Functional inactivation of UDP-N-acetylglucosamine pyrophosphorylase 1 (UAP1) induces early leaf senescence and defence responses in rice

Highlight This study identified the novel gene UAP1 in rice. UAP1 is involved in early leaf senescence and defence responses.


Introduction
Leaf senescence is a complex physiological event that constitutes the last stage of leaf development, integrating multiple developmental and environmental signals (Lim and Nam, 2005;Balazadeh et al., 2011). Although it has been investigated many times, the mechanisms involved are still not well understood. Early leaf senescence has been described as one of the characteristics of lesion mimic mutants (LMMs) (Qiao et al., 2010;Jiao et al., 2012). Leaf senescence in LMMs usually happens rapidly, providing a unique tool for dissecting possible regulatory genes or pathways during the progression of leaf growth and development to senescence. For example, RLS1, encoding an uncharacterized nucleotide-binding-site-containing protein, was identified and shown to be involved in autophagymediated chloroplast degradation during leaf senescence in rice (Jiao et al., 2012). Characterization of more LMMs will help to enrich our understanding of leaf senescence.
Defence responses are important for plants in combating pathogen invasion (Dangl and Jones, 2001). Both resistance genes and many regulatory genes are involved in defence pathways. LMMs develop spontaneous lesions that resemble disease symptoms in the absence of pathogen attack, and interestingly these plants usually have a significantly enhanced resistance to disease (Jung et al., 2005;Wu et al., 2008). It is believed that cloning the genes responsible for LMMs is the key to deciphering the defence pathways and may aid the development of broad-spectrum pathogen resistance in plants.
UDP-N-acetylglucosamine pyrophosphorylase (UAP), also termed UTP:N-acetylglucosamine-1-P uridylyltransferase or N-acetylglucosamine-1-phosphate uridylyltransferase, is widely distributed in living organisms. Researchers have found that both of the human UAP1 isoforms, UAPA and UAPB (also termed AGX1 and AGX2, respectively), can use N-acetylglucosamine-1-phosphate (GlcNAc-1-P) or N-acetylgalactosamine-1-phosphate (GalNAc-1-P) as substrates for producing UDP-N-acetylglucosamine (UDP-GlcNAc) or UDP-N-acetylgalactosamine (UDP-GalNAc) (Peneff et al., 2001). Enzymatic experiments with UAPs from Arabidopsis thaliana also showed that AtUAP1 and AtUAP2 (also termed GlcNAc1pUT-1 and GlcNAc1pUT-2, respectively) catalyse the reversible reactions (Yang et al., 2010): UTP GlcNAc 1 P GalNAc 1 P UDP GlcNAc UDP GalNAc PPi UAP is essential for organisms to grow normally. Some UAP mutants have been described. In Escherichia coli, GlmU has UAP activity, and inactivation of the GlmU gene rapidly reduces peptidoglycan synthesis, leading to alteration in cell shape and cell lysis (Mengin-Lecreulx and van Heijenoort, 1993). In yeast (Saccharomyces cerevisiae), the cells of the ScUAP null mutant displayed aberrant morphology in which most cells were fully swollen and some were lysed (Mio et al., 1998). In Drosophila melanogaster, mutants of DmUAP (also termed mummy) exhibit defects in dorsal closure, central nervous system fasciculation, and eye development (Schimmelpfeng et al., 2006). However, no known phenotype caused by UAP gene mutations has been reported in plants.
In addition, the gene encoding UAP has not been identified in rice.
In order to further elucidate the molecular mechanisms of leaf senescence and defence responses in rice, a new LMM termed spotted leaf 29 (spl29) was isolated. This exhibits spotted leaves and rapid leaf senescence from the seedling stage throughout the rest of its life cycle. The SPL29 gene was identified using a map-based cloning strategy. It was confirmed that SPL29 is UAP1, with UAP activity, and that there is no enzymatic activity in the spl29 mutant. Both early leaf senescence and defence responses were investigated and confirmed in spl29 plants. Our data suggest that UAP1 is probably involved in regulating leaf senescence and defence responses in rice.

Plant material and growth conditions
The LMM mutant line spl29 was isolated in 2008 from regenerated plants derived from a tissue culture of the rice cultivar Zhonghua 11 (ZH11, Oryza sativa spp. japonica). Its mutant phenotype, lesion mimic leaf spots and rapid leaf senescence, was stably inherited over multiple generations. M6 generation seeds of spl29 were used in this study. Germinated wild-type and spl29 seeds were grown in soil in a plant growth chamber (light cycle: 14 h light/10 h dark, 28/26°C) for use at the seedling stage, when the second-emerging leaves were used 28 days after germination. For plants used from the tillering to ripening stages, germinated wild-type and spl29 seeds were grown in an experimental field under natural summer conditions. Leaf samples used at the tillering stage were the third leaves from the top of the main tiller of plants 50 days after germination.

