Figures
Abstract
Grain length is an important quantitative trait in rice (Oryza sativa L.) that influences both grain yield and exterior quality. Although many quantitative trait loci (QTLs) for grain length have been identified, it is still unclear how different alleles from different QTLs regulate grain length coordinately. To explore the mechanisms of QTL combination in the determination of grain length, five mapping populations, including two F2 populations, an F3 population, an F7 recombinant inbred line (RIL) population, and an F8 RIL population, were developed from the cross between the U.S. tropical japonica variety ‘Lemont’ and the Chinese indica variety ‘Yangdao 4’ and grown under different environmental conditions. Four QTLs (qGL-3-1, qGL-3-2, qGL-4, and qGL-7) for grain length were detected using both composite interval mapping and multiple interval mapping methods in the mapping populations. In each locus, there was an allele from one parent that increased grain length and another allele from another parent that decreased it. The eight alleles in the four QTLs were analyzed to determine whether these alleles act additively across loci, and lead to a linear relationship between the predicted breeding value of QTLs and phenotype. Linear regression analysis suggested that the combination of eight alleles determined grain length. Plants carrying more grain length-increasing alleles had longer grain length than those carrying more grain length-decreasing alleles. This trend was consistent in all five mapping populations and demonstrated the regulation of grain length by the four QTLs. Thus, these QTLs are ideal resources for modifying grain length in rice.
Citation: Zeng Y, Ji Z, Wen Z, Liang Y, Yang C (2016) Combination of Eight Alleles at Four Quantitative Trait Loci Determines Grain Length in Rice. PLoS ONE 11(3): e0150832. https://doi.org/10.1371/journal.pone.0150832
Editor: Tongming Yin, Nanjing Forestry University, CHINA
Received: September 24, 2015; Accepted: February 20, 2016; Published: March 4, 2016
Copyright: © 2016 Zeng 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 paper and its Supporting Information files.
Funding: Funded by The National Natural Science Foundation of China (31101004, 31221004), the National High-tech Research and Development Program (863 Program, 2012AA101201), and a fund from the Chinese Academy of Agricultural Sciences to the Scientific and Technical Innovation Team. 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 a staple food crop for more than half of the world’s population. Therefore, rice yield is the primary objective of rice breeding programmes, because food security is continuously challenged by multiple factors, including increasing population, reduced arable land, global climate change, and increasing demand for biofuel production [1]. Grain shape (or size) is a key determinant of grain yield, because it is closely correlated with grain weight, one of the three major components (grain number, panicle number, and grain weight) that determine rice yield [1, 2].
Grain shape (or size) in rice is characterized by a combination of four parameters: grain length, grain width, grain length-to-width ratio, and grain thickness [2]. These four parameters are typical polygenic traits controlled by quantitative trait loci (QTLs). Being a hot point in rice breeding programmes, grain shape has been extensively studied using various genetic-based approaches. Many QTLs for grain shape and grain weight have been mapped, and some of them have been also cloned and characterized. At least six QTLs for grain length have been fine mapped including qGRL1 [3], qSS7 [4], GS7 [5], qGL7 [6], qGL4b [7], and qGL-3a [8]; at least nine QTLs for grain weight have been fine mapped including gw3.1 [9], GW3 [10], GW6 [10], gw8.1 [11], gw9.1 [12], tgw11 [13], GW1-1 [14], GW1-2 [14], and qTGW3.2 [15]; and a total of nine QTLs for grain size have been isolated and cloned including GS3 [16, 17], GS5 [18], qGL3/qGL3.1/OsPPKL1 [19–21], GW2 [22], GW5/qSW5 [23, 24], GW8 [25], GIF1 [26], TGW6 [27], and qGW7/GL7 [28, 29].
The characterization of cloned QTLs suggests that multiple signaling pathways, such as ubiquitination-mediated proteasomal degradation, phytohormones, and G-protein signaling pathways, are involved in the determination of grain length [1]. For instance, GW2 and GW5/qSW5 that encode a RING-type E3 ubiquitin ligase and a nuclear protein that interacts with polyubiquitin [22–24], respectively, participate in the ubiquitination-mediated proteasomal degradation pathway; and TGW6 that encodes a protein with indole-3-acetic-acid (IAA)-glucose hydrolase activity participates in the phytohormone pathway and regulates the level of free IAA in grains [27].
Although the underlying molecular mechanisms of grain size regulation have been elucidated in rice, information regarding the relationship between different QTLs is limited. The relationship of four QTLs for seed size (GS3, GW2, GW5/qSW5, and GIF1) was studied by RNA interference technology using reverse transcription polymerase chain reaction, and it was found that GW2 and GW5/qSW5 positively regulate the expression of GS3, and that GIF1 is positively regulated by GW5/qSW5, but negatively regulated by GS3 and GW2 [30]. Rice plants carrying both GW2 and GW5/qSW5 alleles showed a significant increase in grain width compared to those carrying one of the two alleles, suggesting that GW2 and GW5/qSW5 may participate in independent pathways [31]. The development and study of the gs3/gw8 double mutant in a near-isogenic background [25] showed that GS3 and GW8 also participate in independent pathways [1].
