Maize inbred line B73 was selected as the experimental material. Maize seeds were sown in black pot (7cm×8cm×5cm) soil. Only one corn was sotted in each pot and then placed in a 25 °C thermostatic chamber for cultivation. When maize grew to the trifoliate stage, maize plants were subjected to simulated drought stress and salt stress (Drought stress:20% PEG6000; Salt stress: 200mM NaCl). Samples were then taken at 12 h and 24 h of drought and salt stress treatments, respectively. The samples were flash frozen in liquid nitrogen and then stored in an ultra-low temperature refrigerator at -80 °C.

26 maize pan-genome data sets were derived from the research conducted by Hufford et al (Hufford et al., 2021).Subsequently, the hidden Markov model of ARF (PF06507) was retrieved from the Pfam database (https://www.ebi.ac.uk/interpro/entry/pfam/), and then the ARF domain was searched using HMMER 3.1 software, with the threshold set to 1e ^-5. The preliminary ARF members were submitted to the SMART (http://smart.embl.de/) and Pfam databases (https://www.ebi.ac.uk/interpro/) for identification, and genes lacking the ARF domain were removed. Ultimately, the maize ARF members were characterized, and the protein lengths of the identified ARF members were evaluated using BioPerl (Stajich et al., 2002).

The presence-absence information of ARF members was derived from the research conducted by Hufford et al (Hufford et al., 2021). Subsequently, a built-in script was utilized to generate the list of presence-absence genes, and the ggplot2 package in R scripts was employed to create the presence-absence heat map of ARF members in 26 genomes (Ito and Murphy, 2013).

Arabidopsis protein sequences were retrieved from the Arabidopsis database, and the conserved domain information for Arabidopsis ARF family members was extracted. Then, muscle was used to align the conserved domains of maize and Arabidopsis ARFs, and then the maximum likelihood method (Maximum Likelihood, Bootstrap repeated 1000 times) was used to construct the phylogenetic tree of Arabidopsis and maize ARF family members through IQTREE (Nguyen et al., 2015). The resultant evolutionary tree was submitted to ITOL v6 (https://itol.embl.de/) for enhancement and visualization.

The coding sequences (CDS) and protein sequences of ARF family members were retrieved from 26 maize genomes, and the Ka/Ks values for each ARF family member were computed using KaKs Calculator 2.0 (Wang et al., 2010a). Subsequently, the R software packages ggridges and ggplot2 were employed to create the Ridgeline plot of Ka/Ks, and the R software package pheatmap was used to create the heat map for ARF members with Ka/Ks values exceeding 1.

The structural variation analysis results of 26 maize pan-genomes were derived from the research conducted by Hufford et al (Hufford et al., 2021). the B73 genome was chosen as the reference genome for library construction; subsequently, ANNOVAR was employed to annotate the structural variation (Wang et al., 2010b). Then the information that needs to be retained and structural variation is extracted, and then the structural variation information of ARF family members is extracted. Subsequently, a correlation analysis was performed on the structural variation of ARF family members and gene expression levels, and a histogram was generated for genes exhibiting significant differences.

Download the genome annotation file from the maize database website (http://maize-pange.nome.gramene.org.), then extract the significantly different gene structure information identified in the previous step, and submit the protein sequence to the MEME program (http://meme-suite.org/) for protein conserved motif analysis (parameter settings: the maximum number of motifs is 10, the width of the motif is set to 6-50, and the rest are default parameter values). Use TBtools II to visualize gene structure and conserved motifs (Chen et al., 2023).

Subsequently, the protein sequence of the reference genome B73, along with the protein sequence of the strain exhibiting the greatest overlap with its structural variation (SV), was submitted to the MEME program for conserved motif analysis (parameter settings are the same as above), and webLogos representations of the ARF member protein sequences for both strains were generated.

The reference genome B73 and the 2000bp promoter sequence of the strain with the most overlap with SV were extracted. The promoter sequences were submitted to the PlantCARE database (http://bioinformatics.psb.ugent.be/webtools/plantcare/html/) for cis-acting element analysis (Lescot et al., 2002). Then, Tbtools II software was used to visualize the cis-acting elements (Chen et al., 2023).

