Plant Defensin type 1 (PDF1): protein promiscuity and expression variation within the Arabidopsis genus shed light on zinc tolerance acquisition in Arabidopsis halleri


  • Zaigham Shahzad,

    1. Biochimie et Physiologie Moléculaire des Plantes, Unité Mixte de Recherche Montpellier, SupAgro/CNRS/INRA/Université Montpellier II, Montpellier Cedex 1, France
    Search for more papers by this author
  • Vincent Ranwez,

    1. Montpellier SupAgro, UMR AGAP, Montpellier, France
    Search for more papers by this author
  • Cécile Fizames,

    1. Biochimie et Physiologie Moléculaire des Plantes, Unité Mixte de Recherche Montpellier, SupAgro/CNRS/INRA/Université Montpellier II, Montpellier Cedex 1, France
    Search for more papers by this author
  • Laurence Marquès,

    1. Biochimie et Physiologie Moléculaire des Plantes, Unité Mixte de Recherche Montpellier, SupAgro/CNRS/INRA/Université Montpellier II, Montpellier Cedex 1, France
    Search for more papers by this author
  • Bénédicte Le Martret,

    1. Biochimie et Physiologie Moléculaire des Plantes, Unité Mixte de Recherche Montpellier, SupAgro/CNRS/INRA/Université Montpellier II, Montpellier Cedex 1, France
    Search for more papers by this author
  • Julien Alassimone,

    1. Biochimie et Physiologie Moléculaire des Plantes, Unité Mixte de Recherche Montpellier, SupAgro/CNRS/INRA/Université Montpellier II, Montpellier Cedex 1, France
    Search for more papers by this author
  • Cécile Godé,

    1. Laboratoire de Génétique et Evolution des Populations Végétales, UMR CNRS 8016, Université des Sciences et Technologies de Lille, Villeneuve d'Ascq Cedex, France
    Search for more papers by this author
  • Eric Lacombe,

    1. Biochimie et Physiologie Moléculaire des Plantes, Unité Mixte de Recherche Montpellier, SupAgro/CNRS/INRA/Université Montpellier II, Montpellier Cedex 1, France
    Search for more papers by this author
  • Teddy Castillo,

    1. Biochimie et Physiologie Moléculaire des Plantes, Unité Mixte de Recherche Montpellier, SupAgro/CNRS/INRA/Université Montpellier II, Montpellier Cedex 1, France
    Search for more papers by this author
  • Pierre Saumitou-Laprade,

    1. Laboratoire de Génétique et Evolution des Populations Végétales, UMR CNRS 8016, Université des Sciences et Technologies de Lille, Villeneuve d'Ascq Cedex, France
    Search for more papers by this author
  • Pierre Berthomieu,

    1. Biochimie et Physiologie Moléculaire des Plantes, Unité Mixte de Recherche Montpellier, SupAgro/CNRS/INRA/Université Montpellier II, Montpellier Cedex 1, France
    Search for more papers by this author
    • These authors contributed equally to this work.
  • Françoise Gosti

    Corresponding author
    1. Biochimie et Physiologie Moléculaire des Plantes, Unité Mixte de Recherche Montpellier, SupAgro/CNRS/INRA/Université Montpellier II, Montpellier Cedex 1, France
    Search for more papers by this author
    • These authors contributed equally to this work.


  • Plant defensins are recognized for their antifungal properties. However, a few type 1 defensins (PDF1s) were identified for their cellular zinc (Zn) tolerance properties after a study of the metal extremophile Arabidopsis halleri. In order to investigate whether different paralogues would display specialized functions, the A. halleri PDF1 family was characterized at the functional and genomic levels.
  • Eleven PDF1s were isolated from A. halleri. Their ability to provide Zn tolerance in yeast cells, their activity against Fusarium oxysporum f. sp. melonii, and their level of expression in planta were compared with those of the seven A. thaliana PDF1s. The genomic organization of the PDF1 family was comparatively analysed within the Arabidopsis genus.
  • AhPDF1s and AtPDF1s were able to confer Zn tolerance and AhPDF1s also displayed antifungal activity. PDF1 transcripts were constitutively more abundant in A. halleri than in A. thaliana. Within the Arabidopsis genus, the PDF1 family is evolutionarily dynamic, in terms of gain and loss of gene copy.
  • Arabidopsis halleri PDF1s display no superior abilities to provide Zn tolerance. A constitutive increase in AhPDF1 transcript accumulation is proposed to be an evolutionary innovation co-opting the promiscuous PDF1 protein for its contribution to Zn tolerance in A. halleri.


Defensins represent a large class of small peptides that are widely distributed in Animalia and Plantae kingdoms. They are members of the antimicrobial peptide (AMP) superfamily (Thomma et al., 2002; Ganz, 2003; Brown & Hancock, 2006), and present the archetypal γ-core signature motif of these membrane-active host defence polypeptides, which could be traced back to prokaryotic origins (Yount & Yeaman, 2004, 2006; Yeaman & Yount, 2007). Plant defensins, in particular, are characterized by a cysteine-stabilized α-helix β-sheet three-dimensional structure (CSαβ motif), also conserved over different species (Zhu et al., 2005). Early studies recognized plant defensins for their antifungal activity and abundant expression in seeds; hence they were considered to protect germination and seedling growth from soilborne pathogens (Terras et al., 1995). Later on, a cumulative number of defensins and defensin-like proteins were discovered in different plant species (Mergaert et al., 2003; Graham et al., 2004; Silverstein et al., 2005, 2007). Concomitant studies clearly indicate that defensins are expressed in all plant tissues and that they are involved in a wide range of numerous biological activities and physiological processes in response to different biotic and abiotic stresses. All of these characteristics have been largely documented in recent reviews (Lay & Anderson, 2005; Wong et al., 2007; Carvalho & Gomes, 2009, 2011; De Coninck et al., 2013; Van der Weerden & Anderson, 2013). These data obtained in different organisms locate defensins at the crossroads between biotic and abiotic plant responses, the outcome of which will ultimately allow plants to adapt to environments that are challenging for their survival (Fujita et al., 2006; Atkinson & Urwin, 2012). Despite this increased knowledge and their functional importance, defensins are still puzzling peptides. The mode of action regarding defensin antifungal and antimicrobial activities remains imprecise, with each studied defensin having specific characteristics (Sagaram et al., 2012; Thevissen et al., 2012; De Coninck et al., 2013). The link between the different defensin activities is often established on the basis of sequence congruence and few studies show that the same defensin molecule supports different activities. This is an important point to address, because it could well be that different members of these large gene families ultimately support different defensin activities, as gene duplications are postulated to provide raw genetic material on which adaptive modification can occur (Ohno, 1970).

