http://orcid.org/0000-0001-6623-1256
Figures
Abstract
The source-sink relationship determines the ultimate grain yield of rice. In this study, we used a set of reciprocal introgression lines (ILs) derived from Xuishui09 × IR2061 to map quantitative trait loci (QTLs) that were associated with sink-, source-, and grain yield-related traits. A total of 95 QTLs influencing eight measured traits were identified using 6181 high-quality single nucleotide polymorphism markers. Nine background-independent QTLs were consistently detected in seven chromosomal regions in different genetic backgrounds. Seven QTLs clusters simultaneously affected sink-, source-, and grain yield-related traits, probably due to the genetic basis of significant correlations of grain yield with source and sink traits. We selected 15 candidate genes in the four QTLs consistently identified in the two populations by performing gene-based association and haplotype analyses using 2288 accessions from the 3K project. Among these, LOC_Os03g48970 for qTSN3b, LOC_Os06g04710 for qFLL6a, and LOC_Os07g32510 for qTGW7 were considered as the most likely candidate genes based on functional annotations. These results provide a basis for further study of candidate genes and for the development of high-yield rice varieties by balancing source–sink relationships using marker-assisted selection.
Citation: Wang Y, Wang J, Zhai L, Liang C, Chen K, Xu J (2020) Identify QTLs and candidate genes underlying source-, sink-, and grain yield-related traits in rice by integrated analysis of bi-parental and natural populations. PLoS ONE 15(8): e0237774. https://doi.org/10.1371/journal.pone.0237774
Editor: Pawan L. Kulwal, Mahatma Phule Krishi Vidyapeeth College of Agriculture, INDIA
Received: May 15, 2020; Accepted: August 3, 2020; Published: August 14, 2020
Copyright: © 2020 Wang et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the manuscript and its Supporting Information files.
Funding: This work was funded by the National Key R&D Program of China to KC (grant no. 2017YFD0100100), the National Key R&D Program of China to JX (grant no. 2016YFD0100301), the 863 Key Project from the Chinese Ministry of Science & Technology to JX (grant no. 2014AA10A601), the National Natural Science Foundation of China to YW (grant no. 31671602) and the Agricultural Science and Technology Innovation Program and the Cooperation and Innovation Mission to JX (grant no. CAAS-ZDXT202001). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist
Introduction
Rice (Oryza sativa L.) is one of the most important staple cereals in the world, especially in Asia. Due to continuously increasing population, diminishing cultivated area, and other factors, a nearly 40% increase in grain yield will be necessary to meet global food demand during the next 40 years [1]. Grain yield of rice is mainly determined by the coordination between sources and sinks [2]. Spikelet number per panicle and grain size are regarded as the major sinks that absorb and utilize photosynthate; tissues such as the topmost three leaves (especially the flag leaf) are often defined as the most important primary sources that produce photosynthate. A good balance between source strength and sink strength is a prerequisite for high yield potential.
To date, a number of efforts have been made to identify QTLs underlying source-, sink-, and yield-related traits by using genome-wide association and linkage analyses [3–5]. A large number of genes related to sources, sinks, and the source–sink relationship have been cloned, including sink-related genes, such as Gn1a [6], DEP1 [7], Ghd7 [8], GNP6 [9], and GNP1 [10] for spikelet number, GW5 [11], qTGW3 [12], and GS3 [13] for grain size and grain weight, source-related genes such as NAL9 [14] for leaf area, and source–sink relationship genes NAL1 (SPIKE, GPS, and LSCHL4) [15–18]. Notably, several QTLs influencing source leaves (e.g., flag leaves) have been mapped to chromosomal regions closely affecting sink capacities (e.g., spikelet number per panicle and thousand-grain weight) and yield-related traits (e.g., panicle number), hinting at possible pleiotropy or tight linkage of genes/QTLs affecting source-, sink-, and yield-related traits [3, 4]. Detection of these QTLs is still crucial to gain a better understanding of the genetic basis and the relationship between source- and sink-related traits.
QTL mapping results largely depend on genetic background [19]. Many QTLs detected in mapping populations can’t be detectable in breeding population, thus impeding their application in molecular breeding. So identification of background-independent QTLs or integration of QTL mapping with molecular breeding in the same population will largely minimize effect of background on QTL detection. Genetic analysis using reciprocal introgression lines (ILs) derived from the same parents would be more helpful to understand complex traits [19–22], especially for studying genetic background effect on QTL detection [19–22]. In view of this consideration, we constructed a set of reciprocal introgression lines derived from Xuishui09 × IR2061 and identified background-independent QTLs underlying source-, sink-, and yield-related traits. We selected candidate genes from the important QTL intervals and validated the candidate genes based on gene-based association and haplotype analyses using 2288 rice accessions. Our study has shed new light on the genetic bases of source-sink related traits, and the results provide valuable information for molecular rice breeding by design for high yield potential.
Materials and methods
Materials
Xiushui09 is a high-yield semi-dwarf japonica (geng) rice variety from China while IR2061-520-6-9 (abbreviated as IR2061) is a drought tolerance indica (xian) breeding line introduced from the International Rice Research Institute (IRRI), Los Baños, Philippines. The two parents are genetically significant differences in grain yield, leaf trait, panicle trait, cold tolerance, drought tolerance and heat tolerance which are certainly worth of studying. Therefore, we chose Xuishui09 and IR2061 as the two parents to construct a set of reciprocal introgression lines (ILs) [21]. Specifically, the F1 plants were simultaneously backcrossed to the two parents to develop two BC1F1 populations, with each having around 100 plants. Twenty-five individuals randomly selected from BC1F1 plants were then used as the male parent to backcross with the corresponding parent to produce two BC2F1 populations. The BC progenies were selfed for five consecutive generations without selection, yielding lines from BC2F2 to BC2F6. Finally, two sets of ILs were developed, consisting of 226 BC2F6 lines in the Xiushui09 background (XS-ILs) and 233 BC2F6 lines in the IR2061 background (IR-ILs).
