Genetic Mapping and Molecular Characterization of a Broad-spectrum Phytophthora sojae Resistance Gene in Chinese Soybean

Phytophthora root rot (PRR) causes serious annual soybean yield losses worldwide. The most effective method to prevent PRR involves growing cultivars that possess genes conferring resistance to Phytophthora sojae (Rps). In this study, QTL-sequencing combined with genetic mapping was used to identify RpsX in soybean cultivar Xiu94-11 resistance to all P. sojae isolates tested, exhibiting broad-spectrum PRR resistance. Subsequent analysis revealed RpsX was located in the 242-kb genomic region spanning the RpsQ locus. However, a phylogenetic investigation indicated Xiu94-11 carrying RpsX is distantly related to the cultivars containing RpsQ, implying RpsX and RpsQ have different origins. An examination of candidate genes revealed RpsX and RpsQ share common nonsynonymous SNP and a 144-bp insertion in the Glyma.03g027200 sequence encoding a leucine-rich repeat (LRR) region. Glyma.03g027200 was considered to be the likely candidate gene of RpsQ and RpsX. Sequence analyses confirmed that the 144-bp insertion caused by an unequal exchange resulted in two additional LRR-encoding fragments in the candidate gene. A marker developed based on the 144-bp insertion was used to analyze the genetic population and germplasm, and proved to be useful for identifying the RpsX and RpsQ alleles. This study implies that the number of LRR units in the LRR domain may be important for PRR resistance in soybean.


Introduction
Soybean [Glycine max (L.) Merr] is one of the most important economic crops, accounting for more than half of the global oilseed production [1]. Sustainable soybean cultivation is largely limited by diseases caused by diverse pathogens, including the soil-borne oomycete Phytophthora sojae, which causes Phytophthora root rot (PRR), with annual economic losses of 1-2 billion worldwide [2][3][4][5][6]. In China, PRR was first detected in Heilongjiang province in 1989, and it has since spread to most soybean-producing areas [7]. This disease can occur at any soybean plant developmental stage. In water-saturated soils, P. sojae zoospores can infect soybean plants throughout the growing season, causing damping-off before and after emergence, root and stem decay during the adult stage, and even death [8]. If PRR is established in the field, it is difficult to control with chemical and biological agents, ultimately resulting in considerable or complete yield losses [7]. The most effective way to control PRR currently involves growing soybean cultivars expressing Rps genes, which confer resistance to P. sojae [9]. Compared with the application of biological and chemical agents, growing PRR-resistant

Xiu94-11 Has Broad-spectrum Resistance to Phytophthora sojae Which is Controlled by a Dominant Single Gene
The resistance to 14 P. sojae isolates with varying virulence levels was assessed for the PRR-resistant Xiu94-11 and 22 other cultivars each containing a different identified Rps gene as well as four PRR-susceptible cultivars (Zhonghuang13, Williams, Zhonghuang47, and Jikedou2) as controls. All of the soybean plants of the PRR-susceptible cultivars died after the inoculations with the P. sojae isolates. Moreover, 13 reaction types were observed among the 27 tested soybean cultivars in response to the 14 P. sojae isolates (Table 1). Specifically, Xiu94-11 was resistant to all 14 P. sojae isolates and exhibited the broadest spectrum resistance among the PRR-resistant cultivars. Thus, Xiu94-11 may contain a novel Rps gene or a unique combination of the identified Rps genes.
To analyze the genetic characteristics underlying the resistance of Xiu94-11 and further map the Rps gene(s) involved, an F 2:3 mapping population was derived from a hybridization between Zhonghuang47 and Xiu94-11. Phytophthora sojae isolates PsMC1 and PsJS2 were used to evaluate the phenotypes of 137 F 2:3 families and parental cultivars. All families exhibited consistent responses to the two P. sojae isolates. The 38 homozygous resistant families, 63 segregating families, and 36 susceptible families fit the expected 1:2:1 ratio (Table 2). These results suggested that the resistance of Xiu94-11 is controlled by a single dominant gene, which we tentatively named RpsX.   (Table S1).
The QTL-seq approach was used to identify the genomic region containing RpsX [38]. A total of 1,159,236 high-quality SNPs were obtained, and the delta SNP index of each SNP was calculated. The distribution of the delta SNP index (R30-S30) on 20 chromosomes (Figure S1) at the 99% confidence level revealed a contiguous region exceeding the threshold in the 1.05-3.55 Mb genomic region of chromosome 3 ( Figure 1). Accordingly, this region represented the only candidate region for RpsX. The blue dots represent the delta SNP index corresponding to an SNP obtained by filtering the two bulks. The black, purple, and red lines respectively represent the average value of the delta SNP index as well as the 95% and 99% confidence level thresholds in the corresponding window calculated by the sliding window method. The distribution of the delta SNP index at the 99% confidence level revealed only one contiguous region exceeding the threshold in the 1.05-3.55 Mb genomic region (red) of chromosome 3. The window is 1 Mb, with 10-kb slides.

