Overexpression of a NAC-domain protein promotes shoot branching in rice


  • The GenBank accession number for the OSTIL1 sequence is DQ520641.

Author for correspondence: Ping Wu
Tel:+86 571 88206412
Fax:+86 571 88206617
Email: clspwu@zju.edu.cn


  • • For a better understanding of shoot branching in rice (Oryza sativa), a rice activation-tagging library was screened for mutations in tiller development. Here, an activation-tagging mutant Ostil1 (Oryza sativa tillering1) was characterized, which showed increased tillers, enlarged tiller angle and semidwarf phenotype.
  • • Flanking sequence was obtained by plasmid rescue. RNA-interfering and overexpression transgenic rice plants were produced using Agrobacterium-mediated transformation.
  • • The mutant phenotype was cosegregated with the reallocation of Ds element, and the flanking region of the reallocated Ds element was identified as part of the OsNAC2 gene. Northern analysis showed that expression of OsNAC2 was greatly induced in the mutant plants. Transgenic rice overexpressing the OsNAC2 resulted in recapture of the mutant phenotype, while downregulation of OsNAC2 in the Ostil1 mutant through RNA interfering (RNAi) complemented the mutant phenotype, confirming that the Ostil1 was caused by overexpression of OsNAC2.
  • • Overexpression of OsNAC2 regulates shoot branching in rice. Overexpression of OsNAC2 contributes tiller bud outgrowth, but does not affect tiller bud initiation. This suggests that OsNAC2 has potential utility for improving plant structure for higher light-use efficiency and higher yield potential in rice.


Tiller is an important agronomic trait for rice production, and also a model system to study shoot branching in monocotyledonous plants. Shoot branching is regulated by two distinct steps: the establishment of axillary meristems and the outgrowth of axillary buds. After formation, axillary buds often remain dormant and require one or more of a wide range of cues before outgrowth ensues. Several genes, including rice MOC1, tomato BL and LS, and Arabidopsis REV, have been reported to be involved in the initiation of axillary buds (Talbert et al., 1995; Schumacher et al., 1999; Schmitz et al., 2002; Li et al., 2003), while some other genes, such as Arabidopsis MAX1-4, pea RMS1 and Petunia hybrida DAD1, were involved in axillary bud outgrowth (Sorefan et al., 2003; Booker et al., 2004, 2005; Snowden et al., 2005). Despite the encouraging progress made in the molecular cloning of genes, knowledge of the molecular mechanism for shoot branching is still fragmentary, especially in the monocotyledonous plants.

Several genes affecting the shoot-branching process have been described in rice. MOC1, a GRAS family transcription factor, was reported to control axillary meristem initiation and axillary bud outgrowth (Li et al., 2003). OsTB1, a TCP domain protein, functions in the repression of bud activity. The dormancy of tiller buds was weakened in the fc1 mutant, a loss-of-function mutant of OsTB1. Transgenic rice ectopically expressing OsTB1 led to the suppression of tiller development (Takeda et al., 2003). Several other mutants associated with tiller bud outgrowth were also reported. Among these, two genes were fine-mapped, designated as d3 and htd-1, and suggested to be the rice homologues of MAX2 and MAX3, respectively, in Arabidopsis (Ishikawa et al., 2005; Zou et al., 2005, 2006). Although the homologous relationship was found between MOC1 and LS, HTD-1 and MAX3, D3 and MAX2, further studies to characterize more tillering genes are required to elucidate whether monocot and eudicot plants share similar molecular mechanisms controlling shoot branching.

The NAC (NAM, ATAF1, 2, CUC2) family proteins constitute one of the largest plant-specific families of transcription factors with > 100 members in Arabidopsis (Riechmann et al., 2000; Olsen et al., 2005) and rice (http://www.tigr.org), respectively. The N-terminal region of NAC proteins contains a highly conserved NAC domain, which may form a helix-turn-helix structure that specifically binds target DNA (Aida et al., 1997; Xie et al., 2000; Duval et al., 2002). The C-terminal regions of NAC proteins, on the other hand, are highly divergent, and are thought to act as transcription-activation regions (Xie et al., 2000; Takada et al., 2001; Duval et al., 2002).

Although NAC family proteins are involved in various processes (Olsen et al., 2005), only a small proportion of the NAC proteins have been characterized. One of the best characterized functions of NAC genes is delimiting organs during embryonic, floral and vegetative development (Souer et al., 1996; Aida et al., 1997; Sablowski & Meyerowitz, 1998; Takada et al., 2001; Vroemen et al., 2003; Mitsuda et al., 2005). NAC genes were also reported to be involved in auxin and abscisic acid signal transduction (Xie et al., 2000; Aida et al., 2002) and plant responses to biotic and abiotic stress (Xie et al., 1999; Ren et al., 2000; He et al., 2005; Selth et al., 2005; Hu et al., 2006). Recent reports indicated that NAC family genes are also involved in regulation of senescence (Guo & Gan, 2006; Uauy et al., 2006). However, no NAC gene has been reported to be involved in lateral shoot branching to date.

