To predict candidate substrates of ZmLAZ1 proteins, the amino acid sequences of eight ZmLAZ1 members were submitted to the SWISS-MODEL software¹ to get a protein model structure file in PDB format. Then the PDB file of each ZmLAZ1 was searched against the RCSBPDB software² to get putative substrates. The coding sequences (CDSs) of the ZmLAZ1-4 and ZmLAZ1-8 genes were amplified from pMD19-T-ZmLAZ1-4 and pMD19-T-ZmLAZ1-8 (Liu et al., 2020) by PCR primers CDS1-4F/CDS1-4R and CDS1-8F/CDS1-8R (Supplementary Table 1), respectively, and inserted into His-tagged prokaryotic expression vector pET-32a (Takara, Osaka, Japan) by using ClonExpress II One Step Cloning Kit (Vazyme Biotech, Nanjing, China). The construct was introduced into Escherichia coils BL21 (DE3), screened on Luria-Bertani (LB) plates containing 100 mg/ml ampicillin, and grown in LB medium at 37°C to OD₆₀₀ = 0.6. The His-tagged proteins were induced by 0.1 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) at 16°C overnight, purified by using Ni-TED 1 ml Sefinose (TM) Column (Sangon Biotech, Shanghai, China), detected by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), quantified in NanoDrop One Microvolume UV–Vis Spectrophotometer (Thermo Fisher Scientific, Waltham, MA, United States), and diluted to 1 μg/μl with 10% dimethyl sulfoxide (DMSO).

As described by Huynh and Partch (2015) with minor modification, 2 μl of 10% DMSO, 6 μg of the purified protein, 2 μl of 10 × SYPRO orange, and 0 (blank control) and 200 μM of each predicted substrate were added into each of three wells of a 96-well PCR plate. In CFX Connect Real-Time PCR Detection System (Bio-Rad, Hercules, CA, United States), the sampled plate was equilibrated at 25°C for 5 min and then ramped up to a final temperature of 95°C in increments of 1°C. Fluorescence was read every 0.2°C ramping up. The change rates of relative fluorescence units (RFUs) with time (T) [-d(RFU)/dT] were plotted vs. the temperature to generate melting curves of ZmLA1-4 incubated with each predicted substrate.

The CDS of ZmLAZ1-4 was amplified from pMD19-T-ZmLAZ1-4 plasmid (Liu et al., 2020) by using primers pYES2F/pYES2R (Supplementary Table 1) and used for construction of yeast expression vector pYES2-ZmLAZ1-4 by using ClonExpress^® II One Step Cloning Kit (Vazyme Biotech, Nanjing, China). The pYES2-ZmLAZ1-4 and empty vector pYES2 were transformed into yeast Zn-sensitive mutant Δzrc1 (Miyabe et al., 2001) and wild-type (WT) strain BY4743, respectively. The positively transformed lines were selected on synthetic dropout medium (SD) plates containing 2% galactose (w/v) and without uracil (Ura), identified by PCR amplification with primers LAZ4F/LAZ4R (Supplementary Table 1), grown in SD medium at 30°C for 16 h, diluted to OD₆₀₀ = 0.8, and then diluted by 10-fold serial to 1:10⁴. Subsequently, 5 μl of each dilution were spotted on SD plates containing ZnSO₄ (0 or 2 mM) and 2% galactose with three replicates and incubated at 30°C for 3 days. Meanwhile, 50 μl transformed lines were grown in a 10-ml SD liquid medium containing ZnSO₄ (1 and 2 mM) and 2% galactose, and used for measurement OD₆₀₀ at 0, 6, and 24 h. Then the cultures were washed with 10 μM ethylene diamine tetraacetic acid (EDTA) and used for the determination of Zn concentration by Inductively Coupled Plasma-Mass Spectrometry (Thermo Fisher Scientific, Waltham, MA, United States).

