Identification and fine‐mapping of RppCML496, a major QTL for resistance to Puccinia polysora in maize

Abstract

Southern corn rust (SCR), which is caused by the fungal pathogen Puccinia polysora Underw, is a prevalent foliar disease in maize. Breeding for resistant cultivars is a desirable way for the efficient control of this disease. To identify quantitative trait loci (QTL) for conferring resistance to SCR, a recombinant inbred population including 138 lines (RILs) derived from the SCR‐resistant line CML496 and susceptible line Lx9801 was evaluated for phenotypic reaction to SCR in three trials in two locations over 2 years. The population was genotyped with the maize 9.4K SNP Genotyping Array marker platform. A total of 3 QTL were mapped on chromosomes 6, 9 and 10, respectively. One major QTL on chromosome 10 (bin 10.00/10.01), RppCML496, was consistently detected across environments, which explained 43–78% of the total phenotypic variation. Using a fine mapping strategy, we delimited RppCML496 to an interval of 128 Kb. Genome mining of this region suggests two candidate genes, and a NBS‐LRR gene is the promising one for RppCML496 against SCR. The tightly linked molecular markers developed in this study can be used for molecular breeding of resistance to SCR in maize.

Abbreviations
• CIM
• composite interval mapping
• Kb
• kilobyte
• LOD
• logarithm of odds
• LRR
• leucine-rich repeat
• NBS
• nucleotide-binding site
• QTL
• quantitative trait loci
• RFLP
• restriction length fragment polymorphism
• RIL
• recombinant inbred line
• SCR
• southern corn rust.

INTRODUCTION

Maize (Zea Mays L.) is one of the most important crops used for food and feed. Its productivity is frequently affected by diseases amongst which southern corn rust (SCR) caused by Puccinia polysora Underw is a major one that threatens maize production especially in tropical and subtropical areas (Raid, Pennypacker, & Stevenson, 1988). However, SCR has gradually spread to temperate areas in recent years (Ramirez-Cabral, Kumar, & Shabani, 2017). The disease causes yield losses ranging from 10% to 30% each year and may approach up to 50% of the anticipated yield under epidemic conditions (Raid et al., 1988; Rhind, Waterston, & Deighton, 1952). The development and deployment of resistance varieties to SCR is the most cost-effective way to disease management, which rely on the identification of effective sources of resistance genes (R genes) and their precise mapping in maize genome for effective manipulation of these genes in breeding programs.

Resistance to SCR is mediated by major and minor R genes. Searching for effective R genes in many studies has allowed to identify R genes in tropic and sub-tropic germplasm, in which some of them have been used in maize breeding programs. Genetic investigations on SCR started in the 1950s when two major genes (Rpp1 and Rpp2) were identified from Colombia and Mexican lines, respectively (Storey & Howland, 1957). The following genetic studies led to identify Rpp9 on chromosome 10S, a dominant gene that conferred immunity to physiologic Race 9 in PI186208, a South African yellow flint (Ullstrup, 1965). Rpp9 has been used successfully in northern US to provide effective resistance to the prevalent race P. polysora PP.9 for more than two decades (Brewbaker et al., 2011). The first attempt to map the R gene was made by Holland, Uhr, Jeffers, and Goodman (1998), who mapped one major QTL on chromosome 10 using Restriction Length Fragment Polymorphism (RFLP) analysis of two F_2:3 populations. With the development of molecular markers, more and more major R genes have been mapped on maize chromosomes, which included RppQ, RppD, RppP25 and RppS from Chinese germplasm (Wu, Na, Zhao, Yan, & Wang, 2015; Zhang, Xu, Zhang, Dai, & Wang, 2010; Zhao et al., 2013; Zhou et al., 2007), and major QTL from US germplasm (Holland et al., 1998; Jines et al., 2007). In recent years, QTL with partial resistance to SCR have also been reported by linkage or association mapping (Camacho et al., 2019; Wanlayaporn, Authrapun, Vanavichit, & Tragoonrung, 2013; Zhou et al., 2018; Deng et al., 2019). Although R genes have been identified and mapped on all maize chromosomes except chromosome 7, chromosome 10 contains the largest number of R genes, and molecular genetic analyses have revealed that all the major R genes are distributed on the short arm of chromosome 10 into a cluster consisting of allelic or tightly linked genes.

Core Ideas

• Southern corn rust is a prevalent foliar disease in maize.
• One major QTL, RppCML496, was detected on the short arm of chromosome 10.
• Fine-mapping resolved a 128 Kb interval containing two candidate genes.

