Fuying Ma, Xiaoyan Zhu,Hui Wang, Shiming Wang, Guoqing Cui, Ting Zhang,Zhenglin Yang,Guanghua He, Yinghua Ling, Nan Wang, Fangming Zhao*

Rice Research Institute,Academy of Agricultural Sciences,Southwest University, Chongqing 400715,China

Keywords:Rice Chromosome segment substitution line Increased number of kernels qSP1 QTL mapping for yield traits

A B S T R A C T A chromosome segment substitution line(CSSL)is a powerful tool for combining quantitative trait locus (QTL) mapping with the pyramiding of desirable alleles. The rice CSSL Z1364 with increased kernel number was identified in a BC3F8 population derived from a cross of Nipponbare as the recipient with Xihui 18 as the donor parent.Z1364 carried three substitution segments distributed on chromosomes 1,6,and 8.The mean substitution length was 1.19 Mb.Of 17 QTL identified on the substitution segments, qSP1 for spikelets per panicle, qSSD1 for seed-set density,and qNSB1 for number of secondary branches explained respectively 57.34%,87.7%, and 49.44% of the corresponding phenotypic variance and were all linked to RM6777.Chi-square analysis showed that the increased kernel number in Z1364 was inherited recessively by a single gene.By fine mapping,qSP1 was delimited to a 50-kb region on the short arm of chromosome 1.Based on DNA sequence,a previously uncharacterized rice homolog of Arabidopsis thaliana AT4G32551 was identified as a candidate gene for qSP1 in which mutation increases the number of spikelets and kernels in Z1364. qSP1 was expressed in all tissues,but particularly in 1-cm panicles. The expression levels of OsMADS22, GN1A, and DST were upregulated and those of LAX2, GNP1, and GHD7 were downregulated in Nipponbare. These results provide a foundation for functional research on qSP1.

1. Introduction

Increases in rice yield and quality are needed to meet market demand, and a continued search for favorable alleles for these traits by genetic analysis of mutants or of natural variation is therefore crucial. Kernel number per panicle is an important component of rice yield,and is directly affected by inflorescence development at the shoot apex in rice. A special branching pattern is formed when the shoot apical meristem of the axillary bud enters the reproductive growth phase,when an inflorescence meristem, rather than leaf and bud primordia, is formed [1].However, our understanding of the molecular mechanism that regulates kernel number development is incomplete. Development of the inflorescence meristem is regulated mainly by the FLORAL ORGAN NUMBER 1 (FON1) and FON2 genes [2]. FON1 encodes a leucine-rich repeat receptor-like kinase(LRR-RLK)that is orthologous to CLAVATA 1(CLV1)in Arabidopsis,and regulates the seed germination process by tuning abscisic acid (ABA)signaling in rice[3].Lateral meristems are regulated mainly by the REVOLUTA subfamily of HD-ZIP transcription factors[4].A variety of growth hormones influence the number of spikelets that develop.PLANT ARCHITECTURE AND YIELD 1(PAY1)affects polar auxin transport activity, and overexpression of PAY1 increases mutant panicle branching and grain number [5]. ROLLED AND ERECT LEAF 2 (REL2) encodes a protein containing DUF630 and DUF632 domains that reduces the number of kernels per main panicle via the auxin synthesis and transport pathways[6].If the auxin pathway is inhibited or disrupted,the kernel number per panicle will be lower. Negative regulation of the OsmiR160 target gene OsARF18 (AUXIN RESPONSE FACTOR18) expression by affecting auxin signaling leads to small seeds and reduced seed setting [7]. OsSAUR45 affects the synthesis and transport of auxins and reduces the seed-set rate by inhibiting expression of the OsYUCCA and OsPIN genes [8]. Early Flowering1 (EL1),a casein kinase, regulates GA responses by phosphorylating the DELLA protein SLR1,which is essential for spikelet fertility in grain production [9,10]. GN1a encodes OsCKX2, an enzyme that regulates cytokinin levels. Differential transcription of OsCKX2 causes cytokinin accumulation, resulting in increased kernel number[11].Brassinosteroid(BR)mediates grain development. smg11 (small grain 11) is an allele of DWARF2 (D2), which encodes a cytochrome P450 (CYP90D2) affecting expression of brassinosteroid biosynthesis, and optimizing the expression of smg11 increases kernel number [12]. The CLUSTERED PRIMARY BRANCH 1 (CPB1) gene is an allele of DWARF11 (D11) encoding a cytochrome P450 protein, and the optimized expression of CPB1/D11 increased the yield per plant[13].OsMKK4 and OsBZR1 positively regulate genes associated with the BR signal pathway,leading to increased numbers of rice spikelets and kernels[14,15]. The RECEPTOR-LIKE CYTOPLASMIC KINASE (RLCK) genes OsRLCK57, OsRLCK107, OsRLCK118, and OsRLCK176 reduce the number of spikelets via negative regulation of the BR signal[16]. However, the developmental processes that control kernel number are extremely complex, and the identification of novel genes for increasing kernel number is important and can allow an improved understanding of the molecular mechanisms involved.

