Conrmation of Rpp Genes Conferring Resistance to Asian Soybean Rust and Mapping of Rpp1 Allele from PI 594723

In this study, we aim to develop and validate KASP molecular markers in soybean populations for Asian soybean rust (ASR) resistance gene Rpp1 (PI 200492, PI 594538A, PI 587880A), identify the gene hypothetically present in PI 594723, and validate KASP markers for Rpp2 (PI 230970), Rpp3 (PI 506764), Rpp4 (PI 459025A), and Rpp5 (PI 506764, PI 200487). Ten F 2 soybean (Glycine max (L.) Merrill) populations derived from crosses between rust-susceptible (55I57RSF IPRO, 63I64RSF IPRO) x rust-resistant sources (PI 200492, PI 594738A, PI 587880A, PI 594723, PI 230970, PI 506764, PI 459025A and PI 200487) were evaluated. All F2 plants were individually evaluated in eld conditions for ASR phenotypic reactions, classied according to sporulation level. SNP markers were developed according to markers associated with Rpp genes available at the SoyBase, using KASP methodology. Based on a slight difference in map position and different phenotypic disease reactions of PI 200492, the authors suggest that PI 594723 carries a resistance gene Rpp1-b. The Rpp1-b gene from PI 594723 was mapped in Chr 18 in a 12.4 cM region. The PIs carrying Rpp1-b (PI 594723, PI 587880A, and 594538A) showed strong resistance to ASR compared to the lines carrying Rpp1 (PI 200492). A total of 26 KASP markers were signicantly associated (P < 0.01) with ASR. Among those, M1, M5 and M6 (Rpp1), M13 and M14 (Rpp2), M16, M17 and M20 (Rpp3), M25 and M26 (Rpp4), and M27 and M28 (Rpp5) have the potential to be used in marker-assisted selection strategies.


Introduction
Asian soybean rust (ASR), caused by the biotrophic fungus Phakopsora pachyrhizi Syd. & P. Syd, is considered one of the most damaging soybean (Glycine max (L.) Merrill) diseases worldwide (Langenbach et al. 2016;Godoy et al. 2016). Because of this pathogen dissemination and associated losses to soybean yield, there is a need for greater effort in researching ASR resistance (Meira et al. 2020). The pathogen can infect 31 leguminous species in natural conditions, such as G. max, G. soja, and Vigna unguiculata, and more than 60 different species in controlled conditions (Goellner et al. 2010).
P. pachyrhizi is present in main soybean producer regions, mainly because of the windborne urediniospores dispersion. When the urediniospores reach the leaf, ideal conditions of surface moisture and temperature (17 to 28 ºC) initiate the germination process in a couple of hours. Most rust species enter the leaf through the stomata, but P. pachyrhizi direct penetrates the leaf epidermal cell wall by appressorial peg. After the latent period of ve to eight days, small chlorotic spots on older leaves are observed on the abaxial side. These lesions advance to volcano-shaped uredina, which produce innumerous urediniospores responsible for the new infection cycle (Goellner et al. 2010). ASR causes early defoliation and reduces photosynthetic area, resulting in yield losses and increased costs (Langenbach et al. 2016;Godoy et al. 2016).
Several methods have been developed to control ASR, such as eld monitoring, elimination of secondary hosts, use of soybean-free periods to break the fungus cycle, fungicides, and genetic resistance (Kendrick et al. 2011). In the last years, fungicide e ciency has decreased due to the intense use and lower pathogen sensitivity to different fungicide classes (Goellner et al. 2010;Langenbach et al. 2016). Thus, soybean breeders and geneticists have focused on incorporating genetic resistance or tolerance in high yielding materials, aiming to combine different control methods through an integrated management approach.
Long-term resistance is di cult to achieve due to the diversity of pathogen isolates and race-speci c monogenic resistance of each Rpp gene against ASR isolates (Aoyagi et al. 2020). The same resistance source may present different phenotypic reactions according to the geographic origin of the isolate. In general, the Brazilian ASR isolates are known as the most aggressive. The Rpp1 (PI 200492) is highly resistant to Japanese and Mexican ASR isolates, but a lack of resistance has been reported to Brazilian ASR isolates (Aoyagi et al. 2020;Akamatsu et al. 2017;Hossain et al. 2015). In contrast, the Rpp1-b (PI 587880A, PI 594538A) is resistant to most Brazilian ASR isolates (Ray et al. 2009;Yamanaka et al. 2016;Akamatsu et al. 2017), highlighting a clear allelic difference between Rpp1 and Rpp1-b. Moreover, the PI 594723 hypothetically present Rpp1-b because of the phenotype similarity to PIs carrying Rpp1-b (Li, 2009). Miles et al. (2008) reported high resistance of PI 594723 against Paraguayan ASR isolates.
Introgression of resistance genes from plant introductions (PIs) in elite lines is an e cient way to release varieties with resistance to ASR. Marker-assisted selection (MAS) may guarantee selection in early generations, reducing phenotyping time and selecting only plants with the desirable allele combination. Single nucleotide polymorphism (SNP) markers are considered the best choice among contemporary breeding programs, mainly due to high throughput and low cost. In this way, a genotyping strategy using KASP (Kompetitive Allele-Speci c PCR) methodology, a variant of PCR and based on allele-speci c oligo extension, is a high-throughput and breeder-friendly uorescence-based genotyping platform for SNP markers. KASP offers scalable exibility in applications that require small to moderate numbers of markers. In soybean, the KASP technique has been successfully used to map Rpp7 from PI 605823 in Chr 19 (Childs et al. 2018).

