Characterization of a New Pm2 Allele Conferring Powdery Mildew Resistance in the Wheat Germplasm Line FG-1

Powdery mildew has a negative impact on wheat production. Novel host resistance increases the diversity of resistance genes and helps to control the disease. In this study, wheat line FG-1 imported from France showed a high level of powdery mildew resistance at both the seedling and adult stages. An F2 population and F2:3 families from the cross FG-1 × Mingxian 169 both fit Mendelian ratios for a single dominant resistance gene when tested against multiple avirulent Blumeria tritici f. sp. tritici (Bgt) races. This gene was temporarily designated PmFG. PmFG was mapped on the multi-allelic Pm2 locus of chromosome 5DS using seven SSR, 10 single nucleotide polymorphism (SNP)-derived and two SCAR markers with the flanking markers Xbwm21/Xcfd81/Xscar112 (distal) and Xbwm25 (proximal) at 0.3 and 0.5 cM being the closest. Marker SCAR203 co-segregated with PmFG. Allelism tests between PmFG and documented Pm2 alleles confirmed that PmFG was allelic with Pm2. Line FG-1 produced a significantly different reaction pattern compared to other lines with genes at or near Pm2 when tested against 49 Bgt isolates. The PmFG-linked marker alleles detected by the SNP-derived markers revealed significant variation between FG-1 and other lines with genes at or near Pm2. It was concluded that PmFG is a new allele at the Pm2 locus. Data from seven closely linked markers tested on 31 wheat cultivars indicated opportunities for marker-assisted pyramiding of this gene with other genes for powdery mildew resistance and additional traits.


INTRODUCTION
Powdery mildew of wheat, caused by Blumeria graminis f. sp. tritici (Bgt), is a foliar disease that occurs worldwide, especially in wheat-producing regions with maritime or semi-continental climates (Cowger et al., 2012). In China, wheat yields and quality have been affected by this disease since the 1970s, especially high-yielding cultivars grown with high-inputs of irrigation and fertilization (Bennett, 1984;Sun et al., 2015). Over the past decade, the area affected by powdery mildew in China has ranged from 6 to 8 million hectares each year, resulting in estimated grain losses of 300,000 metric tons 1 .
Although, fungicides can reduce losses in yield and quality caused by powdery mildew, resistant cultivars are the preferred means for control (Horst, 2013;Wang et al., 2015b). More than 70 formally named and about 20 temporarily designated powdery mildew (Pm) resistance genes/alleles have been identified (McIntosh et al., 2013;Hao et al., 2015). However, due to the racespecific nature of resistance and excessive deployment of single resistance genes, the effectiveness of Pm genes is often shortlived as they are defeated by virulent mutants of the pathogen (Hsam and Zeller, 2002;Xiao et al., 2013). Detailed studies have indicated that most current wheat cultivars grown in China have non-effective or no Pm genes . It is therefore urgent to identify effective sources of resistance among germplasm from around the world to increase the existing genetic diversity.
Some of the documented Pm loci have multiple resistance alleles, such as Pm1 Singrün et al., 2003), Pm2 (Ma et al., 2015a;Xu et al., 2015), Pm3 , Pm4 (Schmolke et al., 2012), Pm5 (Hsam et al., 2001;Huang et al., 2003), and Pm24 (Huang et al., 2000;Xue et al., 2012). Although, some alleles at these loci may have lost their effectiveness, further allelic variation may be present in other germplasm. The gene Pm2 was identified several decades ago (Pugsley and Carter, 1953), and was used as an effective resistance source in some countries . Although, the avirulence frequency remains low in some regions of China and other parts of the world after decades of deployment, several new alleles (e.g., Pm2b, Pm2c, PmLX66, and PmW14) were identified, thereby increasing the diversity of available resistance genes (Ma et al., 2015a;Sun et al., 2015;Xu et al., 2015).
Molecular markers are powerful tools for tagging resistance genes. Almost all the designated Pm genes have been mapped to specific chromosomal loci (McIntosh et al., 2013). Microsatellites or simple sequence repeats (SSRs) provide a simple and effective marker system for molecular mapping in wheat (Somers et al., 2004;Sourdille et al., 2004;Xue et al., 2008). To develop high-density marker assays, high-throughput single nucleotide polymorphism (SNP) genotyping platforms based on wheat 9K, 90K, and even 660K SNP chips are now available (Bérard et al., 2009;Lai et al., 2012;Avni et al., 2014;Wang et al., 2014), and these will greatly increase the numbers of markers closely linked to targeted resistance genes.
Molecular markers closely linked to targeted genes controlling valuable traits can be used to rapidly transfer them to other cultivars. Hence, marker-assisted selection (MAS) has been practiced in many parts of the world (USA, Australia, Canada, India, and Europe) to complement conventional breeding programs (Gupta et al., 2010). A number of markers associated with documented QTL/genes for some major economic traits, such as disease resistance, grain protein content and pre-harvest sprouting tolerance, have also been used for MAS in wheat breeding programs (e.g., de Bustos et al., 2001;Davies et al., 2006;Nocente et al., 2007;Badea et al., 2008;Zhang et al., 2009). 1 http: //cb.natesc.gov.cn In this study, the wheat germplasm line FG-1 imported from France showed a high level of powdery mildew resistance in China. To make better use of this resistance resource, the following research was carried out to: (1) determine the inheritance of powdery mildew resistance in FG-1 using an array of Bgt isolates, (2) determine the chromosomal location of the resistance gene using different kinds of molecular markers and allelism tests, (3) compare response spectra of FG-1 and lines carrying documented Pm genes using Bgt isolates, (4) to compare allelic variation between FG-1 and genotypes with documented Pm genes using SNP-derived markers, and (5) investigate the applicability of closely linked markers for MAS.