A large number of QTLs for grain length have been identified using various mapping populations [2]. However, the regulation of rice grain length by different alleles at the grain length-related QTLs is still poorly understood. In this study, we first developed three primary QTL mapping populations (two F2 and a F3) by crossing two rice varieties, ‘Lemont’ and ‘Yangdao 4’. These three mapping populations had been used initially to detect sheath blight resistance QTLs in a previous study [32]. QTL analysis using these primary mapping populations identified four grain length-related QTLs. Interestingly, the grain length was coordinately regulated by the eight alleles at these four QTLs. We then used two permanent mapping populations (a F7 and a F8, developed by crossing ‘Lemont’ and ‘Yangdao 4’) to further test whether the four QTLs act additively across loci, leading to a linear relationship between predicted breeding value of QTLs and phenotype, or act epistatically leading to a clearly nonlinear relationship. Using data collected in five mapping populations, we have thus demonstrated the regulation of rice grain length by the combination of eight alleles at four QTLs.
Materials and Methods
Mapping Populations
‘Lemont’, a U.S. tropical japonica variety, was crossed with ‘Yangdao 4’, a Chinese indica variety, to develop five mapping populations, including two F2 populations, an F3 population, an F7 recombinant inbred line (RIL) population, and an F8 RIL population. ‘Lemont’ and ‘Yangdao 4’ were provided by Dr. Xinghua Wei at China National Rice Research Institute (CNRRI).
The F2 populations with 190 and 182 plants each were sown in May 2011 and May 2012, respectively, in Fuyang, Hangzhou (119°95′E, 30°07′N), at the farm of CNRRI, arranged with 6 plants in each row with spacing of 17 and 20 cm between plants and between rows, respectively. The F3 population with 160 lines that derived from the former F2 population was sown in November 2012 in Lingshui, Hainan (110°02′E, 18°48′N), at the trial station of CNRRI. A total of 18 plants were grown from each of the 160 lines, arranged in three rows of six plants each. The three mapping populations were also used in a previous study to detect sheath blight resistance QTLs [32].
The F7 RIL population with 220 lines was sown in May 2014 in Fuyang, Hangzhou (119°95′E, 30°07′N), at the farm of CNRRI, while the F8 RIL population that derived from the F7 RIL population was sown in November 2014 in Lingshui, Hainan (110°02′E, 18°48′N), at the trial station of CNRRI. A total of 18 plants were grown from each of the 220 lines in both RIL populations, arranged in three rows of six plants each. Field management included all the common agronomic practices in Hangzhou and Hainan.
Measurement of Grain Length
Harvested rice grains were sun-dried and stored at room temperature for at least 1 month before the measurement of grain length using a vernier caliper. Ten fully filled grains were randomly selected from the upper half of the panicle of each individual plant in the F2 populations and measured. Ten plants within each line of the F3 population were randomly selected, and five grains per plant (50 grains in total per line) were measured. In the F7 and F8 RIL population, 16 and 25 grains, respectively, were randomly selected from each line and measured. In all mapping populations, the average grain length was used for analysis.
Molecular Marker Assays and QTL Analysis
DNA extraction and PCR were performed as described by Zhang et al. [33] and Zeng et al. [34], respectively. Briefly, a total of 179 polymorphic markers were used in QTL analysis. All the 179 markers are co-dominant markers, including simple sequence repeat and insertion-deletion markers [34]. Composite interval mapping (CIM) was used to identify QTLs for grain length using Windows QTL Cartographer 2.5 (http://statgen.ncsu.edu/qtlcart/WQTLCart.htm). The threshold of the logarithm of odds (LOD) that declares the existence of a QTL was determined based on 1,000 permutations (P < 0.05). Inclusive composite interval mapping (ICIM) was used to detect digenic epistatic loci using QTL IciMapping v. 4.0.6.0 [35] based on 1,000 permutations (P < 0.05). The digenic epistatic loci were also confirmed by two-way analysis of variance (ANOVA).
A fast and simple method based on CIM was used to identify QTLs for grain length in the five mapping populations. First, a genome-wide linkage map that was constructed using 179 polymorphic markers [32] was used to identify QTLs for grain length in the F2 population grown in 2011 in Hangzhou, and in the F3 population grown in 2012 in Hainan; this revealed three QTLs. Second, the three detected QTLs for grain length were further examined in the F2 population grown in 2012 in Hangzhou, based on a linkage map that was constructed using 44 markers (S1 Table). These 44 markers covered the regions of the three QTLs for grain length that were detected in the F2 population grown in 2011 in Hangzhou, and in the F3 population, and represented a total of 443.3 cM with an average of 13.4 cM between adjacent markers. The three QTLs were not repeatedly detected, but a new QTL was identified, yielding a total of four QTLs in the three primary mapping populations. Third, the four QTLs for grain length detected in the two F2 and the F3 populations were further examined in the F7 and F8 RIL populations using 21 markers (S1 Fig). The 21 markers covered the regions of all the four QTLs for grain length that were detected in the two F2 and the F3 populations, and represented a total of 112.1 cM with an average of 6.6 cM between adjacent markers.