RpsX Was Finely Mapped to the 242 kb Region on Chromosome 3
To validate the accuracy of the RpsX candidate region identified by QTL-seq and further limit the RpsX genomic interval, a genetic mapping approach was used to analyze all 137 F2:3 families. The published simple sequence repeat (SSR) markers in the candidate region and the InDel markers developed based on the InDels identified via whole-genome resequencing were used for screening the polymorphism between the parental cultivars and clarifying the genotypes of the populations [24,25,37]. Among 150 SSR markers, seven identified polymorphisms between Xiu94-11 and The blue dots represent the delta SNP index corresponding to an SNP obtained by filtering the two bulks. The black, purple, and red lines respectively represent the average value of the delta SNP index as well as the 95% and 99% confidence level thresholds in the corresponding window calculated by the sliding window method. The distribution of the delta SNP index at the 99% confidence level revealed only one contiguous region exceeding the threshold in the 1.05-3.55 Mb genomic region (red) of chromosome 3. The window is 1 Mb, with 10-kb slides.

RpsX Was Finely Mapped to the 242 kb Region on Chromosome 3
To validate the accuracy of the RpsX candidate region identified by QTL-seq and further limit the RpsX genomic interval, a genetic mapping approach was used to analyze all 137 F 2:3 families. The published simple sequence repeat (SSR) markers in the candidate region and the InDel markers developed based on the InDels identified via whole-genome resequencing were used for screening the polymorphism between the parental cultivars and clarifying the genotypes of the populations [24,25,37]. Among 150 SSR markers, seven identified polymorphisms between Xiu94-11 and Zhonghuang47 and were closely linked to RpsX (Figure 2A). Additionally, 20 InDels between Xiu94-11 and Zhonghuang47 were developed as PCR-based markers, four of which were associated with polymorphisms and were tightly linked to RpsX (Figure 2A). Moreover, RpsX was mapped between the InDel marker InDelxz6 and the SSR marker BARCSOYSSR_03_0175, with genetic distances of 0.4 and 0.7 cM, respectively, and co-segregated with three SSR markers (BARCSOYSSR_03_0161, BARCSOYSSR_03_0165, and BARCSOYSSR_03_0167). On the basis of the physical position of each marker on chromosome 3, the genomic region comprising RpsX was localized to a 242-kb region (2,910,913-3,153,254 bp) ( Figure 2B), which spans the RpsQ genomic region ( Figure 2C) [20]. Polymorphisms were screened and genetic linkages were analyzed with the published InDel markers (Insert144 and Insert11) that reportedly co-segregate with RpsQ [20]. The resulting data indicated that the two markers also co-segregated with RpX. Therefore, we speculated that RpsX may be an allele of RpsQ.

RpsX and RpsQShared the Same Candidate Gene Model
The available information regarding the annotated soybean genome (Glyma.Wm82.a2.v1) indicates there are 24 gene models in the RpsX mapping interval (https://www.soybase.org/). Fourteen nonsynonymous SNPs (nsSNPs) with a delta SNP index of 1 were identified distributed in eight gene models between the resistant and susceptible bulks ( Figure 3). Among the eight gene models containing nsSNPs, Glyma.03g027200 containing five nsSNPs and annotated as a serine/threonine protein kinase (STK) with leucine-rich repeats (LRRs) is reportedly a type of plant resistance gene. In contrast, there are no reports suggesting the other seven gene models are related to disease resistance in plants. Therefore, Glyma.03g027200 was identified as the most likely RpsX candidate gene. Among the analyzed cultivars, the RpsX mapping interval of Xiu94-11 (RpsX) had the same nsSNP as the corresponding intervals of Qichadou1 (RpsQ) and the parental cultivar of Qichadou1, Ludou4 (Rps9), only for Glyma.03g027200 ( Figure 3).

