To identify candidate TIFY family members in the peach genome, Hidden Markov model (HMM) profiles of the TIFY domain (PF06200) and Jas (CCT-2) domain (PF09425) and the CCT domain (PF06203) were used as queries to search the peach TIFY members in the peach genome v2.0 (Verde et al., 2013, Verde et al., 2017) using hmmsearch (HMMER, v3.2.1) software (Finn et al., 2011). The E-value threshold of the full-length sequence alignment was set as e-05, and only the protein accessions with E-values lower than the threshold were listed as candidate hits.

Candidate domains and motifs were verified by SMART¹ and NCBI CDD² databases with default parameters. Collinearity analysis within the peach genome and synteny analysis between the peach genome v2.0³ and apple genome v1.0, which was downloaded from Phytozome database⁴, were conducted using MCScanX software (Wang et al., 2012). The phylogenetic analysis was conducted using MEGA software (v10.0.5). The full-length deduced amino acid sequences were aligned using Muscle software. The phylogenetic tree was constructed using the maximum likelihood method with a bootstrap test of 1,000 replicates. Gaps were treated with the partial deletion option with site coverage cutoff set at 70% to keep as much potentially phylogenetical information as possible. The final tree was visualized using Interactive Tree of Life (ITOL⁵). The following GenBank or genome sequencing project accession numbers were used: Arabidopsis thaliana AtJAZ1 (At1g19180), AtJAZ2 (At1g74950), AtJAZ3 (At3g17860), AtJAZ4 (At1g48500), AtJAZ5 (At1g17380), AtJAZ6 (At1g72450), AtJAZ7 (At2g34600), AtJAZ8 (At1g30135), AtJAZ9 (At1g70700), AtJAZ10 (At5g13220), AtJAZ11 (At3g43440), AtJA Z12 (At5g20900), AtZML1 (At3g21175), AtZML2 (At1g51600), AtPPD1 (AT4g14713), and AtPPD2 (AT4G14720); Malus × domestica MdJAZ1 (MDP0000187921), MdJAZ2 (MDP00003 01927), MdJAZ3 (MDP0000193833), MdJAZ4 (MDP00001353 75), MdJAZ5 (MDP0000174042), MdJAZ6 (MDP0000718271), MdJAZ7 (MDP0000173534), MdJAZ8 (MDP0000173535), MdJA Z9 (MDP0000889413), MdJAZ10 (MDP0000565690), MdJAZ11 (MDP0000891920), MdJAZ12 (MDP0000452772), MdJAZ13 (M DP0000244580), MdJAZ14 (MDP0000243322), MdJAZ15 (MDP 0000871409), MdJAZ16 (MDP0000285658), MdJAZ17 (MDP000 0241358), MdJAZ18 (MDP0000757701), MdPPD1 (MDP000025 7732), MdPPD2 (MDP0000282472), MdZML1 (MDP00005098 77), and MdZML2 (MDP0000159765); Vitis vinifera VvJAZ1 (XM_002284819), VvJAZ2 (XM_002262714), VvJAZ3 (XM_0036 34778), VvJAZ4 (XM_002272327), VvJAZ5 (XM_002277733), VvJAZ6 (XM_002277769), VvJAZ7 (XM_002277916), VvJAZ8 (CBI30922), VvJAZ9 (XM_002277121), VvJAZ10 (XM_002263 220), VvJAZ11 (XM_002282652), VvPPD1 (XM_002279284), VvPPD2 (CBI25038), VvZML1 (XM_002270325), VvZML2 (XM_002263671), VvZML3 (XM_002283717) and VvZML4 (XM_002283702); Oryza sativa OsJAZ1 (Os04g55920), OsJA Z2 (Os07g05830), OsJAZ3 (Os08g33160), OsJAZ4 (Os09g2 3660), OsJAZ5 (Os04g32480), OsJAZ6 (Os03g28940), OsJAZ7 (Os07g42370), OsJAZ8 (Os09g26780), OsJAZ9 (Os03g08310), OsJAZ10 (Os03g08330), OsJAZ11 (Os03g08320), OsJAZ12 (Os10g25290), OsJAZ13 (Os10g25230), OsJAZ14 (Os10g252 50), and OsJAZ15 (Os03g27900); Physcomitrella patens JAZs (PP00103G00080, PP00442G00070, Pp3c5_11800, Pp3 c6_23650, Pp3c25_6330, Pp3c16_13490, Pp3c5_11730 and Pp3 c25_6300), PpaZML1 (gw1.68.123.1), PpaZML2 (fgenesh 1_pm.scaffold_69000012), PpaZML3 (fgenesh2_pm.scaffold_11 1000001), and PpaZML4 (e_gw1.226.30.1).

