The significance of plastid terminal oxidase (PTOX) in phytoene desaturation and chloroplast function has been demonstrated using PTOX-deficient mutants, particularly in Arabidopsis. However, studies on its role in monocots are lacking. Here, we report cloning and characterization of the rice (Oryza sativa) PTOX1 gene.
Using Ecotype Targeting Induced Local Lesions IN Genomes (EcoTILLING) and TILLING as forward genetic tools, we identified the causative mutation of an EMS mutant characterized by excessive tillering, semi-dwarfism and leaf variegation that corresponded to the PTOX1 gene.
The tillering and semi-dwarf phenotypes of the ptox1 mutant are similar to phenotypes of known strigolactone (SL)-related rice mutants, and both phenotypic traits could be rescued by application of the synthetic SL GR24. The ptox1 mutant accumulated phytoene in white leaf sectors with a corresponding deficiency in β-carotene, consistent with the expected function of PTOX1 in promoting phytoene desaturase activity. There was also no accumulation of the carotenoid-derived SL ent-2′-epi-5-deoxystrigol in root exudates. Elevated concentrations of auxin were detected in the mutant, supporting previous observations that SL interaction with auxin is important in shoot branching control.
Our results demonstrate that PTOX1 is required for both carotenoid and SL synthesis resulting in SL-deficient phenotypes in rice.
Variegated plants produce white/yellow sectors that are characterized by abnormal plastid development as well as chlorophyll and carotenoid deficiency (Miura et al., 2007; Sakamoto et al., 2009). Variegations are induced either by external agents or by mutations in nuclear, plastid or mitochondrial genes (Yu et al., 2007). Mutations resulting in leaf variegation are known in many plant species, but few variegated mutants have been characterized at the molecular level to date. Those that have been characterized include the Arabidopsis thaliana immutans (im) (Carol et al., 1999; Wu et al., 1999; Rosso et al., 2009), var1 (Chen et al., 1999; Rosso et al., 2009), var2 (Chen et al., 1999; Yu et al., 2008; Rosso et al., 2009; Sakamoto et al., 2009), var3 (Næsted et al., 2004), chloroplast mutator (chm) (Abdelnoor et al., 2003) and pds3 (Qin et al., 2007) mutants, the tomato (Solanum lycopersicum) ghost (gh) (Josse et al., 2000; Carol & Kuntz, 2001; Barr et al., 2004), the maize (Zea mays) iojap (ij) (Han et al., 1992), as well as the zebra2 mutant of rice (Oryza sativa) (Han et al., 2012). Most of these mutants are recessive with the mutations residing in nuclear genes, which means that both the green and white sectors have the same genetic background, thus offering excellent opportunities to study nuclear–organelle interactions and chloroplast biogenesis (Rodermel, 2001; Aluru et al., 2006; Sakamoto et al., 2009).
The Arabidopsis im mutant is characterized by arrested chloroplast biogenesis in the white leaf sector (Wetzel et al., 1994) as well as impaired carotenoid biosynthesis leading to accumulation of the intermediate compound phytoene, whose conversion to lycopene is catalyzed by two closely related nuclear-encoded enzymes: phytoene desaturase (PDS) and ζ carotene desaturase (ZDS) (Wetzel et al., 1994; Carol & Kuntz, 2001). IMMUTANS is a nuclear gene that encodes a chloroplast-targeted plastid terminal oxidase (PTOX, also referred to as plastoquinol terminal oxidase). Its primary role in chloroplasts is the specific oxidation of the plastoquinone (PQ) pool by transferring electrons from PQ to molecular oxygen, thus making PQ available for reduction (Aluru et al., 2006; Rosso et al., 2006; McDonald et al., 2011). This activity contributes to a number of important processes in plants including chlororespiration, photosystem I cyclic electron transport, the regulation of photosystem II excitation pressure, as well as providing oxidized PQ to PDS and ZDS for carotenoid synthesis (Rosso et al., 2009; Okegawa et al., 2010; McDonald et al., 2011).
Although the im and gh mutants have been well characterized, previous studies have mainly focused on the role of PTOX in various aspects of chloroplast function. The possible effect of PTOX on other plant processes is not well understood although the indirect effect PTOX might have on plant processes downstream of the carotenoid biosynthesis pathway, such as the synthesis of the plant hormone abscisic acid (ABA), has been suggested (Aluru et al., 2001). Another class of plant hormones that are synthesized from carotenoids is the strigolactones (SLs). These are a group of terpenoid lactones that regulate shoot branching (Gomez-Roldan et al., 2008; Umehara et al., 2008) and are derived from carotenoids through a sequential oxidative cleavage and subsequent oxidation (Xie et al., 2010; Alder et al., 2012). Attempts have been made to elucidate the SL biosynthetic and perception/signaling pathways using several branching mutants characterized from multiple species including Arabidopsis (more axillary growth, max1-max4), pea (Pisum sativum) (ramosus, rms1-rms5), petunia (Petunia hybrida) (decreased apical dominance1, dad1-dad3), and rice (dwarf, high tillering dwarf, d/htd) (for reviews, see Dun et al., 2009; Beveridge & Kyozuka, 2010; Xie et al., 2010). The manner by which SL inhibits branching is being investigated (Seto et al., 2012; Brewer et al., 2013; Kagiyama et al., 2013), and an increasing body of evidence shows the importance of SL interaction with other plant hormones, particularly auxin, in the control of shoot branching (Hayward et al., 2009; Crawford et al., 2010; Shinohara et al., 2013).
