Origin of strigolactones in the green lineage


  • Pierre-Marc Delaux,

    1. Laboratoire de Recherche en Sciences Végétales, Université de Toulouse, UPS, UMR 5546, BP 42617, F-31326, Castanet-Tolosan, France
    2. CNRS, UMR 5546, BP 42617, F-31326, Castanet-Tolosan, France
    Search for more papers by this author
  • Xiaonan Xie,

    1. Weed Science Centre, Utsunomiya University, Utsunomiya 321-8505, Japan
    Search for more papers by this author
  • Ruth E. Timme,

    1. Cell Biology and Molecular Genetics, 2108 Biosciences Research Bldg., and the Maryland Agricultural Experiment Station, University of Maryland, College Park, MD 20742, USA
    Search for more papers by this author
  • Virginie Puech-Pages,

    1. Laboratoire de Recherche en Sciences Végétales, Université de Toulouse, UPS, UMR 5546, BP 42617, F-31326, Castanet-Tolosan, France
    2. CNRS, UMR 5546, BP 42617, F-31326, Castanet-Tolosan, France
    Search for more papers by this author
  • Christophe Dunand,

    1. Laboratoire de Recherche en Sciences Végétales, Université de Toulouse, UPS, UMR 5546, BP 42617, F-31326, Castanet-Tolosan, France
    2. CNRS, UMR 5546, BP 42617, F-31326, Castanet-Tolosan, France
    Search for more papers by this author
  • Emilie Lecompte,

    1. Université de Toulouse, UPS, EDB (Laboratoire Evolution et Diversité Biologique), 118 route de Narbonne, F-31062, Toulouse, France
    2. CNRS, EDB (Laboratoire Evolution et Diversité Biologique), F-31062, Toulouse, France
    Search for more papers by this author
  • Charles F. Delwiche,

    1. Cell Biology and Molecular Genetics, 2108 Biosciences Research Bldg., and the Maryland Agricultural Experiment Station, University of Maryland, College Park, MD 20742, USA
    Search for more papers by this author
  • Koichi Yoneyama,

    1. Weed Science Centre, Utsunomiya University, Utsunomiya 321-8505, Japan
    Search for more papers by this author
  • Guillaume Bécard,

    1. Laboratoire de Recherche en Sciences Végétales, Université de Toulouse, UPS, UMR 5546, BP 42617, F-31326, Castanet-Tolosan, France
    2. CNRS, UMR 5546, BP 42617, F-31326, Castanet-Tolosan, France
    Search for more papers by this author
  • Nathalie Séjalon-Delmas

    1. Laboratoire de Recherche en Sciences Végétales, Université de Toulouse, UPS, UMR 5546, BP 42617, F-31326, Castanet-Tolosan, France
    2. CNRS, UMR 5546, BP 42617, F-31326, Castanet-Tolosan, France
    3. Present address: UMR5245 ECOLAB, ENSAT, Av de l’Agrobiopole, F-31326 Auzeville, Castanet-Tolosan, France
    Search for more papers by this author

Authors for correspondence:
Guillaume Bécard
Tel: +33 5 34 32 38 20
Email: becard@lrsv.ups-tlse.fr

Nathalie Séjalon-Delmas
Tel: +33 5 34 32 39 41
Email: nathalie.sejalondelmas@ensat.fr


  • The aims of this study were to investigate the appearance of strigolactones in the green lineage and to determine the primitive function of these molecules.
  • We measured the strigolactone content of several isolated liverworts, mosses, charophyte and chlorophyte green algae using a sensitive biological assay and LC-MS/MS analyses. In parallel, sequence comparison of strigolactone-related genes and phylogenetic analyses were performed using available genomic data and newly sequenced expressed sequence tags. The primitive function of strigolactones was determined by exogenous application of the synthetic strigolactone analog, GR24, and by mutant phenotyping.
  • Liverworts, the most basal Embryophytes and Charales, one of the closest green algal relatives to Embryophytes, produce strigolactones, whereas several other species of green algae do not. We showed that GR24 stimulates rhizoid elongation of Charales, liverworts and mosses, and rescues the phenotype of the strigolactone-deficient Ppccd8 mutant of Physcomitrella patens.
  • These findings demonstrate that the first function of strigolactones was not to promote arbuscular mycorrhizal symbiosis. Rather, they suggest that the strigolactones appeared earlier in the streptophyte lineage to control rhizoid elongation. They may have been conserved in basal Embryophytes for this role and then recruited for the stimulation of colonization by glomeromycotan fungi.


Strigolactones (SLs) are a family of carotenoid-derived plant secondary metabolites produced by dicots and monocots (Xie et al., 2010) and at least one moss (Proust et al., 2011). Eighteen members in this family have been identified so far, but many more are expected to occur in land plants (Xie et al., 2010). SLs were first characterized as seed germination stimulants of the parasitic plants Orobanche and Striga (Cook et al., 1966, 1972). More recently, they have been implicated as important plant signals for the establishment of arbuscular mycorrhizal (AM) symbiosis (Akiyama et al., 2005; Besserer et al., 2006; Gomez-Roldan et al., 2008). Finally, they have been identified as a new class of plant hormone involved in the inhibition of shoot branching (Gomez-Roldan et al., 2008; Umehara et al., 2008). Recent studies have also suggested that SLs control mesocotyl elongation (Hu et al., 2010), root development by inhibiting lateral root formation and stimulating root hair elongation (Kapulnik et al., 2011; Ruyter-Spira et al., 2011) and the protonema expansion of the moss Physcomitrella patens (Proust et al., 2011).