Phenotypic investigation
All of the ILs and their parents were planted in Sanya (SY, 18.3°N, 109.3°E), Hainan province during December 2016–April 2017 and Hangzhou (HZ, 30.3°N, 120.2°E), Zhejiang province during May–October 2017. Each line was grown in a plot of two rows with 12 plants in each row at a spacing of 25 × 17 cm, with three replicates. The fields were managed according to the local standard practices. At the full heading stage, flag leaf width (FLW, in cm), flag leaf length (FLL, in cm) were measured on the main stems of five uniform plants from each replicate of each line. Flag leaf area (FLA, in cm2) was calculated as FLL × FLW × 0.75 [4]. At maturity, eight uniform plants in the middle row of each replicate of each line were sampled and dried in an oven at 70°C for five days followed by trait measurements of panicle number per plant (PN), total spikelet number per panicle (TSN), filled grain number per panicle (FGN), thousand-grain weight (TGW, in g), and grain yield per plant (GY, in g).
Phenotypic data analysis
To dissect the effects of genotype-by-environment interaction (GEI) on the tested traits, variance components were estimated using multiple-site analysis, with all effects treated as random. Heritability across environments was computed using the estimated variance components as VG/(VG + VGEI/s +Ve/sr), where VG, VGEI, and Ve are the variances of genotype, genotype-by-environment interaction variance, and residual error, respectively; s is the number of environments, and r is the number of replicates. All of the analyses were conducted with the PBTools package (http://bbi.irri.org/products) developed by IRRI. A linear mixed model was used to integrate the phenotypic data from two environments; the line’s genotype effect was treated as the fixed effect, and the environment and replicate within the environment were treated as random effects. The lme4 in R package was used to calculate the best linear unbiased predictors (BLUPs) of each line. BLUP values of each line were used for analysis of phenotypic correlation using the rcorr function implemented in the R package Hmisc [23].
Genotyping
Genomic DNA for SNP genotyping was isolated from approximately 100 mg fresh leaf samples of five-week-old seedlings from the two sets of ILs using a modified cetyltrimethylammonium bromide (CTAB) method [24]. Genotyping was performed using a customized rice 56 K SNP array containing 56,897 SNPs screened from the 3K Rice Genome Project [25, 26]. Target DNA preparation, chip hybridization, and array processing were conducted by Capital Bio-Technology (Beijing, China) according to the Affymetrix Axiom® 2.0 assay protocol. A three-step filtering strategy was applied to select high quality SNP markers from 56K SNP genotypic data for QTL mapping. Firstly, markers that were non-polymorphic between the two parents were removed. Secondly, all heterozygous genotypes were set as “missing” and markers with more than 10% missing values were removed. Thirdly, markers with a minor allele frequency (MAF) less than 3% were removed. Finally, 6181 high quality SNP markers were selected for QTL mapping for the two sets of ILs.
QTL mapping
BLUP values of each line were used for QTL mapping of source-, sink-, and grain yield-related traits. A total of 6181 high-quality SNP markers were used to conduct the QTL identification. Linkage maps of XS-ILs and IR-ILs were constructed separately using the Map functionality in IciMapping [27]. Firstly, SNPs with a known chromosome ID on a physical map were anchored. The unanchored SNP markers were grouped based on a minimum logarithm of odds (LOD) score of 3. Secondly, a combined algorithm of nearest neighbor and two-opt algorithm of traveling salesman problem was used for marker ordering. Finally, the sum of adjacent recombination frequencies calculated with a five-marker window size was used as the rippling criterion. Recombination frequency was converted into map distance by the Kosambi mapping function. QTL mapping was performed using the software IciMapping V4.1 [27]. A probability level of the logarithm of the odds (LOD) value greater than 3.0 was used as the threshold for declaring the presence of a QTL.
Identification of candidate genes
A total of 2288 worldwide accessions, which originated from 89 countries, generated from the 3,000 Rice Genomes Project [26] were collected to conduct gene-based association analysis and haplotype analysis to identify candidate genes. Based on the known population structure and division of subpopulations [26], the panel contained 1361 xian accessions (198 xian-1A, 177 xian-1B, 235 xian-2, 245 xian-3, and 506 xian-adm), 628 geng accessions (207 geng -tmp, 80 geng-sbtrp, 279 geng-trp, and 62 geng-adm), 176 aus/boro, 52 basmati/sadri and 71 highly admixture (adm) accessions (S1 Table). FLL, TGW, and TSN values of 2288 accessions were measured during December 2016–April 2017 in SY (S1 Table) using the same method as described above.
The following four steps were conducted to identify candidate genes for QTLs consistently detected in the two sets of ILs. Firstly, the genes located in the candidate regions for QTLs consistently detected in the two sets of ILs were extracted, and gene functions were annotated by the Rice Annotation Project Database (https://rapdb.dna.affrc.go.jp/). Then, all available SNPs located inside these genes were searched from 4.8 M SNPs data generated from 3K RGP in the Rice SNP-Seek Database [28]. The SNPs with a minor allele frequency less than 0.05 and/or a missing data rate exceeding 20% were removed, and the remaining high-quality SNPs located inside these genes were used to perform gene-based association analyses via compressed mixed linear model using principal component analysis (PCA) and kinship matrix (K). The analyses were performed with the Genome Association and Prediction Integrated Tool (GAPIT), a package of R software [29]. The critical P-value for declaring a significant marker-trait association was 10−5. Thirdly, haplotype analyses were conducted using all of the available non-synonymous SNPs located inside the gene CDS region searched from 18 M SNPs data generated from 3K RGP in the Rice SNP-Seek Database [28]. Candidate genes were determined by testing the significant differences among major haplotypes (containing more than 30 samples) for each important QTL through analysis of variance (ANOVA). Duncan’s multiple mean comparison test was carried out to determine the significance of any differences (5% significance level) using the agricolae package in R. Finally, the most likely candidate genes were selected based on their biologically related functions.
Results
Basic statistics of markers
The same set of 6181 well-distributed polymorphic SNP markers was available for genotyping of the two sets of ILs. The number of SNPs per chromosome ranged from 375 on chromosome 9 to 763 on chromosome 1 (S2 Table). The sizes of chromosomes varied from 23.0 Mb (chromosome 9) to 43.3 Mb (chromosome 1). The full genome size was 373.1 Mb, with an average marker spacing of 61.0 kb ranging from 55.7 kb on chromosome 2 to 72.8 kb on chromosome 12 (S2 Table).