Eight-year-old “Maravilha” and “124 Pan” peach trees are maintained at Wuhan Botanical Garden of the Chinese Academy of Sciences (Wuhan, China). Seven-year-old “Huang Jin Mi,” “Zhong You 18,” and “Qing Feng” peach trees are maintained at Institute of Horticulture, Anhui Academy of Agricultural Sciences (Hefei, China). Fruit samples of “Maravilha” and “124 Pan” were collected at the following developmental stages: S1 (the first exponential growth), S2 (the pit hardening), S3 (the second exponential growth), and S4 (fruit ripening). Fruit samples of “Huang Jin Mi,” “Zhong You 18,” and “Qing Feng” were collected at stage S3. Each fruit sample consisted of three biological replicates from three different trees of the same cultivar, and each biological replicate contained at least five fruits collected from one tree. Fruits were cored, cut into pieces, immediately frozen in liquid nitrogen for at least 5 min and then stored at -80°C until use.

The raw RNA-Seq data were downloaded from the Sequence Read Archive (SRA)⁶ database, with peach transcriptome data of flowers (PRJNA726283, three biological replicates of petal tissues at blooming stage of a 3-year-old F₂ individual generated from self-pollination of a peach accession “05-2-144,” Lu et al., 2021), roots and leaves (PRJEB12334, three biological replicates of root samples and leaf tissues from 15 1-year-old “GF277” peach trees and 15 1-year-old “Catherina” peach trees, respectively, Ksouri et al., 2016), shoots (PRJNA587386, three biological replicates from three different 5-year-old “Soomee” peach trees with three shoots each was used, Yu et al., 2020) and fruits (PRJNA576753, three biological replicates of fruit samples at stages S1, S2, S3, S4, and S5 of cv. “Hujingmilu” with five fruits each were used at each sample time, Cao et al., 2019). Adaptors and low-quality read filtering were conducted using fastp software v0.20.1 (Chen et al., 2018). Then, the clean reads were mapped into the v2.0 peach genome using hisat2 (Kim et al., 2015) version 2.2.1 with default parameters. Quantification of the gene expression levels of the genes was conducted using the R package Rsubread v2.4.3 (Liao et al., 2019). The gene expression levels were quantified using transcripts per kilobase million (TPM). A heatmap of the gene expression levels in various tissues and fruit developmental stages was constructed using TBtools software v1.082 (Chen et al., 2020).

According to the previous reports, the minimum concentration of MeJA used for fruits treatment is usually 50 μM and the maximum concentration is usually lower than 1000 uM (Ju et al., 2016; Wei et al., 2017), therefore in this study, fruits were immersed in solutions with different concentrations of MeJA (50, 200, and 1000 μM) [containing 0.01% dimethylsulfoxide (DMSO)] or distilled water (containing 0.01% DMSO) for 10 min and then transferred to plant growth chambers at 25°C under a light/dark cycle of 16/8 h. Fruit epicarp samples were collected at 2, 4, and 6 days after MeJA treatment.

Measurement of anthocyanin content was conducted according to a previous report (Zhou et al., 2014) with some modifications. Briefly, approximately 0.5 g of epicarp tissue was ground to fine powder in liquid nitrogen, and the anthocyanins were extracted with 15 ml extraction solution (1% HCl in 70% ethanol) for 24 h. After centrifugation, 1 ml supernatant was mixed with 4 ml buffer A (0.2M KCl: 0.2M HCl = 25:67, pH = 1.0) or buffer B (1M sodium acetate: 1M HCl: H2O = 50:30:45, pH 4.5). Absorbance of each mixture was measured at 510 and 700 nm. The anthocyanin content was calculated using the following equation: TA (mg/100 g) = A*MW*75*5*100/(26900*0.5).

Total RNA extraction was conducted using a Total RNA Rapid Extraction Kit (Zomanbio, Beijing, China). First strand cDNA synthesis was performed using the PrimeScript™ RT Reagent Kit with gDNA Eraser (Takara Bio, Inc.). RT–qPCR was conducted using TB Green^® Premix Ex Taq™ (Tli RNaseH Plus, Takara Bio, Inc.), with the following program: one cycle of 30 s at 95°C, followed by 40 cycles of 5 s at 95°C and 30 s at 60°C. The previously reported translation elongation factor gene PpTEF2 was used as the internal reference gene (Tong et al., 2009). Three biological replicates were conducted for each sample. Sequences of the primers used for RT–qPCR are listed in Supplementary Table 1.