Plant material
Ten F 2 populations derived from single crosses between different soybean rust-susceptible cultivars and the resistant sources (PIs) carrying Rpp genes were developed ( Table 1). The susceptible parental used in the crosses were highly cultivated genotypes in Brazil and contained the RR2BT trait. Crosses were performed in two phases, in 2017 for populations 1 to 4 and 2018 for populations 5 to 10.
The resistant sources (PIs) were used as male, and the susceptible soybean cultivars as female. F 1 hybrids were grown in greenhouse conditions, and bulk harvested for the F 2 generation were performed. The eld experiments were designed to evaluate all the F 2 populations and were performed in two phases. First (Table 2). All F 2 plants were evaluated individually, totaling 1990 individuals (Table 3) used to disease rating, and 50 plants from each resistant and susceptible parent were rated. The experiments were planted on a non-preferential date (December) to enable the natural occurrence and development of Asian soybean rust, and no fungicide application was made to control the disease.

KASP markers
The SNP markers used in this study were developed according to markers available at the SoyBase (https://www.soybase.org), linked with Rpp genes (Table 2). Seventeen KASP markers were used to map the Rpp1 gene in the rst experiment (Table S1). KASP markers highly associated with Rpp1 in PI 594723 and markers developed for Rpp2, Rpp3, Rpp4, and Rpp5 were used in the second experiment (Table S2). All consist of SNP markers developed using the KASP methodology (http://www.lgcgroup.com).
DNA analysis DNA was extracted from young leaf tissue of each F 2 plant at the V4 growth stage, using a silica column kit of LGC Genomics (Teddington, UK). Genotyping assays were tested in a 96-well format and set up as 10 µL reactions (4.85 µL of template (50-75 ng of DNA), 5.0 µL of 2 x Kaspar mix, and 0.15 µL of primer mix). PCR was performed according to the protocol: an initial 15 min at 94 ºC; 10 Touchdown cycles of 94 ºC for 20 s, 65 − 57 ºC for 60 s (dropping 0.8 ºC per cycle); 26 ampli cation cycles of 94 ºC for 20 s, 57 ºC for 60 s; with nal extension for 7 min at 72 ºC. The uorescence data were collected in the pre-read and post-read stages (37 ºC for 1 min).
Data were automatically processed using KBioscience Kraken software and visually checked using KBioscience SNPViewer (LGC Limited, UK).