Genome compositions of the two sets of ILs
Most of the genomes of the reciprocal ILs were similar to their recurrent parents, with XS-ILs sharing on average 86.7% of the genome with the recurrent parent Xiushui09 (range from 44.6 to 100%), whereas IR-ILs shared on average 90.7% of the genome with the recurrent parent IR2061 (range from 58.5 to 100%) (S1 Fig). Obviously, the reciprocal introgression line populations were skewed toward one parent or the other, as expected.
Phenotypic variation and correlation
Although Xiushui09 exhibited significantly greater FLW than IR2061, the FLA value did not differ between the two parents, because the former had significantly smaller FLL than the latter in SY (Fig 1). In HZ, Xiushui09 showed significantly larger FLA compared with IR2061, because the FLW of the former was significantly greater than that of the latter, and the FLL value did not differ between the two parents. The TGW of Xiushui09 was significantly higher than that of IR2061, whereas the PN of Xiushui09 was significantly lower than that of IR2061 in the two environments. In SY, Xiushui09 showed significantly lower FGN and similar TSN compared with IR2061, finally resulting in significantly lower GY compared with IR2061. In HZ, Xiushui09 exhibited significantly higher FGN and TSN, thus resulting in significantly higher GY compared with IR2061 (Fig 1). It was observed that the two sets of ILs had wide ranges of segregation, i.e., there were transgressive segregations for the eight traits (Fig 1).
FLL, flag leaf length; FLW, flag leaf width; FLA, flag leaf area; PN, panicle number per plant; TSN, total spikelet number per panicle; FGN, filled grain number per panicle; TGW, thousand-grain weight; GY, grain yield per plant; SY, Sanya; HZ, Hangzhou; XS-ILs, introgression lines in Xiushui09 background; IR-ILs, introgression lines in IR2061 background; *, **, *** indicate significant differences between Xiushui09 and IR2061 at P < 0.05, P < 0.01, and P < 0.001, respectively; ns indicates non-significance.
In the two sets of ILs, five traits (FLL, FLW, FLA, TSN, and TGW) were controlled mainly by genotypes (VG), and the heritabilities of the five traits ranged from 0.38 for FLL to 0.70 for TGW, and from 0.33 for TSN to 0.64 for TGW in XS-ILs and IR-ILs, respectively. FGN, PN, and GY were more sensitive to environmental factors, with gene and environment together potentially affecting phenotypic variation. The heritability of the three traits ranged from 0.04 for GY to 0.23 for PN, and 0.14 for GY to 0.31 for FGN in XS-ILs and IR-ILs, respectively (S3 Table).
In the two sets of ILs, the trends of pairwise phenotypic correlations were similar (Fig 2). The FLA was significantly positively correlated with sink size (TSN) but strongly negatively correlated with PN, and there was no significant correlation with sink capacity (TGW). GY showed significant positive correlations with FGN and PN in the two populations. GY was also significantly positively correlated with TGW in XS-ILs, but there was no significant correlation with TGW in IR-ILs, implying that PN and FGN made larger contributions to GY than TGW in IR-ILs. There were negative or non-significant correlations among the three grain yield component traits (PN, FGN, and TGW) in the two populations. As expected, FLA showed significant positive correlations with its corresponding component traits (FLL and FLW), and it was also significantly positively correlated with GY in the two sets of ILs (Fig 2).
Correlations between the eight evaluated traits in introgression lines (ILs) in the Xuishui09 background (XS-ILs) (A) and the IR2061 background (IR-ILs) (B). FLL, flag leaf length; FLW, flag leaf width; FLA, flag leaf area; PN, panicle number per plant; FGN, filled grain number per panicle; TGW, thousand-grain weight; TSN, total spikelet number per panicle; GY, grain yield per plant. The colors of square correspond to values of the corresponding r. *, **, *** indicate significant correlations at P < 0.05, P < 0.01, and P < 0.001, respectively.
QTL mapping
In total, 95 QTLs were identified for the eight yield-related traits in the two sets of ILs derived from the Xiushui09 × IR2061 cross, ranging from 6 for GY and PN to 17 for TSN. Among these, 45 QTLs were detected only in XS-ILs, and 41 only were found in IR-ILs. Nine QTLs in seven chromosomal regions were commonly identified in the two sets of ILs, with one QTL (qFLW4b) for FLW, two QTLs (qFLL5c and qFLL6a) for FLL, one QTL (qFLA4c) for FLA, two QTLs (qTGW5b and qTGW7) for TGW, and three QTLs (qTSN3b, qTSN4b, and qTSN6b) for TSN. The IR2061 alleles at the qTGW7 and qTSN6b loci were associated with increased TGW and TSN, and they accounted for 4.6% and 4.3% of TGW, and 10.9% and 4.7% of TSN in XS-ILs and IR-ILs, respectively. The Xuishui09 alleles at the remaining loci resulted in increased corresponding trait values, and they accounted for 2.5% to 26.5% of the phenotypic variance (Table 1 and Fig 3).
Red, blue, and black indicate the QTLs affecting source-, sink-, and grain yield-related traits, respectively. Green rectangles indicate the QTL regions simultaneously affecting source-, and sink-/grain yield-related traits in the same population. Asterisks indicate that the QTLs were consistently identified in the two sets of introgression lines. Numbers on the chromosome indicate physical position of each QTL.
