A multi-locus genome-wide association study uncovers candidate genes for alkali tolerance in maize (Zea Mays L.) seedlings

Abstract

Soil salt-alkalization has become a major environmental stress factor limiting the improvement of maize yield. Discovering and deploying genes related to alkali tolerance is of great significance for enhancing maize alkali resistance. Here, we employed an association panel comprising 212 maize inbred lines to identify genetic loci associated with alkali-stress tolerance at the seedling stage using the Maize6H-60 K single-nucleotide polymorphism (SNP) array. A genome-wide association study (GWAS) using six models identified 102 significant SNPs, eight of which showed consistent colocalization across multiple models. These SNPs explained 3.35% to 20.01% of the phenotypic variation. Within the genomic regions covered by the eight co-located SNP loci, we identified a total of 56 candidate genes. Based on functional annotation and homologous gene expression analysis, eight of these genes were considered significantly correlated with alkali tolerance in maize. Subsequent qRT-PCR analysis validated that two candidate genes, Zm00001d014707 and Zm00001d041548, significantly contribute to alkali tolerance in maize seedlings. KEGG pathway and GO enrichment analyses revealed that these genes are potentially involved in stress response and metabolic regulation pathways. The promoter regions of these genes contain regulatory elements associated with stress response and hormone signal transduction. Allelic effect analysis demonstrated that AA and CC were favorable alleles, and their pyramiding constituted a viable strategy to enhance alkali tolerance in maize. These results enhance our understanding of the genetic mechanisms of alkali tolerance in maize and establish a theoretical foundation for generating alkali-tolerant maize lines.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12864-025-12287-1.

Introduction

Soil salt-alkalization is a major abiotic stress that affects plant growth and development, posing threats to agricultural productivity and sustainability [1]. Recently, soil salinization has intensified due to global warming, excessive agricultural fertilization, improper irrigation management, and ecological degradation, reducing crop yields and quality [2–4]. Globally, approximately 1.128 billion hectares of land are affected by saline-alkali stress, accounting for 20% of arable land and 33% of irrigated areas [5, 6]. The effects of soil salinity on plants are categorized into salt stress (NaCl and Na₂SO₄), alkali stress (NaHCO₃ and Na₂CO₃), and combined salt-alkali stress [7–9]. Because of its high pH, alkali stress causes more severe plant damage than other abiotic stresses [10, 11].

Saline-alkali stress induces a series of physiological, biochemical, and morphological changes in plants. It alters the osmotic pressure, disrupts water regulation systems, and affects water absorption and utilization [12]. At the physiological level, saline-alkali stress reduces photosynthetic efficiency while increasing respiratory rates [13]. Biochemically, the alkali environment promotes the accumulation of soluble proteins, free amino acids, and proline [14]. Saline-alkali stress significantly reduces seed viability and germination capacity [15]. Notably, it induces cellular damage in roots, impairing root system development and functionality [16, 17]. Soil salinization disrupts physiological and biochemical homeostasis, thereby compromising stress resistance and reducing crop yield. Additionally, as soil salinity increases, crop yields demonstrate a declining trend [18].

Maize (Zea mays L.) is one of the most vital cereal crops in the world. Among cereals, it is highly susceptible to alkaline stress, particularly during the seedling stage [19, 20]. Therefore, identifying genes related to alkali tolerance is essential for enhancing the alkali tolerance of maize. However, there are currently limited genes associated with alkali tolerance. Currently, research on alkali tolerance mainly focuses on the characterization and analysis of quantitative trait loci (QTLs). In maize, two QTLs related to alkali tolerance were identified in 151 F_2:3 recombinant inbred line (RIL) populations using SSR and SLAF-seq markers [20]. Researchers have also identified a major QTL (qTr6) for alkali tolerance in a rice RIL population (8.5 mM Na₂CO₃) [21]. Genetic mapping under alkaline stress conditions identified seven additive QTLs associated with seedling mortality [22]. Additionally, a major QTL (qAT11) conferring seedling stage alkali tolerance was identified through analysis of 184 RILs in rice [23].

In addition to QTL analysis for identifying alkali stress-related traits, genome-wide association studies (GWAS) have been widely applied to detect trait-associated single-nucleotide polymorphisms (SNPs) and mine candidate genes [24]. Through GWAS analysis, researchers identified a major SNP associated with alkali tolerance in rice on chromosome 11, along with three candidate genes involved in the alkali stress response [23]. The Na⁺ transporter gene ZmHAK4 was identified in maize through GWAS [25]. Association mapping identified two genes, ZmCLCg and ZmPMP3, that regulate salt tolerance in maize [26]. GWAS analysis in maize identified 9 SNPs associated with alkali tolerance, and RNA-seq integration revealed five candidate alkali tolerance genes [2]. Through comprehensive association analysis, seven SNPs and twenty-seven genes were linked to saline-alkali tolerance in wheat [27]. GWAS in rice seedlings identified 121 alkali tolerance QTLs and pinpointed five candidate genes [17]. GWAS has been widely employed to identify genes associated with kernel width, kernel length, and seedling cold tolerance [28, 29].

