One-year-old grafted tree peony plants (Paeonia suffruticosa ‘Yulouchun’), obtained from Luhe Tree Peony Planting Professional Cooperative (Heze, China), were used as the main experimental materials in this study. The plants were planted in plastic pots (16-cm height, 14-cm top diameter, and 11-cm bottom diameter) filled with a soil mixture containing peat moss and perlite (2:1, by vol.) in late September. Each pot was irrigated with 300 mL of tap water containing an appropriate concentration of complex fertilizer (N:P:K = 2:1:1, by wt.) once per week. They were kept in an outdoor environment (a temperature range from 12°C to 21°C, relative humidity of 52%–80%, and natural photoperiod) until early November when the bud dormancy was completely established. The plants were transferred to a cold chamber at 2°C for chilling treatment. The apical buds treated with low temperature at intervals (0 days, 5 days, 10 days, 15 days, 20 days, 25 days, and 30 days) were used for gene expression analysis. The buds at 25 days of chilling treatment, near the time of a complete endodormancy release, were used for a tobacco rattle virus (TRV)-based virus-induced gene silencing (VIGS) experiment. Petunia seeds (Petunia hybrida ‘Mitchell Diploid’), obtained from Goldsmith Seeds (Gilroy, CA, USA), were sown in a tray containing the same soil mixture. The leaflets were harvested as the explants for Agrobacterium-mediated stable transformation (Ji et al., 2023). Wild-type (WT) and transgenic petunia plants were grown in the plastic pots (11-cm height, 11-cm top diameter, and 9-cm bottom diameter) and irrigated as mentioned above. Tree peony plants upon VIGS assay and petunia plants were maintained in a growth chamber at 22°C day/20°C night with a 16-h light/8-h dark photoperiod.

Based on the RNA-Seq data of tree peony buds, the cDNA sequence of PsATL33 containing a 483-bp coding region was isolated. The translation of nucleotides into amino acids was conducted using the ExPASy tool (http://web.expasy.org/translate/). Phylogenetic tree analysis was carried out using MEGA4 software (version 4.0.2). The RING-H2 conserved domain was identified according to a previous report (Serrano et al., 2006). The modeling process of the RING-H2 domain was performed using a Modeling server (version 9.20) based on the sequence alignment. A similar protein structure (Protein Data Bank: 1X4J) of PsATL33 was used as the template. The model was evaluated through Discrete Optimized Protein Energy (DOPE) values and GA 341 scores and finally visualized using the PyMOL tool (version 2.5.4).

The coding region of PsATL33 was amplified with the stop codon removed to construct the pCAMBIA2300-PsATL33–green fluorescent protein (GFP) fusion expression vector. The expression vector containing PIP2A-mCherry was used as the control to mark the plasma membrane. The fusion expression vectors were transformed into Agrobacterium tumefaciens GV3101 through the freeze–thaw method. The Agrobacteria transformed with PsATL33-GFP and PIP2A-mCherry were mixed in equal proportion and injected into 4-week-old Nicotiana benthamiana leaves. The reagent 4′,6-diamidino-2-phenylindole (DAPI) was used to mark the nucleus. The fluorescence was observed under a laser scanning confocal microscope (TCS SP8 SR; Leica, Wetzlar, Germany). GFP, mCherry, and DAPI were excited using 488-, 561-, and 405-nm lasers and detected after passing through 500–560-nm, 590–620-nm, and 410–492-nm band-pass filters, respectively.

To examine the effects of exogenous hormones and abiotic stresses on expression profiles of PsATL33, tree peony buds before artificial chilling treatment were used. For exogenous hormone treatments, the buds were sprayed with solutions containing 100 μM GA₃, 100 μM ABA, 100 μM salicylic acid (SA), 50 μM ethephon (ETH), and 100 μM methyl jasmonate (MeJA). The control buds were treated with deionized water. For abiotic stress treatments, the plants were irrigated with 20% PEG6000 and 100 mM NaCl or placed in a warm room (37°C) and a cold room (−4°C). For each case, the buds with three biological replicates were collected at 0 h, 6 h, 12 h, 24 h, 36 h, and 48 h after treatment. They were frozen in liquid nitrogen and stored at −80°C.

Total RNAs of tree peony buds and petunia leaves were extracted using RNAprep Pure Plant Kit (Tiangen, Beijing, China). To remove DNA contamination, RNA samples were treated with RNase-free DNase I (Promega, Madison, WI, USA) at 37°C for 30 min. RNA concentration and quality were evaluated via 1.2% agarose gel electrophoresis and a spectrophotometer (NanoDrop, Wilmington, DE, USA). First-strand cDNA was synthesized from 2–5 µg of RNA samples using PrimeScript RT reagent (Takara, Otsu, Shiga, Japan). Reverse transcription–quantitative PCR (RT-qPCR) analysis was performed using the synergy brands (SYBR) Green Reagent in a LightCycler480 Real-Time PCR System (Roche Diagnostic, Basel, Switzerland). PsActin and PhEF1α were used as the reference genes in tree peony and petunia, respectively. The primer pairs were designed using Primer3 Input (version 0.4.0) and shown in Supplementary Table S1 . Relative expression levels of the genes were calculated as previously described (Livak and Schmittgen, 2001). Three biological replicates were used in this experiment.