Chromosomal regions simultaneously harboring QTLs for source, sink, and grain yield traits
Comparisons of 43 QTLs affecting source flag leaf size (FLL, FLW, and FLA), 31 QTLs influencing sink capacities (TGW and TSN), and 21 QTLs affecting grain yield (GY) and its related traits (FGN and PN) revealed seven chromosomal regions simultaneously influencing source and sink/yield-related traits in the same population (Table 1 and Fig 3). These seven regions comprised the region of 41.85–42.06 Mb on chromosome 1 harboring qFLA1b for FLA and qPN1 for PN detected in IR-ILs and the Xuishui09 alleles increased FLA at qFLA1b but decreased PN at qPN1; the region of 27.70–28.07 Mb on chromosome 3 harboring qTSN3b, qFLW3b and qFLA3c for TSN, FLW and FLA, respectively, detected in IR-ILs, and the Xuishui09 alleles at the three QTLs consistently resulted in increased corresponding trait values; the region of 24.47–24.92 Mb on chromosome 4 harboring qFLW4a for FLW, qFLA4b for FLA, and qTSN4a for TSN detected in IR-ILs, and the Xuishui09 alleles at the three QTLs consistently increased FLW, FLA, and TSN; the region of 31.15–31.50 Mb on chromosome 4 harboring qFLW4b, qFLA4c, qTSN4b, qFGN4b, and qPN4 for FLW, FLA, TSN, FGN, and PN, respectively, detected in XS-ILs, and the Xuishui09 alleles resulted in decreased PN at qPN4 but increased the corresponding phenotypic trait values at the remaining the four QTLs; the region of 27.73–28.08 Mb on chromosome 5 harboring qFLL5c for FLL, qFLA5b for FLA, and qPN5 for PN detected in XS-ILs, and the Xuishui09 alleles increased FLL at qFLL5c and FLA at qFLA5b but decreased PN at qPN5; the region of 27.40–27.97 Mb on chromosome 6 harboring qTSN6b, qFLW6b, qFLA6a, and qPN6b for TSN, FLW, FLA, and PN, respectively, detected in XS-ILs, and the IR2061 alleles decreased PN at qPN6b but increased the trait values at the remaining three QTLs; the region of 30.19–30.79 Mb on chromosome 6 harboring qFLL6b for FLL, qFLW6c for FLW, and qTGW6c for TGW in IR-ILs, and the IR2061 alleles at three QTLs resulted in increased trait values (Fig 3).
Performance of yield and its related traits of a near-isogenic IL XC10 and Xiushui09
In this study, five chromosomal regions (27.70–28.07 Mb on chromosome 3, 24.47–24.92 Mb on chromosome 4, 31.15–31.50 Mb on chromosome 4, 27.40–27.97 Mb on chromosome 6, and 30.19–30.79 Mb on chromosome 6) simultaneously influencing source- and sink- related traits were identified in the same population (Fig 3). Among these chromosomal regions, the Xuishui09 alleles at the first three were associated with increased corresponding source-, and sink- trait values, whereas the IR2061 alleles at the latter two chromosomal regions increased trait values (Table 1). Based on selection against foreground favorable allele introgressions from the donor parent and background of the recurrent parent, one near-isogenic introgression line, XC10, was chosen from the XS-ILs by marker-assisted selection (MAS). Specifically, the IL XC10 carried the corresponding introgressed IR2061 alleles at qTSN6b/qFLW6b/qFLA6a located in the region of 27.40–27.97 Mb on chromosome 6 but had no IR2061 allele introgressions at the other QTLs for TSN, FLW, or FLA; otherwise the line showed 93.7% of the genome of Xuishui09 (Fig 4A). XC10 showed significant increases in FLA (+19.1% and 15.1%), FGN (+42.7% and 29.3%), and TSN (+28.2% and 26.7%) along with slightly lower but non-significant differences in TGW and PN compared with the recurrent parent Xuishui09 in SY and HZ. Finally, XC10 produced 21.3% and 18.7% higher grain yield than Xuishui09 in SY and HZ, respectively (Fig 4B), indicating that introgression of the IR2061 favorable alleles at qTSN6b/qFLW6b/qFLA6a contributed to increased grain yield due to increases in the corresponding source and sink traits.
(A) Graphical genotype of Xuishui09, IR2061 and XC10, respectively. Yellow green and dark gray bars denote chromosome regions from Xuishui09 and IR2061, respectively. (B) Comparison of grain yield and its related traits between XC10 and Xuishui09 in Sanya and Hangzhou. *, **, *** indicate significant differences at P < 0.05, P < 0.01, and P < 0.001, respectively; ns indicates non-significance.
Candidate gene analysis
In this study, nine QTLs in the seven chromosomal regions were consistently identified in the two sets of ILs. We searched the Oryzabase online resource (https://shigen.nig.ac.jp/rice/oryzabase/) for known rice genes that co-localized with the nine QTLs. Three known genes (APO1, NAL1, and GW5) at five QTLs (qTSN6b, qTSN4b/qFLA4c/qFLW4b, and qTGW5b) were related to the corresponding traits.
We conducted gene-based association and haplotype analyses to identify candidate genes for the remaining four QTLs (qFLL5c, qFLL6a, qTGW7, and qTSN3b) using 2288 accessions of 3K. The 2288 accessions presented substantial variations for the FLL, TGW, and TSN in SY (S2 Fig). A total of 184 annotated genes located in the four QTL regions (qFLL5c, qFLL6a, qTGW7, and qTSN3b) according to the Rice Genome Annotation Project Database (RAP-DB). Of 2366 SNPs available in the 184 genes, a total of 32 significant SNPs located inside the 22 genes were identified to significantly associate with the corresponding traits (FLL, TSN, or TGW) by gene-based association analysis (Fig 5). Among the 22 genes, 15 genes had significant differences in corresponding traits among different haplotypes based on haplotype analyses (Table 2). Finally, the 15 genes were screened as candidate genes.
Gene-based association analyses of targeted genes related to qTSN3b (A), qFLL5c (B), qFLL6a (C) and qTGW7 (D). Each green point show a SNP located in the genes in QTL intervals. Red lines show the threshold employed to determine significant SNPs.
For qTSN3b, this locus was mapped in the region of 27.70–28.07 Mb on chromosome 3 containing 63 annotated genes according to the Rice Genome Annotation Project Database (RAP-DB). A total of 1231 SNPs in 63 genes were used for association analysis. Based on gene-based association analysis results, four genes (LOC_Os03g48600, LOC_Os03g48970, LOC_Os03g49020, and LOC_Os03g49180) were used for haplotype analysis (Table 2). Significant differences in TSN were detected among different haplotypes for the four candidate genes (Table 2). Of these, LOC_Os03g48970 is identical to OsNF-YA4, encoding a Nuclear Factor Y (NF-Y) transcription factor in rice [30]. Two haplotypes of OsNF-YA4 were found, and Hap1 (G) was associated with significantly more TSN than Hap2 (A) (Fig 6A).