Although several putative alkali tolerance QTLs and genes have been identified in maize, research employing GWAS to elucidate genes associated with alkali tolerance remains relatively limited. This study assessed seedling-stage alkali tolerance traits using an association panel of 212 maize inbred lines. Genome-wide association studies were performed for alkali tolerance traits by integrating high-throughput SNP genotyping with six analytical models. This study aimed to identify key candidate genes associated with alkali tolerance in maize. The findings enhance our comprehension of the genetic mechanisms of alkali tolerance in maize, while offering valuable references for breeding tolerant cultivars.

Materials and methods

Plant materials

A panel of 212 maize inbred lines derived from breeding programs in southeastern China, exhibiting significant variation in abiotic stress tolerance, was used in this study. According to previous studies [30], this population was classified into ten groups: Improved Reid, P, Tang Si Ping Tou, Iodent, Lancaster, Early-maturing hard-grain, Lvda Red bone, Reid, X, and Mixed groups (Supplementary Table S1). The maize inbred lines were propagated through controlled manual self-pollination at our experimental base in Sanya (18°N, 109°E), Hainan Province, China. Standardized field management was implemented throughout the growth cycle, with no detectable occurrence of diseases and pests. Harvested seeds were dried and subsequently stored in a 4℃ constant-temperature seed bank for future use.

Seedling alkali stress treatment and phenotypic evaluation

Before germination, seeds were surface-sterilized with 10% sodium hypochlorite and rinsed three times with sterile distilled water. After sterilization, seeds were imbibed at 25℃ for 6 h, then 30 seeds were germinated using the standard between-paper method [31]. After 48 h of germination, uniformly grown seedlings were transplanted into culture pots at a density of five plants per pot, with six replicates (totaling 30 plants) per treatment group. The growth environment was maintained at 60% relative humidity under diurnal cycles of 18 h light phase (25℃) and 6 h dark phase (22℃) [32]. At the three-leaf stage, seedlings in the treatment group were daily treated with an alkaline solution (75 mmol/L, NaHCO₃:Na₂CO₃ = 5:1), while control seedlings were irrigated with distilled water [2]. Three biological replications were conducted for each group. After 7 days of treatment, the root dry weight (RDW) and shoot dry weight (SDW) of each line were measured under both normal and alkali stress conditions. The root-to-shoot ratio (RSR) was calculated as the ratio of RDW to SDW. The RSR values obtained under normal and alkaline conditions were designated as NRSR and ARSR, respectively. The relative root-to-shoot ratio (RRSR) was calculated as the ratio of ARSR to NRSR [33].

Statistical analysis of phenotypic data

Phenotypic data analysis was performed using GraphPad Prism 9 to calculate descriptive statistics, including variance, means, ranges, standard deviations, and coefficients of variation. The repeatability (R) of the phenotypic traits was calculated using the formula:

where σ²_g is the genotypic variance estimated with genotype as a random effect, σ²_e is the residual variance, and n is the number of replicates per genotype [34].

Genome-wide association analysis

The genotypic data of the inbred lines were sourced from previous studies [30]. We genotyped the association population using the Maize 6–60 K SNP array. To ensure data quality, SNPs with a minor allele frequency (MAF) < 0.05 and a missing rate >20% were removed, resulting in the exclusion of 1,545 SNPs. Next, SNPs with a heterozygosity >10% and all remaining heterozygous sites were filtered out [35], removing an additional 3,000 SNPs. After these quality control steps, a total of 55,455 high-quality SNPs were retained for subsequent analyses. We conducted association analysis for two alkali-tolerance traits using a total of six models, including one single-locus model (GLM) and five multi-locus models (MLMM, FarmCPU, BLINK, SUPER, and 3VmrMLM) [36]. Specifically, the GLM model was implemented with TASSEL 5.2, the MLMM, FarmCPU, BLINK, and SUPER models were analyzed using the R package GAPIT 3.5, and the 3VmrMLM model was performed with the IIIVmrMLM software package. Based on our previous study, the first ten principal components were included as covariates in all GWAS models to account for population structure [30]. The kinship matrix (K) was calculated using PLINK 1.9 and incorporated into the multi-locus models (MLMM, FarmCPU, BLINK, SUPER, and 3VmrMLM) to account for relatedness among individuals. Because the Bonferroni correction (0.05/55,455 = 9.02E-7) was overly conservative and yielded too few significant SNPs, we selected -log₁₀(P) >4 as a suggestive threshold for detecting significant association signals [2, 37, 38]. For the SNPs meeting this threshold, the false discovery rate (FDR) calculated in GAPIT was < 0.1 [39, 40]. The Manhattan plot and QQ plots were used to evaluate the significance of the associations.

