Structural motifs of syringyl peroxidases predate not only the gymnosperm–angiosperm divergence but also the radiation of tracheophytes

Authors


Author for correspondence: A. Ros Barceló Tel: +34 968364 945 Fax: +34 968363 963 Email: rosbarce@um.es

Summary

  • • The most distinctive variation in the monomer composition of lignins in vascular land plants is that found between the two main groups of seed plants. Thus, while gymnosperm lignins are typically composed of guaiacyl (G) units, angiosperm lignins are largely composed of similar levels of G and syringyl (S) units.
  • • However, and contrary to what might be expected, peroxidases isolated from basal (Cycadales and Ginkgoales) and differentially evolved (Coniferales and Gnetales) gymnosperms are also able to oxidize S moieties, and this ability is independent of the presence or absence of S-type units in their lignins.
  • • The results obtained led us to look at the protein database to search for homologies between gymnosperm peroxidases and true eudicot S-peroxidases, such as the Zinnia elegans peroxidase.
  • • The findings showed that certain structural motifs characteristic of eudicot S-peroxidases (certain amino acid sequences and β-sheet secondary structures) predate the gymnosperm–angiosperm divergence and the radiation of tracheophytes, since they are found not only in peroxidases from basal gymnosperms, ferns and lycopods, but also in peroxidases from the moss Physcomitrella patens (Bryopsida) and the liverwort Marchantia polymorpha (Marchantiopsida), which, as typical of bryophytes, do not have xylem tissue nor lignins.

Introduction

The xylem constitutes the longest pathway for water transport in vascular plants. It is a simple pathway of low resistance, which enables water to be transported in large quantities with great efficiency from the roots to the leaves. Cell walls of mature xylem elements are impregnated with lignins (Boudet et al., 1995) that confer resistance against the tensile forces of the water columns, and impart water impermeability. Lignins represent the second most abundant organic compound on the earth's surface after cellulose, accounting for c. 25% of plant biomass (Lewis & Yamamoto, 1990). They are found specifically in vascular plants (tracheophyta) and occur in greatest quantity in the secondary cell walls of xylem vessels and tracheids, as well as fibres and sclereids (Ros Barceló, 1997). Lignins have been identified in pteridophytes (ferns, lycophytes and horsetails), widely considered to be the earliest-divergent living vascular plants, and are likely to have played a key role in the colonization of the terrestrial landscape by plants during the Ordovician to Silurian transition, 400–450 million yr ago (Lewis & Yamamoto, 1990).

Lignins are three-dimensional, amorphous heteropolymers that result from the oxidative coupling of three p-hydroxycinnamyl alcohols, p-coumaryl, coniferyl and sinapyl alcohols, in a reaction mediated by both laccases and class III plant peroxidases (Boudet et al., 1995; Ros Barceló, 1997). The cross-coupling reaction produces an optically inactive hydrophobic heteropolymer (Ralph et al., 2004) composed of H (p-hydroxyphenyl), G (guaiacyl) and S (syringyl) units, derived from p-coumaryl, coniferyl and sinapyl alcohols, respectively. The most distinctive variation in lignin monomer composition in vascular plants is that found between the two main groups of seed plants. Thus, in gymnosperms, lignins are typically composed of G units, with a minor proportion of H units, while in angiosperms lignins are mainly composed of similar levels of G and S units (Gross, 1980; Donaldson, 2001). In grasses (Ralph et al., 2004), lignins are even more complex, since they also contain significant amounts of ester-bound p-coumaric acid. In such a scenario, it is generally assumed that the chemical complexity of lignins has increased during the course of plant evolution, from pteridophytes and gymnosperms to grasses. This increase in the heterogeneity of lignin monomer composition might be related to the separation of water transport and support functions.

In gymnosperms, xylem tracheids function in both mechanical support and water transport. Typical gymnosperm tracheids are rich in G lignins (Donaldson, 2001), apparently because they lack the enzymes for sinapyl alcohol synthesis (Peter & Neale, 2004). By contrast, the secondary xylem of angiosperm trees contains two specialized cell types: the vessel elements that conduct water, and the fibre cells, which provide mechanical support. This segregation of function in the angiosperm xylem provides a more efficient and economical architecture. This efficiency is evident in the fewer but larger vessel elements, which show a reduced lignin content (c. 20% of dry matter) compared with gymnosperms (c. 30% of dry matter). Lower lignin and higher carbohydrate contents require significantly less energy, since less carbon is withdrawn from vegetative growth (Amthor, 2003). Interestingly, in angiosperms, the tracheary elements are rich in G lignins, like the tracheids of basal gymnosperms, whereas the sclerenchyma cells are rich in S lignins (Peter & Neale, 2004). This suggests a strong selective pressure to conserve the pathway for G lignin biosynthesis and its regulation in the water-conducting cells of the xylem during seed plant evolution (Peter & Neale, 2004). In fact, molecular evidence suggests that an ancient, predominant pathway for the synthesis of coniferyl alcohol is conserved in seed plants and that a branching pathway for sinapyl alcohol synthesis evolved subsequently in the angiosperms (Peter & Neale, 2004). This divergence apparently occurred within basal angiosperms before the segregation of grasses from eudicots.

As mentioned above, peroxidases (class III plant peroxidases, EC 1.11.1.7) are the main enzymes involved in the process of monolignol assembly leading to lignin biosynthesis. Peroxidases are usually classified into acidic (isoelectric point below 7.0) and basic (isoelectric point above 7.0) peroxidases. Both types of peroxidase are capable of oxidizing p-coumaryl and coniferyl alcohol. By contrast, while basic peroxidases are also able to oxidize sinapyl alcohol, typical acidic peroxidases, with some exceptions (Quiroga et al., 2000), are generally regarded as poor catalysts with this substrate (Nielsen et al., 2001). Since sinapyl alcohol is more prone to oxidation than either coniferyl alcohol or p-coumaryl alcohol (Kobayashi et al., 2005), this observation suggests that substrate accommodation in (exclusion from) the catalytic centre of the enzyme determines the real role played by each peroxidase isoenzyme in lignin biosynthesis (Kobayashi et al., 2001).

The oxidation of sinapyl alcohol by certain G peroxidases is sterically hindered owing to unfavourable hydrophobic interactions between the sinapyl alcohol methoxy atoms and the conserved I-138 and P-139 residues at the substrate binding site of the enzyme (Østergaard et al., 2000), although other factors, such as substrate hydrophilicity should also be considered (Kobayashi et al., 2005). This is not apparently the case for most basic peroxidases (S peroxidases), since the capacity of these enzymes to oxidize S moieties is universally accepted (Quiroga et al., 2000; Ros Barceló & Pomar, 2001; Holm et al., 2003). This observation would explain why antisense suppression of basic peroxidases in transgenic plants produces decreased levels of both G and S lignins (Blee et al., 2003), while antisense suppression of acidic peroxidases produces only decreased levels of G lignins (Li et al., 2003).