Genetic analysis and map-based cloning
For genetic analysis, the leaf phenotypes of F 1 and F 2 plants were observed from hybridizations of Guangzhan63s, 10N056, and Yuehui9113 with spl29, respectively. The F 2 population from the cross between Guangzhan63s and spl29 was used for preliminary and fine mapping of the SPL29 locus. Using 44 mutant plants obtained in the F 2 population, preliminary mapping was performed with simple sequence repeat markers, which were well distributed across all 12 chromosomes and allowed Guangzhan63s and spl29 to be distinguished. The data for preliminary mapping were analysed using Mapmaker 3.0 software. A total of 870 F 2 mutant plants were used for the fine mapping of SPL29. Nine markers (S1, S6, S8, S15, S19, S26, S32, S33, and S40) were developed for fine mapping based on DNA sequence differences between indica and japonica rice varieties. The molecular marker primer sequences are listed in Supplementary Table S1.

Complementation of SPL29
The G-to-T mutation was identified by PCR amplification and sequencing using the primer pair SPL29Mutation (Supplementary  Table S2). For complementation of the mutant phenotype, three DNA fragments of LOC_Os08g10600, the 2214-bp upstream promoter, 4674-bp gene region, and 1000-bp downstream terminator, were first amplified from the genomic DNA of wild-type Zh11 using primer pairs SPL29Pro, SPL29Gene, and SPL29Ter (Supplementary  Table S2), respectively; and then constructed into the binary vector pBWA(V)B (reconstructed from pCAMBIA3300) using SapI sites for the digestion-link, one-step reaction. The resultant seamlessly cloned 7888-bp DNA fragment was sequenced to confirm that it was identical to the wild-type sequence in ZH11. This vector constructed for functional complementation was termed pSPL29C, and the empty pBWA(V)B was renamed pEmvC in this study. Vectors pSPL29C and pEmvC were introduced into Agrobacterium tumefaciens EHA105, and then transformed into calli of the spl29 mutant. Positive transgenic plants were confirmed using Bar178 primers (Supplementary Table S2) specific for the amplification of the phosphinothricin screening gene.

Bioinformatic analysis and construction of the phylogenetic tree
The gene sequence and structure of SPL29 were used to search the Rice Genome Annotation Project (RGAP; http://rice.plantbiology.msu.edu/). A gene family search was conducted using Pfam (http://pfam.sanger.ac.uk/). To identify the specific UDP-Nacetylglucosamine pyrophosphorylase encoded by SPL29, BLASTP searches were performed with the full-length amino acid sequence on NCBI (http://www.ncbi.nlm.nih.gov/). Sequence identities were also analysed by BLASTP.
Multiple full-length amino acid sequences were aligned using ClustalW prior to phylogenetic analysis. MEGA 5 was utilised to construct the consensus phylogenetic tree with 1000 random bootstrap replicates.

Recombinant protein construction, expression, and purification
To generate the glutathione S-transferase (GST) gene fusion constructs GST-SPL29/spl29, the 1470-bp full-length coding sequence (CDS) of SPL29 or spl29 was amplified from the cDNA of ZH11 or spl29 leaves, respectively (primers GST-SPL29/spl29 CDS in Supplementary Table S2). PCR products were inserted into pGEX-6P-1 using the restriction enzyme sites BamHI and EcoRI. The recombinant vectors were transferred into E. coli DH5α and sequenced to check for correct construction. The empty vector pGEX-6P-1 and the recombinant vectors were then transferred into E. coli BL21 for protein expression. Strains harbouring GST, GST-SPL29, or GST-spl29 were cultured in Luria-Bertani medium at 37°C. At an optical density at 600 nm of 0.5, isopropylthiogalactopyranoside was added to a final concentration of 0.5 mM. Strains were then cultured at 18°C for 20 h (200 rpm) to induce gene expression. The cells were centrifuged and resuspended in PBS buffer (137 mM NaCl, 2.7 mM KCl, 10 mM Na 2 HPO 4 , and 2 mM KH 2 PO 4 ; pH 7.4). After sonication, the lysates were centrifuged. Protein purification was conducted using a column with Glutathione Sepharose TM 4 Fast Flow (GE Healthcare) according to the manufacturer's instructions. Proteins were detected by SDS-PAGE analysis and the concentration of purified protein was determined using bovine serum albumin as a standard.

H-nuclear magnetic resonance ( H-NMR) analysis of SPL29/ spl29 activities in situ
The examination of enzymatic activities was performed as described previously (Yang et al., 2010) with some modifications. The forward reactions were carried out in a final volume of 540 μl, and the mixture consisted of 2 H 2 O/H 2 O at a ratio of 8:1 (v/v), Na + /K + phosphate buffer (80 mM K 2 HPO 4 , 20 mM NaH 2 PO 4 ; pH 7.4), 5 mM MgCl 2 , 0.2 mM UTP, 0.2 mM GlcNAc-1-P, 1.5 units of yeast inorganic pyrophosphatase, and recombinant enzyme (1.86 μg of GST, GST-SPL29, or GST-spl29). The reverse reactions were carried out in a similar manner, in a solution containing 2 H 2 O/H 2 O at a ratio of 8:1, Na + /K + phosphate buffer, 5 mM MgCl 2 , 0.2 mM PPi, 0.2 mM UDP-GlcNAc (or UDP-GalNAc), and recombinant enzyme (1.86 μg of GST, GST-SPL29, or GST-spl29). Examination of GlcNAc-1-P/GalNAc-1-P and UDP-GlcNAc/UDP-GalNAc was performed by 1 H-NMR as described previously (Zhang et al., 2011). Data acquisition started at 10 min and 60 min after the addition of enzyme to the reaction mixture.