Candidate gene analysis

Based on previous studies, a linkage disequilibrium (LD) decay distance of approximately 200 kb, corresponding to an R² of 0.15, was used in the present study for candidate gene prediction [30, 41]. Using the maize B73 RefGen_v4 reference genome, candidate genes within 200 kb upstream and downstream of the markers were identified. The MaizeGDB (https://maizegdb.org/) platform was interrogated to pinpoint candidate genes within the flanking regions of genetic loci. Functional annotation of candidate genes was performed using NCBI (https://www.ncbi.nlm.nih.gov/) and UniProt (https://www.uniprot.org/) databases [37]. To better understand the biological functions of candidate genes, we carried out Gene Ontology (GO) enrichment analysis through the AgriGO platform (http://systemsbiology.cau.edu.cn/agriGOv2/). Pathway enrichment analysis of the candidate genes was conducted using the KEGG database (https://www.genome.jp/kegg/) [42].

Quantitative real-time PCR analysis

Based on alkali-tolerance phenotypes and favorable allele frequencies, we selected an alkali-tolerant line (L99-F) and an alkali-sensitive line (M-J244-3), which were subsequently validated through quantitative real-time PCR (qRT-PCR) analysis. Leaf tissues were collected from seedlings at the three-leaf stage, three days after treatment with a 75 mmol/L alkaline solution (NaHCO₃:Na₂CO₃ = 5:1). The Hifair^® III 1 st Strand cDNA Synthesis SuperMix kit (Yeasen Biotech, Shanghai, China) was employed to reverse transcribe RNA into cDNA with integrated gDNA removal. The FQD-96 A Real-Time PCR System (Bioer Technology, Hangzhou, China) was employed for qRT-PCR analysis. The 20 µL reaction mixture consisted of 10 µL 2×HS Taq Universal SYBR Green qPCR Master Mix (Saipu Biotech, Nanjing, China), 1 µL cDNA template, 1 µL each of forward and reverse primers (10 µM), and 7 µL ddH₂O. All qRT-PCR primers were computationally designed with Primer Premier 6.0 software (Supplementary Table S2), with ZmActin1 serving as the reference gene for normalization [43]. Relative expression levels of target genes in each sample compared to controls were calculated using the 2^−ΔΔCT method [24]. The study design incorporated three biological replicates per experimental group, and gene expression was quantified as the mean of three replicate measurements.

Analysis of allelic effects

A haplotype was constructed using 200 kb flanking regions of genome-wide significant SNPs [41]. Haplotype blocks were defined using a linkage disequilibrium (LD) threshold of r² >0.9 [44]. The allelic effects of significant SNPs were assessed through integrative analysis of high-density genotypic and phenotypic datasets.

Results

Evaluation of alkali-tolerance phenotypes

To investigate the phenotypic variation characteristics of alkaline tolerance traits, descriptive analysis and genetic parameter evaluation were conducted. For the NRSR trait, the phenotypic variation ranged from 0.29 to 0.70, with a coefficient of variation of 18.2% (Table 1). For the RSR trait, observed values spanned from 0.09 to 0.48, with a mean of 0.28 and a coefficient of variation of 29.3%. For the RRSR trait, phenotypic values ranged from 0.21 to 0.93, with a mean of 0.56 and a coefficient of variation (CV) of 27.1%. The correlation analysis revealed that ARSR was significantly positively correlated with both NRSR (r = 0.37) and RRSR (r = 0.75), whereas NRSR showed a significant negative correlation with RRSR (r = −0.31; Fig. 1). ANOVA indicated highly significant genotypic differences for all traits (P < 0.001; Supplementary Table S3). Notably, all traits exhibited high repeatability (NRSR: 0.92; ARSR: 0.92; RRSR: 0.91), with substantial phenotypic variation observed within the population (CV > 15%). The high repeatability indicates that these traits were stably expressed across replicates, and their phenotypic variation was largely determined by genetic factors. These traits exhibited approximately normal distribution profiles (skewness and kurtosis < ± 1), consistent with polygenic inheritance patterns and suitable for association mapping.

Table 1.

Descriptive statistics of phenotypic performance under alkali stress and control conditions

Trait	Mean	SD	Skewness	Kurtosis	Range	CV (%)	R
NRSR	0.51	0.093	−0.454	−0.431	0.29–0.70	18.2	0.92
ARSR	0.28	0.082	0.104	−0.504	0.09–0.48	29.3	0.92
RRSR	0.56	0.152	0.166	−0.339	0.21–0.93	27.1	0.91

SD standard deviation, CV coefficient of variation, R repeatability. NRSR: root-to-shoot ratio under normal conditions; ARSR: root-to-shoot ratio under alkali stress; RRSR: relative root-to-shoot ratio. Relative traits were calculated as the ratio of trait values under alkaline stress to those under normal conditions

Fig. 1.

Distribution and correlation of alkali-tolerance phenotypic traits. NRSR: root-to-shoot ratio under normal conditions; ARSR: root-to-shoot ratio under alkali stress; RRSR: relative root-to-shoot ratio. Diagonal density plots illustrate the frequency distributions of phenotypic traits. The upper triangle shows the Pearson correlation coefficients between traits, while the lower triangle presents scatter plots with fitted regression lines. *, **, *** indicate significance levels: P < 0.05, P < 0.01, P < 0.001

GWAS of alkali tolerance traits in maize

We conducted a genome-wide association study using six distinct models to identify genetic variants associated with alkaline tolerance in maize, focusing specifically on the ARSR and RRSR traits. To balance false positives and false negatives, a threshold of P < 1 × 10^− 4 (-log₁₀(P) > 4.0) was selected to evaluate the associated loci. A total of 102 significant SNPs were identified by the six models collectively, distributed across chromosomes 1–10 (Fig. 2, Supplementary Figure S4). Through the association analysis of six models, a total of 102 significant SNPs were identified on chromosomes 1–10 (Fig. 2a, b). Specifically, 18, 14, 17, 3, 16, and 34 significant SNPs were detected by the GLM, MLMM, FarmCPU, BLINK, SUPER, and 3VmrMLM models, respectively. The phenotypic variation explained (PVE) by these 102 SNPs ranged from 3.35% to 20.01%, indicating that they account for 3.35% to 20.01% of the phenotypic variation (Supplementary Table S5). In the 3VmrMLM model, the LOD scores for SNPs ranged from 3.13 to 13.72. Through an integrated analysis of the 102 significant SNPs, we identified eight that were consistently identified by multiple models, which comprise four ARSR loci and four RRSR loci (Fig. 2c, d). These eight co-located SNPs were predominantly distributed on chromosomes 1, 3, and 5. Notably, the SNP AX-247,279,240 on chromosome 1 exhibited a PVE of 20.01%, indicating that this locus accounts for a substantial proportion of the phenotypic variance and suggests a major genetic effect on the target trait. The eight co-located SNPs exhibited strong consistency across multiple models, suggesting they represent stable genetic variants. Further analysis revealed that each of the eight SNPs explained more than 5% of the phenotypic variation, indicating that multiple genetic loci control alkali tolerance in maize. Therefore, we propose that these SNP-associated genes exhibit more stable performance in regulating alkali tolerance in maize.

Fig. 2.

Manhattan plots depicting genome-wide significant SNPs associated with two alkali-tolerance traits under six analytical models. a, b present the distribution of association loci identified by the six models: GLM, MLMM, SUPER, FarmCPU, BLINK, and 3VmrMLM. The x-axis represents the physical positions of SNPs across chromosomes 1–10. The y-axis indicates the association significance -log₁₀(P). The green reference line denotes the genome-wide significance cutoff (P = 1.0 × 10^− 4). c, d depict the overlap of significant SNPs among the different models

Identification of candidate genes related to alkali tolerance in maize

Candidate genes related to SNPs were identified using the B73 RefGen_v4 reference genome database. Through association analysis, eight co-located SNPs were identified. Within a 200 kb upstream and downstream range of these loci, 56 candidate genes were identified, of which 48 have been functionally annotated (Supplementary Table S6). To further investigate the potential biological functions of the candidate genes, we performed GO enrichment and KEGG pathway analyses. The results revealed significant enrichment of these candidate genes in several key biological processes, including regulation of cell communication, regulation of response to stress, metal ion binding, and cellular regulation (Fig. 3, Supplementary Table S7). Additionally, significant enrichment was found in DNA damage repair-related pathways, including mismatch repair, homologous recombination, nucleotide excision repair, DNA replication, and plant hormone signaling transduction. These results demonstrate that coordinated gene networks regulate alkali adaptation through metabolic, stress-response, DNA-repair, and hormonal pathways.

Fig. 3.

Functional enrichment analysis of alkali-tolerant candidate genes. The bubble plot visualizes gene enrichment across GO categories (biological process, molecular function, and cellular component) and KEGG metabolic pathways. Bubble diameter scales with the number of enriched genes, and color intensity represents statistical significance levels

Through integrated analysis of functional annotations and functional enrichment, we identified eight candidate genes associated with alkali tolerance in maize, and homologous genes were subsequently characterized in Arabidopsis and rice (Table 2). Specifically, Zm00001d033053 (molybdate transporter), Zm00001d014707 (late embryogenesis abundant protein), and Zm00001d041407 (pyridoxal kinase) were associated with salt and alkali tolerance in plants. The expression of Zm00001d041405 (E3 ubiquitin ligase) responded to osmotic stress. Furthermore, Zm00001d033056 (purple acid phosphatase), Zm00001d014701 (Myb family transcription factor PHL6), Zm00001d014710 (bZIP transcription factor), and Zm00001d041548 (U-box domain-containing protein) contributed to responses to other abiotic stresses.

Table 2.

Functional annotation of candidate genes associated with alkali tolerance

Trait	SNP	Chr.	Alleles	PVE (%)	Candidate gene	Functional annotation
ARSR	AX-247,279,240	1	C/T	20.01	Zm00001d033053	Molybdate transporter
					Zm00001d033056	Purple acid phosphatase
	AX-86,284,272	1	C/T	11.50	Zm00001d014701	Myb family transcription factor PHL6
	AX-108,047,016	5	C/T	13.38	Zm00001d014707	Late embryogenesis abundant protein
	AX-108,031,615	5	A/G	3.35	Zm00001d014710	bZIP transcription factor
RRSR	AX-107,942,625	3	A/G	8.96	Zm00001d041405	E3 ubiquitin ligase
	AX-90,824,162	3	C/T	10.36	Zm00001d041407	Pyridoxal kinase
	AX-108,088,582	3	C/T	12.27	Zm00001d041548	U-box domain-containing protein

Chr. Chromosome, PVE Phenotypic variance explained

qRT-PCR validation of candidate genes

To further validate the putative genes, four candidate genes (Zm00001d033056, Zm00001d014707, Zm00001d041407, and Zm00001d041548) were selected for qRT-PCR analysis (Fig. 4). After 3-day alkali stress, the two genes (Zm00001d033056 and Zm00001d041407) were significantly upregulated in both the alkali-tolerant line L99-F and the alkali-sensitive line M-J244-3. The expression levels of Zm00001d014707, Zm00001d041407, and Zm00001d041548 were significantly higher in L99-F than in M-J244-3, indicating that these three genes may function as positive regulators in maize response to alkaline stress. Interestingly, Zm00001d014707 and Zm00001d041548 exhibited the most substantial upregulation, suggesting their potential critical roles in alkaline stress signaling or the maintenance of ion homeostasis. In contrast, Zm00001d033056 showed significantly lower expression in L99-F compared to M-J244-3, implying that it may serve as a negative regulator of alkali tolerance, and its suppression under alkaline stress conditions could contribute to cellular homeostasis. These results indicated upregulated expression trends for all four candidate genes under alkali stress conditions. Notably, two candidate genes, Zm00001d014707 and Zm00001d041548, exhibited significant differential expression patterns. These genes were upregulated in L99-F but downregulated in M-J244-3, suggesting their potential key regulatory roles in alkali stress response.

Fig. 4.

qRT-PCR validation of alkali-tolerance-associated candidate genes identified by GWAS. Expression patterns of four candidate genes in alkali-tolerant (L99-F) and alkali-sensitive (M-J244-3) inbred lines under control and alkali-stress conditions. *** denotes statistical significance at P < 0.001

Evaluation of allelic effects and utilization of favorable alleles

To identify allelic variations associated with alkali tolerance in maize, we conducted a linkage disequilibrium (LD) analysis of SNPs within the target genomic region. The results demonstrated significant LD among multiple SNPs across this region (Fig. 5). Notably, a distinct LD block (r² > 0.9) was formed by seven SNPs (AX-247279240, AX-86284272, AX-108047016, AX-108031615, AX-107942625, AX-86268838, and AX-90824162), suggesting that this region likely harbors key genes involved in the regulation of alkali tolerance.

Fig. 5.

Linkage disequilibrium heatmap of genomic regions associated with alkali tolerance in maize. The triangular-shaped blocks represent regions of high LD among SNPs. The color key represents the strength of LD, with darker colors indicating higher linkage

Further analysis of the allele effects of eight co-located SNPs revealed that different allelic variations at these loci were significantly associated with alkali tolerance (Fig. 6a). In the ARSR trait, the CC allele significantly increased phenotypic value by 0.04 compared to TT, while AA similarly showed a 0.04 increment relative to GG. For the RRSR trait, AA carriers showed a significant 0.08 increase in phenotypic value compared to GG, while CC genotypes exhibited a 0.06 improvement over TT. Therefore, the CC and AA alleles are considered favorable for enhancing alkali tolerance in maize.

Fig. 6.

Allelic effects of SNPs and utilization of favorable alleles. a Allelic effect analysis of the eight SNPs. b Distribution of favorable alleles between alkali-tolerant and alkali-sensitive lines. Red and blue bars represent positive-effect and negative-effect alleles, respectively. The x-axis indicates the eight SNPs, and the y-axis lists line names. *, **, *** indicate significance levels: P < 0.05, P < 0.01, P < 0.001

We then analyzed the distribution of favorable alleles for eight significant SNPs in the population to evaluate their potential application in maize breeding. Alleles that enhanced phenotypic traits were classified as positive-effect alleles. The results indicated that MZT-Z and L99-F carried the highest number of positive-effect alleles and exhibited superior alkali tolerance phenotypes (Fig. 6b). These accessions exhibited ARSR values of 0.40 ± 0.02 (MZT-Z) and 0.42 ± 0.03 (L99-F), with corresponding RRSR values of 0.61 ± 0.04 and 0.71 ± 0.06, respectively. In contrast, M-J244-3 and 15H035 contained more negative-effect alleles, exhibiting lower ARSR (0.11 ± 0.01 and 0.10 ± 0.01) and RRSR values (0.21 ± 0.03 and 0.23 ± 0.04). Hybrid progenies were generated by crossing alkali-tolerant (MZT-Z × L99-F) and alkali-sensitive (M-J244-3 × 15H035) germplasms, then treated with alkali stress for tolerance assessment. The results indicated that the hybrid combination (MZT-Z × L99-F) exhibited increases of 0.02–0.04 and 0.09–0.19 in ARSR and RRSR, respectively, compared to its parents (Table 3). In contrast, the hybrid combination (M-J244-3 × 15H035) showed smaller improvements in ARSR and RRSR of 0.01–0.02 and 0.04–0.06, respectively. These findings suggest that pyramiding favorable alleles may contribute to enhancing maize alkali tolerance.

Table 3.

Phenotypic performance of parental lines and their hybrids under alkali stress

Name	ARSR (Mean ± SD)	RRSR (Mean ± SD)
MZT-Z	0.40 ± 0.02	0.61 ± 0.04
L99-F	0.42 ± 0.03	0.71 ± 0.06
MZT-Z × L99-F	0.44 ± 0.02	0.80 ± 0.05
M-J244-3	0.11 ± 0.01	0.21 ± 0.03
15H035	0.10 ± 0.01	0.23 ± 0.04
M-J244-3 × 15H035	0.12 ± 0.02	0.27 ± 0.03

MZT-Z and L99-F were alkali-tolerant lines, while M-J244-3 and 15H035 were alkali-sensitive lines

Discussion

Identification of alkali tolerance QTLs in maize using an association panel

As a typical alkali-sensitive crop, maize exhibits significantly inhibited growth and development under soil alkali stress, characterized by reduced seedling biomass and damaged root system [45, 46]. Previous studies assessed maize alkali tolerance through shoot and root dry weights, establishing significant correlations with stress resistance [2, 26, 47]. In this study, we expanded the phenotypic evaluation by incorporating two additional indices, ARSR and RRSR, to conduct a comprehensive association analysis of alkali tolerance traits in maize. The 212 inbred lines exhibited substantial phenotypic variation for both traits. Notably, a significant positive correlation was observed between ARSR and RRSR (r = 0.75, P < 0.001), suggesting that maize may adapt to alkali stress by modulating root-to-shoot biomass allocation.

GWAS is a critical tool for revealing the genetic architecture of complex quantitative traits, facilitating the efficient identification of trait-associated loci and genes. Related studies have demonstrated that 57 loci associated with salt tolerance have been identified through GWAS, and 49 candidate genes have been predicted [48]. Integration of GWAS and transcriptome sequencing identified five candidate genes involved in alkali stress tolerance regulation [2]. The GWAS approach revealed 149 significant SNPs and identified 13 putative genes related to saline-alkali stress tolerance [26]. Integrative analysis of GWAS and RNA-seq data identified two hub genes involved in salt stress response [49]. This study employed six GWAS models and identified a total of eight SNPs significantly associated with alkali tolerance on chromosomes 1, 3, and 5. The PVE by the eight significant SNPs ranged from 3.35% to 20.01%. These results suggest that alkali tolerance in maize is likely controlled by multiple quantitative trait loci, reflecting a complex genetic architecture.

The reliability of the eight candidate SNPs was evaluated through comparison with established QTLs and published SNPs. Five loci exhibited physical colocalization with published QTL regions. Notably, the SNP AX-247,279,240 on chromosome 1 was located within the previously reported alkali-tolerance QTL intervals qRRDW1 and qSDW1 [47, 48]. In addition, Luo et al. [26] identified locus PZE-101,201,482 at 252,259,061 bp on chromosome 1, located 3.7 Mb from AX-247,279,240. On chromosome 5, the three loci, AX-86,284,272, AX-108,047,016, and AX-108,031,615, were located within the previously reported QTL intervals of qNPH5 and qSPH5 [50]. Integrating these findings, we observed genetic overlap among multiple growth and developmental traits in maize under saline-alkali stress. Notably, the loci identified in this study showed positional variations compared to previous reports. These variations may be attributed to differences in stress conditions, population genetic backgrounds, and phenotypic evaluation criteria. Furthermore, the GWAS results could be influenced by factors such as sample size, statistical models, and population structure [41]. Future studies should account for these factors by evaluating SNP stability and consistency across diverse populations and environments.

In this study, seven SNPs were identified within high LD blocks (Fig. 5). These genomic regions contain stress-responsive candidate genes, such as ACER, LEA, and PUB proteins. Phenotype-genotype association analysis revealed significant differences in alkali tolerance among the different allelic variants at these loci. The CC and AA genotypes demonstrated significant positive effects and were defined as favorable alleles (Fig. 6a). Notably, these favorable alleles showed higher frequencies in the alkali-tolerant germplasms MZT-Z and L99-F compared to the other germplasms. These alleles can serve as molecular markers for screening and evaluating alkali-tolerant germplasm. In maize breeding for alkali tolerance, targeted pyramiding of these favorable alleles could facilitate the development of new varieties with enhanced tolerance.

Comprehensive analysis of candidate genes for alkali tolerance in maize

The genome-wide analysis identified 56 candidate genes within the confidence intervals of the eight significant SNPs. The comprehensive analysis revealed that eight genes were significantly associated with alkali tolerance in maize. Notably, Zm00001d014707 and Zm00001d041548 showed significant differential expression between tolerant and sensitive maize lines. The gene Zm00001d041548 encodes a plant U-box E3 ubiquitin ligase (PUB), with AT2G28830 serving as its orthologous gene in Arabidopsis thaliana. PUB is a crucial regulatory factor in the regulation of abiotic stress tolerance [51]. The PUB protein regulates stress responses through the ubiquitin-proteasome system-mediated degradation of target proteins [52]. In maize, ZmPUB19 and ZmPUB59 are significantly upregulated under salt, drought, and heat stress, potentially regulating stress-related gene expression through transcription factor modulation [53]. In wheat, TaPUB1 overexpression upregulates ion transporter genes, enhancing root Na⁺ efflux while reducing K⁺ loss, thereby maintaining cytosolic Na⁺/K⁺ balance [54]. Similarly, SbPUB13 and SbPUB18 show significant upregulation under salt stress in sorghum. These proteins function to maintain ion homeostasis and regulate intracellular Na⁺/K⁺ balance [55]. Additionally, these genes alleviate oxidative damage by scavenging reactive oxygen species. The Zm00001d041548 protein clusters within the same phylogenetic branch as ZmPWZ34402.1 (NCBI accession: PWZ34402.1) and ZmNP_001336571.1 (NCBI accession: NP_001336571.1; Fig. 7a), suggesting that it belongs to the PUB family of E3 ubiquitin ligases and exhibits a high degree of sequence conservation. Members of the PUB family, such as ZmPUB59, ZmPUB19, and SbPUB26, are known to play key regulatory roles in saline-alkali stress and other abiotic stress responses [53, 56]. Given these known functions, Zm00001d041548 may perform similar roles in plant abiotic stress responses. Multiple cis-acting elements, including ABRE and stress-responsive elements, were identified in the promoter region of Zm00001d041548 (Fig. 7c), suggesting that this gene may be involved in the response to saline-alkali stress.

Fig. 7.

Characterization of candidate genes. a, b Phylogenetic analysis of Zm00001d041548 and Zm00001d014707. c, d Promoter region analysis of Zm00001d041548 and Zm00001d014707

The gene Zm00001d014707, located at the AX-108,047,016 locus on chromosome 5, encodes a late embryogenesis abundant (LEA) protein, which is orthologous to AT5G45320 in Arabidopsis thaliana. LEA proteins are primarily localized in the nucleus and cytoplasm. They play multifaceted protective roles during cellular dehydration, collectively enhancing plant stress tolerance [57]. In barley, LEA proteins have been shown to protect yeast cells against high salt concentrations and freezing [58]. Similarly, expression of a soybean LEA protein in Escherichia coli increased salt tolerance, although it failed to enhance resistance to hyperosmotic stress [59]. The OsLEA3-1 and OsLEA3-2 genes in rice exhibit low basal expression but are highly upregulated under drought and salinity abiotic stresses, and their overexpression confers enhanced tolerance to these conditions [60]. Phylogenetic analysis revealed that the Zm00001d014707 gene clustered with known LEA family members, including ZmONM08574.1 (NCBI accession: ONM08574.1) and ZmPWZ33067.1 (NCBI accession: PWZ33067.1), forming a distinct clade (Fig. 7b). Previous studies have reported that among LEA family members, ZmLEA3 is involved in oxidative and low-temperature stress responses [61, 62], ZmLEA14 contributes to drought tolerance [63], and OsLEA3-2 participates in saline-alkali and osmotic stress responses [60]. Given its clustering with these LEA family members, Zm00001d014707 may also play a role in saline-alkali and osmotic stress responses. Additionally, the promoter of Zm00001d014707 contains ABRE, DRE, and MYB cis-elements (Fig. 7d), suggesting that it may participate in plant responses to abiotic stresses such as saline-alkali and drought stress.

Zm00001d033056 encodes a purple acid phosphatase (PAP), which is homologous to AT3G20500 in Arabidopsis thaliana. PAPs belong to the family of binuclear metallohydrolases, whose catalytic activity depends on heterovalent Fe³⁺- M²⁺ metal centers [64]. In plants, PAPs participate in phosphorus metabolism while also exhibiting significant peroxidase activity [65, 66]. This dual functionality enables them to provide multifaceted protective effects against environmental stresses. Under saline-alkali stress conditions, PAP alleviates the stress-induced reduction in phosphorus availability through enhanced phosphorus recycling [67, 68]. Moreover, PAP collaborates with other oxidases to mitigate oxidative stress resulting from excessive reactive oxygen species (ROS) [69].

The locus AX-90,824,162 on chromosome 3 contains the candidate gene Zm00001d041407, which is homologous to AT5G37850 (SOS4) in Arabidopsis and encodes pyridoxal kinase. The SOS4 encodes pyridoxal kinase, which phosphorylates pyridoxal, a form of vitamin B6, to produce pyridoxal phosphate (PLP). As a universal enzymatic cofactor, PLP plays crucial roles in diverse physiological processes, including amino acid metabolism, antioxidant defense, and ion homeostasis regulation [70]. Under salt-alkali stress, the sos4 mutant experiences metabolic disruption due to PLP deficiency [71]. This deficiency leads to Na⁺ accumulation and K⁺ loss, which further exacerbates its sensitivity to salt-alkali stress [72]. The PLP deficiency further induces ROS overaccumulation, resulting in oxidative cellular damage [73]. Additionally, the sos4 mutant exhibits phenotypic traits including shortened roots, reduced root hairs, and smaller meristems.

To better understand the biological functions of these candidate genes, we analyzed their potential involvement in known stress signaling pathways, transcriptional regulation, and stress-responsive processes [74–77]. By integrating gene annotation, literature reports, homologous gene functions, and expression patterns, we inferred the biological processes in which these candidate genes may be involved (Fig. 8). Under saline-alkali stress, ROS accumulate and ionic homeostasis is disrupted, thereby activating the abscisic acid (ABA) signaling pathway. The activated ABA signaling regulates transcription factors, inducing the expression of downstream stress-responsive genes. During this process, the candidate genes may respond to saline-alkali stress through distinct pathways. Zm00001d014707 contributes to ROS scavenging and the maintenance of membrane integrity. Zm00001d041548 participates in protein ubiquitination, regulating protein degradation, ionic homeostasis, and signal transduction. Zm00001d033056 is involved in phosphate metabolism, alleviating stress-induced phosphate deficiency. Zm00001d041407 phosphorylates pyridoxal to produce PLP, thereby indirectly contributing to antioxidant defense and ionic homeostasis. In summary, these genes may synergistically enhance plant tolerance to saline-alkali stress by coordinating multiple pathways, including ROS scavenging, stress signal transduction, and ionic homeostasis. However, the precise biological functions of each gene and their molecular interactions remain to be further investigated.

Fig. 8.

Hypothetical model of the biological functions of candidate genes under saline-alkali stress

Conclusion

In this study, we analyzed two alkali tolerance-related traits in seedlings using an association panel consisting of 212 maize inbred lines. Genome-wide association analysis using six models identified eight co-located SNPs significantly associated with seedling alkali tolerance. Within the target genomic region, we annotated 56 candidate genes, among which eight were predicted to be involved in alkali tolerance through functional annotation and homologous gene analysis. Further qRT-PCR validation confirmed that two candidate genes, Zm00001d014707 and Zm00001d041548, showed significant differential expression under alkali treatment. The AA and CC were favorable alleles for improving maize alkali tolerance. Pyramiding these beneficial alleles represents an effective strategy to enhance alkali tolerance in maize. These findings advance our understanding of the molecular mechanisms underlying alkali tolerance in maize and provide a theoretical foundation for future breeding strategies targeting alkali stress resistance.

Abbreviations

SNP

Single-nucleotide polymorphism

GWAS

Genome-wide association study

PVE

Phenotypic variance explained

QTL

Quantitative trait loci

RIL

Recombinant inbred line

RDW

Root dry weight

SDW

Shoot dry weight

RSR

Root-to-shoot ratio

NRSR

Root-to-shoot ratio under normal conditions

ARSR

Root-to-shoot ratio under alkali stress

RRSR

Relative root-to-shoot ratio

FDR

False discovery rate

LD

Linkage disequilibrium

Authors’ contributions

Y.Y.: Writing - original draft, Formal analysis, Writing - review & editing. Y.W.: Writing - original draft, Formal analysis, Writing - review & editing. D.W.: Writing - original draft, Writing - review & editing. A.R.: Writing - review & editing. H.Y., C.W. and L.C.: Writing - review & editing, Conceptualization. H.F.: Writing - review & editing, Methodology, Supervision, Conceptualization. X.C.: Writing - review & editing, Methodology, Supervision, Resources, Conceptualization, Project administration. All authors reviewed the manuscript.

Funding

This research was funded by the Emergency Management of the National Natural Science Foundation of China (31440066), Key Discipline Construction Funds for Crop Science of Anhui Sciences and Technology University (No. XK-XJGF001), Research Development Fund of Anhui Science and Technology University (FZ230126), and the Joint Research Program on Elite Maize Breeding in Anhui Province (2021AHMS01).

Data availability

The genotypic datasets generated during the current study are accessible at Genome Variation Map (GVM) under accession number GVM001183. Further inquiries can be directed to the corresponding author.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.