Genetic studies have revealed that the initial steps of SL biosynthesis probably occur in plastids of root cells through the isomerization of all-trans-β-carotene to 9-cis-β-carotene by an iron-containing protein, D27 (Lin et al., 2009; Alder et al., 2012). Moreover, it has been proposed that two CCD (CAROTENOID CLEAVAGEDIOXYGENASE) enzymes, CCD7 and CCD8, cleave the 9-cis-β-carotene to produce the SL precursor carlactone (Matusova et al., 2005; Gomez-Roldan et al., 2008; Umehara et al., 2008; Alder et al., 2012). Several orthologs of the genes encoding these two proteins have been characterized: MAX3 and MAX4 in Arabidopsis thaliana (Sorefan et al., 2003; Booker et al., 2004), RMS5 and RMS1 in pea (Morris et al., 2001; Sorefan et al., 2003), D17 and D10 in rice (Arite et al., 2007), DAD3 and DAD1 in petunia (Snowden et al., 2005; Drummond et al., 2009), and, recently, SlCCD7 in tomato (Vogel et al., 2010; Koltai et al., 2010) and PpCCD8 in P. patens (Proust et al., 2011). The synthesis of SLs in A. thaliana also involves MAX1, a cytochrome P450 (Stirnberg et al., 2002; Booker et al., 2005). In addition, an ABC-transporter, PDR1, has been shown recently to be involved in SL transport (Kretzschmar et al., 2012). Other proteins downstream of the synthesis pathway are expected to be involved in SL perception, including the F-BOX protein MAX2/RMS4/D3 in A. thaliana, pea and rice respectively (Stirnberg et al., 2002; Ishikawa et al., 2005; Johnson et al., 2006), and the α/β-fold hydrolases D14 in A. thaliana and rice and D14-like in A. thaliana (Arite et al., 2009; Waters et al., 2012). Indeed, the mutant phenotypes of the highly branched corresponding mutants could not be rescued by SL application. To date, SLs have been identified only in some Angiosperm species and in the moss P. patens (Proust et al., 2011), with none reported in other taxa of the green lineage (Viridiplantae).

The Viridiplantae is split into two evolutionary lineages, the Chlorophytes (green algae sensu stricto) and the Streptophytes (charophyte algae and Embryophytes). Molecular dating methods place this divergence between 725 and 1200 million years ago (Mya; Floyd & Bowman, 2007). Chlorophytes comprise Prasinophyceae (Ostreococcus, Micromonas), Ulvophyceae (Ulva), Trebouxiophyceae (Chlorella) and Chlorophyceae, including the unicellular model alga Chlamydomonas reinhardtii and the colonial Volvox carteri (Fig. 1). Among the Streptophytes, the unicellular Mesostigmatales and Chlorokybales represent the earliest diverging Charophytes (Lemieux et al., 2007; Finet et al., 2010; Wodniok et al., 2011). Filamentous and unbranched algae are found in Klebsormidiales and Zygnematales, and, finally, Charales and Coleochaetales are characterized by multicellular organization and form with Zygnematales, the advanced charophyte algae (Sørensen et al., 2011; Fig. 1). Molecular phylogenetic analyses and biochemical characterization (glycolate oxidase, superoxide dismutase, presence of sporopollenin), as well as ultrastructural features (phragmoplastic cell division), support the hypothesis that Charophytes have a more recent common ancestor with Embryophytes (Karol et al., 2001; Lemieux et al., 2007; Finet et al., 2010; Wodniok et al., 2011).

Figure 1.

Proposed scheme for the evolution of strigolactone (SL)-related genes in the green lineage. The presence and absence of SLs and of proteins known to be involved in SL synthesis and signaling are indicated in green and red boxes, respectively. Orange boxes indicate putative homologous proteins. White boxes are not determined. Dotted boxes indicate data obtained from expressed sequence tags (ESTs) only. (1) and (2) indicate the two hypotheses for the appearance of SLs in the green lineage. (1) and the red dotted lines show the proposed appearance of SLs in the putative common ancestor of advanced Charophyte and Embryophytes. In this hypothesis, SLs are later lost in Coleochaetales and Zygnematales (red X). (2) and black dotted lines show the proposed appearance of SLs in Charales, with Charales as a sister clade to Embryophytes.

Embryophytes are thought to have emerged from a freshwater aquatic ancestor during the mid-Ordovician and early Silurian periods (480–430 Mya) (Kenrick & Crane, 1997). Phylogenetic analysis implicates liverworts as strong candidates for the first extant descendants of these early Embryophytes (Qiu et al., 1998), although this remains a subject of active study (e.g. Finet et al., 2010). Based on the study of fossils from the Ordovician period, it has been proposed that AM symbiosis appeared concomitantly and could have been crucial for land colonization by plants (Redecker et al., 2000; Humphreys et al., 2010). This association between AM fungi of the genus Glomus and liverworts still persists today (Russell & Bulman, 2005; Fonseca et al., 2006; Wang & Qiu, 2006; Humphreys et al., 2010).

Because of the multiple functions of SLs as both plant hormones and symbiotic signals, the question of the primitive function of SLs during plant evolution remains. Were the first Embryophytes already producing SLs with the primary function of promoting AM symbiosis (Bouwmeester et al., 2007)? Or were SLs involved in some as yet unknown hormonal regulation of ancestral developmental processes? To answer these questions, we analyzed a panel of basal plants belonging to chlorophyte green algae, charophyte green algae and liverworts for their ability to produce SLs using a very sensitive bioassay in combination with mass spectrometry. We report here that SLs are produced and exuded by liverworts of the genus Marchantia, the most basal Embryophytes, but also by charophyte green algae of the order Charales. This first report of the presence of SLs in non-Embryophyte plants suggests that the appearance of SLs predates the colonization of land by plants and the first AM symbiosis. In addition, exogenously applied SLs stimulate rhizoid elongation in mosses, liverworts and Charales. Moreover, based on the short-rhizoids phenotype of the Ppccd8 mutant of P. patens, we conclude that endogenous SLs in basal Streptophytes play a similar role. Together, these results provide evidence that the first selective pressure leading to SLs being widespread in the green lineage was probably more hormonal than symbiotic.

Materials and Methods

Fungal material

Spores of Gigaspora rosea (DAOM 194757) were produced in pot cultures on leeks and collected by wet sieving. They were washed in 0.05% Tween 20, soaked in 2% Chloramine T (Sigma) for 10 min, washed again three times in sterile water and stored in an antibiotic solution containing 100 mg l−1 gentamycin and 200 mg l−1 streptomycin. After 2 d at 4°C, a second treatment with Chloramine T was carried out under the same conditions. Spores were then stored in the antibiotic solution at 4°C before use.

Gigaspora rosea germ tube branching bioassay

Germ tube branching bioassays were carried out according to Buée et al. (2000). Four spores of G. rosea were germinated and incubated under 2% CO2 at 30°C in the dark in M medium (Bécard & Fortin, 1988) supplemented with 10 μM quercetin (Sigma) and gelled with 0.6% Phytagel (Sigma). Seven days after inoculation, each spore produced a single germ tube growing upwards. Two small wells on each side of the germ tube tip were made in the gel with a Pasteur pipette tip and 5 μl of the test solution (10−7 M GR24) in 10% acetonitrile (positive control), crude extract or purified fraction (resuspended in 10% acetonitrile) or 10% acetonitrile (negative control) were injected into each well. After 24 h, germ tube branching was recorded by counting newly formed hyphal tips. Five to eight plates (20–32 spores) were used for each treatment. The mean numbers of germ tube branches for each of the fractions tested were compared by the Kruskal–Wallis test and, when significant, pair comparison was made by the nonparametric Mann–Whitney test (< 0.05). Statistical analyses were performed with R Software. Each experiment was repeated three times.