Statistical analysis
Genotypic and phenotypic data of individual plants evaluated were analyzed. Observed and expected segregations ratios of ASR resistance and KASP markers were tested using Chi-square (χ²) analysis. The expected segregations were 1:2:1 (dominant homozygous, heterozygous, and recessive homozygous) to markers, 3:1 (resistance and susceptible to ASR) to phenotype, and population 7 (Rpp3, 5) with two genes was 15:1. Phenotypic data were converted into resistant (R) summing IM, RB1, and RB2 plants; and as susceptible (S) summing the number of plants with RB3 and TAN lesions.
Linkage map analysis was performed to each mapping population (10 populations) using the MSTmap software (http://mstmap.org/mstmap_online.html), with Single LG to grouping LOD (logarithm of the odds), a threshold of 15 cM to no mapping distance, and Kosambi mapping function to convert recombination values into map distances (cM). QTL mapping was performed using the composite interval mapping (CIM) functionality in the R package qtl (Broman et al. 2003). QTL positions for lesion type were de ned as the peaks of maximum LOD score, and the signi cance thresholds were calculated by a 1000 permutation test analysis at α ≤ 0.05 signi cance level. QTL intervals were estimated via loading function, using 1.5-LOD support con dent intervals. Single marker regression analysis was performed for each marker to test the signi cant association between markers and the ASR phenotypes and determine how much phenotypic variation could be explained for each KASP marker. Additive allelic effects were estimated by the substitution of resistant allele (AA) to susceptible allele (BB).

Phenotype resistance
Most F 2 soybean populations evaluated t the expected segregation ratio of 3:1 (resistant: susceptible) for a dominant resistance gene (Table 3). Populations P1 -Rpp1, P6 -Rpp2, P8 -Rpp4, and P10 -Rpp5 did not t the 3:1 ratio. Also, population 7 -Rpp3, 5 did not follow the expected segregation ratio for the two resistant gene model (15:1). These results can be associated with eld conditions, where the populations were submitted to natural ASR infection, and more than one isolate could be present. The susceptible parents 55I57RSF IPRO and 63I64RSF IPRO produced TAN lesions, con rming the pathogen presence and their susceptibility (Fig. 2, Fig. 3).
Population P8 -Rpp4 showed weak resistance to ASR with 79% of the plants with RB2 or RB3 lesions of 291 plants. This weak resistance to ASR is related to the disease reaction from the resistance source PI 459025A (Fig. 3d). P9 -Rpp5 and P10 -Rpp5, with 284 plants each population, carried the resistant allele Rpp5 (Chr 3) from PI 200487 and showed no sporulation lesion type (RB1) in only 14% of the plants, RB2 in ~ 40%, and RB3 in 25-29.6% (Fig. 3e, 3f).

Mapping of resistance loci to ASR
