Identification and evaluation of quantitative trait loci underlying resistance to multiple HG types of soybean cyst nematode in soybean PI 437655

Key message We performed QTL analysis for SCN resistance in PI 437655 in two mapping populations, characterized CNV of Rhg1 through whole-genome resequencing and evaluated the effects of QTL pyramiding to enhance resistance. Abstract Soybean cyst nematode (SCN, Heterodera glycines Ichinohe) is one of the most serious pests of soybean worldwide. PI 437655 has broader resistance to SCN HG types than PI 88788. The objectives of this study were to identify quantitative trait loci (QTL) underlying SCN resistance in PI 437655, and to evaluate the QTL for their contribution to SCN resistance. Two F6:7 recombinant inbred line populations, derived from cv. Williams 82 × PI 437655 and cv. Hutcheson × PI 437655 crosses, were evaluated for resistance to SCN HG types 1.2.5.7 (PA2), 0 (PA3), 1.3.5.6.7 (PA14), and 1.2.3.4.5.6.7 (LY2). The 1,536 SNP array was used to genotype the mapping populations and construct genetic linkage maps. Two significant QTL were consistently mapped on chromosomes (Chr.) 18 and 20 in these two populations. One QTL on Chr. 18, which corresponds to the known Rhg1 locus, contributed resistance to SCN HG types 1.2.5.7, 0, 1.3.5.6.7, and 1.2.3.4.5.6.7 (PA2, PA3, PA14, and LY2, respectively). Copy number variation (CNV) analysis by whole-genome resequencing showed that PI 437655 and PI 88788 had similar CNV at the Rhg1 locus. The QTL on Chr. 20 contributed resistance to SCN HG types 1.3.5.6.7 (PA14) and 1.2.3.4.5.6.7 (LY2). Evaluation of both QTL showed that pyramiding of Rhg1 and the QTL on Chr. 20 significantly improved the resistance to SCN HG types 1.3.5.6.7 (PA14) and 1.2.3.4.5.6.7 (LY2) in both populations. Our studies provided useful information for deploying PI 437655 as a donor for SCN resistance in soybean breeding through marker-assisted selection. Electronic supplementary material The online version of this article (doi:10.1007/s00122-014-2409-5) contains supplementary material, which is available to authorized users.


3
Quantitative trait loci (QTL) mapping is a powerful tool to identify genomic regions responsible for expression of important agronomic traits. Once the desired QTL are mapped, molecular markers that are tightly linked to the QTL can be applied in marker-assisted breeding to improve and shorten the process of developing resistant cultivars. To date, many QTL conferring resistance to SCN in soybean have been mapped on almost all chromosomes except for Chr. 2 (D1b) Vuong et al. 2010;Winter et al. 2007;Wang et al. 2004;Kabelka et al. 2005;Guo et al. 2005Guo et al. , 2006Wu et al. 2009). Among these, two QTL, Rhg1 on Chr. 18 and Rhg4 on Chr. 8 , were commonly mapped in various sources of resistance, primarily plant introductions (PIs). Recently, these two QTL have been successfully cloned from PI 88788 and cultivar Forrest, respectively (Cook et al. 2012;Liu et al. 2012). The SCN resistance conferred by Rhg1 in PI 88788 was found to be controlled by three genes (Cook et al. 2012). The alteration of expression level caused by copy number variation (CNV) rather than sequence mutation for these three genes explained the phenotypic differences between susceptible and resistant varieties (Cook et al. 2012). Rhg4 encodes a serine hydroxymethyltransferase that is essential for cellular one-carbon metabolism (Liu et al. 2012). Two point mutations in Forrest altered a key regulatory property of this enzyme, which may disturb the folate homeostasis and lead to a nutritional deficiency preventing life of the nematode (Liu et al. 2012).
In addition to cultivated soybean sources, scientists have also explored wild soybean (Glycine soja) germplasm to identify new genes for SCN resistance. Wang et al. (2001) reported three QTL for SCN resistance from Glycine soja PI 468916. Winter et al. (2007) mapped three QTL for SCN resistance from Glycine soja PI 464925B. These novel QTL expanded sources of SCN resistance for breeding SCN-resistant soybean cultivars.
Several sources of SCN resistance, such as PI 88788, Peking, and PI 437654 have been widely used in the development of commercial SCN-resistant soybean cultivars . Continuous use of the same source of SCN resistance may lead to a genetic shift of SCN populations, causing a loss of SCN resistance in soybean. It was reported that some SCN populations recovered from the field where soybean varieties with PI 88788 resistance had been constantly used were virulent on PI 88788 . Lack of genetic diversity for SCN resistance among soybean cultivars has become a concern of soybean breeders. Identifying new sources of SCN resistance is important in controlling this pest. In addition, pyramiding of different QTL for SCN resistance from different sources may provide longer protection against SCN HG type population shifts that reduce the effectiveness of genes already employed in cultivars.
PI 437655 was first reported to be resistant to SCN HG type 0 (race 3) (Anand and Gallo 1984). Then, it was found to be resistant to SCN HG types 1.2.3-and 2.5.7 (races 4 and 1) (Anand et al. 1988;Arelli et al. 1997). In an effort to find new sources of SCN resistance, we evaluated 650 exotic soybean PIs for their resistance to multiple SCN HG types in the greenhouse. We found that PI 437655 had a lower FI for all tested SCN populations and showed broader spectrum resistance to SCN HG types than PI 88788. More importantly, we determined that PI 437655 was moderately resistant to SCN isolate LY2 (HG type 1.2.3.4.5.6.7), which was virulent on PI 437654 (Donald and Young 2004). To date, no PIs, except PI 437655, were reported to be resistant to LY2. The molecular basis underlying broad-based SCN resistance in PI 437655 is unknown. The objectives of this study were to identify the QTL responsible for resistance to multiple SCN HG types in PI 437655, and to evaluate the identified QTL for their contribution to SCN resistance.