Targeting Induced Local Lesions IN Genomes (TILLING) is a high-throughput reverse genetic approach initially developed for screening of induced mutations (McCallum et al., 2000), and further extended to the detection of polymorphisms in natural populations in a technique termed Ecotype Targeting Induced Local Lesions IN Genomes (EcoTILLING) (Comai et al., 2004). In principle, both techniques could be applied as forward genetic tools to improve the efficiency of map-based gene cloning. EcoTILLING polymorphisms detected between mapping parents can be used as markers to fine-map candidate loci. Once the candidate locus has been partially localized, TILLING of this region using DNA pooled from wild-type (WT) and mutant plants in a 1 : 1 ratio provides an alternative to sequencing for identifying the causative mutation. Here, we report the application of both the EcoTILLING and TILLING protocols as forward genetic approaches for fine mapping and identification of the causative mutation of a rice mutant showing a variegated and semi-dwarf phenotype with excessive tillering similar to that observed in SL-related rice mutants (Arite et al., 2007). We show that the causative mutation is present in the rice PTOX1 gene and provide evidence for its indirect involvement in the control of plant stature and tillering through a role in carotenoid-derived SL synthesis.
Materials and Methods
Plant materials and growth conditions
The rice (Oryza sativa L.) Sas1248 mutant was identified among c. 12 000 mutant lines generated by EMS treatment of two elite japonica cultivars Sasanishiki and Hitomebore. The Tos17 insertion knockout line of PTOX1 (ptox1-2) was obtained from the National Institute of Agrobiological Sciences (NIAS) in Japan. The Arabidopsis thaliana im seeds were kindly provided by Prof. Marcel Kuntz.
For experiments involving Arabidopsis, WT Col-0 and im plants were germinated and grown on half-strength Murashige & Skoog (MS) medium for 7 d under low light (c. 4 μmol m−2 s−1) and then transferred to medium light intensity (c. 75 μmol m−2 s−1). All plants were maintained on peat pellets (Sakata Seed; http://www.sakataseed.co.jp) in 14 h : 10 h, light : dark photoperiods at 22°C. Rice plants were germinated and grown on soil under natural light in the glasshouse in the summer.
For EcoTILLING, polymorphic regions between the two mapping parents, Sasanishiki and 144, were first screened using primers designed to amplify 1.3–1.5-kb target regions randomly selected at c. 10–20-kb intervals within the candidate regions delineated by SSR markers using DNA pooled from the two cultivars in a 1 : 1 ratio. Polymorphic EcoTILLING primers were then used for fine mapping of the candidate locus using F2 mutant plants still segregating for the flanking SSR markers. TILLING of the candidate region delineated by EcoTILLING was performed using genomic DNA of WT and Sas1248 pooled in an equal ratio. For both TILLING and EcoTILLING, a two-step PCR was used. In the first step, each primer was tailed with known 20-bp noncomplementary sequences for labeling the target sequence with Infrared dyes (IRD). The PCR product obtained from the initial reaction was purified and diluted, and was then re-amplified using primers labeled with IRD700 and IRD800. This step allowed the labeling of the target locus at both ends so that an unlimited number of loci could be screened using only two dye-labeled oligo-nucleotides with minimum cost.
EcoTILLING/TILLING was performed according to the methods of Rakshit et al. (2007) with the following modifications. The first PCR was carried out in a 20 μl reaction volume containing 9 ng genomic DNA, 4 mM each dNTPs, 0.50 U TaKaRa ExTaq (TaKaRa Bio Inc., www.takara-bio.com) and 2 μl each of the tailed amplification primers using the following thermal cycling profile: 95°C for 2 min/35 cycles of 95°C for 1 min, 60°C for 1 min, 72°C for 1 min 30 s/72°C for 7 min. The PCR products were purified and eluted with 30 μl sterile water. The second PCR was carried out in a 10 μl reaction volume consisting of 2 μl amplified product, 2 mM each dNTPs, 0.25 U TaKaRa ExTaq, and 0.2 μM each of the left and right primer mix using the following PCR profile: 95°C for 2 min/35 cycles of 95°C for 1 min, 60°C for 1 min, 72°C for 1 min 30 s/72°C for 7 min/99°C for 10 min/70°C for 20 s with touch down of 0.3 s/cycl for 70 cycles. For CEL-I digestion, Surveyor™ nuclease from Transgenomic (Omaha, NE, USA) was used at 0.07 μM/reaction. The IRD dye-labeled fragments were resolved on a LI-COR 4300 DNA analyzer (Li-Cor Biosciences, Lincoln, NE, USA) that is fitted with a two-dye channel (700 and 800) detection system.
GR24 was in part kindly provided by Dr Shinjiro Yamaguchi of RIKEN Plant Science Center, Yokohama, Japan, and additionally purchased from Chiralix (http://www.chiralix.com). The mesocotyl elongation assay was performed as previously described by Hu et al. (2010). Ten mM GR24 (dissolved in acetone) was added to the agar medium to a final concentration of 1 and 5 μM GR24. Agar medium with 0.05% acetone was used as a negative control. For complementation of the tillering and semi-dwarf phenotypes of ptox1 by exogenous SL, 1 μM GR24 was supplied with irrigation water at 1-wk intervals starting from 1 wk after sowing. ABA and GA3 (Wako, www.wako-chemicals.co.jp) were added to irrigation water to a final concentration of 5 μM each, and were supplied at 10-d intervals beginning 1 wk after sowing.
Construction of vectors and generation of transgenic lines
Full-length Os04g0668900 cDNA clone (AK067891/J013124H21), obtained from the National Institute of Agrobiological Sciences in Japan, was cloned into the binary vector pCAMBIA1300 (Cambia, www.cambia.org). The resulting construct was introduced into Escherichia coli (strain DH5αa), which was transferred to Agrobacterium tumefaciens (strain EHA101) and used for transforming the ptox1-1 mutant by Agrobacterium-mediated transformation following the methods of Okuyama et al. (2011).
