Quickly-released peroxidase of moss in defense against fungal invaders


Author for correspondence:
Jari P. T. Valkonen
Tel: +358 9 19158387
Email: jari.valkonen@helsinki.fi


  • • Mosses (Bryophyta) are nonvascular plants that constitute a large part of the photosynthesizing biomass and carbon storage on Earth. Little is known about how this important portion of flora maintains its health status. This study assessed whether the moss, Physcomitrella patens, responds to treatment with chitosan, a fungal cell wall-derived compound inducing defense against fungal pathogens in vascular plants.
  • • Application of chitosan to liquid culture of P. patens caused a rapid increase in peroxidase activity in the medium. For identification of the peroxidase(s), matrix-assisted laser desorption/ionization–time-of-flight (MALDI-TOF)/MS, other methods and the whole-genome sequence of P. patens were utilized. Peroxidase gene knock-out mutants were made and inoculated with fungi.
  • • The peroxidase activity resulted from a single secreted class III peroxidase (Prx34) which belonged to a P. patens specific phylogenetic cluster in analysis of the 45 putative class III peroxidases of P. patens and those of Arabidopsis and rice. Saprophytic and pathogenic fungi isolated from another moss killed the Prx34 knockout mutants but did not damage wild-type P. patens.
  • • The data point out the first specific host factor that is pivotal for pathogen defense in a nonvascular plant. Furthermore, results provide conclusive evidence that class III peroxidases in plants are needed in defense against hostile invasion by fungi.


Bryophytes (mosses, hornworts and liverworts) are nonvascular plants that constitute a large part of the photosynthesizing biomass and carbon storage in ecosystems on earth. They play a central role in balancing atmospheric CO2 and buffering against climatic changes, especially in temperate and northern zones (O’Neill, 2000). Therefore, the health of bryophyte vegetation is ecologically important. An abundance of minute parasitic fungi occur on mosses (division Bryophyta) but only a few cause macroscopic symptoms readily recognized by the naked eye (Davey & Currah, 2006). It is not known how bryophytes protect themselves against potentially hostile pests and pathogens (Davey & Currah, 2006). Studies of pathogen defense have been carried out on vascular plants, whose divergence from the bryophyte lineage occurred c. 400 million years ago (Lang et al., 2008). Vascular plants also are exposed to potentially harmful microbes and insects but can control most of them with broadly effective basal or nonhost defense (Collins et al., 2003; Dicke & Hilker, 2003; Mitchell & Power, 2003; Jones & Dangl, 2006).

In addition to living microbes, defense responses in vascular plants are elicited with fungal cell wall extracts (Chivasa et al., 2005) and specific fungal cell wall constituents such as chitin (C8H13O5) and water-soluble chitosan that is obtained from chitin by deacetylation and available commercially for use as an enhancer of fungal defense in plants (Rabea et al., 2003; Miya et al., 2007). This study was initiated based on our observation that chitosan causes a rapid increase in peroxidase activity in the culture medium of treated liquid cultures of Physcomitrella patens, a ‘model moss’ whose genome sequence has been determined (Rensing et al., 2008). Other features of P. patens also encourage its use for molecular study (Quatrano et al., 2007). It is reasonably fast growing and can be easily grown in vitro and transformed genetically. It is haploid for most of the life cycle, and the high frequency of homologous DNA recombination allows using targeted gene replacement (knock-out) technology for assigning genes with functions (Schaefer, 2001; Cove, 2005). Only one previous study has reported secretion of a peroxidase by a nonvascular plant, namely the liverwort Marchantia polymorpha (synonym Marchantia aquatica) that responds in this manner to chemical stress (Hirata et al., 2000). Because peroxidases isolated from tissues of vascular plants can have antifungal activity (Ghosh, 2006) but direct evidence for the role of peroxidases in antifungal defense is lacking (Almagro et al., 2009), we set out to study this further using P. patens.

Materials and Methods

Plant material

Protonemal tissue of P. patens (Hedw.) B.S.G. (family Funariaceae), ecotype Gransden Wood (Ashton & Cove, 1977), was grown in Petri dishes (diameter 9 cm) on cellophane membranes placed on BCD medium (1 mm MgSO4, 1.85 mm KH2PO4 (pH 6.5, adjusted with KOH), 10 mm KNO3, 45 m FeSO4, 0.22 µm CuSO4, 0.19 µm ZnSO4, 10 µm H3BO4, 0.10 µm Na2MoO4, 2 µm MnCl2, 0.23 µm CoCl2, 0.17 µm KI) (Ashton & Cove, 1977) supplemented with 1 mm CaCl2, 45 µm EDTA disodium salt (Na2-EDTA), and 5 mm ammonium tartrate ((NH4)2C4H4O6), and solidified with 0.8% agar. The cultures were grown in a growth chamber (Model 3755; Forma Scientific, Marietta, OH, USA) at 20°C (photoperiod 12 h, light intensity 40 µmol m−2 s−1) and subcultured weekly.

