NAL1 allele from a rice landrace greatly increases yield in modern indica cultivars

Significance

This work reports discovery of a unique gene important for rice agriculture. A significant yield enhancement in rice modern cultivar was achieved by identification of a gene, SPIKELET NUMBER (SPIKE) in Indonesian rice landrace. The SPIKE increased grain yield of an indica cultivar IR64, which is widely grown in the tropics, over four seasons at the field level and improved plant architecture without changing grain quality or growth period, which are important for regional adaptability. These results indicate finding of SPIKE will be extremely valuable for contributing to increase grain production of indica rice cultivars.

Abstract

Increasing crop production is essential for securing the future food supply in developing countries in Asia and Africa as economies and populations grow. However, although the Green Revolution led to increased grain production in the 1960s, no major advances have been made in increasing yield potential in rice since then. In this study, we identified a gene, SPIKELET NUMBER (SPIKE), from a tropical japonica rice landrace that enhances the grain productivity of indica cultivars through pleiotropic effects on plant architecture. Map-based cloning revealed that SPIKE was identical to NARROW LEAF1 (NAL1), which has been reported to control vein pattern in leaf. Phenotypic analyses of a near-isogenic line of a popular indica cultivar, IR64, and overexpressor lines revealed increases in spikelet number, leaf size, root system, and the number of vascular bundles, indicating the enhancement of source size and translocation capacity as well as sink size. The near-isogenic line achieved 13–36% yield increase without any negative effect on grain appearance. Expression analysis revealed that the gene was expressed in all cell types: panicles, leaves, roots, and culms supporting the pleiotropic effects on plant architecture. Furthermore, SPIKE increased grain yield by 18% in the recently released indica cultivar IRRI146, and increased spikelet number in the genetic background of other popular indica cultivars. The use of SPIKE in rice breeding could contribute to food security in indica-growing regions such as South and Southeast Asia.

The world’s population is expected to surpass 9 billion in 2050 (http://esa.un.org/unpd/ppp/index.htm). Most of this increase will occur in the developing countries of Asia and Africa. By 2035, a 26% increase in rice production will be essential to feed the rising population (1, 2). Rice (Oryza sativa L.) is a staple food of more than 3 billion people, mainly in Asia. Predominantly, indica cultivars are grown in southern China, Southeast Asia, and South Asia, occupying approximately 70% of the rice-producing area in the world, whereas japonica cultivars are grown mainly in East Asia (3, 4). Because of urbanization and industrialization, an increase in the yield of indica cultivars is a pressing need in Southeast and South Asia (5). Additionally, good grain quality (influencing market value) and high yield are essential for the adoption of new cultivars in local areas (6).

The grain yield of rice is determined by spikelet number per panicle, panicle number per plant, grain weight, and spikelet fertility. Although many quantitative trait loci (QTLs) for yield components have been identified (www.gramene.org), few have so far been isolated. To date, at least nine genes or loci for yield-related traits in rice have been isolated from natural variation: Gn1a and APO1 for number of grains (7–9); GS3, GW2, and qSW5 for grain size (10–12); DEP1 and WFP for panicle architecture (13, 14); SCM2 for strong culm (15); and Ghd7 for late heading and number of grains (16). APO1, SCM2, and DEP1 increased grain yield in a japonica genetic background in field experiments (9, 13, 15). However, no novel cloned gene has been reported to increase grain yield in indica cultivars (17). Here, we identified a gene in a tropical japonica landrace and used the allele to increase the grain yield of modern indica cultivars at the crop level through a breeding concept developed by International Rice Research Institute (IRRI) breeders more than 20 y ago.