Plant Materials
FG-1 is a common wheat line that was imported from France and maintained in the germplasm bank of Shijiazhuang Academy of Agricultural and Forestry Sciences (Shijiazhuang, Hebei Province, China) of unknown pedigree. It has been grown in Northern China since it was imported from France. It has resistance to powdery mildew at both the seedling and adult growth stages based on observations over many years. The susceptible Chinese cultivar (cv.) Mingxian 169 was crossed with FG-1 to study the inheritance of powdery mildew resistance. Mingxian 169 and Huixianhong were used as susceptible controls for test of powdery mildew resistance. Several stocks with documented Pm genes on chromosome arm 5DS, such as Ulka/8*Cc with Pm2a, KM2939 with Pm2b, Niaomai with Pm2c, Tabasco with Pm48, Liangxing 66 with PmLX66, Wennong 14 with PmW14, YingBo700 with PmYB, Zhongmai 155 with PmZ155, X3986-2 with PmX3986-2, Wanfengjian 34 with PmWFJ, PB3558 with PmPB3558, Brock with MlBrock and D57-5D with PmD57-5D were used in multi-race response comparisons with FG-1. Thirty-one wheat cultivars representing Chinese elite germplasm were tested using molecular markers closely linked to the Pm gene in FG-1 to validate their applicability for MAS.