Plant material and culture

The chlorophyte green alga C. reinhardtii (CEA, Cadarache, France) was grown on Tris-minimal medium (Gorman & Levine, 1965). The Zygnematales Spirogyra sp. and Coleochaete scutata were provided by the Sammlung von Algenkulturen Göttingen collection (SAG, http://www.epsag.uni-goettingen.de/cgi-bin/epsag/website/cgi/show_page.cgi?kuerzel=about) and grown in 200 ml of modified BBM medium (threefold more NaNO3, 0.12 mg l−1 of thiamine hydrochloride and 0.1 mg l−1 of cyanocobalamine; Bischoff & Bold, 1963) under low shaking (120 rpm, Gyrotary shaker, New Brunswick Scientific, Enfield, CT, USA). The Charales Nitella hyalina, Nitella pseudoflabellata and Chara corallina were kindly provided by Professor Ilse Foissner (Salzburg University, Austria) and grown in a 10-l tank containing 1/3 sterilized soil and 2/3 peat covered by 2–3 cm of clean sand and filled with distilled water. Before exudation, apical cells were cut, washed and maintained in sterile water for 1 wk. The liverworts Marchantia spp. and Lunularia cruciata were collected in the Pyrenees Mountains (Ariège, France). Gemmae were sterilized (Fonseca et al., 2006) and grown on KNOP medium (Reski & Abel, 1985). Marchantia polymorpha Takaradaike-1 (male) was kindly provided by Professor T. Kohchi (Kyoto University, Japan). The SL-deficient P. patens Ppccd8 mutant was kindly provided by Dr C. Rameaux (INRA, Versailles, France). Both the wild-type (WT, Gransden 2004) and the Ppccd8 mutant P. patens were grown on BCD medium (Grimsley et al., 1977). Light and temperature conditions were similar for all organisms (22°C, 16 h photoperiod, 18 μE m−2 s−1).

Collection of protein sequences

Protein sequences of D27, CCD7, CCD8, D14, D14-like and MAX2 of A. thaliana, Oryza sativa, Populus trichocarpa, Selaginella moellendorffii (Banks et al., 2011), P. patens (Rensing et al., 2008), C. reinhardtii (Merchant et al., 2007), V. carteri (Prochnik et al., 2010), Chlorella variabilis Shihira & Krauss NC64A (Blanc et al., 2010), Ostreococcus tauri Courties & Chrétiennot-Dinet (Derelle et al., 2006), Ostreococcus lucimarinus (Palenik et al., 2007) and Micromonas pusilla (Worden et al., 2009) were collected from the National Center for Biotechnology Information (NCBI) (http://www.ncbi.nlm.nih.gov/). For CCD trees, sequences of the draft genomes of Medicago truncatula, Glycine max, Ricinus communis, Citrus sinensis, Citrus clementina, Carica papaya, Eucalyptus grandii, Manihot esculenta, Aquilegia coerulea, Linum usitatissimum, Mimulus guttatus, Arabidopsis lyrata, Brassica rapa, Thelungiella halophila, Capsella rubella, Prunus persicus, Sorghum bicolor, Zea mays, Brachypodium distachyon and Setaria italica were collected on http://www.phytozome.net. Charophycean sequences of Nitella hyalina, Nitella mirabilis, Coleochaete orbicularis, Chaetospheridium globosum, Spirogyra pratensis, Penium margaritaceum, Chlorokybus atmophyticus, Klebsormidium flaccidum and Mesostigma viride are from the transcriptome assemblies of Timme et al. (2012). These expressed sequence tags (ESTs) were obtained as described in Timme & Delwiche (2010). ESTs of Gymnosperms, Monilophytes, liverworts and charophyte green algae available on the NCBI website were also screened by tBLASTn. The organisms and sequences used are listed in Supporting Information Table S1.

Sequences of A. thaliana, AtMAX3, AtMAX4, AtMAX2, and of O. sativa, OsD14, OsD14-like and OsD27, were used for BLASTp analyses. Sequences with an E-value < 10−10 were selected for phylogenetic analysis.

Phylogenetic tree construction

Matching sequences from all tested organisms were aligned with MAFFT (http://www.ebi.ac.uk/Tools/mafft/index.html). Before alignment, Actinidia chinensis CCD8 (ADP37984), Pisum sativum RMS1 and Petunia hybrida DAD1 were added to the CCD8 dataset, and Pisum sativum RMS5 and Petunia hybrida DAD3 were added to the CCD7 dataset. Alignment was manually corrected using BioEdit (http://www.mbio.ncsu.edu/BioEdit/). Maximum-likelihood (ML) trees were found with MEGA5 (Tamura et al., 2011), using Jones–Taylor–Thornton (JTT) as the amino acid substitution model and the nearest-neighbor-interchange (NNI) heuristic method. The partial deletion (95%) mode was used to treat gaps and missing data. For each tree, 500 bootstrap replications were performed. ZmVP14 was used as an outgroup for the CCD7 and CCD8 trees. AtTIR1 and AtF-BOX4 were used as outgroups for the F-BOX tree, and bacterial RsbQ for the D14 tree, as proposed by Waters et al. (2012).

All newly generated sequences of CCD7, CCD8, D27, D14-like and MAX2 orthologs from charophyte green algae were deposited in GenBank. For accession numbers, see Supporting Information Notes S1.

Preparation of extracts

The presence of SLs in basal Embryophytes and algae was sought in exudates and tissues. Exudates were usually obtained by soaking each organism in distilled water for 24 h. Exudates of P. patens and C. reinhardtii were also collected in growth media. Exudates in water or growth media were extracted with an equal volume of ethyl acetate. Ethyl acetate was washed with 0.2 M K2HPO4, dried over anhydrous MgSO4 and concentrated in vacuo. For tissue extractions, 10–80 g of fresh algae or liverworts were ground and extracted directly in acetone. Acetone extraction was repeated three times, the extracts pooled, dried in vacuo and dissolved in ethyl acetate. The ethyl acetate extracts were then treated as described above. Each extract was stored at − 20°C until use.