The full-length coding sequences of peach JAZs, ZMLs and MYC2 were amplified using a list of primers (Supplementary Table 1). PCR products were digested with restriction enzymes and inserted into the Y2H vectors pGBKT7 and pGADT7 as bait and prey, respectively. The yeast two-hybrid assay was conducted using the Matchmaker^® Gold Yeast Two-Hybrid System (Clontech, Japan). The bait and prey vectors (empty vectors as negative controls) were transformed into yeast strains “Y2Hgold” and “Y187” using the Frozen-EZ Yeast Transformation II Kit (Zymo RESEARCH), respectively. After mating, the diploid yeast cells were grown on DDO (SD-Trp-Leu) and QDO/A/X (SD-Trp-Leu-Ade-His + AbA + X-α-Gal) media at 30°C. Photos were taken after 3 days following incubation.

Split firefly luciferase complementation assays were conducted on the young Nicotiana benthamiana leaves according to a previous report (Chen et al., 2008). Briefly, whole-coding sequences of PpJAZ3 and PpZML4 (with the native stop codon removed) were both amplified and separately inserted into multiple cloning site (MCS) of the binary vector pCambia1300NLuc, while the whole coding region of PpMYC2 (without ATG translation initiation codon) was cloned and inserted into MCS of the binary vector pCambia1300CLuc. These recombinant constructs and the empty vectors were individually transferred into Agrobacterium strain GV3101 via electroporation and incubated at 28°C for 2 days. The confluent bacteria were suspended in the buffer containing 10 mM MES, 10 mM MgCl₂ and 200 μM acetosyringone and incubated at room temperature for 2 h. Agrobacterium cultures containing the NLuc and CLuc derivatives were mixed in equal ratio and injected into young leaves of Nicotiana benthamiana seedlings grown in the greenhouse using needleless syringes. At the 3rd days after infiltration, the infiltrated leaves were soaked with one millimolar D-luciferin, sodium salt (Yeasen, Shanghai, China) in dark for 6 min to quench the fluorescence. The Luc images were captured using a cooled CCD imaging apparatus (Tanon 5200 Multi-Imaging System, Tanon Science and Technology Inc., Shanghai, China). Leaf discs (1 cm in diameter) adjacent to the infiltration holes were punched to measure firefly luciferase (Luc) activities using Bright-Glo Luciferase Assay System (Promega) on an Infinite M200 luminometer (Tecan, Mannerdorf, Switzerland).

Hidden Markov model profiles of the TIFY, the Jas (CCT-2) and the CCT domains were used as queries to screen the peach genome (v2.0) to identify the putative peach TIFY, JAZ, PPD and ZML subfamily members. In total, 15, 16 and 28 protein hits were found to possess the TIFY domain, the Jas domain and the CCT motif, respectively, with 15 of them containing both the TIFY domain and the Jas domain and 5 of them comprising both the TIFY domain and the CCT motif (Supplementary Table 2). Prupe.1G467500 contains a TIFY domain but not a Jas or CCT motif and was therefore assigned as a member of the TIFY subfamily. Considering the limited functional study and information about the TIFY subfamily, this study focused on TIFY family members containing the Jas or CCT motif, and only one TIFY subfamily member, Prupe.1G467500, was not investigated further. It was interesting that all 5 members comprising both the TIFY domain and the CCT motif were included among the 15 members containing both the TIFY domain and the Jas domain, which meant that 5 members (Prupe.8G071200, Prupe.8G071300, Prupe.1G534400, Prupe.3G019800 and Prupe.1G534300) contained both the Jas domain and the CCT motif. To confirm these results and further classify the proteins, the SMART and NCBI CCD databases were used to examine the conserved domains of the 15 candidate proteins. Indeed, scanning of the SMART database showed the same results as the HMM software: the five proteins contained both the Jas domain and the CCT motif, and the two motifs overlapped with each other; however, the CCT motif showed much smaller E-values than the Jas domain for all five members (Supplementary Table 3). Furthermore, the NCBI CCD database called only the CCT motif but not the Jas domain for the five proteins. Therefore, five proteins (Prupe.8G071200, Prupe.8G071300, Prupe.1G534400, Prupe.3G019800 and Prupe.1G534300) were concluded to contain both the TIFY and CCT motifs and were predicted to belong to the ZML subfamily.

