Identification and mapping of quantitative trait loci associated with soybean rust ( Phakopsora pachyrhizi ) resistance in genotype UG 5

Soybean rust, Phakopsora pachyrhizi, is one of the most serious and widespread foliar diseases of soybean causing high yield losses world-wide. The objective of this study was to identify and map quantitative trait loci (QTLs) resistant to soybean rust in genotype UG 5. Ninety-seven F2 mapping plants, obtained from a cross between Wondersoya and UG 5, were used for this study. Quantitative trait locus analysis using QTL IciMapping software identified three putative QTLs associated with soybean rust (SBR) on chromosomes 6, 9 and 18 with logarithms of odds (LOD) scores ranging from 3.47 to 8.23 and phenotypic variance explained by the QTLs ranging from 18.3 to 25.6%. The putative QTL detected on chromosome 9 is novel and has not been reported elsewhere. The putative QTLs identified in this study could help to facilitate SBR resistance breeding towards efficient markerassisted selection approach and gene pyramiding leading to the development of durable resistance.


INTRODUCTION
Soybean (Glycine max [L.] Merrill) is one of the most important legume crops in the world providing a major source of high-quality protein and oil for human consumption and livestock feeds.However, soybean is attacked by a wide range of pathogens.Soybean rust (SBR), caused by Phakopsora pachyrhizi, is the most severe destructive foliar disease leading to high loss in yield and quality of soybean.Soybean rust was first reported in 1902 in Japan (Hennings, 1903) and subsequently spread from Asia to Africa, South America (Yorinori et al., 2005) and the United States of America (Schneider et al., 2005) through air-borne movement of urediniospores.In Africa, soybean rust was reported in Uganda, Kenya and Rwanda in 1996 (Tukamuhabwa et al., 2001), Zambia and Zimbabwe in 1998, Mozambique in 2000and South Africa in 2001(Levy et al., 2002) and *Corresponding author.E-mail: himeri2004@yahoo.com.
Author(s) agree that this article remain permanently open access under the terms of the Creative Commons Attribution License 4.0 International License further spread in the westward and central direction to Nigeria (Akinsanmi et al., 2001), Ghana (Bandyopadhyay et al., 2007), and Democratic Republic of Congo (Ojiambo et al., 2007).P. pachyrhizi has a unique ability to infect a broad range of legume species that contribute to a diverse and complex virulence pattern (Hartman et al., 2005).Considering the explosive nature of the disease and the high potential yield losses (up to 80%), soybean rust has long been viewed as a serious threat to soybean production worldwide.The development of resistant varieties to soybean rust could reduce the impact of the disease without the expensive, timeconsuming and negative environmental impact of foliar fungicide applications.
So far, several major sources of soybean rust resistance genes (Rpp1 to Rpp7) have been identified in soybean.However, the high virulence and variability of the pathogen isolates pose problems leading to breakdown of resistance.For example, resistance due to Rpp1 to Rpp4 have already been broken in China (Shan et al., 2012) and other three improved soybean rust resistant varieties (Namsoy 4M, Maksoy 1N, and Maksoy 2N) have succumbed to soybean rust in Uganda (Tukamuhabwa et al., 2009), suggesting that the SBR resistance genes are not durable.Therefore, discovering and mapping additional resistance genes in soybean is crucial to further improve the SBR resistance and develop durable SBR-resistant cultivars.
UG 5 is a locally available genotype showing good resistance to SBR in Uganda, which seems to have a unique gene controlling resistance to soybean rust.It was the only line found resistant to different isolates of SBR in Uganda since 1996 among the Ugandan germplasm collections (Maphosa et al., 2013;Kawuki et al., 2003).UG 5 was also found to be SBR resistant to different P. pachyrhizi isolates in Nigeria (Twizeyimana et al., 2009) and USA (Twizeyimana and Hartman, 2012).However, the genes in UG 5 controlling resistance to soybean rust are not yet identified and mapped.The objective of the present study was, therefore, to identify and map quantitative trait loci (QTLs) associated with resistance to soybean rust in UG 5 genotype.