We identified a novel rice chromosome segment substitution line, Z1364, with increased kernel number, derived from Nipponbare as the recipient parent and Xihui 18 as the donor parent,carrying three substitution segments and a significantly increased number of kernels. We performed quantitative trait locus(QTL)mapping of agronomically important traits using a secondary F2population derived from a cross of Nipponbare with Z1364. Finally, we performed fine mapping of qSP1 and identified a candidate gene by sequence and qRT-PCR analysis.

2.Materials and methods

2.1. Plant material and field planting

The rice chromosome segment substitution line Z1364 was developed using Nipponbare as recipient and Xihui 18 as donor parent.After repeated backcrossing and selfing,combined with phenotype-based selection and simple sequence repeat (SSR)marker selection, a genetically stable chromosome segment substitution line with three substitution segments was identified in the BC3F8population and named Z1364.

The plant material used for QTL mapping was a secondary F2population derived from a cross between Nipponbare and Z1364.

Seeds of Z1364, Nipponbare, and the F2population of 207 plants were sown on March 8, 2017 at the experimental station of Southwest University, Chongqing, China. Thirty seedlings of each parent and all F1seedlings were transplanted to the field on April 15, 2017, with 10 plants per row. The spacings between rows and individual plants were 26.4 cm and 16.5 cm, respectively. Finally, thirty seedlings of Nipponbare and progenies(F3)of 3 homozygous single segment substitution lines selected from the F2population by molecular markerassisted selection were planted in 2018 as the same method described in 2017. Conventional management practices were applied.

2.2. Identification of substitution segments in Z1364

First,263 SSR markers polymorphic between Nipponbare and Xihui 18 were selected from 429 markers that covered the entire rice genome with the purpose of developing CSSLs.Molecular marker selection and phenotypic selection were applied from the BC2F1to the BC3F8populations. Finally, a CSSL Z1364 with three substitution segments, which showed the phenotype of increased kernel number,was identified.In the process of SSR marker screening,if the banding pattern of a marker was consistent with that of Nipponbare, it was considered that the DNA fragment was derived from the genome of the recipient parent Nipponbare. If the banding pattern was consistent with that of the donor parent Xihui 18,the DNA fragment was considered to be derived from the genome of Xihui 18. If several consecutive marker fragment patterns were consistent with those of the donor parent, the segment was considered to be a substitution segment [17].Estimation of the length of the substitution segment followed Paterson et al.[18].

2.3. Agronomic trait assessment

Given the similar heading dates of Z1364 and Nipponbare,this trait was not investigated by QTL mapping. Plant height,panicle number,panicle length,number of primary branches,number of secondary branches, kernel length, kernel width,kernel length-to-width ratio,number of spikelets per panicle,number of kernels per panicle, 1000-kernel weight, and yield per plant were recorded for 10 plants of Nipponbare and Z1364 and 207 plants of the F2population. Among these traits,panicle length, number of primary branches, number of secondary branches, number of spikelets per panicle, and number of kernels per panicle were measured using all the effective panicles in the plant. The 1000-kernel weight of Nipponbare and Z1364 was measured from random samples of 3000 kernels,from which 1000-kernel subsets were weighed on an electronic balance, with three repetitions. The 1000-kernel weight of each F2plant was determined as the weight of 200 kernels,multiplied by 5,with three repetitions.The seed-set rate was calculated as kernels per panicle as a percentage of the number of spikelets per panicle. The seed-set density was determined as spikelet number per 10 cm of panicle length.Kernel length-to-width ratio was calculated as kernel length divided by kernel width.The mean value of each trait was used for statistical analysis. Simple statistical analysis including Student's t-test for comparison among ten traits between Nipponbare and Z1364, and examination of the skewness and kurtosis of these traits in the F2population,was performed using statistical functions in Excel 2003.