In order to gain insight into the functional diversity of these bewildering peptides, the study of Plant Defensin type 1 genes (PDF1s) offers a particularly interesting entry point. The PDF1 multigenic family is well described in Arabidopsis thaliana and includes seven members (Thomma et al., 2002), out of which several encoded proteins have been purified and characterized for antifungal activity (Terras et al., 1993; Penninckx et al., 1996; Sels et al., 2007). Interestingly, PDF1s have also been shown to play a role in zinc (Zn) tolerance (Mirouze et al., 2006) after a study of the extremophile species Arabidopsis halleri, which displays high capacities to tolerate and hyperaccumulate Zn and cadmium. These characters are not present in the closest relatives, Arabidopsis lyrata and A. thaliana, from which A. halleri diverged 0.27–0.44  and 5.8–13 million yr ago (MYA), respectively (Koch & Matschinger, 2007; Schranz et al., 2007; Beilstein et al., 2010; Roux et al., 2011). The role of PDF1s in Zn tolerance was documented in wildtype yeast through functional heterologous screening of a cDNA library of A. halleri that revealed four AhPDF1 cDNAs (Mirouze et al., 2006), and in planta where transgenic A. thaliana lines expressing one of the AhPDF1 paralogues exhibited increased Zn tolerance (Mirouze et al., 2006).

Regarding Zn tolerance, PDF1s are novel because, according to their Gene Ontology annotation, they have no documented roles in metal transport or chelation. In order to gain an insight into the functional diversity of defensins in relation to their role in Zn tolerance in planta, we wanted to track the evolution of PDF1s in A. halleri, compared with A. thaliana, in terms of PDF1 functional capacity, gene duplication and transcript accumulation. A limitation of this study was the lack of description of the whole PDF1 family in the A. halleri species, for it could well be that the four AhPDF1 cDNAs identified by functional screening (Mirouze et al., 2006) represented neither the whole PDF1 family in A. halleri nor their functional diversity. As PDF1 in A. thaliana exists as a multigenic family, the identification of orthologous relationships linking PDF1s was a prerequisite. Only then, it becomes possible to investigate functional changes occurring between orthologous genes (i.e. genes related by a speciation event at their most recent point of origin) and those occurring between paralogues (i.e. genes related from duplication events (Fitch, 1970; Koonin, 2005; Kuzniar et al., 2008; Kristensen et al., 2011)).

The present study provides a genomic description of the PDF1 family in A. halleri and specifies the orthologous relationships among PDF1s represented in the Arabidopsis genus. This highlights the evolutionary dynamism of this family and opens the possibility of PDF1 gene retention in the A. halleri lineage. Functional studies show that AhPDF1s and AtPDF1s are promiscuous proteins for antifungal and Zn tolerance properties and that the major characteristic differentiating A. halleri PDF1s from their orthologues in A. thaliana species is their higher constitutive transcript accumulation. We propose that this evolutionary innovation occurred independently of gene amplification in the A. halleri Zn-tolerant species and provided a means to enhance, in this lineage, a PDF1 function that is shared among species of the Arabidopsis genus.

Materials and Methods

BAC clone screening, sequencing and annotation

A bacterial artificial chromosome (BAC) library constructed from a single A. halleri plant (Lacombe et al., 2008) was screened by Southern hybridization as described in Lacombe et al. (2008) with three independent PDF1 probes. Probes were made from PCR products amplified from AhPDF1 cDNA (AY961376), and from AtPDF1.4 (gene ID: 838548) and AtPDF1.5 (gene ID: 841943) genomic DNA using the primer pairs described in Supporting Information, Table S1. Further, restriction fragment length polymorphism (RFLP) analyses of BAC clones were performed as described in (Shahzad et al., 2010). BAC clones 5I19, 5F13, 10J04, 3E16 and 9G01 were fully sequenced using the shotgun method followed by a Sanger-based finishing step (Genoscope, Evry, France). BAC clones 1N20 and 12A7 were sequenced using the 454 Roche multiplexing technology (Titanium kit) without a finishing step (CNRGV, Toulouse, France). For these two BAC clones, only the AhPDF1-harbouring contigs were subsequently considered for analyses. The AhPDF1 sequences present in BAC clones 8B03, 6D13 and 2M24 were determined (GATC Biotech, Konstanz, Germany) after subcloning DNA fragments in the pBSKS+ vector (Statagene®; La Jolla, CA, USA). Genes were predicted using the FGENESH program (Salamov & Solovyev, 2000) under the SOFTBERRY software ( and were annotated based on BLAST similarity searches (Altschul et al., 1990). The presence of metal-related cis-acting motifs in AhPDF1 putative promoter sequences was searched using the last release of PLACE, SOFTBERRY and PlantCARE databases (PLACE,; Rombauts et al., 1999; PlantCARE,; Lescot et al., 2002; SOFTBERRY).

Genetic mapping of AhPDF1 paralogues

Using 62 published markers (Willems et al., 2007; Roosens et al., 2008b; Ruggiero et al., 2008; Frerot et al., 2010; Gode et al., 2012) and three new microsatellite markers defined for this study (Table S2), genotyping was performed using genomic DNAs of the parents of the A. halleri × A. lyrata petraea BC1 population and of 199 plants from this population. The new microsatellite markers were amplified in multiplex according to Gode et al. (2012). Mapping of AhPDF1-harbouring BAC clones was performed by PCR-dominant, cleaved amplified polymorphic sequence (CAPS) or simple sequence length polymorphism (SSLP) analysis, as described in (Willems et al., 2007) using the specific markers generated for this study (Table S3). Genotypes obtained in the BC1 population, for the BAC-derived markers, were combined with the dataset used to construct the A. halleri × A. lyrata petraea linkage map, as described in (Shahzad et al., 2010).

In silico identification of A. lyrata PDF1s, sequence and phylogenetic analysis

Arabidopsis lyrata PDF1s were identified by searching the whole genome shotgun (wgs) nucleotide sequence database using the tblastn program (Altschul et al., 1990) and the A. thaliana (TAIR, The Arabidopsis information resource,; Lamesch et al., 2012) and A. halleri (this study) deduced PDF1 protein sequences as queries (E-value < 1 × e−20). Seven homologous PDF1s (Table S4) were hence identified from the A. lyrata subsp. lyrata (taxid:81972) genome (Hu et al., 2011). All AlPDF1 annotations were controlled manually based on overall PDF1 nucleotide sequence alignments and the position of splicing sites. Consequently, this revealed the presence of a premature stop codon after the 39th amino acid for the AlPDF1.2a gene (Notes S1). The localizations of those AlPDF1s within the A. lyrata genome were obtained using EnsEMBL Plant facilities (Kersey et al., 2010, 2012).

