Development and Evaluation of Primary Introgression Lines of Rice by Advanced Backcross QTL Strategy for Gaozhou Wild Rice (O. rufipogon)

Rice (Oryza sativa L.) is the major cereal crop in China. The modern rice cultivars have narrow genetic base for most of the available traits, wild relatives of rice species contain trait-enhancing genes. The goal of this study was to identify introgression lines that can significantly increase yield in cultivated rice by yield-related traits from an inter-specific advanced backcross population (AB-population) between O. statia×O. rufipogon. AB-population (O. statia ‘Yuexiangzhan’ × O. rufipogon ‘G52-9’) has evaluated using statistical methods for 17 agronomic traits pertaining to yield and yield-related components. Our studies indicated that AB-population had higher values for most traits analyzed in comparison to the recurrent parent, Yuexiangzhan. According to the statistical results, about 33.20% of the BC3F1 lines showed better than Yuexiangzhan in some traits. Grain shape and grain weight have selected for QTL mapping as an illustration of AB-QTL analysis. A total of 23 QTLs were identified for grain length (GL), grain width (GW), Grain length/grain width (LW), and grain weight (GWt). Phenotypic variation of each QTL ranged from 3.77%~28.67%. The grain weight of the recurrent ‘Yuexiangzhan’ was improved 15.10%, 28.67%, 26.29%, 24.28% and 12.90% by five main QTLs of GWt (from Gaozhou wild rice). This study confirms that wild progenitor species present potential donor sources for some complex traits and the advanced backcross population provides an efficient source of gene pool for future breeding programs. Therefore, Gaozhou wild rice is an efficient source of gene pool and can further utilize.


Background
In China, rice is an important food crop. With the trend of population growth and improved standard of living, food shortage is one of the most serious problems of this century. Therefore, to increase rice production is an essential assignment for rice breeding programs.
Wild rice is an important germplasm resource for rice improvement and rice breeding programs. It has formed abundant genetic diversity during the process of evolution and conserved many specific genes presently not available or lost in the cultivated rice (Xiao et al., 1998). Sun et al (2001) studies showed that the number of alleles of cultivated rice was 58.2% of wild rice on 44 RFLP loci. Their results also shown that the number of alleles decreased, the diversity of genetic was reduced, and some excellent of the genes were lost in cultivar rice during the process of evolution.
Tanksley and Nelson introduced the advanced backcross quantitative trait locus (AB-QTL) strategy in 1996, in order to map the favorable QTL alleles of exotic germplasm and introgression of these alleles into elite cultivate lines. Some studies derived from rice and tomato showed that this method is an efficient approach to excavate the favorable gene in wild germplasm resource (Xiao et al., 1998;Moncada et al., 2001;Septiningsih et al., 2003). The introgression of novel alleles from wild germplasm is one effective means for further improvement of agronomic traits, which has successfully used in many cultivated rice varieties as reported in some QTL mapping studies (Xiao et al., 1998;Moncada et al., 2001;Septiningsih et al., 2003;Tian et al., 2006). However, little information has focused on the systemic select methods for the introgression lines of some specific target traits. We aim is to present a systemic method that combines statistical analysis and molecular marker-assisted selection.
Oryza rufipogon accession No. G52-9 is a line of Chinese common wild rice (O. rufipogon Griff.); it grows in Gaozhou city of Guangdong province in China. Located at 21°42′34″~22°18′48″ N latitude and 110°36′48″~111°22′45″ E longitude, which in the transition region from South subtropics to North subtropics. To develop advanced backcross population is the first program of AB-QTL strategy. The purposes of this study were (1) to develop advanced backcross populations; (2) to evaluate yield and yield-related traits under standard farming conditions using statistical methods; (3) to select superior families of introgression lines for construct and cultivar development; (4) to introduce application of the advanced backcross quantitative trait locus (AB-QTL) strategy in rice. Other studies of detailed QTL mapping and introgression lines construction were also done with rice families used in this study.

Principal component based on data of 241 families
Principal component analysis was carried out based on data in Table 1 and the result from regression analysis. Principal component analysis is a n effective multi-analysis that decreases the variable number of results of regression analysis. Table 3 and Table 4 show which of the 17 traits contributed most for grain , PPN (productive panicle number), GL (grain length), GW (1000 grain weight), PSS (percentage seed set), DTH (days to heading), SD (spikelet density).

Selection of superior BC 3 F 1 families
According to the results from Pearson correlation coefficients, regression analysis and principal component analysis, panicle length, grain number per panicle, 1000-grain weight, grain length, percentage seed set, productive panicle number and spikelet density were used to rank BC 3 F 1 families. After ranking, 80 superior families (33.20 %) were selected for genotype analysis. For example, seventeen selected superior families are list in Table 5.

Genotype and introgression segment analysis for selected superior families
Candidate lines of the development of introgression lines were selected based on genotype analysis. The proportion of segment introgression from O. rufipogon ranged from 0.80%~41.50%, with an average introgression of 20.62% in BC 3 F 1 families. The average was different from the expected heterozygous portion of an unselected BC 3 F 1 family, which would be 12.5%. Two markers with extreme proportions are near 0%. The results of genotype analysis by GGT software are show in Figure 1. The selected backcross introgression lines combined with the number of different introgression for any specific region determines the locations of the QTL analysis.  Table 3 Selected eigenvalues in BC3F1 population eight chromosomes for Gaozhou wild rice. The Gaozhou wild rice QTL allele was associated with an improvement of agronomic traits ( Figure 2). For example, at locus rm440 on chromosome 5, the homozygous wild rice genotype can increase grain weight by 28.67%.