To prepare the RNAi construct, a 324 bp PTOX1 cDNA fragment spanning the coding (9th exon) and 3′UTR region was amplified by PCR from a Sasanishiki WT plant, cloned into the Gateway vector pENTR/D-TOPO (Invitrogen, Carlsbad, CA, USA) and transferred into the two recombination sites of the pANDA vector (Miki & Shimamoto, 2004) using LR Clonase (Invitrogen). The resulting vector, pANDA-PTOX1, was introduced into A. tumefaciens and transformed into Sasanishiki by Agrobacterium-mediated transformation (Okuyama et al., 2011).
Gene expression analysis by quantitative RT-PCR
Total RNA was isolated from transformant T0 leaves using the Qiagen RNeasy plant mini kit (Qiagen, Venlo, the Netherlands). cDNA was synthesized with the ReverTra Ace kit (TOYOBO, http://www.toyobo.co.jp), and was then used as a template for qRT-PCR. qRT-PCR was performed using SYBR Green PCR Master Mix (Qiagen) on a Step-One Plus Real-Time PCR system (Applied Biosystems, Foster City, CA, USA). PTOX1 transcript levels were normalized with the Actin1 gene. The primer sequences used in this study were as follows: PTOX1-F: (5′-CATGGAAACCTTCGTTCTCCTC-3′), PTOX1-R (5′-TTTGACACAATCCACGATACCTTC-3′), Actin1-F (5′-CCCCCATGCTATCCTTCGT-3′), Actin1-R (5′-GGCCGTTGTGGTGAATGAGT-3′).
Detection of PTOX protein
One gram of rice WT and mutant leaves were homogenized with 2 ml of GTN buffer (10% glycerol, 25 mM Tris-HCl pH7.5, 150 mM NaCl, 10 mM DTT and cOmplete EDTA-free protease inhibitor cocktail; Roche, Basel, Switzerland). Homogenates were filtered using a metal filter (pore size 10 μm), and 200 μl of the filtrates were centrifuged at 20 000 g at 4 °C for 15 min. The resulting pellet was resuspended with 200 μl GTN buffer. 25 μl 4 × protein sample buffer (40% (w/v) glycerol, 240 mM Tris-HCl pH 6.8, 8% (w/v) SDS, 0.04% (w/v) bromophenol blue, 400 mM DTT) was then added to 75 μl sample solutions, which were used for SDS-PAGE and immunodetection using a polyclonal antibody raised against Arabidopsis PTOX purchased from Uniplastomic (www.biotech-pipeline.fr/index.php/en).
A BLASTP search using the PTOX1 amino acid sequence as a query was performed using the NCBI nonredundant protein database with default settings (http://www.ncbi.nlm.nih.gov). The phylogenetic analysis was conducted using the programs provided in the online tool Phylogeny.fr platform (www.phylogeny.fr; Dereeper et al., 2008). The selected full-length amino acid sequences were aligned using ClustalW v2.0.3 (www.clustal.org), and the neighbor-joining phylogenetic tree was generated by the BioNJ program using the default parameters. The tree was viewed using the TreDyn program v198.3. The bootstrap values for nodes in the phylogenetic tree are from 1000 replications. Protein sequences used: Oryza sativa (NP_001054199.1), Sorghum bicolor (XP_002448728.1), Brachypodium distachyon (XP_003580780), Zea mays (NP_001150780.1), Populus trichocarpa (XP_002317190), Ricinus communis (XP_0025-18411.1), Glycine max (XP_003546171.1), Solanum lycopersicum (NP_001234511.1), Arabidopsis thaliana (NP_567658.1), Capsicum annuum (AAG02288.1), Medicago truncatula (XP_003594164.1), Picea sitchensis (ABK22767.1), Coffea canephora (ABB70513.1), Vitis vinifera (CBI41031.3) and Selaginella moellendoriffii (XP_002973024.1).
Analysis of carotenes
Carotenes in the pigment extracts were analyzed according to the protocols of Takaichi (2000) using an HPLC equipped with a Novapak C18 column (Waters, Milford, MA, USA). Pigments were identified based on the retention times on HPLC and absorption spectra in the eluent. Note that other polar pigments were eluted before 5 min.
The SL ent-2′-epi-5-deoxystrigol was analyzed as described by Xie et al. (2013). Rice seedlings were initially grown on pure water for 1 wk, and then on half-strength Tadano and Tanaka medium (Tadano & Tanaka, 1980) for an additional 5 wk, followed by a growing period of 10 d on a quarter-strength Tadano and Tanaka medium. Finally, seedlings were incubated for 10 d on a phosphate-free medium, which was collected and extracted with ethyl acetate after adding 5 ng of [2H6]-ent-2′-epi-5-deoxystrigol as an internal standard. The ethyl acetate phase was washed with 0.2 M phosphate buffer, pH 9 and passed through a silica gel column according to Xie et al. (2013). ent-2′-epi-5-deoxystrigol was quantified by LC-MS/MS using an Agilent 6460 MS coupled to an Agilent 1200 LC equipped with an Agilent ZORBAX Eclipse XDB C18 column (2 × 50 mm, 1.8 nm; Agilent, Santa Clara, CA, USA). The LC solvent was run at a flow-rate of 0.2 ml min−1 at 40°C and programmed as follows: 0–3 min (30–45% MeOH), 3–8 min (45–50% MeOH), 8–12 min (50–70% MeOH), 12–15 min (70–98% MeOH).
