Thirty-two wheat accessions with wide variations (Zhang et al., 2013) were used to detect the nucleotide polymorphism of the target gene. Three winter wheat populations were employed in this study. Population 323 (P323) was used for GWAS. The accessions include 12 landraces, 36 advanced lines, and 275 modern cultivars, most of whose flowering dates occurred within 1 week, mainly planted in the Yellow and Huai River Valleys Facultative Wheat Zone and the Northern Winter Wheat Zone in China (Li et al., 2019b). Population 157 (P157) (157 landraces) and Population 348 (P348) (348 modern cultivars) collected from all 10 wheat production zones in China were used for haplotype analysis (Wang et al., 2019).

The P323 was planted to investigate agronomic traits at Shunyi (SY) (40°23′N; 116°56′E) and Changping (CP) (40°13′N; 116°13′E) in Beijing, China during the 2016–2017 growing seasons. Two water treatments, rain-fed (drought stress, DS) and well-watered (WW), were applied. The WW plot was irrigated with 750 m³ ha^–1 (75 mm) water at pre-overwintering, flowering, and grain-filling stages, respectively, if the rainfall in the corresponding period was insufficient. The total precipitation in the growing seasons was 143 mm. Heat stress (HS) treatment was conducted in a polyethylene-covered greenhouse at the 1-week post-anthesis stage (Li et al., 2019b). Each accession was planted in four rows, with row spacing of 30 cm, length of 2 m, and 40 seeds in each row. Five plants were randomly selected for phenotyping.

For GWAS, P323 was planted in Changping, Beijing, during the 2019–2020 and 2020–2021 growing seasons, and the LRI of 323 accessions was measured. Two water regimes DS and WW were applied in this study. The width of the flag leaf was detected one-third away from the leaf sheath, and the natural leaf width (Ln) and fully unfolded leaf width (Lw) were measured with three replicates, respectively. The LRI data were measured from 8:00 to 10:00 and from 15:00 to 18:00 at the 2-week after flowering. The LRI of flag leaves was calculated by the formula: “LRI = [(Lw − Ln)/Lw] × 100%,” Lw means the expanded distance of leaf blade margins, Ln means the natural distance of the leaf blade margins, and the low and high of LRI value represent the slight and severe degree of leaf rolling, respectively (Shi et al., 2007).

GWAS of wheat leaf rolling genes was performed by analyzing the wheat natural variation population P323. The overall LRI used for GWAS was calculated as the best linear unbiased predictions (BLUPs), which were based on a mixed linear model and performed using the lme4 package in R 3.3.0. Genotyping was performed by using the 660K SNP Array as previously described (Li et al., 2019b). After removing nucleotide variations with missing rates ≥ 0.2 and minor allele frequency (< 0.05), 395,675 SNPs were used for GWAS in this research. On the whole genome scale, in consideration of the population structure (Q) and familial kinship (K), a mixed linear model (MLM) was applied to analyze the association between molecular markers and traits by using TASSEL 5.0 software (Li et al., 2019c).

Three wheat mutants (tamyb5) were obtained by exome sequencing (data not published) from a population, which was a wheat variety Aikang58 (AK58) mutant population induced by ethyl methanesulfonate (EMS). These mutants were backcrossed two times with wild-type AK58 to eliminate the noise in the genetic background. To observe the leaf phenotype at the flowering stage, wheat plants were grown in artificial climate chambers under long-day conditions (16 h light, 25°C/8 h dark, 20°C) at 60% humidity. Plants were treated with drought stress at the post-anthesis stage. The leaf phenotyping was performed and the LRI was measured after 18 days of drought treatment.

Wheat DNA was extracted from the young leaves by the CTAB method (Stewart and Via, 1993). TaMYB5 was cloned by performing PCR amplification using TransStart FastPfu DNA Polymerase and subsequently ligated into a pEASY-Blunt vector for sequencing (Trans Gen Biotech, Beijing). The PCR conditions were 95°C for 5 min; 35 cycles of 95°C for 30 s, 60°C for 30 s, 72°C for 1 min; followed by a final extension of 72°C for 10 min. Five positive clones were randomly picked for sequencing.