Multiple sequence alignments were computed using MUSCLE v3.8.31 (Edgar, 2004) and visualized with BOXSHADE 3.21 ( run through the Mobyle Portal (MOBYLE,; Neron et al., 2009). Gene tree inference was done using the predicted protein sequences of the 25 Arabidopsis PDF1 homologues together with those of the six outgroup A. thaliana PDF2 paralogues (TAIR; Notes S1). Those sequences were aligned using MUSCLE v3.8.31 (Edgar, 2004) with default options. Poorly aligned regions were removed from this alignment using the ‘automated1’ option of trimAl v1.3 (Capella-Gutierrez et al., 2009), which is specifically designed to trim alignments to be used in maximum-likelihood phylogenetic analyses. The phylogeny of the PDF1 family was then inferred by maximum likelihood, using RAxML v7.2.8 (Stamatakis, 2006). Inference was done starting from 10 distinct randomly chosen maximum parsimony trees. The WAG protein model (Whelan & Goldman, 2001), using empirical base frequencies and a discrete Gamma law with four categories to model heterogeneity of evolutionary rate among sites, was chosen. Branch supports were estimated through full bootstrap analyses, as opposed to the faster RAxML bootstrap approximations. Identification of synteny conservation – that is ‘preserved co-localization of genetic loci and/or genes on chromosomes and/or linkage group of different species’ (Abrouk et al., 2010) – within PDF1 regions was manually done at the macroscopic (c. 20–50 kb) and microscopic (c. 1–5 kb) levels after careful visual examination of the sequence similarity between genomic sequences bordering the PDF1 genomic sequences (Table S5) using blastn with default parameters (Altschul et al., 1990). When possible, synteny was also explored using the viewer facilities offered by TAIR and BRAD (The genetics and genomics database for Brassica plants,; Cheng et al., 2012a) servers.

Antifungal assays

For all AhPDF1s, DNA fragments encoding the C-terminal mature part were cloned in the pET28(a+) vector (Novagen, Darmstadt, Germany) using the primers listed in Table S6. AhPDF1 proteins were produced as already described (Marquès et al., 2009), except that the final high-performance liquid chromatography (HPLC) purification step was omitted. The purity and mass of all the proteins were assessed by sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE) and matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectra. The protein concentration was measured using Bradford assays (Bradford, 2003). Growth inhibition assays were conducted against Fusarium oxysporum f. sp. melonii (INRA Avignon) as described in Marquès et al. (2009). The minimal inhibitory concentration (MIC) was determined as the lowest defensin concentration that induced 100% reduction of growth compared with the control conditions.

Zn tolerance assays in yeast

Entire coding sequences of A. halleri and A. thaliana PDF1s were cloned between EcoRI and XhoI restriction sites of the pYX212 yeast expression vector using the overlap extension (SOEing) PCR-based method (Vallejo et al., 1994) and primers listed in Table S7. The constitutive triose phosphate isomerase promoter controlled PDF1 expression. Only the second exon of AhPDF1.6 was cloned in the pYX212 vector, as only this part of the gene could be identified within the corresponding BAC clone. The recombinant pYX212 vectors were introduced into the BY4741 Saccharomyces cerevisiae strain (MATa, his3∆1, leu2∆0, met15∆0, ura3∆0) together with the pFL38H vector, which harbours a functional HIS3 gene, using the lithium acetate/single-stranded carrier DNA/polyethylene glycol method (Gietz & Woods, 2002). For drop assays, transformed yeast cells from overnight cultures, were washed twice with ultrapure H2O and diluted to OD600nm = 1, 0.1 and 0.01. Ten microlitres of the yeast solutions were dropped onto selective modified yeast nitrogen base (YNB) medium (1.7 g l−1 YNB without amino acids without ammonium sulfate (233520, Difco), 6.4 g l−1 NH4NO3, 2% (w/v) d-glucose, 50 mM succinic acid-KOH, pH 4.5) supplemented with ZnSO4 at various final concentrations, 1.4 μM for the control condition and 25 or 27.5 mM for Zn treatments, and grown at 30°C. At least three independent clones were tested for each construct. A MIC assay against the BY4741 S. cerevisiae clone used was conducted and no inhibitory action was found for values up to 20 μM for all the 11 A. halleri defensins tested as previously found for the AhPDF1.1b defensin (Marquès et al., 2009).

Quantification of transcript accumulation

The SAF2 line of A. halleri used in this study was derived from an individual plant from the metallicolous Auby population (France) through vegetative multiplication. In vitro micropropagation and phenotypic analyses were performed as already described (Shahzad et al., 2010). The A. thaliana (Columbia) plants were grown from seeds in the same sterile hydroponic medium and harvested after 2 wk. All functional analyses were performed from plants cultured in sterile conditions. The presence of each AhPDF1 within the SAF2 plant genome was verified and their sequences were checked (data not shown). Transcripts were quantified from five plant replicates treated independently. Roots and shoots were harvested separately and analysed independently. RNA extraction, cDNA synthesis and real-time reverse transcription polymerase chain reaction (RT-PCR) were performed essentially as described (Shahzad et al., 2010). Specific primer pairs were designed for each different PDF1 (Table S8). Actin was considered as an internal control (Shahzad et al., 2010). The PCR efficiency (E) of each PDF1 specific primer pair was determined after the analysis of five serial 1 : 10 dilutions of BAC clone DNA for AhPDF1s and plasmid DNA for AtPDF1s (Table S8), as previously described (Shahzad et al., 2010). The PCR efficiency of primer pairs used for Actin was determined on the gDNA for A. halleri as well as A. thaliana (Table S8). PCRs were performed on cDNA samples in triplicate and PDF1s vs Actin relative expression levels (RELs) were determined as previously described (Shahzad et al., 2010) and are listed in Table S9. Difference in transcript accumulation upon Zn addition was statistically validated by t-test at a 0.05 confidence threshold.