Activation tagging produces dominant mutations by means of the overexpression of endogenous genes with transcriptional enhancers or promoters that cause the ectopic expression of genes in the vicinity of the T-DNA insertion site (Suzuki et al., 2001). However, no gene was cloned in rice using the activation-tagging system. In this study, an activation-tagging transposon system, which can activate transcription of neighbouring genes by two 35S promoters and/or by four tandem repeats of the enhancer fragment of this promoter (Suzuki et al., 2001), was used to produce activation-tagging T-DNA insertion mutagenesis lines. An activation-tagging mutant with extensive tiller number and enlarged tiller angle was identified and characterized.

Materials and Methods

Isolation of Ostil1 mutant

The rice activation-tagging library of the Nipponbare (Nip) genotype was constructed using the vector pAD100 designed for obtaining gain-of-function mutations (Suzuki et al., 2001). The library was screened for mutations in tiller development. The Ostil1 mutant was found to have increased tiller number and tiller angle, and semidwarf phenotype.

Plant materials and growth conditions

Wild-type (WT) rice (Oryza sativa L.) cv. Nipponbare and mutant Ostil1 were used for the genetic study. Rice plants were mainly grown on soil in a glasshouse at 30°C (day) and 24°C (night) under long-day (15 h light and 9 h dark) conditions. Transgenic rice plants were grown in a safety cabinet under the same conditions as previously described.

At maturity, plants were characterized for the phenotype alteration between the WT and the mutant, and the tiller buds were investigated at different growth stages.

Cosegregation analysis and cloning of OsNAC2

The original T-DNA insertion, excision and reinsertion of the Ds element were analysed as described (Suzuki et al., 2001). The hygromycin-resistant gene located in the Ds element was used for analysis of the cosegregation of the Ds insertion with the mutant phenotype. The Ds insertion was determined by PCR and Southern blot. The primers were designed as follows: GCTTCTGCGGGCGATTTGTGTA; CGGTCGCGGAGGCTATGGATG. To confirm the Ds insertion, a pair of primers was designed across the insert site: pres-up: ACCTGGTGTCTGCTTCCCTAACAG; pres-down: ATGGCTGAGATGTGAATACGGTT. Each of the primers combined with 35S primer was used to confirm the existence of Ds insertion.

The 1192-bp cDNA sequence containing the predicted ORF of OsNAC2 was amplified using RT–PCR with the primers of 5′-AGGAGCAGTTAGCCAGGTAAAG-3′ and 5′-ATTAAGCGAACCTTGGTAGATG-3′. PCR conditions were 94°C for 5 min, followed by 31 cycles, 94°C for 30 s, 60°C for 30 s and 72°C for 1.5 min. The PCR product was cloned into the pUC-T vector and sequenced.

Plasmid rescue

Genomic DNA (20 µg) was digested with HindIII and purified. The digested DNA pellet was then dissolved in water (at a concentration of 20 µg ml−1) and autoligated. After ligation, DNA was purified and dissolved in water at a concentration of 100 µg ml−1. Ligated DNA (1 µl) was transferred into Escherichia coli DH5α cells and plated onto LB plates containing 100 mg ml−1 ampicillin. The positive clones were sequenced using the 35S and M13 primers in the plasmid construct, as described by Suzuki et al. (2001). Sequence was analysed using clustalX ver. 1.81.

Construction of vectors and plant transformation

The overexpression vectors were constructed as indicated. First, the CaMV 35S promoter was subcloned between the EcoRI and SacI sites of pCAMBIA1301; second, the poly(A) addition sequence of pea ribulose 1,5-bisphosphate carboxylase small-subunit rbcS-E9 was also inserted into the site between HindIII and PstI. The resulting plasmid was named 35S-pCAMBIA1301. Then the ORF of OsNAC2 was introduced into 35S-pCAMBIA1301 using the BamHI, EcoRV site.

A 2203-bp promoter was obtained by PCR using primers: TCAGTCGACAACACAGCACTAACCGAGGATTCAG (containing the SalI recognition site) and TCACCCGGG TAGCTAGAGCTTTACCTGGCTAACTG (containing the SmaI recognition site). The resulting DNA fragment was inserted into the 5′ end of the GUS gene (gusA) in pCAMBIA1391Z to create the OsNAC2 promoter::GUS construct.

For the RNAi construct, 192 bp of OsNAC2 cDNA in sense and antisense orientation was constructed into both sides of the second intron of the maize NIR1 gene. The fragment was then inserted into the multiple cloning site of 35S-pCAMBIA2301 (constructed just as 35S-pCAMBIA1301) to construct the OsNAC2 RNAi vector.

The above constructs were used for Agrobacterium-mediated rice transformation of WT or mutant materials (Chen et al., 2003).