Phenotyping Reactions to Powdery Mildew
Forty-nine Bgt isolates with different avirulence/virulence arrays (races) collected from different regions of China (Supplementary  Table S1) were used to inoculate FG-1 and various host lines to determine the breadth of effectiveness of the resistance in FG-1 and compare the response spectrum of FG-1 to wheat stocks with documented Pm genes. Five seedlings per plot were planted in rectangular trays with 128 cells (3 cm × 3 cm) in a growth chamber. The susceptible check Mingxian 169 and Huixianhong were randomly planted in the trays. Three replications were included in each test. When the seedlings reached the one-to twoleaf stage, fresh conidiospores from Mingxian 169 seedlings were dusted on the trays. The trays were then placed in a greenhouse with a daily cycle of 14 h light at 22 ± 2 • C and 10 h of darkness at 18 ± 2 • C. The inoculation operations were performed once a day for three consecutive days. When the disease was fully developed on Mingxian 169, the infection types (ITs) on each plant were assessed on a 0-4 scale based on the IT scale described by Si et al. (1992), with ITs 0, 0, 1, and 2 being regarded as resistant, and ITs 3 and 4 as susceptible. All the phenotypic experiments were repeated three times to confirm their reactions to each Bgt isolate.
The adult plant reactions to powdery mildew of FG-1 and the susceptible controls Mingxian 169 and Huixianhong were evaluated under field conditions using a mixture of the Bgt isolates prevalent in northern China. The tests with the mixture of the isolates were conducted using the procedures described by An et al. (2013) at Luancheng Agro-Ecological Experimental Station, Chinese Academy of Sciences, Shijiazhuang, China. Disease reactions were assessed on a 0-9 scale, where 0-4 was considered resistant and 5-9 susceptible . The adult plant reactions test was repeated over 4 years' growing season using the same procedure.
Bgt race B03 was chosen to inoculate seedlings of the segregating materials and parents for genetic analysis. Twentyfour plants of each F 2:3 family were inoculated with this race. To confirm that the same resistance gene conferred the resistance to all avirulent Bgt isolates, random sets of 10 homozygous resistant and 10 segregating F 2:3 families to the isolate B03 were tested with all of the 42 Bgt isolates that were avirulent on FG-1. Twenty-four plants of each F 2:3 family were inoculated with each tested race. The number of resistant and susceptible plants was counted to confirm the phenotype of each F 2:3 family. Tests on F 2:3 families with intermediate ITs 2 and 3 were repeated to confirm the results of the previous tests for each Bgt isolate.

Molecular Genotyping of the Mapping Population
Genomic DNA of FG-1, Mingxian 169 and the F 2 plants were isolated using the phenol/chloroform method (Sharp et al., 1988) following evaluation of their powdery mildew reactions. Resistant and susceptible DNA bulks were produced using equal amounts of DNA from 10 homozygous resistant and 10 homozygous susceptible F 2 plants, respectively, following progeny testing. To screen for molecular markers potentially linked to the Pm gene in FG-1, the parents FG-1 and Mingxian 169 and resistant and susceptible bulks were assessed for polymorphisms, initially using 50 SSR markers linked to 35 documented Pm genes. To saturate the linkage map, different types of molecular markers were screened further. They included about 100 SSR markers on chromosomes arms 4DL, 5DS, and 7BS from Somers et al. (2004), 25 SNP-derived markers located on chromosome 5DS developed from the wheat 90K SNP array and its sequence alignment with the Aegilops tauschii draft genome sequence (Lu et al., 2015), 25 EST-derived markers located on chromosome arm 5DS and two sequence characterized amplified region (SCAR) markers linked to documented Pm2 alleles Huang et al., 2012). Polymorphic markers were then genotyped on the F 2 population of FG-1 × Mingxian 169. After confirming the response genotypes through progeny testing of F 2:3 families, a linkage map for the resistance gene in FG-1 was then produced using the software MAPMAKER 3.0 with a LOD threshold of 3.0 and a maximum map distance 37.5 cM (Lincoln et al., 1993).
Kosambi function was carried out to calculate the map distances between linked markers and the targeted gene using a total of 214 F 2:3 families from the cross FG-1/Mingxian 169 (Kosambi, 1943). Chi-squared tests were performed to determine the goodness of fit of observed and expected segregation ratios of the F 2 and F 2:3 populations for markers and powdery mildew responses.

Allelism Test
After the Pm gene(s) in FG-1 was mapped on the short arm of chromosome 5D, FG-1 was crossed with the resistant stocks with documented Pm genes on the same chromosome arm to obtain F 2 populations. The Bgt isolate B03, which was avirulent to FG-1 and all the documented resistant stocks, was selected to inoculate the F 2 populations. The susceptible check Mingxian 169 and Huixianhong were randomly planted in the trays. The number of resistant and susceptible plants in each plot was counted after evaluating their phenotyping reactions. Then, the allelic relationships between the Pm gene(s) in FG-1 and documented Pm genes on the same chromosome arm were confirmed based on the ratio of resistant and susceptible F 2 plants.

Allelic Variation of Linked Marker Alleles
To compare the Pm gene(s) in FG-1 and documented allelic and closely linked Pm genes, several markers that were linked with PmFG were selected to amplify FG-1 and resistant stocks with Pm genes at or near Pm2 locus. The allelic sizes of those markers were assessed by the SensiAnsys gel imaging analysis system (Shanghai Peiqing Science & Technology Ltd., Shanghai City, China).