The extracts of exudates were dissolved in 20% acetonitrile in water and loaded onto a solid phase extraction (SPE) C18 cartridge (Varian Bond Elut, 500 mg, 3 ml, Agilent Technologies, Loveland, CO, USA). The SPE C18 cartridges were eluted with 3 ml of 20% (F1), 30% (F2), 40% (F3), 50% (F4), 60% (F5) and 100% (F6) acetonitrile in water. Tissue extracts of algae or liverworts were dissolved in ethyl acetate : hexane (10 : 90) and loaded onto SPE Si cartridges (Thermofisher, 1 g, 6 ml, Waltham, MA, USA). SPE Si cartridges were eluted in ten fractions with increasing concentrations (from 10 to 100%) of ethyl acetate in hexane. These fractions were dried under nitrogen, dissolved in acetonitrile and filtered through a SPE C18 cartridge (Varian Bond Elut, 1g, 6 ml, Agilent Technologies, Loveland, CO, USA). Fractions of both exudates and tissue extracts were dried under nitrogen and stored at − 20°C until use.


For analyses performed on algae and Marchantia spp., purified fractions or crude extracts of exudates were dissolved in 50% acetonitrile in water. SL detection was performed using a 4000 Q Trap mass spectrometer with a Turbo V ESI source in the positive mode, coupled to an Agilent 1100 series high-performance liquid chromatography (HPLC, Agilent Technologies, Loveland, CO, USA) system, as described in Gomez-Roldan et al. (2008), except for the following modifications: HPLC separation was performed using a C18 column (5 μm, 2.1 mm × 250 mm, ACCLAIM 120C18, Dionex, Thermo Fisher Scientific, Waltham, MA, USA). Solutions of formic acid : water [1 : 103 (v/v); A] and formic acid : acetonitrile [1 : 103 (v/v); B] were pumped at 0.2 ml min−1. The gradient was 50% B for 5 min, 50–70% B for 5 min, 70% B for 10 min, 70–100% B for 10 min and 100% B for 5 min. The column was equilibrated at 50% B for 5 min before the next run. For each sample, Multiple Reaction Monitoring (MRM) transitions for known SLs were systematically searched and retention times systematically compared with the corresponding synthetic standard, if available. Co-injection analyses were performed by adding 5 pg of standard to the analyzed sample.

Analyses of M. polymorpha purified tissue extracts were performed with a Quattro LC tandem mass spectrometer (Micromass, Manchester, UK) equipped with an electrospray source, as described previously (Yoneyama et al., 2008). HPLC separation was conducted with an LaChromUltra UHPLC instrument (Hitachi, Tokyo, Japan) fitted with an ODS (C18) column (LaChromUltra C18, 2 mm × 50 mm, 2 μm; Hitachi). The mobile phase was a water–methanol gradient. The gradient was 30–45% methanol for 3 min, 45–50% methanol for 5 min, 50–70% methanol for 4 min, 70–100% methanol for 3 min, 100% methanol for 3 min and 100–30% methanol for 1 min. The column was equilibrated at this solvent composition for 3 min before the next run. The total run time was 22 min. The flow rate was 0.2 ml min−1 and the column temperature was set to 40°C.

Effect of GR24 on Bryophytes and charophyte green algae

Physcomitrella patens gametophores from 6-wk-old colonies were carefully isolated and grown in a 96-well plate. Each well was previously filled with 200 μl of BCD medium (Grimsley et al., 1977) containing 0.01% acetonitrile (control) or 10 nM GR24 (Chiralix, Nijmegen, the Netherlands). After 3 wk, the length of the rhizoid of each gametophore was measured directly under a Leica (Leica, Wetzlar, Germany) RZ 75 stereomicroscope (> 30 gametophores). Comparison of the rhizoid lengths of the WT and Ppccd8 mutant was performed on 14-d-old gametophores grown in BCD medium (= 15).

For assays with Marchantia sp., between 50 and 70 gemmae collected from the same thallus were used for each treatment. They were grown on KNOP medium (Reski & Abel, 1985) gelled with 6 g l−1 Phytagel (Sigma) and containing 0.01% acetonitrile (control) or 10 nM GR24. After 1 wk, an image of each developing thallus (between 50 and 70) was acquired with a Leica RZ 75 stereomicroscope equipped with a Leica DFC320 camera. The lengths of the three longest rhizoids of each thallus were measured (Image Pro Plus, Media Cybernetics, Silver Spring, MD, USA).

For assays with C. corallina, fragments consisting of one node were used. To induce the formation of rhizoids, only one apical cell originating from the node was conserved (the others were cut). The segments were then placed in a glass tube (6 ml) containing 3 ml of sand and 3 ml of artificial pond water (APW; 1 mM NaCl, 0.1 mM KCl, 0.1 mM CaCl2) with 10 nM GR24 (treated) or 0.01% acetonitrile (control).

The mean rhizoid length for each treatment was compared with the corresponding control using the nonparametric Mann–Whitney test (< 0.05) with R software. Each experiment was repeated three times.


Canonical CCD7 and CCD8 seem to be lacking in green algae

To obtain an insight into the evolution of SLs in the green lineage, we performed a wide phylogenetic analysis of three proteins (D27, CCD7 and CCD8) known to be involved in SL synthesis in Angiosperms.

Based on the results of BLASTp, good hits (E-values < 10−40) for OsD27 were found in all genomes examined (Angiosperms, Gymnosperms, Lycophytes, mosses, Chlorophyceae, Prasinophyceae and Trebouxiophyceae) and in transcriptome assemblies of N. hyalina (Charales), C. globosum (Coleochaetales), S. pratensis, P. margaritaceum (Zygnematales), K. flaccidum (Klerbsormidiales) and C. atmophyticus (Coleochaetales) (Fig. 1), confirming the presence of OsD27 putative orthologs across the green lineage, even in charophyte green algae, as postulated previously by Lin et al. (2009).

Potential homologs of AtCCD7 were found in all genomes, with the exception of that of C. variabilis (Trebouxiophyceae), and in ESTs of M. polymorpha (liverworts), C. orbicularis (Coleochaetales) and C. globosum (Coleochaetales) (Fig. 1). Homologs of AtCCD8 were found in all genomes, except those of Prasinophyceae (O. tauri, O. lucimarinus and M. pusilla), in the ESTs of Adiantum capillus-veneris (Monilophyte) and Pinus taeda (Gymnosperm) and in the transcriptome assembly of C. atmophyticus (Chlorokybale) (Fig. 1). Phylogenetic analyses using ML were performed on the putative CCD7 and CCD8 sequences obtained by BLASTp, confirming their assignment as homologous proteins (Supporting Information Figs S1, S2).