Gene expression analysis
Samples were collected and immediately frozen in liquid nitrogen, then stored at -80°C. Total RNA was extracted using TRIzol reagent (Invitrogen). After RNase-free DNase treatment, 5 μg of RNA was used for cDNA synthesis using M-MLV reverse transcriptase (Promega) in a 50-μl reaction mixture. The quantitative, real-time reverse transcription polymerase chain reaction (qRT-PCR) technique was performed using 2×SYBR Green Master Mix reagent (Bio-Rad) in a 96-well plate using a Bio-Rad CFX96 real-time PCR system. Three technological replicates were used for each biological sample. Six rice reference genes (from our unpublished data on the selection of stable reference genes to normalise gene expression in rice) were assessed for their potential as stable internal standards by geNorm as previously described (Vandesompele et al., 2002). These genes were TI (LOC_Os01g05490), ARF (LOC_Os05g41060), EF-1α (LOC_Os03g08020), UBC (LOC_Os02g42314), Profilin-2 (LOC_Os06g05880), and Actin1 (LOC_Os03g50885). As a result of this analysis, UBC, Profilin-2, and Actin1 were selected as internal standards for all leaf samples (Supplementary Figure S1). All primers used for qRT-PCR analysis are listed in Supplementary Table  S3, with good PCR efficiencies (85-105%) assessed using a 10-fold dilution series of total cDNA.
Chlorophyll content and chlorophyll a fluorescence Chlorophyll was extracted from leaves with ice-cold 80% acetone, and the chlorophyll content per gram of leaf fresh weight (FW) was determined as previously described (Lichtenthaler, 1987).
Chlorophyll a fluorescence transients from dark-adapted plants were measured using a Handy Plant Efficiency Analyser (Handy PEA, Hansatech, UK). The description and calculation formulae of parameters are listed in Supplementary Table S4.

Transmission electron microscopy
Leaf segments were fixed with 2.5% glutaraldehyde in sodium phosphate buffer (pH 7.2) for 4 h at 4°C. Ultrathin samples were made as previously described , and viewed using a transmission electron microscope.

Inoculations with bacterial blight pathogen
Xanthomonas oryzae pv. oryzae strain PXO99 (Philippine race 6) was cultured on pressure-sensitive adhesive medium. The second and third youngest fully expanded leaves from both the main and lateral tillers of 5-7 independent wild-type and spl29 plants were inoculated with PXO99 suspensions (optical density of 0.5 at 600 nm) using the leaf clipping method (Manosalva et al., 2011). Lesion lengths on inoculated plants were measured 12 days after inoculation.

Histochemical staining
Leaves in wild-type and spl29 plants were sampled at the tillering stage (about 45 days after germination under natural summer field conditions). Histochemical assays for determining reactive oxygen species (ROS) accumulation were conducted as previously described (Qiao et al., 2010). Three independent experiments were performed.

Lipid peroxidation and ROS-scavenging enzyme assays
Plant samples were ground in liquid nitrogen. Crude protein extraction, measurement of malondialdehyde (MDA) contents, and assays for the activities of superoxide dismutase (SOD) and catalase (CAT) were conducted according to previously described methods .

Quantification of jasmonic acid and abscisic acid
Plant leaves were collected, weighed, immediately frozen in liquid nitrogen, and stored at -80°C. Extraction and measurement of jasmonic acid (JA) and abscisic acid (ABA) were conducted following a previously described method (Chen et al., 2012).

Characterization of the spl29 mutant
Small, dark-brown LM leaf spots (followed by rapid leaf senescence) appeared on spl29 plants from the seedling stage to the ripening stage ( Fig. 1A, B, and C). However, newly emerging leaves showed no difference to wild-type plants. During the seedling stage, abundant leaf spots had appeared on spl29 plants by 28 days after germination, with leaves exhibiting chlorosis beside the spots (Fig. 1D). After the appearance of leaf spots and chlorosis, the leaves withered from the tip to the base in about five days (Supplementary Figure S2). The severity of leaf mutant phenotypes varied for different leaves of a single spl29 plant at the tillering stage 50 days after germination (Fig. 1E). No spots were present on the first new fully expanded leaf, whereas the third leaf from the top of the main tiller had larger and more numerous spots and associated chlorosis than the second leaf, and at this point the fourth leaf was already withered and dead. By contrast, the corresponding leaves on the wild-type plant grew normally. When the last emerging flag leaf was filled with spots ( Fig. 1F), the spl29 plants began to senesce far earlier than wild-type plants. In addition to the obvious leaf phenotype, the plant height, panicle length, total grain number, filled grain number, seed-setting rate, and thousand-grain weight of the spl29 mutants were all significantly decreased when compared with the wild-type plants (Table 1).