Histochemical analysis and GUS assay

GUS analysis was performed as follows. Transgenic plant samples were incubated with GUS staining solution (100 mmol l−1 NaH2PO4 buffer pH 7.0, 0.5% Triton X-100, 0.5 mg ml−1 X-Gluc and 20% methanol) overnight at 37°C. After staining, tissues were rinsed and fixed in formalin acetic acid (FAA) ethanol fixation solution at 4°C overnight, then mounted on slides and photographed (Leica MZ95, Nussloch, Germany).

Shoot bases of 5-d-old seedlings were fixed with FAA fixation solution at 4°C overnight, followed by dehydration and embedding in paraffin (Paraplast Plus, Sigma, MO, USA). Sections (thickness 8 µm) were cut with a microtome (Microm HM325, Walldorf, Germany) and stained with hematoxylin. Sections were observed under bright-field through a microscope (Zeiss AxioCam HRC, Oberkochen, Germany).

Subcellular localization of OsNAC2

CaMV35S-OsNAC2-mGFP4 was subcloned into the binary vector pCAMBIA 1300. The resulting construct was sequenced to verify in-frame fusion and used for transient transformation of onion epidermis using a gene gun (Bio-Rad, Hercules, CA, USA). The GFP was visualized by LSM 510 Laser Scanning Microscope (ZEISS, Oberkochen, Germany).

Yeast system analysis

The Matchmaker Yeast Two-hybrid System (cat. # K1615-1; Clontech, CA, USA) was used for transactivation analysis. The deduced amino acid sequence of OsNAC2 was inserted into the vectors pGBKT7 (BD) and pGADT7 (AD). Two subclones with N-terminal 190 amino acids (1–570 bp) and C-terminal 156 amino acids (562–1029 bp) of OsNAC2 were inserted into pGBKT7 (BD), resulting in fusions with the GAL4 DNA-binding domain and activation domain. The fusion plasmids, BD-OsNAC2, BD-OsNAC2 (1–190 aa) and BD-OsNAC2 (188–343 aa), were transformed into yeast stain AH109. The transformation mixture was plated on synthetic defined media (SD)/-Trp/-His/-Ade medium plates for examination of growth. Quantitative α-galactosidase assays were performed using the method described in the Clontech protocol. The interaction between the pGBKT7 (BD)-p53 and pGADT7 (AD)-SV40 large T-antigen served as a positive control. The empty pGBKT7 (BD) vector was used as a negative control.

RT–PCR analysis

Total RNA was extracted from tissues of 20-d-old seedlings of WT, Ostil1 mutant and transgenic plants, and used for reverse transcription using Superscript II according to the manufacturer's instruction (Invitrogen, Carlsbad, CA, USA). The first-strand cDNA was synthesized from total RNA and used as RT–PCR templates. RT–PCR was performed using gene-specific primers for OsNAC2, forward (5′-AGATCGCCATGTCGTCCGTC-3′) and reverse (5′-GCTGCCCGTACTGCGTGAAG-3′). Amplification of actin cDNA was performed as a control. The PCR products were analysed on 1% agarose gel.

Quantitative real-time RT–PCR was performed with an ABI Prism 7000 Sequence Detection System (Applied Biosystems, Foster City, CA, USA) using the SYBR green I master mix (Applied Biosystems) containing optimized buffer, dNTP and Taq DNA polymerase, and manufactured as described in the user manual.

Southern blot analysis

Genomic DNA was isolated from transgenic plants, WT and mutant plants using the cetyl trimethyl ammonium bromide (CTAB) method. Genomic DNA (5 µg) was digested with restriction enzymes and separated on 0.8% agarose gel. After electrophoresis, the digested DNA was transferred to a Hybond-N+ Nylon membrane (Amersham Pharmacia, Little Chalfont, UK). 32P-dCTP-labelled DNA of hygromycin-resistant gene was used as a probe. The blots were hybridized and washed at 65°C under stringent conditions, and analysed using Typhoon-8600.

Northern blotting

Northern blotting was performed as described previously (Mao et al., 2004).

Computational analysis

The sequence of the full-length cDNA was analysed using blast and motifscan (http://hits.isb-sib.ch/cgi-bin/PFSCAN). Genome annotation was done using rgp (rice genome research program, http://rgp.dna.affrc.go.jp) and TIGR (http://www.tigr.org/tdb/e2k1/osa1/pseudomolecules/info.shtml), and blocks of NAC domain were made using block maker (http://blocks.fhcrc.org/blocks/make_blocks.html).

Multiple sequence alignment of the known NAC family genes was conducted using the clustalX ver. 1.81 program using the default multiple alignment parameters and the weight matrix BLOSUM series. The phylogenetic analysis was carried out by the neighbour-joining method, and no treatment was done for the alignment and gaps. The phylogenetic tree was constructed using phylip ver. 3.5c (http://bioweb.pasteur.fr) with a bootstrap analysis of 1000 resampling replications.