Chitosan treatment

For experiments, 1-wk old protonemal tissue was subcultured on fresh BCD medium without ammonium tartrate and grown for 1 wk. Protonemal tissue was collected from an area of 2 cm2 and transferred to 2 ml of liquid BCD medium containing 5 mm ammonium tartrate and 1% glucose in 15-ml conical tubes (Sarstedt, Nümbrecht, Germany). Tubes were agitated (90 r.p.m.) in a rotary shaker at 25°C and illuminated from above (photoperiod 12 h, light intensity 160 µmol m−2 s−1). Water-soluble chitosan (MW approx. 2000–5000; Wako Pure Chemical Industries Ltd, Osaka, Japan) was dissolved in sterile Milli-Q water (Millipore, Billerica, MA, USA). Stock solution of chitosan (10 mg ml−1) was added to 1-wk old moss liquid cultures (2 ml) to a final concentration of 0.5 mg ml−1 and the moss grown as above. Sterile Milli-Q water was used as a control treatment.

Peroxidase activity assay

To measure peroxidase activity, 50-µl aliquots of medium from moss liquid cultures treated with chitosan for 0, 5, 15, 30, 60, 180 or 300 min were centrifuged at 16 000 g for 1 min. The supernatant was collected with care in order to exclude all debris remaining at the bottom of the tube, transferred to a clean tube and kept on ice. Peroxidase reaction mixtures (1 ml final volume) containing 0.5 mm 2,2′-azino-bis3-ethylbenzthiazoline-6-sulphonic acid (ABTS) (Fluka, St Louis, MO, USA) and 200 µm H2O2 in phosphate-citrate buffer (29 mm phosphate, 23 mm citrate, pH 4.0) were pre-warmed at 35°C followed by the addition of 40 µl ice-cold culture supernatant. Reaction mixtures were incubated for 15–120 s and absorbance at 405 nm was measured using a spectrophotometer (UV-1700; Shimazu, Kyoto, Japan). Activity was calculated based on oxidation of ABTS as katals (mol s−1) per 1 g of fresh moss tissue. The A405 extinction coefficient of ABTS is 36 800 M−1 cm−1. Three replicate samples of the chitosan-treated and nontreated cultures were included from each time-point in each of the repeated experiments. Peroxidase induction and activity measurements were also performed during isolation and identification of the P. patens peroxidase (see later). In these latter cases, peroxidase activity was determined using ABTS, guaiacol (Fluka) or 4-aminoantipyrine (Nakarai Tesque, Inc., Kyoto, Japan) as the peroxidase substrate (data not shown).

The pH in which the peroxidase showed the highest activity was determined using samples taken 15 min after chitosan treatment as above, but the phosphate-citrate buffer used in the reaction mixture was adjusted to pH 2.6, 3.0, 3.6, 4.0, 4.6, 5.0, 5.6, 6.0, 6.6 or 7.0.

Protein isolation and amino acid sequencing

Protonemal tissue of moss (c. 120 mg) was transferred to 10 50-ml conical tubes (Sarstedt) filled with 10 ml of the liquid BCD medium and treated with chitosan as described earlier. The culture medium was collected 3 h after chitosan treatment and filtered (No. 4 paper; Whatman International Ltd, Maidstone, UK). Proteins were precipitated by adding ammonium sulfate ((NH4)2SO4) to 70% saturation followed by centrifugation at 12 000 g for 20 min at 4°C. Pellets were dissolved in 200 µl of phosphate buffered saline (PBS; 130 mm NaCl, 10 mm sodium phosphate (pH 7.0)) and subjected to gel filtration using a Superdex 75 HR 10/30 column (GE Healthcare, Little Chalfont, UK) equilibrated with buffer (150 mm NaCl, 50 mm sodium phosphate (pH 7.4)). Chromatography was done in PBS with a flow rate of 500 µl min−1 and absorbance at 214 nm and 280 nm was monitored. Fractions (0.5 ml) were collected and peroxidase activity was determined as above.

Gel filtration fractions that contained peroxidase activity were subjected to reversed-phase chromatography using a Jupiter C4 column (1.0 mm internal diameter × 150 mm long, 5 µm, 300 Å; (Phenomenex, Torrance, USA) and the SMART system (Pharmacia Biotech, Little Chalfont, UK). A linear gradient of acetonitrile (0–100% in 60 min) in 0.1% (v : v) trifluoroacetic acid was applied at a flow rate of 50 µl min−1. Absorbance at 214 nm was measured and 50-µl fractions were collected. Fractions containing protein were dried in a vacuum centrifuge and subjected to sodium dodecyl sulfate polyacrylamide gel (12%) electrophoresis (SDS-PAGE) (Laemmli, 1970) followed by silver staining (O’Connell & Stults, 1997). For identification and further characterization, protein bands were excised from the polyacrylamide gel and ‘in-gel’ digested using trypsin as described by Shevchenko et al. (1996). Proteins were reduced with dithiothreitol (DTT) and alkylated with iodoacetamide before digestion with trypsin (Sequencing Grade Modified Trypsin, V5111; Promega). A 15-µl aliquot of the recovered peptide mixture (50 µl) was desalted using a Millipore µ-C18 ZipTip pipette tip and subjected to matrix-assisted laser desorption/ionization time of flight mass spectrometric (MALDI-TOF/MS) analysis. The MALDI-TOF mass spectra for mass fingerprinting were acquired using an Ultraflex TOF/TOF instrument (Bruker-Daltonik GmbH, Bremen, Germany) and analysed using the Mascot Peptide Mass Fingerprint program (Matrix Science Ltd., Boston, MA, USA; http://www.matrixscience.com).