Plant materials
Two recombinant inbred line (RILs) populations were developed using the single-seed descent method. Population 1 (Pop1) was a population of 119 F 6:7 RILs derived from a cross of Hutcheson × PI 437655. Population 2 (Pop2) was a population of 192 F 6:7 RILs derived from a cross of Williams 82 × PI 437655. Hutcheson and Williams 82 are two SCN susceptible cultivars (Buss et al. 1988;Bernard and Cremeens 1988). PI 437655 is a SCNresistant plant introduction (Anand and Gallo 1984;Anand et al. 1988;Arelli et al. 1997), originating from China and preserved in the USDA Soybean Germplasm Collection Soybean. Seeds of each RIL were planted for SCN phenotyping. Genomic DNA was extracted from a pooled sample of leaves from five seedlings of each RIL following a previously described protocol (Vuong et al. 2010).
The SCN bioassays were performed in a greenhouse at the University of Missouri-Columbia following a wellestablished method (Arelli et al. 1997;Niblack et al. 2009). In brief, germinated soybean seeds were transplanted into PVC tubes (100 cm 3 ) (one plant per tube). The tubes were filled with steam-pasteurized sandy soil and packed into plastic containers prior to transplanting. Each container held twenty-five tubes and was suspended over water baths maintained at 27 ± 1 °C. Five plants of each indicator line and RIL were arranged in a randomized complete block design. Two days after transplanting, each plant was inoculated with 2000 ± 25 SCN eggs. Thirty days postinoculation, nematode cysts were washed from the roots of each plant and counted using Fluorescence-Based Imaging System (Brown et al. 2010). The female index (FI %) was estimated to evaluate the response of each plant to each HG type of SCN using the following formula: FI (%) = (Number of female cyst nematodes on a given individual/average number of female nematodes on the susceptible check) × 100.

Statistical analysis
Female indexes (%) among RILs of two populations were tested for normality using the PROC UNIVARIATE procedure of SAS 9.3 (SAS institute, Gary, NY, USA). A broadsense heritability was calculated following a described method (Wu et al. 2009).

Linkage analyses and genetic mapping
The universal soybean linkage panel 1.0 (USLP 1.0) containing 1,536 SNP loci (Hyten et al. 2008) was utilized to genotype the two RIL mapping populations using the Illumina GoldenGate assay (Fan et al. 2006). These SNP loci had been mapped onto the integrated molecular genetic linkage map .
Genetic linkage maps were constructed using JoinMap 4.0 (van Ooijen 2006). A likelihood of odds (LOD) threshold score of 3.0 and a maximum genetic distance of 50 cM were used for the initial linkage grouping of markers. The soybean genetic linkage groups (LGs) (Song et al. 2004) were replaced with the new assignments of corresponding chromosome numbers (Chr.) (Grant et al. 2010).
Interval mapping (IM) was initially conducted for QTL prediction. Composite interval mapping was subsequently performed using the multi-QTL method (MQM) with the program MapQTL 5.0 and the appropriate cofactor (van Ooijen 2004). A permutation test (Churchill and Doerge 1994) was performed with 1,000 runs to determine the P = 0.05 genome-wide significance level for declaring a QTL significant. The proportion of the phenotypic variance explained by the QTL effects was estimated at the QTL peaks. Additive (A) effects of significant QTL were estimated from an output of the program MapQTL 5.0. The program QTLNetwork 2.0 was used to predict epistatic interactions between QTL (Yang et al. 2007).

Whole-genome sequencing and copy number variation analysis
Whole-genome sequencing of PI 437655, PI 88788, and cv. Hutcheson was conducted using Illumina technology at the Beijing Genome Institute (BGI), Shenzhen, China, following a described protocol (Xu et al. 2013). The sequencing depth for each sample was about 15× coverage. The CNV analysis was conducted using CNV-seq software (Xie and Tammi 2009).