RpsX and RpsQShared the Same Candidate Gene Model
The available information regarding the annotated soybean genome (Glyma.Wm82.a2.v1) indicates there are 24 gene models in the RpsX mapping interval (https://www.soybase.org/). Fourteen nonsynonymous SNPs (nsSNPs) with a delta SNP index of 1 were identified distributed in eight gene models between the resistant and susceptible bulks ( Figure 3). Among the eight gene models containing nsSNPs, Glyma.03g027200 containing five nsSNPs and annotated as a serine/threonine protein kinase (STK) with leucine-rich repeats (LRRs) is reportedly a type of plant resistance gene. In contrast, there are no reports suggesting the other seven gene models are related to disease resistance in plants. Therefore, Glyma.03g027200 was identified as the most likely RpsX candidate gene. Among the analyzed cultivars, the RpsX mapping interval of Xiu94-11 (RpsX) had the same nsSNP as the corresponding intervals of Qichadou1 (RpsQ) and the parental cultivar of Qichadou1, Ludou4 (Rps9), only for Glyma.03g027200 ( Figure 3). As Glyma.03g027200 was identified as an RpsQ candidate gene, we examined whether Xiu94-11 and Qichadou1 are related and whether RpsX and RpsQ are actually the same gene. We completed a phylogenetic analysis based on the SNPs identified by NGS among 45 cultivars/landraces, some of which are related to Xiu94-11 and Qichadou1 (Table S2). A total of 48,049 SNPs on chromosome 3 identified among 45 cultivars were selected to construct a phylogenetic tree according to the neighbor-joining method. We observed that 45 soybean genotypes formed two subgroups, and Xiu94-11 belonged to a subgroup separate from that of Ludou4 and Qichadou1 ( Figure 4). The cultivar most closely related to Xiu94-11 was Kaohsiung1, which was selected from Chu-tzu-dow, a landrace from Taiwan, which is not related to Qichadou1 [46]. These results implied that Xiu94-11 is not closely related to Qichadou1. As Glyma.03g027200 was identified as an RpsQ candidate gene, we examined whether Xiu94-11 and Qichadou1 are related and whether RpsX and RpsQ are actually the same gene. We completed a phylogenetic analysis based on the SNPs identified by NGS among 45 cultivars/landraces, some of which are related to Xiu94-11 and Qichadou1 (Table S2). A total of 48,049 SNPs on chromosome 3 identified among 45 cultivars were selected to construct a phylogenetic tree according to the neighbor-joining method. We observed that 45 soybean genotypes formed two subgroups, and Xiu94-11 belonged to a subgroup separate from that of Ludou4 and Qichadou1 (Figure 4). The cultivar most closely related to Xiu94-11 was Kaohsiung1, which was selected from Chu-tzu-dow, a landrace from Taiwan, which is not related to Qichadou1 [46]. These results implied that Xiu94-11 is not closely related to Qichadou1.

144-bp Insertion in LRR Domain is Present in Resistant Haplotype of RpsX Locus
To further explore the candidate gene sequence and structural differences between the resistant and susceptible genotypes, the allele sequences of Glyma.03g027200 in Xiu94-11 and Zhonghuang47 were obtained with a Sanger sequencing method. All five nsSNPs identified by NGS were confirmed in Xiu94-11 and Zhonghuang47. A comparison of the obtained and the previously published RpsQ candidate gene sequences [20] revealed a 99% sequence identity between the RpsX and RpsQ alleles. Additionally, we detected a 97% sequence identity between the RpsX allele and the alleles of the Williams82 reference genome sequence and the susceptible control Zhonghuang41 (Supplemental sequence S1). These results indicated that the Glyma.03g027200 sequence is relatively conserved and is highly similar between the resistant and susceptible genotypes. The deduced protein sequences based on the genomic sequences were aligned and analyzed for conserved domains. We detected two functional domains in all aligned sequences, namely the LRR and STK domains. However, the resistant and susceptible alleles differ in the sequence encoding the LRR region. Specifically, 11 LRR motifs are encoded in the RpsX and RpsQ alleles of the resistant genotype, whereas nine LRR motifs are encoded in the corresponding alleles of the susceptible genotypes. Like the RpsQ allele, the RpsX allele contains a 144-bp insertion, resulting in the insertion of 48 amino acid residues comprising two LRR structural units ( Figure 5). The genomic sequence most similar to the 144-bp insertion is a 144-bp sequence upstream of the candidate gene, implying the insertion may have been the result of a replication of the upstream fragment ( Figure 5).