Since the multiple interval mapping (MIM) method may provide greater power and precision for QTL mapping [36], we used it to detect QTLs, and compared it with the CIM method in the five mapping populations using Windows QTL Cartographer 2.5. In all the mapping populations, we used the following procedure to build a MIM model for QTL analysis based on the instructions provided by Silva et al. [37]. First, the MIM forward search method was used to create an initial MIM model because it is more powerful than the other options. The Bayesian information criterion (BIC)-based model selection criteria (BIC-M0 in the software) was used, with a MIM search walk speed of 1 cM. Second, several MIM model refinement rounds was performed to build a final MIM model. The model refinement rounds consisted of (1) searching repeatedly for QTLs with main effects until no main effect QTL were found, (2) searching repeatedly for pairs of epistatic QTLs until no further epistatic QTLs were found, (3) testing epistasis effects and excluding all the epistatic QTLs that were not statistically significant, (4) testing for the main effects of QTLs and excluding all the main effect QTLs that were not statistically significant, and (5) optimizing positions of both main and epistatic QTLs.
Results
Identification and Confirmation of QTLs for Grain Length
The Shapiro-Wilk test was performed to determine whether the phenotypic data were normally distributed (F2 population grown in 2011 in Hangzhou, W = 0.98, P = 0.04; F2 population grown in 2012 in Hangzhou, W = 0.99, P = 0.36; F3 population, W = 0.99, P = 0.58; F7 recombinant RIL population, W = 0.99, P = 0.32; and F8 RIL population, W = 0.99, P = 0.41). The apparent normal distribution of grain length in four of the five mapping populations indicated that grain length was a quantitative trait controlled by polygenes (Fig 1).
(a) F2 population grown in 2011 in Hangzhou, (b) F2 population grown in 2012 in Hangzhou, (c) F3 population grown in 2012 in Hainan, (d) F7 recombinant inbred line (RIL) population grown in 2014 in Hangzhou, and (e) F8 RIL population grown in 2014 in Hainan.
Three QTLs (qGL-3-1, qGL-3-2, and qGL-4) were detected by CIM (Table 1, S2 Fig), using the F2 population grown in 2011 in Hangzhou, while no QTLs were detected using the F3 population. None of the three QTLs were repeatedly detected in the F2 population grown in 2012 in Hangzhou; however, a new QTL, qGL-7, was identified (Table 1, S3 Fig). The putative alleles from Yangdao 4 increased grain length at qGL-3-1 and qGL-4, while in comparison the alleles from Lemont decreased the grain length. The putative alleles from Lemont increased grain length at qGL-3-2 and qGL-7, whereas a relative decrease was observed with the alleles from Yangdao 4 (Table 1).
No QTLs were detected in the F3 population grown in 2012 in Hainan.
We used two RIL populations to further confirm the four QTLs (qGL-3-1, qGL-3-2, qGL-4, and qGL-7) detected in the three primary mapping populations. Three of the four detected QTLs (qGL-3-2, qGL-4, and qGL-7) were consistently identified by CIM in the F7 and F8 RIL populations (Table 1, S4 Fig). Although qGL-3-1 was not detected by CIM, the LOD score curve from marker D309 showed a peak at the qGL-3-1 region in both RIL populations (S4A Fig).
The MIM method then compared with the CIM method. We found that MIM was more sensitive than CIM in detecting QTL; this is shown in Table 2, and indicates that more QTLs were identified using MIM. Neither method detected any QTL in the F3 population planted in 2012 in Hainan. All the four QTLs identified using CIM were also detected using MIM (Table 1, Table 2). The four QTLs detected using CIM (i.e. qGL-3-1, qGL-3-2, qGL-4, qGL-7), corresponded to the four QTLs detected using MIM (i.e. qGL-3-1MIM, qGL-3-2MIM, qGL-4MIM, qGL-7-2MIM), respectively (Table 1, Table 2). The ‘MIM’ followed after the QTL name denoted QTL detected using MIM, and was used to differentiate these QTLs from those detected using CIM. The nearest markers to the four QTLs detected using CIM or MIM were the same except for one QTL, qGL-7 or the corresponding qGL-7-2MIM, which was detected in the F2 population grown in 2012 in Hangzhou. The nearest marker to qGL-7 was D755, while the nearest marker to qGL-7-2MIM was RM234 (Table 1, Table 2). The physical distance between D755 and RM234 is about 1,400 kb according to the Nipponbare reference sequence. qGL-7 and qGL-7-2MIM were both located at the D755-RM234 interval; therefore, qGL-7 and qGL-7-2MIM were considered to represent the same QTL. It indicated that the CIM results were consistent with the MIM results, but MIM was more sensitive than CIM. The following analysis focused on the four QTLs detected using both CIM and MIM, i.e. qGL-3-1MIM, qGL-3-2MIM, qGL-4MIM, and qGL-7-2MIM.
No QTLs were detected in the F3 population grown in 2012 in Hainan.
QTL-By-Environment and QTL-By-Population Interactions
Two-way ANOVA was conducted to examine whether the four grain length-QTLs had significant interaction with the environments or populations. The markers closest to the four QTLs were used to represent the QTL genotypes at the specific populations. The heterozygotes in F2 populations were omitted, and were not used in analysis.