Using structural analyses, Messing et al. (2010) showed that several amino acids of the maize carotenoid oxygenase ZmVP14 are essential for substrate specificity (Phe-171, Phe-411, Val-478 and Phe-589) or cleavage activity (four His). Moreover, they showed that the corresponding amino acids in ZmCCD1 are similarly essential. Based on these results, the authors hypothesized that these amino acids are crucial in the CCD protein family. We aligned the sequence of the putative CCD7 and CCD8 proteins from algae and Embryophytes with those of AtCCD7 and AtCCD8, and compared their respective important amino acids. We found the four histidines essential for cleavage activity. Focusing on the four amino acids proposed to be crucial for substrate specificity (Messing et al., 2010), we found that Leu of AtCCD7, corresponding to Phe-171 of ZmVP14, is generally well conserved, with some exceptions, in green algae (Fig. 2). The two Phe of AtCCD7, corresponding to Val-478 and Phe-589 of ZmVP14, are systematically present in CCD7 proteins of Embryophytes, including M. polymorpha (liverwort), but not in those of algae (Fig. 2). Ile of AtCCD7, corresponding to Phe-411 of ZmVP14, is not conserved in Embryophytes or algae.

Figure 2.

Partial alignment of CCD7 (CAROTENOID CLEAVAGE DIOXYGENASE 7) sequences. Red arrows indicate amino acids corresponding to the positions of Phe-171, Phe-411, Val-478 and Phe-589 of ZmVP14. These amino acids were proposed to be crucial for substrate specificity (Messing et al., 2010). Dashes indicate missing data. Black letters indicate conserved amino acids. Red letters indicate modified amino acids. Vertical lines show: Angiosperms (green), Lycophytes (red), moss (orange), liverwort (yellow), charophyte green algae (dark blue) and chlorophyte green algae (blue).

With regard to CCD8, we found that the four amino acids of AtCCD8 (Phe, Phe, Met, Leu), corresponding to Phe-171, Phe-411, Val-478 and Phe-589 of ZmVP14, are present in all CCD8 proteins of Embryophytes, with a minor modification (Ile instead of Leu) in the two CCD8 proteins of S. moellendorffii (Lycophyte), whereas CCD8 genes from green algae lack at least one of these four amino acids (Fig. 3). We therefore conclude that the canonical CCD7 and CCD8 proteins seem to be specific to Embryophytes.

Figure 3.

Partial alignment of CCD8 (CAROTENOID CLEAVAGE DIOXYGENASE 8) sequences. Red arrows indicate amino acids corresponding to the positions of Phe-171, Phe-411, Val-478 and Phe-589 of ZmVP14. These amino acids were proposed to be crucial for substrate specificity (Messing et al., 2010). Dashes indicate missing data. Black letters indicate conserved amino acids. Red letters indicate modified amino acids. Vertical lines show: Angiosperms (green), Gymnosperms (Brown), Monilophytes (dark orange), Lycophyte (red), moss (orange), charophyte green algae (dark blue) and chlorophyte green algae (blue).

SLs are present in liverworts, the most basal Embryophytes

The presence of canonical CCD7 and CCD8 in basal Embryophytes suggests that these organisms can synthesize SLs. This has already been proven for P. patens (Proust et al., 2011). To confirm this hypothesis, we searched for SLs in exudates of three additional Bryophytes: the liverworts Marchantia spp. (two species) and L. cruciata. These Bryophytes are most likely the earliest diverging lineage of Embryophytes (Qiu et al., 1998). First, purified exudates of the nonmycotrophic moss P. patens were tested as a positive control to validate our exudate purification method. We used the very sensitive germ tube branching assay on the AM fungus G. rosea. Germ tube branching activity was found in fraction F1–2 of purified exudates (Fig. S3). We then focused our analysis on liverworts. Crude extracts of exudates of the liverworts Marchantia spp. and L. cruciata were active in the germ tube branching assay (Fig. 4a,b). To identify the active molecules, specific MRM transitions, comparison of retention times with those of synthetic SL standards and/or co-injection analyses were used. As a result of the low concentration of SLs, even in the Marchantia tissue extracts, only the most abundant MRM transitions were recovered for each SL. The tissue extract of Marchantia spp. contained 5-deoxystrigol (Fig. S4: m/z 331 → 234, Rt = 21.4 min and Fig. 5: m/z 353 → 256, Rt = 14.3 min). Similarly, another specie of Marchantia (M. polymorpha) produced 5-deoxystrigol and five additional SLs identified as solanacol (m/z 365 → 268, Rt = 6.77 min), two new isomers of orobanchol (m/z 369 → 272, Rt = 9.17 min and Rt = 9.72 min), fabacyl acetate (m/z 427 → 270, Rt = 11.4 min) and orobanchyl acetate (m/z 411 → 254, Rt = 12.9 min) (Figs 5, S5). Based on these results, we conclude that SLs are produced by liverworts, the earliest diverging lineage of Embryophytes (Qiu et al., 1998).

Figure 4.

Gigaspora rosea germ tube branching activity in liverworts. Crude extracts of exudates of the liverworts Marchantia spp. (a) and Lunularia cruciata (b) induce strong germ tube branching (= 20–30 spores). Asterisks indicate extracts with significant activity compared with the negative control, according to the Mann–Whitney test; *, < 0.05.

Figure 5.

LC-MS/MS analysis of strigolactones of Marchantia polymorpha. Multiple Reaction Monitoring (MRM) chromatograms of Marchantia polymorpha extracts. Solanacol (m/z 365 → 268), two isomers of orobanchol (m/z 369 → 272), fabacyl acetate (m/z 427.1 → 270.1), orobanchyl acetate (m/z 411.1 → 254.1) and 5-deoxystrigol (m/z 353 → 256) are detected.