To further separate the JAZ and PPD subfamilies, amino acid alignment of all JAZs of peach and Arabidopsis was performed (Supplementary Figure 1) and the annotation of the domains was conducted according to a recent report (Garrido-Bigotes et al., 2020). One member, Prupe.1G020900, lacks the conserved PY motif at its C-terminal region, although it contains the TIFY domain and the Jas domain, which is consistent with the features of PPD proteins (Chini et al., 2009).

Taken together, 1 TIFY, 9 JAZs, 1 PPD and 5 ZML genes were finally obtained in this study. Compared to a previous report of peach JAZs involved in fruit and seed development (Ruiz et al., 2013), 2 more JAZs (Prupe.7G189200 and Prupe.1G578500) were found in this study. Finally, we designated these JAZs following the rules made by Ruiz et al. (2013), with some supplements and modifications (Table 1). Prupe.7G189200 and Prupe.1G578500 were designated PpJAZ2 and PpJAZ6, respectively. In addition, Prupe.1G020900, previously named PpJAZ9, was corrected to the name PpPPD1. To further confirm the structures of these genes, we cloned the genes using fruit cDNAs as templates and verified the gene structures by Sanger sequencing. Finally, 13 full-length CDSs of the 15 genes were successfully cloned, except PpPPD1 and PpJAZ6, possibly due to their low expression levels in fruit. Most of the gene structures were the same as the genome sequencing annotation v2.0; however, some of them were not. For example, PpJAZ3 was identical to the CDS of peach genome annotation v1.0 (ppa008065m) but not v2.0 (Prupe.5G235300). In addition, PpJAZ8 (318 bp) was shorter than the CDS (Prupe.3G188200, 363 bp in length) in the v2.0 annotation. Similarly, compared to the gene structure of Prupe.3G019800 annotated in the v2.0 peach genome, PpZML4 skips one intron, which produces an internal stop codon and results in premature termination of translation. Nevertheless, the deduced protein of PpZML4 still contains the TIFY domain and the CCT motif.

The generation and maintenance of gene families is usually associated with tandem and segmental gene duplication. According to a previous report, tandem and segmental gene duplication play different roles in different gene families (Cannon et al., 2004). The peach JAZ, PPD, and ZML members were mapped to chromosomes 1, 3, 4, 5, 7, and 8 (Figure 1A and Table 1). Based on the positions of these genes and their phylogenetic relationships, we identified two ZML tandem repeat clusters (PpZML1/2 and PpZML3/5) located on chromosomes 8 and 1, respectively. Collinearity analysis of the JAZ, PPD, and ZML members found only one segmentally duplicated pair, PpJAZ3/4 (Figure 1B), which was located on chromosomes 5 and 1, respectively. These results revealed that 2 of the 9 JAZ and 4 of the 5 ZML genes were associated with either tandem or segmental gene duplication events.

Peach and apple share a hypothetical ancestral Rosaceae genome with 9 chromosomes (Illa et al., 2011). Comparison of the JAZ, PPD and ZML genes between peach and apple can provide insights into the evolutionary history of this gene family. A previous report characterized 18 JAZs, 2 PPDs and 2 ZMLs in the apple genome (Li et al., 2015). A comparative syntenic analysis was conducted between peach and apple genomes. The JAZs, PPDs and ZMLs located in/around one identified large-scale synteny are highlighted in Figure 1C. The final syntenies contained homologs from peach and apple, including 7 peach JAZs, 1 peach PPD, 1 peach ZML and 11 apple JAZs, 2 apple PPDs, and 2 apple ZMLs (Supplementary Table 4). One pair of orthologs, PpJAZ8-MdTIFY2, appeared to correspond to a single peach-to-apple JAZ gene. Considering the duplication event in the apple genome, we hypothesized that gene loss events occurred for these genes in the apple genome after whole genome duplication. There are also more instances where a single peach gene corresponded to multiple apple genes, including PpJAZ4-MdJAZ12/18 (MdJAZ18 was unanchored), PpJAZ2-MdJAZ2/9, PpPPD1-MdPPD1/2 and PpZML3-two unidentified apple ZMLs (MDP0000192617 and MDP0000316985). Most of these orthologs followed the whole genome duplication pattern. For example, PpPPD1 is located on the proximal part of chromosome 1 of the peach genome, and MdPPD1 and MdPP2 are located on chromosomes 13 and 16 of the apple genome, which were both derived from the hypothetical ancestral Rosaceae chromosome A1 (Illa et al., 2011). Similarly, the two apple chromosomes 8 and 15, containing the two apple ZMLs MDP0000192617 and MDP0000316985, respectively, also shared the same hypothetical ancestral Rosaceae chromosome A2, with the distal part of peach chromosome 1, where PpZML3 was located. Interestingly, duplication of JAZs (MdJAZ2 and MdJAZ9) on non-homologous chromosomes 2 and 9 of the apple genome, which are orthologs of PpJAZ2, was also observed in our analysis (Figure 1C and Supplementary Table 4).