In accordance with their key role in lignin biosynthesis, cationic (basic) S peroxidases are differentially expressed during the trans-differentiation of Zinnia elegans mesophyll cell cultures, where they act as molecular markers of xylogenesis (Sato et al., 1995; López-Serrano et al., 2004; Gabaldón et al., 2005). Zinnia elegans is a eudicot, but peroxidases capable of oxidizing sinapyl alcohol have also been described in gymnosperms (Tsutsumi et al., 1998; McDougall, 2001), which paradoxically are reported to lack S-type lignins. In this report, we search for homologies between gymnosperm peroxidases and the Z. elegans basic peroxidase, a prototypical S peroxidase, which has been characterized and cloned (Gabaldón et al., 2005). The results showed that certain structural motifs of S peroxidases (amino acid sequences and certain β-sheet secondary structures) not only predate the gymnosperm–angiosperm divergence but also the radiation of tracheophytes, since they are also found in peroxidases from two members of the bryophytes, currently considered as the earliest-diverging land plants (Qiu & Palmer, 1999), and known to be lacking both a xylem tissue and true lignins (Ligrone et al., 2000).

Materials and Methods

Plant species studied

Conifers are the largest and most important class within gymnosperms. In this study, we selected seven conifers species belonging to the families, Araucariaceae (Araucaria araucana, and Araucaria heterophylla), Cupressaceae (Cupressus sempervirens and Tetraclinis articulata), Taxaceae (Taxus baccata and Thuja orientalis) and Pinaceae (Pinus halepensis, Pinus sylvestris and Picea abies) (Fig. 1). Although the xylem of all these conifers contains tracheids, certain differences in xylem structure are well preserved in fossils, for example, wood similar to that of the living Araucarias (monkey-puzzle and its derivatives) first appeared in the Carboniferous, whereas pine-like wood did not appear until the Cretaceous period (Chaw et al., 2000).

Figure 1.

Phylogenetic tree of the land plants used in this study based on molecular data from Qiu et al. (1999). L, land plants; V, vascular plants; E, Euphyllophytes; S, seed plants, F, flowering plants.

Together with conifers, the list includes members of the order Cycadales (Cycas revoluta, Cycas rumphii and Zamia fischeri), which constitute an extant descendant from Pteridosperms, or ‘seed ferns, the earliest seed plants. The oldest cycad fossils date from the Upper Palaeozoic (c. 265–290 million yr ago) (Chaw et al., 2000). Most phylogenies (Qiu et al., 1999; Pryer et al., 2001) place Cycadales as the basal order of gymnosperms, with Ginkgoales (with the only living species, Ginkgo biloba, also considered in this study) and Coniferales as a more advanced order (Fig. 1). Oldest Ginkgo-like fossils date from late Permian to Triassic (c. 260–250 million yr ago) (Chaw et al., 2000).

The list of gymnosperms species also includes members of Gnetales (Ephedra viridis and Welwitschia mirabilis). Gnetales have vessel-like tracheary elements resembling that of angiosperms, although molecular studies suggest that they are closer to conifers and other gymnosperms, and that the anatomical similarities with the angiosperms have arisen independently (Chaw et al., 2000). Molecular studies (Chaw et al., 2000) strongly support the monophyly of all extant gymnosperms, with cycads as the earliest-divergent clade, Ginkgo as the next basal, and a part of the conifers (Araucariaceae, Cupressaceae and Taxaceae in this study), as sister to a clade including the Gnetales and the Pinaceae (Fig. 1).

The list of studied species concludes with five basal land plants: Ceratopteris richardii (Filicopsida, Polypodiidae), Selaginella martensii and Selaginella moellendorfii (Lycopodiophyta, Selaginellaceae), Physcomytrella patens (Bryopsida) and Marchantia polymorpha (Marchantiopsida). Ceratopteris richardii is an aquatic fern that shows both tracheids and vessel elements with helical thickenings (Carlquist & Schneider, 2000), a coexistence that has been reported in Gnetales and dicotyledons but not in monocotyledons. Selaginella spp. belong to Lycopodiophyta, which constitute one of the earliest divergent extant vascular plants (Pryer et al., 2001). These lycopods show vessel elements (Schneider & Carlquist, 2000), whose appearance preceded the evolutionary divergence of lycopodiophytes, ferns and seed plants. Physcomytrella patens (Bryopsida) and M. polymorpha (Marchantiopsida), a moss and a liverwort, respectively, are representative of the most basal land plants. Unlike M. polymorpha, which lacks specialized water-conducting cells, P. patens shows an internal strand of imperforate water-conducting cells (hydroids) and a peripheral ring of thick-walled living cells (stereids) that, however, do not lignify (Ligrone et al., 2000).

Growth conditions

Young branch apices (1–3 cm long) of G. biloba, C. revoluta, C. sempervirens, T. baccata, T. orientalis, P. halepensis, T. articulata, A. araucana, and A. heterophylla were harvested in April and May from trees at least 5 year old growing in the Campus of the University of Murcia. Seedlings of Z. elegans L (cv. Envy) and E. viridis (both from Chiltern Seeds, Ulverston, UK) were grown for 14 d and 90 d, respectively, as described by Ros Barcelóet al. (2000). S. martensii cv. jori was purchased from a local nursery. The taxonomic position, organs selected, stage of growth and vascular bundle organization of the samples are reported in Table 1.

Table 1. Plant species investigated to assess the presence of S lignins and S peroxidases by chemical/biochemical methods, ranked in classes and families according to gymnosperm phylogeny (Pryer et al., 2001)
Class/FamilyPlant speciesOrgan selectedGrowth stageVascular bundle organization
Isoetopsida
 Selaginellaceae Selaginella martensii Twigs30 d oldAmphicribal
Cycadopsida
 Cycadaceae Cycas revoluta Young branch apices15 d oldCollateral
Ginkgoopsida
 Ginkgoaceae Ginkgo biloba Young branch apices15 d oldCollateral
Coniferopsida
 Araucariaceae Araucaria heterophylla Young branch apices15 d oldCollateral
Araucaria araucana Young branch apices15 d oldCollateral
 Cupressaceae Cupressus sempervirens Young branch apices15 d oldCollateral
Thuja orientalis Young branch apices15 d oldCollateral
Tetraclinis articulata Young branch apices15 d oldCollateral
 Taxaceae Taxus baccata Young branch apices15 d oldCollateral
 Pinaceae Pinus halepensis Young branch apices15 d oldCollateral
Gnetopsida
 Ephedraceae Ephedra viridis Young branches90 d oldCollateral

Trans-differentiating Z. elegans mesophyll cell cultures were established from true leaves from 14-d-old seedlings, which were surface-sterilized in 10% (v : v) commercial NaOCl, and rinsed in sterile distilled water. Single cells were isolated and cultured for 3 d in a differentiating medium as described by López-Serrano et al. (2004).

Microscopy

The samples were thoroughly washed with serum saline, fixed in 3% glutaraldehyde at room temperature for 30 min and postfixed for 2.5 h at 4°C with 1% OsO4 in 0.1 m cacodylate buffer, pH 7.5. Following postfixation, the samples were washed, dehydrated in graded series of ethanol and propylene oxide and embedded in Epon. Serial thin sections (1–3 µm) were observed by light microscopy after staining with toluidine blue (Ros Barceló, 2005). Light photomicrographs were taken with a Leica DMRB microscope. When necessary, samples were observed at the electron microscope level to confirm or to exclude the presence of secondary thickenings.