(A) Gene structure (left) and total spikelet number per panicle of different haplotypes (right) of LOC_Os03g48970 at qTSN3b. (B) Gene structures (left) and flag leaf length of different haplotypes (right) of LOC_Os05g48450, LOC_Os05g48590, LOC_Os05g48760, and LOC_Os05g48870 at qFLL5c. (C) Gene structure (left) and flag leaf length of different haplotypes (right) of LOC_Os06g04710 at qFLL6a. (D) Gene structure (left) and thousand-grain weight of different haplotypes (right) of LOC_Os07g32510 at qTGW7. FLL, flag leaf length; TSN, total spikelet number per panicle; TGW, thousand-grain weight. The ** and *** denote significance of ANOVA at P < 0.01 and P < 0.001, respectively. Letters above violins (a, b and c) are ranked by Duncan’s test at P < 0.05.
In the region from 27.73 to 28.08 Mb on chromosome 5 harboring qFLL5c, 750 SNPs in 60 genes were used for association analysis. According to the gene-based association analysis results, six genes (LOC_Os05g48450, LOC_Os05g48560, LOC_Os05g48590, LOC_Os05g48670, LOC_Os05g48760 and LOC_Os05g48870) were selected for the haplotype analysis (Fig 5B). LOC_Os05g48560 had only one haplotype containing more than 30 rice accessions. No significant differences in FLL were found among different haplotypes for LOC_Os05g48670 (Table 2). Highly significant differences in FLL were detected among different haplotypes at the remaining four candidate genes (LOC_Os05g48450, LOC_Os05g48590, LOC_Os05g48760, and LOC_Os05g48870) (Table 2 and Fig 6B).
In the region from 2.02 to 2.31 Mb on chromosome 6 harboring qFLL6a, 249 SNPs in 51 genes were used for association analysis. Based on gene-based association analysis results, a total of 10 genes were used for haplotype analysis (Fig 5C). LOC_Os06g04940 was identified in only one haplotype containing more than 30 rice accessions (Table 2). No significant haplotypes were found in LOC_Os06g04800, LOC_Os06g04960, LOC_Os06g04980, or LOC_Os06g05080 (Table 2). Highly significant differences for FLL among different haplotypes were observed at five genes (LOC_Os06g04690, LOC_Os06g04710, LOC_Os06g04840, LOC_Os06g05060 and LOC_Os06g05160) (Table 2). Among those genes, LOC_Os06g04710 is identical to dwarf and deformed flower 1 (ddf1-1), encoding an F-box protein in rice [31]. Four haplotypes of ddf1-1 were found, and Hap 1 and Hap 3 exhibited significantly larger FLL than the other two haplotypes (Fig 6C).
The locus qTGW7 was mapped in the region of 19.32–19.42 Mb on chromosome 7 containing 10 annotated genes. A total of 136 SNPs in 10 genes were used for association analysis. According to gene-based association analysis results, two genes (LOC_Os07g32470 and LOC_Os07g32510) were used for haplotype analysis (Fig 5D). Significant differences in TGW were detected among different haplotypes at the two candidate genes (Table 2). LOC_Os07g32510 encodes a Dof zinc finger domain protein. Four haplotypes of LOC_Os07g32510 were found, and the TGW values of Hap 1 and Hap 2 were significantly larger than those of Hap 3 and Hap 4 (Fig 6D).
Discussion
Effect of genetic background on detection of grain yield QTLs
QTLs influencing grain yield and its related traits are usually sensitive to genetic background and environment [19]. Among the 95 QTLs affecting yield and its related traits identified in this study, only nine (9.5%) QTLs shared by the two different populations in the two different genetic backgrounds, indicating that most of the QTLs (90.5%) were very specific to the genetic background. The results clearly suggest that most yield QTLs detected in one population would not be identified in another population, i.e., highly genetic background effects existed that influenced the detection of QTLs for yield and related traits [19]. The QTL × genetic background interaction has been considered as one of the major factors limiting the use of QTLs in MAS for quantitatively inherited traits, as evidenced by effects of QTL dep1 (dense and erect panicle1) on grain yield varied with the backgrounds [7, 32–34]. Therefore, it is indispensable to identify background-independent QTLs influencing yield for breeding high-yield rice by MAS, as shown in this study. Here, the nine QTLs (qFLL5c, qFLL6a, qFLW4b, qFLA4c, qTSN3b, qTSN4b, qTSN6b, qTGW5b, and qTGW7) consistently identified in the two different genetic backgrounds might certainly be made a priority for enhancement of rice production via MAS.
Candidate gene identification of important QTLs
The region 27.70–28.07 Mb on chromosome 3 harboring qTSN3b contains four candidate genes, including OsNF-YA4 (LOC_Os03g48970), encoding a Nuclear Factor Y (NF-Y) transcription factors in rice [30]. OsNF-YA4 is involved in regulating the development of the rice panicle, affecting panicle number, grain number per panicle, grain filling, and grain weight. OsNF-YA4 overexpressed transgenic plants exhibited abnormal yield component values, where grain filling rate, TSN, PN, and TGW were significantly decreased compared to the control [30]. In the present study, two haplotypes of OsNF-YA4 were identified, and haplotype G showed significantly more TSN than haplotype A (Fig 6A). Therefore, OsNF-YA4 (LOC_Os03g48970) is considered as the most likely candidate gene for qTSN3b.
For qFLL6a, dwarf and deformed flower 1 (ddf1-1, LOC_Os06g04710) that encodes an F-box protein in rice is the most likely candidate gene. The ddf1-1 mutant exhibited significantly shorter and thinner stems, leaves, and roots than the wild type [31]. Genetic complementation of the ddf1-1 mutant showed that all the mutant phenotypes were rescued, including plants with dwarf stature and smaller stems, leaves, roots, and panicles, as well as aberrant spikelets [31]. In the present study, four haplotypes of ddf1-1 were detected, of which Hap 1 and Hap 3 were associated with significantly larger FLL than the other two haplotypes (Hap 2 and Hap 4) (Fig 6C).
Of the two candidate genes for qTGW7, LOC_Os07g32510 is a Dof zinc finger domain protein. The Dof proteins, a class of zinc finger domain transcription factors, have been reported to participate in regulation of gene expression in seed storage protein synthesis in developing endosperm [35, 36]. It has been reported that seed weight increased in Arabidopisis due to overexpressing one of two soybean Dof transcription factor genes GmDof4 and GmDof11 [37]. In rice, grain size and weight were decreased in the antisense transgenic lines of Dof daily fluctuations 1 (RDD1) encoding a Dof transcription factor gene [38]. In the present study, four haplotypes of LOC_Os07g32510 were detected, with Hap 1 and Hap 2 being associated with significantly larger TGW values than the other two haplotypes (Fig 6D), suggesting that LOC_Os07g32510 is a likely candidate gene of qTGW7 that probably affects TGW in rice.