Identification of PDF1s in A. halleri and A. lyrata

PDF1 genes present in the A. halleri genome (AhPDF1) were identified after a BAC library screening. A total of 26 identified BAC clones (Table 1) were arranged into 10 groups depending on both EcoRI and BamHI RFLP profiles (data not shown). Sequences of BAC clones representing each of the 10 groups identified 13 AhPDF1s (Table 1) that all contained the characteristic PDF two-exon genomic structure (Silverstein et al., 2005), except for AhPDF1.6, which was missing the first exon and intron and which we thus considered as a probable pseudogene. All the remaining AhPDF1s encoded 78- to 80-amino-acid-long predicted proteins, as already reported for plant PDF1s (Carvalho & Gomes, 2011). AhPDF1.7 contains a four-nucleotide-long insertion in its coding sequence, modifying the reading frame and leading to a truncated protein as a result of a stop codon after the 54th amino acid (Notes S1). In some cases, two PDF1s were present on the same BAC, such as AhPDF1.1a and AhPDF1.1b, AhPDF1.2a and AhPDF1.2c, and finally AhPDF1.8a and AhPDF.8b, being separated by a distance of c. 3.4 kb, c. 13.6 kb and > 1.8 kb, respectively. On the basis of a 100% sequence identity, three of the four AhPDF1 cDNAs previously identified by yeast functional screening (Mirouze et al., 2006) were also identified in this study (Table 1). No identity was found for the fourth cDNA (AY961377). Using specific primer pairs, no PCR amplicon could be generated for this gene when assayed on 34 A. halleri plants from the Auby population (data not shown). We considered that the AY961377 PDF1 corresponded to a very rare PDF1 allele, or to a mutated version generated during the cDNA library construction. Hence, this gene was removed from further analyses.

Table 1. Identification of PDF1 paralogues in Arabidopsis halleri
Locus containing PDF1 genesPDF1 genes identified in A. halleriNumber of BAC clones harbouring the AhPDF1 paralogueName of the representative BAC cloneAccession number
  1. a

    BAC, bacterial artificial chromosome. The cDNA of AhPDR1.1a is also known as AY961379 (Mirouze et al., 2006).

  2. b

    The cDNA of AhPDR1.1b is also known as AY961376 (Mirouze et al., 2006).

  3. c

    The cDNA of AhPDR1.2b is also known as AY961378 (Mirouze et al., 2006).

  4. d

    The AhPDR1.5 isolated from BAC clones 5F13 and 2Ms4 correspond to two allelic forms of the same gene.

  5. e

    The BAC clones 5F13 and 2M24 are allelic forms of the same region.

  6. f

    The AhPDF1.8a isolated from either BAC clones 12A07 or 9G01 correspond to two allelic forms of the same gene.

  7. g

    The BAC clones 12A07 and 9G01 are allelic forms of the same region.

  8. h

    The gene AhPDF1.8b was identified only from BAC clone 9G01.

Locus 1 AhPDF1.1a a 48B03 HF545648
AhPDF1.1b b
Locus 2 AhPDF1.2a 41N20 HF545651
Locus 3 AhPDF1.2b c 25I19 HE601752
Locus 4 AhPDF1.4 26D13 HF545647
Locus 5 AhPDF1.5 d 15F13e HE601748
AhPDF1.5 22M24 HF545649
Locus 6 AhPDF1.6 210J04 HE601750
Locus 7 AhPDF1.7 23E16 HE601749
Locus 8 AhPDF1.8a f 312A07g HF545646
AhPDF1.8b h HF563610
AhPDF1.8a 49G01 HE601751
Total 8112610 

Seven PDF1s were identified in the A. lyrata genome through an in silico analysis (AlPDF1 and Table S4). In summary, 13 AhPDF1 and seven AlPDF1 have been identified which reflect the PDF1 genetic diversity described in the A. thaliana genome (AtPDF1, TAIR).

Genetic mapping of PDF1 in A. halleri and orthologous relationships within the Arabidopsis genus

The 10 representative AhPDF1-harbouring BAC clones were mapped at eight loci on the A. halleri × A. lyrata linkage map (Fig. 1, Table 1). BAC clones 5F13 and 2M24 on one side, and 12A07 and 9G01 on another side, were colocalized at loci 5 and 8, respectively (Fig. 1). Interestingly, the AhPDF1s harboured by these BAC couples had high nucleotide sequence identity (99 and 97%, respectively), even when considering noncoding PDF1-flanking regions (data not shown). This suggested that the corresponding colocalized AhPDF1s could be allelic rather than closely linked loci. Using two different allele-specific markers designed for every PDF1 gene present at loci 5 and 8 (Table S10), the distribution of these genes was examined in a set of individual plants of the Auby population. A balanced distribution of these PDF1 markers was observed (Table S11), indicating that PDF1s mapped at locus 5 as well as those mapped at locus 8 indeed corresponded to distinct allelic forms. Interestingly, the BAC clone 12A07 (mapped at locus 8) harboured two PDF1 copies (AhPDF1.8a and AhPDF1.8b) while 9G01 (locus 8 also) harboured only one (AhPDF1.8a). Using a copy-specific marker, AhPDF1.8b was detected in only 59 of the 83 tested plants, indicating that this paralogue was thus probably unfixed in the A. halleri Auby population.

Figure 1.

Genetic mapping of AhPDF1s on the Arabidopsis halleri × Arabidopsis lyrata petraea BC1 linkage map. The mapping was performed from an analysis of 199 plants and the four parents of an A. halleri × A. lyrata petraea BC1 population using Joinmap 3.0. Only linkage groups (LGs) 1, 2, 4, 7 and 8 are shown. The AhPDF1s are indicated in red letters. The markers presented on the map are described in Supporting Information Table S2. They are named according to the gene or bacterial artificial chromosome (BAC) to which the marker is mapped and are indicated after their approximate position in the Arabidopsis thaliana genome on the right side of the LG bar (chromosome number-position in kb). The scale on the left of each LG represents distances in cM. The bars representing the linkage groups are coloured in reference to the conserved synteny between the A. halleri and A. thaliana genomes (Schranz et al., 2006). Regions showing no conserved synteny with A. thaliana are indicated in white.

Knowledge of the conserved genome macrosynteny existing within the Arabidopsis genus (Kuittinen et al., 2004; Clauss & Koch, 2006; Schranz et al., 2006, 2007; Willems et al., 2007; Roosens et al., 2008a,b) permitted the association of ungapped regions of A. thaliana and A. lyrata genomes that showed conserved local macrosynteny with A. halleri BAC sequences containing PDF1s (Table S5, Fig. S1). When PDF1s were found within these regions, they were grouped as orthologous. Then, when two PDF1s paralogues were present at orthologous loci, the one-to-one orthologous relationships were established between the copies showing the strongest micro-synteny conservation (principle of parsimony) around and within the gene sequence, (Fig. S1). From these analyses, AhPDF1s and AlPDF1s genes were named in order to reflect the one-to-one PDF1 orthologous relationships established within the Arabidopsis genus and in compliance with the PDF1 nomenclature already defined in A. thaliana (Fig. 2).

Figure 2.