Overexpression vector (pC2300-35S-ZmLAZ1-4) of ZmLAZ1-4 was constructed as above and introduced into Agrobacterium tumefaciens strain GV3101. Positive strains were identified and used for transformation of WT Arabidopsis thaliana by floral dip. Transgenic lines were screened on kanamycin 1/2 MS plates and identified by PCR with primers LAZ4F1/LAZ4R1 (Supplementary Table 1). Referring to Kawachi et al. (2009), each homozygous line was grouped into three replicates and grown on 1/2 MS zinc deficiency plates (control) with 5 and 50 μM ZnSO₄ at 22°C temperature, 50% humidity, and 16 h light of 120 μE m^–2 s^–1 illumination intensity/8 h dark period for 2 weeks. After photographing for the phenotype, the seedlings were dried at 60°C for 48 h, weighed for biomass, and digested in 80% nitric acid at 250°C overnight. The digested solution was diluted with ddH₂O and used for measurement of Zn²⁺ content by Inductively Coupled Plasma-Mass Spectrometry (Thermo Fisher Scientific, Waltham, MA, United States).

Overexpression vector (pZZ00026-Ubi-ZmLAZ1-4-Tnos) of ZmLAZ1-4 was constructed as above and used to transform embryonic calli isolated from maize inbred line B73 by Agrobacterium mediation. Positive calli were screened on H6 medium with 0.06% (v/v, effective concentration) Basta herbicide. Regenerated plantlets were screened by PAT/bar EPSPS LFD Strip kit (Youlong, Shanghai, China) according to the manufacturer’s instruction and identified by PCR with primers LAZ4F2/LAZ4R2 (Supplementary Table 1) and real-time quantitative PCR (RT-qPCR) with primers LAZ4F3/LAZ4R3 (Supplementary Table 1). Homozygous T₃ lines and WT were grown in vermiculite with Zn dropout Hoagland’s nutrient solution (Coolaber, Beijing, China) at 28°C and 300 μE m^–2 s^–1 illumination intensity for 16 h/20°C dark for 8 h. Referring to Kawachi et al. (2009), at three-leaf stage, the seedlings of each line were grouped into three replicates, treated with 5 and 50 μM ZnSO₄ for 3 weeks, then photographed and dried at 60°C for 72 h, and weighed for biomass and used to measure Zn²⁺ content as above.

The transmembrane domains of LAZ1-4 were predicted by the TMHMM v. 2.0 software.³ The CDS without termination codons of ZmLAZ1-4 and tonoplast maker gene AtTIP2 (Loque et al., 2005) was amplified from pMD19-T-ZmLAZ1-4 plasmid (Liu et al., 2020) and Arabidopsis cDNA using primers Non-Term1-4F/Non-Term1-4R and AtTIP2F/AtTIP2R (Supplementary Table 1) and used for construction of transient expression vector 35S-ZmLAZ1-4-eGFP and 35S-mCherry-AtTIP2 using ClonExpress^® II One Step Cloning Kit (Vazyme Biotech, Nanjing, China), respectively. Subsequently, the 35S-ZmLAZ1-4-eGFP, 35S-mCherry-AtTIP2 of tonoplast maker, 35S-mCherry-OsRac3 of plasma membrane marker (donated by professor Shuangcheng Li, Tao et al., 2021), and empty vector 35S-eGFP (blank control) were introduced into Agrobacterium tumefaciens strain GV3101, respectively.

Maize mesophyll protoplasts were prepared with etiolated leaves and co-transfected with 35S-ZmLAZ1-4-eGFP and 35S-mCherry-OsRac3, as well as 35S-eGFP and 35S-mCherry-OsRac3 as blank control. After incubation at 25°C in dark for 12 h, the protoplasts were used to observe the fluorescence of eGFP and OsRac3 under laser scanning confocal microscope LSM 800 with 488 and 584 nm laser channel (Zeiss, Oberkochen, Germany), respectively.