In 1989, a breeding program for New Plant Type (NPT) rice was launched at IRRI to increase the yields of modern indica cultivars by using genetic material from tropical japonica landraces (18). Several Indonesian tropical japonica landraces—which are characterized by large panicles, large leaves, a vigorous root system, thick stems, and few unproductive tillers—have been used in international breeding programs. However, despite these features, the NPT cultivars yield less than modern indica cultivars, mainly because of low grain fertility and low panicle number (19, 20). To genetically dissect and elicit the valuable traits of NPT cultivars, we backcrossed the NPT cultivars including YP9 (IR68522-10-2-2) against modern indica cultivar IR64 to develop introgression lines (ILs) (Fig. S1). BC₃-derived ILs, which had favorable yield-related traits and few undesirable traits, were selected by field observation (21). Using the ILs, we identified 21 QTLs for yield components such as total spikelet number per panicle (TSN), grain weight, and panicle number. Among the QTLs, qTSN4, for high TSN, was commonly detected on the long arm of chromosome 4 in five NPT lines derived from different tropical japonica cultivars (22). Additionally, a near-isogenic line (NIL) for qTSN4 from YP9, derived from tropical japonica landrace Daringan with an IR64 genetic background, had more spikelets per panicle and more branches than IR64.

In this study, we isolated the gene for qTSN4 through map-based cloning to facilitate its use in breeding. The phenotypic effects of the gene were validated in transgenic plants and by expression analysis. To confirm the effect on practical grain yield in the field, we evaluated yield and related traits by using NILs with genetic backgrounds of popular indica cultivars.

Results

Characterization of NIL-SPIKE.

We characterized qTSN4, designated here as SPIKE, by using an NIL for SPIKE, NIL-SPIKE (Fig. 1A). NIL-SPIKE had larger panicles (Fig. 1B), leaves (Fig. 1C), and panicle necks than IR64 (Fig. 1D). Among yield-related traits, it had higher TSN (Fig. 1E), flag leaf width (FLW; Fig. 1F), root dry weight (RDW; Fig. 1G), and rate of filled grain (Fig. S2A), but had lower panicle number per plant and 1,000-grain weight (Fig. S2 B and C). Notably, along with the rate of filled grain, the grain appearance was improved (Fig. 1H), presumably owing to a strengthening of assimilate supply to the larger number of spikelets by an increase in vascular bundle number (VBN; Fig. 1I). Consequently, the grain yield per m² (GYS) of the NIL was consistently higher than that of IR64 over four cropping seasons, significantly so in three of the four seasons (Fig. 1J). The average GYS of the NIL was 28% higher in the dry season and 24% higher in the wet season than that of IR64 (∼400 g/m²). Therefore, the increase in GYS in the NIL without a decline in grain appearance was achieved through the enlargement of sink size (high TSN), source size (broad FLW and high RDW), and translocation capacity (high VBN). Additionally, days-to-heading was unchanged (Fig. S2D). Thus, SPIKE is highly useful for improving yield without changing locally adapted traits.

Fig. 1.

Characterization of yield-related traits of a NIL for SPIKE. (A) Plant morphologies. (B) Panicle structures. (C) Flag leaves. (D) Cross-sections of panicle neck. (Scale bars: A, 20 cm; B, 10 cm; C, 5 cm; D, 500 µm.) (E–J) Comparisons between IR64 and NIL-SPIKE of spikelet number per panicle (n = 8) (E), flag leaf width (n = 9) (F), root dry weight at maturity (n = 10) (G), rate of chalkiness in brown rice (H), number of vascular bundles in panicle neck (n = 20) (I), and grain weight per m² among two dry (DS) and wet seasons (WS) (J). Percentages above bars in J are yield increases of the NIL relative to IR64. Values are means, with whiskers showing SD. (SEM in J). Significant at ***0.1%; **1%; *5%; n.s., not significant.

High-Resolution Linkage Mapping and Identification of SPIKE.

To identify a gene for SPIKE, we conducted high-resolution linkage analysis by using 7,996 BC₄F₃ plants evaluated for TSN. The candidate region lay between markers Ind4 and Ind12 (18.0 kbp), in which the Rice Genome Annotation Project database at Michigan State University predicts three genes (Fig. 2A). In addition to TSN, the suggested gene was associated with an increase in secondary branch number and leaf width (Fig. S3). Expression analysis in a young panicle revealed that only Os04g52479 (Nal1: NARROW LEAF 1; ref. 23) was expressed (Fig. S4) and, thus, is the most probable candidate for SPIKE. Analysis of the predicted amino acid sequence of SPIKE revealed three amino acid substitutions between IR64 and NIL-SPIKE, one of them in the trypsin-like serine and cysteine protease domain (Fig. S5). Further, the SPIKE protein shows >84% identity with proteins of Brachypodium, wheat, sorghum, and maize, and high similarity in the trypsin-like serine and cysteine protease domain. This similarity suggests conservation of the biochemical function of the SPIKE protein family among these species.

