The indica rice cultivar Jiafuzhan and the japonica rice cultivar Hui1586 were preserved at the Rice Research Institute, Fujian Academy of Agricultural Sciences, China. Hui1586 was derived from a cross between Suxiu867 and Minghui86, and the detailed selection process was as follows: Suxiu867 (Food Crops Research Institute, Jiangsu Academy of Agricultural Sciences), a japonica cultivar, was used as a recipient, and Minghui86 (Rice Research Institute, Fujian Academy of Agricultural Sciences), a restorer indica cultivar, was used as a donor. The F₁ plants were generated from Suxiu867 as the female parent and Minghui86 as the male parent. The F₁ plants were backcrossed to the Suxiu867 parent to produce the BC₁F₁ generation. Six BC₁F₁ plants were backcrossed to the Suxiu867 parent to produce 6 BC₂F₁ plants, which were self-pollinated to produce 72 BC₂F₂ lines (12 individuals from each of the six BC₂F₁ plants was sown). The 72 individuals were self-pollinated for six generations. In this process, shorter plants, higher seed setting rate and better comprehensive agronomic traits were selected for to continue planting single plant and eliminate the remaining lines. Then a stable line designated Hui1586 was obtained.

For the development of the NILs, the indica cultivar Jiafuzhan was used as a recipient, and the restorer japonica cultivar Hui1586 was used as a donor. The F₁ plants were generated from the Jiafuzhan as the female parent and Hui1586 as the male parent. The F₁ plants were back-crossed to the Jiafuzhan parent to produce the BC₁F₁ generation. These BC₁F₁ plants were then backcrossed to the Jiafuzhan parent to produce BC₂F₁ plants. Using the same approach, 118 BC₃F₁ individuals were obtained, which were self-pollinated to produce the BC₃F₂ lines. Based on their characteristics, we selected one or two individual plants from each line. As a result, 176 NILs were obtained (Supplementary Figure 1).

The WinQTLCart 2.5 software (Wang et al., 2012) implementing the composite interval mapping (CIM) was employed for QTL detection (Zeng, 1993). The confidence interval was defined as the 1-LOD reducing region around the locus of a peak LOD value of the identified QTL. LOD value of ≥ 2.5 was set as a threshold for QTL identification.

In autumn 2019, Jiafuzhan, Hui1586 and 176 NILs were planted under natural conditions in the paddy fields of Sanya Experimental Station, Hainan Province, China (18°14′N, 109°31′E, 7^.0 m asl). Forty-eight plants of each of the parents and the NILs were planted in six rows. Four plants from the center of each plot were picked to evaluate their plant height characteristics. Major QTLs associated with plant height were identified on the basis of significant differences in plant height between parents and each of the NILs, as determined by t-test. Moreover, panicle length, effective panicle number, spikelets per panicle, seed setting rate and 1, 000-grain weight were also estimated at maturity stage for the parents, NIL36 and knockout lines.

All plants were planted in accordance with standard commercial practices. Plants were grown in the field at 13.3 cm plant to plant distance and 26.40 cm row to row distance. Agronomical practices were performed according to the normal agricultural practices locally recommended for rice production.

The NIL36 line was crossed with the Jiafuzhan cultivar to develop a mapping population. The F₂ population was constructed by self-crossing of the F₁ hybrid. A primary linkage of the QTLs for plant height was obtained using 45 recessive plants from the F₂ population. Furthermore, 1264 recessive plants from the F₂ population were selected for fine mapping of major plant height QTLs.

The CTAB method (Murray and Thompson, 1980) with minor modifications was used for the extraction of plant DNA from frozen leaves of the rice plants. For PCR amplification, each 20-μl reaction mixture contained 30 ng DNA, 0.4 μm primers and 2× Es Tag MasterMix (Dye). The amplification program includes the following procedures: 2 min at 94°C; 33 cycles of 30 s at 94°C, 30 s at 55°C, and 30 s at 72°C; and a final extension at 72°C for 2 min. The amplified PCR products underwent 3% agarose gel electrophoresis and were stained with ethidium bromide (Panaud et al., 1996).

