Identification and Validation of Candidate Genes Conferring Resistance to Downy Mildew in Maize (Zea mays L.)

Downy mildew (DM) is a major disease of maize that causes significant yield loss in subtropical and tropical regions around the world. A variety of DM strains have been reported, and the resistance to them is polygenically controlled. In this study, we analyzed the quantitative trait loci (QTLs) involved in resistance to Peronosclerospora sorghi (sorghum DM), P. maydis (Java DM), and Sclerophthora macrospora (crazy top DM) using a recombinant inbred line (RIL) from a cross between B73 (susceptible) and Ki11 (resistant), and the candidate genes for P. sorghi, P. maydis, and S. macrospora resistance were discovered. The linkage map was constructed with 234 simple sequence repeat (SSR) and restriction fragment length polymorphism (RFLP) markers, which was identified seven QTLs (chromosomes 2, 3, 6, and 9) for three DM strains. The major QTL, located on chromosome 2, consists of 12.95% of phenotypic variation explained (PVE) and a logarithm of odds (LOD) score of 14.12. Sixty-two candidate genes for P. sorghi, P. maydis, and S. macrospora resistance were obtained between the flanked markers in the QTL regions. The relative expression level of candidate genes was evaluated by quantitative real-time polymerase chain reaction (qRT-PCR) using resistant (CML228, Ki3, and Ki11) and susceptible (B73 and CML270) genotypes. For the 62 candidate genes, 15 genes were upregulated in resistant genotypes. Among these, three (GRMZM2G028643, GRMZM2G128315, and GRMZM2G330907) and AC210003.2_FG004 were annotated as leucine-rich repeat (LRR) and peroxidase (POX) genes, respectively. These candidate genes in the QTL regions provide valuable information for further studies related to P. sorghi, P. maydis, and S. macrospora resistance.


Introduction
Maize is the world's leading cereal in terms of production, with 1093 million metric tons produced on 186 million hectares globally. Maize is grown in both temperate and tropical areas of the world and is largely (around 80%) produced under rainy conditions in sub-Saharan Africa, South and Southeast Asia, and Latin America. It is particularly susceptible to abiotic and biotic stress. Eight major countries growing maize (China, India, Indonesia, Nepal, Pakistan, the Philippines, Thailand, and Vietnam) produce 98% of Asia's and 28% of the world's maize yield. Heavy economic and yield losses have been recorded due to infection by downy mildew (DM) agents in the Philippines, Taiwan, Indonesia, Thailand, India, Japan, Australia, Venezuela, North America, Europe, West Africa, and other parts of the world [1][2][3][4][5][6].
DM is caused by obligate pathogens that cannot be cultured in the laboratory, and its sporulation prefers high relative humidity, a night temperature of 21-23 • C, and light drizzle with cool weather. It
Spreader genotypes were confirmed as having been 100% infected with DM, and the percentage of disease incidence was determined. The resistant plants did not show systemic symptoms of DM (emergence of characteristic chlorotic leaves) [50].

DNA Preparation
Genomic DNA was extracted from the parent plants and 192 F 7 individuals at the four-leaf stage using a modified DNA isolation protocol. The quality and quantity of the genomic DNA were analyzed with 1% agarose gel electrophoresis and Spectrophotometer (Model MN-913, Maestrogen, Hsinchu, Taiwan).

Molecular Marker Assay
A total of 727 SSR and RFLP markers covering all 10 chromosomes were obtained from the Maize Genetics and Genomics Database (MaizeGDB) and publications. The amplification was performed using a Takara PCR Thermal Cycler Dice Touch (Takara, Shiga, Japan) with 25 ng of gDNA and Takara Taq polymerase (Takara, Shiga, Japan). The polymerase chain reaction (PCR) conditions consisted of one cycle of 5 min at 94 • C for the initial denaturation and 30 cycles of 1 min at 94 • C, 1 min at 51-65 • C (based on the annealing temperatures standardized for different primers), and 30 s at 72 • C, with a final extension step for 10 min at 72 • C. The PCR products were separated by electrophoresis on 3% agarose gel with 1× Tris-acetate-ethylenediaminetetraacetic acid buffer at 100 V for 1 h and visualized using ethidium bromide staining. Markers were screened for polymorphism between the parent lines of the RILs. Of the 727 markers, 234 were found to be polymorphic based on agarose gel electrophoresis; these were used for genotyping the 192 RIL individuals (Table S1). Alleles of the parents of B73 and Ki11 were designated as 2 and 0, respectively.