Phenotypic analysis of the Ostil1 mutant

A mutant exhibiting increased number of tillers and enlarged tiller angle was isolated from c. 1000 activation-tagged rice lines of Nipponbare (Nip) (as described in Materials and Methods). According to the phenotype, the mutant was named Ostil1 (Oryza sativa tillering1). To investigate the genetic basis of the mutant, Ostil1 was backcrossed with the WT Nip. The BC1F1 showed the phenotype between the mutant and the WT plant. The phenotype of 108 BC1F2 lines showed a ratio of 1 : 2 : 1 of WT phenotype lines : middle-type phenotype lines : mutant phenotype lines (χ2 = 0.264 < inline image = 5.99), indicating that the mutant phenotype is controlled by a single semidominant gene.

The Ostil1 mutant (the homozygous mutant) showed an increased number of tillers, enlarged tiller angle and reduced plant height. Under field conditions, the average tiller number in Ostil1 plants was 27.5 per plant, more than twice that of the WT plants (11.5 per plant) (Fig. 1; Table 1). Furthermore, the tiller angle of Ostil1 was approx. 41.3°, about fivefold that of the WT (7.4°) (Fig. 1; Table 1). The Ostil1 mutant showed shorter plant height (75% of WT), relatively smaller leaf size and slightly shorter panicle length than the WT plants (Fig. 1; Table 1). Slimmer stem and smaller seed size were also observed in the Ostil1 mutant (data not shown). However, no significant difference was observed in root traits between Ostil1 and WT plants (data not shown).

Figure 1.

Phenotype of Ostil1 mutant. (a,b) Phenotype of Ostil1 (right) and wild-type (WT, left) rice (Oryza sativa) at (a) maximum tillering stage (70 d); (b) mature stage (90 d). (c) Tiller number; (d) tiller angle of Ostil1 and the WT (Nip) under different planting densities (open bars, 20 × 20 cm2; closed bars, 30 × 30 cm2). Error bars, SD of 15 plants. Different letters (a–d) at the top of each column indicate significant difference.

Table 1.  Phenotype of Ostil1 and the wild-type rice plant Nip
LineTillersTiller angle (°)Plant height (cm)Length of sword leaf (cm)Width of sword leaf (cm)Panicle length (cm)
  1. Data are average ± SD of 15 seedlings.

Nip11.5 ± 2.02 7.4 ± 1.4683.2 ± 3.4534.8 ± 4.11.5 ± 0.121.9 ± 0.9
Ostil127.5 ± 2.8341.3 ± 3.2862.8 ± 2.4327.9 ± 3.01.1 ± 0.119.7 ± 0.7

Ostil1 promotes tiller bud outgrowth but does not affect tiller bud initiation

Shoot branching is regulated by two distinct steps: the establishment of axillary meristems, and the outgrowth of axillary buds. To investigate which step was promoted in the Ostil1 mutant, tiller bud initiation and its outgrowth were investigated. Compared with WT plants, the Ostil1 mutant showed no extra axillary buds at each of its leaf axils (Fig. 2a–f). Longitudinal sections showed that the initiation of tiller buds was normal in the mutant (Fig. 2a,b).

Figure 2.

Tiller bud development of wild-type (WT) and Ostil1 rice (Oryza sativa). (a,b) Longitudinal sections of three-leaf stage shoot apexes of (a) Nip; (b) Ostil1. (c–f) Tiller buds of Nip (left) and Ostil1 (right) at different growth stages. (c,d) Tiller buds from the first to third leaf axils, (c) three-leaf stage; (d) four-leaf stage; (e) six-leaf stage; (f) heading stage, second node from top. (g) Growth curve of visible tiller buds and tillers after different time periods. Short arrows, tiller buds; long arrows, outgrowth tiller buds; red arrows in (e) indicate tiller bud in first node. Bar: (a,b) 100 µm; (c–f) 1 mm. Error bar, SD of 15 plants.

At the six-leaf stage, the outgrowth of tiller buds was observed at the first node on the main tiller (stem) in Ostil1 mutant plants, but not in WT plants (Fig. 2e). Altered patterns of tiller development were also found at higher nodes (Fig. 2f). In WT plants, tillers appeared from the second to the seventh nodes. The tiller buds at the upper three to four nodes become arrested. By contrast, tiller buds up to the second node from top could grow out in the Ostil1 mutant (Fig. 2f). The same pattern of enhanced tiller bud outgrowth was observed on secondary and higher-order tillers (data not shown). At the mature stage (90 d), the tiller number of each Ostil1 plant is more than twice that of the WT plant (Fig. 1; Table 1). The tiller number of WT plants reached maximum level at 50 d after germination (Fig. 2g). Nevertheless, the Ostil1 mutant is capable of developing tillers through all its life stages, even at the mature stage (Fig. 2g). As shown in Fig. 2(g), visible tiller buds were almost the same in Ostil1 and WT plants before 30 d after germination. After that, more tiller buds were observed in the mutant than that in the WT. These results suggest that the enhanced tillering capacity in the Ostil1 mutant results from the release of axillary buds from their dormant status, but not from the increase of axillary bud numbers. The tiller bud at the top node was still suppressed in the mutant, indicating that the suppression of tiller bud outgrowth was weakened, but developmental control of tiller bud activity was still functional in the mutant.