We used the obtained phenotypic data and SSR markers for the identification of QTLs. Genetic distance was estimated using MapDraw V2.1 (Liu and Meng, 2003). The genetic linkage map obtained in this study is basically consistent with that reported by Rahman et al. (2007).

The physical map of QTLs for plant height was constructed through a bioinformatics analysis using the published sequences of BAC and P1-derived artificial chromosome (PAC) clones of cv. Nipponbare released by the International Rice Genome Sequencing Project.¹ Target gene linkage markers were used to clone, and sequence alignment was performed using the matching Basic Local Alignment Search Tool. According to the existing sequence annotation database,² candidate genes based on the existing sequence annotation database analysis were identified.

The first exon of the qPH12 gene in the Jiafuzhan cultivar was targeted with one gRNA spacer. Highly specific gRNA spacer sequences were designed using CRISPR plant database and website (gRNA: ggctgacgaccgggagaagaagg; Xie et al., 2014). The primer sequence for vector construction were showed (Supplementary Table 1). Genome editing mutations of target genes in regenerated plants were analyzed. The deletion and insertion within targeted genes were detected by PCR. PCR products were selected from transgenic CRISPR-edited strains for sequencing to identify specific mutations. The degradation sequence decoding method was used to analyze the double peaks (Ma et al., 2015). The primers used in CRISPR/Cas9 experiments are shown in Supplementary Table 1.

Stems of Jiafuzhan (CK), NIL36, qPH12KO-line1, qPH12KO-line2 and qPH12KO-line3 were sampled during the rice jointing stage. The auxin (IAA) contents were measured using MetWare³ based on the AB Sciex QTRAP4500 LC–MS/MS platform. For the determination of auxin content, plants were sampled at the heading and jointing stage of rice with 3 replicates were sampled from each genotype. Samples were treated with methanol dissolved as a solvent and stored at −20°C. Samples were diluted into different gradient concentrations before mass spectrometry. The Ultra Performance Liquid Chromatography was used for separation. The Multiple Reaction Monitoring was implemented for the analysis.

Total RNA was extracted from rice leaves according to the instructions of the extraction kit (TRIzol, Invitrogen, United States), and DNase treated. An aliquot of about 1.5 μg of RNA was reverse transcribed using the first strand cDNA synthesis kit (Bao Bioengineering Co., LTD.), and the cDNA was 10-times diluted for RT-qPCR. Primers for RT-qPCR were designed and optimized to >95% amplification efficiency. Fluorescence of the Real-Time Fluorescence Quantitative kit (Bao Bioengineering Co., Ltd.) was measured in an CFX96 real-time PCR apparatus (Bio-Rad, Munich, Germany) according to the RT-qPCR procedure described by Bustin et al. (2009). Expression levels were measured using three independent biological replicates and three technical replicates and normalized against the reference gene UBIQUITIN. Primer sequences of related genes are shown in Supplementary Table 1.

To evaluate the potential advantages of the NILs for major QTLs detection, the phenotypic variations in plant height were observed in 176 NILs, with the NIL36 line exhibited a lower plant height compared to the Jiafuzhan cultivar, however, no significant differences were observed in plant height among the remaining NILs. QTL analysis using the WinQTLCart 2.5 software showed that the LOD score value of the plant height QTL for the NIL36 reached 7.84, with an explained phenotypic variance (R²) of 17.52%. Further investigations and analyses showed that the plant height of the Jiafuzhan plants was 116.22 cm, whereas that of the NIL36 line was 79.52 cm. The differences in plant height between these lines reached a highly significant level, as revealed by the t-test (Table 1; Figure 1).