Map-based cloning of the SPL29 gene
For genetic analysis, three indica cultivars, Guangzhan63s, 10N056, and Yuehui9113, were selected for hybridisation with spl29. All F 1 plants displayed the wild-type leaf phenotype. In all three F 2 populations, the wild-type and LM mutant phenotypes segregated at a ratio of 3:1 (χ 2 < χ 2 0.05 = 3.84; P > 0.05; see Supplementary Table S5), which suggested that the phenotype of spl29 was controlled by a single recessive nuclear gene.
The F 2 population from the cross between Guangzhan63s and spl29 was used to map the SPL29 gene. Using 44 F 2 mutant plants, the SPL29 gene was closely linked with marker M1077, and mapped inside markers M1037 and M1230 on chromosome 8, with an equal genetic distance of 1.4 cM to each ( Fig. 2A). For fine mapping of the SPL29 gene, more than 3000 F 2 individuals were generated and new markers between M1037 and M1230 were designed according to sequence differences between indica and japonica rice. Nine markers showing polymorphisms between Guangzhan63s and spl29 (Supplementary Table S1) were used to screen the recombinants. Using 870 F 2 mutant plants, the SPL29 gene was eventually limited to a 97-kb region between the new   markers S8 and S26, with one recombinant for each marker (Fig. 2B). Ten putative open reading frames (ORFs) were predicted according to the RGAP website (Fig. 2C). All ten genes were amplified and sequenced. By comparing gene sequences between wild-type and spl29 plants, one point mutation was identified on the fifth gene, LOC_Os08g10600. No DNA sequence changes were found in the other nine genes or the putative promoter region (about 2.2 kb) of LOC_Os08g10600. The single nucleotide substitution of guanine (G) to thymine (T) occurred in the eighth exon of LOC_Os08g10600 (at a position of 712 bp in the CDS), resulting in a single amino acid change from glycine (Gly) to cysteine (Cys) (Fig. 2D). For preliminary confirmation that the G-to-T mutation was responsible for the phenotype of the spl29 mutant, this nucleotide site was examined in different rice cultivars, F 2 non-lesion-mimic (NLM) plants, and F 2 LM plants (Supplementary Table S6). The nucleotide at this site was T for spl29 and all tested F 2 LM plants, but G for 10 different normal rice cultivars. Four of the F 2 NLM plants had a G at this position, while G and T existed together at this site in seven of the F 2 NLM plants. Thus, LOC_Os08g10600 was a likely SPL29 candidate.

Functional complementation with LOC_Os08g10600 in the spl29 mutant
A 7888-bp genomic fragment of wild-type Zh11, containing the 2214-bp upstream promoter, the 4674-bp gene region, and the 1000-bp downstream terminator of LOC_ Os08g10600, was constructed using the plasmid pSPL29C and transferred into the spl29 calli by Agrobacterium tumefaciens-mediated transformation. The empty vector pEmvC was introduced to spl29 calli as the control (Fig. 2E). About 40 days after plant regeneration, all 22 independent transgenic lines containing the wild-type SPL29 gene showed a complete rescue of the mutant phenotype (Fig. 2F), while all 16 independent transgenic lines with the empty vector failed to complement the spl29 mutant (Fig. 2G). Leaf phenotypes of the transgenic plants with pSPL29C or pEmvC are shown in Fig. 2H. Successfully transformed plants were confirmed by PCR amplification of the screening marker Bar gene (Fig. 2I). Sequencing the G-to-T mutation site of the SPL29 gene revealed that plants transformed with pSPL29C had both G and T at the mutant nucleotide site, whereas the pEmvC control plants only had the mutant T nucleotide (Fig. 2J). Therefore, functional complementation with the wild-type candidate gene rescued the mutant phenotype of spl29 plants, and it was concluded that LOC_Os08g10600 was the SPL29 gene.

SPL29 encodes a putative UAP1
As annotated on RGAP, LOC_Os08g10600 (SPL29) encodes a putative 489-amino acid protein. Pfam analysis with its predicted amino acid sequence showed that SPL29 belongs to the UDP-glucose pyrophosphorylase gene family (PF01704). NCBI BLASTP revealed that SPL29 is a putative UAP1 gene in rice (OsUAP1), but not the UDP-glucose pyrophosphorylase (UGP) gene. In the rice genome, one homologue of SPL29 is also annotated as UAP (LOC_Os04g52370, termed OsUAP2). To examine the evolutionary relationships between SPL29 and its homologues in different species, a bootstrap consensus phylogenetic tree was constructed by the neighbour-joining method (Fig. 3). All UAPs in different eukaryotic organisms clustered separately from the UGPs. The protein EcGlmU, which shows UAP activity in E. coli, showed no amino acid homology with SPL29, and was thus located furthest away. UAPs in plants were tightly clustered, with separate groups for monocots and dicots, indicating that gene divergence occurred after diversification of these two clades. In addition, SPL29 showed over 30% amino acid identity with all eukaryotic UAPs analysed by NCBI BLASTP (Fig. 3). These results all suggest that SPL29 encodes a putative UAP1 in rice.
SPL29 shows UAP enzymatic activities but spl29 eliminates these enzymatic functions UAP has not been identified in rice before. In order to reveal whether SPL29 and the mutant spl29 have UAP activity, recombinant proteins of GST-SPL29 and GST-spl29 were produced. Theoretically, the molecular weights are 26 kDa for GST, 54.071 kDa for SPL29, and 54.117 kDa for spl29. GST, GST-SPL29 (about 80 kDa), and GST-spl29 were highly expressed after induction (Supplementary Figure S3, lanes  2-4). Column-purified proteins were used to detect the enzymatic reactions (Supplementary Figure S3, lanes 5-7).
These NMR-based assays provide unambiguous evidence that SPL29 has UAP activity whereas the mutation of SPL29 to spl29 eliminates this enzymatic function. To date, UAP has not been well studied in plants. The spl29 mutant can be used to reveal the biological significance of UAP.