It is well known that lateral branching in rice is modulated by planting density (Hoshikawa, 1989). When grown at a density of 30 × 30 cm, the WT plants generated significantly more tillers, and tiller angles were larger compared with plants grown at a density of 20 × 20 cm (Fig. 1c,d). Similarly, the number of tillers and tiller angle in the Ostil1 mutant at low planting density were greater than at high planting density (Fig. 1c,d), suggesting that lateral branching in Ostil1 is also modulated by planting density. These data indicate that the Ostil1 mutation results in enhanced lateral branching, but its regulation by environmental conditions remains, which is in accordance with other reported tillering mutants (Takeda et al., 2003; Ishikawa et al., 2005).

Ostil1 is caused by activation-tagging of OsNAC2

Among a population of 108 BC1F2 plants, all plants with the Ostil1 phenotype showed the hygromycin-resistance gene detected by PCR and Southern analysis, while the WT phenotype progenies showed no hygromycin-resistance gene (data not shown). Southern hybridization of Ostil1 genomic DNA probed by the hygromycin-resistance gene (located in the Ds element) showed that the mutant contained a single-copy Ds insertion (Fig. 3c). According to the tagging system, the Ds element carried by the T-DNA can jump and insert into a new position and then activate the gene(s) nearby (Suzuki et al., 2001). If the Ds did not jump, amplification products of 678 and 771 bp from the original T-DNA would be obtained by PCR with primer sets a (Ac primer and 35S primer) and b (35S primer and Sp primer), respectively. If the Ds jumped, no PCR products would be obtained. In this case, no PCR product was obtained from the mutants by any of the above two primer combinations and the Ac primer and Sp primer combination (flanking the Ds element). These results indicate that the Ds element jumped and inserted into a new position. Using the spectinomycin-resistance gene in the T-DNA region as a probe, no hybridization band was found in T2 mutant lines (data not shown). This indicates that Ostil1 was caused by insertion of the Ds element (DsAT) and the original T-DNA insertion site was lost during propagation.

Figure 3.

Molecular characterization of Ostil1. (a) Genomic context of Ds insertion in Ostil1 and the structure of DsAT (the latter redrawn from Suzuki et al., 2001). The HindIII recognition sites used for plasmid rescue are listed. IR, duplicate sequence recovered with the Ac element inserted in the maize wx gene; 35S, CaMV 35S promoter; Ampr, β-lactamase gene; Ori, Escherichia coli ColE1 plasmid origin of replication; En4, four-time repeats of enhancer fragments from CaMV 35S promoter; Hgmr, hygromycin phosphotransferase gene with a nopaline synthase promoter and a gene 4 polyadenylation signal; MoaC, MoaC family protein. (b) Chromosome location of Ds insertion. Numbers indicate cM. (c) Ds copy number in Ostil1 revealed by Southern blotting. (d) Northern blotting probed by flanking genome sequence in the right of Ds. R, Root; SB, stem base; S, stem; L, leaf; P, young panicle.

The 5′ flanking sequence of Ds was obtained by plasmid rescue. Nucleotide sequencing analysis showed that the Ds element (DsAT) is inserted in the 59 064–59 065-bp region of the BAC clone OSJNBa0072F16 of chromosome 4 (Fig. 3a,b). A pair of primers (named ‘pres-up’ for left of the insert site and ‘pres-down’ for right of the insert site) were designed to amplify the DNA sequence across the insert site. The 35S primer combining with ‘pres-up’ and ‘pres-down’ was used to amplify the end of DsAT and the flanking genome sequences. A 799-bp fragment was amplified in the WT plant, while there were no PCR products in the Ostil1 mutant using the primer combination of ‘pres-up’ and ‘pres-down’. By contrast, using the primer combination of 35S primer with ‘pres-up’ or ‘pres-down’, a fragment of c. 500 bp or 1 kb was obtained in the mutant, but not in the WT. The results were also confirmed in the BC1F2 population. This confirmed that the mutant phenotype is produced by the insertion of DsAT.

According to the genome annotation provided by the Rice Genome Research Program (RGP, http://rgp.dna.affrc.go.jp) and TIGR (http://www.tigr.org), the DsAT was inserted in the intergenic region between a molybdenum cofactor biosynthesis protein C and an NAC-domain protein (identical to the OsNAC2 registered in the GeneBank) (Fig. 3a). Northern blotting was performed to investigate if these neighbouring genes were upregulated. Using the 3-kb genome sequence flanking the right end of the DsAT as the probe, an overexpressed transcript can be detected in different tissues of the mutant plants, but not in those of the WT (Fig. 3d). Further analysis using OsNAC2 as a probe also detected the same transcript, indicating that overexpression of OsNAC2 may result in the Ostil1 phenotype.