The combination of 35S-ZmLAZ1-4-eGFP and 35S-mCherry-OsRac3, 35S-ZmLAZ1-4-eGFP and 35S-mCherry-AtTIP2, as well as 35S-eGFP and 35S-mCherry-OsRac3, and 35S-eGFP and 35S-mCherry-AtTIP2 plasmid was mixed with 0.1 M spermidine and 2.5 M CaCl₂ and precipitated onto gold particles (φ = 60 μm), respectively. Onion bulbs were surface sterilized with 75% ethanol. The fifth scales without pigment were cut into 2 cm × 2 cm, incubated on MS medium for 4 h, and bombard in helium biolistic gun (Bio-Rad, Hercules, CA, United States) with above gold particles. After filtration at 28°C under dark for 24 h, the bombarded onion scales were used to observe the fluorescence of eGFP, OsRac3, and AtTIP2 under the same microscope.

The Agrobacterium strains harboring 35S-ZmLAZ1-4-eGFP and 35S-eGFP were infiltrated into the abaxial leaf surface of 3-week-old plants of Nicotiana benthamiana, respectively. After incubation at 22°C and 14 light/10 dark for 24 h, the infiltrated leaves were used to observe the fluorescence of eGFP and autofluorescence of chloroplasts under the same microscope.

The transcriptomic data of maize inbred line B73 were downloaded from MazieGDB⁴ and used for co-expression analysis with ZmLAZ1-4 by using a Perl script (Supplementary Data Set 1). The correlation coefficient was set as > 0.9 and < −0.9. Among the candidates, only the ZmBES1/BZR1-11 gene encoded transcription factor, co-expressed with ZmLAZ1-4 (correlation coefficient is −0.93), and used for RT-qPCR analysis. The seeds of inbred line B73 were surface sterilized with 30% H₂O₂, germinated in petri dish, and transplanted into a plastic mesh grid for hydroponic culture at 28°C under a photoperiod of 14 h light/10 h dark. At the three-leaf stage, the seedlings were subjected to the treatment of 5 μM ZnSO₄. At the 0 (control), 1st, 2rd, and 3rd day of treatment, the whole plant was sampled, ground in liquid nitrogen, and used for total RNA extraction by using RNAiso plus kit (TaKaRa, Osaka, Japan). After removing probable genomic DNA contamination by using RNase-free DNase (TaKaRa, Osaka, Japan), these samples were quantified on NanoDrop 2000 (Thermo Fisher Scientific, Waltham, MA, United States) and reverse transcribed into cDNA by using PrimeScript™ reagent kit (TaKaRa, Osaka, Japan). The RT-qPCR was performed as described by Sun et al. (2020). The ZmGAPDH was used as reference. The primers are listed in Supplementary Table 1.

The CDS of ZmBES1/BZR1-11 was amplified with primers pGADT7F/pGADT7R (Supplementary Table 1) and used to construct vector pGADT7-ZmBES1/BZR1-11. The cis-acting elements bound by ZmBES1/BZR1-11 in ZmLAZ1-4 promoter were predicted by PlantCARE.⁵ The sequence (−1 to −1,100 bp) of ZmLAZ1-4 promoter (pZmLAZ1-4) containing cis-acting elements was amplified with primers pAbAiF/pAbAiR (Supplementary Table 1) and used to construct reporter vector pAbAi-pZmLAZ1-4. The pAbAi-pZmLAZ1-4 was restricted with BbsI and transformed into yeast Y1H gold by using a yeast transformation kit (Coolaber, Beijing, China). The transformant was plated onto Ura dropout SD medium and incubated at 30°C for 5 days. The positive clones were identified by PCR across the multiple cloning sites of the pAbAi vector and the ura3-52 gene of Y1H gold with primers Y1HF/Y1HR (Supplementary Table 1), diluted to 10^–1, 10^–2, 10^–3, and 10^–4 folds with 0.9% NaCl and plated onto Ura dropout SD medium containing either 50, 100, 200, or 400 ng/ml aureobasidin (AbA) to inhibit the Y1H gold background. Competent cells were prepared with the positive clones, transformed with prey vector pGADT7-ZmBES1/BZR1-11, plated onto the Leu dropout SD medium containing AbA at an optimal concentration, and incubated at 30°C for 5 days.