Validation of the Closely Linked Markers in Different Genetic Backgrounds
To evaluate the potential of the Pm gene(s) in FG-1 for MAS, several closely linked markers were assayed in 31 Chinese wheat cultivars. The patterns or sizes of the polymorphic bands amplified from these cultivars were compared with those amplified from FG-1 to assess the usefulness of the markers in MAS. If the polymorphic band(s) of a marker were all same for FG-1 and the wheat cultivars, it could not be used for MAS. However, patterns or sizes of the polymorphic bands amplified from FG-1 that differed from those in the wheat cultivars indicate the marker can be used to detect PmFG when it was transferred into those cultivars by hybridization.

Evaluation and Inheritance of Powdery Mildew Resistance in FG-1
FG-1 showed resistance to a mixture of Bgt races in the field over 4 years with disease reaction types 0-2 while adjacent controls Huixianhong and Mingxian 169 were susceptible with disease reaction types 8-9. At the seedling stage, FG-1 was resistant to 43 of the 49 Bgt isolates collected from different regions, indicating an avirulence frequency of 87.8% (Figure 1; Supplementary Table  S1). Compared with several wheat cultivars currently deployed in different regions of China, FG-1 possessed a broader resistance spectrum, and more significantly, it was resistant to several highly virulent Bgt isolates; for example, B29, B38, B50 and B80, which are virulent on several or all of the five cultivars Liangxing 66, Liangxing 99, Wennong 14, Zhongmai 155, and Jimai 22 widely planted in Shandong, Hebei, and Henan provinces and Beijing area. Therefore, FG-1 could serve as a valuable resistance donor to add to the current diversity of Pm genes in different wheat production regions.
When tested with Bgt isolate B03, FG-1 was resistant with an IT 0, while the cultivar Mingxian 169 was susceptible with When tested against 42 other Bgt isolates that were avirulent to FG-1, all the 10 homozygous resistant F 2:3 families for B03 were also homozygous resistant to all avirulent isolates and all the 10 segregating F 2:3 families for B03 again segregated, including isolates B29 and B38 with IT 2 on FG-1. Therefore, PmFG conferred powdery mildew resistance to all the avirulent Bgt isolates.

Molecular Mapping of PmFG
The Pm2-linked SSR marker Cfd81 was polymorphic for the parents and the bulks. Cfd81 was genotyped on the the F 2:3 mapping population (Figure 2). Its linkage with PmFG was estimated at 0.3 cM (Figure 3). Because Cfd81 also can be amplified at loci on chromosome arms 4DL, 5DS and 7BS, 79 SSR markers located on these chromosome arms (Somers et al., 2004) were screened to confirm the location of locus PmFG and to increase the density of the linked markers. SSR markers Cfd40, Cfd78, Gwm159, Wmc608, and Wmc805 on chromosome arms 5DS from the map of Somers et al. (2004) were shown polymorphic and linked to Pm FG (Table 2; Figure 3). Markers reportedly located on chromosome arms 4DL and 7BS (Somers et al., 2004) showed no polymorphisms between the parents and the bulks, indicating that PmFG was not located on those chromosome arms. To saturate the linkage map of the PmFG region, 10 SNP-derived markers on chromosome arm 5DS and the SCAR marker SCAR112 which is linked to a documented Pm2 allele were shown polymorphic and linked to PmFG at genetic distances ranging from 0.3 to 16.9 cM (Table 2; Figure 3). SCAR marker SCAR203 co-segregated with PmFG (Figure 3). The reported locations of some of the SNP markers and the SCAR markers were in the deletion bin 5DS1-0-0.63 (Sourdille et al., 2004;Li et al., 2009;Lu et al., 2015), therefore, PmFG should also be located in this chromosome bin.