Genotypic data revealed signi cant QTLs associated with the phenotypic reaction to ASR for all populations evaluated, except for P1 -Rpp1 (Tables 4 and 5). In P1, no signi cant QTL was identi ed, and no markers showed signi cant association to phenotype reaction (Table S3). This result can be associated with weak resistance (RB3 lesion) of the parent carrying the Rpp1 gene on Chr 18 (PI 200492) (Fig. 2a), lack of phenotypic difference between homozygous resistant (AA) and homozygous susceptible (BB) genotypes (Table S3), or even related to the low number of markers used. The F 2 population showed a distorted segregation ratio that did not t a ratio of 3:1 (Table 3). Aoyagi et al. (2020) and Akamatsu et al. (2017) reported the susceptibility of PI 200492 to ASR isolates from Brazil. P2 and P3, carrying Rpp1-b (Chr 18), presented a QTL in the same region with the LOD peak at marker M6, and an additive effect ranged from 1.70 and 1.77 (Table 4). The QTL explained 68.3% and 57.6% of the phenotypic reaction to ASR for P2 -Rpp1-b and P3 -Rpp1-b, respectively (Table S3). The resistance locus Rpp1-b in P2 was mapped between M5 and M10 (6.5 cM) and between M4 and M10 (8.2 cM) in P3 (Fig. S1). The Chi-square test revealed that all KASP markers mapped in P2 and P3 satisfactorily tted the expected ratio for co-dominant inheritance (1:2:1) (Table S3).
PI 594723 presented strong resistance to ASR ( Fig. 2d; Fig. 3a), and it is hypothesized that PI 594723 carries an unknown Rpp1* gene. A QTL was detected in P4 -Rpp1* between markers M1 and M6 on Chr 18 (Table 4). The QTL identi ed in P4 was validated in the P5 in the same region (M1 to M6). The QTLs explained 42.2% and 27.8% of the phenotypic variation for P4 and P5, respectively (Table S3, Fig. 4). The additive effects of this locus to increase susceptibility ranged from 0.67 to 0.80.
The Rpp2 gene was rst mapped in PI 230970 on Chr 16 by Hartwig and Brom eld (1983). Yu et al. (2015) performed the ne mapping of this gene and de ned the mapping in a physical interval of 188.1 kb on Chr 16. In P6 -Rpp2, the KASP markers used were developed based on the physical position of these markers. A QTL was identi ed on Chr 16 between markers M13 and M14 (3 cM), with the peak at M14 and explained 14.1% of the phenotypic variation for ASR resistance in the population (Table 4, Fig. S1). All markers were signi cantly associated with ASR (Table S3).
Population P7 carries two Rpp genes (Rpp3, Rpp5) from PI 506764. Rpp3 was mapped on Chr 6 between Satt307 and satt460 (Hyten et al. 2009), and Rpp5 on Chr 3 between Sat_275 and Sat_280 (Kendrick et al. 2011). The Rpp3 and Rpp5 genes were con rmed, and all KASP markers used were associated (P ≤ 0.002) with ASR (Tables 4 and 5). The Rpp5 gene was mapped between markers M28 and M27 on Chr 3 and explained 5.8% of the phenotypic variation for ASR. Rpp3 was mapped on Chr 6 between markers M20 and M17 (Fig.  S2) and explained 12.4% of the phenotypic response to ASR resistance.
Genotypic data revealed a signi cant QTL associated with ASR on Chr 18 between the markers M26 and M22 on P8 -Rpp4 (Table 4, Fig. S2), con rming the Rpp4 locus in the PI 459025A (Silva et al. 2008). M26 was associate (P ≤ 0.009) with ASR and explained 3.2% of the phenotypic variation for the trait (Table S3). Populations P9 and P10 have the Rpp5 from PI 200487. The Rpp5 locus was mapped between the markers M27 and M28 (Fig. S2) and explained 5.8-6.4% of the phenotypic variation (Table 4 and Table S3). The additive effect for the QTL ranged from 0.38 to 0.44, dependent on the population.