Comparison of QTLs detected in this study with previously cloned genes
Of the 95 QTLs for source-, sink- and grain yield-related traits, ten QTLs were located in the same or adjacent regions containing previously reported cloned genes in rice (Table 1). For example, qFLW4b for FLW, qFLA4c for FLA, qTSN4b for TSN, qFGN4b for FGN, and qPN4 for PN were mapped in the region of 31.15–31.50 Mb on chromosome 4 that harbored the previously reported NAL1 associated with pleiotropic effects on multiple traits related to the source (e.g., leaf width, leaf chlorophyll content, and photosynthetic efficiency), the sink (e.g., TSN), and the PN in diverse genetic backgrounds [15–18, 39]; qTGW3c in the region 35.37–35.77 Mb on chromosome 3 and qTGW5b in the region 5.29–5.41 Mb on chromosome 5 were co-located with qTGW3 [12] and GW5 [11] associated with rice grain size and grain weight, respectively; qFGN7 in the region 9.11–9.21 Mb and qTSN7d in the region 29.57–29.67 Mb on chromosome 7 were close to Ghd7 [8] and Ghd7.1 [40] regulating spikelet number, plant height and heading date in rice; qTSN6b for TSN was co-located in the region 27.40–27.97 Mb on chromosome 6 harboring APO1 for the number of primary rachis branches that ultimately increase the harvest index and grain yield in rice [41]. Allelism of the above QTLs for yield and its related traits detected in this study with the previously reported genes awaits further verification using fine mapping and QTL cloning.
Application in rice breeding for high yield potential
In the history of rice breeding, the balance between sources (e.g., FLL, FLW, and FLA) and sinks (e.g., TSN and TGW) plays an important role in guiding the breeding of super high-yield rice varieties. The significant positive correlations of FLL, FLW and FLA with TSN in the two populations were in agreement with previous studies [4]. However, no significant correlations were observed between the three flag leaf traits and TGW except between FLW and TGW in XS-ILs in the two sets of ILs (Fig 2). In this study, we identified five QTL clusters (in the regions of 27.70–28.07 Mb on chromosome 3, 24.47–24.92 Mb on chromosome 4, 31.15–31.50 Mb on chromosome 4, 27.40–27.97 Mb on chromosome 6, and 30.19–30.79 Mb on chromosome 6) simultaneously influencing the source- and sink-related traits, hinting at possible pleiotropy or tight linkage. SPIKE (NAL1) located at a QTL cluster (qFLW4b/qFLA4c/qTSN4b/qFGN4b) in the region of 31.15–31.50 Mb on chromosome 4 showed pleiotropic effects in regulating the development of multiple traits related to source (leaf width, leaf chlorophyll content, and photosynthetic efficiency) and sink (spikelet number per panicle) in different genetic backgrounds [15–18]. NILs with the superior NAL1 allele achieved 18.7% and 18% yield increases compared with their controls 9311 and IRRI146 [15, 17], respectively. Therefore, we inferred that it is possible to improve the grain yield by introgression of favorable alleles of the above five QTL clusters (qFLW3b/qFLA3c/qTSN3b, qFLW4a/qFLA4b/qTSN4a, qFLW4b/qFLA4c/qTSN4b/qFGN4b, qFLW6b/qFLA6a/qTSN6b, and qFLW6c/qFLL6b/qTGW6c) into geng or xian varieties to develop super high-yield rice cultivars, as demonstrated by the IL XC10 from this study (Fig 4).
In general, large-panicle cultivars exhibit greater plant size, larger FLL and FLW, and fewer numbers of tillers [42]. In the present study, strong negative correlations of FLL, FLW, FLA and TSN with PN were observed in the two populations (Fig 2). The significant negative correlation between TSN and PN, which is also known as the concept ‘yield component compensation’ can be partly explained by the developmental trade-off between tiller or panicle number and organ size (plant height, leaf size, and panicle size) or physiological competition for nitrogen during spikelet differentiation in rice [43, 44]. In this study, we identified two QTL clusters in the region of 31.15–31.50 Mb on chromosome 4 and 27.40–27.97 Mb on chromosome 6 underlying the development of multiple traits related to source (FLW and FLA), sink (TSN), and grain yield-related traits (PN and FGN) in XS-ILs, and the alleles increasing FLW, FLA, TSN, and FGN decreased PN. In addition, another two QTL clusters in the region of 41.85–42.06 Mb on chromosome 1 and 27.73–28.08 Mb on chromosome 5 were identified that simultaneously influenced leaf size and PN with opposite directions of additive effect. However, only a few overlapping QTL clusters affecting PN and the other source- and sink-related traits were detected in the two populations, suggesting that most QTLs influencing PN and the other source- and sink-related traits are genetically independent. Therefore, one strategy for improvement of yield could be via deploying or pyramiding different favorable alleles of independent QTLs for PN and the other source- and sink-related traits to well coordinate source and sink relationships.
Conclusions
We identified 95 QTLs for the yield- traits, nine of which were consistently detected in seven chromosomal regions in the two set of ILs, and seven QTL clusters that simultaneously affected source-, sink-, and grain yield-related traits were detected in the same population. A total 15 candidate genes for the four QTLs consistently detected in the two populations were identified and the most likely candidate genes for three QTLs (qTSN3b, qFLL6a, and qTGW7) were inferred from gene-based association analysis, haplotype analysis and functional annotation. Favorable alleles of genetic background-independent QTLs for source-, sink, and grain yield-related traits were mined. These results provide an effective guide to developing varieties with high yield potential by balancing source–sink dynamics via marker-assisted selection.