Comparative genomic organization and orthologous relationship among PDF1s in the Arabidopsis genus. Within the Arabidopsis genus, the 25 PDF1 genes were arranged within each lane according to their one-to-one orthologous relationship. The PDF1 grey shadowing indicates genes likely to be nonfunctional as follows: the AlPDF1.2a gene had to be reannotated and the resulting predicted protein had a stop codon at the 39th amino acid (Supporting Information Notes S1), the AhPDF1.6 gene was considered to a pseudogene as the identified sequence lacked the first exon and subsequent intron parts, while the AhPDF1.7 gene had a four-nucleotide insertion, resulting in the translation of a premature stop codon after the 54th amino acid (Notes S1). The species tree is shown on the top. The α and β whole-genome duplications are indicated on the branches of the tree adapted from Beilstein et al. (2010).

A consistent picture could thus be established in which the A. thaliana, A. halleri and A. lyrata PDF1s are distributed on the basis of a one-to-one orthologous relationship over eight syntenic loci (Fig. 2). As each of those synthenic regions was present in the three Arabidopsis spp., the most parsimonious scenario is that these eight regions were also present in the genome of the last common ancestor of the Arabidopsis genus. Overall, except for loci 4 and 5, PDF1s were not similarly distributed over these eight ancestral regions. This prompted us to investigate the evolutionary dynamism of the PDF1 family.

Evolutionary history of the PDF1 family in the Arabidopsis genus

This study gathered a dataset of 25 PDF1 coding sequences and a corresponding dataset of their predicted amino acid translations (Notes S1). Multiple alignment of the latter pinpointed 15 strictly conserved sites (Fig. 3a), eight of which were cysteine residues involved in the four disulphide bonds characterizing the CSαβ superfamily to which PDF1s belong (Zhu et al., 2005). The phylogeny of these 25 PDF1s and six PDF2s from A. thaliana, grouped PDF1s together, as expected (Fig. 3b). This confirmed that the collected sequences indeed encoded PDF1s and that there have been no obvious annotation errors so far. Within the strongly supported clade of PDF1s (≥ 95% bootstrap support), the evolutionary history was less clear. Indeed, only four of the 23 clades were well supported and gathered PDF1s showing orthologous relationships within loci 4, 5, 6 and 7.

Figure 3.

Phylogenetic analysis of the PDF1 family in the Arabidopsis genus. (a) Comparative sequence alignment of predicted amino acid sequences encoded by 25 PDF1 genes from Arabidopsis halleri, Arabidopsis lyrata and Arabidopsis thaliana species. Predicted protein sequences (Notes S1) were aligned with MUSCLE3.8.31 software (Edgar, 2004) and visualized with BOXSHADE 3.21 software. Names of the PDF1 amino acid sequences are shadowed in dark orange, light orange and white according to the species of origin, that is, A. halleri, A. lyrata and A. thaliana, respectively. Stars indicate that the predicted sequence of AlPDF1.2a and AhPDF1.7 genes was corrected to be aligned without taking the premature stop codon into account (as listed in Supporting Information Notes S1). The identical (threshold: 0.55) and similar residues are shadowed in blue and green colours, respectively. Amino acids identical in all PDF1s are shadowed in red and are copied below the alignment under the identity lane. The identity was refined by hand in order to take into account the first two MA (methionine, alanine) amino acids that are missing in AhPDF1.6 and the fourth cysteine residue which is conserved in all proteins except AlPDF1.2a, which correspond here to a reconstructed protein as mentioned earlier. Based on the RsAFP1 structure (Fant et al., 1998), the predicted cleavage site between the N-terminal signal peptide and the mature peptide is indicated by an arrow, while the secondary structure elements (α-helix, cylinder; β-strand, arrow) as well as the disulphide bridge connectivity are shown above and below the alignment, respectively. (b) Phylogeny of the 25 proteins of the PDF1 family using six PDF2 proteins as outgroup. The bootstrap value is indicated for each clade. Each of the four most reliable clades of the PDF1 family (whose bootstrap support is ≥ 95%) is indicated by a red rectangle.

Owing to the poor resolution of the obtained PDF1 phylogeny, it was not possible to rely on reconciliation methods (Doyon et al., 2011) to draw up an evolutionary scenario on PDF1 gains and losses between species belonging to the Arabidopsis genus. We could, however, infer the most parsimonious scenario for those events on the basis of orthologous relationships (Fig. 2). For example, since PDF1s present at locus 8 are specific to A. halleri, the most parsimonious scenario is that these PDF1s were gained in this lineage. Moreover, as representatives of PDF1.2c and PDF1.2b are specifically missing in A. lyrata, these PDF1s were probably lost in this lineage. However, PDF1.3, which is only present in A. thaliana, could have been gained in this lineage or lost in the last common ancestor of A. halleri and A. lyrata. Similarly, since PDF1.1b, PDF1.6 and PDF1.7 were specifically absent in A. thaliana, they could have been lost in this lineage or gained in the last common ancestor of A. halleri and A. lyrata. In order to distinguish between these two equally parsimonious scenarios, syntenic conservation in outgroup species was considered. Hence, we took advantage of the complete sequenced genome information available for the closest relatives, that is, Thellungiella parvula (Dassanayake et al., 2011; Cheng et al., 2012a,b) and Brassica rapa (Wang et al., 2011) to upscale our syntenic analysis (Cheng et al., 2012a,b). Within locus 1, two PDF1 orthologous were identified in T. parvula and B. rapa (Fig. S3 and data not shown), thus making the loss of AtPDF1.1b in the A. thaliana lineage the most parsimonious scenario. A clear synteny conservation was also identified at locus 4 where a single PDF1 was identified in both outgroup species (Fig. S2, and data not shown). For the remaining PDF1 ancestral regions, no clear synteny conservation was identified between genomes of the Arabidopsis genus and the T. parvula or B. rapa outgroup species (data not shown). Hence, the evolutionary scenario underlying the gain or loss of PDF1s in these regions within the Arabidopsis genus (loci 2, 3, 5, 6 and 7, see Fig. 5) remains unresolved.

In summary, this analysis led us to consider that the structure of PDF1 family orthologous is evolutionarily dynamic within the Arabidopsis genus and that PDF1s present at loci 1 and 4 have likely been conserved throughout the Brassicaceae family suggesting that PDF1s present at these loci could be founder from which other PDF1 copies could have been duplicated.