Early leaf senescence in the spl29 mutant is identified by physiological indicators
The decrease of chlorophyll content is commonly used as a physiological indicator related to plant leaf senescence (Jiao et al., 2012). At the seedling stage, chlorophyll content decreased from 1240 μg g -l FW in wild-type leaves to 861 μg g -1 FW in spl29 leaves (Fig. 5A). A similar reduction was also observed at the tillering stage, when chlorophyll contents were 1362 μg g -1 FW in the wild type and 786 μg g -1 FW in spl29 (Fig. 5B). The senescence-induced STAY GREEN (SGR) gene plays an important role in regulating chlorophyll degradation (Park et al., 2007). Analysis by qRT-PCR showed that the expression of SGR was greatly upregulated in spl29 leaves at the seedling and tillering stages (Fig. 5C-D).
Photosystem II (PSII) function is usually reduced during senescence. The chlorophyll a fluorescence transient can be examined to reveal the activities of PSII machinery on light absorption, energy transformation, and electron transfer, thus this technique is commonly used to monitor leaf senescence (Zhang et al., 2012;Panda and Sarkar, 2013). Wildtype and spl29 plants at the seedling stage were analysed. In both lines, chlorophyll a fluorescence transients showed a typical polyphasic rise with the basic steps O-J-I-P, but the fluorescence intensity was reduced in spl29 plants (Fig. 5E). Greater amplitude of ΔW, increased K (300 μs), J (2 ms), and I (30 ms) steps, positive ΔL band , and the maximum amplitude of the IP phase implied PSII reduction (Supplementary Figure S4 However, increased capacities per RC did not rescue the capacity of the PSII machinery in spl29, which was completely reduced as clearly reflected by performance indices (PI ABS , PI CSo , PI CSm , PI tol,ABS , PI tol,CSo , and PI tol,CSm ). These results demonstrated that the PSII units were damaged and could not function properly in spl29 leaves.
The decreased chlorophyll content and reduced PSII capacity indicated that the spl29 plants had entered senescence at the physiological level.

Irreversible degradation of chloroplasts in earlysenescent leaves of spl29 mutants
The main cellular characteristic of leaf senescence is the occurrence of chloroplast degradation (Lim et al., 2007;Jiao et al., 2012). To investigate whether the chloroplasts were affected in the early-senescing leaf of the spl29 mutant, the ultrastructures of wild-type and spl29 cells in 28-day-old seedling leaves were compared using transmission electron microscopy. In wild-type leaves, well-developed mesophyll cells were observed with fully developed and membraneintact chloroplasts (Fig. 6A). However, three different chloroplast states were observed among spl29 mesophyll cells. Fully developed and membrane-intact chloroplasts could be observed in many spl29 mesophyll cells (Fig. 6B), while in some cells the chloroplast envelope was observed to be breaking ( Fig. 6C and Supplementary Figure S5A and B). Unsurprisingly, mesophyll cells which had lost chloroplasts but retained mitochondria were present abundantly in spl29 (Fig. 6D and Supplementary Figure S5C and D). These results suggest that chloroplasts developed normally in spl29 at first, but were completely degraded during early leaf senescence. Chloroplast breakage may be one mechanism causing this degradation. Drastic, irreversible, and complete degradation of chloroplasts in spl29 leaves is coincident with the phenotype of rapid early leaf senescence (leading to death).

Upregulated expression levels of senescenceassociated transcription factors and senescenceassociated genes, and downregulated expression levels of photosynthesis-related genes, are molecular evidence of early leaf senescence in the spl29 mutant
Leaf senescence is a complex process controlled by a large number of different genes. Many genes that encode transcription factors are upregulated to induce leaf senescence . To further confirm that senescence occurred in the spl29 plants, gene expression analysis of three senescence-associated transcription factors (OsWRKY23, OsWRKY72, and OsNAC2) was performed by qRT-PCR. At the seedling stage, transcripts of OsWRKY23, OsWRKY72, and OsNAC2 in spl29 leaves increased 40.7-, 8.5-, and 2.1-fold, respectively, in comparison to those of wild-type leaves (Fig. 7A); at the tillering stage their levels in spl29 leaves were increased 110.0-, 12.0-, and 1.5-fold, respectively, compared to the wild type (Fig. 7B). The expression levels of the senescence-associated transcription factors investigated here were all upregulated, according with the early leaf senescence of the spl29 mutant.