Supporting information
S1 Table. Details of the 2288 accessions investigated and raw data of yield related traits measured in Sanya.
https://doi.org/10.1371/journal.pone.0237774.s001
(XLSX)
S2 Table. Distribution of SNP markers on 12 chromosomes of rice.
https://doi.org/10.1371/journal.pone.0237774.s002
(PDF)
S3 Table. Variance components and heritabilty estimated by multiple-site analysis.
https://doi.org/10.1371/journal.pone.0237774.s003
(PDF)
S1 Fig. Frequency distribution of the Xuishui09 genome in the reciprocal Introgression Lines (ILs) in Xuishui09 (XS) and IR2061 (IR) backgrounds.
https://doi.org/10.1371/journal.pone.0237774.s004
(PDF)
S2 Fig. Box plots of flag leaf length, thousand-grain weight, and total spikelet number per panicle in 2288 accessions.
https://doi.org/10.1371/journal.pone.0237774.s005
(PDF)
References
- 1. Murchie EH, Pinto M, Horton P. Agriculture and the new challenges for photosynthesis research. New Phytol. 2009; 181: 532–552. pmid:19140947
- 2. Xiao Y, Chang TG, Song QF, Wang S, Tholen D, Wang Y, et al. ePlant for quantitative and predictive plant science research in the big data era—Lay the foundation for the future model guided crop breeding, engineering and agronomy. Quant Biol. 2017; 5: 260–271. https://doi.org/10.1007/s40484-017-0110-9
- 3. Cui KH, Peng SB, Xing YZ, Yu SB, Xu CG, Zhang Q. Molecular dissection of the genetic relationships of source, sink and transport tissue with yield traits in rice. Theor Appl Genet. 2003; 106: 649–658. pmid:12595994
- 4. Wang Y, Pang Y, Chen K, Zhai L, Shen C, Wang S, et al. Genetic bases of source-, sink-, and yield-related traits revealed by genome-wide association study in Xian rice. Crop J. 2020; 8: 119–131. https://doi.org/10.1016/j.cj.2019.05.001
- 5. Descalsota-Empleo GI, Amparado A, Inabangan-Asilo MA, Tesoro F, Stangoulis J, Reinke R, et al. Genetic mapping of QTL for agronomic traits and grain mineral elements in rice. Crop J. 2019; 7: 560–572. https://doi.org/10.1016/j.cj.2019.03.002
- 6. Ashikari M, Sakakibara H, Lin S, Yamamoto T, Takashi T, Nishimura A, et al. Cytokinin oxidase regulates rice grain production. Science. 2005; 309: 741–745. pmid:15976269
- 7. Huang X, Qian Q, Liu Z, Sun H, He S, Luo D, et al. Natural variation at the DEP1 locus enhances grain yield in rice. Nat Genet. 2009; 41: 494–497. pmid:19305410
- 8. Xue W, Xing Y, Weng X, Zhao Y, Tang W, Wang L, et al. Natural variation in Ghd7 is an important regulator of heading date and yield potential in rice. Nat Genet. 2008; 40: 761–767. pmid:18454147
- 9. Zhang Z, Sun X, Ma X, Xu B, Zhao Y, Ma Z, et al. GNP6, a novel allele of MOC1, regulates panicle and tiller development in rice. Crop J. 2020; https://doi.org/ 10.1016/j.cj.2020.04.011
- 10. Wu Y, Wang Y, Mi XF, Shan JX, Li XM, Xu JL, et al. The QTL GNP1 encodes GA20ox1, which increases grain number and yield by increasing cytokinin activity in rice panicle meristems. PLoS Genet. 2016; 12: e1006386. pmid:27764111
- 11. Weng J, Gu S, Wan X, Gao H, Guo T, Su N, et al. Isolation and initial characterization of GW5, a major QTL associated with rice grain width and weight. Cell Res. 2008; 18:1199–1209. pmid:19015668
- 12. Hu Z, Lu SJ, Wang MJ, He H, Sun L, Wang H, et al. A novel QTL qTGW3 encodes the GSK3/SHAGGY-Like kinase OsGSK5/OsSK41 that interacts with OsARF4 to negatively regulate grain size and weight in rice. Mol Plant. 2018; 11: 736–749. pmid:29567449
- 13. Mao H, Sun S, Yao J, Wang C, Yu S, Xu C, et al. Linking differential domain functions of the GS3 protein to natural variation of grain size in rice. Proc Natl Acad Sci USA. 2010; 107: 19579–19584. pmid:20974950
- 14. Li W, Wu C, Hu GC, Xing L, Qian W, Si H, et al. Characterization and fine mapping of a novel rice narrow leaf mutant nal9. J Integr Plant Biol. 2013; 55: 1016–1025. pmid:23945310
- 15. Fujita D, Trijatmiko KR, Tagle AG, Sapasap MV, Koide Y, Sasaki K, et al. NAL1 allele from a rice landrace greatly increases yield in modern indica cultivars. Proc Natl Acad Sci USA. 2013; 110: 20431–20436. pmid:24297875
- 16. Takai T, Adachi S, Taguchi-Shiobara F, Sanoh-Arai Y, Iwasawa N, Yoshinaga S, et al. A natural variant of NAL1, selected in high-yield rice breeding programs, pleiotropically increases photosynthesis rate. Sci Rep. 2013; 23: 2149. pmid:23985993
- 17. Zhang GH, Li SY, Wang L, Ye WJ, Zeng DL, Rao YC, et al. LSCHL4 from japonica cultivar, which is allelic to NAL1, increases yield of indica super rice 93–11. Mol Plant. 2014; 7:1350–1364. pmid:24795339
- 18. Wang Y, Zhai L, Chen K, Shen C, Liang Y, Wang C, et al. Natural sequence variations and combinations of GNP1 and NAL1 determine the grain number per panicle in rice. Rice. 2020; 13:14. pmid:32112146
- 19. Zheng TQ, Wang Y, Ali AJ, Zhu LH, Sun Y, Zhai HQ, et al. Genetic effects of background-independent loci for grain weight and shape identified using advanced reciprocal introgression lines from Lemont × Teqing in rice. Crop Sci. 2011; 51: 2525–2534. https://doi.org/10.2135/cropsci2011.05.0259
- 20. Wang Y, Zhang Q, Zheng T, Cui Y, Zhang W, Xu J, et al. Drought-tolerance QTLs commonly detected in two sets of reciprocal introgression lines in rice. Crop Pasture Sci. 2014; 65:171–184. PMID: 22181322
- 21. Cheng L, Wang Y, Meng L, Hu X, Cui Y, Sun Y, et al. Identification of salt-tolerant QTLs with strong genetic background effect using two sets of reciprocal introgression lines in rice. Genome. 2012; 55: 45–55. pmid:22181322
- 22. Mei HW, Xu JL, Li ZK, Yu XQ, Guo LB, Wang YP, et al. QTLs influencing panicle size detected in two reciprocal introgressive line (IL) populations in rice (Oryza sativa L.). Theor Appl Genet. 2006; 112: 648–656. pmid:16331475
- 23.