2.4. QTL mapping and fine mapping of putative qSP1

Leaves of 207 F2plants and the two parents (Nipponbare and Z1364) were sampled to extract DNA at 20 days after transplanting. DNA was extracted using the CTAB method[19]. The procedures for PCR amplification, 10% native polyacrylamide gel electrophoresis, and rapid silver staining followed Zhao et al.[20]Nipponbare bands were scored as“-1”,Z1364 bands were scored as “1”, heterozygous bands were scored as“0”,and the absence of marker bands was scored as“.”The marker assignments of all markers on the substitution segments of Z1364,together with the phenotypic values of each individual in the F2population, were used for QTL mapping.Putative QTL were identified at the significance threshold of P ＜0.05 using the restricted maximum likelihood (REML)estimation method in the HPMIXED procedure of SAS (SAS Institute Inc.,Cary,NC,USA).

The major QTL qSP1 was further fine-mapped using three new SSR markers identified in the same interval as qSP1 using F2recessive plants with multiple kernels.MapMaker(EXP3.0b)software was then used for data analysis.

2.5. Candidate gene prediction and putative qSP1 cloning and sequence

All the genetic information in the QTL mapping regions was analyzed against the Rice Annotation Database (RAD)(http://ricegaas.dna.affrc.go.jp/rgadb/) and the Rice Annotation Project Database (RAP-DB) (http://rapdb.dna.affrc.go.jp/).The genetic information was also aligned with that obtained from Gramene (http://www.gramene.org/rice_mutant/) to identify candidate genes using gene prediction. First, two genes were located in the intervals of qSP1 on chromosome 1.Comparison with reported cloned QTL in the substitution intervals of chromosomes 6 and 8 revealed three cloned genes associated with kernel development. They included PAY1,which affects auxin polar transport,ONAC106,which reduces panicle number and number of kernels per panicle [21],and FON1, which controls vegetative and reproductive development [3]. Primers were designed to amplify the target fragments of these five genes using DNA of Z1364 and Nipponbare as a template,and the amplicons were sequenced by Tsingke Biological Technology Co.,Ltd.(Chongqing,China).Gene mapping was performed using simple-sequence repeat(SSR) markers and the sequence results obtained were compared using Vector NTI software. The fourth exon of putative qSP1 with a differential site was ligated to the T4 vector and transformed into E.coli,and was finally verified by bacterial liquid detection.

2.6. Real-time qRT-PCR analysis of putative qSP1 and other cloned genes related to rice kernel development

Real-time qRT-PCR was performed to characterize the expression pattern of qSP1 in various tissues and at different stages of panicle development in Nipponbare and the expression of cloned genes associated with spikelet development in Nipponbare and Z1364. Total RNA was extracted using the Eastep Super Total RNA Extraction Kit, and reverse transcribed using the GoScript Reverse Transcription System following the manufacturer's instructions,and then analyzed quantitatively on a Bio-Rad CYF96 using real-time PCR Master Mix (TaKaRa Biotechnology (Dalian, China) Co. Ltd.). The rice ACTIN gene was amplified and used as an internal standard to normalize the expression of qSP1 and the other genes tested.The primers used are listed in Table S1.

3. Results

3.1. Identification of substitution segments in Z1364

Six polymorphic SSR markers in substitution segments of Z1364,together with 194 polymorphic SSR markers,were used to detect the substitution fragment and assess the homogeneity of the genetic background of Z1364 using 10 plants of Z1364. All the plants harbored three consistent substitution segments,and no additional chromosomal fragments derived from Xihui 18 were detected. The Z1364 plants were indicated to be homozygous.Three substitution segments of Z1364 originating in Xihui 18 were located on chromosomes 1,6,and 8.The substitution segment on chromosome 1 was RM10177-RM6777-RM1167 and its estimated length was 0.60 Mb.The substitution segment on chromosome 6 was RM412-RM103-RM494-RM20774 and had an estimated length of 0.61 Mb. The substitution segment on chromosome 8 was RM3153-RM8264-RM223-RM284-RM3262 and its estimated length was 2.36 Mb. The total substitution length was 3.57 Mb,the longest substitution length was 2.36 Mb, the shortest was 0.60 Mb,and the mean was 1.19 Mb(Fig.1).

