Introduction

Rice (Oryza sativa L) is one of the major cereal crops feeding more than three billion population and accounts for 20% of the world’s total calorie intake. The doubling of the rice productivity was achieved by adopting semi-dwarf varieties for lodging resistance in the 1960s and hybrid rice technology in the 1970s. However, the rise in the yield levels were led to the excess usage of fertilizers and pesticides causing adverse environmental effects in rice-producing areas besides shooting up cost of the cultivation. Moreover, the demand for rice is continuously increasing to meet the food requirement of the rapidly growing global population (Singh et al. 2018and Wang et al. 2020). Hence, there is an urgent need to substantially enhance the productivity of rice with the existing ever-shrinking cultivable area while addressing climate change.

Rice yield or productivity is a complex trait multiplicatively determined by three main component traits viz., grain number, panicle number, and grain weight (Sakamoto and Matsuoka 2008; Xing and Zhang 2010; Wang et al. 2020). Of these, grain number per panicle is the major yield attributing trait in rice (Wang et al. 2020). Grain number per panicle is highly variable and mainly correlated with the architecture of the panicle that includes the number of primary and secondary branches, panicle length, and percentage of filled grains. Multiple genes are involved in regulating the inheritance of the grain number trait that shows continuous variation in the segregating populations. Hence, identification of the genetic basis responsible for grain number would be of great value in the breeding of high-yielding rice varieties. During the last two decades, many Quantitative Trait Loci (QTLs)/genes viz., Gn1a (Ashikari et al., 2005), DEP1 (Huang et al. 2009), qGN4.1(Deshmukh et al. 2010)OsSPL14 (Jiao et al. 2010), NAL1 (SPIKE, GPS, LSCHL4, and SS1), GNP1 (Wu et al. 2016a, b), qGN1c (Xu et al. 2019) controlling the grain number trait have been mapped, cloned, and functionally characterized in rice. However, the molecular mechanism of grain number trait formation is still far from a clear understanding as DNA marker development and genotyping are the most time-consuming and costly procedures in conventional QTL mapping.

Next-Generation Sequencing (NGS) technologies have facilitated the development of several efficient techniques for gene or locus mapping. Among them, NGS-based QTL-seq approach has successfully identified and mapped many QTLs in rice (Takagi et al. 2013; Daware et al. 2016; Ogiso-Tanaka et al. 2017; Yaobin et al. 2018; Kadambari et al. 2018; Arikit et al. 2019; Lei et al. 2020 and Bommisetty et al. 2020), chickpea (Singh et al. 2017), cucumber (Lu et al. 2014), and tomato (Illa-Berenguer et al. 2015 and Ruangrak et al. 2019).

Identification of QTL region that consists of numerous genes of targeted trait is a challenging task. However, with the availability of high-quality reference genome and advances in DNA sequencing technologies, now it is becoming very easy to map and precisely pinpointing the underpinning genes in QTL regions. Hence, the present research work was formulated to identify genomic regions determining grain number in rice employing QTL-seq approach in F2 mapping population and confirmed the identified QTLs additionally in BC1F2 and F3 populations.

Materials and methods

Development and phenotyping of mapping populations for grain number per panicle

In the present study, BPT5204 (Samba Mahsuri) was crossed with NLR33892 (Pardhiva) to develop an F2 mapping population for mapping of grain number QTLs using the QTL-seq approach. BPT5204 is a fine-grained medium-duration high-yielding variety with moderate grain number developed at Agricultural College, Acharya NG Ranga Agricultural University (ANGRAU), Bapatla, Andhra Pradesh, India, whereas NLR33892 is a long duration, photosensitive variety, with high grain number, developed at Agricultural Research Station, ANGRAU, Nellore, Andhra Pradesh, India.