2.5. Linkage Map Construction and QTL Analysis for P. sorghi, P. maydis, and S. macrospora Resistance QTL IciMapping v. 4.1 software from the Quantitative Genetics Group was used for both linkage map construction and QTL analysis using the MAP and BIP functions, respectively [51,52]. For linkage map construction, groups were ordered with an LOD score of 3.0 and the nearest neighbor combined with the two-opt algorithm (nnTwoOpt). Rippling was ordered with the sum of adjacent recombination frequencies with a window size of 5. Recombination frequencies were transformed into cM distances between linked loci using Kosambi's [53] mapping function. For QTL analysis, inclusive composite interval mapping of the additive (ICIM-ADD) function was applied with 1.0 step and 0.001 probability in the stepwise regression. The data of missing phenotypes were ordered with the Deletion command. The threshold LOD scores were calculated using 1,000 permutations with a type I error of 0.05 [51,54]. Suggestive QTLs with an average LOD value >3.0 in a dataset were noted. A QTL with an average LOD value > 3.0 and average phenotypic variance contribution > 10% was defined as a major QTL [55,56]. The QTLs in the present study were compared with previously published ones [2,21,31,32]. This software program uses an improved algorithm of composite interval mapping with increased power to detect QTLs to reduce false detection rates and create less biased QTL effect estimates by employing stepwise regression followed by QTL scanning [51].

RNA Extraction and Candidate Gene Screening with Quantitative Real-Time PCR (qRT-PCR)
Total RNA of healthy (control) and DM-infected parent lines was extracted from 6-week-old leaves of B73, CML228, CML270, Ki3, and Ki11 using TRIzol (Invitrogen, Carlsbad, CA, USA). For the qRT-PCR, cDNA was synthesized using ReverTra Ace ® qPCR RT Master Mix with gDNA Remover (Toyobo, Osaka, Japan) using 1 µg of total RNA extracted from leaves.
We searched for annotated genes near 14 flanked markers in the reference genome of B73 RefGen_v2 (MaizeGDB 2019). Partially, annotated genes were reported to be associated with biotic and abiotic stress. We obtained 62 genes in seven QTL locations; these were spread over chromosomes 2, 3, 6, and 9. The information and predicted protein of candidate genes were obtained from maizeGDB (B73 RefGen_v2). In addition, 19 DM-related genes from Arabidopsis thaliana, Oryza sativa, and Zea mays were obtained from the NCBI database (2019). The qRT-PCR primer sets were designed based on the 62 candidate genes and 19 DM-related genes using Primer3 [57] to validate the expression value of each candidate genes compared to control and DM-infected maize (Tables S2 and S3). The qRT-PCR was performed by using a gene-specific primer set, the CFX96 Touch Real-Time PCR Detection System (Bio-Rad Laboratories, USA). The qRT-PCR mixture contained 20 ng of cDNA, 0.2 µM of each gene-specific primer, 8 µL of sterile water, and 10 µL of TOPreal ™ qPCR 2× PreMIX (containing SYBR Green with low ROX; Enzynomics, South Korea) in a total volume of 20 µL. The qRT-PCR conditions were 10 min at 95 • C for the initial denaturation, then 45 cycles of 10 s at 95 • C, 10 s at 55-60 • C (based on the annealing temperatures standardized for different primers), and 30 s at 72 • C. The maize 18s rRNA (AF168884.1), UBCP (GRMZM2G102471_T01), MEP (GRMZM2G018103_T01), and LUG (GRMZM2G425377_T01) were used as reference genes (Table S4) [58]. The specificity and efficiency of the amplicon were confirmed through the melting curve analysis from 65 to 95 • C after each qRT-PCR. Three replicates of each experiment were performed for each candidate gene. The relative gene expression compared with the control plant was calculated using the 2 −∆∆Ct method [59].