Plant resistance to the bacterial blight pathogen is enhanced and expression levels of defence signallingrelated genes are induced in the spl29 mutant
The appearance of LM spots in spl29 plants resembles the hypersensitive response (HR), an important resistance mechanism in plants (Dangl and Jones, 2001;Lam et al., 2001). To test their resistance to pathogens, spl29 and wild-type plants were inoculated with the bacterial blight strain PXO99 at the tillering stage. The wild-type plants showed a typical response to bacterial blight disease, while the spl29 plants exhibited significantly enhanced resistance. The average leaf lesion lengths following PXO99 inoculation were 9.87 cm in wildtype plants and 1.42 cm in spl29 plants (Fig. 8A).
Activation of defence-response gene expression has previously been observed during the development of lesions in some rice LMMs (Yin et al., 2000;Manosalva et al., 2011). Therefore, we examined the expression of four defence signalling-related genes (PR1a, PBZ1, PO-C1, and OsWRKY45) in wild-type and spl29 plants by qRT-PCR. At the seedling stage, levels of PR1a, PBZ1, PO-C1, and OsWRKY45 mRNAs in the spl29 mutant were 1627.5-, 96.2-, 419.5-, and 21.1-fold, respectively, the amount of those in wild-type plants (Fig. 8B); at the tillering stage these genes were expressed 71.9-, 21.2-, 11.6-, and 2.6-fold, respectively, as much in spl29 as they were in wild-type plants (Fig. 8C). The upregulation of defence signalling-related genes indicates that the plant defence responses are induced, which may contribute to enhancing disease resistance in the spl29 mutant.

ROS and MDA accumulate in the spl29 mutant accompanied by increased SOD activity and normal CAT activity
ROS generation is one of the earliest responses of plant cells during senescence (Khanna-Chopra, 2012). ROS also play  Table S4). (E-F) Each data point is the mean value of ten independent plants.
important roles in plant-pathogen interactions (Wojtaszek, 1997;Apel and Hirt, 2004). This prompted us to investigate whether ROS accumulation occurred in spl29 plants. Leaves at the tillering stage were used for the ROS assay. The pattern of nitro blue tetrazolium (NBT) staining reflected the formation of blue formazan precipitates and was indicative of O 2 accumulation. There was extensive NBT staining in the leaf area surrounding lesions in spl29 plants, whereas staining was minimal in wild-type leaves (Fig. 9A). When 3,3'-diaminobenzidine (DAB) staining was used as an indicator of H 2 O 2 accumulation, intense brown staining was seen correlating with lesion formation in spl29 mutant leaves, but no such signal was detected in wild-type leaves (Fig. 9B). These results demonstrated that ROS accumulated in spl29 plants.
Lipid peroxidation by ROS was analysed by measuring contents of MDA, an end-product of oxidized lipids. At the seedling stage, the MDA content of 103 nmol g -1 FW in spl29 leaves was more than double the 49 nmol g -1 FW in wild-type leaves (Fig. 9C). At the tillering stage, an MDA content of 378 nmol g -1 FW was generated in the leaves of spl29 mutant plants, much higher than the 209 nmol g -1 FW in wild-type leaves (Fig. 9D). Increased MDA contents suggested that lipid peroxidation occurred in spl29 plants, which provided further evidence of the ROS accumulation and membrane damage in spl29 plants.
During oxidative stress, plants synthesize anti-oxidative enzymes, such as SOD and CAT, to remove ROS (Miller et al., 2010). SOD catalyses the dismutation of O 2 to produce H 2 O 2 , and CAT is the major H 2 O 2 -scavenging enzyme. Thus SOD and CAT activities were examined to explore the ROS metabolic process in spl29 plants. At the seedling stage, SOD activity was 8926 U g -1 FW in spl29 leaves, which was much higher than the 3208 U g -1 FW in the wild-type leaves (Fig. 9E). A similar result was observed in leaves at the tillering stage, with a SOD activity of 12260 U g -1 FW in spl29 leaves compared with 2167 U g -1 FW in the wild type (Fig. 9F). However, there was no obvious difference in CAT activity between spl29 and wild-type leaves at both the seedling stage and the tillering stage (Fig. 9G-H). The increase of SOD activity suggests that the spl29 mutant may actively respond to the O 2 accumulation and produce more H 2 O 2 , while the normal CAT activity may not be enough to scavenge the additional H 2 O 2 , leading to its accumulation in spl29.