Phenotypic comparisons between the NIL36 and the Jiafuzhan plants are presented in Table 1. The results showed some significant differences in major agronomical traits, including plant height, panicle length, spikelets per panicle and yield per plant, between the NIL36 line and the Jiafuzhan cultivar. However, no significant differences observed in the number of effective panicles, seed setting rate, 1,000-grain weight, grain length or grain width (Table 1).

To determine whether plant height in the NIL36 line is controlled by a single gene or not, NIL36 was crossed to the Jiafuzhan cultivar. F₁ plants showed the plant height phenotype of the Jiafuzhan cultivar, meanwhile the F₂ population showed Mendelian segregation (Table 2). The segregation between the Jiafuzhan and NIL36 phenotypes fit the 3:1 segregation ratio in the two F₂ populations (χ² = 0.134 ~ 0.456, p > 0.5). The results showed that the plant height phenotype in the NIL36 is controlled by a single recessive gene.

To identify the gene responsible for the NIL36 plant height phenotype, we located the plant height QTL in the NIL36 and a total of 506 SSR markers from the rice molecular map were selected for polymorphism surveys between Hui1586 and Jiafuzhan (Mccouch et al., 2002). Out of those 506 SSR markers, 296 exhibited polymorphisms between Hui1586 and Jiafuzhan cultivars. Based on the genotypic data of these 296 SSR markers, 45 recessive plants from the F₂ population (NIL36/Jiafuzhan) were used for co-segregation analysis between the SSR markers and the plant height phenotype. One of these SSR markers, RM3326, that located on chromosome 12, showed a complete co-segregation with the plant height phenotype in the selected 45 F₂ recessive individuals. The major QTL was therefore designated qph12.

Based on 296 polymorphisms exhibited between Hui1586 and Jiafuzhan cultivars, the genetic background of NIL36 homozygous material was analyzed using 145 SSR primer pairs distributed evenly on the 12 chromosomes of rice. The results showed that four markers, i.e., RM6832 on chromosome 3, RM3498 on chromosome 6, RM3496 on chromosome 8, and RM2584 on chromosome 12, exhibited the homozygous alleles of the Hui1586 cultivar., whereas the remaining 292 markers showed the genetic background of the Jiafuzhan cultivar (Supplementary Figure 2). Therefore, the NIL36 basically restores the genetic background of the recipient parent.

Publicly available molecular markers around RM3326 marker were used to initially locate the qph12 QTL. A genetic linkage analysis revealed that the qph12 QTL is located between the molecular markers RM2854 and RM235, which are located at a distance of 7.7 cM (Figure 2A). To delimit the genomic region of the qph12, 1264 recessive plants from the Jiafuzhan/NIL36 F₂ population were genotyped using six polymorphic indel markers selected from 18 newly developed indel markers (Table 3). Indel markers from the open rice genome sequences were designed and tested to predict the likelihood of polymorphism between the NIL36 line and the Jiafuzhan cultivar by comparing sequences from Nipponbare⁴ and the indica cultivar 93–11.⁵ The genotyping of all recombinant genes was performed using six polymorphic markers. The results showed that the qph12 QTL was located within a 295 kb region between the molecular markers Indel12-7 and Indel12-9 on chromosome 12 (Figure 2B; Table 3).

For fine mapping of the qph12 QTL, eight polymorphic indel markers were selected from 26 newly developed indel markers (Table 3). Recombinant screening with eight markers located in a more internal position within the target locus detected 13, 11, eight, six, two, one, two, and five recombinant plants, respectively (Figure 2C). Thus, the qph12 locus was precisely located within a 31 kb region between the molecular markers Indel12-29 and Indel12-31.

According to the publicly available sequence annotation databases,^6,7 three annotated genes with a corresponding full-length cDNA are located within the 31 kb region (Figure 2D). Among these genes, LOC_Os12g40860 encodes the leucine-rich repeat family protein, LOC_Os12g40880 encodes the uridine kinase family protein, and LOC_Os12g40890 is the auxin-responsive Aux/IAA gene family member OsIAA30.

