Abbreviations of enzymes are described in Table 1. Type of peroxidase GC content of KatG (%) GC content of whole organism (%) Type of organism BacihalCP 45.06 44.32 Firmicutes BacisteCP 51.85 49.67 Firmicutes DesulfiCP 57.09 50.07 Firmicutes GeobactCP 67.40 61.61 d-Proteobacterium GloeobaCP 65.58 62.86 Cyanobacterium LegipneCP 46.44 39.98 c-Proteobacterium SyncysCP 51.74 48.25 Cyanobacterium SynecspCP 54.14 55.86 Cyanobacterium VibrchoCP 51.18 48.09 c-Proteobacterium AspefumCP 63.64 54.17 Ascomycete AspenidCP 56.49 53.01 Ascomycete BlumgraCP 46.30 44.55 Ascomycete NcrassaCP 58.62 56.13 Ascomycete PenimarCP 55.81 51.43 Ascomycete 3306 M. Za ´ mocky´ (Eur. J. Biochem. 271) Ó FEBS 2004 [...]... repeatedly in the superfamily of bacterial, fungal, and plant heme peroxidases In class I, this event apparently occurred in each important period of evolution, firstly in the ancient evolutionary line allowing differentiation in (a) the progenitor for APX and CCP families and (b) ancestral catalase–peroxidase Furthermore, it acted inside both the APX and catalase–peroxidase families The subsequent event of. .. event of a tandem gene duplication in the already formed family of catalase peroxidases was unique, according to the fact that the duplicons remained fused together in one KatG Functional adaptation resulted in N-domains having the ability to oxidize and reduce hydrogen peroxide (i. e the reaction of catalase) and to remain peroxidatically active (oxidation of substrates with hydrogen peroxide) In contrast,... lines of class I of this superfamily occur, i. e the CCP line and the KatG fusion line, as is the case for Neurospora crassa [47] Evolution by the birth -and- death process in class I of the superfamily of bacterial, fungal, and plant peroxidases Gene duplication is one of the most common evolutionary processes by which new genes arise From the results presented, it is obvious that gene duplication must... contrast, C-domains lost their enzymatic activity and probably remained widespread only to stabilize the whole oligomeric protein In the course of evolution via (repeated) gene duplications, similar nonfunctional copies in the branch of APX and CCP must also have occurred It is difficult to follow the presence of such inactive Ôsecond copiesÕ of APX and CCP genes in the genomes, as mentioned previously [15]... Hence, it is reasonable to suppose a common origin of extant APXs in the ancient line of protists The occurrence and diversity of APX genes in various lineages that descended from the ancestral protist correlate with the various photosynthetic abilities of these organisms, producing various levels of reactive oxygen species The abundant present day APX genes are spread among the genomes of photosynthetically... catalase peroxidases: the catalytically inactive C-domains with as yet unknown or only hypothetical noncatalytic function) So we can conclude that the evolution of this class probably occurred through the death -and- birth process Whether this is also true for class II and class III remains to be proved It is clear that this mode of evolution in heme proteins can create novel reaction specificities, thus... presented in this tree, but in several plant species multiple forms of genes coding for cytosolic and chloroplast APXs exist as a rule, indicating a high level of sequence identity [36] The second subbranch is particularly interesting: the APX gene of a red algae (Galdieria partita) has a common origin with APXs from green algae (Chlamydomonas sp and Chlamydomonas reinhardtii) and all thylakoid APXs from higher... previously [15] However, in staying fused with the N-domains, the C-domains of KatGs represent a unique opportunity to address the mode of evolution of class I of this peroxidase superfamily According to currently established phylogenetic theories [17–20], multigene families can exist in genomes as a consequence of either concerted evolution or evolution by the birth -and- death process From the results presented... possible difference at the protein level In fact, this peroxidase, which is the closest phylogenetic neighbor to fungal CCPs, has a unique cellular localization in the endoplasmic reticulum It also exhibits a unique metabolic role in the trypanothione system [44] Evidence from an evolutionary tree based on 18S rRNA sequences places the origin of the kingdom Plantae among the phyla of the ancestral kingdom... class I of the superfamily of bacterial, fungal, and plant peroxidases the concerted mode of evolution is unlikely to have occurred This mode supposes that polymorphism (genetic diversity) was generated by the introduction of new ´ 3308 M Zamocky (Eur J Biochem 271) ´ variants from different loci through interlocus recombination or gene conversion Multiple sequence alignment of the whole of class I, . evolution, in which the indi- vidual specificities of the peroxidase families distinguished were already formed. Evidence is presented that class I of the heme. Phylogenetic relationships in class I of the superfamily of bacterial, fungal, and plant peroxidases Marcel Za ´ mocky ´ Institute of Molecular Biology,