The promotor sequence (−1 to −1,100 bp) of ZmLAZ1-4 (pZmLAZ1-4) was amplified with specific primers pGreenIIF/pGreenIIR (Supplementary Table 1) and inserted into pGreenII-0800-LUC plasmid to drive firefly luciferase gene (LUC) and generate reporter vector pZmLAZ1-4-LUC. The Renilla luciferase gene REN driven by 35S promoter in pZmLAZ1-4-LUC plasmid was used as internal reference. The CDS of the ZmBES1/BZR1-11 gene was amplified with primers pCAMBIA2300F/pCAMBIA2300R (Supplementary Table 1) and inserted into pCAMBIA2300-35S-eGFP plasmid to create effector vector 35S-ZmBES1/BZR1-11. The reporter and the effector vectors were introduced into Agrobacterium strain GV3101, respectively, and used for co-infiltration of Nicotiana benthamiana leaves. After incubated at 22°C and 14 light/10 dark for 3 days, the leaves were visualized for LUC signal in ChemiDoc™ Imaging System (Bio-Rad, Hercules, CA, United States). The relative activities of LUC and REN were determined in a dual-luciferase reporter assay system (Thermo Fisher Scientific, Waltham, MA, United States) and used to calculate relative LUC activity (LUC/REN).

All experiments were performed with three replicates. The data were shown as mean ± standard deviation and analyzed using Student’s t-test at *p < 0.05 and **p < 0.01 level.

By the RCSBPDB software as described by Wang et al. (2001), substrates of eight ZmLAZ1 members were mainly predicted to be organic solutes including α-D-mannose, β-D-mannose, N-acetyl-D-glucosamine, cholesterol, phosphocholine, ethanesulfonic acid, phosphinic acid, toporphyrin, and octyl-β-octylglucoside. However, only ZmLAZ1-4 (Zm00001d012921) and ZmLAZ1-8 (Zm00001d036361) were predicted to combine inorganic ions containing zinc (Zn²⁺), magnesium (Mg²⁺), and calcium (Ca²⁺) (Table 1). During many times of prokaryotic expression, the ZmLAZ1-8 protein was not successfully purified (Supplementary Figure 1). Therefore, the ZmLAZ1-4 was used for further study.

In the thermal shift assay, melting temperature (Tm) of the ZmLAZ1-4 protein incubated with ZnCl₂ and ZnSO₄ was 4.7 and 4.5°C lower than that of ZmLAZ1-4 incubated alone (blank control), respectively, whereas Tm of ZmLAZ1-4 incubated with other predicted substrates kept same value with blank control (Figure 1). This result suggested that zinc ion was candidate substrate of ZmLAZ1-4.

Under non-stress (0 mM Zn²⁺), the diluted colonies and growth curves of OD₆₀₀ showed no significant difference among Δzrc1 mutant transformed by the ZmLAZ1-4 gene, and Δzrc1 and WT transformed by empty vector pYES2. Under 2 mM Zn²⁺ stress, the difference was significant among these three lines (Figures 2A,B). The complementation of ZmLAZ1-4 significantly inhibited the growth of Zn-sensitive mutant Δzrc1. The Zn²⁺ concentration of Δzrc1 transformed by ZmLAZ1-4 was significantly higher than that of Δzrc1 and WT transformed by empty vector pYES2 (Figure 2C), suggesting that ZmLAZ1-4 could transport Zn²⁺ into cells.