Hormone quantification by HPLC-EMSI-MS/MS
Quantification of auxins, abscisic acid and metabolites, cytokinins, and gibberellins was performed at the National Research Council of Canada (Saskatoon) according to the method for simultaneous analysis of plant hormones described by Chiwocha et al. (2003, 2005). Calibration curves were generated from the MRM signals obtained from standard solutions based on the ratio of the chromatographic peak area for each analyte to that of the corresponding internal standard, as described by Ross et al. (2004). The QC samples, internal standard blanks and solvent blanks were also prepared and analyzed along each batch of samples.
Phenotypic characterization of Sas1248
We screened over 12 000 M3–M5 mutant lines generated by EMS-treatment of two elite japonica rice cultivars Hitomebore and Sasanishiki for a range of morphological phenotypes. Details of the mutagenesis protocol are given elsewhere (Rakshit et al., 2010). Of the mutants we identified in the Sasanishiki background, Sas1248 displayed excessive tillering and semi-dwarfism (Fig. 1a–e). At the mature stage, the mutant produced c. 40 more tillers (Fig. 1c,d) while its height was only c. 65% that of wild-type (WT) plants (Fig. 1c,e). Comparison of internode length showed that the reduced stature of the mutant was the result of all internodes being shorter compared to WT (Fig. 1f,g). The mutant also produced smaller panicles with poor seed set (Fig. 1h). Sas1248 is further characterized by leaf variegation with distinct green and white sectors (Fig. 1b; Supporting Information Fig. S1), although the extent of the variegation varied under our growth conditions (Fig. S1).
Map-based cloning of the causal mutation using EcoTILLING polymorphisms as molecular markers for fine mapping
In order to clone the mutated gene, we used 157 mutant individuals of an F2 progeny derived from the cross of Sas1248 and 144 (indica) that segregated 3 : 1 (WT to mutant phenotype), suggesting a recessive homozygous gene governs the mutant phenotype. We initially mapped the candidate locus to a c. 633-kb region between two simple sequence repeat (SSR) markers RM17597 and RM8220 on the long arm of chromosome 4 (Fig. 2a). For fine mapping, we employed the EcoTILLING protocol as a forward genetic tool (see the 'Materials and Methods' section). Accordingly, EcoTILLING markers (TMs) that detected single nucleotide polymorphisms (SNPs) between Sasanishiki and 144 were used to amplify target regions from genomic DNA extracted from individual F2 plants that were still segregating for the flanking SSR markers (Figs 2a, S2, S3). For instance, the polymorphic markers identified within the genomic region flanked by RM17597 and RM3466 were used to check the segregation pattern of the four F2 plants still segregating for RM17597 (Figs 2a, S3). This allowed further delineation of the causal mutation to a c. 168-kb region between markers TM072 and TM013 that is predicted to contain 36 open reading frames (ORFs) (Fig. 2a).
Identification of the causative mutation by TILLING
In order to identify the mutation responsible for the mutant phenotypes, TILLING of the 168-kb candidate region was undertaken. TILLING primers were designed to amplify 1.3- to 1.5-kb overlapping segments spanning all the predicted gene loci, including at least 1 kb upstream and downstream of the predicted transcribed region, from a DNA pool of WT Sasanishiki and Sas1248 mixed in a 1 : 1 ratio. Heating and cooling of the PCR products produces mixtures of homo- and heteroduplexes, and the mismatches in the latter that have been introduced by EMS treatment are detected and cleaved by the mismatch endonuclease (CEL I) digestion. This cleavage results in two separate fragments whose additive size is equal to the expected size of the entire product (Barkley & Wang, 2008). The two separate fragments can be detected in the 700 and 800 dye channels of the LI-COR 4300 DNA analyzer. Accordingly, we detected a mutation within the target region of one of the primers (34 622 373–34 623 722, Chr. 4, Build 5.0; IRGSP 2005), and the mutation was confirmed both in the 700 and 800 dye channels (Fig. 2b; see the 'Materials and Methods' section).
Sequencing the target region from WT and Sas1248 plants identified a T to A substitution at the 729th nucleotide of the coding sequence (CDS) of Os04g0668900 (Fig. 2a) that introduced a premature stop codon (TAT: Tyr to TAA: Stop) within the 7th exon (Fig. 3a). The gene encodes a protein similar to plastid terminal oxidase (PTOX), which we have named PTOX1 as the rice genome also contains a second gene on chromosome 3, Os03g0847500, with homology to PTOX (Fig. S4). However, Os03g0847500 is predicted to encode a protein of only 154 amino acids and is therefore unlikely to be a functional PTOX (Fu et al., 2005).
PTOX1 is the rice ortholog of IMMUTANS and GHOST
The PTOX1 gene is comprised of nine exons and eight introns (Fig. 3a), and is predicted to encode a 336-amino acid PTOX including a 35-aa chloroplast transit peptide (http://www.cbs.dtu.dk/services/ChloroP/). The mature protein of 301 amino acids has a predicted molecular mass of 35.5 kDa. PTOX1 shares 68% and 74% amino acid sequence identity with two well-characterized members of the PTOX protein family, the Arabidopsis IM and tomato GHOST, respectively (Fig. 3b), suggesting that PTOX1, IM and GHOST are orthologs. To assess the evolutionary relationship between the PTOX proteins, we performed a phylogenic analysis based on the neighbor-joining method. The resulting phylogenetic tree showed that PTOX from monocots form a separate clade from that of dicot PTOXs (Fig. 3c).
