Mapping Quantitative Trait Loci Associated with Toot Traits Using Sequencing-Based Genotyping Chromosome Segment Substitution Lines Derived from 9311 and Nipponbare in Rice (Oryza sativa L.)

Identification of quantitative trait loci (QTLs) associated with rice root morphology provides useful information for avoiding drought stress and maintaining yield production under the irrigation condition. In this study, a set of chromosome segment substitution lines derived from 9311 as the recipient and Nipponbare as donor, were used to analysis root morphology. By combining the resequencing-based bin-map with a multiple linear regression analysis, QTL identification was conducted on root number (RN), total root length (TRL), root dry weight (RDW), maximum root length (MRL), root thickness (RTH), total absorption area (TAA) and root vitality (RV), using the CSSL population grown under hydroponic conditions. A total of thirty-eight QTLs were identified: six for TRL, six for RDW, eight for the MRL, four for RTH, seven for RN, two for TAA, and five for RV. Phenotypic effect variance explained by these QTLs ranged from 2.23% to 37.08%, and four single QTLs had more than 10% phenotypic explanations on three root traits. We also detected the correlations between grain yield (GY) and root traits, and found that TRL, RTH and MRL had significantly positive correlations with GY. However, TRL, RDW and MRL had significantly positive correlations with biomass yield (BY). Several QTLs identified in our population were co-localized with some loci for grain yield or biomass. This information may be immediately exploited for improving rice water and fertilizer use efficiency for molecular breeding of root system architectures.

Introduction associated with root traits adopting genome-wide association mapping using 167 japonica accessions [30].
So far, only four QTLs for rice root morphology have been fine mapped. Sta1, a QTL determining stele transversal area, was delimited in a 359-Kb interval on rice chromosome 9 using BC 2 F 3 and BC 2 F 4 populations [31]. qRL6.1, a major QTL for root elongation, was fine mapped into a 337-Kb region using a CSSL and its derived population [32]. Dro1, controlling rice deep root, was narrowed into a 608.4-Kb segment [33]. Recently, Uga et al. identified qSOR1, a major QTL involved in soil-surface rooting in paddy fields and located it to an 812-kb interval using seven BC 2 F 3 recombinant lines [34]. More recently, Dro1 was cloned into a 6-kb interval [35].
Advanced populations, such as chromosome segment substituted lines (CSSLs), have same genetic background with recurrent parent, except the donor segments. In order to analyze the genetic basis of rice root development and detecting favorable genes related with root traits, we identified 38 QTLs related to seven root traits using a set of resequencing-genotyped CSSLs under hydroponic conditions. These results provide a valuable contribution to the genetic analysis of root morphology.

Materials and Methods
We state clearly that no specific permissions were required for these locations.

Plant materials
A set of 128 CSSLs with a japonica cultivar, Nipponbare, as the donor and an indica cultivar, 9311, as the recurrent parent was generated as previously reported [36]. Each CSSL line was genotyped, and a high quality physical map of ultrahigh-density single nucleotide polymorphisms (SNPs) based on whole-genome re-sequencing data was constructed. Every CSSL had approximately 60,000 SNPs. On the basis of the physical locations and genotypes of these SNPs, each CSSL was genotyped, and a physical map of the 128 CSSLs was constructed [36]. In this study, the CSSLs population was employed for QTLs mapping of rice root traits and other related traits. However, the phenotypic data of three CSSLs were not obtained because of abnormal growth.

Hydroponic culture
The experiment was carried out in Yangzhou University, Yangzhou, China (latitude 32°24' N, longitude 119°26' E) in summer in 2012. The CSSLs and their parents were grown for this experiment under hydroponic conditions, as previously described [37]. All rice seedlings were solution grown in eight concrete ponds with the inner dimensions of 880.0 cm long by 130.0 cm wide by 50 cm deep. Iron tubes at the bottom of the ponds connected the ponds to one another. The ponds were covered with concrete planks (135.0 cm long by 16.7 cm wide by 2.5 cm in deep), and each plank contained 14 holes (4 cm in diameter) for seedling fixation. Plump seeds were surface-sterilized in a 2.5% NaClO solution for 15 min, and then washed three times with distilled water. Seeds were germinated in an illuminated incubator at 30°C. Rice seedlings were transplanted from the soil seed bed 30 days after germination, and fixed into the holes with a sponge, one hole for one seedling. Fifty-six seedlings were prepared for each line in a randomized complete design with two replicates.
The nutrient solution was the mixture of the Epsino nutrient solution and the Arnon microelement nutrient solution. The mixed nutrient solution was filled into the ponds at transplanting and refreshed every 10 days. The pH value of nutrient solution was monitored daily and maintained between 5.5 and 6.5 by adding diluted H 2 SO 4 . A pump was used to keep the solution continuously cycling during the whole growth period of the rice materials, to maintain the uniformity of pH and nutrient concentration in solution ponds and to improve the O 2 supply.