Among the four grain length-QTLs, some QTLs were not detected in some of the mapping populations (Table 1, Table 2). We determined the most likely marker to represent the QTL for those populations by analyzing only the LOD score curves in CIM results, because these results were consistent with those from MIM analysis. For qGL-3-1 (or qGL-3-1MIM) that was detected only in the F2 population grown in 2011 in Hangzhou using either CIM or MIM, D309 was the closest marker (Table 1, Table 2, S2A Fig). Although this QTL was not detected in the F7 and F8 RIL populations by CIM or MIM, the LOD score curve from D309 showed a peak (S4A Fig) at this QTL region. Therefore, the D309 genotype was used to represent the qGL-3-1 genotype for the other four mapping populations. For qGL-3-2 (or qGL-3-2MIM) that was detected in four populations using MIM, both D336B and RM3585 were the closest markers (Table 1, Table 2). We chose D336B as the closest marker to this QTL for the F3 population planted in 2012 in Hainan, because D336B was the most proximal marker in the F7 and F8 RIL populations analyzed using CIM or MIM, and RIL populations are generally considered more reliable than F2 populations. For qGL-4 (or qGL-4MIM) that was detected in four populations using MIM, D463 was the closest marker to this QTL in three populations (Table 2).Therefore, D463 was chosen as the closest marker to this QTL for the F3 population planted in 2012 in Hainan. The closest markers for qGL-7 (or qGL-7-2MIM) were differed according to the mapping populations derived from CIM or MIM analysis (Table 1, Table 2). To identify the most suitable marker, the LOD score curves in the F7 and F8 RIL populations were analyzed. Two peaks were observed in the RIL populations, that is, one peak in the F7 RIL population and another peak in the F8 RIL population, which were co-located at the RM234 position (S4D Fig). Therefore, the RM234 genotype was used to represent the qGL-7 genotype for the F3 population planted in 2012 in Hainan.
ANOVA showed that there was no significant interaction between each QTL and the environment (Table 3). Significant interaction was not detected between each QTL and mapping population (Table 4).
In different mapping populations, the closest Markers to the four grain length-QTLs (qGL-3-1MIM, qGL-3-2MIM, qGL-4MIM, and qGL-7-2MIM) were used to represent the QTL genotypes at the specific populations. Since there were five mapping environments, the degree of freedom for environment is four. There were two genotypes at each QTL (the heterozygotes were not used in analysis), and the degree of freedom for QTL is 1.
In different mapping populations, the closest Markers to the four grain length-QTLs (qGL-3-1MIM, qGL-3-2MIM, qGL-4MIM, and qGL-7-2MIM) were used to represent the QTL genotypes at the specific populations. Since there were three types of mapping populations (F2, F3, and RIL), the degree of freedom for population is two. There were two genotypes at each QTL (the heterozygotes were not used in analysis), and the degree of freedom for QTL is 1.
Digenic Epistasis in Five Mapping Populations
ICIM was used to identify epistatic loci for grain length across all 12 chromosomes (S2 Table); these loci were further confirmed by two-way ANOVA (S3 Table). Four pairs of digenic epistatic loci were detected in the F2 population grown in 2011 in Hangzhou (S5 Fig), while no digenic epistatic loci were detected in the F3 population and the F2 population grown in 2012 in Hangzhou. Digenic epistatic loci were not detected in RIL mapping populations, probably due to the low number of markers used for genotyping.
MIM detected only a pair of epistatic QTLs in the F2 population grown in 2011 in Hangzhou, i.e., qGL-4MIM by qGL-7-1MIM interaction, which explained 1.93% of the epistatic variation, and the dominance-by-dominance effect was -0.24. MIM did not detect other epistatic loci in the other four mapping populations. Overall, these results show that epistasis does play a role in the regulation of grain length.
The Four Grain-Length-QTLs Act Additively across Loci in Regulation of Grain Length
Digenic epistasis was not detected between the four grain-length-QTLs in all the five mapping populations, indicating that the four QTLs acted predominantly in an additive manner. The additive effects of the QTLs were similar within each mapping population (Table 1), revealing that the four QTLs might have similar effects on the regulation of grain length.
In each of the four loci (qGL-3-1, qGL-3-2, qGL-4, and qGL-7), there was an allele from one parent that increased grain length and another allele from another parent that decreased it. The four grain length-increasing alleles were qGL-3-1YD, qGL-3-2LE, qGL-4YD, and qGL-7LE, respectively, in the four loci. The four grain length-decreasing alleles were qGL-3-1LE, qGL-3-2YD, qGL-4LE, and qGL-7YD, respectively, in the four loci. The ‘LE’ or ‘YD’ suffix that follows the QTL name indicates whether the putative allele was from Lemont or Yangdao 4, respectively. To determine whether the eight alleles in the four QTLs act additively across loci, and lead to a linear relationship between predicted breeding values of QTLs and phenotype, linear regression analysis was performed.