Histochemical stains for monitoring lignins, H2O2 and peroxidase

Lignins were detected using the Wiesner test by soaking 0.5 mm-thick sections in 1.0% (w : v) phloroglucinol in 25 : 75 (v : v) HCl-ethanol for 10–15 min (Pomar et al., 2002). The H2O2 localization/accumulation was monitored by staining 250- to 500-µm thick sections with the KI/starch reagent, composed of 4% (w : v) starch and 0.10 m KI (Ros Barceló, 2005), adjusted to pH 5.0 with KOH. Areas of H2O2 localization/accumulation were monitored by observing the development of a dark stain on the cut surface over a period of 1–10 h. Controls were performed in the presence of 200 U ml−1 catalase.

Peroxidase was monitored using the 3,5,3′,5′-tetramethylbenzidine (TMB) endogenous H2O2-dependent method (Ros Barceló, 2005). For this, sections were directly incubated for 10 min at 25°C in a staining solution composed of 0.1 mg ml−1 TMB-HCl in 50 mm Tris-acetate buffer (pH 5.0). Control incubations were performed in the presence of 0.1 mm ferulic acid (Ros Barcelóet al., 2000), a competitive inhibitor of peroxidase, whose oxidation is strictly dependent on H2O2.

Isolation of cell walls, alkaline nitrobenzene oxidation and thioacidolysis analyses

Cell walls were prepared using a Triton X-100 washing procedure (Pomar et al., 2004). Alkaline nitrobenzene oxidation of lignifying cell walls and high-pressure liquid chromatography (HPLC) analyses were performed essentially as described by Pomar et al. (2004). Quantification of p-hydroxybenzaldehyde, vanillin and syringaldehyde was performed at 280 nm using the corresponding standards. Thioacidolysis of lignifying cell walls, which solubilizes the β-O-4 lignin core, and gas chromatography–mass spectrometry (GC-MS) analyses were performed (Pomar et al., 2004) using a Hewlett Packard 5890 Series II gas chromatograph, an HP 5972 mass spectrometer and an HP5 (30 m × 0.25 mm internal diameter) column (Hewlett Packard, Palo Alto, CA, USA). Mass spectra were recorded at 70 eV.

Isolation of intercellular washing fluids (IWFs)

To obtain IWFs, 5-mm thick sections were washed three times with deionized water, and subsequently vacuum-infiltrated for 10 periods of 30 s at 1.0 kPa and 4°C with 50 mm sodium acetate buffer (pH 5.0) containing 1 m KCl and 50 mm CaCl2. Later, the sections were quickly dried and subsequently centrifuged in a 25-ml syringe barrel placed within a centrifuge tube at 900 g for 10 min at 4°C. The IWF samples were desalted and concentrated using the Amicon Ultra-15 system (Millipore, Carrigtwohill, Ireland).

Contamination by cytoplasmic constituents, as monitored by the activity of glucose-6-phosphate dehydrogenase (López-Serrano et al., 2004), was always < 0.1% relative to that found in the cytosolic fraction. Further confirmation of the absence of noticeable symplastic contamination in this apoplastic fraction was obtained by protein fingerprint analysis. Sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS-PAGE) analyses of the major proteins in the symplastic fractions showed the presence of specific proteins, which were almost totally absent from the respective apoplastic fractions (López-Serrano et al., 2004). Using the same method, the recovery of IWFs was extremely high since apoplast-specific proteins were absent from symplastic (IWF-extracted tissue) fractions (López-Serrano et al., 2004).

In the case of Z. elegans tracheary elements, cells were separated from the culture medium by centrifugation at 100 g for 1 min at 4°C, the supernatant constituting the apoplastic protein fraction. This protein fraction was desalted by chromatography on PD-10 Sephadex G-25 (Amersham Bioscience, Piscataway, NJ, USA) columns equilibrated in 50 mm sodium acetate buffer, pH 5.0, containing the protease inhibitors, 1.0 mm phenylmethylsulphonyl fluoride (PMSF) and 1.0 mm benzamidine, and concentrated using Ultrafree-0.5 (Millipore).

Determination of peroxidase activity in IWFs

Peroxidase activities were determined in IWFs in assay media containing 50 mm sodium acetate buffer (pH 5.0) and 500 µm H2O2, using as electron donors 1.0 mm 4-methoxy-α-naphthol or 0.1 mm coniferyl alcohol, sinapyl alcohol, coniferyl aldehyde and sinapyl aldehyde, using the corresponding difference of the extinction coefficient between substrate and product (Δɛ) for each reaction (Ros Barceló & Pomar, 2001).

Isoelectric focusing and visualization of enzymatic activities

Isoelectric focusing under nonequilibrium conditions (NEIEF) was performed as described by López-Serrano et al. (2004). Protein migration was followed at 4°C using cytochrome c as a migration marker. Peroxidase isoenzymes were stained with 4-methoxy-α-naphthol (López-Serrano et al., 2004).

Sequence homology analyses and molecular modelling of peroxidases

Sequence homology analysis was carried out using algorithms of clustalw (http://www.ebi.ac.uk), PROSITE (http://www.expasy.org/prosite) and the integrated software molecular evolutionary genetics analysis (version 3.1, Masatoshi Nei, http://www.megasoftware.net) (Kumar et al., 2004). Protein sequences were withdrawn from the NCBI (http://www.ncbi.nlm.nih.gov/) and the PeroxiBase (http://peroxidase.isb-sib.ch), a class-III plant peroxidase database (Bakalovic et al., 2006). Cluster analysis were performed by the ‘Minimum Evolution’ method, a distance method whereby evolutionary distances are computed for all pairs of analysed sequences, forming a matrix of pairwise distances; the phylogenetic tree is constructed by analysis of the relationships among these distance values (Wolf et al., 2002). Peroxidases were modelled by means of the swiss-model and swiss-pdbviewer molecular graphics modelling packages (http://www.expasy.ch/spdv/), according to the similarities of the modelled sequences to the known structures, available in the Protein Data Bank (PDB). Peroxidases were modelled using 1SCH (peanut peroxidase) and 1QO4 (Arabidopsis thaliana peroxidase A2) as template structures.

Chemicals

3,3′,5,5′-Tetramethylbenzidine-HCl, KI, ferulic acid, coniferyl alcohol, sinapyl alcohol, coniferyl aldehyde, sinapyl aldehyde, and catalase (from bovine liver, EC 1.11.1.6) were purchased from Sigma (Madrid, Spain). The other chemicals were of the highest purity available.

Results

Lignin monomer composition in gymnosperms

Thioacidolysis coupled to GC-MS of the lignifying cell walls of the gymnosperms studied revealed the great heterogeneity of their lignin monomer composition. It demonstrated the presence of thioethylated monomers (erythro and threo isomers) arising from the aryl-glycerol-β-aryl ether (β-O-4) structures derived from p-coumaryl (Figs 2a and 3a) and coniferyl (Figs 2b and 3b) alcohols, typical of gymnosperm lignins, and the presence of aryl-glycerol-β-aryl ether structures derived from sinapyl alcohol (Figs 2c and 3c), typical of angiosperm lignins.