The peach genomic region containing the ZML tandem repeat PpZML1/PpZML2 is homologous with regions on chromosomes 5 and 10 of the apple genome; however, there were no ZML genes in this apple genomic region. For another ZML tandem repeat, PpZML3/PpZML5, only PpZML3 was in a synteny with two apple ZMLs, which implied that PpZML5 was probably generated from PpZML3 by tandem duplication.

To further understand the evolutionary relationship of the JAZ, PPD and ZML genes in different species, a phylogenetic tree was constructed using the maximum likelihood method (Figure 2). The deduced whole-length protein sequences of JAZ, PPD and ZML proteins from the moss Physcomitrella patens (Bai et al., 2011), Arabidopsis (Chini et al., 2009), rice (Ye et al., 2009), grape (Zhang et al., 2012), apple (Li et al., 2015) and peach were extracted and used in this analysis. The phylogenetic analysis results showed that all of the JAZ, PPD and ZML members were classified into 9 clades, comprising 1 PPD clade, 1 ZML clade and 7 JAZ clades. The moss ZMLs were clustered with ZMLs from the other species and grouped into the ZML clade; however, moss JAZs clustered into one clade (JAZ I), distinguishing them from JAZs from eudicots, including peach (Figure 2). These results indicated that JAZs divergent at a faster rate than ZMLs during evolution history. Moreover, similar to a previous report (Zhang et al., 2012), PPD clade members were not present in P. patens but in all dicot plants, indicating the possible late emergence of the PPD domain following the divergence of moss and vascular plants.

When comparing the phylogenetic relationships of these genes between peach and apple, it was found that most of the peach JAZ, PPD and ZML members showed the highest homology with apple genes due to their close evolutionary relationship. For most of the clades, including JAZ I, JAZ III, JAZ IV, JAZ V, and PPD, apple always contained more members than peach (Figure 2). This result is consistent with the genome duplication event that occurred during Rosaceae evolution. However, this rule did not work in the ZML clade, which was in accordance with the syntenic analysis results between peach and apple. These results suggested that gene duplication and loss events occurred during the evolution of Rosaceae species.

Phylogenetic analysis was also conducted using the deduced full-length protein sequences of the 15 peach JAZ, PPD and ZML genes identified in this study (Figure 3A). The resulting topological structure was consistent with the phylogenetic tree constructed from gene sequences from five different species (Figure 2), in which the JAZs and ZMLs clustered into two different clades.

The exon/intron structures were also investigated according to the location of the exon and intron sequences in the peach genome (Figure 3B). The results showed that the genes in the same phylogenetic clades tended to share the same or similar exon/intron structures. PpJAZ2/6, PpJAZ3/4 and PpZML1/5 had identical numbers of exons with similar lengths, which suggested that they may be the products of gene duplication. Interestingly, PpZML2/3 also had similar exon-intron patterns. Considering that tandem duplication events occurred in PpZML1/2 and PpZML3/5, it is reasonable to speculate that the orthologs (PpZML1/5 or PpZML2/3) were generated later than the paralogs (PpZML1/2 or PpZML3/5). Nevertheless, the exon length was not conserved for genes having the same or similar exon numbers and length. For example, the first intron of PpJAZ3 (1.36 Kb in length) is much larger than that of PpJAZ4 (0.46 Kb in length), and the same occurred for the fifth intron of PpZML1 and PpZML5.