First, the genotypic value of each plant in five mapping populations was calculated based on genotypes of the four QTLs, and was used as the predicted breeding value of the four QTLs. Then, linear regression analysis between the genotypic value and the grain length of all the individual plants in the five populations was performed. The genotypic value of each plant was determined by adding up the estimated additive effects of each of the four QTLs in the RIL populations, or by adding up the estimated additive effects and dominance effects of the four QTLs in the F2 populations. The dominance effects of the heterozygotes were used and summed up with the additive effects of the homozygotes across four loci in the F2 populations. When calculating genotypic values, the positive additive effect was used if a locus carried a grain length-increasing allele, and the negative additive effect was used if a locus carried a grain length-decreasing allele. The estimated additive effects or dominance effects of the four putative QTLs in the five mapping populations were determined by the MIM method using Windows QTL Cartographer 2.5.
The linear regression analysis indicated a clear linear relationship between predicted breeding value and phenotype, and yielded five regression equations in the five populations (Fig 2; F2 population grown in 2011 in Hangzhou, F = 86.39, P < 0.0001; F2 population grown in 2012 in Hangzhou, F = 71.04, P < 0.0001; F3 population, F = 42.39, P < 0.0001; F7 recombinant RIL population, F = 91.53, P < 0.0001; and F8 RIL population, F = 70.46, P < 0.0001). The coefficient of determination (R2) was used as an estimate of the cumulative heritability of the four QTLs. The cumulative heritability of the four QTL genotypes was 36%, 28%, 25%, 35%, and 29%, respectively, in the five mapping populations. The results shown in Fig 2 suggest that the grain length was coordinately regulated by the eight alleles in the four QTLs. Plants carrying more grain length-increasing alleles had longer grain length than those carrying more grain length-decreasing alleles.
(a) F2 population grown in 2011 in Hangzhou, (b) F2 population grown in 2012 in Hangzhou, (c) F3 population grown in 2012 in Hainan, (d) F7 RIL population grown in 2014 in Hangzhou, and (e) F8 RIL population grown in 2014 in Hainan. The genotypic value of an individual plant was calculated by adding up the estimated additive effects of all the four QTLs in the RIL populations, or by adding up the estimated additive effects and dominance effects of the four QTLs in the F2 populations. The dominance effects of the heterozygotes were used and summed up with the additive effects of the homozygotes across four loci to determine genotypic values in the F2 populations. When calculating genotypic values, the positive additive effect was used if a locus carried a grain length-increasing allele, and the negative additive effect was used if a locus carried a grain length-decreasing allele.
Discussion
Regulation of Grain Length by Eight Alleles at Four QTLs in Five Mapping Populations
Grain length is one of the most important agronomic traits in rice breeding and production [1, 2, 38] because it is positively correlated with grain weight, and influences both rice yield and other market values [2]. Long, slender grains are generally preferred by consumers in southern China, the USA, and other Southern or Southeast Asian countries; however, consumers in Japan, Korea, and northern China prefer short, round rice grains [2].
In the present study, we evaluated and analyzed five mapping populations grown under different environmental conditions and found that grain length was regulated by at least four QTLs: qGL-3-1, qGL-3-2, qGL-4, and qGL-7. Regression analysis revealed that the eight alleles at the four QTLs act additively in the regulation of grain length, leading to a linear relationship between predicted breeding value and phenotype. These results were consistent in all five mapping populations, demonstrating the regulation of grain length by the four QTLs (Fig 2).
Stability of Four QTLs in Five Mapping Populations
QTL analysis is a genetic research approach that can reveal the underlying genetic mechanisms controlling the agronomic traits [39]. However, QTLs, particularly those with minor effects, are easily affected by environmental factors [40]. In this study, it seemed that qGL-3-1 was hard to detect, because it was only detected in the F2 population grown in 2011 in Hangzhou using both CIM and MIM method, while the other three QTLs (qGL-3-2, qGL-4, and qGL-7) were more easily identified (Table 1, Table 2). However, significant QTL-by-environment interaction was not detected (Table 3); this was also the case regarding QTL-by-population interaction (Table 4). Although qGL-3-1 was not detected easily by CIM or MIM, its effect on grain length was stable in all mapping populations and environments (Fig 2).
Comparison of the Rice QTLs Associated with Grain Shape between Present and Previous Studies
The grain shape-related QTLs that have been fine-mapped or validated on chromosome 3, chromosome 4, and chromosome 7 have been listed in three physical maps, and compared with the grain length-QTLs identified in this study (Figs 3–5). The marker intervals of the four grain length-QTL were determined according to the MIM results (Table 2). The marker intervals for the four grain length-QTLs were D307-D309, D336B-RM3585, D460A-RM1113, and RM234-D755, respectively. We did not find any grain shape-related QTLs that had been fine-mapped at the qGL-3-1 or the qGL-4 region (Fig 3, Fig 4). The qGL-3-2 was co-localized with a thousand-grain weight QTL, qTGW3.2, which was reported by Tang et al. [15]. However, it is not clear whether qGL-3-2 and qTGW3.2 are allelic based on available information. The qGL-7 was co-located with a grain shape-related gene, SRS1 [41]. And it was close to four grain shape-related genes, including GS7 [5], qSS7 [4], qGW7 [28] and GL7 [29] (Fig 5), The SRS1, GL7 and qGW7 have been cloned. GL7 and qGW7 are allelic, and both correspond to the LOC_Os07g41200 gene, which encodes a TONNEAU1–recruiting motif protein [28, 29].