Two homozygous T₃ Arabidopsis lines overexpressing ZmLAZ1-4 were screened on kanamycin 1/2 MS plates and identified by PCR amplification (Supplementary Figure 2). Under 0, 5, and 50 μM ZnSO₄ treatments, the growth phenotype of T₃ lines showed no obvious difference compared to WT (Figure 3A). However, Zn²⁺ content of transgenic lines was significantly higher than WT under 5 and 50 μM ZnSO₄ treatments, while only trace content was measured under 0 μM ZnSO₄ treatment (Figure 3B).

By A. tumefaciens-mediated embryonic calli transformation, from the positive transgenic calli harboring ZmLAZ1-4, ten plantlets were regenerated and four homozygous T₃ maize lines overexpressing ZmLAZ1-4 were identified by PAT/bar EPSPS LFD Strips (Supplementary Figure 3), PCR amplification (Supplementary Figure 4), and RT-qPCR (Supplementary Figure 5). Four homozygous lines and WT were grown in vermiculite with Zn deficient Hoagland’s nutrient solution. After 3 weeks of 5 and 50 μM ZnSO₄ treatments, the growth phenotype and biomass of all transgenic lines showed different compared with WT (Figures 4A,B). However, the Zn²⁺ content of all transgenic lines was significantly higher than WT under 5 and 50 μM ZnSO₄ treatments (Figure 4C). The above results indicated that ZmLAZ1-4 functioned as a Zn transporter.

By the TMHMM software, seven transmembrane domains were predicted during ZmLA1-4 protein (Supplementary Figure 6). As shown in Figure 5, the GFP fluorescence was observed in the cytoplasm and nucleus in maize protoplasts, and onion cells transfected by empty vector 35S-eGFP. However, the GFP fluorescence from the fusion protein (35S-ZmLAZ1-4-GFP) was merged with red fluorescence of the plasma membrane marker OsRac3, tonoplast maker AtTIP2, and autofluorescence of chloroplasts. Especially when AtTIP2 was used as a maker, it could be clearly seen that ZmLAZ-4 was localized on the tonoplast. Furthermore, the ZmLAZ1-4 was also localized to chloroplast (Supplementary Figure 7). These results indicated the subcellular localization of the ZmLAZ1-4 protein on the plasma and vacuolar membrane.

In order to explore the mechanism of ZmLAZ1-4 regulating Zn transport, the co-expression analysis was conducted. The results showed that there were 27 genes co-expressed with ZmLAZ1-4 with correlation coefficient > 0.9 or < −0.9, and only ZmBES1/BZR1-11 among these candidates encoded transcription factor and negatively co-expressed with ZmLAZ1-4 (Supplementary Table 2). The result of RT-qPCR likewise showed that the expression of ZmLAZ1-4 and ZmBES1/BZR1-11 was significantly downregulated and upregulated by Zn deficiency (5 μM), respectively (Figure 6). It was predicted that there were six E-boxes (CAXXTG) of BES1/BZR1 binding element (Yin et al., 2005) during ZmLAZ1-4 promoter by PlantCARE. Hence, the yeast one-hybrid (Y1H) was performed. As shown in Figure 7, on Leu dropout SD medium, the growth of Y1H gold strain co-transformed by empty prey vector pGADT7 and pAbAi-pZmLAZ1-4 harboring six E-boxes elements of ZmLAZ1-4 promoter was inhibited by 200 ng/ml AbA, whereas the Y1H gold strain co-transformed by pGADT7-ZmBES1/BZR1-11 and pAbAi-pZmLAZ1-4 formed few colonies, indicating that the ZmBES1/BZR1-11 transcription factor could bind to ZmLAZ1-4 promoter. The result was further verified by dual-luciferase assay in vivo. The relative LUC activity (LUC/REN) of leaves co-infiltrated by reporter vector ZmLAZ1-4-LUC and effector vector 35S-ZmBES1/BZR1-11 was significantly lower than that of control (Figure 7B). These results indicate that the ZmBES1/BZR1-11 transcription factor binds to ZmLAZ1-4 promoter to inhibit ZmLAZ1-4 transcription.