Discussion
Utilization of wild germplasm as donors in inter-specific crosses is one of the strategies to broaden the genetic diversity of the rice gene pool.
Over the last decade, wild species in rice have successfully utilized for incorporation of yieldenhancing genes. In an earlier study, Xiao et al (1998), Septiningsih et al (2003), and Thomson et al (2003) shown that quantitative trait loci (QTL) alleles from Oryza rufipogon access .no. IRGC105491 had a beneficial effect for yield potential of cultivated rice by as much as 20%, 33%, and 53%, respectively.
Gaozhou wild rice is the largest wild rice population in Guangdong Province, covering about 15 hectares. Gaozhou wild rice grows without obvious winter dormancy; it is evergreen all the year round. Heading once a year, it is a brown grain with small panicles and low seed setting percentage. As a precious rice germplasm resource, it contains some desirable traits, including strong tolerance to disease, insect, cold, and low fertility . Li et al (2006) have studied the genetic diversity of Gaozhou wild rice by using SSR markers, and the results shown that Gaozhou wild rice is probably the largest center of genetic differentiation and diversity of common wild rice from Guangdong province, south China, and even whole China. The result also shown that Gaozhou wild rice may be conserve many specific genes presently not available or has been vanished in the cultivated rice. In this study, we construct an advanced backcross of Gaozhou wild rice for first time in China.
The six traits expressed a discontinuous distribution that might be due to phenotypic variability of ABpopulation and the abundant genetic diversity of Gaozhou wild rice. Chen et al (2008) reported the phenotypic diversity of Gaozhou wild rice was plentiful.  (2003)). 1000-grain weight was negatively associated with both spikelet number per panicle (-0.148) and grain number per panicle (-0.088). These results confirm with those reported earlier for studies involving O.rufipogon (Xiao et al., 1998;Septiningsih et al., 2003;Thomson et al., 2003). In the present study, the correlation between 1000-grain weight and yield was significant as also reported earlier (Xiao et al., 1998;Thomson et al., 2003). However, other studies reported no significant correlation between other yield-related traits and 1000-grain weight. These different results in correlation coefficients among traits in different studies can be due to the different parents and environment.

Rice Genomics and
Not all of the 80 superior families are suitable to construct introgression lines. For example, 17 selected good-performing lines are not always higher than Yuexiangzhan for evaluated traits; the percentage seed set is lower than Yuexiangzhan. In breeding programming, it should adjust according to different goal for cultivar development or gene exhume. For gene finding, the selected families should choose for inferior and superior traits simultaneity. The numbers of QTLs for grain shape and grain weight in this study is higher in comparison to earlier studies with O.rufipogon. This result is 1.77 times as much as Lee et al (2005). The same population is being genotyped for QTL mapping in another study. These studies confirm that wild progenitor species present potential donor sources to improve some complex traits and the AB-population provide an efficient source of gene pool for future breeding programs. Gaozhou wild rice is an abundant source of gene pool and can be utilize further. Therefore, it is necessary to expand ILs and NILs-carrying single wild rice for future project.
Wild species usually show inferior agronomic characteristics compared to cultivated species. For example, to exist badness agricultural character and bad linkage genes; upper adverse gene frequency of wild rice in balanced population and genetic drag affected the utilization presently, and these bring some difficulties to find the elite gene of germplasm resource of wild rice. The present study and earlier some results (Bernacchi et al., 1998a;1998b) indicated AB-QTL analysis is feasible to find the available genes, and shown that wild rice still contain some valuable trait enhancing genes during the process of evolution from common wild rice to cultivated rice.

Detection of introgression and genotyping with markers
DNA was extracted from fresh leaves of 245 field-grown BC3F1 lines. DNA extraction was performed using the protocol of Li et al (2006). A total of 117 polymorphic simple sequence length repeat (SSR) markers from published information were used to analyze the 241 lines (Chen et al., 1997;Temnykh et al., 2001;MoCouch et al., 2002). The polymerase chain reaction (PCR) conditions were as described in Chen et al (2008), with the following modifications: a total volume of 10 μL reaction mixture was composed of 2 ng/μL of template DNA, 1 μmol/L primers, each 1 μL of 10 mM dNTPs, 50 mmol/L KCL, 10 mmol/L Tris-HCL (pH=9.0), 1.5 mmol/L MgCl 2 , and 0.75 U Taq polymerase. PCR amplification conditions were as follows: denaturing at 94℃ for 5 min, followed by 35 cycles of 94℃ for 30 s, 55℃ for 1 min, and 72℃ for 2 min, and lastly, 8 min at 72℃. SSR analyses were performed on 6% polyacrylamide denaturing gels.

Data analysis
The data of this paper were analyzed in SAS/STAT User's Guide Release 8.2 in 1992. Various elementary descriptive statistics were calculated using PROC MEANS for each trait (parental and progeny). Analysis of BC 3 F 1 population phenotype variance was conducted using the General Linear Model (GLM) procedure. Normal distribution test of all traits were determined using PROC GCHART. The trait Pearson correlation coefficients were calculation at P<0.05, P<0.01 and P<0.001 using PROC CORR. Regression analysis was performed using PROC REG. Principal component analysis was carried out based on PROC FACTOR. Genotype and introgression segment of wild were analyzed using GGT software (http://www.plantbreeding.wur.nl/). Detection of QTL was conducted using QTL Cartographer Ver.2.5 (Basten et al., 1997).

Author Contributions
ZBJ and CL conceived the overall study, performed the experiment designs, and drafted the manuscript. ZBJ, CY and DJP took part to the data analysis and the writing. YYQ, ZLF and JYC obtained and analyzed the data and were involved in the writing. All authors read and approved the final manuscript.