Phenotypic Data for the DM Analysis
In this study, the DM incidence of DM-resistant genotypes (CML228, Ki3, and Ki11) and DM-susceptible genotypes (B73 and CML270) are reported as phenotypically similar between April and September 2015 (Table 1). Both B73 and CML270 showed 100% DM disease incidence within 6 weeks after inoculation in April and September, while comparable figures for CML228, Ki3, and Ki11 were 0%/25%, 0%/22.2%, and 5%/25%, respectively. According to the weather records, September was wetter and cooler than April. We evaluated resistance to P. sorghi, P. maydis, and S. macrospora using 192 F 7 families derived from B73 (susceptible) × Ki11 (resistant) since we decided that the approach of polygenic control for the various DM strains would offer more effective pathogen resistance ( Figure S2).
We performed DM screening of F 7 families in September 2017 because DM development is favorable under high humidity and low temperature conditions [60]. The mean of the DM incidence for the F 7 families ranged from 0% to 100%, and the distribution of the F 7 families was skewed toward the susceptible lines (a RILs mean of 86.49% for DM incidence) ( Figure 1 and Table 2).

Marker Data Analysis and Linkage Mapping
We screened 691 SSR and 36 RFLP markers to identify polymorphisms between B73 and Ki11. A total of 228 SSR and 6 RFLP markers were used to construct a linkage map of F 7 families. The map covered around 2042.51 cM at an average marker interval of 9.12 cM and 79 bins among the approximate 120 bin locations on 10 chromosomes. Around 85.04% of the markers were within 20 cM of the nearest interval. Chromosome 2 had the highest number of markers (34), while chromosome 7 had the lowest (14) (Table S5). Genes 2020, 11, x FOR PEER REVIEW 6 of 24

Marker Data Analysis and Linkage Mapping
We screened 691 SSR and 36 RFLP markers to identify polymorphisms between B73 and Ki11. A total of 228 SSR and 6 RFLP markers were used to construct a linkage map of F7 families. The map covered around 2042.51 cM at an average marker interval of 9.12 cM and 79 bins among the approximate 120 bin locations on 10 chromosomes. Around 85.04% of the markers were within 20 cM of the nearest interval. Chromosome 2 had the highest number of markers (34), while chromosome 7 had the lowest (14) (Table S5).

qRT-PCR Validation of the Candidate Genes for P. sorghi, P. maydis, and S. macrospora Resistance
We performed an analysis of the expression levels of reference genes (18s rRNA genes, UBCP, MEP, and LUG) [58]. Nineteen DM-related genes from Arabidopsis thaliana, Oryza sativa, and Zea mays (LOC4345959, LOC107275863, SGT1B, LOC103650325, LOC100382073, LOC103647182, LOC4351808, HSK, LOC4324025, LOC103654479, LOC4345309, DMR6, IDC1, LOC103632498, LOC100191339, LOC107275878, LOC103648264, LOC103642860, and EDM2) were validated by comparing control and DM-infected plants that were DM-resistant (CML228, Ki3, and Ki11) or DM-susceptible (B73 and CML270) genotypes (Table 4). Five genes (LOC107275878, LOC103632498, LOC103647182, HSK, and LOC4345309) were upregulated by DM in the DM-resistant genotypes, while another five genes (LOC103648264, LOC4324025, LOC100191339, LOC100382073, and IDC1) were partially upregulated ( Figure 3). The other nine genes were not expressed in the control or DM-infected genotypes. We considered that gene expression levels showed a wide range of variation because resistant genotypes (CML228, Ki3, and Ki11) differ in origin and pedigree. umc1231-umc2343, and dupssr29-umc1505 presented 0.75%, 0.48%, 2.90%, 2.85%, 0.77%, and 0.47% of the phenotypic variation, respectively.   The candidate genes related to P. sorghi, P. maydis, and S. macrospora resistance were obtained by searching for all of the flanked markers in the QTL regions in MaizeGDB (B73 RefGen_v2) (Tables 5  and S6). The genes were flanked by 14 markers (umc1165, bnlg1297, umc2363, phi098, umc1030, phi243966, mmc0022, bnlg420, bnlg345, umc1859, umc1231, umc2343, dupssr29, and umc1505). Three markers (bnlg1297, umc2363, and phi098) were identified by the same physical location in B73 RefGen_v2, and so we suggest that qDM1 and qDM2 are likely the same. qDM1 was analyzed as a major QTL, thus we supposed that DM-related genes were possible to locate in qDM1 region. qDM1 and qDM2 are located at 4,051,998-4,407,701; 37 genes were obtained in the region. Of the 37 genes, 7 were identified as short sequence length or without genetic information. The relative expression levels of the 62 genes by qRT-PCR were validated by comparing the control and DM-infected plants from B73, CML228, CML270, Ki3, and Ki11. However, 17 genes were not expressed in the control or DM-infected genotypes. The data are presented as the mean ± standard error (SE). Student's t-tests were used in the statistical analysis (* p < 0.05; ** p < 0.01; *** p < 0.001). The