Recapitulation of the Ostil1 mutant phenotype

To confirm that the overexpression of OsNAC2 results in the Ostil1 phenotype, the OsNAC2 RNA interfering (RNAi) and overexpressing constructs were transformed to the Ostil1 mutant line and the WT plant, respectively. The expression level of OsNAC2 in those transgenic lines was detected by quantitative real-time RT–PCR analysis (Fig. 4e). Six independent RNAi transgenic lines, which the expression of OsNAC2 reduced almost to the level of WT, recovered the WT phenotype (Fig. 4b,c,e; Table 2). On the other hand, altered tiller development including increased tiller numbers, enlarged tiller angle and reduced plant height was observed in seven independent OsNAC2 overexpression transgenic lines. Although the severity varied among the seven lines compared with the Ostil1 mutant, all showed the same phenotypic spectrum (Fig. 4d,e). The correlation between the severity of mutant phenotypes and the level of OsNAC2 expression confirmed that the observed phenotypes were indeed caused by the ectopic overexpression of OsNAC2 (Fig. 4e; Table 2).

Figure 4.

Recapitulation of the rice Ostil1 mutant phenotype. (a–c) RNA-interfering transgenic lines from the Ostil1 mutant. (a) Ostil1 mutant; (b,c) RNA-interfering transgenic lines 1 and 2 (Ri1 and Ri2). (d) Overexpression of OsNAC2 in the wild-type (WT) plant recapitulating the mutant phenotype. Left plant, WT; two right plants indicate OsNAC2 overexpression transgenic lines 1 and 2 (OV1 and OV2). (e) Quantitative RT–PCR analysis of OsNAC2 in the RNA-interfering and overexpression transgenic lines. Expression levels were normalized with respect to the internal control ACTIN and plotted relative to the expression of Ostil1 mutant. Nip, WT. Bar, 2 cm. Error bar, SD of three replicates.

Table 2.  Phenotypes of transgenic rice plants
LineTiller numberPlant height (cm)Tiller angle (°)
  1. Data are average ± SD of 15 seedlings. Ri1, 2, OsNAC2 RNA-interfering transgenic mutant line1 and 2; OV1, 2, OsNAC2 overexpression transgenic line1 and 2.

Nip11.5 ± 2.0283.2 ± 3.45 7.4 ± 1.46
Ostil127.5 ± 2.8362.8 ± 2.4341.3 ± 3.28
Ri111.8 ± 1.9883.0 ± 1.7310.7 ± 0.85
Ri211.4 ± 1.6883.1 ± 1.6310.3 ± 0.65
OV118.2 ± 2.1070.0 ± 0.5729.9 ± 3.04
OV218.8 ± 0.5466.9 ± 0.4130.8 ± 2.84

OsNAC2 is expressed in all tissues

RT–PCR was performed to investigate the spatial expression of OsNAC2 in rice. The results showed that OsNAC2 is expressed in all rice tissues tested except for seed, and the transcript accumulated more in roots and stem bases than in other tissues (Fig. 5g). To confirm the expression pattern, an expression vector of udiA (β-glucuronidase, GUS protein) driven by the OsNAC2 promoter was constructed and transformed into the WT plants. The transgenic plants showed GUS staining in the roots, tiller buds, stem, leaf, lamina joint and the young husks (Fig. 5a–f). The expression pattern revealed by GUS staining is similar to that from RT–PCR analysis.

Figure 5.

Expression pattern of OsNAC2. (a–f) GUS staining of different tissues from OsNAC2::GUS transgenic rice plants, (a) root tip; (b) middle part of root; (c) lamina joint; (d) stem base; (e) leaf blade; (f) young husks; (g) expression pattern revealed by RT–PCR. Bar, 1 mm.

OsNAC2 is a transcription activator

The coding sequence of OsNAC2 consists of three exons of 175, 296 and 561 bp separated by two introns of 134 and 123 bp (Fig. 6a). The NAC domain, which spans amino acids 12–168, is encoded by most of the first, the second, and the beginning of the third exon. Sequence alignment showed that OsNAC2 is a member of the NAC family protein, which is estimated to have about 105 members in Arabidopsis (Riechmann et al., 2000) and c. 125 members in rice (http://www.tigr.org/tdb/e2k1/osa1/pseudomolecules/info.shtml). A phylogenetic tree of the whole protein of OsNAC2 and function identified NAC-domain proteins revealed that OsNAC2 belongs to a branch involved in the CUC/NAM subclade, and is most similar to the Arabidopsis CUC2, CUC3 and petunia NAM (Fig. 6b). The whole protein of OsNAC2 showed 62.4% similarity with CUC2 and 61.3% with CUC3.