Figure 2.

Gas chromatography profiles of the thioethylated monomers (erythro and threo isomers) arising from aryl-glycerol-β-aryl ether (β-O-4) structures derived from (a) p-coumaryl (b) coniferyl and (c) sinapyl alcohols from gymnosperm lignins.

Figure 3.

Mass spectra of the thioethylated monomers (erythro and threo isomers) arising from aryl-glycerol-β-aryl ether (β-O-4) structures derived from (a) p-coumaryl (b) coniferyl and (c) sinapyl alcohols from gymnosperm lignins. The ions of m/z (a) 239, (b) 269 and (c) 299 correspond to the (CH3)3SiOC6H4CH(SCH2CH3) (CH3)3SiOC6H3(OCH3)CH(SCH2CH3), and (CH3)3SiOC6H2(OCH3)2CH(SCH2CH3) radicals, respectively.

As representative of most gymnosperms, a detailed fingerprint of the thioethylated monomers arising from thioacidolysis analysis in P. halepensis is shown in Table 2, the main features being the presence of aryl-glycerol-β-aryl ether structures derived from coniferyl alcohol (peak 5). Thioacidolysis also revealed significant amounts of monomers arising from 4-O-linked coniferyl alcohol (peak 3) and 4-O-linked coniferyl aldehyde (peaks 4 and 6) end groups, as well as thioethylated monomers arising from C6C2-enol ether structures (peak 2) and stilbene structures (peak 7). Furthermore, thioacidolysis analysis revealed the presence of the 4-O-linked dihydroconiferyl alcohol end unit (peak 1), a monomer typical of gymnosperm lignins (Lapierre et al., 1995).

Table 2. Monomeric degradation product fingerprint obtained by thioacidolysis of lignifying Pinus halepensis cell walls and assignment of the original fragment in lignins
Peak No.(Rt, min)TIC (× 108) m/z (relative intensity)Original fragment in lignins
  1. Rt, retention time; TIC, total ionic current; G, guaiacyl; Tr, trace.

1(19.39)0.1273 (100), 179 (29), 206 (72), 209 (14), 221 (9), 236 (10), 311 (8), 326 (20)4-O-linked dihydroconiferyl alcohol
2(26.28)0.2473 (61), 135 (100), 179 (9), 192 (11), 209 (16), 329 (2), 344 (3)C6C2 enol ether structure (G-CH = CHOAr)
3(27.59)Tr73 (61), 75 (100), 205 (5), 209 (4), 235 (13), 297 (9), 343 (1), 358 (3) O-4-linked coniferyl alcohol
4(31.89)0.1073 (100), 135 (38), 192 (6), 209 (5), 239 (5), 269 (59), 295 (7), 327 (3), 356 (3), 403 (1), 418 (1) O-4-linked coniferyl aldehyde end group
5(32.30)2.4473 (33), 75 (22), 235 (8), 269 (100), 418 (1)β-O-4′-linked coniferyl alcohol
6(32.41)0.1073 (100), 192 (6), 209 (2), 239 (1), 269 (19), 418 (1) O-4-linked coniferyl aldehyde end group
7(34.67)0.1173 (30), 269 (90), 417 (2), 463 (1), 478 (1)Stilbene structure (G-CHR-CH2-G)

Table 3 summarizes the lignin monomer composition of each gymnosperm species studied, as seen from thioacidolysis or nitrobenzene oxidation. The results show that, in most species, lignins are mainly composed of H/G units with trace amounts, if any, of S units. The results obtained by thioacidolysis are consistent with those obtained by nitrobenzene oxidation. However, one feature of the thioacidolysis analysis in some species (e.g. C. revoluta and P. halepensis) was the absence of the thioethylated monomer arising from p-coumaryl alcohol, while p-hydroxyphenyl units were revealed by nitrobenzene oxidation product analyses (Fig. 4 illustrated the case of P. halepensis). This result, nevertheless, is to be expected since it is known that p-coumaryl alcohol is mainly incorporated in the condensed lignin fraction deposited in primary cell walls (Terashima & Fukushima, 1988), which is not susceptible to thioacidolysis cleavage (Lapierre et al., 1995).

Table 3. Monomer composition of lignins as revealed by nitrobenzene oxidation (NBO) and thioacidolysis (TA) in the cell walls of lycopods and gymnosperms
Plant speciesH (%)G (%)S (%)
NBOTANBOTANBOTA
  • tr, Trace; nd, not determined.

  • a

    Values are taken from Gross (1980).

Araucaria araucana 3a296a971a1
Araucaria heterophylla nd7nd93ndtr
Cupressus sempervirens nd0nd100nd0
Cycas revoluta 1a098a> 991a< 1
Ephedra viridis nd1nd39nd60
Ginkgo biloba 6a1193a881a1
Picea abies 2and97and1and
Pinus halepensis 609410000
Pinus sylvestris 2and98andtrand
Selaginella martensii 22a534a2444a71
Taxus baccata 4a395a971atr
Tetraclinis articulata 6 (8a)nd60 (61a)nd34 (30a)nd
Thuja orientalis nd7nd93nd0
Figure 4.

High-pressure liquid chromatography profile of alkaline nitrobenzene oxidation products obtained from lignifying Pinus halepensis stems showing the presence of p-hydroxybenzaldehyde (H) and vanillin (V).

Among the gymnosperms, the species T. articulata (Cupressaceae) and E. viridis (Ephedraceae, Gnetales) deserve special attention. Much as in angiosperms, these species contain significant amounts of S units (i.e. c. 30–60% of the lignin building blocks; Table 3). In the case of E. viridis, this observation is congruous with the notion that members of the Gnetales possess vessels and angiosperm-type lignins (Gross, 1980). This is also the case of S. martensii (Table 3), a representative of the Lycopodiophyta and one of the earliest divergent clade of extant vascular plants (Pryer et al., 2001). The presence of S units in the lignins of Selaginella spp. has been also detected by ozonation, acidolysis, infrared spectroscopy and 1H-nuclear magnetic resonance (NMR) (Jin et al., 2005), so that the revelation of S units in S. martensii lignins by thioacidolysis is unlikely to be a false positive of the technique.

Histochemical localization of H2O2/peroxidase in the gymnosperm xylem

Gymnosperm lignins are rich in 4-O-linked coniferyl aldehyde (Table 2, peaks 4 and 6) end-groups, and therefore are suitable for staining with phloroglucinol by means of the Wiesner reaction (Pomar et al., 2002). Phloroglucinol staining revealed that lignification in the stem was restricted to the wood, whereas the bark tissue was not stained (Fig. 5a,e). Staining was independent of the type of lignin synthesized, since similar results were found both in gymnosperms possessing G-type lignins (Fig. 5a) and G,S-type lignins (Fig. 5e).

Figure 5.