The ptox1 loss-of-function mutation is responsible for all mutant phenotypes
The high similarity between PTOX1, IM and GHOST amino acid sequences suggested that the mutation detected in PTOX1 is at least responsible for the Sas1248 variegated phenotype. To verify this, and also determine whether the same mutation is responsible for the additional mutant phenotypes, we carried out RNA interference (RNAi) analysis targeting the PTOX1 gene in WT Sasanishiki plants. RNAi transformants displayed semi-dwarf and tillering phenotypes and had reduced levels of PTOX1 transcripts (Fig. 4a–c). Additionally, we obtained a retrotransposon Tos17 insertion line (FT930204) with an insertion in the Os04g0668900 gene. The line was identified among the mutant lines induced by rice retrotransposon Tos17 in the Japonica cultivar Nipponbare background and currently maintained at the National Institute of Agrobiological Sciences in Japan (Miyao et al., 2003). First, we identified individual plants heterozygous for the insertion among the plants established from the M2 seeds received. The M3 plants generated from seeds of heterozygous individuals further segregated for WT and mutant phenotypes that were similar to phenotypes of the Sas1248 mutant (Fig. 4d). By sequencing, we confirmed that the Tos17 insertion site is close to the start of exon 9 (Fig. S5). We accordingly named Sas1248 as ptox1-1 and the Tos17 insertion line as ptox1-2.
We further carried out a complementation test by transforming ptox1-1 plants with the full-length PTOX1 cDNA driven by the cauliflower mosaic virus (CaMV) 35S promoter. Phenotypic analysis of six independent transgenic lines (T0) confirmed that PTOX1 fully complemented all mutant phenotypes including leaf variegation, dwarfism and excessive tillering (Fig. 4e). In addition, we performed immunoblot analysis using an anti-PTOX antibody raised against Arabidopsis IM, which we confirmed recognizes a protein of the correct size in WT Arabidopsis that is absent in the im mutant (data not shown). The antibody cross-reacted with a number of other proteins in rice, but detected a band that corresponded with the predicted size of PTOX1 in WT rice (Fig. 4f). This protein was absent in ptox1-1, but was detected at high levels in the complemented line shown in Fig. 4e, consistent with the strong expression of PTOX1 driven by the CaMV 35S promoter. The ptox1-2 mutant also did not accumulate PTOX1, while a protein of the correct size was detected in WT Nipponbare (Fig. 4f). The ptox1-1 mutant has a premature stop codon in exon 7 and thus lacks a highly conserved 16 amino acid sequence (corresponding to exon 8) that is required for PTOX protein stability (Fig. 3b; Fu et al., 2005). The Tos17 insertion at the start of exon 9 results in ptox1-2 lacking two of the six conserved iron-binding sites that are also required for PTOX activity (Fig. 3b; Fu et al., 2005). However, in neither case did we observe the accumulation of a truncated PTOX1 protein. Taken together, our data demonstrate that all of the ptox1 phenotypes result from a loss-of-function mutation identified in the PTOX1 gene.
PTOX1 deficiency leads to phytoene accumulation and reduced SL synthesis
Disruption of the carotenoid biosynthetic pathway at phytoene desaturation causes accumulation of phytoene, the C40 colorless intermediate in carotenoid biosynthesis, and leads to β-carotene deficiency (Wetzel et al., 1994; Qin et al., 2007). To test whether this is also the case in ptox1 mutant, we analyzed contents of the various carotenes in predominantly white leaf sectors of ptox1-2. Consistent with the role of PTOX in carotene biosynthesis, we detected phytoene accumulation and a corresponding reduction in β-carotene contents in the white leaf sector of ptox1-2 plants (Fig. S6). Because ptox1 has reduced β-carotene contents, we hypothesized that the mutant would be deficient in carotenoid-derived SLs. To investigate the possible role of PTOX1 in SL biosynthesis, we measured 2′-epi-5-deoxystrigol, one of the major rice SLs (Umehara et al., 2008; Xie et al., 2013), in root exudates of WT and the ptox1-1 mutant. As shown in Fig. 5, we detected 2′-epi-5-deoxystrigol in WT, but not in root exudates from the ptox1-1 mutant. These results show that the ptox1 mutation affects SL synthesis and that this is likely to be due to a direct effect on β-carotene synthesis.
The ptox1 phenotype is related to strigolactone deficiency
Dark-grown seedlings of SL-deficient rice mutants show accelerated mesocotyl elongation, which can be complemented by the synthetic SL analog GR24 (Hu et al., 2010). We used this assay to evaluate the possibility that the ptox1 mutant phenotype is associated with strigolactone deficiency. WT and ptox1-1 seeds were grown on agar medium in darkness for 8 d with and without GR24, and the mesocotyl length was measured. Mesocotyls of ptox1-1 seedlings were c. 10-fold longer than those of WT seedlings (Fig. 6). Moreover, the elongated mesocotyl phenotype of ptox1-1 was rescued by GR24 in a dose-dependent manner with application of 5 μM GR24 sufficient to reduce the ptox1-1 mesocotyl length to that of WT (Fig. 6). By contrast, and consistent with previous studies (Hu et al., 2010), application of GR24 did not have an effect on coleoptile growth (Fig. 6b).
Rice mutants with reduced SL biosynthesis or signaling such as d3, d10, d14, d17 and d27 also display reduced stature and enhanced tillering phenotypes that are similar to phenotypes of the ptox1 mutants described here (Arite et al., 2007; Umehara et al., 2008; Hu et al., 2010). We therefore tested whether the growth and tillering phenotypes of ptox1 plants was due to inhibition of SL synthesis by examining the effect of GR24 treatment. As shown in Fig. 7, application of 1 μM GR24 fully inhibited bud outgrowth as well as complemented the semi-dwarf phenotype of 5-wk-old ptox1-2 plants, while WT plants were unaffected. These results demonstrate that the mesocotyl elongation, tillering and semi-dwarf phenotypes of ptox1 are all due to SL deficiency.