3.2. Phenotypes of Z1364

The number of spikelets per panicle, number of kernels per panicle,panicle length,number of primary branches,number of secondary branches, and seed-set density of Z1364 were significantly greater than those of Nipponbare(Fig.2,Table 1).Plant height, kernel length, kernel width, kernel length-towidth ratio, and 1000-kernel weight of Z1364 were increased significantly compared with those of Nipponbare.In contrast,panicle number and seed-set rate were significantly lower in Z1364.However,the panicle number was 9.8 per plant and the seed-set rate was 87.69%(Table 1).

3.3.Genetic analysis and distribution of kernel traits in the F2 population

Fig.1- Substitution segments of Z1364.Physical distances (Mb)are shown to the left of each chromosome and marker names are shown to the right.Substitution segments together with markers are enlarged to the left of each chromosome and identified QTL are marked on them.qPN,QTL for panicle number;qNPB,QTL for number of primary branches;qNSB,QTL for number of secondary branches;qSP,QTL for spikelets per panicle;qGP,QTL for kernels per panicle;qSSR,QTL for seed-set ratio;qSSD,QTL for seed-set density;qGW,QTL for kernel width;qRLW,QTL for ratio of kernel length-width;qGWT,QTL for 1000-kernel weight.

Fig.2- Phenotypes of Nipponbare and Z1364.a.Plants of Nipponbare and Z1364 at maturation stage;b. Main panicles of Nipponbare and Z1364.Scale length is 20 cm.

Table 1-Agronomic traits of Nipponbare,Z1364,and the F2 population.

Number of spikelets per panicle and seed-set rate displayed a bimodal distribution(Fig.S1).The two peaks for the number of spikelets per panicle were distributed in the ranges 93.67-156.99,with 156 plants and 157.00-228.08, with 51 plants. The ratio of the number of plants with fewer kernels (156) to the number of plants with multiple kernels(51)fitted a 3:1 segregation(χ2=0.002 ＜χ2(0.05,1)= 3.84) (Fig. S1-a). The peak values for seed-set density were distributed in the ranges 46.07-65.99, with 143 plants, and 66.00-87.30, with 64 plants. The ratio of number of plants with low (143) to high seed-set density (64) fitted a 3:1 segregation (χ2= 3.56 ＜χ2(0.05,1)= 3.84) (Fig. S1-b). Additionally,for spikelets per panicle,the ratio of the number(158)of plants of genotypes qSP1/qSP1 and qSP1/qsp1 to that (49) of genotype qsp1/qsp1 at the RM1167 locus in the F2population also fitted a 3:1 segregation(χ2= 0.13 ＜χ2(0.05,1)= 3.84)(Fig.S1-c).

3.4. QTL mapping of agronomic traits in Z1364

Seventeen QTL for agronomic traits were identified in Z1364 substitution segments. Nine QTL were linked to the RM6777 marker on chromosome 1: qSP1 for number of spikelets per panicle, qGP1 for number of kernels per panicle, qSSD1 for seed-set density, qNPB1 for number of primary branches,qNSB1 for number of secondary branches, qGW1 for kernel width, qPN1 for panicle number, qGWT1 for 1000-kernel weight, and qSSR1 for seed-set rate. The first six QTL showed positive additive effects, and their genetic effects from Xihui 18 increased number of spikelets per panicle by 15.29,number of kernels per panicle by 8.43, spikelets per 10 cm panicle by 6.75, number of primary branches by 0.18, number of secondary branches by 3.64, and kernel width by 0.03 mm.These QTL explained respectively 57.34%, 17.31%, 87.7%,4.34%, 49.34%, and 15.7% of the corresponding phenotypic variance.The genetic effects of qPN1,qGWT1,and qSSR1 from Xihui 18 reduced the number of panicles per plant by 1.02,the 1000-kernel weight by 0.6 g, and the seed-set rate by 3.19%,explaining respectively 14.53%, 13.21%, and 11.54% of the corresponding phenotypic variance. Only one QTL, qSSR6 for seed-set rate, was detected on the substitution segment of chromosome 6, increasing seed-set rate by 4.48% and explaining 22.83% of phenotypic variance. Four QTL were linked to RM8264 on chromosome 8:qNPB8,qNSB8-1,qRLW8-1,and qSSR8-1. The genetic effects of these QTL reduced the number of primary branches by 0.25,the number of secondary branches by 1.69,and the kernel length-to-width ratio by 0.04,and increased the seed-set rate by 3.79%; the QTL explained respectively 7.94%, 10.60%, 10.25%, and 16.29% of phenotypic variance. Three QTL were linked to RM284, namely qNSB8-2 for number of secondary branches,qRLW8-2 for kernel lengthto-width ratio, and qSSR18-2 for seed-set rate. The genetic effects of these QTL were estimated to increase number of secondary branches by 1.58 and kernel length-to-width ratio by 0.03, and reduced seed-set rate by 4.29%, explaining respectively 9.28%, 6.20%, and 20.94% of phenotypic variance(Table 2).