During Kharif (June to October), 2016, BPT5204 was crossed with high grain number male-parent NLR33892 to develop F1 seeds (Fig. 1). The F1 plants were screened with polymorphic simple sequence repeat (SSR) markers and true F1 plants were identified (Supplementary Fig. 1). During rabi (November to April), true F1s were selfed and simultaneously backcrossed with BPT5204 to generate F2 and BC1F1 populations, respectively. The F2 mapping population along with parents was evaluated for yield and its component traits during Kharif, 2017 while F2:3 and BC1F2 plants along with parents were raised during rabi (November to April), 2017–18 at wetland farm, S. V. Agricultural College, Acharya NG Ranga Agricultural University, Tirupati, Andhra Pradesh, India. A total number of grains including chaffy and filled grains per panicle was counted from each panicle in each plant.

Fig. 1
figure 1

Schematic representation of development of mapping population for identification of grain number QTLs in rice employing QTL-seq analysis

Full size image

Preparation of bulks and DNA isolation for whole-genome resequencing

In this experiment, 297 F2 plants of the BPT5204 x NLR33892 cross were evaluated. Based on grain number per panicle, 15 plants with extremely low grain number and 15 plants with extremely high grain number were selected as Low grain number bulk (L-bulk) and High grain number bulk (H-bulk), respectively. The genomic DNA from 100 mg fresh rice leaves of the parents and the selected extreme F2 individuals for a number of grains per panicle were extracted using DNeasy Plant Mini Kit (QIAGEN Sciences), and quantification of DNA was performed using Quant-iTPicoGreen dsDNA reagent and kits (Invitrogen). The bulk DNA samples were prepared by mixing equimolar concentrations of extracted DNA from 15 individuals of extreme phenotypic traits and pooled together for whole-genome resequencing.

QTL-seq and QTL validation

QTL-seq and QTL validation was followed as per Bommisetty et al. (2020). All generated whole-genome sequencing data of BPT5204 were used from our previous study (Lachagari et al. 2019) and NLR33892 with bulks, whole-genome sequencing data are available in the SRA database under the BioProject ID: PRJNA484092.

Statistical analysis

Single-marker analysis is the simplest method to identify the marker–trait association. Linear regression is the most commonly used single-marker analysis method as the coefficient of determination (R2) from the marker explains the phenotypic variation arising from the QTL linked to the marker (Collard et al. 2005).

Results

Phenotypic evaluation of parents

Parents (BPT5204 and NLR33892) were analyzed for test of significance using paired t test(Fig. 2A, Fig. 2B, Fig. 3a and Table1) for plant height, number of tillers per plant, number of panicles per plant, panicle length, number of grains per panicle, number of filled grains per panicle, number of chaffy grains per panicle, spikelet fertility, biological yield per plant, grain yield per plant, harvest index, grain length, grain width, the ratio of grain length to width, and 1000 grain weight. It revealed highly significant differences between BPT5204 and NLR33892 for all the characters except for harvest index and grain size, indicating the presence of variability between the parental genotypes.

Fig. 2
figure 2

A. Morphology of BPT5204 and NLR33892 at maturity grown in paddy field B. Panicle length (BPT5204:24.52 cm and NLR33892:29.23 cm)

Full size image
Fig. 3
figure 3

QTL-seq approach adopted for mapping genomic regions responsible for grain number per panicle (a) and (c) Image shows the morphological difference of BPT5204 (Low grain number parent) and NLR33892 (High grain number parent) (b) Frequency distribution of grain number per panicle of 297 F2 plants. (d) ∆SNP index for the chromosome number 12, plotted using a sliding window of 4 Mb with a step of 10 kb. The significant genomic regions are highlighted in shaded color (23–27 Mb). The statistical confidence interval under the null hypothesis of no QTLs is presented in the graphs (orange, P < 0.01 and green P < 0.05). (e) Grain size picture of the parents

Full size image
Table 1 Summary of Illumina sequencing results of parental lines and bulks for grain number in F2 population of the cross between BPT5204 and NLR33892
Full size table

