The Malus domestica genome used in this study was the “Golden Delicious” apple GDDH13 v1.1 reference genome (GDDH13_1-1,https://iris.angers.inra.fr/gddh13/) (Daccord et al., 2017). Blastp was used to identify all MdPRE members. A. thaliana PRE protein sequences with sequence numbers referenced from Siefers et al. and downloaded from the TAIR database (https://www.Arabidopsis thaliana.org/) (Siefers et al., 2009) were used as the query sequence for protein homology alignment. The six AtPRE sequences are AtPRE1 (At5g39860), AtPRE2 (At5g15160), AtPRE3 (At1g74500), AtPRE4 (At3g47710), AtPRE5 (At3g28857), and AtPRE6 (At1g26945). The searched apple members were submitted to SMART (http://smart.embl-heidelberg.de/) for conservative structural domain confirmation (Letunic and Bork, 2018), resulting in the candidate MdPREs.

The MdPRE gene length was obtained from the GFF3 annotation file of the “Golden Delicious” apple GDDH13 v1.1 reference genome. The length, isoelectric point (pI), molecular weight, and charge at pH 7.0 of the PRE protein sequences were predicted using DNAstar software (DNASTAR 7.1, http://www.dnastar.com). The bHLH structural domain position of MdPREs was analyzed using the Pfam database.

The chromosome localization information of apple PREs was downloaded from the GDR database (https://www.rosaceae.org/; gene_models_20170612.gff3). MdPRE chromosomal localization was mapped using MG2C (http://mg2c.iask.in/mg2c_v2.1/) (Chao et al., 2015). GSDS 2.0 service (http://gsds.gao-lab.org/index.php) was used to analyze the exon–intron structure of MdPREs (Hu et al., 2015).

Clustal Omega (https://www.ebi.ac.uk/Tools/msa/clustalo/) was used to perform multiple sequence alignment and structural domain analysis of MdPRE protein sequences, and the results were visualized in Jalview 2.11.1.4 (Waterhouse et al., 2009). Genetic evolution analyses of maximum likelihood and neighbor-joining trees (MdPREs) were performed in MEGA_X with the step test set to 1,000 times (Kumar et al., 2018). The PRE sequences of Arabidopsis thaliana were referenced from Petroni et al. (2012). Based on the available literature, a combination of SGN (Solanaceae Genomics Network, solgenomics.net, Zhu et al., 2017), RGAP (Rice Genome Annotation Project, rice.uga.edu, Guo Pengyu, 2021), and COTTONGEN (www.cottongen.org, Zheng, 2020) databases, a total of thirty-seven related genes were found for five PREs in tomatoes, seven ILIs in rice, and twenty-five PREs in cotton.

For the conserved motif analysis of MdPRE genes, the MEME5.1.1 server (http://memesuite.org/tools/meme) was used (Bailey et al., 2009), where the Zoops site distribution was chosen, and the maximum module length was set to 60.

The 3D structure analysis of MdPRE proteins was performed using the homology modeling service Phyre² (http://www.sbg.bio.ic.ac.uk/phyre2/html/page.cgi?id=index) (Kelley et al., 2015).

The entire apple genome sequence was downloaded from the GDR database, and the 2.0-kb-long sequences upstream of the transcription start site of the nine MdPRE genes were extracted. The cis-acting elements related to stress responsiveness and plant hormones in the promoter regions of the MdPRE genes were analyzed using PlantCARE (http://bioinformatics.psb.ugent.be/webtools/plantcare/html/) software. The results were visualized using TBtools (V 1.068; https://github.com/CJ-Chen/TBtools).

The interaction network of MdPRE proteins in the apple was predicted using the online STRING database (version 11.5; http://stringdb.org), using Arabidopsis thaliana as the specified organism. The network was constructed using the method described by Mao et al. (2017).

In 2020, different apple tissues (primary roots, annual shoots, fully developed mature leaves, flowers at the first flowering stage, and fruits 150 days after flowering) were obtained from “Royal Gala” apple trees at the Experimental Station of Shandong Agricultural University, immediately frozen in liquid nitrogen, and stored at −80°C for studying the expression pattern of MdPREs.