Electrospray ionization quadrupole TOF tandem mass spectra for de novo sequencing were generated from the rest of the recovered peptide mixture using a Q-TOF instrument (Micromass, Manchester, UK) connected to an Ultimate nano liquid chromatograph (LC-Packings, Amsterdam, the Netherlands) essentially as described by Poutanen et al. (2001).

Sequence analysis

The resultant peptide sequences of the peroxidase were subjected to tblastn (The National Center for Biotechnology Information, NCBI; Bethesda, MD, USA: http://www.ncbi.nlm.nih.gov) analysis to determine the corresponding peroxidase gene in the moss genome. The P. patens genomic sequence from the Joint Genome Institute (JGI), US Department of Energy (http://genome.jgi-psf.org/Phypa1_1/Phypa1_1.home.html) database was searched for putative peroxidases using hmmer software (Washington University School of Medicine, St Louis, MO, USA; http://hmmer.janelia.org/) and a peroxidase hidden Markov model (HMM; PF0014). The HMM search did not reveal any new peroxidase models compared with those already predicted by the JGI. Short and abnormal sequences were removed from the alignment. Predicted class III peroxidases were given a running number by their ascending location in the scaffolds. Thirty out of the 45 predicted class III peroxidases have been assembled from expressed sequence tags (ESTs) by the PeroxiBase database for peroxidases (University of Geneva, Geneva, Switzerland; http://peroxibase.isb-sib.ch/index.php) (see the Supporting Information, Table S1).

Phylogenetic analysis

Amino acid alignments of the 45 predicted class III peroxidases of P. patens were assembled as described by Duroux & Welinder (2003) and included two separate alignments using either 73 class III peroxidases from Arabidopsis thaliana (L.) Heynh. (Tognolli et al., 2002) or 138 class III peroxidases from rice (Oryza sativa L.) (Passardi et al., 2004). Briefly, the sequences were truncated according to mature horseradish peroxidase C (HRPC; Uniprot P00433) residues 1–305 in the clustalx (Thompson et al., 1997) alignment using the default parameters and then realigned without HRPC using the following parameters: Gonnet protein weight matrix, ‘Gap opening penalty’ = 10, ‘Gap extension penalty’ = 1, and ‘Delay divergent sequence’ = 70%. The same gap penalties were used for pairwise alignments. Phylogenetic analyses using the neighbour-joining and maximum parsimony methods were performed using the mega version 3.1 programme (Kumar et al., 2004) and subjected to 1000 bootstrap replicates.

RNA extraction, cDNA synthesis and quantitative PCR

Moss tissue frozen under liquid nitrogen was ground to powder. Total RNA was extracted using a Trizol-like reagent (Caldo et al., 2004) and the concentration determined using a GeneQuant 1300 spectrophotometer (Amersham). Total RNA (1 µg) was treated with RQ1 RNase-free DNase (Promega) according to the manufacturer's instructions. Expression levels of Prx34A/34B were estimated using quantitative PCR (qPCR). Briefly, 1 µg of the DNase-treated RNA (11 µl) and 1 µl (200 ng) of random hexamers (Sigma Genosys, Woodlands, TX, USA) were preincubated at 70°C for 10 min and then added to reaction mixtures containing 4 µl Moloney murine leukaemia virus (M-MLV) reverse transcription (RT) buffer (250 mm Tris-HCl (pH 8.3 at 25°C), 375 mm KCl, 15 mm MgCl2, 50 mm DTT), 2 µl 0.1 m DTT, 1 µl 10 mm dNTPs, 0.5 µl (20 U) RNasin (Promega) and 2 µl (400 U) M-MLV RT (Promega) in a total final reaction volume of 21.5 µl. Reaction mixtures were incubated at 37°C for 1 h, followed by 70°C for 10 min, after which 80 µl of nuclease-free water was added. qPCR using 5 µl of the cDNA as a template was performed using the LightCycler 480 SYBR Green system according to the manufacturer's instructions (Roche, Basel, Switzerland).

Primers for amplification of Prx34A/34B transcripts were designed (Table S2) to span the second intron of the predicted gene. The cDNA was checked by PCR amplification with these primers to verify absence of genomic DNA (expected product size 73 bp without the intron). The same primers were used for qPCR. Primers were also designed for the RUBISCO (RbcS) small subunit (GenBank accession no. AB120708) to be used as an internal control (Table S2). Target-specificity of the primers was verified by sequencing the amplification products. The normalized relative ratio of Prx34A/34B transcripts to RbcS transcripts was calculated with LightCycler 480 software (Roche) using the formula inline image, where E = primer efficiency, T = target (Prx34A/34B), R = reference (RbcS), C = control, S = treated sample, and Cp = threshold value (region of exponential amplification). Primer efficiencies were determined from a standard series (ET = 1.851 and ER = 1.937).