Fig. 2.

Map-based cloning and expression analysis of SPIKE. (A) A high-resolution map narrowed the SPIKE locus to an 18.0-kbp region between Ind4 and Ind12. The candidate gene is indicated in red. The squares indicate an artifact of gene model prediction. Numbers below the map show the number of recombinants. (B) Semiquantitative expression analysis of SPIKE in culm, leaf, leaf sheath, and root of IR64 and NIL-SPIKE (NIL). (C and D) Production of GUS driven by the NIL-SPIKE promoter in cross-sections of crown roots and lateral roots (C) and young panicles (D). (Scale bars: C, 50 µm; D, 2 mm.) (E) Quantitative expression analysis of SPIKE in 3- to 5-, 6- to 10-, 11- to 20-, and 21- to 50-mm stages of young panicle of IR64 and NIL-SPIKE. Expression is calibrated to the 3- to 5-mm panicle stage of IR64. Values are means of three replications, with whiskers showing SEM. Significant at *5%; n.s., not significant.

Expression Analysis of SPIKE.

SPIKE was consistently expressed in several organs (Fig. 2B). To analyze the expression of SPIKE during plant development, we expressed the β-glucuronidase (GUS) reporter gene under the control of the native SPIKE promoter in transgenic IR64 plants. Histochemical analysis revealed GUS activity in the coleoptile, vascular bundle at the panicle neck and culm, leaves (Fig. S6 A–C), crown roots, lateral roots (Fig. 2C), and young panicles (Fig. 2D). Aside from the coleoptile, the pattern of GUS expression coincided with the organs enlarged in NIL-SPIKE. Quantitative RT-PCR (qRT-PCR) revealed that the expression of SPIKE in young panicles at various stages was consistently higher in NIL-SPIKE than in IR64, and double that of IR64 at the 21- to 50-mm stage (P = 0.05; Fig. 2E). The results suggest that the increase in SPIKE expression at the young panicle stage increased spikelet number.

Gene Validation of SPIKE Through Transgenic Analysis.

To validate our results and to gain insight into the function of SPIKE, we generated overexpressor lines (using a constitutive promoter) and silencing lines (using artificial microRNA: amiRNA). A DNA fragment containing the cDNA of SPIKE from NIL-SPIKE fused with the ubiquitin promoter (Ubi:SPIKE) was introduced into IR64 by transformation. The overexpressor transgenic plants showed a similar phenotype to NIL-SPIKE, including large panicles and broad flag leaves (Fig. 3 A and B). Plants carrying a single copy had significantly greater TSN and FLW than IR64 (Fig. 3 C and D). Plants carrying multiple copies had significantly greater TSN and FLW than those with a single copy, suggesting increasing TSN and FLW along with expression of SPIKE. A significant higher transcript in the event carrying multiple copies of SPIKE cDNA was observed (Fig. S7A). This result suggests the gene dosage effect of SPIKE on the plant phenotype. Concomitantly, TSN and FLW of T₀ Ubi:SPIKE plants increased with copy number (Fig. S7 C–E). In contrast, transformation of two amiRNA precursors that targeted the first (amiRNA1) and fourth exons (amiRNA4) of SPIKE into NIL-SPIKE to down-regulate SPIKE (Fig. 3 E and F and Fig. S7B) produced transgenic plants with significantly lower TSN and narrower leaves than NIL-SPIKE (Fig. 3 G and H). The nal1 (loss-of-function) mutant Fn188 similarly showed reduced TSN and FLW relative to its wild type, Taichung 65 (Fig. S8). These results demonstrate that SPIKE (allele of Nal1 from tropical japonica) enlarges the panicle and flag leaf in correspondence with expression. Although Nal1 was reported to relate to auxin polar transport (23), we observed no differences in indoleacetic acid (IAA) biosynthesis or transport between IR64 and NIL-SPIKE (Fig. S9).