Inheritance pattern of grain number per panicle in the F2 mapping population

The F2 mapping population comprising 297 plants along with BPT5204 and NLR33892 was evaluated for grain number per panicle trait during Kharif2017. The grain number per panicle varied from 102 to 582 in the F2 population, with a mean, median, and mode of 303.53, 316, and 326, respectively. Grain number per panicle recorded a skewness value of − 0.25 and a kurtosis value of 0.26. The continuous variation as well as the normal frequency distribution of grain number per panicle was observed in the mapping population (Fig. 3b). Further, grain number per panicle showed a bi-directional transgressive segregation beyond that of parental genotypes in the F2 mapping population. High variability and normal distribution of grain number per panicle in the F2 mapping population inferred the polygenic inheritance of the character.

Identification of QTLs for grain number per panicle by QTL-seq approach

The DNA sequencing outputs are summarized in Table 2. From whole-genome resequencing, 93,272,796, 10,6482,020, 124,559,926, and 91,651,766 raw reads were obtained for BPT5204, NLR33892, H-bulk and L-bulk, respectively. H-bulk and L-bulk had 43.179 and 44.39% of GC content, respectively. The mapped reads covered 92.37, 89.94, 90.02, and 89.53% of rice genome in BPT5204, NLR33892, H-bulk, and L-bulk, respectively. Later, the raw data were pre-processed using Adapter removal (version: 2.2.0). After processing, a total of 63,084,068 clean reads with 75.36% high-quality bases were obtained for H-bulk, while L-bulk recovered 63,084,068 clean bases with 78.01% high-quality base. The aligned samples and the reference genome sequence are used for variant calling and was performed using Samtools v0.1.18.

Table 2 Test of significance for 15 yield and its component traits in BPT5204and NLR33892 rice varieties
Full size table

The H-bulk contained 8339 SNPs while L-bulk had 8337 SNPs. Homozygous SNPs present in H-bulk and L-bulk were 2171 and 2091, respectively. A total of 2945 SNPs were present in the gene region and of them, only 1541 SNPs fall in the exonic regions of the gene. The SNP index was calculated for each SNP (Abe et al. 2012). SNPs with SNP index < 0.3 in both bulks were removed as they could be spurious SNPs. The SNP index of remaining SNPs calculated from each bulk was physically plotted throughout 12 rice chromosomes (Fig. 4). The Δ(SNP index) was calculated by subtracting the SNP index values in HGW-bulk by those in LGW-bulk together with the sliding windows of average SNP indices of SNPs located within a 2-Mb region and 1 Kb stepwise were also plotted. Statistical confidence intervals of ∆(SNP index) for all the SNP positions with given read depths under the null hypothesis of no QTLs were calculated and plotted along with ∆(SNP index).

Fig. 4
figure 4

Plots of SNP index of two bulks (High grain number (HGN) bulk and Low grain number (LGN) bulk) and Δ(SNP index) compared between them. (A) Read coverage of HGN bulk. (B) Read coverage of LGN bulk. (C) SNP index of HGN bulk. (D) SNP index of LGN bulk. (E) Delta SNP index (F) SNP density with 50 Kb step with P < 0.05. (G) Candidate genomic regions containing QTLs for number of grains per panicle

Full size image

Identification of genomic regions for grain number

In total, five genomic regions viz., qGN1, qGN3, qGN7, qGN9, and qGN12 on chromosomes 1, 3, 7, 9, and 12 were identified for grain number at P < 0.01 as shown in Fig. 4 and Table 3 (Supplementary Fig. 2).