144-bp Insertion in LRR Domain is Present in Resistant Haplotype of RpsX Locus
To further explore the candidate gene sequence and structural differences between the resistant and susceptible genotypes, the allele sequences of Glyma.03g027200 in Xiu94-11 and Zhonghuang47 were obtained with a Sanger sequencing method. All five nsSNPs identified by NGS were confirmed in Xiu94-11 and Zhonghuang47. A comparison of the obtained and the previously published RpsQ candidate gene sequences [20] revealed a 99% sequence identity between the RpsX and RpsQ alleles. Additionally, we detected a 97% sequence identity between the RpsX allele and the alleles of the Williams82 reference genome sequence and the susceptible control Zhonghuang41 (Supplemental sequence S1). These results indicated that the Glyma.03g027200 sequence is relatively conserved and is highly similar between the resistant and susceptible genotypes. The deduced protein sequences based on the genomic sequences were aligned and analyzed for conserved domains. We detected two functional domains in all aligned sequences, namely the LRR and STK domains. However, the resistant and susceptible alleles differ in the sequence encoding the LRR region. Specifically, 11 LRR motifs are encoded in the RpsX and RpsQ alleles of the resistant genotype, whereas nine LRR motifs are encoded in the corresponding alleles of the susceptible genotypes. Like the RpsQ allele, the RpsX allele contains a 144-bp insertion, resulting in the insertion of 48 amino acid residues comprising two LRR structural units ( Figure 5). The genomic sequence most similar to the 144-bp insertion is a 144-bp sequence upstream of the candidate gene, implying the insertion may have been the result of a replication of the upstream fragment ( Figure 5).
Because there are currently no soybean cultivars derived from Xiu94-11, to further verify whether the detected 144-bp insertion is important for the resistance to P. sojae, soybean genotypes related to Qichadou1 based on pedigrees were analyzed regarding their reactions to 12 P. sojae isolates. Additionally, their alleles corresponding to Glyma.03g027200 were sequenced (Supplemental sequence S1). A phylogenetic tree was constructed according to the neighbor-joining method using the Glyma.03g027200 allelic sequences of 30 genotypes. Four cultivars, namely Xiu94-11 (RpsX), Qichadou1 (RpsX), Ludou4 (Rps9), and Kexin5, were clustered in one subgroup. Moreover, the same 144-bp insertion was detected in their Glyma.03g027200 allelic sequences ( Figure 6). Qichadou1 was derived from a cross between Ludou4 and Peking, whereas Kexin5 was the result of the chemical mutagenesis of Ludou4. These four cultivars all exhibited excellent resistance to 10-12 P. sojae isolates, suggesting that the 144-bp insertion in the sequence encoding the LRR region may be important for the observed resistance to P. sojae.  Because there are currently no soybean cultivars derived from Xiu94-11, to further verify whether the detected 144-bp insertion is important for the resistance to P. sojae, soybean genotypes related to Qichadou1 based on pedigrees were analyzed regarding their reactions to 12 P. sojae isolates. Additionally, their alleles corresponding to Glyma.03g027200 were sequenced (Supplemental sequence S1). A phylogenetic tree was constructed according to the neighbor-joining method using the Glyma.03g027200 allelic sequences of 30 genotypes. Four cultivars, namely Xiu94-11 (RpsX), Qichadou1 (RpsX), Ludou4 (Rps9), and Kexin5, were clustered in one subgroup. Moreover, the same 144-bp insertion was detected in their Glyma.03g027200 allelic sequences  Because there are currently no soybean cultivars derived from Xiu94-11, to further verify whether the detected 144-bp insertion is important for the resistance to P. sojae, soybean genotypes related to Qichadou1 based on pedigrees were analyzed regarding their reactions to 12 P. sojae isolates. Additionally, their alleles corresponding to Glyma.03g027200 were sequenced (Supplemental sequence S1). A phylogenetic tree was constructed according to the neighbor-joining method using the Glyma.03g027200 allelic sequences of 30 genotypes. Four cultivars, namely Xiu94-11 (RpsX), Qichadou1 (RpsX), Ludou4 (Rps9), and Kexin5, were clustered in one subgroup.