Table 2-QTL identified for agronomic traits in rice.

3.5. Fine mapping of putative qSP1 and sequence analysis of candidate genes

qSP1 was initially mapped to a 0.60-Mb region between markers RM10177 (3.64 Mb) and RM1167 (4.24 Mb), and linked with RM6777(4.22 Mb).Four pairs of new SSR primers targeting this interval were designed, and two displaying polymorphism were used for further fine mapping of qSP1 using 53 recessive multi-kernel plants.Finally,qSP1 was delimited within a 50-kb interval between the molecular markers RM10224 and RM1329 on chromosome 1 (Fig. 3-a). Progeny testing (genotype and phenotype of progeny lines) with RM6667 in the F3generation was performed (Fig. 3-b). The mean spikelets per panicle(133.89) of five plants of a single-segment substitution line(SSSL) lacking the locus was greater than that (103.26) of the recipient parent Nipponbare harboring the locus(Fig.3-b).Two candidate genes in this area were then cloned and sequenced.The first of these was BZR1, one of the immediate targets of the brassinolide-activated and BRI1-mediated pathway, which participates in the signal transduction pathway of brassinolide by dephosphorylation. However, there was no sequence difference between Z1364 and Nipponbare. The second was a previously uncharacterized rice homolog of AT4G32551 in Arabidopsis thaliana, annotated as the transcriptional corepressor LEUNIG.Sequencing revealed a single base change from G in Nipponbare to A in Z1364 at base 349 in the fourth exon and resulting in mutation of amino acid 202,Ala,to Thr(Fig.3-a).It was accordingly assigned as a primary candidate gene for qSP1.

Fig.3- Fine mapping of putative qSP1 and progeny testing of genotype and phenotype of SSSL at locus RM6667.a.Fine mapping of putative qSP1.b.Mean number of secondary branches,spikelets per panicle,and seed setting density of five SSSLs with putative qsp1 at locus RM6667 and Nipponbare with putative qSP1 at 6667 RM locus in F3 generation.

Three cloned genes associated with yield development traits were identified in the other QTL mapping intervals of chromosomes 6 and 8: ONAC106 (LOC_Os08g33670), FON1(LOC_Os06g50340), and PAY 1(LOC_Os08g31470). These genes were amplified with the designed primers using DNA of Z1364 and Nipponbare as templates (Table S2). For ONAC106(LOC_Os08g33670), seven sites differed between Z1364 and Nipponbare. A single-base mutation in Z1364 occurred at bases 320 and 323, both of which were changed from C to G,resulting in the mutations of amino acids 107 and 108 from Gly to Ala. The three bases CAG after base 555 and the six bases GACGAC after base 584 were inserted in Z1364,resulting in the insertion of the three amino acids Gln, Glu, and Val after amino acid 185, amino acid 191 changing from Val to Thr, and amino acid 215 changing from Glu to Thr. Bases 643 and 646 both changed from C to G, resulting in the mutation of amino acids 218 and 219 from Pro to Ala. Finally, base 794 changed from C to T, resulting in the mutation of Pro at position 268 to Ser (Fig. S2). Compared with Nipponbare, the reading frame of FON1 (LOC_Os06g50340) showed two base changes in Z1364.Base 742 changed from C to G,resulting in the change of amino acid 248 from Leu to Val, and a single base change at position 1349 of C to T resulted in amino acid 450 changing from Thr to Ile (Fig. S3). In PAY1 (LOC_Os08g31470),which contained 5239 bases and four exons, there were no differences between Z1364 and Nipponbare.

