We downloaded the sequence of 477 RING-containing proteins that have been reported to be present in Arabidopsis (Stone et al., 2005) from The Arabidopsis Information Resource (TAIR)¹ and constructed their domains in Pfam². Six different Hidden Markov model (HMM) motifs were collected. HMM-Blast was applied to the protein sequences in the database downloaded from TPIA³ with HMMer⁴ (Mistry et al., 2013). For the prediction and analysis of the domains of these proteins, we used InterPro⁵ for the initial appraisal. Pfam (see text footnote 2) and SMART⁶ for a secondary verification and determination of redundant domains.

We downloaded the expression levels of these RING family genes under different stresses from TPIA (Supplementary Table 1). The gradient of cold treatment was CA1-6h (10°C for 6 h), CA1-7d (4–10°C for 7 days), CA2-7d (0–4°C for 7 days), and DA-7d [recovery afterward at room temperature (20–25°C) for 7 days] (Wang et al., 2013). Salt and drought treatments were conducted for 24, 48, and 72 h (Zhang et al., 2017). MeJA treatment was applied in gradients of 12, 24, and 48 h (Shi et al., 2015). We calculated the relative ratio of these expression levels to the control, which was normalized by log2 (Supplementary Figure 1). When the expression of H2-type genes under stress at any time is upregulated or down-regulated twice as much as the control (Supplementary Figure 2), we selected and collected it for Venn diagram analysis (Figure 2).

Tea plants (Camellia sinensis var. Sinensis cv. Shuchazao) were collected from the Tea Plant Cultivar and Germplasm Resource Garden in Guohe Town, Anhui Agricultural University. Tea branches having the same growth conditions and divided into four groups, after which they were placed in erlenmeyer flasks containing the same amount of water. After 24 h of adaptation, the water was replaced with 200 mM NaCl solution for Group 1 and with 25% PEG4000 solution for Group 2. Group 3 was treated by smearing 0.25% MeJA on each leaf. Group 4 was treated at 10°C, respectively. The 2nd leaves tissues were collected at seven time points: 0, 6, 12, 24, 36, 42, and 60 h for Group 1; 0, 4, 8, 16, 24, and 36 h for Group 2; at 0, 6, 12, 24, and 36 h for Group 3; 0, 6, 12, 24, 36, 72, and 96 h for Group 4. At each time point, three repeats were performed. Samples were immediately frozen in liquid nitrogen and kept at −80°C for RNA extraction.

Total RNA was extracted using the FastPure Plant Total RNA Isolation Kit (polysaccharides and polyphenolics-Rich) (Vazyme Cat. RC401-01) according to the manufacturer recommendations. The cDNA were synthesized using PrimeScript RT Reagent Kit Perfect Real Time (TaKaRa, Dalian, China; Code: DRR037A) and stored at −20°C for later use. Verifying PCR product-amplification specificity was determined using the fusion curve (55–95°C). The housekeeping gene was glyceraldehyde-3-phosphate dehydrogenase. The RT-qPCR mixture consisted of 10 μL of CHAMQ SYBR qPCR mixture (Vazyme), 7.4 μL of ddH2O, 0.8 μL of upstream and downstream primers, and 1 μl of cDNA. RT-qPCR reacted in the 96-well optical reaction plates at 95°C for 30 s, followed by 40 cycles at 95°C for 5 s and reaction at 60°C for 30 s PCR product amplification specificity was determined using the melting curve (55–95°C). Three biological replicates and three experimental replicates were applied for each sample. The relative expression values were calculated using the 2^–ΔΔCt method. In each group, untreated branches were used as control. All primers used for RT-qPCR analysis were recorded in Supplementary Table 4.

The steps of subcellular localization analysis were referred to Yoo et al. (2007), and some modifications were made. CsMIEL1 was constructed on pUC19-GFP vector and transformed into Arabidopsis protoplasts of leaves.

The entire CsMIEL and CsMIEL1-C192S open reading frame was cloned into the pEGX vector and expressed in Escherichia coli, the primers are displayed in Supplementary Table 5. The in vitro ubiquitination assays were performed as described elsewhere (Zhao et al., 2012). For the E3 ubiquitin ligase activity assay, recombinant wheat (Triticum aestivum) E1 (GI: 136632), human E2 (UBCH5B; 100 ng), and purified Arabidopsis ubiquitin with HIS-tag (UBQ14, AT4G02890; 500 ng) were mixed and used for the assay. The mixture was incubated at 30°C for 2 h, boiled at 100°C for 5 min, and frozen at −20°C until the SDS-PAGE analysis. After Western blotting, the reactants isolated by SDS-PAGE were incubated with GST antibody (1:10,000) and observed using ECL chromogenic solution.