Study site
Phenotypic evaluation and molecular work were carried out at Makerere University Agricultural Research Institute-Kabanyolo (MUARIK), Uganda, from 2017 to 2018 in screen-house and biotech laboratory, respectively.

Development of mapping population and P. pachyrhizi isolate evaluation
The parental soybean genotypes used for the development of mapping population were soybean rust-susceptible genotype (Wondersoya) from Nigeria and Uganda local SBR resistant genotype UG 5.The susceptible genotype Wondersoya as a female parent was crossed to the resistant genotype UG 5 to develop a mapping population consisting of 97 segregating F2 (Wondersoya x UG 5) plants for SBR.The F2 mapping population and the parental genotypes were grown in plastic pots and scored for SBR resistance.Three plants were maintained in each pot filled with soil from the field in order to have adequate plants for leaf sampling and phenotypic evaluation.The P. pachyrhizi pressure was readily available in the screen-house which had favorable conditions for the pathogen and was maintained on SBR susceptible soybean genotypes (Wondersoya and Nam-II).Therefore, the parental genotypes and the F2 progenies were evaluated against P. pachyrhizi urediniospores in a screen-house under natural infestation.The data was recorded when the plants reached R6 reproductive stage (full-seed stage).Plants were evaluated for soybean rust reactions by examining disease severity (DS) based on a 1 to 5 scale (Miles et al., 2008) and lesion types.Reddish brown (RB) lesion types are associated with resistance while TAN lesions are indicators of susceptibility.Plants with DS score of 1 to 3 were considered resistant, while those with DS score of 4 or 5 were considered susceptible (Souza et al., 2014).

DNA extraction and marker analysis
Genomic DNA was extracted from young leaves of the parental genotypes and 97 individual F2 plants using cetyltrimethyl ammonium bromide (CTAB) method (Lemos et al., 2011).The concentration of DNA samples was determined using a nano-drop spectrophotometer from the absorbance data of DNA sample at 260 nm.The purity of the DNA sample was determined by A260/A280 ratio (1.8 to 2.0 of pure DNA).The integrity of the extracted DNA was estimated on 0.8% agarose gel electrophoresis.Subsequently, DNA was diluted to a final concentration of 50 ng/l for polymerase chain reaction (PCR).Out of the 97 F2 leaf samples taken, the DNA of 86 samples was with good quality and was used for genotyping.
For the linkage analysis, a total of 122 SSR markers were chosen based on their distribution throughout the integrated molecular linkage map of soybean (Song et al., 2004) including those markers flanking the previously mapped Rpp genes and were PCR amplifications were performed in Thermo Cycler Block (96 universal gradient, Thermo Scientific ® ) in 10 μl final volume containing 5 l premix (AccuPower ® PCR Master Mix containing 100 mM dNTPs, 1.0 U Taq DNA polymerase; BiONEER C&D Center, South Korea), 0.25 l of each primer (10 pM), 1 l of template DNA (50 ng) and 3.5 l of ddH2O.The PCR thermo-cycler was programmed with an initial denaturation step at 95°C for 5 min (preheating) and 35 cycles each with 30 s DNA denaturation at 95°C, 30 s annealing at 55°C and 40 s extension at 72°C followed by a final extension step at 72°C for 5 min (to fill in the protruding ends of the newly formed PCR products) and a 4°C soak (for preservation till the products are taken out from the machine).The PCR products were finally separated on 3% (w/v) agarose gel for 2 h at 120 V in 1 X TAE buffer using a gel electrophoresis apparatus (Model V16.2 Gibco BRL, Gaithersburg, MD, USA).Gels were visualized under UV trans-illuminator (M-15 UVP Upland, CA 91786 USA) and photo-documented with a digital camera.DNA fragment sizes were determined based on a 100 bp DNA standard ladder (BiONEER C&D Center, South Korea) and marker alleles of SSRs were scored manually.