3.6. Expression pattern of putative qSP1 and other cloned genes associated with spikelet development

Real-time qRT-PCR was used to quantify the expression of putative qSP1 at different tissues and various stages of panicle growth in Nipponbare. Putative qSP1 was expressed in all tissues, including root, stem, leaf sheath, and panicle, but particularly in 1-cm spikes(Fig.4-a).

The expression levels of seven cloned genes associated with number of kernels were measured in Z1364 and Nipponbare. The expression levels of OsMADS22, GN1A, and DST were upregulated highly significantly in Nipponbare compared with Z1364. The expression levels of LAX2, GNP1,and GHD7 were downregulated highly significantly in Nipponbare. However, the expression of DEP1 showed no difference between Nipponbare and Z1364(Fig.4-b).

4. Discussion

4.1. Increased kernel number of Z1364 was caused by increasing primary branches and secondary branches, as well as seed-set density

Rice yield is determined mainly by three factors: number of effective panicles, number of kernels per panicle, and 1000-kernel weight[22].Among these traits,the number of kernels per panicle is one of the most critical factors. It is therefore important to identify genes for the multikernel phenotype to increase rice yield.In this study,we identified a chromosome segment substitution line, Z1364, exhibiting a multikernel phenotype, with three substitution segments. Phenotypic analysis indicated that the multikernel phenotype of Z1364 reflects mainly the increased number of primary branches,the number of secondary branches, and the seed-set density(Fig. 2, Table 1). Many QTL associated with kernel development were linked to marker RM6777 on chromosome 1,including qSP1 for number of spikelets per panicle, qSSD1 for seed-set density, and qNSB1 for number of secondary branches. Progeny testing (genotype and phenotype of progeny lines in the F3generation)showed that the mean spikelets per panicle (133.89), number of secondary branches (26.23)and seed-set density(61.57)of five plants of a single segment substitution line (SSSL) lacking the qSP1 locus were greater than those (103.26, 20.58, and 51.47) of the recipient parent Nipponbare carrying the locus (Fig. 3-b). This phenomenon may be thought of as “multifactorial linkages” followed by natural selection favoring co-adapted traits [23], as confirmed by Zhao et al. [20] using single-segment substitution lines.Moreover,qSP1,qSSD1,and qNSB1 were all major QTL explaining respectively 57.34%, 87.7%, and 49.44% of the corresponding phenotypic variance. In particular, the many-kernels vs fewerkernels traits, and high- vs low-density seed-set followed a bimodal distribution in the secondary F2population and fitted a 1:3 segregation. These results indicate that the increased kernel number and high seed-set density phenotypes of Z1364 were controlled mainly by a single major QTL with recessive inheritance.

4.2. QTL for agronomic important traits identified in Z1364

Compared with the reported QTL, qSP1 and qGP1 may be the QTL qGNPP1-1 for kernel number per panicle (RM3521-RM8111) mapped by Tian et al. [24], but it has not yet been cloned.We have now identified a candidate gene for qSP1 as a novel rice homolog of AT4G32551 in Arabidopsis. However, its functions in rice are unknown. The QTL qSSR6 for seed-set rate may be associated with the FON1 gene, which controls vegetative and reproductive development. FON1 encodes a receptor kinase homologous to Arabidopsis CLAVATA1[25]and increases seed set by regulating the ABA pathway [3]. In the present study, the reading frame of FON1 (LOC_Os06g50340)contained two base changes in Z1364, resulting in the mutation of amino acid 248 from Leu to Val and the amino acid at position 450 changing from Thr to Ile. The QTL qNPB8 for primary branches per panicle and qNSB8-1 for number of secondary branches may be associated with PAY1,which was located 365 Kb from RM8264 (a marker linked to qNSB8-2).PAY1 affects auxin polar transport, and changes in endogenous indole-3-acetic acid distribution lead to increased numbers of branches and spikelets per panicle [5]. However,there were no sequence differences in PAY1 between Z1364 and Nipponbare, indicating that PAY1 did not account for the increase in branch number per panicle in Z1364 and did not show the allelic variation of qNPB8 and qNSB8-1. The QTL qNSB8-2 for number of secondary branches may be associated with the cloned gene ONAC106, localized 105 kb from RM284(a marker linked with qNSB8-2). ONAC106 reduces panicle number, number of kernels per panicle, and 1000-kernel weight by binding to the promoter regions of SGR, NYC1,OsNAC5,and LPA1,modulating the target genes of their signal pathways [21]. There were seven site differences in ONAC106(LOC_Os08g33670) between Z1364 and Nipponbare, including five single-base changes and the insertion of the three bases CAG after base 555 and the six bases GACGAC after base 584.These findings suggest that variation in ONAC106 may account for the increased secondary branches per panicle of Z1364. qRLW2 for kernel length-to-width ratio may be identical to Rlw-8-1 mapped by Zhao et al., which was in a same substitution interval (PSM394-RM556), it have the common linked marker RM284 in SSSL[26].These QTL may be relatively stable and detectable in different genetic backgrounds and environments, suggesting their potential for use in QTL pyramiding for rice breeding.