candidate genes related to P. sorghi, P. maydis, and S. macrospora resistance were obtained by searching for all of the flanked markers in the QTL regions in MaizeGDB (B73 RefGen_v2) ( Table 5 and Table S6). The genes were flanked by 14 markers (umc1165, bnlg1297, umc2363, phi098, umc1030, phi243966, mmc0022, bnlg420, bnlg345, umc1859, umc1231, umc2343, dupssr29, and umc1505). Three markers (bnlg1297, umc2363, and phi098) were identified by the same physical location in B73 RefGen_v2, and so we suggest that qDM1 and qDM2 are likely the same. qDM1 was analyzed as a major QTL, thus we supposed that DM-related genes were possible to locate in qDM1 region. qDM1 and qDM2 are located at 4,051,998-4,407,701; 37 genes were obtained in the region. Of the 37 genes, 7 were identified as short sequence length or without genetic information. The relative expression levels of the 62 genes by qRT-PCR were validated by comparing the control and DM-infected plants from B73, CML228, CML270, Ki3, and Ki11. However, 17 genes were not expressed in the control or DM-infected genotypes.

Discussion
DM, caused by Peronosclerospora species in maize, causes severe yield loss despite the use of chemical pesticides and has spread in many tropical and subtropical regions throughout the world. The lack of suitable studies for gene diversity in DM has been a major constraint in tropical Asia, especially in the maize-growing environments of South and Southeast Asia [61]. The results from a great many studies on P. sorghi (sorghum DM) resistance have been reported because it is widely distributed throughout Asia, Africa, and America, where it is easier to perform experiments than in . Relative expression levels (log 2 DM-infected/control) of 15 candidate genes for P. sorghi, P. maydis, and S. macrospora via qRT-PCR in the control and infected plants. The analysis was conducted on three independent plants as biological replicates. The data are presented as the mean ± SE values. Student's t-tests were carried out in the statistical analysis (* p < 0.05; ** p < 0.01; *** p < 0.001).

Discussion
DM, caused by Peronosclerospora species in maize, causes severe yield loss despite the use of chemical pesticides and has spread in many tropical and subtropical regions throughout the world. The lack of suitable studies for gene diversity in DM has been a major constraint in tropical Asia, especially in the maize-growing environments of South and Southeast Asia [61]. The results from a great many studies on P. sorghi (sorghum DM) resistance have been reported because it is widely distributed throughout Asia, Africa, and America, where it is easier to perform experiments than in other regions. However, we need to evaluate various DM species because the fungus can easily spread to other regions through the movement of spores in the air.
We screened resistance for P. sorghi (sorghum DM), P. maydis (Java DM), and S. macrospora (crazy top DM) using 192 F 7 families derived from B73 (susceptible) × Ki11 (resistant) in Phnom Penh, Cambodia. The distribution of the F 7 families was skewed toward the susceptible lines. According to previous studies, it is not uncommon that phenotype values of mapping populations do not follow a normal distribution [2,21,30]. The phenotype data for the DM of the F 7 families leaned more toward susceptible because of the response to the three DM pathogens. We used 691 SSR and 36 RFLP markers from MaizeGDB (B73 RefGen_v2) for parental polymorphism. We performed polymorphism analysis by electrophoresis using 3% agarose gel with 1× TAE buffer at 100 V for 1 h. It is necessary that a difference of greater than 20 bp between PCR products could be detected by using 3% agarose gel and loading for 1 h. Additionally, we analyzed flanked markers in the QTL regions by using QiaXcel advanced system (Qiagen, Hilden, Germany). The constructed linkage map covered around 2042.51 cM at average intervals of 9.12 cM between markers for 228 SSR and six RFLP markers on 10 chromosomes. Around one-third of the markers showed to be polymorphic; this result was considered an average ratio when compared to references [21,[32][33][34]62]. These markers were distributed over 79 bins among around 120 bin locations. Approximately 85.04% of the markers were within 20 cM of the nearest interval (Table S5). In QTL analysis, it is likely that RILs (F 7 families (B73 × Ki11)) made a significant contribution because the use of RILs is more powerful than F 2 or backcross populations in QTL analysis [25][26][27]. However, in this study, phenotype data for P. sorghi, P. maydis, and S. macrospora resistance showed recessive trait. In analysis of recessive traits, backcross inbred lines (BILs) derived from resistant genotype have more power to detect recessive QTLs because they have an advantage in segregation ratio compared to F 2 or RILs [63].