To further confirm the evolutionary relationships among the peach JAZ, PPD and ZML genes, the distribution of the conserved domains and motifs was assessed (Figure 3B). All members in the same clades shared the same conserved domains and motifs. For example, all JAZs have the TIFY domain and Jas domain, and all ZMLs contain the TIFY, CCT and GATA Zn finger domains. It is interesting that both JAZs and ZMLs have the TIFY domain, but ZMLs showed a more conserved location and distribution of this domain than JAZs. The TIFY domain of ZMLs always spans the first and second exons but does not always exhibit this pattern for JAZs. These results suggested that JAZs showed more divergent structures than ZMLs, probably implying more divergent functions of JAZs.

To investigate the tissue-specific expression patterns of the JAZ, PPD and ZML genes of peach, RNA-Seq data was downloaded from NCBI SRA database, including peach transcriptome data of flowers, roots, leaves, shoots and fruits. A heatmap was constructed according to gene expression levels (TPMs) in various tissues (Figure 4 and Supplementary Table 5). The heatmap results showed that most of the genes in this gene family were expressed in a tissue-specific pattern. For example, the expression of PpJAZ10 was much higher in the fruit than in the other tissues investigated. PpJAZ5 showed lower mRNA levels in shoots and leaves than in flowers, roots and fruits. PpZML2 was preferentially expressed in young shoots but not in roots or fruits. However, some members showed similar expression levels in nearly all investigated tissues. For example, PpJAZ1 and PpJAZ2 showed constitutive expression features, and their expression was always maintained at high levels. In contrast, PpJAZ6 and PpPPD1 were always expressed at extremely low levels in all or nearly all tissues investigated.

When we examined the expression levels of these genes in different developmental stages. On the whole, the expression levels of TIFY members could be classified into high (TPM > 100), medium (20 < TPM < 100) and low (TPM < 20) levels. These highly expressed genes either showed constitutively high expression levels at all fruit developmental stages (PpJAZ1 and PpJAZ2) or were highly expressed at both the exponential growth (stages S1 and S3) and ripening (stages S4 and S5) stages (PpJAZ5 and PpJAZ10), except at the pit-hardening stage (S2). For those members with medium expression levels, including PpJAZ3, PpJAZ4, PpJAZ7 and PpZML1, only PpJAZ3 was highly expressed at the fruit ripening stages in cv. Hujingmilu (Figure 4). However, the expressional of PpJAZ3 varied at the fruit ripening stage in different peach accessions. For example, PpJAZ3 showed much lower expression level at the fruit ripening stage in cv. “Zao Huang Pan Tao” (Supplementary Table 6). For TIFY members with low expression levels, PpJAZ6 was undetectable at the mRNA level at all fruit developmental stages, and PpPPD1, PpZML2 and PpZML4 showed extremely low expression levels throughout the entire fruit developmental stages. Interestingly, the expression of PpJAZ8 was nearly undetectable from stage S1 to fruit ripening stages in “Hujingmilu” (Figure 4) but was active at the E1 and early S1 stages in another cultivar “Zao Huang Pan Tao” (Supplementary Table 6), indicating the relevance of this gene to the fruit set process. These results indicated that the mRNA levels of JAZ, PPD and ZML genes did not seem to be affected by fruit ripening cues.

To further confirm the expression patterns of the genes during fruit developmental and fruit ripening stages, RT–qPCR was conducted with two different peach cultivars, “Maravilha” and “124 Pan” (Figure 5). Overall, the RT–qPCR results were consistent with the RNA-Seq results mentioned above. The most abundantly expressed genes, PpJAZ1, PpJAZ2 and PpJAZ5, showed the highest expression levels in the RT–qPCR tests. Similar to the RNA-Seq results, the expression of PpJAZ6 was nearly undetectable in fruits of both cultivars (Figure 5). A slight difference from the RNA-Seq results was that the expression of PpJAZ8 was also detected at stage S2 (Figure 5), probably due to the variation in fruit collection time and fruit developmental process in different accessions.

In addition, all ZMLs showed relatively medium or low expression levels, which also agreed with the heatmap results. Many of the genes showed relatively higher expression levels in juvenile fruits (stages S1 and S2) than in the pre-ripening and ripening fruits. For instance, the expression levels of all ZML genes showed a decreasing trend during fruit development and ripening processes. Nevertheless, the gene expression patterns of these genes varied between the two cultivars. For most genes investigated, their expression levels were higher in “124 Pan” than in “Maravilha,” indicating the divergence of JAZs in the fruit developmental process on different genetic backgrounds.