[Malus hupehensis (Pamp.) Rehd. pingyiensis] seeds were collected from the experimental station of Shandong Agricultural University. The seeds and sand were soaked in a low-concentration potassium permanganate solution for 2 h for surface disinfection. Then the seeds and wet sand were uniformly mixed and laminated and then stored in a refrigerator at 4°C for 45 days. M. hupehensis laminated seeds were planted in vermiculite cavity trays for about 75 days. When M. hupehensis seedlings had about 7-8 true leaves, they were transferred to water for 5 days. The lower part of the ground of the seedlings at the same growth state was treated with various hydroponic solutions: NaCl (100 mmol L⁻¹), ABA (150 μmol L⁻¹), IAA (50 μmol L⁻¹), BR (15 nmol L⁻¹), and GA (100 μmol L⁻¹), sampled after 0, 1, 2, 3, 6, and 12 h. After treatment, whole seedlings were immediately frozen in liquid nitrogen and stored at −80°C for subsequent analysis.

Malus domestica “Orin” apple calluses were grown on MS medium with 0.45 mg L⁻¹ 6-BA, 1.6 mg L⁻¹ 2,4-D, 20 g L⁻¹ sucrose, and 6.0 g L⁻¹ agar powder and adjusted to pH 5.9 with 1.0 mol L⁻¹ sodium hydroxide. Calluses were cultured in the dark at 26°C and subcultured every 18 days.

RNA was extracted from apple tissues using the RNA Plant Plus kit (TIANGEN, Beijing, China), and cDNA was obtained by using the PrimeScript RT reagent kit with gDNA Eraser kit (TaKaRa, Dalian, China). The iCycler iQ5 System (Bio-Rad) was used for quantitative real-time PCR assays. The 2^−ΔΔCt method was used to analyze the data. In addition, three independent replicates were also performed. Md18s was used as an internal reference gene. The sequences of the primers used for quantification are shown in Supplementary Table S1.

The full-length fragments of MdPRE4.3 were amplified from Malus domestica “Gala” apple using the polymerase chain reaction (PCR). The primers used were MdPRE4.3-F 5′ATG​TCA​AGT​AGA​AGA​CCA​TCA3′ and; MdPRE4.3-R 5′ATG​CTG​CAA​AAG​TCT​TCT​AA3’.

The full-length DNA fragment of MdPRE4.3 was cloned into the pCAMBIA1300-cLuc plant expression plasmid downstream of the cauliflower mosaic virus (CaMV) 35S promoter. Subsequently, the pCAMBIA1300-cLuc plasmid and resulting constructs were transformed into Agrobacterium tumefaciens strain LBA4404 using the heat-shock method.

The wild-type (WT) and overexpressing (OE) MdPRE4.3 transgenic apple callus were obtained using the Agrobacterium-mediated transformation method (Zhao et al., 2016).

The 18-day-old WT and OE MdPRE4.3 transgenic apple callus were subcultured on medium containing 100 mmol L⁻¹ NaCl, 150 μmol L⁻¹ ABA, 50 μmol L⁻¹ IAA, 15 nmol L⁻¹ BR, and 100 μmol L⁻¹ GA, respectively, for 21 d in the dark. Growth was monitored using a fresh weight assay.

Three technical replicates with three biological replicates each were performed per experiment. Analysis of variance (ANOVA) was performed using SPSS. DPS software was used for significant difference analysis, and a p-value < 0.05 was considered a significant difference (*, p < 0.05; **, p < 0.01; ***, p < 0.001). Error bars represent the standard deviation.

Apple has 188 reported bHLH genes (Mao et al., 2017). To identify PRE members in apples, BLASTp analysis was performed using A. thaliana PRE protein sequences, and a total of nine MdPRE members were identified in the GDDH13 v1.1 reference genome (Table 1). They were named MdPRE2.1, MdPRE2.2, MdPRE3.1, MdPRE3.2, MdPRE4.1, MdPRE4.2, MdPRE4.3, MdPRE6.1, and MdPRE6.2, based on their homology with the A. thaliana PRE genes (Table 1). Sequence analysis showed that the predicted molecular weights of the MdPRE proteins ranged from 10229.41 (MdPRE3.2) to 11023.27 (MdPRE6.2) Da. Their predicted pI values ranged from 6.06 (MdPRE4.1) to 9.18 (MdPRE3.2).