To identify the gene responsible for the observed plant height phenotype, we sequenced the three candidate genes in the Jiafuzhan cultivar and the NIL36 line. A deletion of only 1 bp (120:C) was found in the LOC_Os12g40890 gene in the NIL36 (Figure 2E), and no further difference was observed between the sequences of the remaining two genes in the two genotypes. Thus, we hypothesized that the recessive gene corresponds to the major plant height QTL qph12 in the NIL36 line, and the dominant allele (LOC_Os12g40890) of the WT Jiafuzhan was designated qPH12. Since the fragment of qph12 was probably derived from the parent Hui1586, we sequenced the qph12 of Hui1586, and the results showed that the sequences of qph12 from both Hui1586 and NIL36 are identical (Figure 2E).

The analysis of the open reading frame (ORF) region showed that the qPH12 gene has five exons. qph12 exhibited a 1 bp deletion in the 120th bp of the first exon, which resulted in premature termination of the qPH12 (Figure 2E).

To confirm that qph12 confers the plant height phenotype, we examined whether the knockout of qPH12 in the Jiafuzhan cultivar would lead to the NIL36 phenotype. One sequence-specific guide RNA (sgRNA) was designed to knock out qPH12 using the CRISPR/Cas9 gene editing system. After resistance screening, 20 strains were randomly selected for sequencing. The results revealed four main types of mutations, i.e., insertion, deletion, complex variant and no mutation, from which insertion and deletion T₁ homozygous genotypes were self-pollinated for two generations to produce the T₃ generations, designated as qPH12KO-line1 and qPH12KO-line2, respectively. The complicated variant T₁ replaced the heterozygous genotype, and the T₂ generation was sequenced and screened, and self-pollinated for three generations to produce the T₄ generation, qPH12KO-line3. A total of three plants from three independent events (qPH12KO-line1, qPH12KO-line2 and qPH12KO-line3) were obtained. Sequencing of the qPH12 gene in the three independent knockout lines confirmed that these plants carry mostly insertion or deletion in the targeted sites (Figure 3A). Evaluation of the plant height phenotype of these three homozygous lines at maturity and found that all three lines showed the NIL36 phenotype (Table 1; Figure 3B). Therefore, the targeted mutation of the qPH12 gene led to the NIL36 plant height phenotype, indicating that the loss of function of qPH12 was responsible for the reduced plant height phenotype. Notably, the three knockout lines qPH12KO-line1, qPH12KO-line2 and qPH12KO-line3 showed shorter panicle length, fewer number spikelets per panicle and lower plants yields compared to the Jiafuzhan cultivar (Table 1; Figures 3C–F). Therefore, we hypothesized that the qph12 QTL not only affects plant height but also underlies panicle length, spikelets per panicle and plant yield in rice. According to the standard commercial practices, qPH12KO-line1, qPH12KO-line2, qPH12KO-line3 and Jiafuzhan were grown in a paddy field under natural environmental conditions at transgenic experimental base in Fuzhou, China.

To analyze whether the qph12 QTL affects the changes in the Aux/IAA levels, we measured the Aux/IAA content in the Jiafuzhan (CK) and the qPH12KO knockout lines. The results showed that the Aux/IAA content in qPH12KO-line1, qPH12KO-line2, qPH12KO-line3 and NIL36 line was significantly lower than that of the Jiafuzhan (CK; Figure 4).

To further investigate the expression abundance of qPH12 gene in Jiafuzhan, NIL36 and the three qPH12KO knockout mutant lines, we analyzed the expression profile of qPH12 in these materials using RT-qPCR. The qPH12 expression was significantly higher in the Jiafuzhan that showed higher plant height and grain yield, longer panicles and more spikelets per panicle compared to the NIL36 line and the three knockout mutant lines qPH12KO-line1, qPH12KO-line2 and qPH12KO-line3 that showed a reduced plant height phenotype (Figure 5).