Phenotypic variation and genetic linkage analysis
The FI (%) data of Pop1 and Pop2 showed large genetic variation when assayed with each of the four SCN HG types ( Table 2). The normality test by the Shapiro-Wilk (w) indicated that the FI data of HG types 1.2.5.7 (PA 2) and 1.2.3.4.5.6.7 (LY2) in Pop2 were normally distributed, while others were not normal (Table 2; Fig. 1). Broad-sense heritability of the FI for each HG type was calculated based upon the analysis of variance of family means. The values ranged from 0.46 to 0.68 in Pop1 and from 0.41 to 0.71 in Pop2 (Table 2).
In Pop1, 670 markers were polymorphic between the two parents and were incorporated into linkage analysis. A linkage map spanning 3,029.50 cM across 20 chromosomes was constructed (Supplement S1). In Pop2, 768 markers were found to be polymorphic and a linkage map spanning 2,387.09 cM across 20 chromosomes was constructed (Supplement S2).

Effects of confirmed QTL to SCN resistance
We evaluated the Rhg1 locus and the QTL on Chr. 20 for their contribution to the resistance to different SCN     (Table 4). This indicated that pyramiding different QTL for resistance could significantly confer more stable and higher resistance in soybean to multiple HG types of SCN.

Discussion
Most QTL for SCN resistance reported in literature so far were based on a single bi-parental mapping population ). Although the same resistant parent was utilized to develop different mapping populations, variable QTL can be detected. For instance, in the population from a Hamilton × PI 438489B cross, Yue et al. (2001) identified QTL for SCN HG type 2.5.7 (PA5) resistance on Chrs. 11, 6, 1, and 18. However, Vuong et al. (2011) only identified a QTL on Chr. 4 for resistance to SCN HG types 2.5.7 (PA5) using the population from a Magellan × PI 438489B cross. The inconsistency of identified QTL may be attributed to the differences among genetic backgrounds.
To produce more robust QTL results, two or more mapping populations have an advantage over a single population. In our study, QTL analyses were conducted in two RIL populations derived from cv. Hutcheson crossed with PI 437655 (Pop1) and cv. Williams 82 crossed with PI 437655 (Pop2). Like the previous studies mentioned above, several QTL detected in one population were not confirmed in the other population. For instance, a QTL was mapped on Chr. 5 in Pop1 and different QTLs on Chrs. 16,20,3,4,and 15 were mapped for SCN resistance in Pop2. In contrast, QTL were mapped on Chrs. 18 and 20 in both backgrounds, indicating significant QTL conferring SCN resistance from PI 437655. Apparently, these two QTL regions may be very useful as new gene sources leading to the development of new soybean varieties with improved SCN resistance.
The Rhg1 locus has been widely mapped in various sources of SCN resistance Guo et al. 2005Guo et al. , 2006Vuong et al. 2011;Wu et al. 2009). It was shown that there were two different types of Rhg1, PI 88788-type Rhg1 and Peking-type Rhg1. These two Rhg1 were believed to be different functional alleles (Meksem et al. 2001;Brucker et al. 2005). PI 88788-type Rhg1 was recently cloned (Cook et al. 2012). However, the Rhg1 allele from Peking has not been cloned. To date, many sources of SCN resistance contain Rhg1 . Because the level of SCN resistance conferred by the Rhg1 gene was associated with CNVs (Cook et al. 2012), it is important to determine CNVs of Rhg1 among those sources of resistance. In our study, we predicted that PI 437655 might carry PI 88788-type Rhg1 because Rhg4 was not detected in PI 437655. CNV analysis showed that PI 437655 had the same high number of copies of Rhg1 as PI 88788, which indicated PI 437655 and PI 88788 might have the same level of SCN resistance or the same Rhg1 allele.
Currently, PI 88788 has been widely employed as a predominant donor source in breeding for soybean cultivars with SCN resistance. The most important gene underlying SCN resistance in PI 88788 is Rhg1 (Cook et al. 2012). However, several studies have reported that more and more SCN populations overcame the SCN resistance conferred by PI 88788 (Diers and Arelli 1999;Faghihi et al. 2008), which highlighted the need to develop soybean cultivars with broader and more stable resistance to multiple HG types. In addition to Rhg1, PI 437655 also contained a QTL on Chr. 20, not found in PI 88788 (Glover et al. 2004). This additional QTL might be the reason why PI 437655 showed better SCN resistance than PI 88788 to the four HG types evaluated in this study. RILs with the combination of the Rhg1 locus and the QTL on Chr. 20 in each of two populations had enhanced SCN resistance than genotypes with the Rhg1 allele alone. Thus, pyramiding different resistance QTL will likely be an effective approach to reduce the problem of increased virulence on resistance sources like PI 88788 due to genetic shifts in SCN populations. Moreover, it has been reported that there was no linkage drag for yield when either cv. Fayette or its derived cultivars were utilized in soybean breeding for Northern cultivars. The newly identified QTL in PI 437655 can be introgressed into either to cv. Fayette or its derived cultivars for broadening resistance to SCN.