SLs are present in Charales

Based on the sequence comparison of CCD7 and CCD8, we postulate that SLs are probably lacking in green algae. To test this hypothesis, we examined the germ tube branching activity of exudates of several green algae: the model unicellular green alga C. reinhardtii, a Chlorophyte and five green algae belonging to the Charophytes: Spirogyra sp. (Zygnematales), C. scutata (Coleochaetales), C. corallina (Charales) and two Nitella species (Charales). Neither crude extract nor purified fractions of C. reinhardtii exudates were able to induce germ tube branching (Fig. 6). Similarly, no activity was found in the exudate extracts (even purified) of Spirogyra sp. or C. scutata (Fig. 6). Tissue extracts of these two algae were also not active (data not shown). In addition, no known SLs were found by MRM analysis of any of the three extracts (data not shown).

Figure 6.

Gigaspora rosea germ tube branching activity in green algae. Purified fractions of exudates of Chlamydomonas reinhardtii, Spirogyra sp. and Coleochaete scutata have no effect. Fractions 1 and 4 of Nitella hyalina are highly active. (= 20–30 spores). Asterisks indicate extracts with significant activity compared with the negative control, according to the Mann–Whitney test; *, < 0.05.

Unexpectedly, several purified fractions of the exudates of the three Charales, N. hyalina (Fig. 6), N. pseudoflabellata (Fig. S6a) and C. corallina (Fig. S6b), stimulated significantly germ tube branching. We also tested the activity of tissue extracts of the two Nitella species purified on an SPE Si column, and found activity in fractions SiF4, SiF5, SiF7 and SiF8 (Fig. S6c). Tissue extracts usually contain a larger amount of SLs and were preferred for LC-MS/MS analyses. The active molecules of SiF7 and SiF8 could not be characterized. For fractions SiF4 and SiF5, we identified, by MRM, two sorgolactone isomers (m/z 317 → 97, Rt  =  18.2 min and Rt = 18.5 min) as the major SLs produced by these organisms (Fig. 7). Retention times and MS/MS spectra were similar to those of the synthetic mix of sorgolactone stereoisomers, clearly confirming the identity of sorgolactones as the major SLs present in N. hyalina (Fig. 7) and N. pseudoflabellata (Fig. S7). Together, our results demonstrate that SLs are present in Charales and Embryophytes, and that their structures are not different.

Figure 7.

LC-MS/MS analysis of strigolactones of Nitella hyalina. (a) Multiple Reaction Monitoring (MRM) chromatogram (m/z 317 → 97) (left) and MS/MS spectrum of synthetic sorgolactone (right). (b) MRM chromatogram (m/z 317 → 97) of fractions SiF4 (black) and SiF5 (blue) of Nitella hyalina. (c) MS/MS spectra of sorgolactone isomer 1 (left) and 2 (right) detected in Nitella hyalina extracts.

D14-like appears with charophyte green algae

Interestingly, the SLs identified in these basal Streptophytes are similar to those found in Angiosperms. Thus, the evolution of SL function in the green lineage was probably the result of the emergence of new signaling pathway components, rather than the emergence of new SL structures. It has been proposed that MAX2/RMS4/D3 (F-BOX protein), D14 (α/β-fold hydrolase) and D14-like play crucial roles in SL perception in Angiosperms (Stirnberg et al., 2002; Ishikawa et al., 2005; Johnson et al., 2006; Arite et al., 2009; Waters et al., 2012). We looked for homologs of these three proteins in the same organisms as above.

We did not find potential orthologs of AtMAX2 in the Chlorophytes (Fig. 1). We identified closely related proteins in Picea sitchensis (Gymnosperm), Ceratopteris richardii (Monilophyte), S. moellendorffii (Lycophyte), P. patens (moss) and the charophyte green algae C. atmophyticus (Chlorokybales) and N. mirabilis (Charales) (Fig. 1). However, according to ML analyses, the putative F-BOX proteins of the two charophyte green algae are more closely related to AtFBL4, a different F-BOX protein of A. thaliana, than they are to AtMAX2 (Fig. S8).

We identified putative orthologs of D14 and D14-like in all Embryophyte genomes, in the ESTs of P. sitchensis and Picea glauca (Gymnosperms) and in the transcriptome assemblies of C. atmophyticus (Chlorokybales), K. flaccidum (Klerbsormidiales), S. pratensis (Zygnematales) and N. mirabilis (Charales) (Fig. 1). To identify each potential ortholog, we constructed ML trees of the D14 family using the putative D14 and D14-like BLAST hits and the dataset used by Waters et al. (2012) to resolve the phylogeny of these proteins in Embryophytes (Fig. 8). As expected, two distinct groups were recovered: one with the core D14 and D14L2 clades, and the other with the D14-like proteins (Fig. 8). Sequences of Angiosperms were found in each clade. Sequences of P. sitchensis and/or P. glauca (Gymnosperms) were found in D14-like and D14L2 clades and, by contrast with previous analyses (Arite et al., 2009; Waters et al., 2012), also in the core D14 clade (Fig. 8). Finally, the D14-like clade encompasses sequences of Angiosperms, P. glauca and P. sitchensis (Gymnosperms), P. patens (moss), M. polymorpha (liverworts) and charophyte green algae (Fig. 8). Interestingly, the sequences of C. atmophyticus (Chlorokybales), K. flaccidum (Klerbsormidiales) and S. pratensis (Zygnematales) clustered with a divergent clade of P. patens sequences, whereas the N. mirabilis (Charales) sequence was clearly found at the basis of the core D14-like clade (Fig. 8).

Figure 8.

Phylogeny of D14 and D14-like proteins. Maximum-likelihood (ML) tree of D14 and D14-like proteins. Bootstrap values above 50 are shown. The tree is rooted with bacterial RsbQ sequences as proposed by Arite et al. (2009) and Waters et al. (2012).

These results suggest that D14-like appears in the charophyte green algae, whereas D14 is specific to Gymnosperms and Angiosperms. In addition, we hypothesize that only Embryophytes contain MAX2 orthologs.