QTL mapping and statistical analysis
Chi-square ( 2 ) analysis was used to test Goodness-of-fit between observed and expected segregation ratios of soybean rust phenotypes and genotypes of SSR markers in the F2 population.Analysis of variance and regression analysis were used to test the significance of the association between SBR phenotype and flanking markers and to estimate how much phenotypic variation could be explained by flanking markers, respectively.QTL IciMapping version 4.1 software (http://www.isbreeding.net)was used for linkage and QTL mapping.IciMapping uses inclusive composite interval mapping (ICIM), which is a modified algorithm of composite interval mapping (CIM) (Li et al., 2007).The SSR marker orders and distances were calculated using the MAP functionality in QTL IciMapping with a LOD score of 3.0 as a linkage threshold and a recombination frequency value of 0.30.The linkage map was constructed with genetic distances (cM) calculated using the Kosambi function (Kosambi, 1943) and linkage groups were named based on the chromosome information of the genomic sequence (http://soybase.agron.iastate.edu).Genomic regions significantly associated with disease severity were detected as QTLs using BIP functionality in QTL IciMapping with a significance logarithms of odds (LOD) threshold of 3.0.The estimated order of markers determined by the QTL IciMapping software was used for QTL analysis.The QTL positions for the disease severity were defined as the peaks of maximum LOD score.

Reaction of parental lines and progenies to soybean rust
Phenotypic evaluation of soybean parental lines showed variation in their reaction to SBR (Table 2).The resistant parental genotype, UG 5 produced typical RB lesions with a mean disease severity (DS) score of 1.8, while the susceptible parental genotype, Wondersoya produced TAN lesions with a mean DS score of 4.0.The DS score ranged from 1.2 to 2.8 in genotype UG 5 and 3.0 to 5.0 in genotype Wondersoya.The disease severity range for the F 2 progenies was 1.3 to 5.0 with a population mean of 2.8.The F 2 plants with RB lesions had a mean disease severity score of 2.3 ranging from 1.3 to 2.8, while those with TAN lesions had a higher mean severity score of 4.0 ranging from 2.9 to 5.0.
Segregation of the F 2 mapping population is shown in Table 3.The F 2 mapping population showed segregation of 69 plants with resistant phenotype and 28 plants with susceptible phenotype based on disease severity score.Moreover, based on the lesion types, 70 plants showed RB lesions and 27 plants showed TAN lesions among the 97 F 2 plants of the mapping population.A chi-squared test in both cases revealed that the observed segregation fitted well with the expected segregation ratio of a single dominant resistance gene, 3:1 (resistant: susceptible) in the F 2 generation.

Marker screening and segregation distortion
A total of 122 SSR markers were tested for polymorphism between the resistant and susceptible parental genotypes, out of which 33 SSRs were found to be polymorphic showing 27.05% of polymorphism.The distribution of the different genotypes among the F 2 populations is shown in Table 4.The majority of markers showed an excess of the heterozygote genotypes with the exception of four markers (Satt294, Satt444, Satt288 and Satt440), which showed an excess of homozygote alleles for susceptibility to soybean rust and five markers (Satt309, Satt100, Satt229, Satt442 and Sat_263), which showed an excess of homozygote alleles for soybean rust resistance.The Chi-squared test analysis for the polymorphic markers showed 42.4% segregation distortion which deviated significantly (critical  2 = 5.99; d.f.= 2; P ≤ 0.05) from the 1:2:1 Mendelian segregation ratio in the progeny mapping population.

Genotyping and linkage mapping analysis
Linkage analysis was performed using QTL IciMapping 4.1 software with 86 F 2 individuals and 33 polymorphic SSR loci.This resulted in the formation of eight linkage groups (LGs) comprising 25 SSR loci (Figure 1).The number of SSR markers in an individual chromosome or linkage group (LG) varied from two (LGs-K, B1, J and L) to six (LG-G).The remaining 8 SSR markers were found to be unlinked.