The Arabidopsis im mutant also shows a SL-deficient phenotype
In order to assess whether the SL-deficient phenotype is common to PTOX mutants in other species, we analysed the phenotype of the Arabidopsis im mutant. Consistent with an SL-deficient phenotype in Arabidopsis (Stirnberg et al., 2002; Cazzonelli et al., 2009), the im mutant is also partially dwarf (Fig. S7a–e) with increased branching (Fig. S7f).
ptox1 is ABA-deficient, but accumulates auxin
We next undertook an analysis of endogenous concentrations of four plant hormones (Table S1) to gain a more complete understanding of the ptox1 phenotype. In addition to SL synthesis, β-carotene serves as a precursor in the biosynthesis of the hormone abscisic acid (ABA). In agreement with its low β-carotene contents (Fig. S6), the ptox1-2 mutant showed a significant reduction in endogenous ABA, ABA glucose ester (ABAGE) and phaseic acid (PA) contents compared to WT (Fig. 8a; Table S1). We therefore investigated the response of ptox1-2 to exogenous ABA application. Consistent with the role of ABA as a negative regulator of early growth in rice (Kim et al., 2012), ABA application reduced growth both in ptox1-2 and WT plants in a similar manner, but had no effect on tillering (Fig. 8b). This result confirms that the reduction in ABA content is not the cause of the ptox1 developmental phenotypes observed here. An Arabidopsis T-DNA insertion mutant lacking PDS3, the target of PTOX, has previously been described. This mutant, which is also deficient in β-carotene synthesis, is characterized by an albino phenotype and extreme dwarfism with small rosette leaves, and it was reported that the dwarf phenotype could be partially rescued by application of the gibberellin GA3 (Qin et al., 2007). We therefore tested whether GA deficiency could account for the phenotype of ptox1. However, the ptox1-2 mutant showed no inhibition of endogenous GA concentrations (Table S1) and GA3 application failed to rescue the semi-dwarf phenotype of ptox1 (Fig. S8). Another plant hormone that can regulate plant height is cytokinin, however, the ptox1 mutation also had no major effects on active cytokinins (Table S1).
Previous studies have shown the importance of the interaction between SL and auxin in the control of shoot branching, and increased auxin content has been reported in SL-deficient mutants (Arite et al., 2007). Consistent with this, quantification of endogenous auxin contents revealed a significant up-regulation of the biologically active indole acetic acid (IAA) in ptox1-2 compared to WT (Fig. 8c; Table S1).
TILLING and EcoTILLING for forward genetics
TILLING has rapidly become a potent reverse genetic tool with applications so far in a wide range of plant and animal species (for reviews, see Barkley & Wang, 2008; Bush & Krysan, 2010). Extending the approach to the detection of polymorphisms in natural populations (EcoTILLING) provided an improved single nucleotide polymorphism (SNP) detection system (Comai et al., 2004; Till et al., 2006a,b). The availability of whole genome sequences and dense marker systems means that map-based cloning (MBC) has become a straightforward procedure, at least in model systems such as Arabidopsis and rice, in which SSR markers with defined genomic positions are widely used. However, linkage analysis in some genomic regions is difficult either because there are no SSR markers or those available are not polymorphic between mapping parents. SNPs are considered the ultimate forms of molecular markers and have a major advantage of being the most abundant in the genome. The availability of high-throughput methods such as TILLING for the detection of SNPs has enabled their application in genetic mapping, diversity analysis and marker assisted selection. In principle, TILLING and EcoTILLING protocols can be applied as forward genetic tools to facilitate MBC (Rakshit et al., 2010). The SNPs detected by EcoTILLING between two mapping parents can be employed as markers for fine mapping using individual F2 plants that are still segregating for the flanking SSR markers. Once the candidate locus is delineated to a small chromosomal region by EcoTILLING, TILLING of the candidate region with overlapping primers using DNA pooled from WT and mutant plants in a 1 : 1 ratio provides an alternative to sequencing for identifying the causative mutation.
In the present study, we have successfully applied TILLING- and EcoTILLING-based strategies as forward genetic tools for fine mapping of the PTOX1 locus and for identification of the causal mutation of the Sas1248 mutant. Our approach combines the power of TILLING and the abundance of SNPs to substantially improve the efficiency of forward genetics. In our case, the use of SSR and EcoTILLING markers enabled the localization of the locus of interest to a region of c. 168 kb. TILLING of the candidate region with overlapping primers identified the candidate mutation, which was later confirmed by complementation tests and RNAi analysis. This step provides a cheaper alternative to sequencing for rapidly identifying the causative mutation. As more whole-genome sequences are made available, we believe that this strategy could easily be applied to any species and become a routine protocol in many laboratories.
The rice PTOX1 gene
We located the causative mutation of the Sas1248 mutant phenotypes in ORF Os04g0668900 on Chromosome 4, which we have named PTOX1 based on its strong similarity to PTOX genes in other plant species. A second rice gene exists with homology to PTOX1 on Chromosome 3 (Os03g0847500) (Fig. S4). Although this gene is predicted to encode a protein of 154 amino acids, it lacks many of the conserved residues/domains that are required for PTOX protein function and stability, and is therefore unlikely to be functional as a PTOX (Fu et al., 2005). Previously, a salt-responsive gene identified by differential display analysis in rice, OsIM1, that was mapped to a distance of 9.3 cM on Chromosome 3 between two restriction fragment length polymorphism (RLFP) markers G164 and RG756 was reported as a PTOX (Kong et al., 2003). Comparison of the DNA coding and predicted amino acid sequences of PTOX1 and IM1 (Accession No. AF085174) suggests that IM1 is the same as PTOX1 and that the chromosomal location reported by Kong et al. (2003) is incorrect. The proposed location of IM1 is within the genomic region where the second PTOX gene reported in the current study (Os03g0847500) resides, and we hypothesize that IM1 was incorrectly mapped to this region. Because designation of the gene as IM1 was based on the name of the Arabidopsis mutant immutans (Rédei, 1963), we propose that the name PTOX1 more closely reflects its function in rice.
Previously, the im (Carol et al., 1999; Wu et al., 1999) and ghost mutants (Carol & Kuntz, 2001; Barr et al., 2004) were used to clearly establish the role of PTOX in carotene synthesis. The rice ptox1 mutant reported here is deficient in β-carotene while accumulating phytoene, suggesting disruption in carotene synthesis (Fig. S6). The ptox1 mutant is also deficient in the carotene-derived hormones ABA (Fig. 8a) and the SL ent-2′-epi-5-deoxystrigol (Fig. 5), and thus the molecular consequences of PTOX disruption on carotenoid biosynthesis are conserved. To date, most of what we know about PTOX function comes from studies utilizing the Arabidopsis mutant im. The isolation of a ptox1 mutant in an important monocot crop species should be a valuable addition to the molecular genetic tools available for studying these important processes.
PTOX affects strigolactone-mediated growth responses in rice
The rice ptox1 mutants show a wide range of altered developmental responses in addition to the leaf variegation phenotype that has been extensively described for PTOX-deficient mutants in other species (Wetzel et al., 1994; Carol et al., 1999; Wu et al., 1999; Aluru et al., 2001; Shahbazi et al., 2007). These altered developmental responses include, most noticeably, semi-dwarfism and excessive tillering. All of these phenotypes were observed in both ptox1 mutant alleles as well as in PTOX1 RNAi lines, and the mutant phenotypes were fully rescued by expression of PTOX1. Shoot branching control is the most widely studied role of SLs in plants (Brewer et al., 2013), and the enhanced tillering of ptox1 is similar to phenotypes of rice mutants with altered SL responses (Arite et al., 2007; Umehara et al., 2008; Hu et al., 2010). Because SL synthesis is compromised in ptox1, we hypothesize that the enhanced tillering phenotype is the result of SL deficiency. In support of this hypothesis, application of the synthetic SL analog GR24 was able to fully rescue the tillering phenotype of ptox1 (Fig. 7). Moreover, another SL response, inhibition of mesocotyl elongation (Hu et al., 2010), was also lacking in ptox1, but rescued by GR24 (Fig. 6).
Although strigolactones are derived from carotenoids, a direct link between carotenoid deficiency and SL-mediated control of shoot branching is not well established, probably because of the severe effect of carotenoid deficiency on plant viability. However, the increased branching phenotype of the ramosus1 (rms1) and rms4 mutants of pea could be phenocopied by treatment with Norflurazon that inhibits PDS, an established target of PTOX-deficiency (Cazzonelli et al., 2009). Our observation that the Arabidopsis im mutant also shows an increased branching phenotype supports a role for PTOX in SL synthesis and is consistent with numerous reports on the role of carotenoid-derived SLs in shoot branching in Arabidopsis (Umehara et al., 2008; Leyser, 2009; Crawford et al., 2010). For the rice ptox1 mutant, we have shown that β-carotene synthesis is inhibited with a concomitant increase in phytoene accumulation and a reduction in SL production. This is consistent with the model shown in Fig. 9 in which reduced PTOX activity inhibits carotene synthesis and consequently SL synthesis leading to the developmental phenotypes observed. However, the reduction in β-carotene was only clearly observed in white leaf sectors, and ptox1 mutant plants that are entirely or predominantly green also show the tillering phenotype (Fig. S1). This suggests the possibility that PTOX may also have a more direct role in SL biosynthesis (Fig. 9). Interestingly, the ppd5 mutant of Arabidopsis with reduced NADPH dehydrogenase activity required to supply plastoquinol to the cytochrome b6f complex during cyclic electron transfer also shows SL-deficient phenotypes (Roose et al., 2011). In this case, loss of this PsbP domain protein appears to have little impact on carotenoid content, but leads instead to a strong lateral root branching phenotype that could be complemented by GR24, as well as increases branching in aerial tissues. Altered electron transport in ppd5 chloroplasts may therefore directly affect SL biosynthesis and it is possible that ptox mutations, although apparently reducing rather than oxidizing the plastoquinone pool, could somehow affect SL biosynthesis by a similar mechanism.
Another characteristic phenotypic trait of ptox1 is semi-dwarfism; a trait also observed clearly for the im mutant (Figs 1c,e, S7). Semi-dwarfism is a characteristic of SL-deficient mutants (Stirnberg et al., 2002; Arite et al., 2007, 2009; Cazzonelli et al., 2009), and our finding that application of GR24 can rescue the growth inhibition of ptox1 seedling supports the hypothesis that SL deficiency accounts for the semi-dwarfism of this mutant (Fig. 7). The Arabidopsis PDS loss-of-function mutant pds3 also showed a dwarf phenotype (Qin et al., 2007). In their study, Qin et al. proposed that changes in PDS activity led to changes in expression of GA biosynthesis genes that resulted in growth inhibition. They were also able to partially rescue growth by exogenous application of GA3 (Qin et al., 2007). However, their analysis was complicated by the very severe phenotype of the pds3 mutant and elongation growth was not explicitly tested. In our case, the ptox1 mutant is neither GA deficient (Table S1) nor shows an altered response to GA application (Fig. S8). Furthermore, although the ptox1 mutant is deficient in ABA, consistent with the role of PTOX in carotene synthesis, exogenous application of ABA could not complement the mutant developmental phenotypes (Fig. 8a,b). We therefore propose instead that SL deficiency is likely to be a significant contributor to the dwarf phenotype of both pds3 and im plants. The higher concentration of IAA detected in the ptox1-2 mutant is also in agreement with previous reports for SL-related rice mutants (Arite et al., 2007), and supports the importance of SL and auxin interaction in the control of shoot branching (Hayward et al., 2009; Crawford et al., 2010; Shinohara et al., 2013).