Exogenous and endogenous SLs control rhizoid elongation in basal Streptophytes

The presence of D14-like orthologs in charophyte green algae and basal Embryophytes supports the hypothesis of a hormonal function of SLs in these algae. Because charophyte green algae have been poorly studied, we lack genetic and molecular tools. To identify a potential hormonal role of SLs in these algae, our best possible strategy was to examine the effect of exogenously applied SLs on algal development. We treated three advanced charophyte algae, C. scutata (Coleochaetales), Spirogyra sp. (Zygnematales) and C. corallina (Charales), with 10 nM GR24. Chara corallina was preferred to Nitella because rhizoid formation in this species can be easily induced at the node after cutting off the branchlets. Then, we monitored several growth parameters (thallus size, rhizoid length and cell length). The development of C. scutata and Spirogyra sp. was not modified by the addition of GR24 (data not shown). By contrast, GR24 treatment stimulated significantly the rhizoid elongation of C. corallina (+ 50%, Fig. 9a). To test whether this effect was conserved in liverworts and moss, we also treated Marchantia sp. and P. patens with 10 nM GR24. Both Marchantia sp.-treated thalli (+ 42%) and P. patens-treated gametophores (+35%) exhibited a significant increase in basal rhizoid length (Fig. 9b,c). In contrast with charophyte green algae, P. patens is suitable for genetic studies. Proust et al. (2011), using a knockout (KO) mutant, demonstrated that SL synthesis in P. patens, as in Angiosperms, is partly dependent on PpCCD8 (Proust et al., 2011). To investigate the endogenous role of SLs on rhizoid elongation, we compared the rhizoid length of P. patens WT and Ppccd8 mutant. Rhizoids of Ppccd8 gametophores were significantly (20%) shorter than those of WT (Fig. 10). CCD enzymes may be involved in the biosynthesis of a large variety of apo-carotenoids. To confirm that the rhizoid phenotype was caused by the lack of SLs in the Ppccd8 mutant, complementation assays with exogenous GR24 were performed. Treatment with 10 nM GR24 restored the normal rhizoid length in the Ppccd8 mutant (Fig. 10).

Figure 9.

Effect of exogenously applied GR24 on Bryophytes and Charales. Rhizoid length of apical cells of Chara corallina (= 10 cells per treatment) (a), thalli of the liverwort Marchantia spp. (= 50–70 gemmae per treatment) (b) and gametophores of the moss Physcomitrella patens (= 30–40 gametophores per treatment) (c), treated with 10 nM GR24 (closed bars) or not treated (open bars). Scale bars, 1 mm. Error bars correspond to SEM. Asterisks indicate significant differences between the two treatments, according to Student’s t-test; *, < 0.05.

Figure 10.

Rhizoid growth of Physcomitrella patens wild-type (WT) and Ppccd8 mutant. The rhizoids of the Ppccd8 mutant are shorter than those of the WT. GR24 treatment (10 nM) of Ppccd8 restored the WT phenotype. Error bars correspond to SEM. Asterisk indicates significant difference according to the Kruskall–Wallis test (*, < 0.05) and box plot analysis.


SLs probably appeared in Charales

Our results show that exudate or tissue extracts of C. reinhardtii (Chlorophyte), Spirogyra sp. (Zygnematales) and C. scutata (Coleochaetales) neither activated germ tube branching of G. rosea nor contained MRM transitions for known SLs (Fig. 6). MRM analyses and retention time comparisons using synthetic or natural standards are powerful ways to detect the presence of SLs. However, the sensitivity of this approach remains low (10−9 M). By contrast, the germ tube branching assay on the AM fungus G. rosea (Buée et al., 2000) has been proven to be extremely sensitive for the detection of SLs at concentrations as low as 10−13 M (Besserer et al., 2006). It has been successfully used in recent years to detect SLs in various tissues and exudates of mycotrophic (Akiyama et al., 2005; Besserer et al., 2006; López-Ráez et al., 2008; Yoneyama et al., 2008; Balzergue et al., 2011) or nonmycotrophic plants (P. patens, this study). It can also respond to all natural SLs characterized to date (Akiyama et al., 2010; data not shown). Thus, we can confidently conclude that the analyzed extracts of C. reinhardtii (Chlorophyceae), Spirogyra sp. (Zygnematales) and C. scutata (Coleochaetales) did not contain SLs.

In Angiosperms, SL synthesis is controlled by environmental conditions. In particular, high concentrations of phosphate or nitrate abolish SL production (Yoneyama et al., 2008, 2011; Balzergue et al., 2011;). To limit such potential inhibitory conditions, all extractions were performed from tissues previously soaked in distilled water for 24 h. These conditions allowed us to identify SLs in Angiosperms as well as in moss, liverworts and Charales. Thus, although we cannot exclude the occurrence in SL-deficient algal extracts of an unknown inhibition of SL synthesis, our results suggest that, among the green algae, only Charales can produce SLs.

SL synthesis in Charales and liverworts supports the occurrence of a CCD8-independent SL biosynthesis pathway

In Angiosperms, the first steps in SL synthesis require enzymatic cleavage of a C40 carotenoid by CCD7 and CCD8 (Schwartz et al., 2004; Gomez-Roldan et al., 2008; Umehara et al., 2008; Xie et al., 2010). In the moss P. patens, CCD8 is also involved in SL biosynthesis (Proust et al., 2011). We found orthologs of CCD8, with the expected critical amino acids, in the Lycophytes, Monilophytes and Gymnosperms (Fig. 1), suggesting a strong conservation of this metabolic pathway from Embryophytes to mosses.

Here, we found that liverworts and Charales also produce SLs. However, no BLAST hits were found for CCD8 in the transcriptomes of N. hyalina (33 106 contigs), N. mirabilis (90 000 contigs) or Chara vulgaris (13 615 contigs, Professor B. Becker, pers. comm.). In addition, no ortholog of CCD8 was found in the ESTs or draft genome of M. polymorpha (Professor J. L. Bowman, pers. comm.). Moreover, we were unable to amplify CCD8 sequences from the genomic DNA of N. hyalina, C. corallina, C. scutata, Marchantia sp. or L. cruciata using degenerated primers (iCODEHOP, Boyce et al., 2009) designed on consensus sequences of algal and/or Embryophyte CCD8. These results suggest that M. polymorpha and Charales species, which produce SLs, could lack CCD8. Interestingly, the ccd8 mutants of A. thaliana (max4) and P. patens (Ppccd8) still produce a low, but detectable, amount of SLs (Kohlen et al., 2010; Proust et al., 2011), suggesting the presence of another biosynthesis pathway.

Taken together with the presence of SLs in some ccd8 mutants, these results suggest the appearance in Charales of an earlier CCD8-independent SL biosynthesis pathway conserved in Embryophytes and predating the CCD8-dependent pathway. The involvement of CCD7, present with the expected amino acids in the ESTs (Fig. 2) and draft genome (Professor J. L. Bowman, pers. comm.) of M. polymorpha, in this alternative biosynthesis pathway will require further investigation, including the analysis of SL production in ccd7/ccd8 double mutants and complete sequencing of a Charale genome.