QTLs associated with resistance to soybean rust in UG 5
The putative QTLs associated with soybean rust resistance detected on genotype UG 5, their respective Homozygous for "UG 5" allele : Heterozygous : Homozygous for "Wondersoya" allele;  2 = calculated Chi-squared value according to the expected Mendelian genotypic segregation ratio of 1:2:1 (Critical  2 = 5.99); d.f.: degree of freedom; b Significance level of 5%.
positions and effects are shown in Table 5.Three QTLs, explaining 63.13% of the total phenotypic variation in the population, were detected on three different linkage groups associated with SBR resistance by QTL IciMapping with a genome-wide LOD threshold of 3.0 (Table 5 and Figure 2).The QTL with the highest peak was located on chromosome 18 (LG-G) at a LOD score of 8.18 and accounted for 25.71% of the phenotypic variation in the population.This QTL was flanked by markers Sat_064 at a distance of 6.52 cM and Sat_372 at a distance of 30.31 cM with additive and dominance effects of -0.7336 and -0.5066, respectively.Two other QTLs were detected on chromosome 6 (LG-C2) and chromosome 9 (LG-K) at a LOD score of 3.47 and 7.36, respectively.The phenotypic variance explained by these two QTLs was 18.27 and 19.15%, respectively.The QTL on chromosome 6 was located at distance of 21.5 cM from Satt643 and 39.0 cM from Satt281 with additive and dominance effects of -0.626 and -0.2639, respectively.On chromosome 9, the QTL was flanked by SSR markers Satt264 and Satt337 at a distance of 4.0 and 3.99 cM, respectively, with additive effect of -0.4293 and dominance effect of -0.7738.

DISCUSSION
Host-plant resistance and/or tolerant is one of the best strategies for soybean improvement to soybean rust.Importance of introgression of resistance genes into soybean crops is increasing as fungicides lose efficacy due to adaptation of the pathogen as well as the concern for environmental pollution causing human health problems and increased production costs of chemicals.In many cases, pyramiding genes into elite cultivars is required for sustained resistance to soybean rust which requires identification and mapping of additional genes resistance to soybean rust.
In the current study, the skewed distribution towards the resistance parent for soybean rust severity score suggested dominance over susceptible parent.Rust-infected lines in majority of the F 2 plants of this study developed the type of RB lesion associated with resistance (Table 3).In previous genetic studies of resistance to soybean rust, dominant (Rpp), recessive (rpp), and incompletely dominant resistance genes have been reported in crosses with various sources of resistance (Li et al., 2012;Ray et al., 2011;Chakraborty et al., 2009;Calvo et al., 2008;Garcia et al., 2008;Monteros et al., 2007).
The Chi-squared ( 2 ) test for disease severity scores and lesion type was 0.569 and 2.832; P = 0.451 and  0.092, respectively and suggested a single dominant resistant gene associated with resistance to soybean rust.This was reflected in the  2 value that fitted the Mendelian segregation ratio of 3 (Resistance):1 (Susceptible) (Table 3).However, the marker analysis indicated that UG 5 carries more than one putative soybean rust resistance loci (Table 5 and Figure 2).This difference could be likely due to the smaller size of the F 2 mapping population used in the study and the number of F 2 plants used to assess both phenotypic and genotypic evaluation.This observation calls for further studies with increased number of markers and mapping population to confirm the number of genes associated with resistance to soybean rust in UG 5.
The inclusive composite interval mapping (QTL IciMapping 4.1) showed three QTLs in association with DS on three different LGs (Figure 2).The maps created from this population were in good agreement with the consensus map created by Song et al. (2004) regarding markers" order but differed with regard to the distances between each marker.Probably, the small size of the population used in this study could be the cause for this discrepancy.
The putative QTL with the highest peak (LOD = 8.18) and highest phenotypic variance which accounted for PVE = 25.71% in association with SBR resistance was mapped to the genomic location of Rpp1-b locus (Chakraborty et al., 2009) flanked by the same markers (Sat_064 and Sat_372).This could be the dominant QTL controlling resistance to soybean rust in genotype UG 5.This most probably indicated that UG 5 carries the same allele as PI594538A, the source of the original Rpp1-b, on this locus.Allelism tests, however, will be required to confirm whether this locus is identical with the Rpp1-b gene located on chromosome 18 or not.
The second putative QTL detected in association with SBR resistance (LOD = 7.36 and PVE = 19.15%;Table 5) on chromosome 9 (LG-K; Figure 2) could carry a novel Rpp gene as no other Rpp gene was previously reported on this chromosome.The third putative QTL, with a LOD score of 3.47 and PVE of 18.27%, detected in association with SBR resistance was located on chromosome 6 (LG-C2; Figure 2) where two dominant (Rpp (Hyuuga) and Rpp3) and one recessive (rpp3) genes were previously reported from three different sources of resistance (Ray et al., 2011;Hyten et al., 2009;Monteros et al., 2007).The SSR markers flanking the previously reported genes were included in this study, for which none of them was found to be linked to the current putative QTL suggesting that this putative QTL in UG 5 could be a different allele as compared to the previously reported Rpp genes.To verify this, the relationship between UG 5 and the PIs containing the known Rpp genes will require allelism tests.The high phenotypic variance and negative effects (additive and dominance) of the QTLs indicated their involvement in resistance to SBR.The negative values for the additive and dominance effects of the QTLs (Table 5) were also evidences that both additive and dominance effects are important in the inheritance of resistance to SBR (Bassi et al., 2017).
UG 5 was found to be resistant to different isolates of P. pachyrhizi in different countries (Maphosa et al., 2013;Twizeyimana and Hartman, 2012;Twizeyimana et al., 2009;Kawuki et al., 2003).For instance, the genotype expressed an RB reaction when inoculated with field isolates from Nigeria and Uganda (Hailay et al., 2018;Maphosa et al., 2013;Twizeyimana et al., 2009), whereas, it showed an immune (no visible reaction) for 72 P. pachyrhizi isolates in USA as compared to the other six soybean genotypes with the known resistance genes (Rpp1, Rpp2, Rpp3, Rpp-Hyuuga, Rpp4 and Rpp5 (Twizeyimana and Hartman, 2012).The resistance of UG 5 to diverse isolates of SBR across wider agro-ecologies could, therefore, be due to the presence of more than one SBR resistance gene.