To test the effect of MeJA on different peach cultivars, three cultivars, “Huang Jin Mi,” “Zhong You 18,” and “Qing Feng,” showing no red pigmentation in fruit skin at stage S3 were selected for the exogenous MeJA (50, 200, and 1,000 μM) treatment experiments. The skin color of “Zhong You 18,” was notably affected by exogenous MeJA treatment (Supplementary Figures 2A,B), along with the significantly increased anthocyanin content (Supplementary Figure 2C). Therefore, fruits of “Zhong You 18” were collected to study the effects of MeJA on the expression levels of peach JAZ, PPD, and ZML genes 2, 4, and 6 days after treatment (DAT). As shown in Figure 6 and Supplementary Figure 2D, PpJAZ6, PpJAZ8, PpPPD1, PpZML2, and PpZML4 showed no or nearly no expression in this experiment. PpJAZ10, PpZML1, PpZML3, and PpZML5, with average relative expression values below 0.2, were characterized as genes with low expression. Only the genes with the average relative expression values higher than 0.2 (including PpJAZ1, PpJAZ2, PpJAZ3, PpJAZ4, PpJAZ5, and PpJAZ7) were submitted for one-way ANOVA (Figure 6). Overall, most of the highly expressed genes showed higher expression levels at 6 DAT than at 2 or 4 DAT (Figure 6). However, PpJAZ2 showed no significant changes in mRNA abundance after exogenous MeJA treatment, and the expression level of PpJAZ7 was significantly lower at 4 or 6 DAT than at 2 DAT (Figure 6). When comparing the effects of different concentrations of exogenous MeJA with the control check (CK), significant differences were found at 2 DAT for PpJAZ7 but not found at 4 or 6 DAT. Moreover, no significant expression level changes were found for PpJAZ3 between the treatment groups and the CK at 6 DAT, which indicated that PpJAZ3 might be not affected at the mRNA level under exogenous MeJA treatment. Nevertheless, the remaining three JAZs, PpJAZ1, PpJAZ4 and PpJAZ5, which showed upregulated expression levels at 6 DAT compared with at 2 or 4 DAT, were also significantly upregulated at the mRNA level under exogenous MeJA treatment compared with CK at 6 DAT, although they were differently affected by different MeJA concentrations (Figure 6). For example, the mRNA abundance of PpJAZ1 was significantly affected by exogenous treatment with 50 μM and 200 μM MeJA compared with CK at 6 DAT; however, the effective MeJA concentrations for PpJAZ4 were 200 and 1000 μM. All the tested concentrations of MeJA (50, 200, and 1000 μM) had significant effects on the expression level of PpJAZ5 at 6 DAT (Figure 6). Overall, four TIFY members, PpJAZ1, PpJAZ4, PpJAZ5 and PpJAZ7, were significantly affected at the mRNA level by exogenous MeJA treatment in the peach epicarp.

Pearson correlation analysis of anthocyanin contents and gene expression levels of TIFY family members under exogenous MeJA treatment was conducted (Supplementary Table 7). The results showed that the expression levels of PpJAZ1, PpJAZ3, PpJAZ4, and PpJAZ5 positively correlated with anthocyanin contents at a significant level. In contrast, the expression levels of PpJAZ7, PpZML1 and PpZML3 negatively correlated with anthocyanin contents with significant p-value. Combined with the previous one-way ANOVA results, it was hypothesized that PpJAZ1, PpJAZ4, and PpJAZ5 were likely induced at the mRNA level by exogenous MeJA treatment, and were involved in JA signal transduction and the regulation of secondary metabolite biosynthesis, such as anthocyanins. Nevertheless, according to a previous report, JAZs respond to JA stimulation mainly at the protein level (Howe and Yoshida, 2019); therefore, it would also be possible for the other TIFY members to play important roles in the JA signaling pathway in peach fruits.

The TIFY (ZIM) domains participate in mediating the homo- and heteromeric interactions between JAZs (Chini et al., 2009), which are also included in the peach PPD and ZML subfamily members (Figure 3). To reveal the protein interaction abilities of peach JAZs and ZMLs, yeast two-hybrid analysis was conducted for the 13 members we cloned, except PpPPD1 and PpJAZ6 (Figure 7). To investigate whether these genes exhibit auto-activation activity, the full-length CDSs of the 13 genes were inserted into the yeast pGBKT7 vector and transformed into the “Y2Hgold” strain. Yeast hybrids of these yeast transformants with the yeast strain “Y187” transformed with empty pGADT7 vector were conducted, and the final results showed that only PpZML2 had auto-activation activity among the 13 members (Figure 7).