Multiple sequence alignments of the full-length RING proteins were performed by DNAMAN (Version 6.0; Lynnon Corporation, Quebec City, QC, Canada) with default parameters. The phylogenetic tree was constructed with molecular evolutionary genetics analysis (MEGA) software (Version 6.0)⁷ (Tamura et al., 2013), using the neighbor-joining (NJ), minimal evolution (ME), and maximum parsimony (MP) methods and the bootstrap test carried out with 1,000 iterations to test the significance of the nodes.

Arabidopsis ecotype Columbia (Col-0) plants were grown in Murashige and Skoog (MS) media at 22°C in long day conditions (16 h of light and 8 h of darkness) and used as wild types and for genetic transformation and other analyses. The Agrobacterium tumefaciens GV3101 strain was grown in LB media supplemented with 50 μg/mL kanamycin and 20 μg/mL rifampicin. The CsMIEL1-MYC and CsMIEL1-C192S-MYC construct consisted of the coding sequence under the control of a 35S promoter. Transgenic Arabidopsis thaliana organisms were generated using floral-dip transformation (Clough and Bent, 1998). All T0 generation seeds were sown on MS medium containing 15 μg/mL glyphosate for selection. RNA was extracted and reverse transcribed into cDNA for semi-quantitative PCR, and finally four CsMIEL1-MYC transgenic lines and CsMIEL1-C192S-MYC transgenic lines were obtained. Among them, Line 2, 4, and 6 of CsMIEL1 have high expression level, and Line 8 is low. Line 2, 3, and 4 of C192S are high, and Line 1 has low expression level (Supplementary Figure 3). Through the glyphosate selecting of each generation until the T3 generation, PCR verification assays have confirmed that the overexpressing line is homozygous.

Wild-type, CsMIEL1-MYC, and CsMIEL1-C192S-MYC overexpressing Arabidopsis thaliana were sown simultaneously in Petri dishes containing the same volume of MS medium under common growth conditions. After 10 days of pre-culture, the seedling of the same size was selected and treated on the MS medium with 100 mM NaCl (Zhang et al., 2008; Kuki et al., 2020), 200 mM mannitol (Malefo et al., 2020), 100 μm of MeJA (Leon-Reyes et al., 2010), and 10°C (Zhang et al., 2008) for another 10 days. Plants were grown for 10 days under normal conditions after pre-culture as controls (Mock). In all biological replicates, at least 50 plants were used for each treatment, and the error bars represent the standard deviation of replicates.

Extraction of anthocyanin was extracted according to Jiang et al. (2017) with some modifications. In total, 0.2 g of plant tissue was quick-frozen in liquid nitrogen for preparation for subsequent experiments. During extraction, plant tissues were fully ground with a ball mill and extracted four times through the addition of a 500 μL extraction buffer. The ratio of the buffer was 80% methanol, 1% hydrochloric acid, and 19% water. After each extraction, the homogenate was centrifuged at 12,000 rpm for 10 min and combined. The absorbance values were measured at 530 nm, and each genotype was repeated at least three times.

The plants were stripped from the MS medium, and various growth indicators were measured manually. We recorded the data of all biological replicates for calculating the average and standard deviation. The measurement was all done by one person to control the error.

For statistical analyses, IBM SPSS Statistics version 19 were use. The Tukey’s and least significant difference (LSD) multiple comparison tests were used to analyze significant differences between pairs. Differences were considered statistically significant when ^∗P < 0.05 and ^∗∗P < 0.01. All the results were based on the average of three parallel experiments.

