Introduction

Stripe rust, caused by Puccinia striiformis f. sp. tritici (Pst), is a serious fungal disease for wheat, especially in hypothermal and moist environments (Chen et al. 2014). To control this disease genetically, numerous Yr (yellow rust) genes have been found (McIntosh et al. 2016). The formally designated Yr genes were up to Yr76 (Dracatos et al. 2016; Xiang et al. 2016). Yr15, Yr24/Yr26, Yr35, Yr36, Yr53, Yr64 and Yr65 were derived from tetraploid wheat, Yr8 from Aegilops. comosa (Niu et al. 2004), Yr9 from Secale cereale (Mago et al. 2005), Yr17 from Ae. ventricosa (Jia et al. 2011), Yr28 and Yr48 from Ae. tauschii (Singh et al. 2000; Lowe et al. 2011), Yr37 from Ae. kotschyi (Heyns et al. 2011), Yr38 from Ae. sharonensis (Marais et al. 2010), Yr40 from Ae. geniculata (Kuraparthy et al. 2009), Yr42 from Ae. neglecta (Marais et al. 2009), Yr50 from Thinopyrum intermedium (Liu et al. 2013), Yr70 from Ae. umbellulata (Bansal et al. 2016), and others mainly from hexaploid landraces (McIntosh et al. 2016).

To date, Yr10, Yr18, Yr36 and Yr46 have been cloned (Liu et al. 2014; Krattinger et al. 2009; Fu et al. 2009; Moore et al. 2015). Yr10 encodes a NBS–LRR protein; Yr18 encodes an ATP-binding cassette (ABC) transporter; Yr36 encodes a wheat Kinase-START (WKS) protein; Yr46 encodes a hexose transporter. Fine mapping of Yr9, Yr15 and Yr26 was also conducted (Mago et al. 2005; Zhang et al. 2013; Abdollahi Mandoulakani et al. 2015). Moreover, a lot of functional genes involved in the wheat-stripe rust responses have been identified, such as TaHLRG, TaMDHAR and TaADF7 (Liu et al. 2008a, b; Feng et al. 2014; Fu et al. 2014). Concerning coevolution of plant and pathogen, the effectiveness of single gene to the resistance is limited and short-term. Gene pyramiding will be an effective way to improve plant durable resistance (Ellis et al. 2014).

When proposing pyramiding strategies, distribution of Yr genes needs to be identified in wheat germplasms and breeding lines. Marker-assisted detection (MAD) is the most commonly used method to identify the presence of Yr genes (Goutam et al. 2015). Since most markers are linkage markers instead of gene markers, the validity of them remains to be assessed. The presence of Yr5, Yr9, Yr10, Yr15, Yr17, Yr18, Yr26 or Yr36 has been identified in a limited number of wheat cultivars and breeding lines (Tabassum et al. 2010; Yuan et al. 2012; Zeng et al. 2014). There are a great number of Pst resistance genes, especially newly reported ones (McIntosh et al. 2016), having not been detected in diverse germplasms and breeding lines. Moreover, the effectiveness of most Yr genes against the newly emerged Pst races, such as PST-V26 (Tian et al. 2016), is not very clear. The function of resistance gene also depends on the genetic background (Ellis et al. 2014). So, it is meaningful to evaluate the effectiveness of Yr genes in diverse germplasms not just in the near-isogenic lines (NILs).

Many researches have indicated that additive effects exist extensively among Yr genes. For example, complex additive interactions were observed by Yang et al. (2013). Yr31 suppressed the additive effect of Yr30 and a 3D locus, but not of Yr18 in Mexican. The 3D and 5BL loci were generally not additive with each other, but were additive when combined with other loci in China. Additive effects were also observed between Yr58 and Yr46. The recombinant inbred lines (RILs) carrying both genes showed a lower IT than those carrying Yr58 or Yr46 individually (Chhetri et al. 2016). Studies also showed that rust resistance can be enhanced by combining all stage resistance (ASR) or seedling resistance (SR) genes with adult plant resistance (APR) or slow rusting resistance (SLR) genes (Chen et al. 2013; Ellis et al. 2014). It is necessary to evaluate the interactions between different Yr genes pyramided together.