Photomicrographs of transversal stem sections of (a–d) Pinus halepensis, a gymnosperm bearing G-type lignins, and (e–h) Tetraclinis articulata, a gymnosperm bearing G,S-type lignins, after staining with phloroglucinol (a,e) to reveal p-hydroxycinnamaldehyde-rich lignins, with KI-starch (b,f) to reveal sites of H2O2 production and with 3,3′,5,5′-tetramethylbenzidine (TMB) (c,g) to reveal sites of peroxidase localization. Controls in the presence of 1.0 mm ferulic acid (d,h) support the peroxidase/H2O2-dependent nature of the staining reaction described in (c) and (g). Bar, 300 µm.

The independence of the pattern of lignin deposition from monomer composition was confirmed by the localization of H2O2 (Fig. 5b,f) and peroxidase activity (Fig. 5c,g). In both cases, staining was observed into the three- to eight-cell rows in which the new xylem was being produced (shown in detail in Fig. 6 after staining with toluidine blue). Controls in the presence of ferulic acid (Fig. 5d,h) support the peroxidase-dependent nature of the staining reaction shown in Fig. 5c,g. From these results, it may be concluded that H2O2 production and peroxidase activity are specifically colocalized in the tissue area where lignin is being synthesized.

Figure 6.

Detailed view of the zone of the vascular bundle where peroxidase is located. Plates show the tissue area stained with the TMB reagent in (a) Cycas revoluta, (b) Ginkgo biloba, (c) Taxus baccata, (d) Thuja orientalis, (e) Cupressus sempervirens, (f) Pinus halepensis and (g) Ephedra viridis. This zone comprises the three- to eight-cell rows in which the new xylem (X) is being produced, as viewed after further staining with toluidine blue. All the gymnosperms studied show a collateral vascular bundle organization, in which the cambial cells are located towards the outside (top in the plates), while the xylem is located towards the inside (bottom in the plates). In this vascular bundle organization, the growth of the new xylem results from the periclinal (tangential plane) division of elongated fusiform cambial initial cells, which generate the wood elements on the inner side (bottom).

S peroxidases in gymnosperms

It might be expected that if, in some gymnosperms, lignins are rich in S units while in others S units are present in low or trace amounts (Table 3), the presence of a peroxidase capable of oxidizing both coniferyl and sinapyl alcohol would be of great advantage for the expensive process of lignin assembly. To test this hypothesis, peroxidase activity was measured in IWFs. The results (Table 4) showed that, in all the species, IWF peroxidases were capable of oxidizing not only coniferyl alcohol and coniferyl aldehyde but also sinapyl alcohol and sinapyl aldehyde, independently of the presence of S units in their corresponding lignins (Table 3). Peroxidases capable of oxidizing syringyl moieties have also been reported in P. abies (Marjamaa et al., 2006), Cryptomeria japonica (Tsutsumi et al., 1998), and Picea sitchensis (McDougall, 2001), which also lack or nearly lack S-type lignins.

Table 4. Peroxidase activity measured with 4-methoxy-α-naphthol, coniferyl alcohol, coniferyl aldehyde, sinapyl alcohol and sinapyl aldehyde in IWF of lycopods and gymnosperms
Plant speciesPeroxidase activity (nkat g−1 FW)
4-Methoxy-α-naphtholConiferyl alcoholConiferyl aldehydeSinapyl alcoholSinapyl aldehyde
  1. Values are means ± SE.

Araucaria heterophylla 8.49 ± 0.4655 ± 39 ± 14 ± 10.21 ± 0.87
Cupressus sempervirens 0.26 ± 0.001.03 ± 0.960.73 ± 0.020.10 ± 0.010.02 ± 0.01
Cycas revoluta 5 ± 147 ± 1.033 ± 313 ± 23 ± 0
Ephedra viridis 0.29 ± 0.041.80 ± 0.171.62 ± 0.160.91 ± 0.060.24 ± 0.01
Ginkgo biloba 0.20 ± 0.070.55 ± 0.070.45 ± 0.020.06 ± 0.000.02 ± 0.00
Pinus halepensis 0.34 ± 0.280.84 ± 0.070.31 ± 0.070.11 ± 0.000.00 ± 0.00
Selaginella martensii 201 ± 30956 ± 9799 ± 4981 ± 253 ± 6
Taxus baccata 18 ± 0130 ± 2664 ± 511 ± 13 ± 0
Thuja orientalis 19 ± 9125 ± 1376 ± 52 ± 00.92 ± 0.11

The best characterized S peroxidase is the Z. elegans peroxidase since this has been purified and cloned (Gabaldón et al., 2005), and the polymers resulting from the oxidation of sinapyl alcohol catalysed by this peroxidase have been characterized (Gabaldón et al., 2006). With this in mind, a rapid method was developed for screening peroxidase isoenzymes homologous to the Z. elegans peroxidase in the apoplast of basal and evolved gymnosperms. Isoenzymes were analysed by isoelectric focusing, since the isoelectric point is determined by the balance between the acidic and the basic amino acids of the protein, and thus reflects similarities at the amino acid level of the isoenzymes.

Peroxidase isoenzyme patterns in these species are shown in Fig. 7. All the species showed the common presence not only of analogues (isoenzymes of similar mobility in the isoelectric focusing (IEF) gels) but also of homologues (isoenzymes of identical mobility in IEF gels) of the basic peroxidase isoenzyme isolated from differentiating Z. elegans tracheary elements (Fig. 7, lane a). This result agrees with previous reports that showed that basic peroxidases constitute the main population of peroxidase isoenzymes in some gymnosperms, such as A. araucana (Riquelme & Cardemil, 1993), P. abies (Koutaniemi et al., 2005) and P. sylvestris (Tarkka et al., 2001). Homologous isoenzymes were also found in the IWFs of the lycophyte S. martensii (Fig. 7, lane b). This agrees with the fact that S. martensii shows peroxidases capable of oxidizing S moieties (Table 4) and S-type lignins (Table 3). Cell wall-located basic peroxidases with isoelectric points (pI) around 10.5, such as the Z. elegans enzyme (Gabaldón et al., 2005), have also been described in mosses, such as Sphagnum magellanicum (Tutschek, 1979), and liverworts, such as Marchantia polymorpha (Hirata et al., 2000). These results led us to search for the protein sequence data base of the NCBI and the PeroxiBase to verify the presence in gymnosperms and basal land plants, including ferns, lycophytes, mosses and liverworts, of peroxidases homologous to the Z. elegans basic isoenzyme.

Figure 7.

Peroxidase isoenzyme patterns of tracheary elements of (a) Zinnia elegans, (b) Selaginella martensii, (c) Cycas revoluta, (d) Ginkgo biloba, (e) Araucaria heterophylla, (f) Cupressus sempervirens, (g) Thuja orientalis, (h) Taxus baccata, (i) Pinus halepensis and (j) Ephedra viridis obtained by nonequilibrium isoelectric focusing of the apoplastic protein, and stained with 4-methoxy-α-naphthol and H2O2 as substrate. Arrows indicates the position of homologous peroxidases to the Z. elegans basic peroxidase.