Sequence polymorphism of the TaMYB5 gene was detected in 32 wheat accessions with wide variations. Based on a SNP (C/A) located at 544 bp of the gene coding region, a dCAPS marker was developed to distinguish the C/A genotype by mismatching nucleotide to create the Sal I restriction site. The first-round PCR products were obtained by genome-specific primers and used as the templates for the second round of PCR. Second-round PCR products were digested by Sal I restriction enzyme at 37°C for 2 h and then target fragments were separated by electrophoresis in 4% agarose gel (Wang et al., 2020). Association analyses were conducted by SPSS 16.0 software. P < 0.05 was considered a significant correlation. The primers are shown in Supplementary Table 1.

Gene expression analysis was performed using wheat variety Hanxuan 10 as the plant material. Tissue-specific gene expression was analyzed in leaf, root base, and root of seedling at the two-leaf stage, as well as in leaf, spike, penultimate node, root base, root tissues at different depths of jointing stage plants. Wheat seedlings were cultured by hydroponics. Seedlings at two-leaf stage were treated with PEG-6000 (−0.5 MPa) and 50 μM abscisic acid (ABA), and the leaves were collected at 0, 0.5, 1, 2, 3, 6, 12, 24, 48, and 72 h under the treatments (20 h light/4 h dark). Samples at 0 h served as the control. Total RNA was extracted with a TRIzol reagent (TIANMO, Beijing). cDNA was synthesized with a FastQuant RT Kit (TIANGEN, Beijing). Gene expression levels in plant samples were detected using quantitative real-time PCR (qRT-PCR), which was performed with a SYBR Green PCR Master Mix Kit (TAKARA, Beijing). Five plants were harvested for each replicate, and the three biological replicates were set for each treatment.

Full-length ORF of TaMYB5 was cloned and inserted into the pCAMBIA1300GFP vector to generate the p35S-TaMYB5-1300GFP construct. The recombinant plasmid was transformed into Agrobacterium tumefaciens strain EHA105 and then transiently expressed into Nicotiana benthamiana leaves (Zhuang et al., 2021). After 48 h of culture, the green fluorescence signal was detected by a confocal microscope (Carl Zeiss, Germany).

To detect the transcription activity, the full-length cDNA of TaMYB5 was fused into p35S-GAL4DB-NOS to generate the p35S-GAL4-TaMYB5-NOS fusion plasmid. The p35S-5 × GAL4-LUC-NOS, p35S-REN-NOS vectors served as the reporter plasmids, and p35S-GAL4DB-NOS served as the effector plasmids (Liu et al., 2020). The TaMYB5-GAL4DB, the firefly luciferase (LUC), and Renilla luciferase (REN) plasmids were co-transformed into tobacco protoplasts to detect the LUC activity ratios (LUC/REN).

The p35S-TaMYB5-1-1300GFP and p35S-TaMYB5-2-1300GFP were used as effectors. proTaNRL1 was fused into the pGreenII 0800-LUC vector to generate the reporter constructs. pCAMBIA1300-GFP served as vector control. They were transformed into Agrobacterium tumefaciens strain GV3101 and then transiently expressed into tobacco leaves (Li et al., 2022). The luciferase activities were tested using Dual-Glo Luciferase Assay System (Promega, United States).

The CDS sequences of TaMYB5-1 and TaMYB5-2 were cloned and fused into the pB42AD as the effector vector. The promoter fragment of TaNRL1 containing the “ACCTAAC” element was cloned and inserted into the pLacZ vector as a reporter. The effector and reporter plasmids were co-transformed into yeast strain EGY48. Empty pB42AD and pLacZ were used as negative controls. Yeast strain was grown on SD/-Trp/-Ura medium containing X-Gal to detect the interaction at 29°C.

The cDNA sequence of TaMYB5 was fused into vector pGEX-4T1 to construct GST-TaMYB5. The empty GST vector pGEX-4T1 and GST-TaMYB5 were transformed into the Escherichia coli expression strain BL21 (Wang et al., 2020). TaMYB5 protein was induced by 0.2 mM IPTG at 16°C for 10 h and was purified using glutathione-sepharose 4B (GE Healthcare, United States). Biotin-labeled ACC-box and mutant ACC-box, unlabeled ACC-box and mutant ACC-box, and their reverse complementary sequences were synthesized and used as probes. EMSA assay was performed using a LightShift^® Chemiluminescent EMSA Kit (Thermo, United States).