Allelism between PmFG and Previously Documented Pm2 Alleles
Because PmFG was mapped on the chromosome arm 5DS, FG-1 was crossed with several powdery mildew resistant stocks with documented Pm genes on the same chromosome arm to determine the allelic relationship between PmFG and previously documented Pm genes. These included Ulka/8 * Cc with Pm2a, Liangxing 66 with PmLX66 and TD114 with Pm2a+Pm6. Bgt isolate B03 which is avirulent on FG-1 and the other lines was used to inoculate the large F 2 populations. The phenotyping reactions of those  F 2 populations are shown in Table 3. No susceptible plants were recovered from the F 2 populations of FG-1 and the three documented resistant stocks with Pm2 alleles including reciprocal crosses. Therefore, PmFG seems to be an allele of the Pm2 locus.

Multi-race Comparisons of FG-1 and Lines with Reported Resistance Alleles at or Near the Pm2 Locus
FG-1 showed a clearly different response spectrum to the Bgt isolates from those of other lines with possible Pm2 alleles and Pm48 (Supplementary Table S1). Compared to the lines with Pm2a, Pm2b, Pm2c, PmLX66, PmW14, PmZ155, PmX3986-2, PmWFJ, MlBrock, and PmD57-5D and Pm48, there were differences in response for 6, 5, 8, 5, 6, 7, 8, 12, 5, 4, 6, and 7 Bgt isolates, respectively (Supplementary Table S1; Figure 1). FG-1 also showed a different response pattern from the putative Agropyron cristatum derivative PB3558 that carried PmPB3558 in the Pm2 chromosome region for the response to Bgt isolate B07 and B30 (Supplementary Table S1). Therefore, PmFG was different from the other Pm genes located at or near the Pm2 locus.  Figure 4). For example, the two Bwm21 amplicons in FG-1 were different from those in Ulka/8*Cc (Pm2a), Tobasco (Pm48) and Brock (MlBrock); and the 194 bp Bwm13 amplicon was unique to FG-1 (Table 4). Therefore, FG-1 showed a diversity of alleles of the PmFG-linked markers compared to other resistant stocks with Pm genes at or near the Pm2 locus. From our results of allelism tests, allelic variation of the linked markers and response spectrum analysis, we find that PmFG is a new allele of Pm2.

Potential of Closely Linked Markers for MAS
To investigate the usefulness of the markers linked to PmFG in MAS, eight closely linked markers flanking PmFG were assayed on 31 Chinese elite cultivars (Table 5; Figure 5). The Bwm20, Bwm21, Bwm25, Cfd81, SCAR112, and SCAR203 alleles in all tested cultivars, except Jimai 22, were different from those in FG-1, demonstrating that the six markers could be used in MAS if PmFG was transferred to these cultivars by conventional hybridization. Other markers such as Bwm13 and Bwm16 also could be used, but only in certain cross combinations.