Developed Marker Insert144 Is Able to Efficiently Detect Resistant Haplotypes at RpsX Locus
Because the 144-bp insertion is a key variant of the RpsX locus, Insert144, which is an InDel marker based on this insertion, was developed as a marker that co-segregates with RpsQ and was used to distinguish RpsQ from the Rps1 alleles. In the present study, Insert144 was further validated in all cultivars carrying currently identified Rps genes as well as in the susceptible controls [20]. Only three soybean cultivars, namely Xiu94-11 (RpsX), Ludou4, and Qichadou1 (RpsQ), contained this insertion (Figure 7). The screening and identification of RpsX and its alleles among 177 soybean germplasms revealed that six soybean cultivars (Ludou2, Qihuang9, Fendou78, Fendou79, Qihuang12, and Qihuang13) had the same genotype as Xiu94-11. The phenotypic responses to eight P. sojae isolates indicated that they were all resistant to PRR, meaning the eight cultivars are likely to contain RpsX or PRR-resistance alleles at the RpsX locus. These results suggested that Insert144 can serve as a functional and diagnostic marker for RpsX and the PRR-resistance alleles during soybean breeding.
the LRR region may be important for the observed resistance to P. sojae.

Developed Marker Insert144 is Able to Efficiently Detect resistant Haplotypes at RpsX Locus
Because the 144-bp insertion is a key variant of the RpsX locus, Insert144, which is an InDel marker based on this insertion, was developed as a marker that co-segregates with RpsQ and was used to distinguish RpsQ from the Rps1 alleles. In the present study, Insert144 was further validated in all cultivars carrying currently identified Rps genes as well as in the susceptible controls [20]. Only three soybean cultivars, namely Xiu94-11 (RpsX), Ludou4, and Qichadou1 (RpsQ), contained this insertion (Figure 7). The screening and identification of RpsX and its alleles among 177 soybean germplasms revealed that six soybean cultivars (Ludou2, Qihuang9, Fendou78, Fendou79, Qihuang12, and Qihuang13) had the same genotype as Xiu94-11. The phenotypic responses to eight P. sojae isolates indicated that they were all resistant to PRR, meaning the eight cultivars are likely to contain RpsX or PRR-resistance alleles at the RpsX locus. These results suggested that Insert144 can serve as a functional and diagnostic marker for RpsX and the PRR-resistance alleles during soybean breeding.

Discussion
In this study, a rapid approach combining high-throughput sequencing and traditional genetic mapping was deployed to identify a novel allele, RpsX, at the RpsQ locus in a small F2:3 population. The most prominent advantage of this approach is that it enables the genotyping and mapping of a resistance gene in a relatively early generation like F2:3. Additionally, the SNPs and InDels identified based on high-throughput sequencing can be used for further fine mapping, a haplotype analysis, and the identification of candidate genes [24,25,[38][39][40][41][42][43][44][45]. Phytophthora sojae virulence involves a complex mechanism and can rapidly change, enabling this pathogen to quickly overcome the resistance conferred by most Rps genes, ultimately leading to severe yield losses. Thus, whole-genome resequencing represents a fast, efficient, accurate, and relatively simple method for identifying novel Rps genes [24,25].
Many of the currently known Rps genes have been mapped on the short arm of soybean chromosome 3, including some genes that have been finely mapped such as Rps1k, RpsYD29, RpsQ, RpsHC18, RpsWY, RpsHN, and RpsUN1 [11,20,[24][25][26][27]30]. Moreover, most of these genes were mapped in the interval containing typical plant resistance genes encoding a nucleotide-binding site (NBS) and an LRR domain. Previous studies indicated that some genes encoding an NBS-LRR structure were candidate genes for these Rps genes [14,18,21,24]. The Rps1k gene, which is associated with broad-spectrum and durable resistance, has been cloned, and the tandemly arranged NBS-LRR genes Rps1k-1 and Rps1k-2 were functionally validated as responsible for