Production of gene replacement and site-directed insertional lines

The two identical copies (Prx34A and Prx34B) of the peroxidase gene were knocked out using a gene replacement technique (Strepp et al., 1998). For total DNA isolation, 100 mg P. patens protonema tissue frozen under liquid nitrogen was ground to powder, extracted with 500 µl hexadecyl trimethyl ammonium bromide (CTAB) buffer (1.4 m NaCl, 20 mm Na2-EDTA, 100 mm Tris-HCl, pH 8.0, 2% (w : v) CTAB) (Ausubel et al., 1995) containing 1 µg ml−1 RNaseA (Sigma) and incubated at 60°C for 1 h followed by chloroform extraction and centrifugation at 16 000 gfor 5 min at room temperature. The aqueous layer was transferred to a new tube. DNA was precipitated by adding 500 µl isopropanol and pelleted by centrifugation at 16 000 g for 4 min. The pellet was washed with 70% ethanol and re-suspended in 100 µl of TE buffer (10 mm Tris-HCl, pH 8.0; 1 mm Na2-EDTA).

The primers Prx5F and Prx5R (containing 5′NcoI and EcoRI restriction endonuclease sites, respectively) and Prx3F and Prx3R (containing 5′EcoRI and NotI restriction endonuclease sites, respectively) (Table S2) were designed to amplify a 579 bp fragment of the upstream flanking region (UFR) and a 576 bp fragment of the downstream flanking region (DFR), respectively, immediately adjacent to the open reading frames (ORFs) of Prx34A and Prx34B (Fig. S1). These flanking sequences are identical in the Prx34A and Prx34B loci, as are the ORFs. For construction of the gene replacement cassette and vector, the amplified UFR fragment was digested with NcoI and EcoRI and ligated to NcoI–EcoRI digested pGEM-T easy vector (Promega). The resultant vector was digested with EcoRI and NotI and the EcoRI–NotI digested downstream fragment was ligated in the vector. The resulting vector was digested with EcoRI. Cohesive ends were filled in using the Klenow fragment of DNA polymerase I (Fermentas, Vilnius, Lithuania) and subsequently dephosphorylated using calf intestine alkaline phosphatase (Fermentas). A plasmid containing a hygromycin resistance gene (aphIV gene from Escherichia coli) under control of the Cauliflower mosaic virus (CaMV) 35S promoter and followed by the CaMV polyadenylation signal was obtained from the John Innes Centre, Norwich, UK (http://www.pgreen.ac.uk/). The hygromycin selection cassette was released from the plasmid with EcoRV and inserted via blunt-end ligation into the modified pGEM-T easy vector containing the UFR and DFR fragments.

Moss protoplasts were prepared according to Schaefer et al. (1991). Protoplasts (300 µl, 1 × 106 protoplasts ml−1) were transformed with 30 µg of the NcoI–NotI linearized plasmid, as previously described (Schaefer et al., 1991). Transformed protoplasts were initially selected on cellophane placed on BCD medium supplemented with 1 mm CaCl2, 45 µm Na2-EDTA, 5 mm ammonium tartrate, 6.6% mannitol and 0.5% glucose in plates under constant light as above for 6 d. The regenerating protoplasts were then transferred on the cellophane membranes to plates containing BCD medium and hygromycin (30 µg ml−1 56685; Fluka, Seelze, Germany) and incubated for 2 wk. The protoplasts on cellophane were subsequently transferred to plates containing fresh medium and antibiotic and incubated for an additional 2 wk. This initial selection period was followed by picking the individual colonies on culture medium without the antibiotic for 2 wk and then transferring to antibiotic-containing medium for a final 2-wk selection period.

To verify replacement of the Prx34A/34B genes by the hygromycin resistance cassette in knockout transformants surviving the selection, PCR amplification of the 5′ integration site of the hygromycin cassette was performed using primer 5′HygR and either primer 5′Prx34AF or 5′Prx34BF (Table S2, Fig. S1) to generate a 826-bp or a 798-bp fragment, respectively. In addition, amplification of the 3′ integration site of the hygromycin cassette was performed using primers 3′HygF and 3′PrxR to generate a 744-bp fragment. Absence of the Prx34A and Prx34B ORFs was verified by PCR amplification (Fig. S1) using primers 5′Prx34AF and 5′Prx34AR, 5′Prx34BF and 5′Prx34BR, or 3′PrxF and 3′PrxR (Table S2). Gene replacement lines were those in which the absence of both copies of the gene was confirmed. Site-targeted insertion lines were those in which integration of the gene replacement cassette was detected in place of the 5′ or 3′ part of Prx34A/B but the opposite end of the gene could be detected by PCR. In all lines tested, the PCR products were also sequenced to verify their authenticity.