Comparative analysis of structural motifs of the primary structure of eudicot S peroxidases

Structural motifs of S peroxidases were determined by alignment of the Z. elegans basic isoenzyme with two tomato (Lycopersicon esculentum) S peroxidases, one of acidic (X15854) (Roberts & Kolattukudy, 1989) and the other of basic (L13653) (Quiroga et al., 2000) nature, and with a basic peroxidase from asparagus (Asparagus officinalis) (AB042103) (Takeda et al., 2003) (Table 5). The capacity of these four peroxidases to oxidize S moieties is well known (Quiroga et al., 2000; Takeda et al., 2003). Structural determinants of S peroxidases were deduced by comparing amino acid sequence motifs common to these four eudicot peroxidases with amino acid sequence motifs of two typical G peroxidases, the Arabidopsis thaliana (ATP) A2 peroxidase (X99952) (Østergaard et al., 2000) and the Armoracia rusticana (horseradish) A2 peroxidase (HRP A2, P80679) (Nielsen et al., 2001) (Table 5). All these peroxidases show an N-terminal propeptide but lack a C-terminal extension, supporting the view that they are cell-wall located enzymes, with a putative role in lignification. Sequence-homology analyses were restricted to the region between the distal (H41 in ATP A2) and the proximal (H169 in ATP A2) histidines that either surround or determine the conformation (relaxation capacity) of the catalytic centre of the enzymes. This region also comprises the amino acids P69, I138, P139, S140 and R175 (numbered according to ATP A2), which determine the conformation and hydrophobicity of the substrate-binding site (Østergaard et al., 2000). The region considered is extremely distinctive of class III plant peroxidases since it is codified by exon 2 in most peroxidase genes (Gabaldón et al., 2005), and delimited by type 2 and type 3 introns (two introns actively involved in recombinations between plant peroxidase genes) (Gabaldón et al., 2005). Finally, this region has been shown to be one of the most highly conserved in class III plant peroxidases during land plant evolution (Valerio et al., 2004).

Table 5. Accession number (AN), number of amino acids (NAA), theoretical Mr (TM), and theoretical isoelectric point (TIP) of the peroxidase sequences studied
PeroxidasePlant speciesANNAATMTIP
  • GP, guaiacyl peroxidases; SP, syringyl peroxidases.

  • a

    The sequences deposited in the data banks are incomplete and lack the C-ends.

  Ceratopteris richardii BE643121312342165.82
Cycas rumphii CB092730a   
Ginkgo biloba CB074963300339779.02
Marchantia polymorpha BAB97197298321115.17
Physcomitrella patens AW561204a   
Selaginella moellendorffii AC158187306329328.64
Zamia fischeri ZfiPrx02a   
SP Picea abies AJ809340294313098.90
Pinus sylvestris AAG02215321350024.93
Welwitschia mirabilis WmPrx06a   
GP Arabidopsis thaliana X99952305319364.82
GP Armoracia rusticana P80679305318994.72
SP Asparagus officinalis AB042103303334128.85
SP Lycopersicon esculentum L13653304329428.91
SP Lycopersicon esculentum X15854294311244.49
SP Zinnia elegans AJ880395291308638.47

As may be expected, alignments of this zone for the peroxidases selected revealed highly conserved amino acid residues common to all peroxidases (shaded in yellow in Fig. 8), and partly conserved amino acid residues of ATP A2 and HRP A2 present also in some of the resting peroxidases (shaded in green in Fig. 8). Structural motifs, differentially present in all eudicot S peroxidases and absent from the G peroxidases, ATP A2 and HRP A2, are shaded in red (Fig. 8). These motifs are I76 (in Z. elegans peroxidase) (V80 in ATP A2), A94 (in Z. elegans peroxidase) (S98 in ATP A2), 101-ARD (in Z. elegans peroxidase) (105-SEA in ATP A2), and K151 (in Z. elegans peroxidase) (V155 in ATP A2). Some of these amino acids, such as the A94 (in Z. elegans peroxidase) (S98 in ATP A2), form part of the structural motif VSCAD in S peroxidases which contrast with that found in G peroxidases, VSCSD (Fig. 8). The same structural motifs common to eudicot S peroxidases were found in peroxidases from Welwitschia (WmPrx06), which has G,S-lignins; in peroxidases from Picea (AJ809340) (Koutaniemi et al., 2005) and Pinus (AAG02215) (Tarkka et al., 2001), both lacking S-lignins, and in peroxidases from basal gymnosperms, such as Cycas (CB092730), Zamia (ZfiPrx2) and Ginkgo (CB074963), which also significantly lack S-lignins (Table 3). Structural motifs common to all eudicot S peroxidases were also found in peroxidases from Selaginella (AC158187) and Ceratopteris (BE643121), the first a terrestrial lycopod containing G,S-lignins (Table 3), and the latter an aquatic fern. Surprisingly, structural motifs common to all eudicot S peroxidases were also found in peroxidases from Physcomitrella (AW561204) and Marchantia (BAB97197) (Table 5, Fig. 8), a moss and a liverwort.

Figure 8.

Amino acid sequence alignment of eudicot G and S peroxidases with basal land plant peroxidases. Structural motif characteristics of S peroxidases were determined by alignment of the Zinnia elegans basic isoenzyme (AJ880395) with two tomato (Lycopersicon esculetum) S peroxidases (X15854 and L13653), and with one asparagus (Asparagus officinalis) basic peroxidase (AB042103) – a set of peroxidases whose capacity for oxidizing S moieties is well known. Structural determinants of S peroxidases (shaded red) were deduced by comparing amino acid sequence motifs common to these four eudicot peroxidases with the amino acid sequence motifs of two typical G peroxidases, the Arabidopsis thaliana A2 peroxidase (X99952) and the Armoracia rusticana A2 peroxidase (P80679). Sequence homology analyses were restricted to the region between the distal (H41 in ATP A2) and the proximal (H169 in ATP A2) histidines that determine the conformation of the catalytic centre of the enzymes. This region also comprises the amino acids, I-138, P-139 and S-140 (numbered according to ATP A2, black arrowheads), which determine the conformation and hydrophobicity of the substrate-binding site (Østergaard et al., 2000). Alignments of this zone for the peroxidases selected revealed highly conserved amino acid residues common to all peroxidases (shaded in yellow), and partly conserved amino acid residues of ATP A2 and HRP A2 presents also in some of the resting peroxidases (shaded green). Database accession numbers are as in Table 5. Consensus symbols: *, amino acid residues in the column are identical in all sequences of the alignment; :, conserved substitutions have been observed; ., semiconserved substitutions are observed.

Similar to the eudicot peroxidases used as a template, these basal peroxidases have a N-terminal propeptide but lack a C-terminal extension, suggesting that these also are cell-wall-located enzymes. In the case of Welwitschia, Zamia and Physcomitrella peroxidases, the cell wall localization could not be confirmed since the sequences deposited in the PeroxiBase data bank are incomplete and lack the C-ends. The reliability of the molecular approach followed in this study is confirmed by the fact that the P. abies peroxidase PAPX5 (AJ809340) (Koutaniemi et al., 2005), which shows all the structural motifs common to S peroxidases (Fig. 8), is capable of oxidizing syringaldazine (Marjamaa et al., 2006), that is the prototype substrate for S peroxidases (Ros Barcelóet al., 2000).