DISCUSSION
FG-1 was a common wheat line introduced from France and provided by the germplasm bank of Shijiazhuang Academy of Agricultural and Forestry Sciences. It was resistant to many Bgt isolates originating from different wheat-producing regions in  '+'amplified marker allele differs from that in FG-1 implying that the marker might be useful for MAS, and '−'marker is similar to that in FG-1. Table S1). Compared with the currently deployed cultivars in China, FG-1 is resistant to several highly virulent Bgt isolates that defeated several of the popular wheat cultivars that are currently grown in China. This suggests that FG-1 is a valuable resistant germplasm, which could be used to complement the resistant genes currently deployed in cultivars in China. However, FG-1 was not resistant to all the races present in certain regions of China. Therefore, the gene in FG-1 needs to be combined with other effective resistance genes to increase the durability of resistance. Genetic analysis demonstrated that a single dominant gene, designated PmFG, conferred resistance to powdery mildew in FG-1 at the seedling stage. Unlike in previous studies (e.g., Gao et al., 2012;Huang et al., 2012;Xue et al., 2012;Xiao et al., 2013;Lu et al., 2015), PmFG was investigated for its resistance to all the avirulent races tested in this study. Based on this information, the single dominant gene in FG-1 was more thoroughly shown to confer powdery mildew resistance to all the avirulent races. Using molecular markers, PmFG was mapped in the Pm2 region on the short arm of chromosome 5D. Its allelic relationship with Pm2 was confirmed by allelism tests.
In this study, PmFG was distinguished from these documented genes by its response spectrum and the allelic variation of the linked marker alleles of the SNP-derived markers. Previous studies indicated that the documented Pm2 alleles shared several similar markers, and the allelic variation of the linked markers could not be detected. However, the advent of the next generation in sequencing technologies significantly reduces sequencing costs, making SNP markers increasingly important due to their abundance in the genome and their very simple genetic mode of inheritance (bi-allelic). Therefore, to further distinguish PmFG from the Pm genes at or near the Pm2 locus, SNP-derived markers of PmFG were screened to distinguish the allelic sizes. For the first time, Pm genes at or near Pm2 locus were studied for their allelic variation using SNP-derived markers. This will contribute to differentiate the variation in the Pm2 locus.
Multiple allelism in disease resistance genes is not uncommon. In the case of powdery mildew resistance in wheat, multiple resistance alleles have been identified at Pm1, Pm2, Pm3, Pm4, Pm5, and Pm24 (McIntosh et al., 2013). So far, 17 functional alleles have been identified at the Pm3 locus, making it one of the largest allelic series of plant resistance genes (Bhullar et al., 2010). For other wheat disease resistance, multi-allelic loci have also been identified. For example, using physical mapping, mutation and complementation, the stem rust resistance gene Sr50 locus revealed extensive diversity, and holds promise for the mining other effective resistance alleles (Mago et al., 2015). These types of genetic diversity may contribute to the genetic improvement of crops and detection of variation in the pathogen and host-pathogen interactions (Prada, 2009;Wicker et al., 2013). In this study, the gene PmFG was identified as a new allele located at the Pm2 allelic cluster. Like the Pm3 allele cluster, more and more Pm2 alleles with different response spectra to Bgt isolates and allelic variation have been identified, increasing the diversity at this locus. However, to further distinguish these alleles, more research should be undertaken in the future, such as fine mapping of all the Pm2 alleles, re-sequencing of the Pm2 region, and even development of functional markers based on the cloning of functional genes of these alleles.
In order to transfer PmFG into the susceptible cultivars or to pyramid multiple R-genes effectively, MAS should be a high-priority in wheat breeding programs. In the previous studies, the SSR marker Cfd81 was shown to be an effective marker for differentiating several Pm2 alleles in MAS (Ma et al., 2015a,b,c). However, Cfd81 serves as only a one-sided marker of Pm2 alleles, and other markers like SCAR112 and SCAR203 also have limited roles because of their dominant characters, which do not allow homozygous and heterozygous genotypes to be distinguished (Ma et al., 2015a,b,c). More closely linked markers need to be screened to increase the density of applicable molecular markers for breeding. SNP markers are based on the variation of a specific nucleotide at a given sequence position between individuals, and therefore their numbers in the wheat genome should be much higher and their detection can be facilitated by cost-efficient based on chips or other array techniques (Colasuonno et al., 2014;Wang et al., 2014Wang et al., , 2015a. In this study, five co-dominant SNP markers were evaluated for their applicability in MAS. The SNP-derived markers Bwm20, Bwm21, and Bwm25 flanked PmFG at genetic distances of only 0.3, 0.3, and 0.5 cM respectively, and they were diagnostic in 96.8% of the tested cultivars in this study. Therefore, these SNP-markers can be used effectively in wheat breeding in the future. One interesting observation was that the marker alleles of FG-1 by the PmFG-linked markers were all same as those of the cultivar Jimai 22, although PmFG is located on a significantly different genetic locus from that of the powdery mildew resistance gene in Jimai 22, which was mapped on chromosome arm 2BL (Yin et al., 2009). Therefore, PmFG cannot be distinguished in Jimai 22 genomic backgrounds by the PmFGlinked markers. This may ascribe to the insufficient marker density at this locus. Future fining mapping and characterization of the haplotype of this locus may contribute to clarify this issue.

AUTHOR CONTRIBUTIONS
PM: experimental implementation, data analysis, and manuscript preparation. HX: production of the mapping population and the genetic map. LL: data analysis. HZ: experimental implementation. GH: MAS analysis. YX: production of the genetic map. XZ: germplasm creation and field investigation. DA: study concept and design.

ACKNOWLEDGMENT
We are grateful to Dr. R. A. McIntosh, University of Sydney, Australia for critically reviewing and revising the drafts of this paper.