The Sichuan Province in China is an important overwintering area for stripe rust races. The prevalent Pst races are PST-CYR32 and PST-CYR33 (Zhou et al. 2014a, b). New virulent races also emerged frequently, such as PST-G22 (Xiang et al. 2013). If stripe rust could not be controlled in the Chengdu Plain, it would put a threat on the wheat production of the middle and lower reaches of the Yangtze River. To control epidemics of stripe rust, the evaluation for the efficacy of Yr genes in wheat germplasms and breeding lines should be timely assayed.

Materials and methods

Plant materials

A total of 672 wheat accessions (Table S1) were collected, including 17 Yr gene NILs of Avocet “S” (AvS), 21 cultivars and 147 breeding lines of “Chuanyu” wheats, 170 landraces and 140 cultivars in China, 148 accessions from all over the world, and 29 synthetic wheats. Among these materials, the NILs were received from the Sichuan Academy of Agricultural Sciences (SAAS), the 200 China core collections and 84 foreign germplasms were obtained from the Chinese Academy of Agricultural Sciences (CAAS, http://www.cgris.net/), and 23 accessions containing specific Yr genes were acquired from USDA-ARS (http://www.ars-grin.gov/).

Field testing

All 672 wheat accessions were evaluated for stripe rust reaction at Shuangliu, Shifang and Jitian in Sichuan Province in 2013, 2014 and 2015. Twenty seeds of each accession were planted in a row, and susceptible wheat strains “Minxian169” and “Chuanyu12” were inserted after every 9 rows. Mixed Pst spores of races PST-CYR32, PST-CYR33, PST-SU11, PST-Hybrid46 and PST-G22 (provided by SAAS), were suspended in 0.05% Tween 20 and were sprayed on wheat seedling leaves at trefoil stage. The infection types (ITs) were recorded at the adult plant stages (twice for each environment at 150 and 164 days after seeding) by a modified method according to the standard classification system from 0 to 4 (McIntosh et al. 1995).

Molecular markers

Two closely linked (usually flanking) markers of each Yr gene were chosen to identify its presence/absence in the wheat accessions, except a few genes for which only one closely linked marker was reported. As shown in Table 1, a total of 77 markers (37 SSR markers, 15 STS markers and 25 EST or gene based markers) for 39 stripe rust resistance genes were employed in this study. The primer sequences were synthesized by GENEWIZ Biotech (China).

Table 1 The information of molecular markers for the Yr genes used in this study
Full size table

DNA extraction

A total of 100 mg fresh leaf tissue was collected from each accession, frozen in liquid nitrogen, and ground to powder with a high-throughput tissue grinder. Then, 2× hexadecyltrimethylammonium bromide (CTAB) extraction buffer containing 1.4 M NaCl, 100 mM Tris–HCl (pH 8.0), 2% CTAB and 20 mM EDTA was added to extract the genomic DNA according to Riede and Anderson (1996). Finally, the DNA was dissolved in 100 μl TE buffer with 10 mM Tris–HCl (pH 8.0), 1 mM EDTA and 20 μg/ml RNase, and was incubated for 1 h at 37 °C before storing at −20 °C.

PCR and electrophoresis

Polymerase chain reaction (PCR) was performed using a Master Cycler Pro PCR System (Eppendorf, Germany). A 10 μl PCR mixture consisted of 100 ng template DNA, 5 μl 2× Es Taq MasterMix (CWBIO Biotech, China), 0.3 μl 10 μM forward primer and 0.3 μl 10 μM reverse primer. Amplifications were programmed with 5 min of denaturation at 94 °C; 35 cycles of 30 s at 94 °C, 30 s at 40–60 °C (depending on the primer pair) and 30 s at 72 °C; and 72 °C for 10 min followed by a 4 °C hold step. PCR products were separated by 1.5% agarose gel, 8% non-denaturing polyacrylamide gel or 6% denaturing polyacrylamide gel, and were stained with ethidium bromide or silver nitrate (An et al. 2009), respectively. For the primer pairs STS-7/8, STS-9/10 and URIC/LN2, the PCR products were digested with DpnII (New England Biolabs, USA) according to Chen et al. (2003) before electrophoresis.