In order to characterize genes associated with wheat leaf rolling, we assembled a natural population of P323 consisting of 323 winter wheat accessions from 10 major wheat regions in China, which had been used for GWAS (Li et al., 2019c; Wang et al., 2021). The leaf rolling degree of each accession was analyzed by measuring and calculating the LRI at the flowering stage. There was a large variation of LRI in 323 accessions, ranging from 0.17% to 70.63%, which indicates that the leaf rolling at the flowering stage is suitable for genetic analysis. Two extreme phenotypes of leaf rolling in the population are shown in Supplementary Figure 1. On the whole genome scale, in consideration of the population structure (Q) and familial kinship (K), mixed linear model (MLM) with the correction function was applied to analyze the association between markers and traits. A haplotype block (479,471,884 bp–480,575,035 bp) on chromosome 3A was significantly associated with LRI and contained 12 candidate genes (Figure 1 and Supplementary Table 2). It has been reported that the MYB family genes were involved in regulating leaf rolling in plant; therefore, TraesCS3A02G257800 was selected as a candidate gene for regulating leaf rolling. Protein structure analysis showed that TraesCS3A02G257800 is an MYB protein with two incomplete MYB repeats, belonging to the R2R3-MYB family (Supplementary Figure 2A). Further phylogenetic analysis showed that MYB5 was the closest homologous to TraesCS3A02G257800, hence the gene was named TaMYB5-3A (Supplementary Figure 2B).

To verify the relationship between TaMYB5 and wheat leaf rolling, we selected three tamyb5 mutants (Supplementary Table 3) and backcrossed the mutants with the wild type (WT, AK58) two times to eliminate background noise. Under drought treatment, the leaves of tamyb5 mutants were flatter than that of WT at the flowering stage (Figure 2A), and the LRI index of WT was more than 60%, while that of mutant plants was less than 10% (Figure 2B). The phenotype of the tamyb5 mutant further confirmed that TaMYB5 was involved in the regulation of wheat leaf rolling.

To further clarify the relationship between TaMYB5 and leaf rolling, we analyzed the polymorphism of TaMYB5 coding regions in 32 wheat cultivars with wide variations (Supplementary Table 4). A SNP (C/A) was detected at 544 bp of the TaMYB5-3A coding region, resulting in a Thr to Lys substitution, corresponding to two alleles SNP-3A-1 and SNP-3A-2 (Figure 3A). Based on the SNP (C/A), a marker of TaMYB5-3A-dCAPS was developed to distinguish the two alleles (Figure 3B). The marker contained a base mismatch in the upstream primer that generated the recognition site for the restriction enzyme Sal I (Figure 3C). TaMYB5-3A-dCAPS was significantly correlated with leaf rolling, and the LRI of SNP-3A-1 was significantly lower than that of SNP-3A-2 (Figure 4A). The results confirmed that TaMYB5 played a role in wheat leaf rolling.

Previous studies suggest that leaf rolling is correlated with canopy temperature (CT) under moderate drought stress (Garrity and O’Toole, 1995). Association analysis was conducted between TaMYB5-3A and agronomic traits in the wheat population P323 and found that TaMYB5-3A was significantly correlated with CT under five environmental conditions. The CT of SNP-3A-1 was significantly lower than that of SNP-3A-2 (Figure 4B).

The frequency distributions of these two alleles in different wheat zones were determined in wheat populations P157 (157 landraces) and P348 (348 modern cultivars). The frequency of SNP-3A-1 in P157 was much higher than SNP-3A-2 in all 10 zones except for Zone VI, especially in Zones I, II, III, V, and X where all accessions were SNP-3A-1 (Figure 5A). The frequency of SNP-3A-1 in P348 was still higher than SNP-3A-2 in all 10 zones (Figure 5B). However, from landraces to modern cultivars in China, SNP-3A-1 with flat leaf and low CT was positively selected in semi-arid areas, including the wheat Zones VI, VII, VIII, and IX, which might have improved drought tolerance of wheat accessions. SNP-3A-2 with rolling leaf and high CT was positively selected in humid and semi-humid areas, such as the wheat Zones I, II, III, IV, and V, which selected for moderate leaf rolling might contribute to photosynthesis in plant development (Figure 5 and Supplementary Figure 3; Zeng et al., 2011). Therefore, the favorable genotype was positively selected in areas with different rainfall levels during the process from landraces to modern cultivars in China.