We downloaded the sequence of the RING family in Arabidopsis for analysis of this huge family in the tea plant. After deduplication and comparison, 335 RING-contain proteins were identified. On the basis of the differences in amino acid sequences at these binding sites and the number of amino acids between each site, these 335 domain in tea plant were divided into six categories: RING-H2, RING-HCa, RING-HCb, RING-C2, RING-v, and RING-G. In all types, the amino acid residues at positions 1, 2, 3, 6, 7, and 8 were conserved and consisted of Cys, whereas the amino acid residues 4 and 5 were not conserved. According to the aforementioned classifications, 64% (214) of the RING domain in tea plant were distributed into the RING-H2 group. The metal-ligand positions 4 and 5 are two His residues, which accounted for the name of the RING-H2 group. According to the characteristics of the amino acid residues of metal ligands 4 and 5, the remaining RING types were classified into RING-HCa, HCb, C2, V, and G groups. The differences between HCa and HCb groups corresponded to the number of amino acids between positions 7 and 8. Between metal ligands 7 and 8, the HCa group comprised two and four amino acids, respectively. Few members belonged to the HCb and G groups (only three members and one member, respectively). Similar to Arabidopsis, most proteins were the H2 type followed by the HCa type. In contrast to Arabidopsis thaliana, among the 477 RING proteins in Arabidopsis thaliana, 41 (8.6%) HCb-type proteins were present, whereas only 3 HCbs were present in the tea plants. We downloaded data from the Grape Genome Browser⁸ database and performed the same manual search and identification of the RING domain in grapes to compare the RING-containing protein classifications of different species (Supplementary Table 2). Apple’s data in the table comes from previous reports (Zhang et al., 2008). Among the four species, the distribution of RING domains was consistent; the H2-type domains were the most numerous followed by the HCa-type domains. The number of V-type and C2-type domains in each species was nearly identical, but the number of HCb-type domains differed substantially. Research has speculated that the RING-D domain may only exists in Arabidopsis thaliana (Stone et al., 2005). Accordingly, in our search of tea plants and grapes, we did not identify this type of RING domain. But in Apple, a D-type RING domain was found (Zhang et al., 2008). However, no C2-type grape RING domains were found in this selection. Perhaps due to the problem of genome assembly and annotation, the identification in tea plant did not involve S/T-Type, D-type, mH2-type and mHC-type, and the structure of these types was not explained in detail (Figure 1).

We predict other motifs of RING finger proteins in tea plants. The results indicated that 235 members exhibited no other motifs and contained only the RING finger domain. An additional 17 domains were identified in the remaining RING-containing proteins, such as the WD40 domain, IBR domain, and other motifs with unknown functions (Supplementary Table 3). The WD40 domain is cited as a binding module for pSer/Thr. Additionally, when present in ubiquitin ligase, the WD40 domain may also be related to the phosphorylation of the relevant substrate (Deshaies, 1999; Koepp et al., 1999; Nakayama et al., 2000). The IBR domain is a common Cys-rich region located between two RING domains, but its function is unknown (van der Reijden et al., 1999). The identification of numerous domains suggests that the RING-containing proteins in tea plants may have multiple regulation or regulated mechanisms and perform diverse functions.

To figure out the RING finger genes related to stress responses in tea plants, we analyzed the induced expression difference data of 335 genes provided by TPIA under different stresses. Abiotic stresses included cold, salt, and drought treatments. Responses to biological stress were simulated by detecting the response of plants to exogenous methyl jasmonate (MeJA). Differentially expressed H2-type RING genes were individually compared and analyzed, and the results are presents in Supplementary Figure 2. Figure 2A reports the number of genes responding to various stresses. In total, 128, 134, 121, and 82 genes responded to cold stress, salt stress, drought stress, and exogenous MeJA treatment, respectively. A Venn diagram indicates that 53 genes responded to all stresses (cold stress, drought stress, salt stress, and MeJA treatment), and 31 genes respond to only one form of stress. A total of 26 genes responded only to salt stress, and five responded only to cold stress. However, no gene responded only to drought stress or MeJA treatment (Figures 2B,C). These findings are reported in Figure 2D.

To verify the response of these genes to stress, we treated the branches of tea plants with a low degree of lignification with NaCl, PEG4000, a low-temperature condition, and exogenous MeJA to simulate the four selected stresses. After the treatments were completed, RNA was extracted from the second leaf and reverse transcribed into cDNA for Quantitative real-time RT–PCR analysis (RT-qPCR) (Figure 3A). In the process of designing and RT-qPCR primers, we found that for many genes, the target fragments could not be cloned, or the amplified products were not specific, and most genes could not design primers with an efficiency of 90–110%. This may be caused by genome assembly and annotation issues. Therefore, we selected five genes that met the requirements of amplification efficiency for RT-qPCR: TEA031033, TEA023239, TEA28518, TEA030031, and TEA018024. The results of the RT-qPCR analysis indicated that gene responses to salt and MeJA were relatively low (Figures 3B,E), but responses to drought and cold were more substantial (Figures 3C,D). Among the selected genes, TEA031033 had the strongest response when treated with NaCl for 12 h: the response was 3.28 times that of the control (Figure 3B). After 24 h of MeJA treatment, the expression level of TEA031033 was approximately three times that of the control (Figure 3E), which also represented the most considerable gene response to exogenous MeJA. TEA030031 and TEA018024 responded significantly to cold stress; their strongest response to cold stress were 9.6 and 6.6 times that of the control, respectively (Figure 3C), but TEA030031 had a weak response to salt, drought, and MeJA treatment. Under drought stress, the strongest response values of TEA28518, TEA031033, and TEA018024 were 8.23, 7.55, and 4.8 times that of the control, respectively. These five genes exhibited different levels of response to biotic and abiotic stresses, but TEA031033 had a relatively high level of response to all four types of stress. Therefore, we conducted an in-depth investigation on this gene to identify its specific function in tea plants.