SLs were probably first selected as plant hormones rather than as symbiotic signals

In Angiosperms, SLs have two distinct biological activities: a hormonal role in the control of shoot and root branching and a rhizospheric signaling role for parasitic weeds and AM fungi. The hormonal role on shoot and root branching is not expected to occur in basal Streptophytes because shoots and roots are specific traits of vascular plants. Moreover, two genes, D14 (Arite et al., 2009) and FC1 (OsTB1, Minakuchi et al., 2010), which act downstream of SLs to inhibit bud outgrowth, are specific to Spermatophytes (Navaud et al., 2007; Arite et al., 2009; Waters et al., 2012; Figs 1, 8). In the moss P. patens, SLs control the expansion of protonema (Proust et al., 2011). This hormonal regulation also cannot exist in liverworts, as they lack this developmental stage. By contrast, most Embryophytes, including Angiosperms, mosses and liverworts, share the ability to live symbiotically with AM fungi (Smith & Read, 2008). Because SLs are present in basal Embryophytes, a primitive signaling function for SLs has been hypothesized: the signaling role for the promotion of the AM symbiosis (Bouwmeester et al., 2007). However, this hypothesis is challenged by the fact that Charales, which do not participate in AM symbiosis, also produce SLs.

We show that Charales can not only synthesize SLs, but can also exude them into the medium (Fig. 6). Interestingly, an ABC-transporter of petunia has been shown recently to be involved in SL cellular export and exudation (Kretzschmar et al., 2012). ABC-transporters are widely distributed proteins (Dassa, 2011) and their involvement in SL exudation in basal Streptophytes needs to be examined. In Charales, we speculate that SLs could be exuded to promote another beneficial plant–microorganism association. However, only one endophytic association between a Chytridiomycete fungus and Charales has been reported (in a fossil record of Paleonitella) and is more probably pathogenic than symbiotic (Taylor et al., 1992). An association between extant Charales and epiphytic nitrogen-fixing bacteria has also been reported (Ariosa et al., 2004); however, SLs seem unable to stimulate bacterial growth (Soto et al., 2010). Thus, we postulate that the primitive function of SLs in Charales was probably hormonal. Supporting this hypothesis, a gene coding for a D14-like homolog is present and expressed in Charales (Figs 1, 8). Very recently, Waters et al. (2012) demonstrated that, in A. thaliana, D14-like is required for SL signaling. The D14-like orthologs present in Charales could play a similar role.

Menand et al. (2007) demonstrated that a basic helix–loop–helix transcription factor (AtRHD6/PpRSL1) regulates root hair development in A. thaliana and rhizoid development in P. patens, suggesting that root hairs and rhizoids share a similar developmental pathway. In a recent study, Kapulnik et al. (2011) showed that exogenous treatment with the SL analog GR24 stimulates the elongation of A. thaliana root hairs. Our results show a similar effect on the rhizoids of the model moss P. patens, the liverwort Marchantia sp. and the Charales C. corallina (Fig. 9). We hypothesize that SLs appeared with these organisms, where they played a role in rhizoid elongation and thus increased their anchorage ability, a positive trait for plant terrestrial colonization. Later, SLs could have been recruited to facilitate symbiotic fungal interaction. SLs would have been positively conserved during land plant evolution for these two underground functions, and then used as a root hormone to control root hair elongation and, more recently, as a shoot hormone to control the aerial architecture of flowering plants (Gomez-Roldan et al., 2008; Umehara et al., 2008).

Are Charales the closest relatives to Embryophytes?

The presence in Charales, as in Embryophytes, of SLs, rhizoids and D14-like proteins, and the fact that we could not find any of these characteristics in Coleochaetales and Zygnematales, is consistent with the hypothesis that Charales are the closest relatives to land plants (Karol et al., 2001; Turmel et al., 2003, 2006; McCourt et al., 2004). However, some recent phylogenetic studies (including ours) place Coleochaetales and/or Zygnematales, rather than Charales, as the sister group of Embryophytes (Finet et al., 2010; Wodniok et al., 2011). Clearly, the identity of the sister taxon to Embryophytes remains an open question. Some interesting hypotheses have been developed around the notion that Zygnematales and Coleochaetales are secondarily reduced. Stebbins & Hill (1980) proposed that the early evolution of Charophytes took place on land on moist soil surfaces. They postulate that the more derived green algal lineages (Coleochaetales, Zygnematales and Charales) then returned to aquatic life, inducing large losses of developmental processes and previously acquired structures (Stebbins & Hill, 1980). If, as recent evidence shows, either the Coleochaetales (Finet et al., 2010) or Zygnematales (Wodniok et al., 2011), rather than the Charales, are sister to Embryophytes, the lack of SLs, rhizoids and D14-like proteins in these lineages would be the result of such losses, and the ideas of Stebbins & Hill (1980) should be revisited (hypothesis 1, Fig. 1). However, multi-gene studies from large datasets involve complex and subtle analytical challenges, and the phylogeny proposed by Karol et al. (2001), with Charales as the sister group to land plants, which is easiest to reconcile with the SL and rhizoid data presented here, remains a viable hypothesis (hypothesis 2, Fig. 1).


The authors thank Dr J-P. Vors (Bayer CropScience) and Dr J. Chave (Toulouse University, France) for stimulating and helpful discussions, and Dr S. Rochange (Toulouse University, France) and K. Forshey (University of Wisconsin Madison, WI, USA) for critical reading of the manuscript. They are grateful to Professor I. Foissner (Salzburg University, Austria) for providing Nitella species, to Dr C. Rameau (INRA Versailles, France) for the Physcomitrella patens strains, to Professor T. Kohchi (Kyoto University, Japan) for the Marchantia polymorpha strain, to Professor J. L. Bowman (Monash University, Australia) and Professor B. Becker (Koln University, Germany) for BLAST analyses of the unpublished data of Marchantia polymorpha and Chara vulgaris, to Jay Thierer for unpublished data, and to Dr G. Concepcion and T. Gibbons for technical help. Mass spectrometry experiments were carried out on the Metabolomics and Fluxomics platform of Toulouse (MetaToul). P-M.D. was funded by grant award No. CIFRE0391/2008 from Bayer CropScience. X.X. was supported by the Japanese Society for the Promotion of Science and a Post-Doctoral Fellowship for Foreign Researchers. Part of the work of K.Y.’s group was supported by the Program for the Promotion of Basic and Applied Researches for Innovations in Bio-oriented Industry. Part of the work of G.B.’s group was supported by the LABEX TULIP project.