AhPDF1s are promiscuous, being antifungal and able to provide Zn tolerance

PDF1 proteins are structured with an N-terminal secretory signal peptide that is cleaved to release the mature functional part (Bendtsen et al., 2004) used to perform in vitro antifungal assays. This cleavage site was clearly located within all AhPDF1 predicted proteins except for AhPDF1.4 and AhPDF1.5. In that case, the first amino acid of the mature peptide had to be determined considering the alignment of the predicted protein sequences (Fig. 3a). Each of the 11 AhPDF1 mature recombinant proteins was assayed in vitro for their ability to inhibit the growth of Fusarium oxysporum f. sp. melonii. Seven of the AhPDF1s showed the same 2.5 μM PDF1 minimal inhibitory concentration (MIC) (Table 2). With a MIC of 5 μM, AhPDF1.4 and AhPDF1.6 presented lower antifungal activity. AhPDF1.8b and AhPDF1.7 were by far less active, with a MIC of 5–10 μM and > 10 μM, respectively (Table 2). These latter results were likely due: to the existence of a premature stop codon in the AhPDF1.7 coding sequence that removes half of the mature protein (Notes S1); and to the numerous amino acid changes occurring in the AtPDF1.8b mature part, this protein being the most divergent as compared to other PDF1s (Fig. 3a).

Table 2. Antifungal activity of AhPDF1s as tested against the fungus Fusarium oxysporum f. sp. melonii
Name of the PDF1 testedMIC (μM)a
  1. a

    Calculated minimal inhibitory concentrations (MICs) correspond to the lowest protein concentrations causing 100% fungal growth inhibition.

AhPDF1.7> 10.0

The comparative ability of AtPDF1s and AtPDF1s to induce cellular Zn tolerance was analysed upon expression in yeast. Thirteen of the 18 PDF1s similarly induced the highest Zn tolerance activity (rectangle in Fig. 4). Lower Zn tolerance ability was observed for AhPDF1.4, AhPDF1.5 and for AtPDF1.5 (rounded rectangle in Fig. 4). Meanwhile, AhPDF1.6 and AhPDF1.7 did not provide Zn tolerance (encircled in Fig. 4), which was not surprising since these genes encode a truncated version of PDF1 missing the first exon (AhPDF1.6) – the translation of which has been shown to be mandatory for providing Zn tolerance (Oomen et al., 2011) – or having a premature STOP codon (AhPDF1.7) (Notes S1). Interestingly, we noticed that AtPDF1.4 had a slightly higher Zn tolerance than AhPDF1.4 although these two proteins differed at only two amino-acid positions. The Ala28 and Ser54 of AhPDF1.4 are changed to Gly and Arg in AtPDF1.4, respectively. The Ala28Gly and Ser54Arg double substitution was necessary to increase the degree of Zn tolerance provided by AhPDF1.4 to that of AtPDF1.4 (Fig. S3). So this study did not permit us to identify a specific motif or domain associated with Zn tolerance.

Figure 4.

In vitro functional characterization of the ability of Arabidopsis halleri and Arabidopsis thaliana PDF1s to confer cellular zinc (Zn) tolerance. Serial dilutions of the Saccharomyces cerevisiae BY4741 strain expressing the pYX212 empty vector (EV) or a pYX212 harboring one of the A. halleri or A. thaliana PDF1s were spotted on medium supplemented with 1.4 μM (control), 25 mM or 27.5 mM ZnSO4, as indicated above the panels. Each spot was made with 10 μl of a yeast culture diluted at the OD600nm mentioned below the drops. Pictures were taken on day 2 for control and day 11 for Zn treatments; they are representative of the three experiments, which have been performed with three independent yeast transformants. Rectangles, ovals and rectangles with rounded corners indicate the names of the PDF1 proteins presenting the standard, decreased and no-Zn-tolerance capacities, respectively.

In summary, the overall AhPDF1 family showed antifungal properties, and PDF1 family members coming from A. thaliana or A. halleri showed similar in vitro cellular Zn-tolerance properties. In most cases, PDF1 antifungal activity and Zn tolerance were associated, thus giving promiscuous characteristics to this protein. Note, however, that AhPDF1.5 had standard antifungal activity but was not as efficient as other PDF1s for Zn tolerance. Conversely, AhPDF1.8b had lower antifungal activity but standard Zn tolerance.

PDF1 genes are constitutively more highly expressed in A. halleri than in A. thaliana

Expression of PDF1s was measured at the transcript level in roots and shoots of A. halleri and A. thaliana plants grown in axenic hydroponic conditions. Within each species, the transcript abundances varied by two to three orders of magnitude between different PDF1 paralogues (Fig. 5, Table S9). For a given PDF1, transcript abundances were, on average, higher in shoots than in roots. This was particularly obvious in A. thaliana, in which no PDF1 transcripts could be detected in roots. This was also observed in A. halleri where PDF1 transcripts were, on average, 300 times more abundant in shoots than in roots (Table S9). In A. halleri, the greatest organ differences were observed for PDF1 transcripts located at locus 1 (AhPDF1.1b and AhPDF1.1a) and locus 3 (AhPDF1.2b). Transcripts of these PDF1s were the most highly accumulated in shoots. Note that this observation is in agreement with previous functional findings (Mirouze et al., 2006), which led to the cloning of the three corresponding cDNAs from the screening of a leaf cDNA library (see Table 1). Some AhPDF1s, however, displayed a different pattern: AhPDF1.4, AhPDF1.5 and AhPDF1.7 transcripts were accumulated at similar abundances in roots and shoots, whereas AhPDF1.8a transcripts were c. 30 times more highly accumulated in roots than in shoots. In response to increasing Zn concentrations in the culture medium, the relative abundances of AhPDF1 transcripts remained unchanged, apart from a significant increase for AhPDF1.1a (×2.5) and AhPDF1.2b (×4), which was observed in shoots and roots. Overall, and most importantly, transcript analysis revealed that PDF1 transcripts were c. 1000 times more highly accumulated in A. halleri than in A. thaliana (Fig. 5 and Table S9). Searches performed in the 2-kb-long regions upstream of the AhPDF1 coding sequences for putative metal cis-responsive elements or other particular motives, which could potentially be correlated with the level of expression, were unsuccessful (data not shown).

Figure 5.

Accumulation of PDF1 transcripts in Arabidopsis halleri and Arabidopsis thaliana plants. Shoots and roots were collected from individual plants of the A. halleri SAF2 genotype derived from the Auby accession that had been exposed to 10 (control), 100 or 1000 μM ZnSO4 for 4 d. Quantitative PCR was performed using the gene copy specific primer pairs listed in Supporting Information Table S8. A. halleri and A. thaliana PDF1 transcript abundances are expressed relative to Actin. Five biological replicates were analysed using three technical Q-PCR repeats. Error bars correspond to confidence intervals at the 0.05 threshold. Stars indicate PDF1s for which a significant difference in transcript accumulation was observed upon zinc (Zn) addition in the media. Regarding the A. halleri species, transcripts of AhPDF1.2c, AhPDF1.6 and AhPDF1.8b genes could not be detected in either shoots or roots. For the A. thaliana species, transcripts of the AtPDF1.1, AtPDF1.2c and AtPDF1.3 genes could not be detected in shoots, and none of the AtPDF1 transcripts could be detected in roots.