Figure 6.

Structure of OsNAC2 and alignment of its putative translation product with other NAC-domain proteins. (a) Schematic diagram showing the genomic structure of the OsNAC2 gene. Open triangles, introns; numbers on top, length of intron. Five conserved regions within NAC family members are indicated by A–E. AA, amino acid. (b) Unrooted phylogenetic tree of known NAC transcription factors. Numbers between branches indicate bootstrap values based on 1000 replications. Names and references for other NACs: Arabidopsis thaliana, ATAF1, ATAF2 (Aida et al., 1997), AtNAC2 (He et al., 2005), AtNAC3 (Takada et al., 2001), AtNAM (Duval et al., 2002), CUC1, CUC2 (Takada et al., 2001), CUC3 (Vroemen et al., 2003), NAC1 (Xie et al., 2000), NAC2, NAP (Sablowski & Meyerowitz, 1998), NST1, NST2 (Mitsuda et al., 2005), TIP (Ren et al., 2000); rice, OsNAC6, ONAC300 (Kusano et al., 2005; Ohnishi et al., 2005); petunia, NAM (Souer et al., 1996); tomato, SENU5 (John et al., 1997); wheat, GRAB1, GRAB2 (Xie et al., 1999).

It has been reported that NAC proteins are nuclear-located, and the NAC family proteins have been proposed to function as transcription factors (Xie et al., 2000; Vroemen et al., 2003). To test whether OsNAC2 is localized in the nucleus, OsNAC2 was fused in-frame to the N terminus of the mGFP4 and transiently expressed in onion epidermal cells. The green fluorescent signal was detected only in the nucleus (Fig. 7a–c), suggesting that OsNAC2 is a nuclear protein. To determine whether OsNAC2 has transactivation activity, the OsNAC2 protein, the fragment of OsNAC2 without the NAC domain (from 188 to 343 aa) and the N-terminal fragment from 1 to 190 aa were fused to GAL4 DNA-binding domain and transformed to yeast strain AH109. The result showed that the yeast cells containing pBD-OsNAC2, pBD-OsNAC2-188-343, pBD-OsNAC2-1-190 and the negative control plasmid pBD all grew well on SD medium, while the cells containing pBD-OsNAC2-1-190 and the negative control plasmid pBD could not grow in Trp-, His- and Ade-deficient SD medium (Fig. 7g). The α-galactosidase activity of yeast with BD-OsNAC2 (188–343 aa) was more than twice as much as that with BD-OsNAC2 (Fig. 7h). The results indicate that the part of OsNAC2 from 188 to 343 aa is a putative transcriptional activator, and the activation may depend somewhat on the unmasking of a repressor domain (the N-terminal NAC domain). α-galactosidase activities of the yeast with AD-OsNAC2 + BD-OsNAC2-1-190 showed no significant difference from that of the yeast with AD-OsNAC2 + BD, suggesting that OsNAC2 does not form a homodimer, at least in the N-terminal region (data not shown).

Figure 7.

Nuclear localization and transactivation activity of OsNAC2. (a–f) Nuclear localization of the OsNAC2 protein in onion epidermal cell. Photographs were taken in dark field for green fluorescence (a,d); in bright light for morphology of the cell (b,e); and in combination (c,f). (a–c) Transformed cell expressing OsNAC2–GFP fusion. (d–f) Transformed cell expressing GFP control. (g,h) Localization of the transactivation domain of OsNAC2. (g) Transactivation analysis of OsNAC2 in yeast. (h) α-galactosidase activity assays in the yeast two-hybrid system. The α-galactosidase activity obtained from different constructs was normalized to that obtained from the pBD and plotted to compare transactivation activity in different parts of the OsNAC2 protein. BD, pBD; 1–343, pBD-OsNAC2; 1–190, pBD-OsNAC2 (1–190 aa); 188–343, pBD-OsNAC2 (188–343 aa). Activity values are means of three measurements. Bar: (a–f) 50 µm; (g) 1 cm. Error bar, SD of three replicates.


Overexpression of OsNAC2 promotes shoot branching in rice

In this study, an activation-tagging tillering mutant Ostil1 was identified. OSTIL1 showed pleiotropic phenotypes including increased tiller numbers, enlarged tiller angle and semidwarf phenotype. Molecular biological analysis showed that the mutant was cosegregated with the Ds insertion and the overexpression of OsNAC2. The OsNAC2 overexpression transgenic lines showed increased tiller number, enlarged tiller angle and reduced plant height in a dose-dependent way, which is in accordance with the phenotype of the Ostil1 mutant (Fig. 4d). Furthermore, the WT phenotype could be recovered by reduction of the expression of OsNAC2 in the Ostil1 mutant (Fig. 4b,c). These results indicate that overexpression of OsNAC2 causes an alteration of shoot branching in rice, suggesting that OsNAC2 has potential utility for improving the rice plant structure for higher light-use efficiency and higher yield potential.