To explore the role of TaMYB5 in plant growth and development, we detected the expression levels of TaMYB5 in wheat tissues at the seedling stage and jointing stage by real-time quantitative PCR. The results showed that TaMYB5 was widely expressed in all tissues, and was mainly expressed in the leaf (Figure 6A), suggesting that TaMYB5 has a function in wheat leaf.

We next performed qRT-PCR to determine whether TaMYB5 is affected by PEG and ABA. The results showed that both ABA and PEG treatments reduced the expression of TaMYB5 in leaves. Under PEG treatment, the expression of TaMYB5 was down-regulated at 0.5 h, and there was no significant change in subsequent expression (Figure 6B). The expression of TaMYB5 decreased significantly after 0.5 h ABA treatment (Figure 6B). Taken together, these results suggest that TaMYB5 may play an important role in plant drought tolerance.

To examine the subcellular localization of TaMYB5, we constructed the fusion protein of 35S-TaMYB5-GFP and transiently expressed it in tobacco leaves. The laser scanning confocal microscope images showed that the fluorescence signal of the 35S-TaMYB5-GFP fusion protein was only detected in the nucleus; nevertheless, the fluorescence of the 35S-1300GFP vector was detected across the whole cell (Figure 7A). Moreover, TaMYB5-2 was also localized in the nucleus (Supplementary Figure 4). The result provided that TaMYB5 encoded a nuclear localization protein, which was consistent with its potential localization as a transcription factor.

We further used a dual-luciferase reporter system to detect the transcriptional activity of TaMYB5. The empty GAL4DB vector was used as the negative control, and the fusion vector of VP16 (Herpes Simplex Virus16) was used as the positive control. As shown in Figure 7B, TaMYB5 exhibited stronger transcriptional activity than the empty GAL4DB vector. Together, these data indicate that TaMYB5 is a transcriptional activator of nuclear localization.

To determine the downstream target gene of TaMYB5, TaMYB5-1 and TaMYB5-2 were separately constructed into the pB42AD vector, and the promoters of predicted downstream genes were cloned into the pLacZ vector for yeast one-hybrid assays. We examined whether TaMYB5 binds to the promoter of genes involved in leaf rolling regulation, namely, ACL1, CLD1, ROC5, CESAs, RL14, NRL1, PSL1, and SRL2. Yeast one-hybrid assay showed that both TaMYB5-1 and TaMYB5-2 could bind to and activate the proNRL1 (Figure 8A). A previous study has demonstrated that NRL1 is involved in leaf rolling regulation in rice, and the loss-of-function mutant nrl1 displayed leaf rolling by affecting bulliform cell size (Hu et al., 2010). Yeast one-hybrid assay suggested that TaMYB5 could bind to the proTaNRL1, and the A/C nucleotide variation of the TaMYB5 coding region did not affect the binding to proNRL1.

The AC box (ACCTAAC) is a specific cis-acting element regulated by the R2R3-MYB transcription factor (Prouse and Campbell, 2012). Notably, the TaNRL1 promoter region contains the AC element (Figure 8A). After the induction and purification of the TaMYB5 protein (Supplementary Figure 5), electrophoretic mobility shift assay (EMSA) was performed to examine whether the TaMYB5 protein binds to the AC element in the TaNRL1 promoter (Figure 8B). The migration rate of the biotin-labeled AC probe was significantly reduced with the addition of TaMYB5-GST protein, whereas the addition of GST protein alone did not affect the migration rate of the AC probe. In the competition experiment, the unlabeled AC probe could compete with the biotin-labeled AC probe, while the mutated probe did not affect the binding of the biotin-labeled probe to TaMYB5 protein, suggesting that TaMYB5 could specifically bind to the AC cis-acting elements.

We also performed a dual-luciferase reporter assay to further verify that TaMYB5 binds to the TaNRL1 promoter and regulates its transcription. The p35S-TaMYB5-1-1300GFP and p35S-TaMYB5-2-1300GFP were constructed as effectors, and the TaNRL1 promoter was inserted into the LUC vector as a reporter. Compared with the negative control, the LUC activities were higher when the effector and reporter were co-expressed in tobacco leaves (Figure 8C). However, the LUC activity between the two alleles of TaMYB5 was not a significant difference. Taken together, the results suggest that TaMYB5 directly binds to its promoter to promote TaNRL1 expression through AC cis-acting element, while the A/C variation of the TaMYB5 coding region does not affect the binding to the TaNRL1 promoter.