Phenotyping and genotyping

The immune plant with no visible symptoms was scored as IT 0, the highly resistant plant with little necrotic flecks and no sporulation was scored as IT 1, the moderately resistant plant with a few necrotic flecks and trace sporulation was scored as IT 2, the moderately susceptible plant with necrotic blotches and moderate sporulation was scored as IT 3, the highly susceptible plant with chlorotic stripes and abundant sporulation was scored as IT 4. The average ITs from the scores of all environments was used to represent the resistant/susceptible (R/S) phenotype of a wheat accession. Few wheat accessions in some environments were not scored because of their absence, but it did not affect the scores collected from other environments. While analyzing genotyping results, the wheat accession was counted for carrying a specific Yr gene only when both flanking markers were presented, except for a few genes where one closely linked marker has been reported.

Data analysis

All data of phenotypes and genotypes were recorded in Microsoft office excel 2010 for statistical analysis. To compare the effective of each gene in diverse genetic backgrounds, the data was divided into two groups (presence of a gene and absence of a gene). To evaluate the effective of different combinations of two resistance genes, the data was divided into four groups (group one, where both selected genes were present; group two and group three, where only one of the selected genes was present; group four, where none of the two genes was present). One-way analysis of variance (ANOVA) was conducted to evaluate the variance and significance among these groups.

Results

Detection of stripe rust resistance genes in 672 wheat accessions

The accuracy of MAD is affected by the distance between the target gene and the linkage markers. So, closely linked markers are desired. For example, a recent report indicated that YrSP was different from Yr5/Yr7 (Feng et al. 2015). Both of them were located on the 2BL chromosome, and were hard to be distinguished. Seven published markers were screened in CS, Avocet and Yr-NILs of Yr5, Yr7 and YrSP (Fig. 1). S19M93 and STS7/8 can be used to identify Yr5; gwm526 and barc349 can be used to identify Yr7; dp269 and WC5/WC6 can be used to identify YrSP. Moreover, some markers were easy to be distinguished, such as STS (S19M93 for Yr5) and SCAR (SC385 for Yr17), and were helpful for high-throughput detection. By using the NILs or donors of the corresponding gene as the positive control, the detection of 36 Yr genes in 672 wheat accessions was accomplished (Fig. S1).

Fig. 1
figure 1

Electrophoretogram of seven PCR based markers for the Yr5/Yr7–YrSP locus. An 8% non-denaturing polyacrylamide gel was used to separate the PCR products. A molecular weight standard of base pairs was listed in the right. Right arrow indicates the polymorphism bands of each molecular marker among the five wheat accessions

Full size image

Distribution of 36 Pst resistance genes in different taxon groups

Totally, Yr10, Yr17 and Yr18 were identified positively in numerous accessions, while Yr8, Yr36, Yr61 and YrZH84 were detected positively only in a few accessions. As shown in Table 2, Yr5, Yr9, Yr15, Yr17, Yr30, Yr48, Yr65, Yr67, YrSP and TaHLRG showed high frequency in “Chuanyu” breeding lines; Yr10, Yr18, Yr33, Yr51, Yr59 and Yr62 showed high frequency in Chinese landraces; Yr5, Yr9, Yr29, Yr30, Yr39, Yr41, Yr48, Yr49, Yr57 and Yr59 showed high frequency in Chinese modern cultivars; Yr4, Yr5, Yr7, Yr9, Yr29, Yr30, Yr33, Yr41, Yr52, Yr57, Yr60 and YrSP showed high frequency in introduced foreign germplasms. These results revealed broad diversity of resistance genes in different wheat taxon groups. Moreover, the accessions carrying multiple Yr genes identified in this study might be useful as parental lines for diversifying Pst resistance sources in wheat breeding.