Table 3 Genomic regions identified as QTLs from QTL-seq analysis in F2 population of BPT5204 x NLR33892 cross
Full size table

Confirmation of the QTL-seq-derived QTLs in F2, F3, and BC1F2 populations

To verify the QTLs identified in the present study, traditional QTL mapping was carried out using single-marker analysis for randomly selected 167 F2 plants of the total population. Twenty markers designed in the identified QTL regions and 20 previously reported genomic SSR markers (physically/genetically mapped on selected QTL regions of chromosomes 1, 3, 7, 9, and 12) were selected to screen parents (Supplementary Table 1). No polymorphism was observed between BPT5204 and NLR33892 for selected genomic regions on chromosomes 1, 3, 7, and 9 using InDel and SSR markers; this clearly explains that variation present in the segregation population was not under the control of genes present in these selected regions. One polymorphic marker, RM6953 located in the QTL region of chromosome 12, was used for genotyping of F2, F3, and BC1F2 population, and the resulting data were subjected to single-marker analysis (Fig. 5). Single-factor analysis of RM6953 revealed the significant association of the marker with grain number with a phenotypic variance of 24.58, 22.10, and 17.20% (Fig. 5 and Table 4) in F2, F2:3, and BC1F2 populations, respectively, at P value < 0.0001 and NLR33892 was the donor parent for high grain number trait (Table 4 and Fig. 5).

Fig. 5
figure 5

Representative gel picture showing segregation pattern of F2 (A) and BC1F2 (B) populations of the cross between BPT5204 and NLR33892. Linear regression analysis for RM6953 and number of grains per panicle in F2(C), F2:3(D), and BC1F2(E) populations of the cross between BPT5204 and NLR33892. Frequency of marker RM6953 genotype alleles in the F2, F2:3, and BC1F2 populations of the cross between BPT5204 and NLR33892 for number of grains per panicle (F)

Full size image
Table 4 Validation of qGN12 QTL governing grain number per panicle in F2, F2:3, and BC1F2 population of BPT5204 and NLR33892 cross
Full size table

Discussion

Rice is the major food crop feeding more than half of the global population. There is an urgent need to raise the productivity of rice to meet global food demand for the estimated nine billion people by 2050. Therefore, understanding the genetic basis, identification, and incorporation of major QTLs governing yield and its component traits into elite cultivars is essential to increase yield levels with existing agricultural land, water, and climatic conditions. Keeping this in view, the present investigation was conducted to identify QTLs governing grain number per panicle using QTL-seq analysis and their further confirmation by conventional linkage analysis.

The parents, BPT5204 and NLR33892 varieties, showed significant differences for grain number per panicle. In addition to continuous variation, grain number per panicle showed biparental transgressive segregation in the F2, F2:3, and BC1F2 population (Fig. 3b, Supplementary Fig. 3a, b). All these collectively inferred the quantitative genetic inheritance pattern of grain number per panicle in rice. Hence, this population was selected for mapping grain number QTLs with the help of NGS-based QTL-seq analysis.

In the present study, QTL-seq analysis identified five genomic regions viz., qGN1, qGN3, qGN7, qGN9, and qGN12 on chromosomes 1, 3, 7, 9, and 12, respectively, as candidate QTLs for a number of grains per panicle in rice based on ∆SNP index values at P value < 0.01 in an F2 population derived from BPT5204xNLR33892 cross. In the selected genomic region of chromosome 1, five yield traits QTLs viz., grains per panicle QTL gp1a (Hua et al. 2002), plant height QTL ph1.1 (Marri et al. 2005), panicles per plant QTL qPN1 (Tian et al. 2006), total number of spikelets per panicle QTL qTNSP-1–1 (Zhuang et al. 2002), and pollen sterility QTL Rf3 (Sattari et al. 2007) were reported previously (Supplementary Table 2). Chromosome 3 harbors two yield traits QTLs namely pollen sterility QTL S34(t) (Zhuang et al. 2005) and grain length QTL gl3a. Chromosomes 7, 9, and 12 had no previously reported yield QTLs in the selected genomic regions, suggesting they are novel and found in the present study for the first time for grain number in rice.