CONCLUSION AND RECOMMENDATION
This research provides evidence for the presence of three putative loci on chromosomes 6, 9 and 18 for soybean rust resistance in genotype UG 5.The QTL on chromosome 9 was novel for which no soybean rust resistance genes were previously reported.The putative QTLs identified in this study will help to facilitate SBR resistance breeding toward a more efficient markerassisted selection approach and gene pyramiding leading to the development of durable resistance.The identified loci on this genotype need to be further screened on larger population size and increased number of markers from each linkage group to precisely locate and identify the putative genes.The structural and functional roles of the putative genes need to be determined.
lesions; RB: reddish brown lesions; DS: disease severity.† Mean disease severity score on a scale of 1 to 5: 1 = no visible lesions, 2 = light infection with few lesions present, 3 = light to moderate infection, 4 = moderate to severe infection, and 5 = prolific lesions.ffiThe number indicates the sum of the number of homozygous RB and segregating lines.

Figure 1 .
Figure 1.Linkage map of 25 SSR markers.C2, M, K, B1, F, J, G and L are the linkage groups formed; the numbers to the left-side of the map are positions of the SSR markers in cM.

Figure 2 .
Figure 2. Genomic regions and SSR markers significantly associated with resistance to SBR detected using QTL IciMapping software in Wondersoya × UG 5.

Table 1 .
Marker names, forward and reverse primer sequences of the polymorphic SSR markers.

Table 2 .
Soybean rust lesion type and disease severity in Wondersoya x UG 5 population and their parents.

Table 3 .
Segregation of F2 population to soybean rust resistance and lesion type in UG 5.

Table 4 .
Chi-squared analysis of 33 polymorphic SSR markers in the F2 population.

Table 5 .
Summary of putative QTLs for the resistance character detected in UG 5. Phenotypic variation explained by each QTL; LG: linkage group; LOD: logarithms of odds; cM: centi-Morgan.ffi Nearest flanking marker. PVE: