The maize mitogen‐activated protein kinase kinase kinase gene ZmMAPKKK45 is associated with multiple disease resistance

Summary

• Southern leaf blight (SLB), caused by the necrotrophic fungus Cochliobolus heterostrophus, is a major foliar disease of maize (Zea mays) world‐wide.

• A genome‐wide association study was performed to dissect the genetic basis of SLB resistance in maize. Functional validation was performed using mutant and transgenic analyses. Molecular experiments provided preliminary insights into the underlying disease resistance mechanisms.

• Association analyses identified 14 single nucleotide polymorphisms (SNPs) linked to SLB resistance, 13 of which overlapped with known quantitative resistance loci, highlighting 10 candidate genes. Functional studies confirmed ZmMAPKKK45, encoding a mitogen‐activated protein kinase kinase kinase (MAPKKK), is the causal gene at a resistance locus on chromosome 3. ZmMAPKKK45 also enhanced resistance to northern leaf blight and gray leaf spot and promotes reactive oxygen species (ROS) accumulation during defense responses.

• Our results indicate that ZmMAPKKK45 functions outside canonical MAPK cascades and likely enhances disease resistance by upregulating maize respiratory burst oxidase homolog (ZmRBOH) genes, thereby increasing ROS production and contributing to broad‐spectrum foliar disease resistance in maize.

Introduction

Maize (Zea mays ssp. mays L.) is one of the most important staple foods world‐wide and plays a crucial role in sustaining food security and satisfying energy requirements (Ranum et al., ²⁰¹⁴). Maize diseases are among the most important yield‐limiting factors, and disease resistance is essential for reliable maize production (Zhu et al., ²⁰²¹). Identifying disease‐resistance loci and breeding disease‐resistant maize varieties is the most fundamental and effective method for preventing and controlling maize diseases.

Mitogen‐activated protein kinase (MAPK) cascades are widely conserved eukaryotic signaling pathways that consist of three protein kinase components: MAPK kinase kinases (MAPKKKs), MAPK kinases (MAPKKs) and MAPKs (Sun & Zhang, 2022). They generally transduce and amplify signals received externally to the cell by membrane‐bound receptors, usually receptor‐like kinases (RLKs) or receptor‐like proteins (RLPs). Signal reception leads to the activation of a MAPKKK which then phosphorylates and activates specific MAPKKs. The activated MAPKKs phosphorylate and activate their target MAPKs which, in turn, phosphorylate their target substrates to regulate their activities (Zhang & Zhang, 2022; Xie et al., ²⁰²³). MAPK cascades are important in a remarkable array of plant developmental and stress responses. These include meristem growth and maintenance, various aspects of fertilization and gamete production, hormone response and, in many cases, response to pathogens (Sun & Zhang, 2022).

In the area of defense response, MAPK cascades are probably best known for transducing the pattern‐triggered immunity (PTI) response (Zhang & Zhang, 2022). The PTI response occurs upon the perception of a pathogen‐associated molecule, such as flagellin, in the apoplast by a specialized RLK or RLP known as a pattern recognition receptor (PRR). This signal is transduced into the cytosol through a series of phosphorylation events leading to the activation of receptor‐like cytoplasmic kinases which induce the production of reactive oxygen species (ROS) through the activation of the NADPH oxidase protein respiratory burst oxidase homolog D and induction of defense gene expression through MAPK cascades activation (Bigeard et al., ²⁰¹⁵).

PTI, which is activated outside the cell, is intimately connected with another defense response known as effector‐triggered immunity (ETI). ETI is activated by the recognition of pathogen‐derived proteins, known as effectors, inside the cell (Ngou et al., ^{2021, 2022}). MAPK cascades are also an important component of the ETI response. In Nicotiana benthamiana, a MAPK cascade has been reported to activate an ROS burst via activation of WRKY transcription factors which induced RBOHB expression (Adachi et al., ²⁰¹⁵; Yoshioka et al., ²⁰²³). ROS production during ETI may often require prior or simultaneous activation of the PTI response (Yuan et al., ²⁰²¹).

MAPKKKs can be divided into three subfamilies: the MAPK ERK KINASE KINASE (MEKK) subfamily, the RAPIDLY ACCELERATED FIBROSARCOMA (Raf) subfamily and the ZIPPER INTERACTING PROTEIN KINASE (ZIK) subfamily (Xie et al., ²⁰²³). Several MAPKKKs have been reported to function in the plant defense response. AtMEKK1 functions downstream of FLAGELLIN‐SENSING 2 (FLS2), an RLK that detects bacterial flagellin (Asai et al., ²⁰⁰²). AtMKD1, a Raf‐like MAPKKK, is required for full immunity against bacterial and fungal infection (Asano et al., ²⁰¹⁹). EDR1, a member of the Raf subfamily in Arabidopsis, acts as a negative regulator of plant defense, with loss‐of‐function mutants exhibiting enhanced resistance to fungal, bacterial and oomycete pathogens (Frye & Innes, 1998; Frye et al., ²⁰⁰¹). EDR1 appears to suppress the Arabidopsis immune response through a number of distinct mechanisms. It negatively regulates defenses and directly modulates the MKK4/MKK5‐MPK3/MPK6 cascade, a pathway that is triggered after the activation of several PRRs including FLS2 (Asai et al., ²⁰⁰²; Zhao et al., ²⁰¹⁴). EDR1 is thought to limit the accumulation of MKK4 and MKK5 by suppressing the phosphorylation and consequent degradation of KEG, an E3 ligase that itself ubiquitinates MKK4 and MKK5 and targets them for degradation (Gao et al., ²⁰²¹). EDR1 has also been reported to interfere with the interaction of EDS1 and PAD4, related lipase‐like proteins that modulate immunity, and to suppress the immune responses associated with their activity (Neubauer et al., ²⁰²⁰). The rice EDR1 is also a negative regulator of immunity, inhibiting OsMPKK10.2 (Kim et al., ²⁰⁰³; Shen et al., ²⁰¹¹; Ma et al., ²⁰²¹). In rice, the Raf‐like MAPKKK gene OsILA1 (OsMAPKKK43) suppresses the OsMAPKK4‐OsMAPK6 cascade to negatively regulate resistance to Xanthomonas oryzae pv. oryzae (Chen et al., ²⁰²¹). Similarly, in cotton, GhMAP3K65 also acts as a negative regulator of disease resistance (Zhai et al., ²⁰¹⁷). Although MAPKKKs in maize defense remain poorly characterized, several MAPKs (Zhang et al., ²⁰¹⁴; Jiang et al., ²⁰²³) and MAPKKs (Cai et al., ²⁰¹⁴) contribute to disease resistance.

Southern leaf blight (SLB, causal agent Cochliobolus heterostrophus) is a common necrotrophic fungal disease of maize world‐wide (Yoder, 1988). Host resistance to SLB is largely quantitative (Lim & Hooker, 1976; Burnette & White, ¹⁹⁸⁵; Holley & Goodman, ¹⁹⁸⁹; Kump et al., ²⁰¹¹). Numerous quantitative resistance loci (QRL) for SLB resistance have been mapped (Carson et al., ²⁰⁰⁴; Balint‐Kurti et al., ^{2006, 2007, 2008}; Zwonitzer et al., ²⁰⁰⁹; Kump et al., ²⁰¹¹; Belcher et al., ²⁰¹²; Bian et al., ²⁰¹⁴; Chen et al., ^2023b) and several genes underlying resistance/susceptibility to SLB have been identified. The ascorbate peroxidase ZmAPX1 positively regulates SLB resistance by activating the JA‐mediated defense signaling pathway (Zhang et al., ²⁰²²). ChSK1, ZmNANMT and ZmAGO18b are all negative regulators of maize SLB resistance (Chen et al., ^2023a; Dai et al., ²⁰²³; Li et al., ²⁰²³). ChSK1 likely functions as an inhibitory co‐receptor in the PTI response (Chen et al., ^2023a). ZmNANMT encodes a homolog of proteins that convert nicotinate to trigonelline, a function conserved across various plant species (Li et al., ²⁰¹⁷). The enhanced resistance observed in ZmNANMT knockout lines might be caused by the accumulation of nicotinic acid, which could increase plant disease resistance (Li et al., ²⁰²³). ZmAGO18b, an argonaute protein, is thought to regulate small RNA production, though the mechanism by which it modulates resistance to SLB is unclear (Dai et al., ²⁰²³). The multiple disease resistance gene ZmCCoAOMT2, positively regulates resistance to SLB likely by influencing metabolites within the phenylpropanoid pathway and/or by regulating programmed cell death (Yang et al., ²⁰¹⁷).

In this study, we used a maize association mapping population consisting of 270 diverse maize inbred lines to identify SLB resistance loci through a genome‐wide association study (GWAS). We report the associated loci, compare them with previously identified QRLs associated with SLB resistance and identify candidate genes. We demonstrate that one of the candidate genes, ZmMAPKKK45, positively regulates resistance to SLB as well as to northern leaf blight (NLB) and gray leaf spot (GLS) diseases and modulates the accumulation of ROS.

Materials and Methods

Plant materials and field trials

A set of 270 diverse lines (Supporting Information Table S1) from a maize (Zea mays L.) diversity panel, henceforth ‘the diversity panel’, representing global public maize inbreds, was used in this study. This population has been used in previous studies (Flint‐Garcia et al., ²⁰⁰⁵). SLB trials were performed at five environments: the summers of 2004, 2005 and 2006 at Central Crops Research Farm in Clayton, NC (referred to as NC04, NC05 and NC06), the summer of 2006 in Tifton, GA (GA06), and the winter of 2003 in Homestead, FL (FL03). Each location contained two replications in a randomized complete block design, except NC04, which had one replicate. In the NC environment, plots were 2 m in length with a 0.6 m alley between plots and inter‐row spacing was 0.97 m. In FL and GA, plots were 0.91 m in length with 0.97 m between rows and a 0.61‐m alley at the end of each plot. Part of the data was used in a previously published study (Wisser et al., ²⁰¹¹). In this study, more lines (270 vs 253) were included in the GWAS analysis.

The UMFu‐06499 mutant line, which contains a mutator insertion (mu1047777) at the 5′‐UTR of GRMZM2G080499 and is in a background of the maize inbred W22, was derived from the UniformMu insertional mutant resource and provided by the Maize Genetics Cooperation Stock Center (McCarty et al., ²⁰¹³). It was selfed to obtain lines homozygous for the mutation of interest. The homozygous mutant was crossed to W22 and selfed for three generations to generate F_3:4 families. DNA from bulked leaf samples of five individual plants per family was used for genotyping with two pairs of primers (F2/R2 and TIR6/R2). If F2/R2 shows a band while TIR6/R2 does not, the genotype is wild‐type (W). If both F2/R2 and TIR6/R2 show bands, the genotype is heterozygous (H). If F2/R2 does not show a band but TIR6/R2 does, the genotype is mutant (M). All the primers used are detailed in Table S2.

The full‐length cDNA of ZmMAPKKK45 was obtained from W22 by using Q5 high‐fidelity DNA polymerase (New England BioLabs, M049). The fragment was inserted into the pMCG1005 vector under control of the maize ubiquitin promoter. A confirmed clone was designated pMCG1005‐ZmMAPKKK45 and was transformed into the maize inbred line B104 through an Agrobacterium‐mediated transformation system at the Plant Transformation Facility at Iowa State University. B104 is susceptible to SLB. All the transgenic events were selfed or backcrossed to B104. Vector construction and transgene identification primers are shown in Table S2.

SLB and NLB trials to evaluate mutant and transgenic lines' phenotype plots were planted at Central Crops Research Farm Calyton, NC. Plots were 2 m in length with 0.97 m between rows and a 0.6‐m alley between ranges. In Andrews, NC, GLS trial plots were planted using the same configuration as the SLB and NLB trials, except with a 4 m plot length. Twelve seeds were planted in each row for SLB and NLB trials, whereas 15 seeds per row were planted for GLS trials at Andrews. Each disease evaluation included at least two replicates in a randomized complete block design, except for GLS in Andrews, which had one replicate.

Artificial inoculation and evaluation in the field

In the field, plants were artificially inoculated with isolate 2‐16Bm (Carson, 1998) of the causal agent of SLB, C. heterostrophus (Drechsler), or with a mixture of race 0, race 1, race 2, 3 and race 2, 3, N of the causal agent of NLB, Exserohilum turcicum (Passerini). GLS was developed by natural infection by its causal agent, Cercospora zeae‐maydis (Tehon and Daniels) in Andrews, where there is sufficient natural disease pressure. Resistance to SLB, NLB and GLS was scored based on a 9‐point scale that emphasized symptoms on the ear leaf, with 1 being completely dead and 9 being asymptomatic (Sermons & Balint‐Kurti, 2018). Multiple measurements were taken at c. 7 d intervals, beginning at 3–4 wk after anthesis, with 3–5 measurements per experiment. Days to anthesis (DTA) at five environments was also measured as the number of days between planting and 50% pollen shed in a row on each plot (Table S1). The DTA data had been used in a previous study (Wisser et al., ²⁰¹¹).

Growth‐chamber experiment and SLB inoculation

A 2.4 × 3.7 × 2.1 m growth chamber at the North Carolina State University (NCSU) Phytotron was maintained at a temperature of 25°C daytime and 18°C nighttime, with a 16 h light : 8 h dark cycle. Plants were grown in 6‐inch pots in standard substrate composed of a 1 : 2 peat‐lite/gravel mixture and watered with standard NCSU Phytotron nutrient solution once daily in the afternoon. One seed was planted per pot. A split–split–plot design was used with three replicates for each experiment with different genotypes as the whole plot factor, inoculation or not as the subplot factor, and different inoculation times as the sub–subplot factor. Three biological replicates were analyzed for gene expression, pathogen biomass quantification and 3,3′‐diaminobenzidine (DAB) staining experiments.

SLB inoculation was carried out at the three‐ or four‐leaf stage in the growth chamber. Briefly, the inoculum consisted of C. heterostrophus strain C5 (Leach et al., ¹⁹⁸²) spores suspended in sterile water with 0.05% Tween‐20 and 0.05% agar, at a titer of 1–5 × 10⁵ spores ml⁻¹. For gene expression, western blotting experiments and DAB staining assays, the whole plants were inoculated by spraying c. 1 ml of the spore suspension per plant and the middle third portion of the inoculated leaves was sampled for the experiments. Subsequently, the inoculated plants were placed in clear plastic bags for 12 h to promote spore germination. For the evaluation of SLB resistance, a piece of 4 cm by 2 cm sterile filter paper (110 mm diameter, Whatman), completely soaked with the inoculum, was placed in the middle of the third leaf. The inoculation site was then covered with tinfoil to keep it dark and moist for 12 h. After tinfoil removal, the plants were maintained under normal conditions for 2 d, and the lesion areas were calculated with APS Assess 2.0 software (Lamari, 2009).

Statistical analysis of phenotypic data for GWAS

For GWAS in the diversity population, the level of SLB resistance was calculated as the area under the disease progress curve (AUDPC), standardized by the total time during which measurements were taken. To minimize the effect of environmental variation, the best linear unbiased predictors (BLUPs) across environments were calculated for each line using ASRSeml 3.0 software (http://www.vsni.co.uk/software/asreml). Since SLB is a late‐season disease that infects maize plants after pollination, special care was taken to avoid confounding effects of maturity on disease resistance. The mixed model analysis fit DTA as a covariate (fixed effect), while inbred lines, environment, replications nested with environment and line by environment interactions were fitted as random effects. Since the total data missing rate for SLB phenotype across the five environments was low (3.21%, Table S1), we consider the data to be sufficiently balanced to permit the use of BLUPs for GWAS in this case (Holland & Piepho, 2024). The heritability was calculated using the equation as:

$${\widehat{H}}_c=1-{\sigma }_{\mathit{PPE}}^2/2{\widehat{\sigma }}_G^2$$

where ${\sigma }_{\mathit{PPE}}^2$ is the average prediction error variance for all pairwise comparisons of the lines, and ${\widehat{\sigma }}_G^2$ is the estimated genetic variance (Cullis et al., ²⁰⁰⁶).

These phenotype data have been reported in a previous study by using a much smaller genotypic dataset of 858 molecular markers (Wisser et al., ²⁰¹¹). Here, we conducted additional analyses on this phenotypic data, including correlation and phenotypic distribution analyses (Table S1).

Genome‐wide association analysis

Genotypic data for the diversity panel were derived from two different datasets including 57 838 single nucleotide polymorphisms (SNPs) using Illumina MaizeSNP50 BeadChip and 681 257 GBS SNP markers. SNPs with a minor allele frequency (MAF) value < 0.05 and > 20% missing data across the 270 lines were removed, leaving a total of 246 497 SNP markers for the GWAS (Table S3). This dataset has been used for several other traits except SLB (Olukolu et al., ^{2016a, 2016b}; Xie et al., ²⁰¹⁸).

GWAS was conducted using BLUPs as input for each line. We used the efficient mixed linear model (MLM) as implemented in Tassel v4.1.8 for association analysis with 246 497 SNPs with MAF ≥ 0.02. A kinship matrix (K) was included in the analysis to minimize confounding effects of population structure. The K matrix was previously calculated in the diversity panel using a subset of 4000 SNPs uniformly distributed across the entire genome without missing data. The model is as follows:

$$Y=\mathit{X\beta }+\mathrm{Zu}+e$$

where Y is the vector of best linear unbiased predictor values (BLUPs) of SLB; β is the vector of fixed effects including SNP marker effects; u is the vector of random additive genetic background effects for lines, whereas X and Z are design matrices, and e is a vector of random residuals. The variances of the u vector are modeled as Var(u) = 2KV _g, where K is the kinship matrix, and V _g is genotypic variance.

The P values for all the marker association tests between one SNP and SLB resistance within the analysis were used to estimate the corresponding positive false discovery rate (FDR) using QVALUE v1.0 in R. Based on this analysis, significant SNPs were declared at Q value <0.2.

Data resampling analysis

In order to investigate the robustness of SNPs detected in the full diversity panel, a second GWAS was performed based on a subsampling approach (Valdar et al., ²⁰⁰⁹). We generated 1000 subsample datasets, with each subsample containing a random sample of 80% of the inbred lines. GWAS was conducted on each of the 1000 subsamples using the same model as the full dataset. A significance level of P < 0.0001 was used for each analysis. The resample model inclusion probabilities (RMIP) were calculated for each variant as the frequency across 1000 data samples with which the variant association test had a P value of <0.0001. SNPs with Q value <0.2 and RMIP ≥0.2 were retained.

Candidate gene identification

The candidate genes located within or adjacent to the significant SNPs (Q value <0.2; RMIP ≥0.2) were identified from the maize B73 filtered gene set (AGP_v5) using the MaizeGDB genome browser.

Phylogenetic analysis

The amino acid sequences of reported immune‐related Raf MAPKKKs were downloaded from the National Center for Biotechnology Information (NCBI) database (https://www.ncbi.nlm.nih.gov/). The ZmMAPKKKs amino acid sequences were retrieved from the Gramene database (http://www.gramene.org/). The full‐length sequences were used for the phylogenetic analysis, and the phylogenetic trees were constructed using the neighbor‐joining method in MEGA 7.0 (http://www.megasoftware.net). Bootstrap values from 1000 pseudo‐replicates were used to provide support for the nodes in the phylogenetic tree. The conserved domains were predicted using the NCBI Conserved Domains Database (https://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi).

Subcellular localization

The full‐length coding sequence of ZmMAPKKK45 was cloned into the Gateway pDONR207 vector through BP reactions. Subsequently, the confirmed clone was transferred to pDEST‐EGFP vector via LR recombination, generating the fusion construct pDEST–ZmMAPKKK45‐EGFP. The construction was transiently expressed in Nicotiana benthamiana leaves by Agrobacterium‐mediated infiltration. EGFP fluorescence was visualized under a confocal microscope (Zeiss LSM 880).

Reverse transcription‐quantitative polymerase chain reaction

Total RNA was extracted using an RNeasy Plant Mini Kit (QIAGEN, 74904) according to the manufacturer's protocol. First‐strand cDNA was synthesized from 1 μg total RNA using the QuantiTect Rev. Transcription Kit (QIAGEN, 205311). The maize housekeeping gene ZmGAPDH served as an internal control. The 2^−∆∆CT method was employed to calculate relative transcript levels (Schmittgen & Livak, 2008). To quantify fungal biomass, the relative transcript level between ChACT1 and ZmGAPDH was calculated with the 2^−∆∆CT method. For each sample, quantitative polymerase chain reaction was carried out in three biological replicates, with three technical replicates for each biological replicate, using a CFX96 Bio‐Rad real time PCR machine. Each sample contained at least three individual plants. Primers used are detailed in Table S2.

ROS burst detection

ROS production was measured using a luminol‐based assay as described previously (Wang et al., ²⁰²¹). Leaf discs were excised from mutant and transgenic plants at the two‐leaf stage using a 4‐mm diameter punch and incubated in 100 μl water overnight. At least eight plants per line were sampled. Next, water was replaced by 2× L‐012 (FUJIFILM Wako Pure Chemical Corporation, Richmond, VA, USA, Cat # 120‐04891) in 0.05% Silwet L‐77 (Plantmedia, Dublin, OH, USA, Cat # 30630216‐1; 15 mg L‐012 in 1 ml 200 mM KOH is 100× L‐012). Then two kinds of 2× Horseradish peroxidase (HRP) buffer (Sigma‐Aldrich; 25 mg HRP in 2.5 ml is 100× HRP) were prepared. One was used for PAMP treatment with 2.5 mM flg22 (GenScript, Cat # RP19986) or 40 μg ml⁻¹ chitohexaose (Accurate Chemical and Scientific Corporation, Cat # BCR56/11); the other was used for mock treatment. The luminescence signal was monitored immediately after the treatment and continuously recorded every 60 s for 60 min with a PHERAstar FSX (BMG LABTECH) plate reader. The data were analyzed to display the changes in ROS production over time.

MAPK assay

For the analysis of MAPK activation induced by C. heterostrophus, maize plants were inoculated using the same method as described above for the gene expression and DAB staining assays. To assess MAPK activation in response to chitohexaose, the fully expanded fourth leaves were harvested, and ~10 cm segments from the middle portion of each leaf were collected. The leaf segments were treated with 2 μg ml⁻¹ chitohexaose for designated time intervals and immediately frozen in liquid nitrogen. To facilitate the penetration of chitohexaose into maize leaf tissue, 0.02% Silwet L‐77 was added to the treatment solution.

Plant total protein was extracted from maize leaves. The tissue sample was ground in liquid nitrogen and resuspended in an equal volume (1 : 1 fresh weight/volume) of the extraction buffer (25 mM Tris–HCl, 10 mM NaCl, 10 mM MgCl₂, 5 mM DTT, 10 mM PMSF; pH 7.5) on ice for at least 30 min, and then centrifuged at 12 000  g for 30 min at 4°C to obtain the supernatant. The phosphorylation of MAPK proteins was detected by western blot with anti‐phospho‐p44/42 MAPK antibody (Cell Signaling Technology, Cat # 4370T). Endogenous actin was used as a loading control and detected with an anti‐actin polyclonal antibody (Abcam, ab197345). Gel loading was further verified by Ponceau S staining. Western blots were performed as previously described (Huang & Rojas‐Pierce, 2024).

DAB staining

The histochemical analysis of H₂O₂ accumulation was performed using DAB staining as described previously with slight modifications (Daudi & O'Brien, 2012). The leaf samples were vacuum‐infiltrated with DAB staining buffer (1 mg ml⁻¹) for 2 h and then incubated at 30°C in darkness for 12–16 h. Following the incubation, replace the DAB staining solution with bleaching solution (ethanol:acetic acid:glycerol = 3:1:1) and place it in a boiling water bath for 20 min. After 20 min of boiling, replace the bleaching solution with fresh bleaching solution and allow it to stand for 30 min. Then, the leaves can be directly visualized for DAB staining. The ImageJ v1.8.0 was used to analyze the mean gray value of the staining leaves.

Statistical analysis

P‐values and sample sizes (n) are indicated in individual figures and figure legends. Statistical analyses of disease phenotypes, gene expression levels and ROS experiments were performed by IBM SPSS Statistics SV26. Differences between two groups were analyzed by a two‐sided Student's t‐test. Statistical significance of differences between more than two groups was analyzed based on one‐way ANOVA with Tukey's test or Fisher's least significant difference test; means identified with a common lowercase letter were not significantly different (P > 0.05).

Results

Phenotypic variation and heritability

The maize diversity panel was evaluated for SLB resistance across five environments with two replications in each location except NC04 with one replication (Table S1). The pairwise Pearson's correlations of SLB AUDPC BLUPs were highly significant between all pairs of environments, ranging from 0.62 to 0.95 with P < 0.00001 (Table 1). The SLB AUDPC BLUP values for all five environments and individual environments followed an approximately normal distribution (Figs 1a, S1). Significant (P < 0.0001) genetic variation for SLB was observed (Table 2) and the entry mean heritability estimate across all five environments was high (0.95, Table 2). The BLUPs for SLB resistance among the 270 entries of the maize diversity panel ranged from 3.47 to 7.70 with an average of 5.85 (Table S1) on the 1 (susceptible) to 9 (resistant) scale used.

Table 1.

Pearson's correlation coefficients between area under disease progress curve (AUDPC), best linear unbiased predictions (BLUPs) for Southern leaf blight (SLB) in the five environments tested.

Location	NC04	NC05_BLUP	NC06_BLUP	GA06_BLUP	SLB_BLUPs
FL03_BLUP	0.70*****	0.74*****	0.78*****	0.62*****	0.88*****
NC04		0.86*****	0.90*****	0.85*****	0.91*****
NC05_BLUP			0.94*****	0.81*****	0.93*****
NC06_BLUP				0.84*****	0.95*****
GA06_BLUP					0.84*****

Significance of correlation coefficients (r) is shown as *****P < 0.00001. FL: 27 Farms, Inc., Homestead, FL; NC: North Carolina State University's Central Crops Research Station, Clayton, NC; GA: University of Georgia‐Tifton's Gibb's Farm, Tifton, GA. AUDPC, area under the disease progress curve; BLUP, best linear unbiased prediction.

Fig. 1.

SLB genome‐wide association and data resampling analyses. (a) Distribution of southern leaf blight (SLB) best linear unbiased prediction (BLUP) values in the maize diversity panel. (b) Manhattan plot for genome‐wide association study (GWAS) of SLB BLUPs in the maize diversity panel using days to anthesis as a covariate. Significance threshold at FDR = 0.20 (−log₁₀(P) = 4.01) is indicated. (c) Quantile–Quantile (Q–Q) plots from GWAS results for SLB BLUPs in the maize diversity panel. Expected null distribution of P‐value assuming no associations, represented as solid red line; distribution of observed P‐values were indicated by the black scatter. (d) Data resampling analysis. GWAS results from 1000 random subsamples, sampling 80% of the 270 lines of the diversity panel each time. The vertical axis indicates the resample model inclusion probability (RMIP), whereas the horizontal axis indicates chromosomes and physical map positions of SNPs. The dashed line depicts the RMIP threshold of 0.20. SNPs with a P‐value < 0.0001 from the GWAS were used to calculate RMIP. PPR, pentatricopeptide repeat.

Table 2.

Summary statistics and heritability estimate for Southern leaf blight (SLB) resistance in the maize diversity panel.

Parameter	Variance component estimate a	P‐value
Environment	0.57 ± 0.41	NS
Replication within environment	0.01 ± 0.00	NS
Inbred line	1.02 ± 0.09	< 0.0001
Environment by line	0.15 ± 0.09	< 0.0001
Residual	0.15 ± 0.01	< 0.0001
Heritability	0.95

NS, not significant.

^a Estimate of variance and its SE.

Genome‐wide association study

We carried out GWAS using BLUP values and 246 497 high‐quality SNPs at an average density of 1.1 SNPs per 10 kb. An MLM, which took genome‐wide patterns of genetic relatedness into account, was used to identify association signals. A Manhattan plot shows positive associations were identified from the panel (Fig. 1b). The distribution of P‐values quantile–quantile plot (Fig. 1c) is consistent with the highly associated SNPs being caused by bona‐fide genetic effects.

At a 20% FDR (Q value <0.20), we identified 22 significant SNPs associated with SLB resistance (Fig. 1b; Table S4). The phenotypic variation (R ²) explained by each locus ranged from 5.92% to 10.19% (Table S4). RMIP values were calculated by conducting GWAS analyses with a set of 1000 random data samples of 80% of the diversity panel with P < 0.0001. A total of 16 variants were identified above the RMIP threshold of 0.2 (Fig. 1d; Table S5).

To ensure that the identified SNPs had robust associations with SLB variation, we only considered the 14 SNPs that cleared the FDR and RMIP thresholds (Q value <0.20, RMIP ≥0.20, Table 3). Their allelic effects ranged from −0.82 to 1.02 on our 1–9 scoring scale and their minor allele frequencies ranged from 5% to 47% (Table 3). All 14 SNPs were located within promoter regions or coding sequences of predicted genes from the filtered predicted gene set. Three SNPs on chromosome 3 c. 16.1 Mbp were mapped to a single gene. Additionally, two SNPs near 16.3 Mbp and two other SNPs near 215.3 Mbp on chromosome 3 were each mapped to a common gene, respectively. In total, 10 candidate genes underlying 14 variants were identified from the B73 reference genome (AGP_v5) (Table 3).

Table 3.

Chromosome physical positions, allele effect estimates, closest candidate genes and other summary statistics for the 14 single nucleotide polymorphisms (SNPs) significantly associated with SLB resistance from the maize diversity panel association analysis.

SNP	Chr a	SNP position (bp) (AGP_v5)	P‐value	Q value	RMIP b	R 2 (%) c	Allele d	Minor allele frequency (%)	Allele effect e	Genic position f	Gene ID	Annotation
1	2	205 203 908	6.6 × 10−6	0.049	0.375	7.89	A/G	43	−0.479	Exonic	Zm00001eb104750	C2H2 zinc finger protein
2	3	16 101 604	3.9 × 10−6	0.049	0.799	9.89	G/T	42	−0.529	Intronic	Zm00001eb123970	Pentatricopeptide repeat (PPR) superfamily protein
3	3	16 105 163	3.2 × 10−6	0.049	0.846	9.87	C/T	42	−0.527	Exonic
4	3	16 105 185	1.3 × 10−5	0.059	0.566	8.72	G/T	40	−0.489	Exonic
5	3	16 313 658	8.3 × 10−6	0.049	0.695	7.98	G/A	38	−0.519	Promoter	Zm00001eb124020	Aspartyl‐tRNA synthetase
6	3	16 314 281	6.4 × 10−6	0.049	0.709	9.35	A/C	40	−0.501	Intronic
7	3	32 130 373	5.5 × 10−5	0.166	0.295	7.56	T/G	37	−0.548	Exonic	Zm00001eb126930	C‐type lectin receptor
8	3	214 117 471	2.1 × 10−6	0.049	0.84	10.19	A/G	21	0.556	Promoter	Zm00001eb156690	Cytochrome b561
9	3	215 308 542	5.6 × 10−5	0.166	0.315	8.33	A/G	14	0.849	Intronic	Zm00001eb157070	Protein kinase superfamily protein/ZmMAPKKK45
10	3	215 308 565	5.6 × 10−5	0.166	0.315	8.33	T/C	14	0.849	Intronic
11	5	57 903 372	8.3 × 10−6	0.049	0.481	7.89	A/G	43	0.498	Intronic	Zm00001eb226800	Pentatricopeptide repeat (PPR) superfamily protein
12	5	196 501 936	9.7 × 10−5	0.194	0.202	5.92	T/A	10	0.653	Exonic	Zm00001eb249600	Expansin precursor
13	6	9730 130	7.7 × 10−5	0.167	0.238	7.34	T/C	5	1.023	Intronic	Zm00001eb260870	RNA recognition motif containing protein
14	8	138 057 719	6.1 × 10−5	0.167	0.265	6.32	T/C	7	−0.816	Intron	Zm00001eb356010	Nodulin MtN21/EamA‐like transporter

^a Chromosome.

^b Resampling model inclusion probability (RMIP).

^c R ², proportion of phenotypic variance explained by SNP after accounting for background polygenic variance.

^d Allele is reported as: ‘minor allele/major allele’.

^e Allele effect is reported as effect in reference to the minor allele. A positive effect implies increasing resistance conferred by the major allele.

^f The position of the associated SNP relative to the gene. The promoter is defined as 2 kb upstream of the start codon. All associated SNPs were within a transcribed gene or in a promoter region.

Colocalization of associated SNPs with previously reported SLB QRL in maize

Most of the 14 SNPs identified in this study were colocalized with SLB QRLs reported in previous studies (Table S6). SNPs 2–6 and SNP 7, located on chromosome 3 at c. 16 Mbp and c. 32 Mbp (AGP_v5), fell within a major QRL region identified previously in seven different populations, respectively (Balint‐Kurti et al., ^{2006, 2007, 2008}; Zwonitzer et al., ^{2009, 2010}; Bian et al., ²⁰¹⁴; Lennon et al., ²⁰¹⁷; Kaur et al., ²⁰¹⁹; Lopez‐Zuniga et al., ²⁰¹⁹). SNP 14, located on chromosome 8 at c. 138 Mbp, was situated within an SLB QRL region detected in five populations (Balint‐Kurti et al., ^{2006, 2007, 2008}; Zwonitzer et al., ²⁰¹⁰; Lennon et al., ²⁰¹⁷). Additionally, SNP 1 (chromosome 2, 205.2 Mbp) and SNPs 8–10 (chromosome 3, 214.1–215.3 Mbp) overlapped with QRLs identified in two populations each (Balint‐Kurti et al., ^{2006, 2008}; Kaur et al., ²⁰¹⁹). Notably, SNP 13 on chromosome 6 did not colocalize with any previously reported SLB QRL (Table S6).

Molecular characterization of ZmMAPKKK45

SNPs 9 and 10 were located within the intronic region of Zm00001eb157070 on chromosome 3 and defined two haplotypes in the maize diversity panel (Fig. S2a). Statistically, lines carrying haplotype 1 exhibited significantly higher SLB resistance compared to those carrying haplotype 2 (Fig. S2b). Zm00001eb157070 encodes a predicted Raf subfamily MAPKKK protein named ZmMAPKKK45 (Kong et al., ²⁰¹³), containing a STKc_MAP3K‐like domain (Figs 2a, S3a). Phylogenetic analysis was conducted using the amino acid sequences of 46 previously identified Raf subfamily ZmMAPKKKs proteins from B73 (Table S7) (Kong et al., ²⁰¹³), in combination with Raf MAPKKKs from Arabidopsis thaliana, rice (Oryza sativa) and cotton (Gossypium hirsutum) that have reported functions in plant immunity (including EDR1 (Tang et al., ²⁰⁰⁵) and MKD1 (Asano et al., ²⁰¹⁹) from Arabidopsis, OsEDR1 (Shen et al., ²⁰¹¹) and OsILA1 (Chen et al., ²⁰²¹) from rice and GhMAPKKK65 (Zhai et al., ²⁰¹⁷) from cotton). The resulting phylogeny showed that the proteins cluster into four distinct clades (Fig. S3b). ZmMAPKKK45 shares the highest sequence similarity with ZmMAPKKK44 and resides in the same clade as EDR1 and OsEDR1 (Fig. S3b). However, certain other maize MAPKKK family members exhibit even greater sequence similarity to EDR1 and OsEDR1 (Fig. S3b). Analysis of publicly available transcriptome data (Plant Public RNA‐seq Database, https://plantrnadb.com/) indicates that ZmMAPKKK45 is broadly expressed across maize tissues (Fig. S4a). Notably, the transcript levels of ZmMAPKKK45 were significantly induced in maize leaves upon SLB infection, whereas the expression of the closest homolog gene ZmMAPKKK44 was not (Fig. S4b). When ZmMAPKKK45 was tagged with EGFP at the C‐terminus and transiently expressed in N. benthamiana leaves, we observed both nuclear and cytoplasmic localization (Fig. S4c).

Fig. 2.

Functional validation of the role of ZmMAPKKK45 in southern leaf blight (SLB) resistance in maize. (a) Schematic illustration of the Mu element (mu1047777) insertion site in the 5′‐UTR of ZmMAPKKK45 in the maize UniformMu line UFMu‐06499, designated as W22^MAP3KMu. The inverted triangle marks the position of the Mutator transposon insertion. Horizontal arrows indicate the locations of the primers used for mutant genotyping, and vertical arrows denote the positions of associated SNPs. (b) The relative transcript level of ZmMAPKKK45 in W22 and the W22^MAP3KMu mutant. This experiment was repeated three times, and the results combined. (c, d) SLB symptoms (c) and area under the disease progress curve (AUDPC) values (d) of W22 and the W22^MAP3KMu mutant grown in the field. The numbers in the bar chart represent the sample size. AUDPC, area under the disease progress curve. SLB severity was rated on a 1 (susceptible) to 9 (resistant) scale. Bar, 10 cm. This experiment was repeated twice in the field in two different years, yielding similar results. (e–g) The symptoms (e), relative fungal biomass (f) and percent lesion area (g) of W22 and W22^MAP3KMu mutant in maize seedlings grown under controlled environment conditions. Bar, 2 cm. This inoculation experiment was repeated twice, and the phenotypic results were consistent. (h) The SLB phenotype of F₄ plants from F_3:4 families derived from a W22 × W22^MAP3KMu cross, grouped by family. The genotype with respect to the Mu element mu1047777 of the parent F₃ plant for each F_3:4 family is indicated: W, wild‐type (Mu element absent); H, heterozygote; M, Mu element homozygous. The numbers in the bar chart represent the sample size. (i) The data shown in (h) were categorized by genotype rather than by family. The numbers in the bar chart represent the sample size. Data in (b), (d), (f) and (h) show as means ± SD. Data in (g) and (i) displayed as box and whisker plots with individual data points. The horizontal line denotes the median, the box shows the interquartile range (IQR), and the whiskers extend to 1.5 × IQR. In (b), (d), (f) and (g), statistical significance was determined by a two‐sided Student's t‐test. In (i), different lowercase letters indicate a significant difference (P < 0.05) based on one‐way ANOVA with Fisher's least significant difference test.

ZmMAPKKK45 is an SLB resistance gene

The maize UniformMu resource is a collection of lines carrying sequence‐indexed insertions of the Mutator (Mu) transposon throughout the genome in a background of the W22 inbred line (Settles et al., ²⁰⁰⁷). We identified a maize Mu insertion line, UFMu‐06499, carrying a Mu insertion (mu1047777) in the 5′‐UTR region of ZmMAPKKK45 (Fig. 2a). We self‐pollinated this line to develop a line homozygous for mu1047777 which we called W22^MAP3KMu. In this line, the transcript level of ZmMAPKKK45 was 3.5‐fold lower than that in the wild‐type W22 (Fig. 2b). In the field, W22^MAP3KMu showed significantly lower resistance to SLB compared with W22 (Fig. 2c,d). At the seedling stage, W22^MAP3KMu exhibited increased SLB susceptibility, with higher fungal biomass and larger lesion area than the wild‐type (Fig. 2e–g).

Since the original maize Mu insertion line, UFMu‐06499, carried 11 other known insertions and since other undetected genomic modifications might be present in UFMu‐06499, we could not be sure from these data alone that the observed reduction in resistance was due to the presence of mu1047777 or to one of these other genomic modifications. To address this possibility, we generated 17 F_3:4 families from the crosses between W22^MAP3KMu and W22, among which all the genomic modifications (assuming no genetic linkage) would be expected to segregate independently and tested their SLB resistance in the field and growth chamber. In growth chamber assays, 33 individual F₄ plants, derived from the five F_3:4 families homozygous for the mu1047777 insertion at ZmMAPKKK45 were, on average, about twice as susceptible to SLB as the 16 individual F₄ plants derived from three F_3:4 families that were wild‐type at this locus (c. 80% necrotic tissue compared to c. 40% necrotic tissue, Fig. 2h,i). Similarly, in the field, in two independent replicates in 2024, the homozygous mutant was significantly more susceptible to SLB than the homozygous wild‐type (Fig. S5a–d). The same results were observed in 2025 (Fig. S5e). These results indicate that the mu1047777 insertion cosegregates with the SLB phenotype, and that the more susceptible phenotype of the UFMu‐06499 homozygous plants is caused by the transposon insertion at ZmMAPKKK45. Furthermore, in both field and growth chamber assays, the distribution of SLB resistance among plants from F_3:4 families derived from F₃ plants heterozygous for the mu1047777 insertion was intermediate between the distributions of wild‐type and mutants (Figs 2i, S5b,d), suggesting that the SLB resistance conferred by ZmMAPKKK45 was partially dominant.

We cloned full‐length cDNA of ZmMAPKKK45 from W22 and transformed it into the SLB‐susceptible maize inbred line B104 using an expression cassette driven by the maize ubiquitin promoter (Fig. 3a). We generated two independent transgenic events in B104 (A702B1, A702B6), backcrossed these events to B104 for three generations to produce T₁BC₃ and selfed one time to generate T₁BC₃F₂ populations. Transgenic plants carrying the A702B1 and A702B6 events displayed c. 1.5‐fold and c. 12.5‐fold higher ZmMAPKKK45 transcript levels, respectively (Fig. 3b). In field tests with segregating families, plants carrying the A702B6 event exhibited significantly higher SLB resistance compared to their nontransgenic siblings (Fig. 3c–e). In families segregating for the A702B1 event (which showed lower levels of ZmMAPKKK45 overexpression (OE)) transgenic plants were also more resistant to SLB on average, although the difference between them and their nontransgenic siblings was not statistically significant (Fig. 3b–e). To test whether ZmMAPKKK45 OE could complement the SLB‐susceptible phenotype of the mu1047777 mutant, we crossed the heterozygous OE events A702B1 and A702B6, as well as the recipient line B104, with homozygous mu1047777. Field inoculation of the resulting F₁ progeny showed that plants harboring the A702B6 event exhibited significantly enhanced resistance to SLB compared to their nontransgenic siblings (Fig. S6). Taken together, these results indicate that ZmMAPKKK45 confers resistance to SLB, and its expression level is positively associated with SLB resistance.

Fig. 3.

Transgenic functional validation of ZmMAPKKK45 in B104 maize inbred background. (a) Schematic diagram of the overexpression construct pMCG1005‐ZmMAPKKK45. The ZmMAPKKK45 coding sequence was cloned from W22 cDNA. Horizontal arrows indicate the positions of the two primer pairs used for transgenic line identification. (b) The relative transcript levels of ZmMAPKKK45 in the overexpressing transgenic lines A702B1 and A702B6 and nontransgenic siblings. The quantitative reverse transcription‐polymerase chain reaction experiment was conducted twice with consistent results, and the results from the second time are presented here. (c–e) Southern leaf blight (SLB) symptoms (c) and disease scores (1 = susceptible, 9 = immune) in segregating T₁BC₃ (d) and T₁BC₃F₂ (e) transgenic plants evaluated in the field, evaluated in 2018 and 2024, respectively. Due to rapid SLB progression in the B104 background, only one scoring dataset is presented (shown as ‘SLB scale’). NT, nontransgenic plant; T, transgenic plant. The numbers in the bar chart represent the sample size. Bar, 10 cm. In (b), (d) and (e), data are presented as means ± SD and the statistical significance was determined by a two‐sided Student's t‐test.

ZmMAPKKK45 does not appear to function in canonical MAPK cascades

MAPK cascades are involved in signaling multiple defense responses, including ROS generation, defense gene activation and other defense responses (Meng & Zhang, 2013). ZmMAPKKK45 may function as a component of the MAPK cascades. Since MAPK3/6 are terminal kinases in the cascades that play key roles in immune signaling and are rapidly activated in response to diverse biotic and abiotic stresses, their activation serves as a reliable indicator of MAPK cascades activation and can therefore be used to assess whether ZmMAPKKK45 functions within this pathway (Zhang & Zhang, 2022). To assess this, we treated both ZmMAPKKK45 OE line A702B6 and its nontransgenic sibling, as well as W22 and the W22^MAP3KMu mutant line, with C. heterostrophus and collected samples 0, 12, 24 and 36 h post‐inoculation (hpi) to analyze the phosphorylation status of MAPK3/6. In all genotypes, C. heterostrophus infection led to increased activation of MAPK3/6; however, neither OE nor mutation of ZmMAPKKK45 altered the phosphorylation levels of MAPK3/6 (Fig. 4a,b).

Fig. 4.

ZmMAPKKK45 is not involved in MAPK3/6 activation or transcriptional regulation of defense‐related genes in maize. (a, b) Inoculation with Cochliobolus heterostrophus enhances MAPK3/6 activity in ZmMAPKKK45 overexpression A702B6 line and its nontransgenic sibling (a), as well as in W22 and W22^MAP3KMu (b). hpi, hours post‐inoculation. Three leaves were pooled for each time point. MAPK3/6 activations were analyzed by immunoblot with an Phospho‐p44/42 MAPK (Erk1/2) antibody, and equal loading is shown by α‐actin immunoblot and ponceau S staining. (c, d) Transcript levels of salicylic acid responsive resistance genes (ZmNPR1, ZmPR1 and ZmPR5) (c) and jasmonate responsive genes (ZmAOC, ZmAOS and ZmLOX1) (d) in the fourth leaves of ZmMAPKKK45 overexpression lines and their nontransgenic siblings at the seedling stage. (e) The transcript levels of SA‐responsive genes ZmNPR1, ZmPR1 and ZmPR5 48 h in the fourth leaves after inoculation or mock inoculation of W22 and the W22^MAP3KMu plants. Plants inoculated with the SLB pathogen (C. heterostrophus) are denoted as ‘C.h’ and mock‐infected plants are denoted as ‘Mock’. (f) The transcript levels of JA‐responsive genes ZmAOC, ZmAOS and ZmLOX1 48 h in the fourth leaves after inoculation or mock inoculation of W22 and the W22^MAP3KMu plants. In (c–f), data are presented as mean ± SD. In (c) and (d), statistical significance was determined by a two‐sided Student's t‐test. In (e) and (f), different lowercase letters indicate a significant difference (P < 0.05) based on one‐way ANOVA with Tukey's test. All these experiments were conducted twice with similar results; one representative result is shown.

To further investigate whether ZmMAPKKK45 is involved in canonical MAPK cascades, we examined the transcript levels of several defense‐related genes known to be downstream of the MAPK cascades (Balmer et al., ²⁰¹³), including genes responsive to salicylic acid (SA‐ ZmNPR1, ZmPR‐1 and ZmPR5) and jasmonate (JA‐ ZmAOC, ZmAOS and ZmLOX1), in ZmMAPKKK45 OE lines and the W22^MAP3KMu mutant line. The transcript levels of the SA and JA responsive genes in uninfected OE lines were similar to those in nontransgenic plants, except for ZmPR5, whose transcript level was significantly increased in the OE lines (Fig. 4c,d). The expression levels of all the SA and JA responsive genes were also not significantly different between W22 and W22^MAP3KMu (Fig. 4e,f). Forty‐eight hours after C. heterostrophus inoculation, the transcript levels of all these genes were upregulated in maize seedlings (Fig. 4e,f). However, there were no significant differences in transcript levels between infected W22 and infected W22^MAP3KMu (Fig. 4e,f). In addition, we examined the expression patterns of ZmPR1 and ZmLOX1 at multiple time points following inoculation in W22 and W22^MAP3KMu. The expression of ZmPR1 was continuously induced by C. heterostrophus, whereas ZmLOX1 was rapidly upregulated, peaking at 12 hpi, and gradually declining thereafter. Apart from the difference in ZmLOX1 expression at 24 hpi, the transcript levels of both genes were largely comparable between W22 and W22^MAP3KMu (Fig. S7). Collectively, these results suggest that ZmMAPKKK45 is unlikely to function through the classical MAPK cascade pathway.

ZmMAPKKK45 regulates ROS accumulation

ROS are important secondary messengers that play a key role in plant immunity and disease resistance by mediating downstream immune responses (Wu et al., ²⁰²³). We characterized whether ZmMAPKKK45 affects ROS production mediated by two commonly used elicitors: flg22, an epitope derived from bacterial flagellin, and chitohexaose, a chitin‐derived hexamer, using a luminol‐based assay (Felix et al., ¹⁹⁹⁹; Samira et al., ²⁰¹⁹). Flg22‐induced ROS accumulation was significantly higher in transgenic OE plants than in nontransgenic plants (Fig. S8a). Despite repeated efforts, we were not able to detect any flg22‐mediated response in W22 or W22^MAP3KMu using the luminol assay (Fig. S8b). Sequence and conserved domain analyses revealed that the W22 allele of the presumed flg22 receptor ZmFLS2^W22 lacks a leucine‐rich repeat N‐terminal domain compared to AtFLS2 and ZmFLS2^B104, which may explain why flg22 fails to trigger a ROS burst in the W22 background (Fig. S8c). By contrast, chitohexaose treatment induced ROS bursts in both transgenic OE plants and mutants (Fig. 5a,b). Compared to nontransgenic plants, ROS accumulation was significantly enhanced in the transgenic OE lines (Fig. 5a), whereas the mutant, relative to the W22, exhibited markedly reduced ROS levels (Fig. 5b). However, despite the ability of chitohexaose to activate MAPK3/6, we detected no difference in MAPK3/6 activity between W22 and W22^MAP3KMu at different time points, further indicating that ZmMAPKKK45 does not function through the canonical MAPK cascades pathway (Fig. 5c).

Fig. 5.

ZmMAPKKK45 regulates reactive oxygen species (ROS) accumulation in maize. (a, b) Chitohexaose‐induced ROS production in the third leaves of the ZmMAPKKK45 overexpression line A702B6 and its non‐transgenic sibling (a), as well as in W22 and W22^MAP3KMu (b) at the four‐leaf stage. ROS production was measured as relative light units (RLU) in a luminol‐based assay. This experiment was repeated twice with similar results. (c) Chitohexaose treatment enhances MAPK3/6 activity in W22 and W22^MAP3KMu. Three leaves were pooled for each time point. MAPK3/6 activations were analyzed by immunoblot with a phospho‐p44/42 MAPK (Erk1/2) antibody, and equal loading is shown by α‐actin immunoblot and ponceau S staining. This experiment was repeated two times, yielding consistent results. (d) The 3,3′‐diaminobenzidine (DAB) staining shows H₂O₂ accumulation in W22 and W22^MAP3KMu plants with or without pathogen treatment at four‐leaf stage. Histochemical detection of H₂O₂ by DAB staining at 24 h post‐inoculation (hpi). The staining results are shown 24 h after treatment with either infection buffer (Mock) or infection with Cochliobolus heterostrophus. Bar, 1 cm. (e) The transcript levels of ZmRBOHs in ZmMAPKKK45 overexpression lines compared to nontransgenic siblings. The third leaf of four‐leaf stage seedlings was analyzed. This experiment was performed twice and similar results were obtained. (f) The transcript levels of ZmRBOHs after inoculation or mock inoculation of W22 and the W22^MAP3KMu plants. The third leaf of four‐leaf stage seedlings was analyzed. This experiment was performed twice with similar results. In (a) and (b), data are presented as mean ± SE. In (e) and (f), data are shown as mean ± SD. In (e), the statistical significance was determined by a two‐sided Student's t‐test. In (f), different lowercase letters indicate a significant difference (P < 0.05) based on one‐way ANOVA with Tukey's test.

In these lines, we also visualized H₂O₂ accumulation differences using DAB staining after C. heterostrophus treatment. We observed that the leaves of W22 accumulated higher levels of H₂O₂ than W22^MAP3KMu at 16 and 24 hpi (Figs 5d, S9). Similarly, plants carrying the A702B6 ZmMAPKKK45 OE event, accumulated more H₂O₂ than their nontransgenic siblings at 24 hpi (Fig. S10). These data show that ZmMAPKKK45 regulates the production of ROS, and we suggest that this may be related to its function in conferring higher levels of SLB resistance.

In N. benthamiana, MEK2 has been shown to activate the ROS burst via interaction with WRKY transcription factors that bind to the promoter of RBOHB, and which, presumably, modulate its expression (Adachi et al., ²⁰¹⁵). We hypothesized therefore that the effect of ZmMAPKKK45 on ROS production might also be mediated by its effect on the expression levels of ZmRBOHs. The maize genome contains six ZmRBOHs (ZmRBOH1‐6) (Zhong et al., ²⁰²⁴). Publicly available transcriptome data show that SLB treatment induces the expression of ZmRBOH1, ZmRBOH3 and ZmRBOH4 (Fig. S11a). Since ZmRBOH5 and ZmRBOH6 are expressed at very low levels and are primarily expressed in anthers (Fig. S11), we focused on examining the expression of ZmRBOH1‐4. In families segregating for the ZmMAPKKK45 OE construct, significant increases in the transcript levels of ZmRBOH1, ZmRBOH2 and ZmRBOH4 are detected in transgenic A702B6 plants relative to their nontransgenic siblings, whereas A702B1 plants did not show significant changes in the expression of these genes (Fig. 5e). In the corresponding experiments, the basal transcript level of ZmRBOH1 was slightly higher in W22 than in W22^MAP3KMu (Fig. 5f). Transcript levels of ZmRBOH3 and ZmRBOH4 could be increased after C. heterostrophus inoculation, and the induction levels were higher in W22 than in W22^MAP3KMu (Fig. 5f). While these increases in expression are not dramatic, these results are consistent with the hypothesis that ZmMAPKKK45 might modulate ROS accumulation, in part at least, by regulating the expression of ZmRBOHs, ultimately affecting resistance to SLB.

ZmMAPKKK45 contributes resistance against multiple foliar fungal diseases

Since ROS accumulation is a general resistance component, we were interested in whether ZmMAPKKK45 could confer resistance to other foliar diseases. We characterized NLB and GLS resistance in field trials. For these trials, we assessed W22 and W22^MAP3KMu along with the same F_3:4 families described above that were segregating for mu1047777 at ZmMAPKKK45. W22^MAP3KMu showed significantly decreased resistance to both NLB and GLS compared with W22 (Fig. 6) and the F_3:4 families homozygous for the mu1047777 insertion at ZmMAPKKK45 were also more susceptible to NLB and GLS than F_3:4 family lines homozygous for the absence of the insertion (Figs S12, S13). The NLB resistance level of the mu1047777 heterozygote was intermediate between that of the wild‐type and the mutant (Fig. 6c). However, the heterozygote exhibited a similar GLS resistance to the wild‐type, both of which were significantly more resistant than the mutant (Fig. 6f). Considering these findings along with the previous SLB results, we concluded that ZmMAPKKK45 provides quantitative resistance to multiple foliar fungal pathogens.

Fig. 6.

ZmMAPKKK45 affects maize resistance to northern leaf blight (NLB) and gray leaf spot (GLS) in the field. (a, b) The NLB area under the disease progress curve (AUDPC) values (a) and symptoms (b) of W22 and W22^MAP3KMu mutant. The numbers in the bar chart represent the sample size. 1, susceptible‐9, resistant. (c) NLB resistance phenotype of plants from F_3:4 families derived from a W22 × W22^MAP3KMu cross. The genotype with respect to the Mu element mu1047777 of the parent F₃ plant for each plant is indicated: H, heterozygote; M, Mu element homozygous; W, wild‐type (Mu element absent). Each dot represents a single plant. The numbers below the scatter plot represent the sample size. (d, e) The GLS scale values (d) and symptoms (e) of W22 and W22^MAP3KMu mutant. The numbers in the bar chart represent the sample size. Due to rapid GLS progression in the field, only one scoring dataset is presented (shown as ‘GLS scale’). (f) GLS resistance phenotype of plants from F_3:4. Each dot represents a single plant. The numbers below the scatter plot represent the sample size. Data are shown as mean ± SD in (a) and (d). Data in (c) and (f) are presented as violin plots overlaid with boxplots and individual data points. The box represents the interquartile range (IQR), the horizontal line indicates the median, whiskers extend to 1.5 × IQR, and individual points show all observations. The violin outline illustrates the kernel density distribution of the data. In (a) and (d), statistical significance was determined by a two‐sided Student's t‐test. In (c) and (f), different lowercase letters indicate a significant difference (P < 0.05) based on one‐way ANOVA with Fisher's least significant difference test.

Discussion

In this study, we performed a GWAS analysis and identified SNP variants associated with SLB resistance in maize. Heritability observed across environments was very high (H ² = 0.95), indicating that most of the variation among inbred lines was attributable to genetic variation, and that sufficient genetic variation was present in the panel to perform association analysis. Based on combined FDR and RMIP thresholds, we identified 14 highly associated SNPs and 10 associated candidate genes (Table 3). All the associated SNPs were in previously identified QRL regions for SLB resistance, except for one SNP on chromosome 6 (Table S6). The 14 significantly associated SNPs were all located in promoter regions or coding sequences of annotated genes (Table 3). A previous genome‐wide association study using the powerful maize nested association mapping population identified a set of SNPs associated with SLB resistance (Bian et al., ²⁰¹⁴). Somewhat surprisingly, none of the 14 associations from the current study were reported as significantly associated with SLB in this previous study. The closest pair of associations between the current study and the NAM study were the SNPs at 16314281 bp (AGP_v5) (this study) and 16 593 803 bp (AGP_v5) (NAM) on chromosome 3 (Bian et al., ²⁰¹⁴).

The closely linked SNPs 5–6 were located in the gene Zm00001eb124020 with high homology to an Arabidopsis thaliana gene encoding an aspartyl‐tRNA synthetase that catalyzes the biosynthesis of aspartyl‐tRNA through esterification of L‐aspartic acid (L‐Asp) to cognate tRNAs^Asp (Guo et al., ²⁰¹⁰). In Arabidopsis thaliana, OE of this gene enhances resistance to Hyaloperonospora arabidopsidis infection (Luna et al., ²⁰¹⁴). Using a combination of fine mapping and analysis of both insertional mutant and edited lines, we recently identified ChSK1, which encodes a leucine‐rich repeat RLK and appears to act as a suppressor of the defense response (Chen et al., ^2023a). ChSK1 spans from 16 581 024 to 16 589 870 on chromosome 3 (AGP_v5), very close to the associated SNP identified in the NAM population (Bian et al., ²⁰¹⁴) but c. 276 kb from SNPs 5–6 identified in this study (Fig. S14). Linkage disequilibrium (LD) in diverse maize usually decays within much shorter distances than 276 kb (Remington et al., ²⁰⁰¹; Yan et al., ²⁰⁰⁹). Sometimes anomalous long‐range disequilibrium can occur, especially if the MAF is low. However, in this case, the MAF of all the SNPs is c. 40% (Table 3). We estimated LD and found that these two regions are not in the same LD block in the maize diversity panel (Fig. S14), which suggests that there are two adjacent relatively large effect SLB quantitative resistance genes in this area of chromosome 3. Interestingly, in our previous studies, we also identified two closely linked QTLs in this region (Balint‐Kurti et al., ²⁰⁰⁷). We have proven that the functional gene at one of these two QTLs is ChSK1 (Chen et al., ^2023a). The aspartyl‐tRNA synthetase identified in the current study may be the causal gene for the other QTL; however, its potential role in SLB resistance requires further investigation.

Differences between mapping results for the same trait from the diversity panel and NAM have been observed in several studies for traits such as maize kernel composition (Cook et al., ²⁰¹¹), hypersensitive defense response (Olukolu et al., ²⁰¹⁴), and plant height (Peiffer et al., ²⁰¹⁴). In these cases, none, one, and none of the associated SNPs were shared between the two populations. This lack of correspondence may be due in part to different population sizes and sampling of alleles. The 26 parents of the NAM population are a subset of the maize diversity panel, whereas a substantial amount of the diversity present in the diversity panel is not included in NAM. It is also probably a reflection of the highly complex genetic basis of these traits and the sensitivity of these mapping approaches (Bian et al., ²⁰¹⁴). The highly polygenic nature of SLB resistance with relatively small effects at each associated locus as well as rapid LD decay in the diversity panel could result in substantial differences in the particular SNPs declared as significant from each study.

Another example of this is that the current study did not identify a significant association close to a glutathione S‐transferase (GST) gene (Zm00001eb315490) on chromosome 7 that was previously reported by Wisser et al. (2011) based on analysis of the same phenotypic data on a subset of 253 lines analyzed with a much smaller and largely distinct set of 858 SNPs. The SNP used in the present study that was nearest to the GST gene (from 133 809 802 bp to 133 814 250 bp, AGP_v5) was located at 133815060 bp (AGP_v5) on chromosome 7 and had a P‐value of 2.5 × 10⁻⁴, which was relatively low but did not pass the FDR threshold that we used. The fact that we used many more SNPs in this study necessitates a much higher FDR threshold. As with every study of this nature, it is inevitable that a relatively stringent significance threshold will result in some false negative assignments, and this may be a case in point.

SNPs 9 and 10 were located in a previously identified QRL identified in two different populations (Table S6) and were within the intronic region of Zm00001eb157070, which encodes a predicted Raf subfamily MAPKKK protein named ZmMAPKKK45 (Kong et al., ²⁰¹³). We confirmed the SLB resistance function of ZmMAPKKK45 by using mutant and transgenic complementation analyses. MAPK cascades are an important component of the PTI and ETI defense responses, which are intimately connected and which both induce ROS production (Yoshioka et al., ²⁰²³). ROS itself is toxic to microbes and ROS‐mediated processes are involved in plant defense signaling (Wang et al., ²⁰²⁴). A number of quantitative disease resistance genes have been reported to be associated with ROS production. For example, the SLB susceptibility gene ChSK1 suppresses ROS production during the defense response, likely by acting as an inhibitory PRR co‐receptor (Chen et al., ^2023a). The maize GLS resistance gene ZmWAKL ^Y positively regulates GLS resistance by regulating ROS accumulation through ZmRBOH4 (Zhong et al., ²⁰²⁴). The maize FERONIA‐like receptors (FLRs) confer broad‐spectrum disease resistance and are required for PTI‐mediated ROS production (Yu et al., ²⁰²²). Here, we show that ZmMAPKKK45 was able to modulate flg22‐, chitohexaose‐ and C. heterostrophus‐induced ROS production possibly via regulating the expression level of ZmRBOHs. This suggests an association of ROS with SLB, GLS and NLB resistance. However, the signaling pathway by which ZmMAPKKK45 regulates the expression of ZmRBOHs and the role ROS plays in this resistance has not been clearly defined and is likely not through the SA pathway, as discussed above.

ZmMAPKKK45 has 48.11% homology at the amino acid level to the well‐characterized Arabidopsis MAPKKK EDR1 and 48.29% homology to its rice ortholog OsEDR1. It is noteworthy, therefore, that while EDR1 and OsEDR1 are negative regulators of disease resistance, ZmMAPKKK45 has a positive effect on resistance. This is one of many examples of the complex nature of plant defense signaling pathways in which positive and negative inputs, often mediated by very similar proteins, must be integrated to achieve the optimal output that confers resistance while not unduly impacting other processes (Aerts et al., ²⁰²¹). Our future work will investigate the connections that ZmMAPKKK45 makes with other components of the maize MAPK signaling machinery and its direct outputs.

Competing interests

None Declared.

Author contributions

TZ and QY planned and carried out experiments, wrote and edited the manuscript; PB‐K planned experiments, wrote and edited the manuscript; BO, SX, YB, RJW and JH analyzed datasets and edited the manuscript. PSO administered grants and edited the manuscript. TZ and QY contributed equally to this work.

Disclaimer

The New Phytologist Foundation remains neutral with regard to jurisdictional claims in maps and in any institutional affiliations.

Supporting information

Fig. S1 Phenotypic distribution of SLB AUDPC BLUPs in different environments.

Fig. S2 Haplotype analysis of SNPs significantly associated with SLB resistance.

Fig. S3 Phylogenetic tree and domain architecture of putative plant Raf MAPKKKs.

Fig. S4 Characterization of ZmMAPKKK45.

Fig. S5 The mutation in ZmMAPKKK45 reduces maize resistance to SLB in the field.

Fig. S6 The overexpression of ZmMAPKKK45 confers enhanced resistance in heterozygous mutant plants.

Fig. S7 Relative transcript levels of ZmPR1 and ZmLOX1 in W22 and W22^MAP3KMu after inoculation with C. heterostrophus in the growth chamber.

Fig. S8 Flg22‐induced reactive oxygen species burst in ZmMAPKKK45 overexpressing and mutant lines.

Fig. S9 ZmMAPKKK45 is involved in regulating H₂O₂ accumulation mediated by C. heterostrophus.

Fig. S10 Overexpression of ZmMAPKKK45 increases H₂O₂ production in leaves infected with C. heterostrophus.

Fig. S11 The expression pattern of ZmRBOHs.

Fig. S12 ZmMAPKKK45 confers resistance to NLB in the field.

Fig. S13 ZmMAPKKK45 confers resistance to GLS in the field.

Fig. S14 LD analysis of SNPs between Zm00001eb124020 and ChSK1.

Table S1 SLB phenotype (AUDPC) and DTA in the maize diversity panel.

Table S2 Primers used in this study.

Table S3 Number of SNPs on different chromosomes.

Table S4 All the significant GWAS variants (Q < 0.20) using the full maize diversity panel.

Table S5 All the significant GWAS variants (RMIP ≥ 0.20) from the 1000 subsample analyses.

Table S6 Significantly associated SNP hits that overlap with QRL to maize SLB in independent mapping populations.

Table S7 The Raf subfamily of ZmMAPKKKs in maize genome.

Please note: Wiley is not responsible for the content or functionality of any Supporting Information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office.

Data availability

The genotype and source data that support the findings of this study are openly available in FigShare at https://doi.org/10.6084/m9.figshare.30614891.v1.