The grain number per panicle QTL, qGN12 (23 Mb to 27 Mb) was further confirmed through conventional QTL mapping in F2, F2:3, and BC1F2 populations of BPT5204 and NLR33892 cross. From the grain number trait mean values of the marker classes, the NLR33892 marker allele was identified as the source of the favorable allele. A total of 713 annotated genes were present in the qGN12 QTL region (23-27 Mb) as per the rice genome annotation project database (RGAP-DB). Among them, only seven genes viz., LOC_Os12g37690 (MYB family transcription factor), LOC_Os12g37970 (MYB family transcription factor), LOC_Os12g38400 (MYB family transcription factor), LOC_Os12g39330 (AP2 domain-containing protein), LOC_Os12g41060 (AP2 domain-containing protein), LOC_Os12g39640 (MYB family transcription factor) and LOC_Os12g40860 (Leucine-Rich Repeat family protein) were found to be potential candidate genes controlling grain number as similar kind of genes in different regions were found to control grain number in rice and other crops based on previous literature(Table 5).

Table 5 The shortlisted candidate genes based on previous reports in the qGN12 QTL region controlling grain number per panicle in rice
Full size table

A Myb-1 factor (Blind gene), which regulates the development of secondary meristems and inflorescence in tomato, has been isolated by positional cloning. Deshmukh et al. (2010) reported downregulation of Myb-1 transcription factor in Pusa 1266 (compact panicle and high grain number) as compared to Pusa Basmati 1(low grain number) in the qGN4-1 region. The APETALA-2-Like transcription factor OsAP2-39 regulates plant growth and seed production by maintaining abscisic acid and gibberellin balance in rice (Yaish et al. 2010). The thick tassel dwarf1 (td1) encoding leucine-rich repeat receptor-like kinase (LRR-RLK) results in extra rows of kernels in maize ear and also increases spikelet density in the tassel (Bommert et al. 2005 and Liu et al. 2019).

Genes controlling MYB family transcription factor (LOC_Os12g37690, LOC_Os12g37970, LOC_Os12g38400, and LOC_Os12g39640), AP2 domain-containing protein (LOC_Os12g39330 and LOC_Os12g41060), and leucine-rich repeat family protein (LOC_Os12g40860) were found to be expressed in the panicle, flag leaf, and flower from RicevarmapV2.0 (Fig. 6 and Supplementary Table 3). LOC_Os12g38400, LOC_Os12g39640, and LOC_Os12g40860 genes were found to be highly expressed in the panicles. Sequence differences of the potential candidate genes between the parents revealed the presence of missense variance in the exonic region of LOC_Os12g39330 gene (Table 6). Among the seven candidate genes, the highest number of missense variants was observed in LOC_Os12g41060 through Ricevarmap2.0 (Supplementary Table3). Further it was found that both the genes were involved in the production of AP2 domain-containing protein which plays an important role in panicle development and seed development in rice (Yaish et al. 2010). Hence, these genes could be considered as candidate genes controlling high grain number in rice.

Fig. 6
figure 6

Gene Expression Atlas of the shortlisted candidate genes governing grain number per panicle. Intensity of the color represents the expression value of the genes in different tissues of rice genotypes

Full size image
Table 6 Sequence differences of the potential candidate genes identified through comparing genome resequencing data of parents
Full size table

The grain number QTL identified in the present study exhibited as much as 20% phenotypic variance consistently in the F2, F2:3, and BC1F2 populations. Hence, this QTL can be considered as not only novel but also a major QTL, which can be targeted for map-based cloning to identify the underlying candidate gene. Moreover, this QTL can be transferred to low grain number varieties to enhance the yield through marker-assisted selection. Further, NLR33892 provided favorable allele, hence can be used as a potential donor for yield enhancement.

However, further validation of novel QTL is warranted before exploiting the genes predicted in the present study for deploying into elite cultivars. The novel QTLs identified in the present study undoubtedly enhance our understanding of the complex nature of the yield component traits.