Fig. 3.

Transgenic analysis for SPIKE through overexpression and gene silencing. (A) Morphologies of IR64 plant and Ubi:SPIKE plant in which SPIKE is overexpressed by the ubiquitin promoter. (B) Panicle structures of IR64 and Ubi:SPIKE. Spikelet number per panicle (C) and flag leaf width (D) of IR64 (n = 17) and Ubi:SPIKE plants carrying a single copy (n = 20) and five copies (n = 13). (E) Morphologies of NIL-SPIKE plant and transgenic plant in which SPIKE is silenced by amiRNA. (F) Panicle structures of NIL-SPIKE and transgenic plants. Spikelet number per panicle (G) and flag leaf width (H) of NIL-SPIKE (n = 5) and of amiRNA1 (n = 4) and amiRNA4 transgenic plants (n = 3). Values are means, with whiskers showing SD. Results of Tukey–Kramer test for multiple comparisons at the 5% level are shown in C, D, G, and H. (Scale bars: A and E, 20 cm; B and F, 5 cm.) Means labeled with different letters (a, b, c) differ significantly.

Enhancing Grain Yield in indica Cultivars Through SPIKE.

To evaluate the efficacy of SPIKE at increasing yield in different genetic backgrounds, we introgressed it into a high-yielding indica cultivar, IRRI146 (released as “NSIC Rc158” in the Philippines) (24). Recurrent backcrossing to IRRI146 and marker-assisted selection (MAS) produced the IRRI146-SPIKE NIL (Fig. 4 A and B). Because IRRI146-SPIKE has 98% genetic identity to IRRI146, the pleiotropic effects of SPIKE in IRRI146 were similar to those in NIL-SPIKE. GYS, TSN, and FLW of IRRI146-SPIKE were significantly higher than those of IRRI146 (Fig. 4 C–E). SPIKE from YP9 was similarly introduced into five popular indica cultivars with different genetic and geographic backgrounds. Its effects were confirmed on the different genetic background of popular indica cultivars: PSBRc18 (IR51672-62-2-1-1-2-3) from Philippines, Ciherang from Indonesia, TDK1 from Laos, BR11 from Bangladesh, and Swarna from India. The plants homozygous for SPIKE had significantly higher TSN (Fig. 4F) than the recurrent parent.

Fig. 4.

SPIKE increases grain yield in indica genetic background. (A and B) Gene location (blue ellipses) and plant morphology of New Plant Type cultivar YP4 (A) and IRRI146 and IRRI146-SPIKE (B). (Scale bars: 20 cm.) (C–E) Comparison between IRRI146 and IRRI146-SPIKE of grain weight per m² (C), spikelet number per panicle (D), and flag leaf width (E) (n = 10). (F) Comparison of spikelet number per panicle between indica cultivars with and without SPIKE—PSBRc18 (from Philippines, n = 10), TDK1 (from Laos, n = 10), Ciherang (from Indonesia, n = 13), Swarna (from India, n = 17), and BR11 (from Bangladesh, n = 27)—characterized in the field at IRRI, Philippines. Values are means, with whiskers showing SE. Significant at ***0.1%; **1%; *5%.

Discussion

During the last decade, many genes for yield-related traits in rice have been isolated on the basis of genome sequence information. Such genes associated with enlarging sink size have been identified by evaluating yield-related traits such as spikelet number, panicle branch number, and grain weight (7–16). However, grain yield is a complex trait controlled by many genetic factors related to yield components (25, 26). It has been unclear whether a single gene could increase grain yield in the genetic backgrounds of different elite cultivars, and no genes have been reported to increase grain yield of indica rice at the crop level in the field (17).