Evaluation of phenotypes
Plants were sampled at the mature stage. Seven root traits, including root number (RN), total root length (TRL), root dry weight (RDW), maximum root length (MRL), root thickness (RTH), total absorption area (TAA) and root vitality (RV), were evaluated. For RN, only the first ramification was counted. For MRL, the length of the root was measured from the base of the plant to the tip of the longest root. TRL was the total length of the whole roots and calculated as MRL Ã RN/2. RTH was the average diameter of roots at the middle position of the root per plant. For RW, after deactivating enzymes at 105°C for 10 min, the roots were dried at 75°C to constant weight, and then weighed. For RV, we adopt the method of α-NA to measure root vitality [38]. The measure of TAA was by reference to the method of Xiao et al. [39].
We also measured several yield traits, including panicle number per plant (PN), number of spikelets per panicle (SPP), grain weight (GW), seed setting rate (SSR), biomass yield (BY) and grain yield (GY) during the mature stage. PN was the number of effective panicles with 10 or more grains. SPP was measured as the total number of spikelets of the whole plant divided by its total number of panicles. Grain weight was calculated on the basis of 100 grains and converted to 1000-grain weight. SSR was calculated as the number of filled grains per panicle divided by the total number of spikelets per panicle. For BY, after deactivating enzymes at 105°C for 10 min, the aboveground organ were dried at 75°C to constant weight, and then weighed. For GY, total grains per plant were dried naturally after harvesting and stored at room temperature for at least 1 month before weighing.

QTLs analysis
QTLs analysis was conducted according to a previously published method [36]. The multiple linear model was used as the main effect model. The mean value of the ith line of population is defined as y i , the mean of the population is defined as the b 0 , in the whole genome, overall number of bins is defined as m, the main effect related to bin k is defined as b k , donor parent bin denotes x ik = 1, recurrent parent bin denotes x ik = -1. e i denotes the residual error:

Phenotypic variation
The phenotypic value of the seven root traits of the parents and CSSL population were summarized in Table 1 and Fig 1. The two parents, 9311 and Nipponbare, showed highly significant differences in all the root traits examined. 9311 produced more, longer and heavier roots than Nipponbare (Table 1). Extensive variations and a normal distribution among CSSLs were observed for the most traits, which is in accord with the characteristics of quantitative traits (Fig 1). Except MRL, the mean phenotypic values of other six root morphology in the CSSLs population were closer to those of parent 9311 (Table 1). We can explain that the CSSLs have the 9311 genetic backgrounds.
Correlations among root morphological traits. Table 2 showed the pair-wise correlation coefficients among the seven root traits. Three among them, RN, TRL and RDW, had significant positive correlation with each other, indicating that these three traits promote each other. In addition, MRL had significant positive correlation with TRL. TRL had significant positive correlation with RDW, RTH and MRL. Positive correlation was also found between RTH and RV. Except this, a negative correlation was found between RN and MRL.
We also carried out the correlation analysis between the seven root traits and yield traits ( Table 3). BY had significant positive correlation with TRL, RDW and MRL, respectively. Positive correlations were also found between GY and TRL, RTH and MRL. These data suggested that these three root traits are closely related to grain and biomass yield. Positive correlations were also found between TRL and PN, RTH and SSR, GW and MRL, SSR and TAA, respectively.

QTLs identification for root morphological traits
Thirty-eight QTLs were initially detected for the seven root traits on all chromosomes, except chromosome 12.
RTH. Four putative QTLs associated with RTH were detected on chromosomes 1, 2, 5 and 10, respectively. These QTLs were qRTH1.1, qRTH2.1, qRTH5.1, and qRTH10.1, which were located in X 24 on chromosome 1, X 64 on chromosome 2, X 193 on chromosome 5 and X 347 on chromosome 10, respectively. The QTL with the largest effect was mapped to X 347 and occupied to the physical position of 17,928,505 bp to 18,055,332 bp (Table 8). TAA. Two QTLs for TAA were detected on chromosomes 2 and 11, respectively. These QTLs were qTAA2.1 and qTAA11.1, which were located in X 66 on chromosome 2 and X 387 on chromosome 11, respectively. The QTL with the largest effect was mapped to X 387 and occupied to the physical position of 25,559,185 bp to 26,317,711 bp (Table 9).
RV. Five putative QTLs associated with RV were detected on chromosomes 3, 4 and 5 respectively. These QTLs were qRV3.1, qRV4.1, qRV5.1, qRV5.2 and qRV5.3, which were located in X 134 on chromosome 3, X 160 on chromosome 4 and X 218 , X 219 , X 225 on chromosome 5, respectively. The QTL with the largest effect was mapped to X 160 and occupied to the physical position of 16,578,162 bp to 26,317,711 bp (Table 10).
Among these QTLs, we found that some bins controlling one trait was simultaneously detected for other traits. qTRL10.1 and qRTH10.1 were simultaneously located in bin X 347 . qTRL11.1 and qMRL11.1 were detected in X 388 . qRDW6.1 and qMRL6.1 were detected in X 233 .