Table 2 The distribution percentage of each resistance gene in different wheat panels
Full size table

Stripe rust reactions in field testing

By analyzing the IT scores collected from three places and three years, environmental variation was non-significant (p = 0.87), but significant variances were observed among the average ITs of different wheat accessions from all environments (p < 0.01). The overview for the percentages of each Pst infection phenotype and the average ITs of whole wheat accessions and four taxon groups are shown in Fig. S2. The ratio of resistant to susceptible phenotype in the total wheat accessions is near to 1:1, and the average IT of the total wheat accessions is between 2.5 and 3.

Relationship between the number of pyramided genes and Pst resistance

To obtain a realistic relationship between Pst resistance and the pyramided gene number, the results of MAD and field testing were combined for analysis. As shown in Fig. 2, a significant positive correlation (R2 > 0.8, p < 0.01) between the number of Yr genes and the Pst resistance was observed in the 672 wheat accessions. This high correlation suggested that pyramiding of Yr genes will be an effective way to improve stripe rust resistance in wheat breeding. Moreover, the accessions carrying multiple Yr genes might be useful as intermediate materials for introducing Pst resistance sources into commercial varieties.

Fig. 2
figure 2

The relationship of stripe rust resistance and pyramided gene number. The trend line was added based on the sum percentages of MR and HR, a or the average ITs, b of each pyramided gene number. HS highly susceptible phenotype, MS moderately susceptible phenotype, MR moderately resistant phenotype, HR highly resistant phenotype. Each bar represents the mean value of the stripe rust IT (0, 1, 2, 3, 4) and standard deviation (SD; n ≥ 17) in b

Full size image

Contribution of Yr genes individually to Pst resistance

The contribution of each Yr gene to stripe rust resistance was evaluated in diverse genetic backgrounds. As shown in Fig. 3, the wheat accessions carrying Yr15, Yr17, Yr18, Yr65 and Yr67 had significantly lower ITs than those without the corresponding gene, whilst the situation for Yr29 was opposite. The other genes did not meet a significant level to stripe rust resistance in this study. Moreover, the contribution of eleven genes to Pst resistance in the four taxon groups was shown in Fig. S3. These results revealed some Yr genes contributed effective resistance to the current Pst races. The other genes that did not show significance but still had lower ITs, such as Yr46 and Yr60, might contribute to partial resistance in pyramids, which will be analyzed in the following section.

Fig. 3
figure 3

Contribution of each gene to Pst resistance. Each bar represents the mean value of the stripe rust IT (0, 1, 2, 3, 4) and SD (n ≥ 15). One-way ANOVA was used to determine the significance level between the presence and absence groups of each Yr gene. Significance at *p < 0.05. Significance at **p < 0.01

Full size image

Comparing different combinations of two genes with respect to Pst resistance

A total of 183 combinations of each two genes were suitable for the statistical analysis, and partial results of them were shown in Fig. 4. 30.23% combinations showed a similar average ITs with those carrying either one gene, 27.13% combinations showed a higher average ITs than those either one gene present, and 42.64% combinations showed a lower average ITs than those either one gene present. The results revealed that some combinations of two genes, one conferring significant resistance and another conferring non-significant resistance (Fig. 3), such as Yr17 + Yr26 and Yr9 + Yr18, effectively improved the Pst resistance in the field trails. The results also showed that some combinations of two genes conferring non-significant resistance, such as Yr30 + Yr46, effectively improved the Pst resistance when pyramided together. But in some other combinations, there was a reduction of resistance compared with those carrying either of the two genes, such as Yr48 + Yr67. These results revealed the additive effects or epistatic effects between the resistance genes when pyramided.