Ploidy levels of the moss lines were determined based on their nuclear DNA content (Schween et al., 2003), which was measured using flow cytometry at Plant Cytometry Services, Schijndel, the Netherlands.

Fungal isolates and inoculation

Fungi were collected from a moss, Racomitrium japonicum (Hedw.) (Grimmiaceae) in Japan and isolated on to malt agar (MA; 2% malt extract, 2% glucose, 0.1% peptone) and replated four subsequent times onto MA, each time using a small amount of hyphae from an edge of the colony to obtain a pure culture. DNA was extracted using DNeasy Plant Mini Kit (Qiagen) according the manufacturer's instructions. The internal transcribed sequences (ITS1 and ITS2) of ribosomal genes were amplified by PCR using the primers (White et al., 1990) ITS1 (5′-TCCGTAGGTGAACCTGCGG-3′) and ITS4 (5′-TCCTCCGCTTATTGATATGC-3′). The PCR reaction mixture (total volume 50 µl) contained 50 ng fungal DNA, 2.5 mm MgCl2, 200 µm dNTPs, 1 mm of each primer, 0.05 µg µl−1 bovine serum albumin, and 1.25 U of Taq DNA polymerase (Fermentas), in NH4SO4 buffer (75 mm Tris-HCl (pH 8.8 at 25°C), 20 mm (NH4)2SO4, 0.01% (v : v) Tween 20). The PCR cycling parameters were as follows: 94°C for 5 min; 34 cycles at 94°C for 30 s, 58°C for 30 s, and 72°C for 1 min; 72°C for 10 min. The amplification products were directly sequenced in both directions using the ABI 3730xl DNA Analyzer at Biomedicum, Molecular Medicine Sequencing Laboratory, University of Helsinki, Finland. The resulting sequences (530 nucleotides) was subjected to blast search at NCBI (Bethesda, MD, USA) and were 98% identical to the sequenced of Irpex lacteus isolate WB 832 (accession no. AY569562.1) and 98% identical to Fusarium avenaceum (accession no. EU255801.1). The shape and size of conidia that were produced by the Fusarium sp. on inoculated knockout moss colonies were typical of F. avenaceum. However, because further characterization was not performed, the fungal isolates were designated as Irpex sp. and Fusarium sp. until further study.

Colonies of wild type and gene replacement lines of P. patens were grown for inoculation as described for plant material, except that cultures inoculated with Fusarium sp. were grown on media without ammonium tartrate. When 1 month old, the moss colonies were inoculated with fungi using a small amount of hyphal mass scraped from the edge of a fungal culture actively growing on potato dextrose agar (PDA; Biokar Diagnostics, Pantin Cedex, France). Culture of inoculated moss colonies was continued under the same conditions. The fungi were re-isolated from moss colonies to PDA and identified. No other microbes were detected.


Identification of a secreted peroxidase upon elicitation with chitosan

We observed that treatment of an axenic culture of P. patens with chitosan caused an immediate, substantial level of peroxidase activity detected in the culture medium in the liquid cultures in vitro. By contrast, little peroxidase activity was detected in untreated moss cultures followed for up to 5 h (Fig. 1a,b). Similar preliminary results were obtained with an axenic culture of another moss, R. japonicum (Fig. S2).

Figure 1.

Peroxidase activity in culture medium following chitosan treatment. Peroxidase activity was detected in the liquid culture medium following addition of chitosan to the liquid cultures of Physcomitrella patens. (a) Mean peroxidase activity 5 h following application of chitosan at different concentrations (n = 3). Error bars indicate standard deviation. (b) Mean peroxidase activity as a function of time (Basal level, 0, 5, 15, 30, 60, 180 and 300 min) following treatment with chitosan at 0.5 mg ml−1 (n = 3). Closed squares, mean values obtained following chitosan treatment; open squares, values of untreated controls. Error bars indicate standard deviation. (c) pH dependence of peroxidase activity. Peroxidase activity was measured in reactions at the indicated pH (adjusted using phosphate-citrate buffer). Error bars indicate standard deviation of the means (n = 3).

Measurement of the peroxidase activity in reaction mixtures with pH adjusted to 2.6–7.0 showed the highest peroxidase activity at pH 4.0 (Fig. 1c), which was used in further experiments. For characterization of the peroxidase, proteins were precipitated from the chitosan-treated liquid cultures and subjected to gel filtration. Peroxidase activity eluted in a single fraction and only traces of activity were detected in the previous and subsequent fraction (Fig. 2a). In the peroxidase-containing fraction (Fig. 2a) subjected to reversed-phase chromatography a single protein peak was detected (Fig. 2b). The estimated peroxidase activity in this fraction was 4500 nkat mg−1 protein.

Figure 2.

Isolation of the peroxidase from chitosan-induced cultures of Physcomitrella patens. (a) Elution curve of gel filtration and detection of peroxidase activity in the 0.5-ml fractions of a protein isolate from the liquid culture medium of chitosan-induced moss. Line indicates absorbances at 214 nm. Open bars indicate peroxidase activity. (b) Purification of the peroxidase. Reversed-phase elution curve of the active peroxidase fraction is shown. Only a single protein peak is detected. The insert shows the corresponding 35-kDa protein analysed by sodium dodecyl sulfate–polyacrylamide gel electrophoresis.

The putative peroxidase migrated corresponding to a molecular weight of c. 35 kDa in sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE; insert in Fig. 2b). For identification and further characterization, the peroxidase band was cut out from the gel and ‘in-gel’ digested with trypsin. Peptide mass fingerprint analysis of the tryptic peptides by MALDI-TOF MS and Mascot search gave no significant matches to any peroxidases among accessions of the Mass Spectrometry Protein Sequence Data Base (MSDB; http://www.matrixscience.com/help/seq_db_setup_msdb.html). At the time of this search the genome sequence of P. patens was not available. For further characterization of the protein the tryptic digest was subjected to partial sequence analysis by liquid chromatography-electrospray ionization tandem mass spectrometry (LC-ESI-MS/MS) with results shown in Table 1.

Table 1.  Identification of the Physcomitrella patens peroxidase Prx34 by peptide mass fingerprinting and tandem mass spectrometry
Determined protonated mass (Da)Predicted neutral mass (Da)Position in mature proteinPeptide sequence
  1. The determined monoisotopic masses in the tryptic matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF/MS) peptide mass fingerprint are shown together with masses, positions and sequences of tryptic peptides predicted from Pp-Prx34. Sequences in bold type were confirmed by LC-ESI-MS/MS de novo sequencing. Cysteine residues were detected as a carbamidomethyl derivative.


The five resultant peptide sequences (Table 1) and the tblastn program were used to detect the peroxidase-encoding gene in the genome sequence of P. patens (Rensing et al., 2008) at the Joint Genome Institute (JGI), US Department of Energy (http://genome.jgi-psf.org/Phypa1_1/Phypa1_1.home.html). All five sequences showed a perfect match with a putative peroxidase for which two identical genes exist in the P. patens genome (JGI protein IDs 144819 and 144797). The aforementioned tryptic peptide mass fingerprint now also matched to this putative peroxidase with a sequence coverage of 60% (peak intensity coverage of 72%). The peptide sequences obtained by tandem mass spectrometry covered 19% of this protein. The two copies of the gene were designated as Prx34A and Prx34B, and the peroxidase protein is referred to as Prx34.

Comparison of Prx34 with class III peroxidases in vascular plants

The Prx34 protein is 330 amino acids (aa) long, including an N-terminal 26-aa signal peptide, and is targeted to the secretory pathway according to analysis with the signalp (Bendtsen et al., 2004) and targetp (Emanuelsson et al., 2000) programs. One of the peptide masses (1574.812 Da) in the tryptic peptide mass fingerprint corresponds to the N-terminal tryptic peptide (NSYNLRPNYYSGK, 1574.748 Da) of the mature protein (Table 1), which confirmed the suggested cleavage site of the 26-aa signal sequence. Analysis using blastp indicated that Prx34 is similar (E-Value 7e-91) to the class III secretory peroxidase (cd00693) of the plant heme-dependent peroxidase superfamily and contains the conserved domains characteristic of this family of peroxidases (Fig. 3). The blastp analysis also indicated that Prx34 is similar to many other plant peroxidases of class III, or their precursors (Fig. 3).

Figure 3.

Conserved domains among class III peroxidases. Amino acid sequence alignment of the Physcomitrella patens mature peroxidase Prx34 with selected members of class III peroxidases. Conserved residues important for catalytic activity are in bold type. Shading indicates the substrate access channel lining residues (according to horseradish peroxidase HRPC1A; Gajhede et al., 1997). Cysteine residues forming disulfide bridges are boxed. The Ca2+-binding residues are indicated by *, and the Asp–Arg salt bridge residues are indicated by +. The peroxidases are: HRPC1A, horseradish (Armoracia rusticana) (UNIPROT accession P00433); TAP1, tomato (Solanum lycopersicum) (P15003); WP1, wheat (Triticum aestivum) (Q43212); BP1, barley (Hordeum vulgare) (Q40069); TOP A, tobacco (Nicotiana tabacum) (P11965); Os10g0536700, rice (Oryza sativa) (Q8W2X2); POD6, cotton (Gossypium hirsutum) (Q8RVP4); MpPOD1, liverwort (Marchantia polymorpha) (Q8H958).

Analysis of the genome sequence of P. patens revealed 48 putative peroxidase genes. Among these, sequences of 45 genes were unique, whereas the gene (Prx34) for the peroxidase characterized in this study and two other peroxidase genes (Prx8 and Prx28) were present as two copies (designated as A/B) (Table S1). In Arabidopsis thaliana, the majority of peroxidase genes contain a conserved ‘classical’ exon–intron pattern consisting of four exons and three introns (Tognolli et al., 2002) (Fig. 4). Variations from this pattern in A. thaliana include loss of one or more introns (patterns A–E) or a gain of an intron (pattern D+3″) (Fig. 4). In P. patens, the ‘classical’ pattern was the most common followed by pattern C. The other patterns of Arabidopsis were also present. Additional novel patterns in P. patens resulted from modulation of patterns A and C by one additional intron at the 3′-proximal end of the gene (A+3″ and C+3″) or a variant of D (D+5″+3″) containing an additional intron at both ends of the gene (Fig. 4, Table S1). The exon–intron pattern of Prx34A/B predicted based on the genomic sequence (Table S2) was in agreement with the Prx34 amino acid sequences (60%) obtained by mass fingerprinting (Table 1).

Figure 4.

Schematic representation of the predicted peroxidase gene exon–intron patterns from Physcomitrella patens. Thick bars represent exons and lines indicate introns. The names of the exon–intron patterns are indicated at left, and the number of genes contained in each pattern is indicated at right. Patterns A+3″, A+5″+3″, C+3″ and D+5″+3″ are found only in P. patens, whereas the remaining patterns are found also in Arabidopsis thaliana (Tognolli et al., 2002). Prx34 exhibits pattern C.

The deduced peroxidase aa sequences of P. patens were subjected to phylogenetic analysis in comparison with the 73 and 138 putative class III peroxidases whose genes are predicted in the genomes of A. thaliana (Tognolli et al., 2002) and rice (Oryza sativa) (Passardi et al., 2004), respectively. Eleven peroxidases of P. patens including Prx34 formed a cluster different from the peroxidases of A. thaliana (Fig. 5) and rice (Fig. S3).

Figure 5.

Phylogenetic analysis of class III peroxidases. Phylogenetic tree based on comparison of the deduced amino acid sequences of 45 peroxidases from Physcomitrella patens and 73 peroxidases of Arabidopsis thaliana. The tree was derived from neighbor-joining analysis using 1000 bootstrap replicates.

Induction of the Prx34 gene following chitosan treatment

Quantification of the Prx34A/34B mRNA levels by real-time-PCR indicated a low level of constitutive expression that was not altered for 15 min following treatment with chitosan (Fig. S5). Subsequently, at 30, 60 and 180 min, expression levels were increased by 1.7-, 4.0- and 3.5-fold, respectively, in the chitosan-treated cultures compared with the untreated controls (P < 0.05) (Fig. 6), which however occurred later than the release of Prx34 and the increase of peroxidase activity in the culture medium. These data were consistent with storage of class III peroxidases in vacuoles and cell walls from which they can be rapidly released (Passardi et al., 2005) and induction of peroxidase genes following pathogen infection (Van Loon et al., 2006).

Figure 6.

Real-time-PCR (qPCR) analysis of Prx34A/34B expression following treatment with chitosan. Expression levels of mRNA from chitosan-treated tissues of Physcomitrella patens (closed bars) and nontreated controls (open bars) grown in liquid culture. Samples were collected at the indicated time-points following addition of chitosan (final concentration 0.5 mg ml−1) and subjected to qPCR analysis. The Rubisco small subunit (AB120708) was used as an internal control for normalization. Error bars represent standard deviation (control, n = 2; chitosan-treated, n = 4).

Prx34 knockout mutants do not release peroxidase upon chitosan treatment

The two identical copies of the peroxidase gene (Prx34A/34B) were knocked out in P. patens using a gene replacement technique (Fig. S1). Eight stable gene replacement lines and three site-targeted insertion lines were regenerated under selection for hygromycin resistance, as described by Schaefer et al. (1991). Integration of the gene replacement cassette was verified by PCR, as exemplified in Fig. S1b, and sequencing of the products. All lines were found to be haploid similar to the wild type P. patens, as determined by measurement of nuclear DNA content. Peroxidase activity was not detected in the liquid culture medium of the knock-out mutants as followed for up to 3.0 h after chitosan treatment, whereas in the cultures of wild type P. patens, peroxidase activity was substantially elevated as before (Fig. S4).

Prx34 knock-out mutants are susceptible to fungi

The Prx34A/34B mutant lines were phenotypically (morphology, growth rate) indistinguishable from wild-type P. patens (Fig. 7). They were all tested by inoculation with pure cultures of a saprophytic fungal isolate of genus Irpex (Basidiomycota; Polyporales; family Steccherinaceae) and a pathogenic isolate of Fusarium sp. (Hyphomycetes), both isolated from R. japonicum growing in the field, in four and two experiments, respectively. Both fungi grew on wild-type P. patens (Figs 3k, S5) but caused only some mild discoloration of stems (both isolates), a few small lesions on basal parts of the shoot and in rhizoids in vitro (Irpex) and death of a few shoots (Fusarium) as observed 14 d and 90 d post-inoculation, respectively (Fig. 3e,g,i,k). However, the growth of hyphae was abundant and symptoms were severe in all gene replacement and site-targeted insertion lines and most shoots died within 2-wk post-inoculation (Fig. 3f,h,j,l). A total of 17 additional gene replacement lines targeting genes involved putatively in cell wall synthesis or organogenesis (our unpublished materials) were also inoculated with these fungi. These gene replacement lines were not damaged more than was wild-type P. patens (data not shown), consistent with the altered response to fungal infection being caused by mutation of the Prx34A/B locus. In all experiments, the inoculated fungal isolates were reisolated from the moss cultures to fulfill Koch's postulates. No other microbes but the inoculated fungal isolate were detected.

Figure 7.

Altered response of Physcomitrella patens to fungal infection following removal of the genes for chitosan-responsive peroxidase (Prx34). (a) Growth of healthy, axenic, 6-wk-old cultures of the wild-type (wt) P. patens (five colonies) and (b) eight independent gene replacement lines (colonies 3–8, 10 and 11) and three site-targeted insertion lines (colonies 1, 2, and 9). (c) A close-up of the gametophores in the wt P. patens and (d), a knock-out line (colony 7 shown in b). (e–h) Moss colonies inoculated with Irpex sp. 14 d post-inoculation (dpi), and (i–l), with Fusarium sp. 90 dpi. (e,g,i,k) Some discoloration in the basal part of gametophores and rhizoids (both fungi), and (k) death of a few shoots (Fusarium) in wt P. patens. (f,h,j,l) Death of the mutant lines caused by the fungal isolates.


Here, we report that the peroxidase activity detected in cultures of P. patens was caused by a single peroxidase enzyme quickly released to the culture medium as a response to chitosan. Following purification of the peroxidase and its identification by mass spectrometric methods, we generated gene replacement lines of P. patens in which the two copies of the gene for the chitosan-responsive peroxidase were knocked out. No peroxidase was released from these lines upon treatment with chitosan. Furthermore, these gene replacement lines of P. patens died following inoculation with fungi that naturally colonize moss but that cause only mild or negligible damage to wild-type P. patens grown in vitro. The data identify a peroxidase that functions as a resistance factor used against invading fungi in a nonvascular plant.

The results show that the moss, P. patens, can recognize chitosan and responds by releasing a single class III peroxidase, Prx34, which is one of 45 different putative peroxidases encoded by the moss genome. Prx34 belongs to a group of peroxidases that is unique to P. patens according to phylogenetic analysis and has probably evolved after radiation of plants to different lineages such as bryophytes, monocots and dicots 400 million years ago (Shaw & Goffinet, 2000; Lang et al., 2008). Data suggest that P. patens is armed with Prx34 to protect itself against chitin-containing invaders, such as fungi. Besides this first peroxidase isolated and characterized from a moss, the liverwort M. polymorpha (synonym M. aquatica (Nees) Burgeff) is known to secrete a peroxidase in response to chemical stress (Hirata et al., 2000).

In vascular plants, cultured suspension cells can secrete peroxidases (Chibbar et al., 1984; Chivasa et al., 2005, 2006), and class III peroxidases isolated from plant tissues can have antifungal activity (Ghosh, 2006). Furthermore, elicitation of pathogen defense by chitosan has been widely studied in vascular plants (Rabea et al., 2003; Miya et al., 2007). However, direct evidence for the role of peroxidases in antifungal defense has been lacking (Almagro et al., 2009), whereas this study shows that chitosan treatment induces the release of a peroxidase that is needed for antifungal defense.

Further details of the mechanism by which chitosan or chitin is recognized and Prx34 is released require further study. However, the results of this study are consistent with the location of class III peroxidases in vacuoles and cell walls, hence providing a reservoir that can be released quickly upon induction (Almagro et al., 2009). The data demonstrate that Prx34 is pivotal in protecting P. patens against hostile invasion by saprophytes (Irpex) and fungi such as Fusarium that are pathogenic on other mosses. Hence, Prx34 may be a component of a basal defense system (Mitchell & Power, 2003) that protects the moss against many potential pathogens and pathogens such as Fusarium that are able to overcome defense in another moss species, R. japonicum. Previous studies have provided indications that mosses contain resistance-like genes (Akita & Valkonen, 2002) and P. patens can be induced to resist a bacterial pathogen of vascular plants (Pectobacterium carotovorum; formerly Erwinia carotovora subsp. carotovora) by treatment with the plant defense signaling compound, salicylic acid, as also occurs in vascular plants (Andersson et al., 2005). Strains of P. carotovorum also need similar virulence factors for infecting P. patens and vascular plants (Andersson et al., 2005). Induction of homologs of the defense-related genes CHS, LOX, PAL and PR-1 has been observed in P. patens following treatment with cell-free culture filtrates of P. carotovorum and infection with the gray mold fungus Botrytis cinerea (De Bary) Whetzel, a pathogen of vascular plants (Ponce de Leon et al., 2007). However, the present study singles out the first host factor which nonvascular plants need in defense against fungi.


We are grateful to Anna Kärkönen for helpful discussions about peroxidase assays, Jaakko Hyvönen and Soili Stenroos for collaboration, and Anssi Vuorinen for the help with qRT data analysis. Financial support from the Finnish Graduate School of Plant Biology and the Academy of Finland (grant 1118766) is gratefully acknowledged.