JA and ABA accumulate in the spl29 mutant
Plant hormones, such as JA and ABA, are reported to play roles in both leaf senescence and defence responses (Lim et al., 2007;Robert-Seilaniantz et al., 2011). This prompted us to investigate the contents of JA and ABA in spl29. At the seedling stage, the content of JA was 110.6 ng g -1 FW in spl29, whereas JA in the wild type was hardly detectable, with a content of 0.3 ng g -1 FW (Fig. 10A). Similar results were also found at the tillering stage, with a JA content of 53.5 ng g -1 FW in spl29, but 0.2 ng g -1 FW in the wild type (Fig. 10B). The content of ABA at the seedling stage was 49.6 ng g -1 FW in spl29, compared with 8.0 ng g -1 FW in the wild type ( Fig. 10C). At the tillering stage, the content of ABA was 34.7 ng g -1 FW in spl29, higher than the 11.0 ng g -1 FW in the wild type (Fig. 10D). These results show that JA and ABA accumulate in spl29 plants, probably playing a role in early leaf senescence and defence responses in this mutant.

SPL29 is the UAP1 gene in rice and functional inactivation of OsUAP1 is responsible for the mutant phenotypes
From the seedling stage, the spl29 mutant showed LM spots on every leaf. The leaf then exhibited chlorosis and soon withered ( Fig. 1 and Supplementary Figure S2). It was decided that examining spl29 might show a novel gene involved in early leaf senescence and defence responses. The SPL29 gene was identified as LOC_Os08g10600 by map-based cloning (Fig. 2), and an NCBI BLASTP search revealed it to be a putative UAP1 gene. Phylogenetic analysis showed that SPL29 belongs to the cluster of UAPs, which are separate from the cluster of UGPs (Fig. 3). SPL29 was confirmed as UAP1 in rice, able to perform UAP enzymatic activity (Fig. 4). The mutant phenotype of spl29 results from its UAP1 functional inactivation (Fig. 4). UAP is an important enzyme, widely distributed in living organisms; UAP mutants have been discovered in E. coli, S. cerevisiae, and D. melanogaster, and all show abnormal development (Mengin-Lecreulx and van Heijenoort, 1993;Mio et al., 1998;Schimmelpfeng et al., 2006). Until now, there has been no report of UAP mutants in plants, nor has the UAP gene been identified in rice. This study identified the UAP1 gene in rice and showed that the functional inactivation of OsUAP1 induces early leaf senescence and defence responses.

Early leaf senescence is induced by functional inactivation of OsUAP1
Only a few studies describe senescence as a phenotype of spl mutant plants, and the senescence syndrome has not been fully analysed in these mutants (Qiao et al., 2010). Additionally, senescence has not been reported in UAP mutants in other species. Therefore, it was necessary to determine that early leaf senescence does indeed occur in spl29 plants.
The major physiological change during leaf senescence is the gradual loss of chlorophyll (Jiao et al., 2012;Zhang and Zhou, 2013). After the mutant phenotypes appeared, chlorophyll contents were indeed decreased in spl29 (Fig. 5A-B).
Senescence-induced SGR expression in spl29 ( Fig. 5C and D) may play an important role in regulating chlorophyll degradation (Park et al., 2007). Reduced PSII capacity is another manifestation of leaf senescence at the physiological level (Zhang et al., 2012;Panda and Sarkar, 2013). The PSII machinery in spl29 was degraded and its function was fully depressed (Fig. 5E and F; Supplementary Figure S4). Surprisingly, spl29 had increased capacity in each RC (ABS/ RC, TR o /RC, ET o /RC, and DI o /RC), although this was previously seen in the senescence-induced PSII alterations of Cucumis sativus cotyledons (Prakash et al., 2003).
The earliest and most significant change during leaf senescence at the cellular level is chloroplast breakdown (Lim et al., 2007). Since the leaves initially grew normally in spl29, well-developed chloroplasts could be observed in many spl29 mesophyll cells even when early leaf senescence occurred (Fig. 6B). However, chloroplasts were completely degraded in abundant dying spl29 mesophyll cells ( Fig. 6D; Supplementary Figure S5C and D), with chloroplast breakage potentially being involved in this degradation process ( Fig. 6C; Supplementary Figure S5A and B). Chloroplast degradation in spl29 is different from the autophage-mediated process described in previous reports (Jiao et al., 2012), but the drastic and irreversible degradation process is nevertheless coincident with the rapid leaf senescence phenotype in spl29.
At the molecular level, leaf senescence is mediated by a large number of genes. The senescence-associated transcription factors OsWRKY23, OsWRKY72, and OsNAC2 are all upregulated during leaf senescence, caused by knockdown of the H subunit gene of the glycine decarboxylase complex . Four SAGs (Osl2, Osl30, Osl43, and Osl85) are upregulated during darkinduced or natural leaf senescence (Lee et al., 2001). Also the expression of photosynthesis-related genes usually decreases during leaf senescence (Lim et al., 2007). Upregulation of senescence-associated transcription factors ( Fig. 7A and B) and SAGs ( Fig. 7C and D) and downregulation of photosynthesis-related genes ( Fig. 7E and F) in spl29 together support the hypothesis that early leaf senescence in this mutant is not a passive and unregulated degenerative process, but an active and well-controlled genetic programme.
The physiological, cellular, and molecular evidence all suggest that early leaf senescence happens in spl29 plants. Functional inactivation of OsUAP1 results in the appearance of early leaf senescence, suggesting a role for OsUAP1 in the regulation of this process.