Fig. 4
figure 4

Contribution of different combinations of two Yr genes to stripe rust resistance. The difference value (D value) of the ITs between the combination group and the higher or lower group (indicated by ) are shown on the column to represent the resistance improvement (negative D value) or reduction (positive D value). The results of a absolute D value >15% are shown here. Each bar represents the mean value of the stripe rust IT (0, 1, 2, 3, 4) and SD (n ≥ 10). One-way ANOVA was used to determine the significance between the combination group and the higher or lower group of each pair of Yr genes. Significance at *p < 0.05, significance at **p < 0.01

Full size image

Discussion

Although numerous Yr genes have been reported, Yr11, Yr12, Yr13, Yr14, Yr19, Yr20, Yr22, Yr23 and Yr25 have not been mapped, Yr27, Yr31, Yr32, Yr37, Yr38, Yr40, Yr42, Yr45, Yr50, Yr54, Yr56, Yr63, Yr66, Yr70, Yr73 and Yr74 have no appropriate markers, Yr2, Yr3, Yr21, Yr34, Yr43, Yr47, Yr55 and other temporarily named Yr genes have no closely linked markers. So, 36 Yr genes were finally screened and used to evaluate their contribution to the current Pst resistance in this study.

When conducting marker-assisted selection (MAS) in breeding, validity and convenience of a molecular marker should be considered. Here, PCR based markers were employed (Table 1) for identification of the corresponding gene (Fig. S1). There are still some markers used to map a gene but not suitable for detecting the gene in diverse genetic backgrounds, such as Owm45F3R3 for Yr51 (Randhawa et al. 2014). It might be because the polymorphism of a marker showed in the mapping population could not be distinguished in some wheat germplasms. Some RGAP markers also have this problem, such as the RGAP markers for Yr44 (Sui et al. 2009), Yr45 (Li et al. 2011) and Yr59 (Zhou et al. 2014a, b). So, it is hard to apply these markers directly for MAD, even with the donor lines. Moreover, the brightness of the target band of some markers was too weak to be distinguished from polymorphic bands even with a touchdown PCR program, such as stm673acag for Yr1 (Bansal et al. 2009) and gwm508 for Yr35 (Dadkhodaie et al. 2011).

To enhance the reliability of molecular markers, much improvement has been made from polymorphic markers to specific markers, to gene specific markers, and to functional markers. There were many types of marker conversion having been conducted, such as amplified fragment length polymorphism (AFLP) to SCAR (SC-OPD11, Niu et al. 2004), SSR to SCAR (gwm415 to SC-372, Jia et al. 2011), SSR to STM (gwm533 to stm559tgag, Hayden et al. 2004), EST to SSR (bu099658, Hasancebi et al. 2014), DArT to STS (sun104, Randhawa et al. 2014) and RGAP to STS (wgp5467 to STS5467, Zhou et al. 2014a, b). Still, fine mapping of some Yr genes is required for valid detection, such as distinguishing the closely linked loci Yr5/Yr7-YrSP (Fig. 1). Moreover, sequence specific marker could indicate the presence of the cloned Yr10, Yr18 and Yr36 in the genomic DNA accurately (Fig. S1). But, whether it could be used to indicate the functional type still needs to be considered. Because there are many haplotypes existed in the hexaploid wheat, such as the resistant (Lr67res) and susceptible types (Lr67sus) of Yr46 (Moore et al. 2015). So, function specific markers were developed to indicate the functional mutant. For example, the marker THR1 developed in the non-coding region was used to identify the association of TaHLRG and stripe rust resistance (Liu et al. 2008a, b). Two markers cib-Yr28M1 and cib-Yr28M2 developed in the gene coding region were used to distinguish the resistance associated Yr28 (unpublished). To utilize these cloned Pst resistance-associated genes, such as TaMDHAR (Feng et al. 2014) and TaADF7 (Fu et al. 2014), more functional markers are required.