The seven QTLs for P. sorghi, P. maydis, and S. macrospora resistance were identified on 4 of 10 chromosomes (Table 3). A major QTL was identified in bin 2.01 (qDM1) on chromosome 2, while the other QTLs were detected in bins qDM2 (2.02), qDM3 (3.04), qDM4 (3.05), qDM5 (6.05/6.06), qDM6 (9.05), and qDM7 (9.07). All of the DM-resistant alleles were obtained from Ki11. According to previous studies, the QTLs for DM resistance have been detected in chromosomes 2, 3, 6, and 9, and several flanked markers of QTL regions have been reported. We considered that these seven QTLs affected P. sorghi (sorghum DM), P. maydis (Java DM), and S. macrospora (crazy top DM) from Cambodia. The QTL regions found in the present study were similar to those in previous studies for P. sorghi (sorghum DM) and P. heteropogoni (Rajasthan DM) resistance [21,[32][33][34]49] (Table S7). qDM1 and qDM2 shared the same flanked markers (umc1165 and umc2363 in bins 2.01 and 2.02, respectively) [34]. qDM3 had flanked marker umc1030 in bin 3.04 [62]. qDM4 and qDM5 shared the same flanked markers (bnlg420 and umc1859 in bins 3.05 and 6.05/6.06, respectively) [21]. In addition, qDM6 and qDM7 shared the same flanked markers (umc2343 and dupssr29 in bins 9.05 and 9.07, respectively) [33]. The bin locations (3.04 and 3.05, respectively) of qDM3 and qDM4 matched with other reports on QTLs [21,32,33,49]. The other QTLs were identified from five different DM strains (P. sorghi and P. heteropogoni from India, P. zeae from Thailand, P. philippinensis from the Philippines, and P. maydis from Indonesia) and sorghum DM on bin 6.05 using RILs (derived from Ki3) and backcross populations [2,21]. Ki3 and Ki11 were developed as DM-resistant lines derived from the Suwan1 strain from Thailand. These results suggest that it should be possible to detect candidate genes for P. sorghi, P. maydis, and S. macrospora resistance near these bin positions. Although the QTL analysis was performed using different mapping populations, the results of this study are well matched with previously reported ones.
We analyzed 19 DM-related genes from Arabidopsis thaliana, Oryza sativa, and Zea mays (Table 4): EDM2, SGT1B, LOC100191339, HSK, DMR6, IDC1, and LOC103642860, among others. Among the 19 genes, five were upregulated in the DM-resistant genotypes, two of which (LOC103632498 and LOC103647182) are located on chromosomes 2 and 7 of maize, respectively. LOC100191339 was located on nearby qDM4 and was highly upregulated in DM-infected CML270 (susceptible). Two genes (LOC107275878 and LOC4345309) and HSK, originating from Arabidopsis thaliana and Oryza sativa, were significantly upregulated in the DM-infected DM-resistant genotypes. Hence, we considered that these genes are conserved in monocotyledon and dicotyledon plants [64,65].