1. Fan et al. [16]; 2. Wan et el. [8]; 3. Li et al. [9]; 4. Guo et al. [10]; 5. Zhang et al. [21]; 6. Tang et al. [15]; 7. QTL mapped in this study. Numbers to the left of the chromosome bar indicate physical position (Mb) of the corresponding markers.
1. Wang et al. [26]; 2. Kato et al. [7]; 3. Tanabe et al. Plant Cell. 2005; 17:776–790; 4. QTL mapped in this study. Numbers to the left of the chromosome bar indicate physical position (Mb) of the corresponding markers.
1. Abe et al. [41]; 2. Shao et al. [5]; 3. Qiu et al. [4]; 4. Bai et al. [6]; 5. Wang et al. [29]; 6. Wang et al. [28]; 7. QTL mapped in this study. Numbers to the left of the chromosome bar indicate physical position (Mb) of the corresponding markers.
Epistasis Influencing Rice Grain Length
Epistasis or interactions between non-allelic genes play complex roles in the control of quantitative traits in plants [42]. However, it is difficult to identify the complete epistatic networks, because most of the available software for mapping epistatic loci only detects digenic epistatic loci. Epistatic effects have rarely been observed for grain length [2]. In this study, digenic epistasis did play a role in the determination of grain length, although the four QTLs did not interact with each other. It suggested that the genetic regulation network for grain length is complicated; and the regulation of grain length by eight alleles at the four QTLs revealed only part of the regulatory mechanisms.
The additive phenotypes of the four QTLs indicated that these four loci did not act in a simple linear pathway. Since an interaction between these four loci was not found, we infer that they may participate in four distinct genetic pathways.
Development of Cultivars with Different Grain Length by Pyramiding QTLs Carrying Suitable Alleles
The four grain length-increasing alleles and the four grain length-decreasing alleles found in the present study are ideal resources for modifying grain length in rice. Marker-assisted selection using the nearest markers to these QTLs may be applied for developing new rice cultivars with longer or shorter grain length by pyramiding different number of grain length-increasing or decreasing alleles. Breeding by design [43] using these four QTLs is still not applicable due to the limited understanding of the genetic regulation network, because it is still unclear whether other genes interact with these four QTLs. Further study is needed to investigate whether the four QTLs have the same pyramiding effect when introduced into cultivars with different genetic backgrounds.
Supporting Information
S1 Fig. Marker linkage map used for detecting QTLs in an F7 RIL population and an F8 RIL population.
https://doi.org/10.1371/journal.pone.0150832.s001
(DOCX)
S2 Fig. LOD score curves of QTLs for grain length detected in an F2 population grown in 2011 in Hangzhou.
https://doi.org/10.1371/journal.pone.0150832.s002
(DOCX)
S3 Fig. LOD score curves of QTLs for grain length detected in an F2 population grown in 2012 in Hangzhou.
https://doi.org/10.1371/journal.pone.0150832.s003
(DOCX)
S4 Fig. LOD score curves of QTLs for grain length detected in an F7 RIL population and an F8 RIL population grown in 2014 in Hangzhou and Hainan, respectively.
https://doi.org/10.1371/journal.pone.0150832.s004
(DOCX)
S5 Fig. Digenic epistatic loci detected in an F2 population grown in 2011 in Hangzhou.
https://doi.org/10.1371/journal.pone.0150832.s005
(DOCX)
S1 Table. Markers used to construct a linkage map that was used for detecting QTLs in an F2 population grown in 2012 in Hangzhou.
https://doi.org/10.1371/journal.pone.0150832.s006
(DOCX)
S2 Table. Pairs of digenic epistatic loci detected in an F2 population grown in 2011 in Hangzhou.
https://doi.org/10.1371/journal.pone.0150832.s007
(DOCX)
S3 Table. Two-way ANOVA used to confirm the digenic epistatic loci detected in an F2 population grown in 2011 in Hangzhou.
https://doi.org/10.1371/journal.pone.0150832.s008
(DOCX)
Acknowledgments
The authors thank Fengqiu Liu, Anfu Zheng, and Huilan Xu for the help in the field and laboratory.
Author Contributions
Conceived and designed the experiments: YZ CY. Performed the experiments: YZ ZW. Analyzed the data: ZJ YL. Wrote the paper: YZ CY.
References
- 1. Zuo JR, Li JY. 2014. Molecular genetic dissection of quantitative trait loci regulating rice grain size. Annu Rev Genet. 2014; 48: 99–118. pmid:25149369.
- 2. Huang RY, Jiang LR, Zheng JS, Wang TS, Wang HC, Huang YM, et al. Genetic bases of rice grain shape: so many genes, so little known. Trends Plant Sci. 2013; 18: 218–226. pmid:23218902.
- 3. Singh R, Singh AK, Sharma TR, Singh A, Singh NK. Fine mapping of grain length QTLs on chromosomes 1 and 7 in Basmati rice (Oryza sativa L.). J Plant Biochem Biotechnol. 2012; 21: 157–166.