Through MAD, the distribution of 36 resistance genes in 672 wheat accessions was illustrated. The result showed consistence with previous studies. Yr9, Yr10, Yr17 and Yr18 were the mostly identified genes, while Yr8 and Yr36 were identified in a few accessions (Tabassum et al. 2010; Yuan et al. 2012; Zeng et al. 2014). Many “Chuanyu” cultivars and advanced breeding lines were released in Sichuan and other provinces of China. They showed a good adaptability to the local Pst races (Fig. S2). This might be related to the introduction of new Yr genes (such as Yr15, Yr65 and Yr67) and the reduction of previously widely applied Yr genes (such as Yr9, Yr10 and Yr24/Yr26). However, there are two prominent problems. One problem is the over-use of a few Yr genes, such as Yr17 presenting in 74.40% of “Chuanyu” breeding lines (Table 2). This will lead to diversity reduction of resistance genes in breeding population and is unfavourable for breeding durable resistance varieties. The other problem is that the durable genes have not been widely deployed in “Chuanyu” wheat. The Chinese landraces maintaining rust resistance chronically (Fig. S2) showed a high frequency of Yr18 to 59.41% (Table 2). Many reports revealed the importance of durable resistance genes in combining with other Yr genes (Krattinger et al. 2009; Yang et al. 2011). The enhancement effect was also confirmed in this study, such as Yr9 + Yr18 (Fig. 4). Therefore, through MAD, the overuse of a few Yr genes can be avoided, and durable genes with low frequency can be introduced intentionally. Additionally, the newly reported heterogenous genes, such as Yr37 (Heyns et al. 2011), Yr40 (Kuraparthy et al. 2009) and Yr50 (Liu et al. 2013), should also be applied properly. To perform resistance breeding for a long period, extensive resistance resources are required to broaden the genetic basis of breeding materials.

As prevalent Pst races were variant with environments, it is necessary to understand the effective of Yr genes under a specific condition. Here, a case study was performed in Sichuan province, and we hope that it would attract enough attention to the breeders and researchers in the world-wide. Some Yr genes identified in this study showed significant effects under current conditions, such as Yr15 and Yr65, while some once massively used Yr genes did not confer significant resistance, such as Yr9 (Fig. 3). These results were basically consistent with the rust testing results of some available NILs. Only the NILs of Yr5, Yr15 and Yr18 showed resistance, others (such as the NILs of Yr9 and Yr26) were susceptible to Pst in this study. But variances existed in the results of Yr17. Yr17 contributed significantly to the current Pst resistance in this study (Fig. 3), but Yr17-NIL was susceptible. Further analysis showed that the contribution significance of Yr17 varied with different taxon groups, non-significant in “Chuanyu” breeding lines but significant in Chinese modern cultivars and introduced foreign germplasms (Fig. S3). Other genes, such as Yr67, also showed similar results. The function of one gene depends on the genetic background, so it is meaningful to evaluate the contribution of Yr genes to stripe rust resistance in diverse genetic backgrounds rather than in their NILs. MAD combined with rust testing results could be used to select the most effective genes against current Pst races when breeding high resistant varieties.

The genes conferring significant resistance will be applied widely in breeding, such as Yr15, Yr17 and Yr65 (Table 2; Fig. 3). But other genes showing non-significant effectiveness under current conditions still need to be utilized in gene pyramiding. Because the results identified in this study (Fig. 2) and the presence of multi-QTLs in resistant wheat (Lowe et al. 2011; Rosewarne et al. 2013; Yang et al. 2013) suggested that gene pyramiding could improve the durability of rust resistance. Additive effects and epistatic effects exist extensively in Yr gene pyramids (Fig. 4). And 42.64% of the combinations of two Yr genes improved Pst resistance than those either one present. Although showing non-significance, these combinations with substantially lower ITs (negative D value, Fig. 4) than those carrying either one gene still need to be considered. Moreover, some durable resistance genes conferring partial resistance to Pst showed enhancing effect to race-specific genes in this study, such as Yr17, Yr18, Yr30 and Yr46 (Fig. 4). Taking advantage of the positive interactions and avoiding the negative interactions should be carefully considered in resistance gene pyramiding.