From phytozome (JGI) database (https://phytozome.jgi.doe.gov) to download the members of the family of IAA corn protein sequences, it was uploaded to the STRING database together with the ARF family member sequences of B73 (Szklarczyk et al., 2023), the interaction between ARF family and IAA family was analyzed, and the results were visualized by cytoscape software (Shannon et al., 2003).

Data were downloaded from the NCBI database based on the RNA-seq data (PRJNA237837), with B73 selected as the reference genome, and Transcripts Per Million (TPM) were utilized as the metric to quantify the expression level. The expression profiles of ZmARF family members in embryos and seeds across various developmental stages were analyzed, and TBtools II was utilized to generate the expression heatmap for ZmARFs (Chen et al., 2023).

Total RNA was extracted from the samples using Promega (LS1040), cDNA was synthesized by reverse transcription using Kangwei Century RT gDNA (CW2020M), RT-qPCR was performed on CFX96 (Bio-Rad, USA) using the US EVERBRIGHT (AugeGreen qPCR Master Mix S2008L) (Three biological replicates and three technical replicates were used). Relative expression was calculated using the 2^-ΔΔct method (Maren et al., 2023). The primers used were synthesized by Sangon Bioengineering (Shanghai) Co., LTD.

In total, 35 ZmARF family members were identified from the maize pan-genome. B73 was utilized as the reference genome to compare the quantity of ARF family members and the lengths of protein sequences across various maize lines, revealing significant differences among the lines. Regarding the number of ARF family members, CML52 and MS71 both identified 36 ARF members, the highest number among the lines, whereas Ki1 identified 33 ARF members, the lowest count ( Figure 1A ). The presence-absence analysis of the identified ZmARF family members revealed that the ZmARF family comprises 21 core genes, 10 near-core genes, and 4 non-essential genes. No private genes were identified within the ZmARF family ( Figure 1B ). A further analysis of the protein lengths of ARF family members across different strains indicated that only ARF12, ARF16, and ARF20 were consistently present in all strains with identical protein lengths, whereas ARF1 was absent in CML69 and Tzi8, although the protein lengths of ARF1 in the other strains remained consistent. With the exception of these four genes, the protein lengths of the other family members exhibited variability ( Figure 1C ). Complete information of ARF family members is provided in Supplementary Table S1 .

A phylogenetic tree was constructed based on the domain sequences of maize and Arabidopsis and classified into six subtribes according to the study by Xing et al (Xing et al., 2011). The G1 subtribe includes only one near-core gene, the lowest count, while the G3 subtribe comprises 13 ARF family members, the highest count, comprising 10 core genes, 2 near-core genes, and 1 non-essential gene. The G2, G4, G5, and G6 subtribes comprise 7, 6, 5, and 3 ARF family members, respectively, and all ARF family members in the G6 subtribe are core genes ( Figure 1D ).

The Ka/Ks ratio serves as a significant metric for investigating the mechanisms of genetic evolution (Hurst, 2002). To investigate the selection pressure on ARF family members across 26 maize genomes, the Ka/Ks values for each ARF family member were computed. The results indicated that the Ka/Ks values for ARF1 and ARF35 could not be determined. With the exception of the peak values for ARF8 and ARF11, which exceeded 1, the Ka/Ks peak values of the majority of other ARF family members were below 1 ( Figure 2A ). Some maize strains exhibit positive selection effects on ARF8 and ARF11, whereas the other genes are subject to purifying selection. Further analysis of the heatmap for Ka/Ks ratios greater than 1 revealed that only ARF8, ARF11, and ARF18 exhibited high Ka/Ks ratios, indicating that these genes were subjected to selection pressure during maize development ( Figure 2B ).

The variation data of ARF family members were extracted from the maize structural variation file; subsequently, the Pearson correlation coefficients of the expression levels of genes both overlapping and not overlapping with SV were calculated. The results indicated that ARF2, ARF3, ARF4, and ARF25 exhibited significant differences in expression levels between the presence of structural variation (SV) and its absence. These significant differences suggest that SV markedly influences the expression of these four genes ( Figure 3A ). Subsequently, we calculated the number of genes exhibiting structural variation (SV) overlap between the reference genome B73 and other genomes, and selected the strain Ki3, which exhibited the highest degree of overlap, to generate a conserved domain alignment diagram ( Figure 3B ). The results indicated that in the Ki3 strain genome, three conserved domains were identified as corresponding to the reference genome B73, whereas the remaining seven did not align. Furthermore, the amino acids within the corresponding conserved domains of each group do not exhibit complete alignment, suggesting that structural variation (SV) substantially influences the conserved domains of ZmARF. In the gene expression analysis, ARF4 exhibited the most significant difference; therefore, we delineated its gene structure and conserved domains across 26 maize genomes. Our analysis revealed that the gene structure and conserved domains in the remaining genomes were in alignment with those of the reference genome B73, showing no significant discrepancies ( Figure 3C ).