Yeast hybrids were conducted between these JAZs and ZMLs pair by pair. All diploid hybrids grew on DDO (SD-Trp-Leu) medium, but only those containing positive dimers grew on QDO/A/X (SD-Trp-Leu-His-Ade/AbA/X-α-Gal) medium. First, we focused on the homodimers. In total, 6 homodimers were found, including PpJAZ1, PpJAZ3, PpJAZ4, PpJAZ5, PpZML4 and PpZML5. Next, for the heterodimers, we found that JAZ proteins seldom interacted with ZMLs. JAZ-JAZ heterodimers, including PpJAZ1/3, PpJAZ1/4, PpJAZ1/5, PpJAZ1/7, PpJAZ3/4 and PpJAZ4/5, were found in reciprocal transformations (i.e., with the pair of constructs either as prey or as bait) of Y2H assays (e.g., PpJAZ1BD × PpJAZ3AD and PpJAZ3BD × PpJAZ1AD). The combinations of “PpJAZ10BD × PpJAZ1AD,” “PpJAZ7BD × PpJAZ5AD,” and “PpJAZ10BD × PpJAZ5AD” were found to be positive in one-way Y2H tests (Figure 7), which was similar to reports in other species, such as Arabidopsis (Chini et al., 2009), Hevea brasiliensis (Hong et al., 2015), cotton (Li et al., 2017), and Mentha canadensis (Xu et al., 2021). For the ZMLs, PpZML3/4, PpZML3/5 and PpZML4/5 were able to form heterodimers, which were confirmed by reciprocal Y2H assays, whereas “PpZML4BD × PpZML1AD” was found to be positive in the one-way Y2H test. It is worth mentioning that Y2H tests of “PpZML3BD × PpZML2AD,” “PpZML4BD × PpZML2AD,” and “PpZML5BD × PpZML2AD” were positive but could not be verified in the opposite direction due to the auto-activation activity of PpZML2. Finally, we found an interesting result in which a protein interaction between PpJAZ3 and PpZML4 was also detected in reciprocal Y2H assay. This result indicated that JAZ and ZML proteins are also able to form heterodimers.

In Arabidopsis, AtJAZs function as repressors of JA signaling by interacting with AtMYC2 (Pauwels and Goossens, 2011). The sequence of AtMYC2 was used as a query to blast against the peach genome to search for its homologs. One peach gene, Prupe.5G035400, that showed the highest similarity with AtMYC2 was cloned and designated PpMYC2. Deduced amino acid sequence alignment of PpMYC2, AtMYC2 and AtMYC2 homologs from apple, pear, strawberry and tomato showed that these MYCs not only shared conserved sequences at the MYC N-terminal region and bHLH domain, but also at the C-terminus (Supplementary Figure 3), implying possibly conserved protein functions. To investigate whether JAZ-MYC2 dimers also exist in peach, yeast two-hybrid-based protein interaction tests were conducted between PpJAZs and PpMYC2 and between PpZMLs and PpMYC2 (Figure 8A). JAZs (not including PpJAZ6) and ZMLs (not including PpZML2 for its auto-activity) were fused with GAL4BD as the bait, and PpMYC2 was fused with GAL4AD as the prey. The empty vectors were used as the controls. After hybridization, all positive diploid transformants for each combination of JAZ/ZML-MYC2 grew well on DDO medium but showed divergent growth status on QDO/A or QDO/A/X media, which indicated that all JAZs tested interacted with PpMYC2, but none of the ZMLs was able to bind with PpMYC2, very likely due to the lack of the Jas domain.

To further validate the protein interactions between JAZs and MYC2 of peach, PpJAZ3 and PpZML4 were selected to test the interaction with PpMYC2 using the split firefly luciferase complementation assay (Figure 8B). Transient over-expression of “PpJAZ3-NLuc + CLuc-PpMYC2” structures combination in tobacco leaves was able to rescue an intense luciferase activity, which was not exhibited in the combination “PpZML4-NLuc + CLuc-PpMYC2” and the empty control groups “PpJAZ3-NLuc + CLuc” and “NLuc + CLuc-PpMYC2” (Figure 8C). The rescued luciferase activity by co-expression of PpJAZ3-NLuc and CLuc-PpMYC2 was also validated by measurement of firefly luciferase activities using a luminometer (Figure 8C). These results indicated that PpJAZ3 but not PpZML4 could interact with PpMYC2, which was in line with the Y2H results.