Discussion
In this study, a rapid approach combining high-throughput sequencing and traditional genetic mapping was deployed to identify a novel allele, RpsX, at the RpsQ locus in a small F 2:3 population. The most prominent advantage of this approach is that it enables the genotyping and mapping of a resistance gene in a relatively early generation like F 2:3 . Additionally, the SNPs and InDels identified based on high-throughput sequencing can be used for further fine mapping, a haplotype analysis, and the identification of candidate genes [24,25,[38][39][40][41][42][43][44][45]. Phytophthora sojae virulence involves a complex mechanism and can rapidly change, enabling this pathogen to quickly overcome the resistance conferred by most Rps genes, ultimately leading to severe yield losses. Thus, whole-genome resequencing represents a fast, efficient, accurate, and relatively simple method for identifying novel Rps genes [24,25].
Many of the currently known Rps genes have been mapped on the short arm of soybean chromosome 3, including some genes that have been finely mapped such as Rps1k, RpsYD29, RpsQ, RpsHC18, RpsWY, RpsHN, and RpsUN1 [11,20,[24][25][26][27]30]. Moreover, most of these genes were mapped in the interval containing typical plant resistance genes encoding a nucleotide-binding site (NBS) and an LRR domain. Previous studies indicated that some genes encoding an NBS-LRR structure were candidate genes for these Rps genes [14,18,21,24]. The Rps1k gene, which is associated with broad-spectrum and durable resistance, has been cloned, and the tandemly arranged NBS-LRR genes Rps1k-1 and Rps1k-2 were functionally validated as responsible for conferring complete resistance to P. sojae [47,48]. Unlike the Rps genes in tandemly arranged NBS-LRR gene clusters, RpsX was mapped to an interval on the short arm of chromosome 3 lacking NBS-LRR genes. Only one plant resistance gene, Glyma.03g027200, which encodes an STK-LRR structure, was detected in the RpsX region. Additionally, Glyma.03g027200 contains an RpsX-specific nsSNP, which was identified by QTL-seq (Figure 3). These results imply that Glyma.03g027200 is the likely candidate gene for RpsX, and may represent another gene type conferring resistance to P. sojae.
Interestingly, Glyma.03g027200 is also the candidate gene for RpsQ, which we previously identified and mapped [20]. The majority of the mapping intervals with RpsX coincided with the mapping intervals with RpsQ ( Figure 2B,C). A pedigree analysis revealed that Qichadou1, which contains RpsQ, was the result of a hybridization between Ludou4 and Peking, which is a PRR-susceptible cultivar [20]. In contrast, Xiu94-11 was obtained from a cross between lines 89-6 and Dandou806. Both of these lines were derived from landraces originating in northeastern China, whereas the ancestors of Qichadou1 were landraces originating in Shandong province, China, a Peking landrace, and the American cultivar Magnolia. Consequently, Xiu94-11 and Qichadou1 are not related. Moreover, our phylogenetic analysis based on the homozygous SNPs on chromosome 3 revealed a distant genetic relationship between Xiu94-11 and Qichadou1 and Ludou4. Interestingly, however, the Glyma.03g027200 allele sequences of RpsX and RpsQ are highly similar and carry the same SNPs and InDels. Therefore, we speculated that RpsX and RpsQ evolved independently via the same molecular mechanism in different ecological regions. The Glyma.03g027200 locus should be a mutation hotspot in the soybean genome, and additional PRR-resistance alleles may exist in soybean germplasm.
The Glyma.03g027200 sequence comprises an STK-LRR gene, which is another type of plant resistance gene. The STK-LRR resistance gene encodes an extracellular receptor-like protein kinase with an extracellular LRR and an intracellular STK. A few plant resistance genes have been cloned and verified to be STK-LRR genes, including rice genes conferring resistance to bacterial blight (Xa21, Xa21D, and Xa3/Xa26), a wheat leaf rust resistance gene (Lr10), and an apple scab resistance candidate gene (Rvi12_Cd5) [49][50][51][52][53]. In the present study, the candidate gene sequences were highly similar (97%) between the PRR-resistant genotypes Xiu94-11 (RpsX) and Qichadou1 (RpsQ) and the PRR-susceptible genotypes Zhonghuang47 and Williams82. Most of the sequence differences between the resistant and susceptible genotypes were detected in the sequence encoding the LRR domain, which plays a key role in the disease resistance mechanism of plants. The LRR domain determines the specificity of the recognition of pathogen effectors [54,55]. The LRR units containing approximately 20-30 amino acid residues consist of repeating core xxLxLxx motifs (L = leucine or other aliphatic amino acids, x = any amino acid) [56]. Allelic mutations due to SNPs and InDels in the sequence encoding the LRR domain may result in the generation of new R genes [57,58]. Because of the high sequence identity between the sequences encoding the LRR units in the LRR domain, unequal crossing-over and illegitimate recombinations are prone to occur, resulting in new R gene specificities due to the differences in the number of encoded LRR units [56,59,60]. Therefore, in this study, the 144-bp insertion of the RpsX and RpsQ candidate genes was caused by an unequal exchange with the adjacent 144-bp fragment, resulting in a new Rps gene.
InDel marker Insert144 is a co-segregated marker developed to detect 144-bp insertions in RpsQ. Additionally, this marker can distinguish RpsQ from the alleles at the Rps1 locus. Thus, this marker co-segregates with and can be used to detect RpsX. We used this marker to analyze diverse soybean genotypes and determined that it can detect PRR-resistance alleles at the RpsX/Q locus. Therefore, Insert144 may be applied for molecular marker-assisted selection. The PRR-resistant genotypes containing the 144-bp insertion are all highly resistant to P. sojae, indicating this insertion is important for the resistance to P. sojae.
In conclusion, we identified a novel allele, RpsX, at the RpsQ locus using a QTL-seq method involving high-throughput sequencing and traditional genetic mapping. A cluster analysis with homozygous SNPs on chromosome 3 and an analysis of the allelic sequences of the candidate gene confirmed that the genetic background of RpsX and RpsQ varies considerably, but both candidate alleles have the same 144-bp insertion in the sequence encoding the LRR domain. Therefore, changes to the LRR-encoding region play an important role in the development of novel Rps genes. The method described herein represents a rapid and efficient procedure for identifying novel Rps genes and may be useful for the cloning of RpsX and RpsQ as well as the application of RpsX and RpsQ functional markers for marker-assisted selection.