MdPRE genes were mapped to seven chromosomes by analyzing genomic location information obtained from the GDR database. Chromosomes 00, 09, 13, 16, and 17 each contain one MdPRE gene, while chromosome 06 (containing MdPRE3.1 and MdPRE4.1) and chromosome 14 (containing MdPRE3.2 and MdPRE4.2) contain two MdPRE genes (Figure 1A). IDs and genomic positions of the identified MdPRE genes are summarized in Table 1.

To distinguish differences in the MdPRE gene structures, the exons and introns in the MdPRE gene sequences were analyzed. The analysis of the gene structures showed that the MdPRE gene coding region had a similar intron–exon gene structure, regardless of intron size. Except for MdPRE4.3, which does not contain a UTR, MdPRE genes contain two exons and one intron with similar distribution (Figure 1B), reflecting the relative stability of the MdPRE gene structure during evolution.

To obtain the MdPRE taxonomic and evolutionary relationships, a maximum likelihood phylogenetic analysis was performed with all A. thaliana, tomato, Gossypium hirsutum, and ricePRE/ILI members, which shows that MdPRE2.1/2.2 are closely related to SlPRE2, MdPRE3.1/3.2/4.1/4.2/4.3 are more closely related to GhPREs, and MdPRE6.1/6.2 are closely related to OsILI6 (Figure 2). Subsequently, we named the family members of apple MdPREs based on their close relatives to the model plant Arabidopsis thaliana. This protein nomenclature for PREs has been applied to many other sequenced plants, such as tomatoes (SlPREs) (Zhu et al., 2017), strawberries (FaPREs) (Laura et al., 2019), and Gossypium hirsutum (GhPRE1) (Zheng, 2020).

Figure 3A shows that MdPREs have a conserved structural domain, the helix–loop–helix (H-L-H) structural domain, but no typical basic structural domain. The HLH core conserved structural domain consists of three typical segments, which are two helix segments and one loop segment. The first α-helix is near the amino terminus and is the recognition helix, which recognizes and binds to specific protein sequences. The second α-helix is near the carboxyl terminus and is parallel to the double helix chain, and together with the first α-helix and the intermediate linker loop, they form the helix–loop–helix spatial structure (Murre et al., 1989). As shown in Figure 3B, all PRE members in apples and A. thaliana had the three typical HLH conserved regions, which constitute the core conserved structural domains of PRE, suggesting the functional conservation of MdPRE proteins. Then the high-level structure of MdPRE proteins was predicted using homology modeling. Figure 3C shows that the 3D structures of the nine MdPREs are similar. Moreover, the best templates exactly matched the core conserved domain region and showed a typical HLH structure, consistent with the A. thaliana results (Guo Pengyu, 2021).

To initially investigate the temporal and spatial expression patterns of MdPRE genes in different tissues, MdPRE expression levels in primary roots, annual shoots, fully developed mature leaves, flowers at the first flowering stage, and fruits 150 days after flowering were examined using qRT-PCR (Figure 4). The results showed that the MdPRE gene family was expressed in all the tissues examined. MdPRE3.1, MdPRE3.2, and MdPRE4.1 had the highest expression in leaves, stems, and roots, respectively, and they may be mainly involved in growth and development stages. Moreover, MdPRE3.1 was significantly increased in all tissues tested, except for fruit. MdPRE2.1, MdPRE2.2, and MdPRE4.2 had the highest expression in fruit. MdPRE6.1/6.2 had the highest expression in flower, and MdPRE4.3 was exclusively expressed in flower. Therefore, these genes might be mainly involved in the reproductive growth stage. MdPRE2.1/2.2 in the stem, MdPRE3.2 and MdPRE4.1 in the fruit, and MdPRE6.1/6.2 in the stem and leaf also showed high expression levels.