- 4. Qiu XJ, Gong R, Tan YB, Yu SB. Mapping and characterization of the major quantitative trait locus qSS7 associated with increased length and decreased width of rice seeds. Theor Appl Genet. 2012; 125: 1717–1726. pmid:22864386.
- 5. Shao GN, Wei XJ, Chen ML, Tang SQ, Luo J, Jiao GA, et al. Allelic variation for a candidate gene for GS7, responsible for grain shape in rice. Theor Appl Genet. 2012; 125: 1303–1312. pmid:22772587.
- 6. Bai XF, Luo LJ, Yan WH, Kovi MR, Zhan W, Xing YZ. Genetic dissection of rice grain shape using a recombinant inbred line population derived from two contrasting parents and fine mapping a pleiotropic quantitative trait locus qGL7. BMC Genet. 2010; 11:16. pmid:20184774.
- 7. Kato T, Segami S, Toriyama M, Kono I, Ando T, Yano M, et al. Detection of QTLs for grain length from large grain rice (Oryza sativa L.). Breed Sci. 2011; 61: 269–274.
- 8. Wan XY, Wan JM, Jiang L, Wang JK, Zhai HQ, Weng JF, et al. QTL analysis for rice grain length and fine mapping of an identified QTL with stable and major effects. Theor Appl Genet. 2006; 112: 1258–1270. pmid:16477428.
- 9. Li JM, Thomson M, McCouch SR. Fine mapping of a grain-weight quantitative trait locus in the pericentromeric region of rice chromosome 3. Genetics. 2004; 168: 2187–2195. pmid:15611185.
- 10. Guo LB, Ma LL, Jiang H, Zeng DL, Hu J, Wu LW, et al. Genetic analysis and fine mapping of two genes for grain shape and weight in rice. J Integr Plant Biol. 2009; 51: 45–51. pmid:19166493.
- 11. Xie XB, Song MH, Jin FX, Ahn SN, Suh JP, Hwang HG, et al. Fine mapping of a grain weight quantitative trait locus on rice chromosome 8 using near-isogenic lines derived from a cross between Oryza sativa and Oryza rufipogon. Theor Appl Genet. 2006; 113: 885–894. pmid:16850315.
- 12. Xie XB, Jin FX, Song MH, Suh JP, Hwang HG, Kim YG, et al. Fine mapping of a yield-enhancing QTL cluster associated with transgressive variation in an Oryza sativa × O. rufipogon cross. Theor Appl Genet. 2008; 116: 613–622. pmid:18092146.
- 13. Oh JM, Balkunde S, Yang P, Yoon DB, Ahn SN. Fine mapping of grain weight QTL, tgw11 using near isogenic lines from a cross between Oryza sativa and O. grandiglumis. Genes & Genomics. 2011; 33:259–265.
- 14. Yu SW, Yang CD, Fan YY, Zhuang JY, Li XM. Genetic dissection of a thousand-grain weight quantitative trait locus on rice chromosome 1. Chinese Sci Bull. 2008; 53: 2326–2332.
- 15. Tang SQ, Shao GN, Wei XJ, Chen ML, Sheng ZH, Luo J, et al. QTL mapping of grain weight in rice and the validation of the QTL qTGW3.2. Gene. 2013; 527: 201–206. pmid:23769924.
- 16. Fan CC, Xing YZ, Mao HL, Lu TT, Han B, Xu CG, et al. GS3, a major QTL for grain length and weight and minor QTL for grain width and thickness in rice, encodes a putative transmenbrane protein. Theor Appl Genet. 2006; 112: 1164–1174. pmid:16453132.
- 17. Mao HL, Sun SY, Yao JL, Wang CR, Yu SB, Xu CG, 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.
- 18. Li Y, Fan CC, Xing YZ, Jiang YH, Luo LJ, Sun L, et al. Natural variation in GS5 plays an important role in regulating grain size and yield in rice. Nature Genet. 2011; 43: 1266–1269. pmid:22019783.
- 19. Hu ZJ, He HH, Zhang SY, Sun F, Xin XY, Wang WX, et al. A kelch motif-containing serine/threonine protein phosphatase determines the large grain QTL train in rice. J Integr Plant Biol. 2012; 54: 979–990. pmid:23137285.
- 20. Qi P, Lin YS, Song XJ, Shen JB, Huang W, Shan JX, et al. The novel quantitative trait locus GL3.1 controls rice grain size and yield by regulating Cyclin-T1;3. Cell Res. 2012; 22: 1666–1680. pmid:23147796.
- 21. Zhang XJ, Wang JF, Huang J, Lan H, Wang C, Yin C, et al. Rare allele of OsPPKL1 associated with grain length causes extra-large grain and a significant yield increase in rice. Proc Natl Acad Sci USA. 2012; 109: 21534–21539. pmid:23236132.
- 22. Song XJ, Huang W, Shi M, Zhu MZ, Lin HX. A QTL for rice grain width and weight encodes a previously unknown RING-type E3 ubiquitin ligase. Nat Genet. 2007; 39: 623–630. pmid:17417637.