Phenotyping for PRR Resistance
Phytophthora root rot-resistant cultivar Xiu94-11 along with 22 cultivars with a single Rps gene and four PRR-susceptible cultivars were analyzed regarding their resistance to 14 P. sojae isolates that varied in terms of virulence. Additionally, 28 soybean cultivars and landraces related to Qichadou1 as well as Xiu94-11 and Zhonghuang47 were inoculated with 12 P. sojae isolates ( Figure 6) to identify the RpsX PRR-resistance alleles. Soybean plants were grown and inoculated with P. sojae isolates as previously described [24,25].
A mapping population was constructed with Zhonghuang47 as the female parent and Xiu94-11 as the male parent to derive F 1 seeds. An F 1 seed was used to produce 137 F 2:3 families by self-crossing. For each family, 20-25 seeds were sown in paper cups filled with vermiculite to evaluate the responses of the resulting plants to P. sojae isolates PsMC1 and PsJS2. The plants were inoculated and their phenotypes were evaluated as previously described [24,25].

Next-generation Sequencing and QTL-seq Analysis of Resistant and Susceptible Bulks
On the basis of the phenotypic evaluations, 30 homozygous resistant and 30 susceptible families were respectively used to construct PRR-resistant (R30) and PRR-susceptible (S30) bulks for a subsequent NGS with an Illumina system. The generated data underwent a QTL-seq analysis [25,38]. The DNA of the extreme-phenotype bulks and the Xiu94-11 and Zhonghuang47 cultivars were isolated with the Plant Genomic DNA Kit (Tiangen, Beijing, China). The raw read data generated for the Illumina libraries were filtered to produce clean reads, which were then aligned with the Glyma.Wm82.a2.v1 reference genome (http://phytozome.jgi.doe.gov/pz/portal.html) with the genome alignment software BWA [61]. The SNPs and InDels in the two extreme-phenotype bulks and parental cultivars were detected and filtered with the variation analysis software GATK [24,61].
The SNP index of each bulk and the delta SNP index were calculated based on an SNP detected in the two extreme-phenotype bulks during a previous filtering analysis [38]. Three confidence levels (P < 0.1, 0.05, and 0.01) were set for the delta SNP index [38], which was calculated for each SNP position with the following formula: delta SNP index = SNP index (PRR-resistant bulk) − SNP index (PRR-susceptible bulk).