Besides their involvement in plant responses to pathogen attack, the roles played by defensins in response to multiple abiotic stresses are diverse (Carvalho & Gomes, 2011). These include the role played by PDF1s in providing cellular Zn tolerance, as revealed through a study of Zn-tolerant and Zn-hyperaccumulating A. halleri species (Mirouze et al., 2006). Protein promiscuity is the phenomenon in which multiple functions may be associated with a single peptide or protein structure (Nobeli et al., 2009). It was initially proposed for defensins as an evolutionary concept by Franco (2011), based on the idea that if protein and peptides possess a structure directly related to a single function it would hamper their ability to adapt and develop new functions. This study describes the PDF1 family as represented by 25 members encoding fairly similar proteins and forming a quite homogeneous multigenic family within the Arabidopsis genus (Fig. 3a). So far, PDF1 functional studies focused mainly on A. thaliana where three PDF1s (out of seven) were studied and showed in vitro activity against different fungi (Terras et al., 1993; Penninckx et al., 1996; Sels et al., 2007, 2008). Having in mind that PDF1 promiscuity could contribute to A. halleri's evolutionary walk towards Zn tolerance, the present study focused on the roles played by PDF1s in antifungal activity and cellular Zn tolerance. This is one of the few studies in which multiple functional assays are conducted on the same defensin molecule. Indeed, in most cases, the diversity of defensin functions is inferred indirectly from sequence similarity to a protein mask typical of the large multigenic defensin family. The findings of the functional assays presented here showed that none of the 11 identified A. halleri PDF1s provided a specific increase in Zn tolerance as compared with the seven PDF1s originating from A. thaliana nontolerant species. In addition, the AhPDF1 antifungal activity and Zn-tolerance properties were globally correlated across the members of the AhPDF1 family. No obvious amino acid stretch has been found to be specifically associated with one of the activities studied. So, from our results, we cannot correlate a function to a structure. Regarding the in planta data, we know that the overexpression of AhPDF1.1b provides A. thaliana with Zn tolerance (Mirouze et al., 2006); and that the overexpression of AtPDF1.1 leads to a reduction in pathogen symptoms (De Coninck et al., 2010). Confronting the in vitro assay and in planta phenotypic effects suggests that a single PDF1 molecule is likely responsible for both antifungal activity and cellular Zn tolerance. These two functions could be assumed, for example, in different physiological or protein concentration conditions as proposed in the case of pure protein promiscuity (Franco, 2011). It should be noted that no information is currently available to enable us to determine whether the two PDF1 properties that are considered here (i.e. antifungal activity as a response to biotic stress and Zn-tolerance capacity as related to abiotic stress) originate from the same mode of action. However, our current research is now focused on the search for common properties that could explain their dual role. From sequences analyses, very few amino acids appear to be common to all the PDF1s. Among these are the eight cysteines supposedly involved in four disulphide bonds. One of our working hypotheses is that the common properties of defensins would rely on these cysteines. As far as Zn tolerance is concerned, we hypothesize that PDF1s could chelate Zn in some special conditions in which the protein is not fully oxidized, as discussed previously (Marquès & Oomen, 2011). The results of the exhaustive functional analysis presented here will be useful for guiding future research on the PDF1 mode of action so as to elucidate how these two functional properties closely dovetail for the benefit of plant responses to both biotic and abiotic stresses.

As there was no obvious functional discrimination in PDF1s originating from A. halleri as compared with A. thaliana, evolutionary processes undergone by PDF1s were investigated at the genomic level. Interspecies transcriptomic experiments performed in comparison to A. thaliana revealed a global high constitutive accumulation of PDF1 transcripts in A. halleri (Talke et al., 2006). The analyses reported here provide a particularly striking insight into each of the 11 AhPDF1s. The between-species difference in PDF1 shoot transcript accumulation was particularly high for specific AhPDF1s (AhPDF1.1a-AhPDF1.1b and AhPDF1.2b) as compared with their orthologues in A. thaliana (AtPDF1.1 and AtPDF1.2b-AtPDF1.3; Figs 2, 5, Table S9). Within A. halleri, these copies also showed the greatest contrast in transcript accumulation between shoots and roots. As an indication, when measured on the same A. halleri line, cultivated in the same conditions and when quantified with respect to the same reference gene, the shoot relative transcript abundance of the highest accumulated PDF1s (AhPDF1.1b in Fig. 5) was in the same range as the shoot relative transcript abundance of another cellular Zn-tolerance gene, that is, the vacuolar Zn transporter Metal Tolerance Protein 1 (AhMTP1-As and AhMTP1-B in fig. 7 of Shahzad et al., 2010).

Higher constitutive transcript accumulation of metal homeostasis-related genes is a hallmark of A. halleri evolutionary innovation, as compared with phylogenetically related species (i.e. A. thaliana and A. lyrata), which are Zn-nontolerant and Zn-nonaccumulators (Talke et al., 2006; Hanikenne et al., 2008; Shahzad et al., 2010; Deinlein et al., 2012). This increase is considered as a way of enhancing a function already present in the Arabidopsis ancestor (Hanikenne & Nouet, 2011). It occurs between orthologous genes present in nontolerant species (Hanikenne et al., 2008; Shahzad et al., 2010), but it could also be linked to gene amplification in A. halleri spp., although this is not always the case (Talke et al., 2006). For instance, increased transcript accumulation seems to be the main evolutionary change for genes that remain as a single copy throughout the Arabidopsis genus, for example, Iron-Regulated Transporter 3 (IRT3), Ferric Reductase Deficient 3 (FRD3) and Zinc Iron Permease 10 (ZIP10) (Talke et al., 2006). In other cases, such as Heavy Metal ATPase 4 and MTP1, high constitutive transcript accumulation is associated with gene duplication in the A. halleri lineage, as compared with a single orthologous gene represented in A. thaliana (Hanikenne et al., 2008; Shahzad et al., 2010). The orthologous relationships documented here between PDF1s of multigenic families occurring in the three species of the Arabidopsis genus highlight a new configuration in comparison to those mentioned earlier. Indeed, the high PDF1 transcript deregulation is an evolutionary innovation, which most probably might have occurred in the A. halleri lineage independently on each PDF1 locus and on each harboured gene, thus contributing to the A. halleri adaptive walk towards Zn tolerance (Orr, 2005) independently of PDF1 gene amplification in this lineage. Indeed, although there are 11 PDF1s in A. Halleri, compared with seven in A. thaliana or A. lyrata, only AhPDF1.8a and AhPDF1.8b are specific to the A. halleri lineage (Fig. 2). Hereafter, it is therefore unlikely that the 11 AhPDF1s are the result of specific gene expansion in A. halleri. Rather, the PDF1 family should be considered as evolutionarily dynamic. The orthologous structure of the family might be a footprint of genome rearrangement in A. lyrata or might reflect genome reduction, which is known to have occurred in A. thaliana (Johnston et al., 2005; Proost et al., 2011). The latter is more likely, as A. halleri and A. lyrata are the two species that show the most similar organization of PDF1s at the genome level, as expected for species that diverged only c. 300 000 yr ago (Roux et al., 2011). It would thus be interesting to investigate whether PDF1s harboured by loci 2, 3 and 8 were actually specifically retained in A. halleri to a greater extent than in the A. lyrata lineage. Interestingly, overall, PDF1 transcripts are also constitutively highly accumulated in Noccaea caerulescens (formerly Thlaspi caerulescens) as compared to the A. thaliana model plant (Hammond et al., 2006; van de Mortel et al., 2006). When available, it would be very interesting to analyse the N. caerulescens genome sequence in order to determine whether PDF1s have a similar genomic organization in this other metal-tolerant and hyperaccumulating species and whether different PDF1s have independently also been the focus of evolutionary transcriptional innovation.