Deployment of major resistance genes to SCR has long been adopted in maize breeding programs. However, due to the rapid evolution of the pathogen, Rpp9 was overcome by a new race of P. polysora in southern US (Brewbaker et al., 2011). Gene RppQ is the more recent example to gradually lose the stable resistance to SCR in China, where it has been effectively deployed for over two decades (Wang et al., 2019). Therefore, it is necessary to continuously discover new resistance genes to protect maize from SCR. In the present study, the QTL RppCML496 with a major resistance effect on SCR was identified and fine-mapped to a 128 Kb region, and is a desirable source of SCR resistance in maize breeding programs.

MATERIALS AND METHODS Population formation

A genetic mapping population comprising 138 BC₂RILs was developed by single-seed descent from the cross CML496 × Lx9801 using Lx9801 as recurrent parent. The resistant parental line CML496 is an elite tropical inbred line from CIMMYT, and susceptible parent line Lx9801 is a temperate inbred line widely used in maize breeding programs in China (Sun et al., 2014).

Field conditions and disease phenotyping

Phenotypic disease evaluations were conducted at two locations over two years: Ledong station in Hainan province (18°45′5.38″ N, 109°10′10.22″ E) in 2015 and 2017, and Changge station (34°12′54.36″ N, 113°46′21.99″ E) in 2017. All of the RILs were grown in single 3-m rows, spaced 0.67 m apart in a randomized incomplete block design with two replications in the field. The standard agricultural practices were followed throughout the growing season. For the nature of airborne conidiospores like P. polysora, natural infection has been confirmed as an effective way to ensure an efficient spread of the pathogen to neighboring plants in fields (Zhao et al., 2013). Disease scores were recorded at 30 days post-pollination. The infection types were visually divided into 9 categories from 1 (highly resistant) to 9 (highly susceptible) using the Stakman infection type scale (Stakman, Stewart, & Loegering, 1962).

Statistical analysis

Analysis of variance (ANOVA) was conducted and means were separated by pairwise t tests (α = 0.05) using type III analysis in the Proc Mixed function of SAS (SAS Institute, 2017). The model for variance analysis was Y = μ + G_i + E_y + GE_iy + R_yr + ε_iyr, where G_i is the random effect of the ith family, E_y is the random effect of the yth environment, GE_iy is the family × environment interaction, R_yr is the environment × replication interaction, and ε_iyr is the residual. The broad-sense heritability (h²) was estimated using the formula: h² = σ_g²/(σ_g² + σ_ge²/n + σ_e²/nr) (Knapp, Stroup, & Ross, 1985), where σ_g² is the genotypic variance, σ_ge² is the genotype × environment interaction, σ_e² is the error variances, n is the number of environments, and r is the replications in each environment. All the variances were acquired from variance analysis.

Linkage map construction and QTL detection

Genotypic data was derived from the MaizeSNP9.4K BeadChip, which including 9,433 SNPs evenly distributed across the B73 reference genome (http://www.cgmb.com.cn/index.php/Home/Index). Polymorphic SNPs in RILs were selected to construct the genetic map. Heterozygous SNPs were treated as missing data, and SNPs for each line with > 20% missing data were removed. Markers with a severe segregation distortion (P ≤ .001) were also excluded from the analysis. Recombination frequencies within linkage groups were calculated using the nnTwoOpt mapping algorithm of the IciMapping software (Meng, Li, Zhang, & Wang, 2015), and linkage groups were determined using the LOD threshold of 3.0. Recombination frequencies were converted to centimorgans (cM) by the Kosambi mapping function (Kosambi, 1944).

QTL mapping was conducted using Windows QTL Cartographer version 2.5 (Wang, Basten, & Zeng, 2012). The composite interval mapping (CIM) (Zeng, 1994) method was applied for QTL mapping with a window size of 10 cM. A significant threshold for declaring a putative QTL was obtained from 1,000 permutations at a genome-wide 0.05 level for each data set and a LOD cut-off score of 3.5 in all environments.

Fine mapping of RppCML496

Based on the RppCML496 region mapped by Windows QTL Cartographer, a F₂ population from the cross between CML496 and Lx9801 was screened for recombinants. This was followed by self-pollination to produce F_2:3 family. Then eight plants from each of the F_2:3 families were selected for genotyping using markers in the QTL region, and homozygous plants for each recombinant event were selfed to produce enough seeds for the evaluation of SCR resistance in a field trial (2017 at Changge). To obtain unbiased assessment of the phenotype for a given recombinant corresponding to the genotype, two-tailed Student's t-test was used to test for the significance of disease score between genotypes. A significant difference (P < .01) between homozygote and resistance parental line CML496 indicated the absence of RppCML496 in the donor region, and the corresponding phenotype was deduced to be susceptible. The lack of a significant difference (P > .01) between the two genotypes suggested the presence of RppCML496 in the donor region, and the corresponding phenotype was deduced to be resistant. Analysis of both the deduced phenotype and the donor region for all recombinants enabled us to fine map RppCML496.