The Ki3 strain with the most overlap with the reference genome SV was selected, analysis of cis-acting elements in the promoter, the top 20 elements with the largest number of cis-acting elements were selected for statistics and drawing. It was found that in the reference genome B73 and Ki3, there are differences between cis-acting elements related to light response and cis-acting elements related to plant hormones, B73 contains 8 cis-acting elements related to light response and 5 cis-acting elements related to plant hormones ( Figure 4A ); Ki3 contains 7 cis-acting elements related to light response and 6 cis-acting elements related to plant hormones ( Figure 4B ). Both B73 and Ki3 contained four stress-related cis-acting elements and three growth-related cis-acting elements. It is concluded that structural variation alters the cis-acting element composition of ARF members in different strains, which may further affect physiological processes such as photosynthesis and hormone response in plants. The composition of ARF cis-acting elements in B73 and Ki3 is detailed in Supplementary Figure S1 .

By analyzing the interaction between IAA family and ARF family in maize, it was found that they were closely related ( Figure 5 ). There are 9 members of the ARF family, ARF3, ARF6, ARF11, ARF13, ARF15, ARF16, ARF17, ARF26 and ARF33, play key roles in protein-protein interactions with 28 members of the IAA family. It is concluded that these nine members of the ARF family are more closely related to auxin during auxin signaling and thus play an important role in auxin signaling and regulating plant growth and development.

Transcriptome data pertaining to maize embryos and seeds across various developmental stages were retrieved from the SRA database, and the expression profiles of ARF family members at these stages were subsequently analyzed. The results showed that 11 ARF family members were highly expressed in E1 (embryo-10day), 12 ARF family members were highly expressed in E2 (embryo-16day), 16 ARF family members were highly expressed in S1 (seed-0day), 7 ARF family members were highly expressed in S2 (seed-6day), and only 2 ARF family members were highly expressed in S5 (seed-32day), while the expression of ARF family members in the rest of the developmental stages was at a low level ( Figure 6 ). These findings suggest that ARF family members play a crucial role in the early development of maize embryos and seeds.

The response of the ZmARF gene to drought and salt stress was further analyzed. The results showed that under 20% PEG6000 drought stress treatment, ZmARF gene was significantly differentially expressed at different treatment times ( Figure 7A ). The expression levels of ARF2, ARF6, ARF8, ARF9, ARF15, ARF19, ARF23, ARF29, ARF30 and ARF32 reached the peak at 12 h of stress, and then decreased to different degrees at 24 h of stress. ARF4 and ARF25 peaked at 24 h after stress treatment. Compared with CK, ARF4 was up-regulated at 12 h of stress, while ARF25 was down-regulated at 12 h of stress and then up-regulated at 24 h of stress. Under the treatment of 200 mM NaCl, the expression of ZmARF gene was also significantly different under different treatment time ( Figure 7B ). ARF2 and ARF6 reached the peak at 12 h, and then were down-regulated at 24 h compared with 12 h. ARF4, ARF15, ARF19, ARF23, ARF30 and ARF32 peaked at 24 h of stress treatment, Among them, ARF4 and ARF15 were up-regulated, while ARF19, ARF23, ARF30 and ARF32 were down-regulated after 12 h of stress compared with CK; The expression levels of ARF8, ARF9, ARF25 and ARF29 were down-regulated under salt stress. Among them, ARF8 and ARF29 were up-regulated at 24 h compared with 12 h, while ARF9 and ARF25 were gradually down-regulated over time under salt stress. In conclusion, ZmARF members responded to drought stress and salt stress to different degrees. Among them, only ARF4 expression increased gradually with the increase of stress time under drought stress and salt stress, suggesting that it plays an important role in the response to abiotic stress.