Comparative analysis of structural motifs of the secondary structure of eudicot S peroxidases

The secondary structural elements of ATP A2 consist of 12 α-helices, helix A (A14-L26), helix B (I32-C44), helix C (F77-A90), helix D (C97-L111), helix D′ (L131-S-137), helix E (L145-A154), helix F (T159-T170), helix F″ (S198-L207), helix G (N231-Q237), helix H (Q244-S251), helix I (T257-S266) and helix J (Q268-G283), and two short β-strands, the β-strand I (R173-R175) and the β-strand II (I217-N219) (Østergaard et al., 2000). It is interesting to note that the position of the F and F″α-helices is a unique feature of class III plant peroxidases within the plant peroxidase superfamily (Veitch, 2004). With this in mind, a modelling of the secondary structures of both basal and eudicot peroxidases was performed using the available programs (http://www.expasy.ch/spdv/) to investigate whether conserved and nonconserved amino acid sequences may generate similar or different secondary structures.

The amino acid region analysed in Fig. 8, belonging mainly to the exon 2 and the neighbouring coding region of most peroxidase genes, is displaced by 35–40 amino acids in Fig. 9 to include the region from the helix C (hC) to the helix F″ (hF″), two common structural elements for all peroxidases. Although all the peroxidases analysed showed as common features the invariable positioning of the helices C, D, E, F and F″, some significant differences emerged. For example, in comparison with G peroxidases, the helix D′ (hD′) was absent from all the eudicot S peroxidases and from all the basal peroxidases analysed (Fig. 9). This observation may have strong implications since, in ATP A2 peroxidase, helix D′ relies on the haem prosthetic group (Østergaard et al., 2000). The absence of helix D′ from eudicot S peroxidases probably represents a relaxation factor for the haem crevice, allowing the docking of S moieties, and this would be extrapolable to all the basal peroxidases (Fig. 9), including peroxidases from a moss (AW561204) and a liverwort (BAB97197). A relaxation of the substrate-binding site in eudicot S peroxidases, owing to the absence of the helix D′, is supported by the observation that the oxidation of sinapyl alcohol by certain G peroxidases is sterically hindered by unfavourable hydrophobic interactions between the sinapyl alcohol methoxy atoms and the conserved I-138 and P-139 residues (Østergaard et al., 2000). It is interesting to note that these residues form part of the IPS motif (black arrowheads, Fig. 9) which is, in turn, sterically fixed by the immediately upstream helix D′ in G peroxidases, a helix which is absent from all the eudicot S peroxidases. No variation in helix structures was found in other parts of the proteins.

Figure 9.

Secondary structural elements of eudicot G and S peroxidases and basal land plant peroxidases. The amino acidic region analysed in Fig. 8, belonging mainly to the exon 2 and the neighbouring coding region of most peroxidase genes, is displaced 35–40 amino acids in this Figure to include from the helix C (hC) to the helix F″ (hF″), two common structural elements for all peroxidases. This region also comprises the amino acids, P-69, I-138, P-139, S-140 and R-175 (numbered according to ATP A2, black arrowheads), which determine the conformation and hydrophobicity of the substrate-binding site (Østergaard et al., 2000). All the peroxidases analysed show as common features the invariable positioning of the helices C (hC), D (hD), E (hE), F (hF) and F″ (hF″) (shaded green). The β-strand I (R173–R175, ATP A2) (shaded yellow) was absent in all the eudicot S peroxidases, in which novel β-strands emerged (shaded red), some of them are upstream the proximal histidine (arrow). Peroxidases were modelled by means the swiss-model and swiss-pdbviewer molecular graphics modelling packages (http://www.expasy.ch/spdv/), using 1SCH (peanut peroxidase) and 1QO4 (Arabidopsis thaliana peroxidase A2) as template structures. Database accession numbers are as in Table 5. Consensus symbols: *, amino acid residues in the column are identical in all sequences of the alignment; :, conserved substitutions have been observed; ., semiconserved substitutions are observed.

As for the helix D′, the β-strand I (R173–R175, Fig. 9) in ATP A2 peroxidase was found to be absent from all the eudicot S peroxidases analysed, in which instead novel β-strands emerged (Fig. 9). Among these, special attention should be paid to the β-strand sited upstream of the proximal histidine (arrow in Fig. 9), which probably influences the catalytic centre of the enzymes.

Phylogenetic tree of S peroxidases

Cluster analysis by the ‘Minimum Evolution’ method (Fig. 10) suggests that S peroxidases constitute an ancestral branch, whose origins go back to ancestral (nonvascular) plant lineages within the monophyletic origin of land plants (Qiu & Palmer, 1999). In this cluster analysis, both the A. thaliana (X99952) and the A. rusticana (P80679) G peroxidases (Cruciferae/Brassicaceae) appeared as a more evolved branch within class III plant peroxidase evolution.

Figure 10.

Phylogenetic tree of eudicot S peroxidases (closed triangles) using the ‘Minimum evolution’ method. Database accession numbers are as in Table 5. In this cluster analysis, both the Arabidopsis thaliana (X99952) and the Armoracia rusticana (P80679) G peroxidases (open triangles) appeared as a more evolved branch within class III plant peroxidase evolution.

Discussion

The lignification pattern of Z. elegans seedlings is unique in that, at a certain developmental stage, it offers simultaneously two models of lignification that closely resemble those occurring in gymnosperms and angiosperms. That is, S units predominate in the hypocotyl, while G units predominate in the stem (Ros Barcelóet al., 2004). In this regard, Z. elegans hypocotyl lignins are typical of angiosperms, while the lignins of the young stem partly resemble those of gymnosperms, since (H + G) alone constitute 78% of the lignin building blocks. This contrasts with the fact that roots, hypocotyls, stems and trans-differentiating Z. elegans mesophyll cell cultures express the same basic peroxidase isoenzyme (López-Serrano et al., 2004; Gabaldón et al., 2005).

The Z. elegans basic peroxidase isoenzyme has been purified and cloned (Gabaldón et al., 2005), and is capable of oxidizing not only the three p-hydroxycinnamyl alcohols but also the three p-hydroxycinnamyl aldehydes and the three p-hydroxycinnamic acids (Ros Barcelóet al., 2004; Gabaldón et al., 2006), a fact that suggest that the peroxidase isoenzyme complement used by Z. elegans to synthesize S lignins is the same as that used to synthesize G lignins. Given these extraordinarily catalytic properties, it is expected that this peroxidase isoenzyme should have been strongly conserved during plant evolution as lignins were.