We cloned TEA031033 and compared its sequence with that of other species. Our results indicated that TEA031033 has a RING finger domain at the C-terminus and a Zinc-finger domain at the N-terminus (Figure 4A). The phylogenetic analysis results indicated that AtMIEL1 had the highest homology with TEA031033 (Figure 4B). Consequently, TEA031033 was named CsMIEL1. Subcellular localization results show that CsMIEL1 is localized in the nucleus (Supplementary Figure 4).

First, to verify the specific function of CsMIEL1 in tea plants, we investigated the potential presence of E3 ubiquitin ligase activity, an in vitro ubiquitination assay was conducted. We cloned the open reading frame of the CsMIEL1 and C192S site-directed mutagenesis, the 192th residue of C192S was changed from Cys to Ser (Figure 5A). Polyubiquitinated CsMIEL1 were found in the presence of ATP, ubiquitin, E1, E2, and CsMIEL1-GST fused proteins using an anti-GST antibody, an absence of any factor had cause CsMIEL1 to not be ubiquitinated. However, even in the presence of all the auxiliary factors, no polyubiquitin CsMIEL1-C192S was detected in assays (Figure 5B). As a result, we concluded that the RING finger family protein CsMIEL1 exhibited E3 ubiquitin ligase activity and that this ubiquitination activity was lost after mutation at position 192. Moreover, we concluded that the complete RING domain is necessary to preserve the E3 ubiquitin ligase activity of the RING finger protein.

To investigate the functions of CsMIEL1, we obtained 35S: CsMIEL1-MYC and 35S: CsMIEL1-C192S-MYC overexpressing Arabidopsis plants. The four selected stress treatments were performed on the wild-type, CsMIEL1-MYC, and CsMIEL1-C192S-MYC, respectively (Figure 6A). For the transgenic plant seedlings and wild-type seedlings grown at room temperature for 10 days without any exogenous hormones, no significant differences existed across the growth indicators, including leaf length (LL) and width (WL), hypocotyl length (LH), and root number and length (Figure 6B).

Under salt stress conditions, the growth of both transgenic plant seedlings and wild-type (WT) seedlings was inhibited. The growth of CsMIEL1-MYC transgenic plants was less pronounced than that of wild-type and CsMIEL1-C192S-MYC plants, which was reflected in the significantly longer length of the petiole (LP) and primary root (LPR) compared with the control (Figure 6C). Similarly, under drought stress, the growth of the transgenic plant and wild-type seedlings was also inhibited, but the length of the primary root (LPR) and hypocotyl (LH) and the number of lateral roots (NL) of CsMIEL1-MYC transgenic plants were significantly greater than those of the control and CsMIEL1-C192S-MYC (Figure 6D).

The effect of cold treatment on the growth of all seedlings was not particularly notable. The difference in growth indicators between CsMIEL1-MYC transgenic plants and WT did not reach a significant level, although its primary root length (LPT) was lower than these of wild-type and CsMIEL1-C192S-MYC transgenic plants (Figure 6F). Under low-temperature conditions, the anthocyanin levels of CsMIEL1-MYC transgenic plants were significantly lower than these in the control and CsMIEL1-C192S-MYC transgenic plants (Figures 6G,H).

The result in Figure 6E showed that CsMIEL1-overexpressing plants were hypersensitive to MeJA and that the growth of the root system of the CsMIEL1-MYC transgenic plant was significantly inhibited. Additionally, after these three genotypes were treated with exogenous MeJA, the primary root length (LPT), hypocotyl length (LH), and lateral root number (NL) of the CsMIEL1-overexpressing plants were shorter than those of the wild-type and CsMIEL1-C192S-MYC transgenic plants.