Harrell Jr FE. Hmisc: Harrell Miscellaneous, R package version 3. 2015; https://cran.r-project.org.
- 24. Murray MG, Thompson WF. Rapid isolation of high molecular weight plant DNA. Nucleic Acids Res. 1980; 8: 4321–4325. pmid:7433111
- 25. Meng L, Wang B, Zhao X, Ponce K, Qian Q, Ye G. Association mapping of ferrous, zinc, and aluminum tolerance at the seedling stage in indica rice using MAGIC populations. Front Plant Sci. 2017; 8:1822. pmid:29123537
- 26. Wang W, Mauleon R, Hu Z, Chebotarov D, Tai S, Wu Z, et al. Genomic variation in 3,010 diverse accessions of Asian cultivated rice. Nature. 2018; 557:43–49. pmid:29695866
- 27. Meng L, Li H, Zhang L, Wang J. QTL IciMapping: integrated software for genetic linkage map construction and quantitative trait locus mapping in biparental populations. Crop J. 2015; 3: 269–283. https://doi.org/10.1016/j.cj.2015.01.001
- 28. Alexandrov N, Tai S, Wang W, Mansueto L, Palis K, Fuentes RR, et al. SNP-Seek database of SNPs derived from 3000 rice genomes. Nucleic Acids Res. 2015; 43: D1023–1027. pmid:25429973
- 29. Lipka AE, Tian F, Wang Q, Peiffer J, Li M, Bradbury PJ, et al. GAPIT: genome association and prediction integrated tool. Bioinformatics. 2012; 28:2397–2399. pmid:22796960
- 30. Lee DK, Kim HI, Jang G, Chung PJ, Jeong JS, Kim YS, et al. The NF-YA transcription factor OsNF-YA7 confers drought stress tolerance of rice in an abscisic acid independent manner. Plant Sci. 2015; 241:199–210. pmid:26706071
- 31. Duan Y, Li S, Chen Z, Zheng L, Diao Z, Zhou Y, et al. Dwarf and Deformed Flower 1, encoding an F-box protein, is critical for vegetative and floral development in rice (Oryza sativa L.). Plant J. 2012; 72: 829–842. pmid:22897567
- 32. Wang J, Nakazaki T, Chen S, Chen W, Saito H, Tsukiyama T, et al. Identification and characterization of the erect-pose panicle gene EP conferring high grain yield in rice (Oryza sativa L.). Theor Appl Genet. 2009; 119: 85–91. pmid:19407986
- 33. Yi X, Zhang Z, Zeng S, Tian C, Peng J, Li M, et al. Introgression of qPE9-1 allele, conferring the panicle erectness, leads to the decrease of grain yield per plant in japonica rice (Oryza sativa L.). J Genet Genomics. 2011; 38: 217–223. pmid:21621743
- 34. Zhou Y, Zhu JY, Li ZY, Yi CD, Liu J, Zhang HG, et al. Deletion in a quantitative trait gene qPE9-1 associated with panicle erectness improves plant architecture during rice domestication. Genetics. 2009; 183: 315–324. pmid:19546322
- 35. Mena M, Vicente-Carbajosa J, Schmidt RJ, Carbonero P. An endosperm-specific DOF protein from barley, highly conserved in wheat, binds to and activates transcription from the prolamin-box of a native B-hordein promoter in barley endosperm. Plant J. 1998; 16:53–62. pmid:9807827
- 36. Qi X, Li S, Zhu Y, Zhao Q, Zhu D, Yu J. ZmDof3, a maize endosperm-specific Dof protein gene, regulates starch accumulation and aleurone development in maize endosperm. Plant Mol Biol. 2017; 93:7–20. pmid:27709320
- 37. Wang H, Zhang B, Hao Y, Huang J, Tian A, Liao Y, et al. The soybean Dof-type transcription factor genes, GmDof4 and GmDof11, enhance lipid content in the seeds of transgenic Arabidopsis plants. Plant J. 2007; 52:716–729. pmid:17877700
- 38. Iwamoto M, Higo K, Takano M. Circadian clock- and phytochrome-regulated Dof-like gene, Rdd1, is associated with grain size in rice. Plant Cell Environ. 2009; 32: 592–603. pmid:19210638
- 39. Yano K, Yamamoto E, Aya K, Takeuchi H, Lo PC, Hu L, et al. Genome-wide association study using whole-genome sequencing rapidly identifies new genes influencing agronomic traits in rice. Nat Genet. 2016; 48: 927–934. pmid:27322545
- 40. Yan W, Liu H, Zhou X, Li Q, Zhang J, Lu L, et al. Natural variation in Ghd7.1 plays an important role in grain yield and adaptation in rice. Cell Res. 2013; 23:969–971. pmid:23507971
- 41. Terao T, Nagata K, Morino K, Hirose T. A gene controlling the number of primary rachis branches also controls the vascular bundle formation and hence is responsible to increase the harvest index and grain yield in rice. Theor Appl Genet. 2010; 120: 875–893. pmid:20151298
- 42. Wu LL, Liu ZL, Wang JM, Zhou CY, Chen KM. Morphological, anatomical, and physiological characteristics involved in development of the large culm trait in rice. Aust J Crop Sci. 2011; 5:1356–1363. https://doi.org/10.1016/j.fcr.2011.04.016
- 43. Grafius JE. Multiple Characters and Correlated Response1. Crop Sci. 1978; 18: 931–934. https://doi.org/10.2135/cropsci1978.0011183X001800060004x
- 44. Adriani DE, Lafarge T, Dardou A, Fabro A, Clement-Vidal A, Yahya S, et al. The qTSN positive effect on panicle and flag leaf size of rice is associated with an early down-regulation of tillering. Front Plant Sci. 2016; 6:1197. pmid:26779230