To exploit genes for complex traits, trait-specific ILs that have been developed through backcross breeding with phenotypic selection offer a powerful approach (27). In our previous study, genetic factors underlying grain yield were effectively detected by using trait-specific ILs with an IR64 genetic background in a new approach that combines breeding selection with genetic analysis (21). The 334 ILs with high yield potential had been selected from backcrossed populations and had a few segments introgressed from NPT sources. Through selection of backcrossed progeny based on field observation, we eliminated unfavorable agronomic traits such as low panicle number and low grain fertility inherited from the NPT sources. The selected ILs, which had large numbers of favorable genes for grain productivity, tended to have higher grain yields than IR64. Approximately 29% of the ILs had SPIKE, suggesting that NPT SPIKE is a major factor for increasing grain yield among the ILs (21). Additionally, NIL-SPIKE, which was developed from IL YTH326, stably increased grain yield in the IR64 genetic background at the crop level in the field. This approach opens the way to efficiently identifying genes that increase grain yield.

SPIKE had pleiotropic effects on several traits, increasing TSN, FLW, RDW, panicle neck diameter, and VBN, but decreasing the 1,000-grain weight and panicle number. Among components of sink size, SPIKE notably increased secondary panicle branches as well as Gn1a (7). The consequent increase in spikelets on secondary branches usually increases competition for assimilate supply, resulting in abortion (28, 29). To our surprise, however, the grain fertility of NIL-SPIKE was improved compared with that of IR64. This improvement might be explained by increased VBN contributing to assimilate supply to spikelets on the secondary branches. NPT breeding aims at an ideal plant architecture characterized by fewer, larger panicles and large leaves (18). These morphological effects of SPIKE were mostly corresponding to ideal plant architecture in NPT breeding. The introduction of SPIKE into indica elite cultivars gave the desired morphological effects. This study reports a gene that increases grain yield of indica rice through enlarging sink size, source size, and translocation capacity.

Many QTLs for morphological traits, including leaf size, spikelet number, panicle number, and root volume, occur in a cluster on the long arm of chromosome 4 (http://qtaro.abr.affrc.go.jp). In this region, where SPIKE was identified, several QTLs for TSN have been detected, among which the alleles from japonica or tropical japonica cultivars increased TSN (30–38). Hitherto, no gene for TSN or grain yield in this cluster has been isolated (39). Through high-resolution mapping, we identified SPIKE, which influences TSN and FLW. Transgenic analysis revealed that SPIKE was identical to Nal1, which affects vein patterning in leaves and polar auxin transport (23). SPIKE, identified from natural variation, is a unique allele from tropical japonica, whereas nal1, identified from a mutant line, is a loss-of-function mutation. The nal1 mutant was reduced in TSN compared with wild type, whereas the unique allele from tropical japonica in Nal1 showed increased TSN. The fact suggested that the activity of auxin transport at panicle initiation stage might be related to TSN. Through increase in TSN, the grain yield of NIL-SPIKE was increased as a consequence. Therefore, SPIKE was a unique isolated allele for grain productivity from natural variation and located on the QTL cluster region of chromosome 4.

The effect of a gene is generally influenced by both the genetic background and the environment. To understand the effect of SPIKE in different genetic backgrounds, we introduced it into five indica cultivars popular in South and Southeast Asian countries and some of them (Swarna, BR11, PSBRc18, and Ciherang) are grown in millions hectares (24, 40). The improvement of grain yield of these cultivars is an important breeding objective in this region. SPIKE increased TSN in all of these cultivars, confirming its effectiveness in various genetic backgrounds. Additionally, SPIKE would increase grain yield in all popular cultivars because at least TSN in these popular cultivars were increased by SPIKE. To confirm effects of SPIKE in the broad area, multienvironment trials using NIL-SPIKE are ongoing in the Philippines, Laos, Indonesia, and subtropical Japan. In future study, the breeding lines with SPIKE needs to be characterized at different locations and management levels for understanding increase of yield potential in field. SPIKE would increase grain yields of further indica cultivars in South and Southeast Asia through MAS breeding, thus contributing to food security in these regions.

Materials and Methods

Plant Materials.