To explore the potential regulatory factors of MdPREs, a prediction analysis of their promoter cis-acting elements was performed (Figure 5). The sequences 2000-bp upstream of each gene ATG were obtained from the GDR database, and cis-acting elements were analyzed by the online database PlantCARE website. The results showed many hormone-responsive elements (TGA-element, ABRE, CGTCA-motif/TGACG-motif, AuxRR-core, P-box, TCA-element) (Michal et al., 2019), stress-responsive elements (WUN-motif, TC-rich repeats ARE, LTR, MBS), and light-responsive elements (G-box, GA-box, GT1-motif, GATA-motif, Box4, TCT-motif) (Bastian et al., 2010; Porto et al., 2014; Ning et al., 2017). In addition, meristem expression elements (CAT-box) and tissue expression-specific regulatory elements (e.g., O2-site) were contained in some members, indicating that these genes may be influenced by multiple factors (An et al., 2019). For example, the MdPRE4.3 promoter region contains three types of response elements mentioned earlier, and in addition, it contains HD-Zip1 (Sessa et al., 1993), a related element in fenestrated chloroplast differentiation. Taken together, the presence of these cis-acting elements suggests that MdPREs may be involved in multiple responses, which require further studies to elucidate.

To investigate which of the nine MdPRE genes are significantly responsive to abiotic stresses and phytohormones, the changes in MdPRE transcript levels under NaCl, ABA, BR, IAA, and GA treatments were investigated. Consistent with the results of Figure 5, most MdPRE genes responded to different treatments. For example, the response pattern of MdPRE genes to NaCl treatment was complex and diverse, with most genes (MdPRE2.1, MdPRE2.2, MdPRE3.1, MdPRE3.2, MdPRE4.2, MdPRE6.1, and MdPRE6.2) showing upregulated expression levels after 1–2 h of treatment (Figure 6A). Under ABA (Figure 6B), IAA (Figure 6C), and BR (Figure 6D) treatments, the expression pattern of most MdPRE genes showed an increasing trend followed by a decrease. However, the highest expression level was reached at different time points for each gene. Except for MdPRE4.3, at the late stage of GA treatment, the expression levels of MdPRE genes were consistently significantly higher than the pretreatment expression levels (Figure 6E).

The expression levels of most MdPRE genes were responsive to different stress conditions. To characterize the function of MdPRE4.3 in apples, we obtained MdPRE4.3 overexpressed transgenic apple calluses MdPRE4.3-OE-1 and MdPRE4.3-OE-4. The transgenic callus generated much higher transcription levels of MdPRE4.3 than the WT control, suggesting that MdPRE4.3 was successfully transformed into the callus (Figure 7A). Under normal conditions, the fresh weight of MdPRE4.3 overexpressed transgenic calluses was not significantly different from WT (Figures 7B,C). Under NaCl, ABA, and IAA treatment conditions, the fresh weight of MdPRE4.3-overexpressing calluses decreased more, especially under ABA treatment (Figure 7C). Under BR treatment, MdPRE4.3-overexpressing calluses exhibited a less sensitive phenotype than WT calluses (Figures 7B,C). The fresh weights of MdPRE4.3-overexpressing and WT calluses under GA stress treatment were significantly reduced (Figure 7B). However, the amount of their fresh weight reduction was not significantly different (Figure 7C).

The potential functions of apple PRE proteins were explored by mapping MdPREs to A. thaliana homologs using the protein–protein interaction database (Figure 8). In the protein function annotation step, MdPRE2.1/2.2 mapped to AtPRE2/BNQ2, MdPRE3.1/3.2 mapped to AtPRE3/BS1, MdPRE4.1/4.2/4.3 matched to AtPRE4/BNQ3, and MdPRE6.1/6.2 matched to AtPRE6/KDR. According to the predicted results, AIF1, HFR1, IBH1, and HBI1 interact with all PRE proteins and are involved in various biological functions, such as response to hormone signaling (Singh et al., 2017; Gruszka, 2018). In the network, AIF1 negatively regulates BR signaling (Ikeda et al., 2013); HFR1 is involved in phytochrome signaling (Gruszka, 2018); IBH1 negatively regulates cell and organ elongation in response to GA and BR signaling (Lu et al., 2018); and HBI1 acts as a positive regulator of cell elongation downstream of multiple external and endogenous signals. In addition, PRE proteins interact with HUB1 and AT3G06590 belonging to the BRE1 family in the network and are involved in the BR signaling pathway, respectively, indicating that PRE family members are closely related to BR signaling pathway members (Oh et al., 2014). SIEL and AGL21 are involved in plant root growth and developmental processes and interact with PRE3/BS1 (Li, 2010). This suggests that multiple protein interactions mediate the different regulatory processes involved in PREs.