Discussion
Rice has the greatest water requirement, compared with other cereals [40].Rice production also needs a supply of exogenous nutrients, especially nitrogen fertilizer [41]. Rice roots play an important role in water and nutrient uptake [42]. More water and nutrient uptake of rice depend on faster and more extensive root growth. [43]. Additionally, some studies found that rice root traits have positive relationship with grain yield. [44]. Among different rice varieties, there are large variation associated with root traits. [45]. Generally speaking, indica rice varieties have fewer roots, smaller root absorbing areas, lower root absorbing area density, lower root/ shoot ratio, larger root volume and root length, and show an earlier decline in root parameters than japonica rice variety. The upland rice variety has a greater rooting capacity, root length, root absorbing area density and root/shoot ratio [46]. The development of crop cultivars with thicker and deeper roots is expected to increase water and nutrient availability. Therefore, improving our understanding of the genetics of rice roots could have a significant impact on yield production and water saving. Most of the root traits showed polygenic inheritance. A more thorough knowledge of genetic mechanism of root traits is very important for root breeding program, and there are few examples of the introduction in breeding programs of root traits [5,42,47]. Although hundreds of QTLs associated with rice root traits have been found by different populations, including F 2 , DHs, and mostly RILs, only four QTLs have been fine mapped.
Primary populations such as F 2 , DHs and RILs are very easy to develop; however, due to their genetic background noise, they are difficult for further study [48]. Therefore, advanced mapping populations, like CSSLs, have been developed. CSSLs have same genetic background with female, except the donor segments. Thus, CSSLs make QTLs analysis easier than before and many QTLs of important agronomic traits have been detected in this way [36,[49][50][51][52][53][54][55]. In our previous work, we developed a set of 128 CSSLs generated using an indica cultivar, 9311, as the recurrent and a japonica cultivar, Nipponbare, as the donor. We detected accurately the lengths of the substituted segments and provided more accurate background information using the resequencing-based map [36].
In this study, 38 QTLs associated with root morphology were identified using the CSSLs population mentioned above. Among these QTLs, qRV4.1 explained the largest variance of phenotypic effect (37.08%). qMRL6.1 explained the least variance of phenotypic effect (3.42%). Four QTLs identified for all root traits was found for more than 10% phenotypic variation explained by a single QTL. Several of them were located on the same chromosome segment reported in previous studies (Fig 2). For example, qRTH1.1 was on bin X 24 close to the QTL for maximum root thickness detected in the Bala × Azu cross [56]. qRN6.1 was identified on bin X 241 close to the QTL for lateral root number detected in the IR1552 × Azu cross [28]. qMRL7.1, identified on bin X 262 , was locating on a overlapping region containing a pleiotropic QTL for seven root traits on chromosome 7 [57]. qRN9.1, locating on bin X 321 , was very close to a fine mapped QTL, Sta1, which determines the stele transversal area on chromosome 9 [31]. Additionally, qMRL8.1 on bin X 291 , qRTH5.1 on X 193 and qRDW8.1 on X 278 co-localized on the same region with qRL8.1, qRT5.1 and qRDW8.1, associated with root length, root thickness and dry root weight, respectively, using an ARB25 × Pusa Basmati 1460 F 2:3 population [58].
A well-developed root system increases biomass and yield in different treatments and cultivars in paddy fields [6]. We are interested in whether or not the QTLs for root traits are located on the overlapping region containing QTLs regulating grain yield and biomass. Actually, several QTLs associated with grain yield and biomass yield found in other studies also close to the bins controlling root traits found in our research. For example, qMRL11.1 was located on the chromosome segment containing a QTL for shoot biomass detected in a DH population from the cross of CT9993 × IR62266 [9]. qRTH1.1 on bin X 24 and qRDW3.1 on X 137 had the similar positions with two QTLs, yd1b and yd3, respectively, which were involved in grain yield found in RILs from the cross of Zhenshan97 and Minghui63 [59]. Both of qRV4.1 on bin X 160 and qYLD4-1, a QTL for grain yield per plant identified in the DH population from IR64 and Azucena [60], was located in a overlapping region. In addition, qRTH2.1 on bin X 64 shared same region with two QTLs, DTY2.1 [61] and qDTY2.1 [62], regulating grain yield under drought conditions. This result implied that qRTH2.1 maybe a valuable loci to improve the ability to avoid drought stress and beneficial for stable rice production.
In this study, a total of 38 QTLs associated with root morphology were identified using a resequencing-genotyped CSSLs population under hydroponic culture conditions. Six of the detected QTLs were narrowed in very small regions. For instance, qTRL5.1 on bin X 205 , qRDW3.1 on bin X 137 , qMRL7.1 on bin X 262 , qRN1.1 on bin X 8 , qRN3.1 on bin X 131 , and qRV3.1 on bin X 134 were delimited into chromosome segments less than 100-Kb in physical distance. These results would be helpful for cloning of these QTLs and improving rice water and fertilizer use efficiency by molecular breeding of root system architectures.