Linkage Analysis and Genetic Mapping of the Candidate Region
After identifying the RpsX candidate region via QTL-seq, a genetic linkage mapping approach was used to further limit the RpsX candidate region. Publicly available SSR markers in the RpsX candidate region were selected to screen for polymorphisms between the parental cultivars and the genotype mapping population [47]. High-quality InDels identified by whole-genome resequencing were further developed to PCR markers to map the candidate region (Table S4) [24,25]. Combining phenotypic and genotypic results, the genetic linkage analysis of RpsX was completed with the MAPMAKER/EXP (version 3.0) program [62]. The genetic linkage map was constructed with the MapDraw program [63].

Phylogenetic Analysis of Soybean Genotypes
Forty-five soybean genotypes, including landraces/cultivars, were selected to analyze the genetic relationship between Xiu94-11 and Qichadou1/Ludou4. Moreover, in addition to Xiu94-11 and Zhonghuang47, 43 soybean genotypes were subjected to whole-genome resequencing (unpublished data). The DNA of these genotypes was extracted and sent to Annoroad Gene Technology (Beijing, China) for the construction of Illumina sequencing libraries. Raw data were filtered, sequences were aligned with reference genome sequences, and homozygous SNPs were identified and annotated as previously described [24,25]. The homozygous SNPs identified on chromosome 3 for each genotype were used to construct a phylogenetic tree according to the neighbor-joining method of the MEGA 6.0 program [64].

Analysis of the Allelic Sequences of the Candidate Gene Locus
The allele sequences of the Glyma.03g027200 locus for the cultivars related to Qichadou 1 were determined by PCR-based Sanger sequencing. The sequencing primers were the same primer pairs used to amplify the RpsQ candidate allele and overlapped new primers designed with the NCBI Primer-BLAST tool (https://www.ncbi.nlm.nih.gov/tools/primer-blast/) (Table S5). A PCR assay was completed with PrimeSTAR™ HS DNA Polymerase (Takara Biotechnology, Dalian, China). The PCR product obtained for each sample was sequenced by Sangon Biotech (Beijing, China). The resulting sequences were assembled with ContigExpress to obtain the complete Glyma.03g027200 allele sequences. Each sample was amplified and sequenced three times to avoid errors generated during amplification and sequencing.
A multiple sequence alignment involving the obtained sequences was completed with ClustalW, after which a phylogenetic tree was constructed according to the neighbor-joining method (with 1000 bootstrap replicates) of the MEGA 6.0 program [64]. The sequence coding region was confirmed based on the cDNA sequence of the RpsQ allele and the published annotated Williams82 reference transcripts (https://www.soybase.org/). The coding region sequence was converted to a protein sequence with ExPASy (https://web.expasy.org/translate/). Conserved domains were predicted for the Xiu94-11, Zhonghuang47, Qichadou1, Ludou4, and Williams82 sequences with the Conserved Domain Database (http://www.ncbi.nlm.nih.gov/cdd/) [65] and SMART (http://smart.embl-heidelberg.de/) [66].

Validation and Screening of RpsX and Its Alleles in Soybean Genotypes Using the Functional Marker Insert144
Cultivars containing single Rps genes were genotyped with the Insert144 marker that co-segregated with RpsQ. This marker was developed based on the 144-bp insertion in the RpsQ allele. Insert144 was also used to detect the RpsX and RpsQ/9 resistance alleles in 177 soybean cultivars and landraces whose reactions to eight P. sojae isolates had been analyzed. The PCR assay was completed as previously described.

Acknowledgments:
We thank Lijuan Qiu from the Institute of Crop Sciences, Chinese Academy of Agricultural Sciences, and Ran Xu from the Shandong Academy of Agricultural Sciences for supplying the soybean cultivars tested in this study.

Conflicts of Interest:
The authors declare that they have no conflict of interest.

LRR
Leucine-rich repeat PRR Phytophthora root rot NGS Next-generation sequencing SNP Single nucleotide polymorphism InDel Insertion/deletion STK Serine/threonine protein kinase NBS Nucleotide-binding site