Discussion
Several factors could lead to inconsistent results in the segregation ratio of F 2 soybean populations evaluated in this study. The trials were conducted in eld conditions, where the combination of different ASR isolates, natural infection and weather conditions could promote high inoculum pressure. In addition, the ASR isolates presented in Brazil are considered more virulent than ones found in Japan, Argentina, and Paraguay (Yamanaka et al. 2010;Aoyagi et al. 2020 Hossain et al. (2015) and Akamatsu et al. (2017) corroborate these differences in reactions to different isolates between accessions with the Rpp1 locus and genotypes with Rpp1-b.
The phenotypic reaction and mapping location suggested that PI 594723 carries the locus Rpp1-b. Ray et al. (2009) identi ed the gene Rpp1-b on Chr 18 from PI 587880A in the same region we identi ed the Rpp1* locus on P4 and P5. In a few studies performed with this resistance source, Miles et al. (2008) reported resistant RB lesion type, with reduced sporulation level and low severity, in greenhouse and eld conditions in Paraguay. However, Li (2009) observed moderate resistance of PI 594723 to Mississippi isolates. Pedley et al. (2019) studying the source of resistance at the Rpp1 locus (PI 200492), in the previous locus region mapped between markers Sct_187 and Sat_064 (Hyten et al. 2007) (Gm18: 56,182,230 to 56,333,803), identi ed a leucine-rich repeat (NBS-LRR) protein with a novel ubiquitin-like-speci c protease 1 (ULP1) domain associated to the Rpp1. The NBS-LRR domain is usually associated with R proteins that detect the pathogen and start signaling pathways.
Regarding the eight R genes reported in the Rpp1 locus, R1 and R2, R6 to R8 are located before and after the physical position of Rpp1, respectively (Pedley et al. 2019). This fact can explain the immunity, and hypersensibility reaction to ASR observed in this study to PI 587880A, PI 594538A, and PI 594723 carrying Rpp1-b (Fig. 1). Chakraborty et al. (2009) mapped Rpp1-b on PI 594538A between markers Sat_064 and Sat_372, which agree with the physical position of R6 to R8. The Rpp1-b mapped on PI 587880A was located between markers Sat_191 and Sat_187 (Ray et al. 2009), which is in the same physical position of R1 and R8 resistance genes. Thus, according to the mapping proposed of PI 594723, eight homologous to the NBS-LRR family of disease R genes could be present (Pedley et al. 2019).
The markers previously mapped to Rpp1-b in PI 594723 (M1, M6, and M11) were con rmed in the P5 -Rpp1-b. This population avoids pathogen infection through hypersensibility reactions, resulting in lesions without sporulation, known as RB1 (Fig. 1). This resistant source has great potential to be used in breeding for ASR resistance, especially in South America. The anking and interval KASP marker used in these populations (P4 and P5 -Rpp1-b) allows it to select plants with strong resistance.
The PI 230970 carries the dominant gene Rpp2 (Hartwig and Brom eld, 1983;Silva et al. 2008). In addition, Yu et al. (2015) ne mapped Rpp2 from PI 230970 into a 188.1 kb region. Our results con rmed the Rpp2 gene from PI 230970 on Chr 16. However, the phenotypic segregation ratio for P6 -Rpp2 did not follow the expected 3:1 ratio. A good explanation for that may be that multiple isolates could be presented in the area since we had a natural infestation. Garcia et al. (2008) observed a similar trend when they used a different ASR isolate than previously used by Brom eld & Hartwig (1980) to map Rpp2 from PI 230970.
Despite PI 506764 natural gene pyramiding (Rpp3 and Rpp5), the phenotype reaction to ASR by P7 (Rpp3, 5) corroborates with results obtained by Walker et al. (2014). These authors described the PI 506764 as presenting a medium level of sporulation and RB lesions to ASR (Walker et al. 2011;Walker et al. 2014;Kendrick et al. 2011). Aoyagi et al. (2020) genotyped soybean landraces (WV51 and WC61) carrying Rpp3 and reported different phenotypic reactions to isolates, presenting slight resistance to Brazilian isolates, agreeing with our results. The Rpp5 (PI 200487), mapped in Chr 3 between markers Sat_275 and Sat_280 (Garcia et al. 2008) showed great potential to be used in breeding programs. Our results demonstrated the contribution of Rpp5 to increase levels of resistance in pyramided-lines contain Rpp3 + Rpp5 (Fig. S3).
The resistant source carrying Rpp4 (PI 459025A) presented satisfactory resistance to ASR. According to , this PI showed strong resistance against 80% of isolates from Bangladesh and Japan. However, when this source was submitted to South American isolates (Brazil, Argentina, and Paraguay), the PI 459025A showed resistance to 50% of the isolates. Rpp4 and Rpp1 were mapped in the same linkage group on Chr 18, and this region is considered a hotspot for ASR resistance in soybean (Hyten et al. 2007;Silva et al. 2008). Previous studies with Rpp4 veri ed a biphasic response to ASR, proposing that the gene detect effectors in the haustoria developing stage due to one or more of the multiple TIR-NBS-LRR candidate genes in the region (Meyer et al. 2009). These authors support the hypothesis that susceptibility to ASR can be associated with small amino acid differences responsible for playing a key role in resistance.
Pyramiding resistant genes in a single line can confer more durable and broad-spectrum resistance to a pathogen.  observed higher resistance to ASR when combined in one line multiple Rpp genes depending on the isolate. The KASP markers validated in this study might be used in MAS strategies to pyramiding different Rpp genes in one single line.
In conclusion, based on a slight difference in map position and a different reaction to ASR of PI 200492, the data suggested that PI 594723 carries a resistance gene Rpp1-b. The PIs carrying Rpp1-b (PI 594723, PI 587880A, and 594538A) showed strong resistance to ASR and generated high resistance plants when crossed with susceptible commercial cultivars. A total of 26 KASP markers were signi cantly associated (P < 0.01) with ASR and successfully mapped the resistant loci Rpp1, Rpp2, Rpp3, Rpp4, and Rpp5. Among the 26, M1, M5, and M6 (Rpp1), M13 and M14 (Rpp2), M16, M17 and M20 (Rpp3), M25 and M26 (Rpp4), and M27 and M28 (Rpp5) have the potential to be used in marker-assisted selection strategies.

Declarations
Authors' Contributions DM and VB conducted eld evaluations and data analysis; MP, LM, FB and AB performed data collection and genotyping; DM, VB, LW, AM, EB wrote the paper; GB, GM and SJ supervised the research; GB, GM, SJ, TF, EB revised and edited the manuscript. All authors read the manuscript.