RESULTS Phenotypic data analysis

Phenotypic data was analyzed in each and across the environments. The parent CML496 showed complete resistance to P. polysora with a mean disease severity value of 1.1, while Lx9801 was highly susceptible to P. polysora with a mean disease severity value of 6.3 (Table 1). Significant correlations were found among the environments ranging from 0.61 to 0.83 (Supplemental Table S1). The heritability across all environments was 0.89, indicating that much of the phenotypic variance in the RIL population is genetically controlled.

TABLE 1 Means (± standard deviation [SD]), variation ranges, variance components (σ_g², σ_ge², σ_e²), and broad-sense heritabilities (H²) for the SCR response in the parents and recombinant inbred lines (RILs) population in three environments

	CML496	Lx9801	RILs
Environments	Mean ± SD	Mean ± SD	Mean	Range	σg2	σge2	σe2	H2
Ledong in 2015	1.2 ± 0.2	6.7 ± 0.3	7.2 ± 1.4	3.0–8.0	1.43**		0.38	0.88
Ledong in 2017	1.2 ± 0.2	6.7 ± 0.3	8.3 ± 0.9	3.0–9.0	1.40**		0.32	0.90
Changge in 2017	1.0 ± 0.1	5.9 ± 0.2	5.9 ± 1.6	1.0–7.0	1.37**		0.35	0.89
Combined	1.1 ± 0.2	6.3 ± 0.4	6.5 ± 1.9	2.0–8.0	1.36**	0.15	0.36	0.89

**

Significant at P < .01.

Genetic linkage map construction and QTL mapping

The genetic linkage map was constructed using the MaizeSNP9.4K BeadChip. Among 9,433 SNPs, 1,049 polymorphic SNPs in RILs were selected to construct the genetic map. The map spanned a total length of 1,840.4 cM with an average distance of 1.75 cM. Markers are evenly distributed on chromosomes and there is no obvious gap region, which is suitable for QTL mapping.

Combined with phenotypic data collected from the RILs, QTL conferring SCR resistance were identified in each of the three environments. In total three resistance QTL were detected, each accounting for 4–78% of the total phenotypic variation, and all the resistance alleles were derived from the resistant parent CML496 (Table 2). Among the three QTL, only one major QTL, named RppCML496 located on Chr.10 (bin 10.00/10.01; 2,365–3,662 Kb based on B73 reference genome), was detected across all three environments, and it accounted for 43–78% of the total phenotypic variation. The remaining two QTL were detected in only one of the field trials and each of these could account for no more than 6% of the total phenotypic variation, suggesting that they are environmentally sensitive, small-effect QTL.

TABLE 2 QTL associated with SCR scores detected by composite interval mapping (CIM) in three environments

		Marker	Ledong in 2015			Ledong in 2017			Changge in 2017
Bin a	Position (cM) b	Interval (bp) c	LOD	Add d	R2 e	LOD	Add d	R2 e	LOD	Add d	R2 e
6.05	97.31	151,999,143–153,935,606	5.65	0.45	0.04
9.03	54.51	63,978,028–77,727,443							4.04	0.65	0.06
10.00/10.01	16.41	2,365,414–3,662,794	54.00	2.07	0.78	20.95	1.00	0.43	24.83	1.88	0.60

a

Chromosome bin where QTL located.

b

The peak of the LOD of a QTL.

c

Physical position interval between two markers flanking the QTL in version 4 of the B73 reference genome.

d

Positive value indicates the allele from parent CML496 increases resistance.

e

Percentage of phenotypic variation explained by QTL.

Fine mapping of RppCML496

In the initial QTL mapping, RppCML496 was mapped to an interval of ∼1,297 Kb in bin 10.00/10.01 (Table 2). For fine-mapping RppCML496, a F₂ population of 2,944 plants was screened for recombination events between the two SNPs C06824-1 and C06838-1 flanking RppCML496. Then 471 recombinants were identified and selfed to produce F_2:3 families. From randomly selected 176 F_2:3 families eight plants were further genotyped with five newly developed markers (i.e. MZA9535-6, C01857-1, G047, CSR_2924602 and C06836-1) in the RppCML496 region (Supplemental Table S2), in which homozygous plants for each recombinant event were selfed to produce enough seeds for the evaluation of SCR resistance in a field trial (2017 at Changge).