Through backcross breeding, we developed 334 BC₃-derived ILs, which have variation in agronomic traits inherited from NPT cultivars, in the genetic background of indica cultivar IR64 (21). We selected an IL with high TSN: YTH326 (IR84640-11-110-6-4-2-2-4-2-2-3-B), derived from NPT cultivar YP9 (IR68522-10-2-2), which was derived from a cross between indica cultivar Shennung 89-366 and tropical japonica landrace Daringan (Fig. S1). Using a BC₄F₂ population derived from a cross between IR64 and YTH326, we identified qTSN4, for high TSN, between SSR markers RM3423 and RM17492 on the long arm of chromosome 4 (22). NIL-SPIKE was developed by self-pollination of a plant selected from the BC₄F₂ population and was used for evaluating agronomic traits, transformation, and expression.

Line Fn188, carrying nal1, was provided by Kyushu University under the National Bioresource Project (www.nbrp.jp). Fn188 had been developed from BC₃ progeny derived from a cross between an nal1 mutant as the donor parent and japonica cultivar Taichung 65 as the recurrent parent. The nal1 locus has been mapped between markers C1100 and C600 on the long arm of chromosome 4 (41). Fn188 was used for agronomic characterization to compare with the effects of SPIKE, because we considered Nal1 to be same to SPIKE.

Development of IRRI146-SPIKE.

A high-yielding indica cultivar, IRRI146 (IR77186-122-2-2-3), has been released as “NSIC Rc158” in the Philippines (24). Progeny of a cross between NPT IR65564-22-2-3 from tropical japonica Bali Ontjer and IRRI146 were backcrossed to IRRI146 three times. In each generation, MAS was conducted by using SPIKE-flanking markers RM5503 and RM6909. A whole-genome survey of 96 BC₃F₁ plants using 116 polymorphic SSR markers that covered all chromosomes was conducted. One BC₃F₁ plant was selected and self-pollinated to develop a NIL for SPIKE in the IRRI146 genetic background. This IRRI146-SPIKE was compared with the recurrent parent for agronomic traits and grain yield.

Development of indica Cultivars with SPIKE.

SPIKE was introgressed into five popular cultivars through backcrossing and MAS: PSBRc18 (IR51672-62-2-1-1-2-3) (Philippines), Ciherang (Indonesia), TDK1 (Laos), BR11 (Bangladesh), and Swarna (India) (24). Progeny of the cross between YP9 and each cultivar were backcrossed to the popular cultivar twice. In each generation, MAS was conducted by using the SPIKE-flanking markers Ind2 and RM17487. Plants homozygous for SPIKE were selected from each BC₂F₂ population and evaluated for TSN in a field at IRRI.

Phenotypic Evaluation of SPIKE.

All plants were grown in a field at IRRI, Los Baños, Laguna, the Philippines, and evaluated for 1,000-grain weight, PN, FLW, and TSN at maturity as described (21). The panicle rachis was sectioned at 1 cm below the neck, and VBN were counted under a stereomicroscope. RDW of plants that were grown in pots was measured at maturity.

To evaluate grain yield, we grew IR64, NIL-SPIKE, IRRI146, and IRRI146-SPIKE in a randomized plot with four replications per line. The area of each plot was at least 4.8 m²; three plants were transplanted per hill at 21 d after sowing at 20 cm between hills and 25 cm between rows. As a basal dressing, 30 kg/ha each of N, P, and K were applied the day before transplanting, and 30 kg/ha of N was applied twice as a topdressing at 2 and 4 wk after transplanting. At maturity, 1.0 m² of rice plants (20 hills in each plot) was harvested, and plants were dried in an oven at 70 °C for 5 d. GYS was calculated on a 14% moisture content basis. Grain chalkiness was evaluated with a Grain Inspector (Cervitec 1625 Grain Inspector, FOSS Analytical) with four replications per line.

High-Resolution Linkage Map.