Defence responses are induced by functional inactivation of OsUAP1
LMMs frequently have upregulated defence response gene expression and spontaneous appearance of HR-like lesions, both of which may contribute to enhanced resistance to pathogens (Yin et al., 2000). A point mutation in OsUAP1 results in the induction of HR-like lesions and enhanced resistance against the bacterial blight pathogen (Fig. 8A). Induced . Data represent the mean ± SD from 5-7 independent plants at the tillering stage (Student's t-test: ***, P < 0.0005). (B, C) Relative expression of four defence signalling-related genes in wild-type and spl29 plants at the seedling and tillering stages analysed by qRT-PCR. The expression level of each gene in the wild type was normalised to 1; data represent the mean ± SD of three biological replicates (Student's t-test: ***, P < 0.0005).
defence responses can also be reflected by the upregulation of defence response-associated genes. PR1a and PBZ1 have commonly been used as molecular markers for rice defence responses (Campbell and Ronald, 2005;Tang et al., 2011). Constitutively overexpressing rice PO-C1 enhanced disease resistance and induced transcript levels of pathogenesisrelated genes in transgenic carrot (Wally and Punja, 2010). Overexpression of OsWRKY45 markedly enhanced fungal and bacterial disease resistance (Shimono et al., 2012). These four defence response-associated genes (PR1a, PBZ1, PO-C1, and OsWRKY45) were all upregulated in spl29 ( Fig. 8B and C), and probably play important roles in its enhanced resistance to bacterial blight. Defence responses are induced in spl29, which implies a role for OsUAP1 in disease resistance by regulating the defence response.

ROS (O 2
and H 2 O 2 ) accumulated in spl29 ( Fig. 9A and B), and the mutant also had elevated levels of MDA ( Fig. 9C and D), a product of lipid peroxidation. Increased SOD activity, but normal CAT activity (Fig. 9E-H), may result in the H 2 O 2 accumulation in spl29 plants. Despite increased SOD activity to remove O 2 -, O 2 still accumulated in the mutant, and therefore the ROS accumulation in spl29 can probably be attributed to the overproduction of O 2 -. ROS may play several important roles in the leaf senescence of spl29. ROSmediated chloroplast degradation happens during leaf senescence (Khanna-Chopra, 2012), while lipid peroxidation by ROS has been proposed as the major cause of membrane deterioration in plant senescence (Pauls and Thompson, 1980). Thus, ROS accumulation is the most likely reason for the lipid peroxidation ( Fig. 9C and D) and chloroplast membrane breakage ( Fig. 6C; Supplementary Figure S5A and B) in spl29 leaves. The onset of plant senescence can be promoted by ROS (Navabpour et al., 2003). Expression patterns of the senescence-related transcription factors OsWRKY23, OsWRKY72, and OsNAC2, as well as the SAG Osl85, were all upregulated after H 2 O 2 treatment . Upregulation of these four genes was also found in spl29 plants (Fig. 7A-D). ROS probably also play a key role in the defence responses of spl29. The oxidative burst, a rapid and transient production of huge amounts of ROS, is one of the earliest observable aspects of a plant's defence strategy (Wojtaszek, 1997). ROS have been implicated not only in anti-pathogen roles, but also in cellular signalling associated with the induction of defence gene expression (Desikan et al., 2000). In spl29, the bacterial blight resistance was enhanced (Fig. 8A) and defence response-associated genes were upregulated ( Fig. 8B and C).
In addition, JA and ABA levels were significantly elevated in spl29 plants (Fig. 10). The JA and ABA signalling (G, H) CAT activities. For C-H, leaf samples were analysed at the seedling and tillering stages; data represent the mean ± SD of three biological replicates (Student's t-test: ***, P < 0.0005). FW, fresh weight.
pathways are widely considered to play a role in modulating leaf senescence and defence responses (Lim et al., 2007;Robert-Seilaniantz et al., 2011;Zhang and Zhou, 2013). The accumulated JA and ABA in spl29 (Fig. 10) are probably involved in its early leaf senescence and defence responses.
In conclusion, functional inactivation of OsUAP1 induces early leaf senescence and defence responses in spl29. ROS, JA, and ABA are all likely to be involved in these two biological processes or pathways in the mutant. It is suggested that OsUAP1 may play an important role in regulating leaf senescence and defence responses in rice.

Supplementary material
Supplementary data can be found at JXB online.
Supplementary Table S1. List of PCR-based molecular markers used for map-based cloning of SPL29.
Supplementary Table S2. Primers for the detection of the mutation site, construction of the functionally complementary vector, and confirmation of positive transgenic plants.
Supplementary Table S3. All primers used for qRT-PCR analysis.
Supplementary Table S4. Formulae and glossary of terms used in the analysis of chlorophyll a fluorescence transients.
Supplementary Table S5. Segregation of F 2 populations from three crosses. Supplementary Table S6. Sequencing of the mutation site in different rice cultivars and F 2 plants.
Supplementary Figure S1. Determination of reference genes for normalisation in qRT-PCR analysis.
Supplementary Figure S2. Phenotype of the secondemerged leaf after germination in wild-type and spl29 plants.
Supplementary Figure