We obtained 62 candidate genes for P. sorghi, P. maydis, and S. macrospora resistance near the flanked markers in the QTL region; these genes were validated by comparing their relative expression levels in the control and DM-infected groups. The physical locations of candidate genes were continually updated from B73 RefGen_v1 to B73 RefGen_v4, but updating the genetic information of the genomic markers is slower than the transcripts. Hence, we used B73 RefGen_v2 to set the physical locations of the genomic markers and candidate genes. The annotations of 45 genes were analyzed to predict their functions using the Pfam database. The annotations and functions of these genes were identified as being related to DM resistance ( Table 6 and Table S8). There are various factors involved in DM resistance, such as POX (peroxidase), MYB (transcriptional activator Myb), GSO1 (LRR receptor-like serine/threonine protein kinases (STKs), (LRR-STKs)), plant RLKs, polygalacturonase inhibitor protein (PGIP), polyphenol oxidase (PPO), NAC, WRKY transcription factors, and pathogenesis-related (PR) protein [66][67][68][69][70][71][72].
Three genes (AC210003.2_FG004, AC191071.3_FG001, and GRMZM2G039345) were annotated as POX, FMO, and RuBisCO (ribulose-1,5-bisphosphate carboxylase/oxygenase), respectively. The activities of POX and PPO, along with b-1,3-glucanase, are associated with DM resistance in sunflowers [73]. Plants can implement DM post-infection mechanisms such as an increase in localized callose deposition to fortify plant cell walls [74,75], reactive oxygen species (ROS), peroxidase activity, and hypersensitive response activation [74,76]. When a resistant grapevine is infected with DM, it produces high concentrations of resveratrol that can be oxidized by induced peroxidase [77]. In broccoli, BoAPX (ascorbate peroxidase) genes contribute enhanced both DM and heat tolerance, and play important roles in cellular defense against ROS-mediate oxidative damage [68]. In previous studies, high POX activity has been associated with resistance to PM [78] in lettuce [79] and melon [80] and to Verticillium dahlia in tomatoes [81]. PM is very similar to DM in that both are caused by fungi and are widespread under humid and low temperature conditions; the disease symptoms are also similar. FMO1 positively regulates the enhanced disease susceptibility1 (EDS1) pathway in Arabidopsis thaliana. The EDS1 pathway controls defense activation and programmed cell death against pathogens [82]. In addition, a defect in FMO1 partially disables toll interleukin 1 receptor nucleotide binding sites leucine-rich repeat (TIR-NB-LRR) resistance and basal defense. In Arabidopsis thaliana, RPP4, which has been identified as a TIR-NB-LRR protein coupled with its dependence on signaling components in leaves, confers resistance to DM [83]. RPP4-mediated resistance is regulated by interaction between EDS1 and NDR1 signaling in cotyledons.
Two genes (GRMZM2G342564 and GRMZM2G040095) were identified as lipoxygenase (LOX)-producing. LOX is known to play a role in disease resistance for many host pathosystems. In earlier reports, LOX activity was found to increase in resistant plants and to decrease in susceptible plants. Also, LOX was reported to affect DM and PM resistance of pearl millet and wheat, respectively [84][85][86][87][88]. GRMZM2G028643, GRMZM2G128315, and GRMZM2G330907 (identified as LRR) and GRMZM2G363066 (identified as nonspecific STK) were associated with defense reactions against pathogens. PR protein are encoded by disease resistance (R) genes, responding to pathogenic microorganisms and signaling cascades that activate defense reactions [89,90]. The largest family of PR proteins is defined by the presence of 12 to 21 LRRs (it has been speculated that LRRs bind pathogen-derived ligands). A glutamate-to-lysine substitution in LRR partially compromises the function of R genes against DM [91]. Several kinases (protein kinases, wall-associated kinases, calcium-dependent protein kinases, STKs, LRR-STKs, and mitogen-activated protein kinases (MAPKs)) are strongly associated with signal transduction mediated by Ca 2+ permeable channels [90,[92][93][94]. It has been shown that the activation of the PR protein and phenylpropanoid pathway enzymes such as LRR-STKs and MAPKs responds to DM infection in pearl millet by inducing molecules for signal transduction [95]. Also, nucleotide-binding site (NBS)-LRR and receptor-like proteins (RLP) were included in five classes of R genes, these have been reported as resistance against DM [70,71]. RLKs are well known to play a role in many important signaling process such as plant growth, development, hormone signaling, and stress response. LRR-RLK family proteins regulate plant innate immunity and defense [69]. PGIP is a defense protein consisting of an extra-cytoplasmic LRR which specifically binds the invading fungus cell wall to the host tissue [96]. In previous studies, the transcription of PGIP and a cell wall glycoprotein were induced in DM-resistant strains of pearl millet [38] and grape [97]. In Arabidopsis thaliana, homoserine accumulation in the chloroplast triggers a novel form of DM resistance that is independent of known immune responses [98]. In pearl millet, the role of PGIP in resistance against DM pathogen (Sclerospora graminicola) was reported by differential gene expression analysis between resistant and susceptible genotypes [72].