Fig.4- Expression pattern of putative qSP1 in Nipponbare and other cloned genes associated with spikelet development in Nipponbare and Z1364.Spikelets of Nipponbare and Z1364 with a length of 1 cm in Fig.4-b were taken at booting stage.

4.3.Putative qSP1 is a previously uncharacterized gene in rice

Putative qSP1 is a previously uncharacterized gene in rice,whose gene annotation is a transcription repressor LEUNIG, a homolog of Arabidopsis AT4G32551, which regulates floral organ identity [27-29]. In Z1364, the variation in qSP1 may regulate the development of the inflorescence meristem of the spikelets to increase the number of spikelets and kernels per panicle in rice. In particular, qSP1 was expressed more highly in 1-cm panicles in Nipponbare, supporting this hypothesis. In the 1-cm panicle stage, spikelet meristem development in rice is critical. Flower organ differentiation occurs mainly in the 0.15-4.00-cm panicle stage, and rapid elongation of rachis and branches occurs mainly during the 4-20 cm stage [30]. Of the genes whose expression was measured in 1-cm panicles, OsMADS22 encodes a hypothetical protein [31], which is a MADS-box gene in the STMADS11 subfamily[32]whose over-or ectopic expression may result in loss of function similar to IDS1, leading to indeterminate cell fate in the spikelet meristem [33]. Gn1a encodes an oxidase/dehydrogenase (OsCKX2) that degrades the phytohormone cytokinin[11].DST,a zinc finger protein,directly regulates the expression of OsCKX2, enhanced panicle branching and increased kernel number [34]. In Z1364, these genes were all downregulated, suggesting that reduction of their function increased spikelet number. LAX2 encodes a nuclear protein with a plant-specific conserved domain and interacts with LAX1, which maintains axillary meristem (AM) development and positively regulates branching [35]. Grain number per panicle 1 (GNP1) encodes rice GA20ox1, actively regulating cytokinin activity [36]. Ghd7 encodes CCT (CO, CO-LIKE, and TIMING OF CAB1) domain protein functions, expressed mainly in juvenile meristems, such as apical meristems,and its expression is regulated by photoperiod [37]. In Z1364,these genes were all upregulated, suggesting that their high expression increases spikelet number.The factors that govern the regulation of qSP1 genes invite further study.

5. Conclusions

A three-segment substitution line Z1364 for rice with increased kernel numbers was identified, carrying 17 QTL, of which qSP1, qSSD1, and qNSB1 were major QTL. They were inherited recessively in Z1364. qSP1 was subjected to finemapping using a 50-kb interval on chromosome 1. Putative qSP1 is a previously uncharacterized gene in rice,whose gene annotation is as a transcription repressor LEUNIG, and was expressed mainly in 1-cm panicles. Genes associated with number of kernels, OsMADS22, GN1A, and DST, were upregulated highly significantly in Nipponbare compared with Z1364 and LAX2, GNP1, and GHD7 were downregulated highly significantly in Nipponbare.

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2018.12.009.

Conflict of interest

The authors declare that they have no competing interests.

Acknowledgments

The study was supported by the National Key Research Plan Project (2017YFD0101107), the Chongqing Science and Technology Commission Special Project (cstc2016shmsztzx0032),and the Southwest University Innovation Team Project(XDJK2017A004).Professor Shizhong Xu(University of California,Riverside,USA)wrote the stem program for QTL mapping.