The genomic DNA of 7996 BC₄F₃ plants generated from BC₄F₂ plants heterozygous for SPIKE was extracted from fresh leaves by a simple method. The genomic DNA of 1,073 BC₄F₃ plants with recombination between flanking markers RM17450 and RM3836 was individually extracted from freeze-dried leaves by the cetyl trimethylammonium bromide method. Selected 41 BC₄F₃ plants that occurred with recombination between RM3423 and AGT3 were self-pollinated to generate BC₄F₄ lines to be used for a progeny test. Among the BC₄F₄ lines, homozygous plants from representative recombinants were selected and evaluated for TSN and FLW. Twenty-two DNA markers were used for map construction (Table S1).

Transformation of SPIKE.

A fragment encompassing the full-length coding region of SPIKE was amplified from cDNA derived from young panicles of NIL-SPIKE by using primer pair 8M17-c1. The fragment was ligated into the binary vector pCAMBIA1300int-prUbi1-tNOS (provided by Emmanuel Guiderdoni, Centre de Coopération Internationale en Recherche Agronomique pour le Développement, Montpellier, France) between the maize ubiquitin promoter and the nopaline synthase terminator to generate the overexpression vector. Using Agrobacterium-mediated transformation, we introduced the vector into IR64 (42). The regenerated plants were evaluated for transgene copy numbers by Southern blot analysis. For gene silencing of SPIKE, the amiRNA approach was used (43). Two 21-bp amiRNA sequences—amiRNA1 (TATAAGAAGTATGCTGCGCTA, for the first exon of SPIKE) and amiRNA4 (TTAATATCAAGTTCCAGACGC, for the fourth exon)—were designed by using Web MicroRNA Designer 3 software (http://wmd3.weigelworld.org/cgi-bin/webapp.cgi). The amiRNA precursors were generated through site-directed mutagenesis by using overlapping PCR (Table S1) with plasmid pNW55 as a template, as described (43). The precursors were ligated into the binary vector pCAMBIA1300int-prUbi1-tNOS to generate the silencing vectors. Using Agrobacterium-mediated transformation, we introduced the vectors into NIL-SPIKE (42). The transgenic plants (T₀) were transplanted into pots, and T₁ plants were transplanted in a screenhouse at 20 cm between hills and 30 cm between rows. These plants were evaluated for TSN and FLW.

To generate the promoter:GUS vector, we amplified a 1,918-bp fragment upstream from the ATG codon of SPIKE by using primer pair UP6-1. The amplified fragment was ligated into the binary vector pCAMBIA0380 (Cambia) upstream of the GUS reporter gene. This vector was introduced into IR64 by Agrobacterium-mediated transformation (42). Organs of the regenerated plants were sampled to analyze GUS activity as described (44).

Expression Analysis and IAA Transport.

Total RNA from each organ was extracted by using an RNeasy Plant Mini Kit (Qiagen). To identify a candidate gene for SPIKE, we performed RT-PCR by using 1 μg of total RNA. PCR was performed by using 1 µL of cDNA with the gene-specific primers for each candidate (Table S1). For comparison of expression in different organs, total RNA of young panicle, culm, leaf sheath, leaf, and root was extracted at the panicle initiation stage. RT-PCR was performed with 500 ng of total RNA by using primer pair seq8M17-56 and a ReverTra Ace qPCR RT Kit (Toyobo). qRT-PCR reactions were carried out with one-fifth cDNA mixtures by using primer pair seq8M17-56 with LightCycler 480 SYBR Green I Master Mix on a LightCycler 480 System (Roche Applied Science). The data were normalized to the expression of a house hold gene, Ubiquitin (Os01g22490).

The rate of IAA biosynthesis in IR64 and NIL-SPIKE coleoptiles was investigated by measuring the amount of IAA that was transported from cut coleoptiles to an agar block (described in Fig. S9) by gas chromatography–selected ion monitoring–mass spectroscopy (GC-SIM-MS) as described (45, 46). To investigate polar IAA transport in IR64 and NIL-SPIKE coleoptiles, we applied 3 µM IAA to the top of coleoptile sections (1.5–3.0 mm) for 30 min, then incubated the coleoptiles on an agar block for 10 min and measured the transported IAA by GC-SIM-MS as above.