We identified homozygous recombinants from randomly selected 24 F_2:3 families, which can be classified into 9 haplotypes. The number of recombination events in type 2, 3, 7 and 8 is 3, 8, 2 and 6, respectively, while each of the remaining 5 types included only one recombinant event. Four haplotypes (Type 4, 5, 6 and 7) were deduced to be resistant based on the resistant phenotype of their progenies, and 5 haplotypes (Type 1, 2, 3, 8 and 9) were deduced to be susceptible based on the phenotype of their progenies (Figure 1). Analysis of both the deduced phenotype and the donor region for all recombinants enabled us to narrow RppCML496 from 1,297 Kb to a 128 Kb region based on B73 reference genome, flanked by the markers C01857-1 and G047 (Figure 1). In the target region, genes Zm00001d023309 and Zm00001d023311 were identified, which encode for a putative cytochrome P450 superfamily protein and a putative disease resistance RPP13-like protein, respectively (Supplemental Table S3)

FIGURE 1. RppCML496 was fine-mapped to a 128 Kb interval. The recombinants were classified into 9 types. For each type, the chromosomal composition at RppCML496 is depicted as the solid bars with black or gray color, corresponding to resistant CML496 genotype and susceptible Lx9801 genotype, respectively. The total number of plants refers to all progeny of a given type. Based on analysis of the deduced phenotype and the donor region for all recombinants, the RppCML496 region was refined to a 128 Kb region with flanking markers C01857-1 and G047, which contain two functional annotated genes

DISCUSSION

SCR is a serious foliar disease in maize, and the deployment of resistance varieties is the most desirable means. However, the resistance may be lost with the deployment of the R genes due to the high variability of the pathogen P. polysora, which are exemplified by Rpp9 and RppQ being overcome due to the emergency of the new strains of P. polysora. Therefore, it is necessary to continue to search for new R genes, especially from tropical SCR resistant resources, because of the lack of effective resistance genes in temperate germplasm (Brewbaker et al., 2011). We identified the tropical line CML496 to be highly resistant to SCR and as the source of QTL RppCML496. The QTL RppCML496 is important since it is a stable QTL across environments with a major resistance effect. Therefore, this QTL is a good resistance resource to be introduced to temperate germplasm widely used in maize resistance breeding program.

There are two functional genes in RppCML496 region based on the 128 Kb annotation of the maize B73 reference genome (AGPv4), in which only one gene (Zm00001d023311) annotated as NBS-LRR gene (Supplemental Table S3). The candidate gene Zm00001d023311 with NBS-LRR domains seems to be the promising one since its molecular function causes resistance to biotrophic pathogens (Dangl & Jones, 2001; Jones & Dangl, 2006). However, we evaluated B73 for its resistance to SCR in Ledong and Changge and found B73 is highly susceptible to SCR. Therefore, other candidate genes not included in the B73 reference genome cannot be ruled out. Considering the complexity of the maize genome and high diversity of the plant R genes, it is necessary to do a de novo assembly of the CML496 genome for cloning the disease resistant gene to P. polysora in RppCML496 region.

Most of the R genes reside in clusters rather than being equally distributed on chromosomes in maize. For the R genes resistance to SCR, many of them have been mapped on the short arm of chromosome 10 in diverse maize germplasm (Holland et al., 1998; Jines et al., 2007; Wu et al., 2015; Zhang et al., 2010; Zhao et al., 2013; Zhou et al., 2007) including RppCML496 identified in the present study. We compared RppCML496 in the present study with those published in the literature and found that RppCML496 was mapped to the same region as RppP25, a gene from resistance line P25 used in Chinese maize breeding programs. However, a comparison of RppCML496 with other SCR resistance genes (e.g. RppQ, RppD, and RppS) is not available because of the differences in the used mapping populations and molecular markers. Allelism test among the R genes in the cluster on chromosome 10S has shown that there may be different alleles (Zhang et al., 2010). Therefore, map-based cloning the R genes in the cluster will help to elucidate R genes diversity and their functional diversification to different races of P. polysora.

SCR frequently develops in tropical and subtropical areas. However, the infrequent occurrence of SCR in warm temperate areas has resulted in the absence of constant disease pressure necessary for phenotypic selection in breeding. Alternatively, the use of marker-assisted selection (MAS) for the resistance would be a desirable approach. The QTL RppCML496, which causes resistance to SCR, could be a potential target for MAS in maize breeding programs. For this purpose, we further fine-mapped RppCML496 to an interval of 128 Kb. Flanking markers tightly linked with RppCML496 were developed, which will accelerate R gene incorporation in maize resistance programs.

ACKNOWLEDGMENTS

This work was supported by a grant from the National Key Research and Development Program of China (2016YFD0101003), the National Natural Science Foundation of China (31501326), Innovative Talents in Colleges and Universities of Henan Province (19HASTIT010) and the Shenzhen Science and Technology Research Funding (JSGG20160429104101251).

CONFLICT OF INTEREST DISCLOSURE

The authors declare that there is no conflict of interest.