Phylogenetic analysis indicates that OsNAC2 is grouped into the same subclade with the Arabidopsis CUC2, CUC3 and petunia NAM proteins (Fig. 6b). CUC2, CUC3 and NAM were known to function in the development of shoot apical meristem (SAM) and cotyledons, and to act in boundary specification and SAM formation (Souer et al., 1996; Vroemen et al., 2003). However, no phenotypic defect in SAM and organ boundary formation is observed in the Ostil1 mutant and the OsNAC2 overexpressing transgenic lines. This suggests that overexpressed OsNAC2 has no abnormal effects on SAM development and the boundary specification. On the other hand, reduction of the expression of OsNAC2 in WT plants using RNAi showed no significant phenotype alteration (Fig. S1 in Supplementary Material). This could be a reflection of the redundancy of the NAC genes, or could indicate that parts of the accumulated transcripts would be enough to maintain normal development. Further studies are needed to investigate the functional alteration of T-DNA insertional line in which the OsNAC2 was knocked out.

Overexpression of OsNAC2 contributes tiller bud outgrowth while not affects tiller bud initiation

Shoot branching is regulated by two distinct steps. The first step is the initiation of axillary meristems in the axils of leaves. Subsequently, the bud either remains active or goes dormant until outgrowth is triggered (Ward & Leyser, 2004). Tiller buds are normally developed in each leaf axil in the Ostil1 mutant and the WT plant; no extra tiller bud or axillary meristem was found by longitudinal section in the Ostil1 mutant or the OsNAC2 overexpressing transgenic lines (Fig. 2a,b). However, more tiller buds are developed into tillers in the Ostil1 mutant (Fig. 2e,f). These results indicate that the establishment of axillary meristems is normal, but subsequent suppression of tiller bud activity is weakened in the Ostil1 mutant. The activity of tiller buds is controlled by a variety of factors, including their position on the axis, the developmental status of the primary SAM, the age of the plant, and various environmental conditions (Hoshikawa, 1989). In WT plants, the growth of the tiller bud at the first node of the main culm is usually suppressed, and the bud often degenerates. Buds on higher nodes, which arise at a later stage of the vegetative phase, also tend to become dormant in the WT plants. In the Ostil1 mutant, dormancy of the bud at first node and buds at higher-order nodes up to the second node from the top can develop into tillers (Fig. 2e,f). This indicates that overexpression of OsNAC2 promotes tiller bud activity or weakens bud dormancy.

Overexpression of OsNAC2 might regulate tiller bud outgrowth independently of known pathways

Three known pathways that regulate shoot branching have been proposed, including repression of cytokinin synthesis by auxin signalling, regulation of auxin transport by the MAX pathway (and orthologous pathways: Ward & Leyser, 2004; Bennett et al., 2006; Beveridge, 2006; Lazar & Goodman, 2006), and the action of Tb1 (Takeda et al., 2003). To investigate if the first two pathways are involved in shoot branching regulated by the OsNAC2 overexpression, the levels of endogenous auxin IAA and cytokinin zeatin in the junction of root and shoot were measured. No significant difference was observed between the Ostil1 mutant and WT (Table S1). In addition, the expression of D3, HTD-1 and other rice homologues of MAX1 to MAX4 were not changed in the Ostil1 mutant compared with the WT plant (Fig. S2b). These results suggest that shoot branching regulated by overexpression of OsNAC2 could be different from the known auxin-regulated pathways.

OsTB1 was a negative regulator of tiller bud outgrowth. The OsTB1 knockout mutant showed increased tillers by weakened tiller bud outgrowth repression (Takeda et al., 2003). RT–PCR showed that the expression of OsTB1 was not changed in the Ostil1 mutant (Fig. S2b). Yeast two-hybrid analysis revealed that OsTb1 and OsNAC2 do not interact physically (Fig. S2a). These results show that overexpression of OsNAC2 should not regulate shoot branching through OsTB1. For the absence of OsTB1 knockout and overexpression materials, whether OsNAC2 is regulated by OsTB1 was not investigated. However, the enlarged tiller angles were not reported in the OsTB1 knockout material (Takeda et al., 2003). These results suggest that OsNAC2 might not act as a downstream component of OsTB1.

Taken together, overexpression of OsNAC2 in the regulation of shoot branching in rice might involve a pathway that is independent of the known pathways. Further analysis of upstream and downstream genes is needed for a comprehensive understanding of the control of shoot branching by overexpression of OsNAC2.


This work was supported by the Special Program of Rice Functional Genomics of China (2002AAZZ1003), Zhejiang Bureau of Science and Technology and Zhejiang Bureau of Education. The authors thank Yoshihito Suzuki for kindly providing the vector pAD100.