GRMZM2G005984 was classified as a photosystem II protein. In grapevine, PM-responsive proteins are involved in photosynthesis, metabolism, disease/defense, protein destination, and protein synthesis [99]. These proteins are associated with the plant defense response and slow down disease progression against Erysiphe necator. GRMZM2G314171 was classified as DEAD-box RNA helicase, one of which (OsBIRH1) has been found to modulate the defensive response to infection and oxidative stress in rice [100]. The DEAD-box RNA helicase family functions in chloroplast biogenesis in maize [101]. GRMZM2G178880 was classified as a GTF (general transcription factor); these proteins are probably involved in defense-related processes [102]. It has been shown that GTF expression increases during the interaction between PM and barley [103], and its expression level is induced after wounding and phytopathogenic bacteria attack [104]. GRMZM2G024293 was identified as a conserved hypothetical adenosine triphosphate (ATP)-binding protein. Its downregulation brings about a decrease in ATP binding in DM-infected plants, which makes it possible to estimate DM susceptibility. Meanwhile, further study is needed to identify the functional characteristics of the uncharacterized genes.
Hence, we suggest that 10 genes (AC210003.2_FG004, AC191071.3_FG001, GRMZM2G039345, GRMZM2G028643, GRMZM2G128315, GRMZM2G330907, GRMZM2G363066, GRMZM2G005984, GRMZM2G178880, and GRMZM2G314171) are related to P. sorghi, P. maydis, and S. macrospora resistance. The downregulation of some of the P. sorghi, P. maydis, and S. macrospora resistance genes occurs in DM-infected plants. Moreover, it was shown that 30 genes are partially upregulated in DM-infected plants of resistant genotype. From these results, we predict the possibility that the phenotype of resistant plants shows resistance (R) or moderate resistance (MR).
Experiments on P. sorghi, P. maydis, and S. macrospora resistance have been limited to QTL analysis in previous studies. However, we focused on their great importance through the screening of candidate genes using qRT-PCR. The results of this study can serve as fine-mapping for DM and marker-assisted selection (MAS) in maize breeding. In further research, our approach could be used to screen other DM strains in different environments using RILs and to validate candidate genes using CRISPR/Cas9. In addition, the genes and pathways associated with resistance could be identified via comparative transcriptome profiling using RNA-seq.
Supplementary Materials: The following are available online at http://www.mdpi.com/2073-4425/11/2/191/s1, Figure S1: Field layout using the spreader-row technique in Phnom Penh, Cambodia; Figure S2: Maize phenotypes and symptoms of downy mildew (DM) 6 weeks after inoculation using the spreader-row technique; Figure S3: The logarithm of odds (LOD) values of the P. sorghi, P. maydis, and S. macrospora resistance quantitative trail loci (QTLs) between the flanked markers; Figure S4: Relative expression levels (log 2 DM-infected/control) of 30 candidate genes for P. sorghi, P. maydis, and S. macrospora via quantitative real-time polymerase chain reaction (qRT-PCR) on the control and infected plants; Table S1: Primer information of SSR and RFLP markers for QTL analysis; Table  S2: Primer information of P. sorghi, P. maydis, and S. macrospora related genes for qRT-PCR analysis; Table S3: Primer information of candidate genes for qRT-PCR analysis; Table S4: Primer information of reference genes for qRT-PCR analysis; Table S5: The numbers of markers and bins covered in each chromosome; Table S6: List of the P. sorghi, P. maydis, and S. macrospora resistance candidate genes between the left and right flanked markers in QTL regions; Table S7: Comparison of QTLs identified in the present study with previous reports for resistance to P. sorghi, P. maydis, and S. macrospora; Table S8: Pfam (EBI 2019) domain analysis of 47 candidate genes.