- 23. Shomura A, Izawa T, Ebana K, Ebitani T, Kanegae H, Konishi S, et al. Deletion in a gene associated with grain size increased yields during rice domestication. Nat Genet. 2008; 40: 1023–1028. pmid:18604208.
- 24. Weng JF, Gu SH, Wan XY, 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.
- 25. Wang SK, Wu K, Yuan QB, Liu XY, Liu ZB, Lin XY, et al. Control of grain size, shape and quality by OsSPL16 in rice. Nat Genet. 2012; 44: 950–954. pmid:22729225.
- 26. Wang ET, Wang JJ, Zhu XD, Hao W, Wang LY, Li Q, et al. Control of rice grain-filling and yield by a gene with a potential signature of domestication. Nat Genet. 2008; 40: 1370–1374. pmid:18820698.
- 27. Ishimaru K, Hirotsu N, Madoka Y, Murakami N, Hara N, Onodera H, et al. Loss of function of the IAA-glucose hydrolase gene TGW6 enhances rice grain weight and increases yield. Nat Genet. 2013; 45: 707–711. pmid:23583977.
- 28. Wang SK, Li S, Liu Q, Wu K, Zhang JQ, Wang SS, et al. The OsSPL16-GW7 regulatory module determines grain shape and simultaneously improves rice yield and grain quality. Nat Genet. 2015; 47:949–954. pmid:26147620.
- 29. Wang YX, Xiong GS, Hu J, Jiang L, Yu H, Xu J, et al. Copy number variation at the GL7 locus contributes to grain size diversity in rice. Nat Genet. 2015; 47:944–948. pmid:26147619.
- 30. Yan S, Zou GH, Li SJ, Wang H, Liu HQ, Zhai GW, et al. Seed size is determined by the combinations of the genes controlling different seed characteristics in rice. Theor Appl Genet. 2011; 123: 1173–1181. pmid:21805338.
- 31. Ying JZ, Gao JP, Shan JX, Zhu MZ, Shi M, Lin HX. Dissecting the genetic basis of extremely large grain shape in rice cultivar “JZ1560”. J Genet Genomics. 2012; 39: 325–333. pmid:22835979.
- 32. Wen ZH, Zeng YX, Ji ZJ, Yang CD. Mapping quantitative trait loci for sheath blight disease resistance in Yangdao 4 rice. Genet Mol Res. 2015; 14: 1636–1649. 10 pmid:25867306.
- 33. Zhang XQ, Zou JS, Zhu HT, Li XY, Zeng RZ. Genetic analysis and gene mapping of an early flowering and multi-ovary mutant in rice (Oryza sativa L.). Hereditas (Beijing). 2008; 30: 1349–1355. (in Chinese with English abstract) pmid:18930897.
- 34. Zeng YX, Ma LY, Ji ZJ, Wen ZH, Li XM, Shi CH, et al. Fine mapping and candidate gene analysis of LM3, a novel lesion mimic gene in rice. Biologia. 2013; 68: 82–90.
- 35. Meng L, Li HH, Zhang LY, Wang JK. QTL IciMapping: integrated software for genetic linkage map construction and quantitative trait locus mapping in biparental populations. The Crop Journal. 2015; 3:269–283.
- 36. Kao CH, Zeng ZB, Teasdale RD. Multiple interval mapping for quantitative trait loci. Genetics. 1999; 152: 1203–1216. PMCID: PMC1460657. pmid:10388834
- 37.
Silva LDCE, Wang SC, Zeng ZB. Composite interval mapping and multiple interval mapping: procedures and guidelines for using Windows QTL Cartographer. In: Rifkin SA, editor. Quantitative Trait Loci (QTL)-Methods and Protocols. Humana Press: Springer Science + Business Media New York; 2012. pp. 75–119. https://doi.org/10.1007/978-1-61779-785-9_6
- 38. Bai XF, Wu B, Xing YZ. Yield-related QTLs and their applications in rice genetic improvement. J Integr Plant Biol. 2012; 54:300–311. pmid:22463712.
- 39. Doerge RW. Mapping and analysis of quantitative trait loci in experimental populations. Nature Rev Genet. 2002; 3: 43–52. pmid:11823790
- 40. Liu GF, Zhang ZM, Zhu HT, Zhao FM, Ding XH, Zeng RZ, et al. Detection of QTLs with additive effects and additive-by-environment interaction effects on panicle number in rice (Oryza sativa L.) with single-segment substitution lines. Theor Appl Genet. 2008; 116: 923–931. pmid:18274724.
- 41. Abe Y, Mieda K, Ando T, Kono I, Yano M, Kitano H, et al. The SMALL AND ROUND SEED1 (SRS1/DEP2) gene is involved in the regulation of seed size in rice. Genes Genet Syst. 2010; 85:327–339. doi: https://doi.org/http://doi.org/10.1266/ggs.85.327 pmid:21317545.
- 42. Carlborg Ö, Haley CS. Epistasis: too often neglected in complex trait studies? Nat Rev Genet. 2004; 5: 618–625. pmid:15266344.
- 43. Peleman JD, Voort JR (2003) Breeding by design. Trends Plant Sci. 2003; 8: 330–334. pmid:12878017.