From an evolutionary perspective, lignins are uniformly distributed from primitive pteridophytes and gymnosperms to highly evolved monocotyledons (Boerjan et al., 2003). With very rare exceptions (see Table 3), the monolignol pathway, which provides lignins in pteridophytes and gymnosperms, only utilizes p-coumaryl and coniferyl alcohol, whereas in angiosperms and Gnetales the lignin biosynthetic pathway is further branched to use sinapyl alcohol as substrate. However, this assumption is not always true since lycopods, such as Selaginella spp. (Table 3), placed in the most basal orders within tracheophytes (Qiu & Palmer, 1999), are able to synthesize S-lignins, and this could also be the case of Ceratopteris, an aquatic fern, which, in common with angiosperms, Gnetales, and lycopods such as Selaginella, shares the presence of vessels which coexist with tracheids (Carlquist & Schneider, 2000). The lignification pattern would have been an adaptive trait of great significance during plant evolution, since both gymnosperms and angiosperms share an ancient, conserved set of enzymes, which are regulated by conserved transcription factors, and which are responsible for the formation of G lignins (Peter & Neale, 2004). As far as is known, angiosperms and Gnetales, nevertheless, have evolved two enzymes that catalyse the production of S lignins and, given the presence of S-lignins in lycopods, it is probable that the pathway for sinapyl alcohol has evolved (independently or not) at least three times during the evolution of tracheophytes. An alternative hypothesis might be that sinapyl alcohol was incorporated into lignin in the earliest tracheophytes, the mechanism for sinapyl alcohol recruitment having been secondarily lost in certain gymnosperms and in ferns, so that sinapyl alcohol recruitment to lignin is the ‘primitive state’.

The results obtained apparently support the latter hypothesis since, in seven of the 12 gymnosperms studied, S lignins may be found in either significant or trace amounts (Table 3). All them contain peroxidase isoenzymes capable of oxidizing S moieties (Table 4) and possess peroxidase isoenzymes homologous, as regards pI, to the Z. elegans basic peroxidase isoenzyme (Fig. 7), independently of the nature (either G or S/G) of their lignins (Table 3). This is also true for the main traits of the lignification pattern, as shown by the presence of an unique pattern of H2O2 production/peroxidase localization, which is superimposed on the lignification front, and which is common to gymnosperms bearing S/G lignins (Fig. 5e–h) and gymnosperms bearing G lignins (Fig. 5a–d).

These results also indicate that some structural motifs of S peroxidases were strongly conserved during the early evolution of vascular plants. In fact, certain structural motifs (certain amino acid sequences and certain β-sheet secondary structures) of peroxidases capable of oxidizing sinapyl alcohol are also present in the peroxidases of gymnosperms which lack S-type lignins (Figs 8 and 9), an observation which suggests that the evolutionary gain of the monolignol branch leading to the biosynthesis of S lignins, was made possible because the enzymes responsible for its polymerization evolved very early during the evolution of land plants. These results also suggest that these peroxidases were present in an ancestral plant stock before the radiation of tracheophytes, a suggestion confirmed by the presence of the same structural motifs in peroxidases from mosses and liverworts. At this point, it is necessary to mention that mosses and liverworts have no xylem and do not lignify (Erickson & Miksche, 1974; Wilson et al., 1989). Taken together, the present results strongly support the idea that some structural motifs and/or structural elements of eudicot S peroxidases not only predate the gymnosperm–angiosperm divergence, being present in basal tracheophytes, but also predate the radiation of tracheophytes.

The findings also indicate that the gene codifying this peroxidase isoenzyme evolved before the radiation of tracheophytes and, as such, it should be considered as an ancestral gene within the monophyletic origin of flowering plants (Fig. 10). This result is not surprising since gymnosperms share c. 70% apparent gene homology with angiosperms (Ujino-Ihara et al., 2005) and, in the case of multigene families such as peroxidase (Passardi et al., 2004; Valerio et al., 2004), it is quite likely that that multiplicity arose before the divergence of seed plants (Nishiyama et al., 2003).

In other words, our phylogenetic studies on S peroxidases have shown that the genes codifying these enzymes could predate the appearance of vascular plants themselves (Fig. 10), and might even be contemporaneous with the acquisition of the most primitive short-distance water and nutrient transport systems that coevolved with mosses and liverworts (Ligrone et al., 2000; Sperry, 2003). This observation would agree with the fact that an ancestral phenylpropanoid pathway is still present in bryophytes, where peroxidases of a versatile nature have been involved in the metabolism of cell wall-located p-hydroxycinnamic acid derivatives (Rudolph & Samland, 1985; Wächter et al., 1987). Since the most characteristic property of S peroxidases is the absence of steric restrictions at the substrate binding site for oxidizing p-hydroxyphenylpropanoid metabolites (Ros Barcelóet al., 2004; Gabaldón et al., 2006), this observation agrees with the existence of an ancestral basic model of cell wall architecture and function which would have evolved before the evolutionary divergence of bryophytes, ferns and seed plants (Ligrone et al., 2002; Popper & Fry, 2004; Carafa et al., 2005), and which have been finely tuned in response to specific evolutionary pressures. In this model, G peroxidases would constitute the ‘evolved state’ (Fig. 10).

In such a scenario, these results support the hypothesis of an initial evolutionary independence of lignification and water-conducting cell morphology, as is suggested by microchemical analysis of lower Devonian fossils (Friedman & Cook, 2000; Boyce et al., 2003), which provide evidence that lignification could have originated in the peripheral tissues of protracheophytes and was only later coopted for the strengthening of tracheids in eutracheophytes. At this point, we should mention the knowledge acquired from the study of some rhyniophyte lineage fossils, such as Aglaophyton, the sister group to the eutracheophyte clade that includes all the living vascular plants. Aglaophyton is a protracheophyte fossil which appears to lack lignified thickenings (Boyce et al., 2003), but whose perforated tubes resemble the hydroids of some mosses and may have a common ancestry (Sperry, 2003). It is known that versatile peroxidases, with a broad range of substrate specificity for p-hydroxyphenylpropanoid metabolites (Tutschek, 1979; Speicher et al., 2003), are located in the primary cell wall of liverworts (Ishida et al., 1985) and in the unlignified thick cell wall of living chlorocytes, which protect from the collapse to the adjacent dead hydrocytes in peat mosses (Wächter et al., 1987). In this context, the observation that, in flowering plants, S peroxidases are putatively located in lignifying thick-walled living xylem fibres, which are not involved in water transport (Ros Barcelóet al., 2000), seems to represent an inheritance from their origin in unlignified thick-walled living cells of an ancestral land plant lineage – an enzymological character which survives in the unlignified thick-walled living cells from peristomate mosses.

The results described above are not surprising since cell wall-located class III plant peroxidases are found in most vascular plants, including early land plant lineages, such as ferns, mosses and liverworts (Duroux & Welinder, 2003), although they are absent from green algae. Molecular clocks suggest that land plants are monophyletic, and that they diverged from green algae c. 700 million years ago, the charophytes (freshwater green algae) being the closest lineage to land plants (Qiu & Palmer, 1999). Phylogenetic studies also suggest that class III plant peroxidases appeared with the emergence of land plants (Duroux & Welinder, 2003), and it is likely that this class of enzymes conferred important adaptive traits to plants for their new life on land, the most characteristic gain being the acquisition of vascular tissues whose cell walls are impregnated with lignins. In this scenario, it is not surprising that the enzymes responsible for lignin building construction appeared early in the evolution of land plants, and that these enzymes, similar to all highly expressed proteins that evolved slowly (Drummond et al., 2005), have been largely conserved during plant evolution.

Acknowledgements

This work was supported by grants from the Fundación Séneca (project #00545/PI/04) and MCYT (BOS2002-03550). L.V.G.R. and C.G. hold fellowships (FPI and FPU, respectively) from the MCYT and MECYD, respectively.

Ancillary