Gene duplication introduces genome plasticity, so that, upon evolution, paralogous sequences tend to diverge over time to perform different functions via non-, sub- or neofunctionalization routes, ultimately resulting in gain or loss of gene copies upon the selection process (Ohno, 1970; Zhang, 2003; Lynch & Katju, 2004). As the strongest criterion differentiating PDF1 paralogues is the wide variation in the range of constitutive expression levels, one might expect that this could be a criterion for discriminating the genes playing bigger roles than others. In A. halleri, locus 8 specifically harbours PDF1s. However, these paralogues do not contribute mainly to the constitutive high transcript abundance, which is reinforced for AhPDF1.8b as no transcripts were detected; it encodes proteins with lower antifungal activity, and most importantly the gene copy is not fixed in the Auby population. It seems also that Zn tolerance in A. halleri did not select for the functional expression of PDF1 at loci 2, 6 and 7 based on the low transcript accumulation and/or pseudogene structure (AhPDF1.2a-AhPDF1.2c and AhPDF1.6) and the presence of a premature stop codon (AhPDF1.7). Overall, out of the 11 AhPDF1s identified, only three genes – AhPDF1.1a, AhPDF1.1b and AhPDF1.2a – located in loci 1 and 3, clearly retain fully the functional and genomic characteristics to be completely operational in A. halleri. They thus represent good candidates to decipher the molecular mechanisms enabling PDF1s to provide Zn tolerance using overexpression and/or RNAi approaches.

The response of A. thaliana PDF1s to Zn treatment has been studied (Mirouze et al., 2006) after infiltration into the leaves of plants grown in nonsterile conditions, but these findings cannot really be compared with the present results obtained after Zn application in the culture medium of plants grown under axenic conditions. In A. thaliana, the response to pathogens initially targeted AtPDF1.2a, inducible in leaves following an ethylene/jasmonate pathway (Penninckx et al., 1996, 1998; Manners et al., 1998; De Coninck et al., 2010; Niu et al., 2011). Only recently, expression studies revealed similar up-regulation in the context of nonhost resistance for AtPDF1.2a-AtPDF1.2c and AtPDF1.2b-AtPDF1.3 (Hiruma & Takano, 2011). Hereafter, it would be interesting to investigate the extent to which the regulation of different PDF1 transcripts is shared over, or specific to, different loci. For example, in A. lyrata locus 2 (Fig. 2), the AlPDF1.2c copy was probably lost, with the remaining AlPDF1.2a being mutated by the presence of a premature stop codon, thus likely nonfunctional, and no PDF1 was present at locus 3. If we refer to the earlier-cited description of AtPDF1 harboured in loci 2 and 3, then A. lyrata plants would be quite deprived of the PDF1 responsiveness to pathogens unless other AlPDF1 members were to take over. In the same vein, it would be interesting to know if AhPDF1.2b (one of the greatest constitutively accumulated AhPDF1s) and AhPDF1.2a and AhPDF1.2c (some of the least constitutively accumulated AhPDF1s) are responsive to jasmonic acid and/or pathogen attack. Additional studies are thus necessary to compare the evolutionary conserved genome structure with the functional contribution of each PDF1, and it would be mostly interesting to include other species such as N. caerulescens in such an analysis.

Understanding how gene function and gene expression contribute to the acquisition of adaptive traits is necessary to gain further insight into the molecular evolution (Jacob, 1977; Orr, 2005; Mitchell-Olds et al., 2007; Conant & Wolfe, 2008). Cooption occurs when natural selection finds new uses for existing traits (True & Carroll, 2002). Genes can be coopted to generate developmental and physiological novelties by changing their regulation patterns or the functions they encode, or both (Babbitt et al., 2007; Hittinger & Carroll, 2007). The research presented here revealed that A. halleri PDF1s are antifungal proteins also displaying Zn-tolerance properties and that high constitutive transcript accumulation in A. halleri targets different PDF1s independently of gene duplication. We propose that the high increase in AhPDF1 transcript accumulation is an evolutionary innovation coopting promiscuous PDF1s for their contribution to Zn tolerance in the A. halleri sp. Moreover, the evolutionary dynamic orthologous structure of the PDF1 family is not in favour of PDF1 amplification in A. halleri, but it paves the way for possible PDF1 retention in this lineage. Finally, given that PDF1s are at the crossroads of the plant response to biotic and abiotic stresses, our results provide a strong basis for further genetic diversity and evolutionary studies questioning, in particular, the potential emergence of Zn tolerance as a defence mechanism against biotic attack from pathogens or herbivores (Poschenrieder et al., 2006; Boyds & Martens, 2007; Rascio & Navari-Izzo, 2011).


Z.S. was supported by a scholarship from the Higher Education Commission, Islamabad-Pakistan. This work was partially supported by a French Agence Nationale de la Recherche grant ANR-10-BINF-01-02 ‘Ancestrome’. We are thankful to all BPMP technical and administrative staff for their kind and generous support. We are greatly indebted to Jérôme Salse for pertinent advice concerning synteny conservation searches. We are grateful to Hatem Rouached for fruitful comments on the manuscript. We gratefully acknowledge each of the anonymous referees for their helpful comments and constructive suggestions, which greatly improved the presentation of this article.