PTOX is involved in carotenoid biosynthesis at an early stage of leaf development and accordingly provides photo-protection to developing plastids (Aluru et al., 2006; Rosso et al., 2009; Putarjunan et al., 2013). Nevertheless, its physiological role in mature leaves, particularly in regulating photosynthesis, is not well established (McDonald et al., 2011; Trouillard et al., 2012). It has been suggested that PTOX plays a role in modulating the excitation pressure on photosystem II (PSII) through its involvement in an alternate electron pathway that mediates the electron flow from plastoquinone (PQ) to O2 to prevent the over-reduction of the PQ pool (McDonald et al., 2011). This has led to the hypothesis that PTOX can act as a safety valve under stress conditions, which has been demonstrated in a few studies (Ivanov et al., 2012; Laureau et al., 2013). However, the role of PTOX as a safety valve appears to happen only under limited conditions (Rosso et al., 2006; Okegawa et al., 2010; Trouillard et al., 2012; Laureau et al., 2013). In this study, PTOX deficiency clearly results in poor seed set and reduced aboveground biomass in the ptox1 mutant (Figs 1h, S9). While it is possible that some of the reduced final biomass in ptox mutants is due to the role of PTOX in protecting photosynthesis, we clearly show that the dwarf phenotype, at least in younger plants, is the result of SL deficiency. PTOX1 is expressed in all tissues and organs throughout development in rice (Fig. S10), as also observed in Arabidopsis (Aluru et al., 2001) and tomato (Barr et al., 2004). Interestingly, the highest expression level is found in the anther (Fig. S10), suggesting the possibility of a direct involvement of PTOX1 in regulating reproductive processes in rice.
It was noticeable that the SL-related phenotypes of tillering and semi-dwarfism reported in this study were relatively homogenous (Figs 1d,e, S9c,d) while a high degree of variation was observed for the variegated phenotype (Fig. S1). In fact, we have repeatedly tested the performance of the ptox1 mutant under low and high light conditions, and although very few variegated leaves were observed under low light, there was still variation among mutant plants with respect to their degree of variegation irrespective of the light intensity under which they were grown. By contrast, mutant plants showed the excessive tillering and semi-dwarf phenotypes under all light conditions. How can we explain this observation? PTOX1 is expressed in all tissues and this could certainly account for the strong effect on SL biosynthesis in the root as well as the seedling morphology phenotypes. It is likely that these effects are essentially independent of the light environment. By contrast, the role of PTOX in phytoene desaturation in leaf tissue results in a light-dependent bleaching phenotype with high light conditions generally enhancing white sector formation due to deficiency of the photoprotective carotenoids (Wetzel et al., 1994; Baerr et al., 2005; Shahbazi et al., 2007). The photobleaching phenotype will depend on the balance of chlorophyll synthesis and degradation and will be observed as an all-or-none phenomenon when the loss of chlorophyll due to photo-oxidative destruction exceeds new synthesis. Thus, this aspect of the phenotype is closely linked to the environmental conditions under which the plants are grown. However, it should also be noted that the apparent impact of the ptox1 mutation on carotene (Fig. S6) and ABA (Fig. 8a) synthesis is less severe than the impact on SL synthesis (Fig. 5). This could also account for respective differences in the observed phenotypes and would also be consistent with an additional independent role for PTOX in SL biosynthesis as discussed above (Fig. 9).
Mutants lacking functional PTOX proteins are among the best-characterized chloroplast mutants with studies on im, in particular, contributing enormously to our understanding of redox regulation within the chloroplast (McDonald et al., 2011), chloroplast biogenesis (Aluru et al., 2006), chloroplast regulation of nuclear gene expression (Aluru et al., 2009) and stress signaling (Sun & Wen, 2011). The isolation of a rice ptox1 mutant will enable these studies to continue in an important model crop species. However, our results showing that many of the morphological phenotypes of PTOX-deficient mutants are attributable to SL deficiency demonstrate that care must be taken in interpreting ptox mutant phenotypes, especially as SLs can induce the expression of light-harvesting genes (Mayzlish-Gati et al., 2010). In addition to SLs and ABA, the recent observation that carotenoid-oxidation products are involved in chloroplast-mediated stress signaling (Ramel et al., 2012) further illustrates the need to obtain a thorough understanding of the consequences of carotenoid deficiency.
We thank Prof. Marcel Kuntz (Centre National de la Recherche Scientifique and Universite′ Joseph Fourier, Grenoble, France) for the kind gift of Arabidopsis im seeds, Dr Shinjiro Yamaguchi (RIKEN Plant Science Center, Yokohama, Japan) for GR24, and Prof. Koichi Yoneyama (Weed Science Center, Utsunomiya University, Japan) who kindly provided the [2H6]-ent-2′-epi-5-deoxystrigol internal standard. The Tos17 insertion line (FT930204) and the full-length cDNA clone (AK067891) of PTOX1 were obtained from the National Institute of Agrobiological Sciences in Japan. This work was partly supported by the Ministry of Agriculture, Forestry, and Fisheries of Japan (Genomics for Agricultural Innovation PMI-0010), by the Program for Promotion of Basic and Applied Researches for Innovations in Bio-oriented Industry, Japan, by Grant-in-aid for Scientific Research from the Ministry of Education, Cultures, Sports and Technology, Japan (Grant-in-Aid for Scientific Research on Innovative Areas 23113009), and by JSPS KAKENHI (grant no. 24248004) to R.T. T.Y. is funded by the MEXT-supported Program for the Strategic Research Foundation at Private Universities, 2008–2012.