The samples were ground into powder in liquid nitrogen and extracted with 80% (v/v) methanol and 1 mM butyl hydroxytoluene. The extract was transferred into a tube with 20 mg of polyvinylpolypyrrolidone and fully mixed. The mixture was centrifuged at a relatively low temperature at 5,000 rpm for 20 min. Under the condition of 40°C, the extract was almost concentrated into a water phase containing ammonia. After filtration with a 0.45-µm filter, the samples were dried by vacuum freezing. The particles were then dissolved in 50% (v/v) methanol and analyzed through high-performance liquid chromatography using an Agilent chromatograph (Model 1100, Agilent Technologies, Santa Clara, CA, USA). The detection wavelength used in this assay was 210 to 280 nm, and the flow rate was 1 mL/min. The standard hormones were purchased from Sigma-Aldrich (St Louis, MO, USA). The peak areas were analyzed to quantify the levels of bioactive GAs and ABA. Three biological replicates were used for each hormone measurement.

The coding region of PsATL33 was PCR-amplified and cloned into a modified pCAMBIA1300 vector to generate the overexpression construct (Sun et al., 2020). The recombinant plasmid was transformed into A. tumefaciens GV3101. The leaf disc method was used for stable genetic transformation of petunia plants (Yuan et al., 2024). Briefly, young leaves from petunia ‘Mitchell Diploid’ plants were collected and cut into 1 cm × 1 cm leaf discs, which were infected with Agrobacteria harboring the recombinant plasmids. The inoculated leaves were placed on a co-cultivation medium at room temperature for 2 days and then transferred to a fresh regeneration medium for the selection of positive transformants. Regular PCR amplification was conducted when the plants reached the four-leaf stage to confirm the integration of PsATL33. Transcript abundances of PsATL33 in leaves from transgenic petunia lines were examined by RT-qPCR. Three lines with higher transcription of PsATL33 were selected for functional investigation.

To generate the TRV-PsATL33 construct, a 347-bp fragment of PsATL33 was PCR-amplified and introduced into the TRV2 vector between KpnI and XhoI sites. TRV2 empty vector and TRV-PsATL33 plasmids were transformed into A. tumefaciens GV3101. The inoculum was prepared according to a previous description (Mao et al., 2022). The transformed Agrobacteria were cultured in Luria-Bertani (LB) media (40 mg/L kanamycin, 20 mg/L gentamicin, 10 mM MES, and 20 μM acetosyringone) at 28°C for 48 h. When the OD600 reached 2.0, the cells were harvested and resuspended in the infiltration buffer (10 mM MgCl₂, 10 mM MES, and 200 μM acetosyringone). After a gentle shaking for 3 h, Agrobacterium cultures bearing TRV1 and TRV2 empty vectors or TRV-PsATL33 were mixed together in equal volumes. The bacterial solution was used to inoculate tree peony buds under a vacuum pressure at 0.7 MPa for 20 min. The inoculated plants were transplanted into the soil mixture and maintained in a cold room at 10°C for 2 days, which would help increase viral accumulation. Next, the plants were transferred to a growth chamber for phenotype observation. Three tree peony plants were used for each inoculation.

To investigate the function of PsATL33 in response to freezing stress, tree peony plants inoculated with TRV empty vector and TRV-PsATL33 were placed in a cold room at −4°C with light illumination. A normal condition at 22°C was used as the control. After cold treatment, the survival rate of tree peony buds was recorded at 0 and 6 h post-freezing. The buds were collected at the same time points for measurement of malondialdehyde (MDA) content and ion leakage rate. Three biological replicates were used with a pool of three seedlings per replicate. The detection of MDA content was performed as previously described (Shin et al., 2012). In brief, the bud tissues were homogenized with 0.1% (w/v) trichloroacetic acid on ice. The mixture was centrifuged at 12,000 rpm for 10 min at 4°C. The resulting supernatant was evenly blended with 0.25% (w/v) thiobarbituric acid. The mixture was heated in a water bath at 95°C for 10 min and cooled rapidly to 4°C for further centrifugation at 4,500 rpm for 10 min. The supernatant was detected by measuring A₄₅₀, A₅₃₂, and A₆₀₀ values, with deionized water being the blank control. MDA content was analyzed based on the following formula: 6.45 (A₅₃₂ − A₆₀₀) − 0.56 A₄₅₀. The ion leakage rate was measured following a previously described method (Wu et al., 2017a). First, the buds were soaked in 0.4 M mannitol at 20°C for 3 min, and the initial conductivity was measured using a meter (Leici, Shanghai, China). The samples were heated in a water bath at 85°C for 20 min, and the total conductivity was measured thereafter. The ion leakage rate was determined as the percentage of initial conductivity to total conductivity. Three biological replicates were used for each measurement.

All experiments reported here were performed using three biological replicates. Statistical significance was evaluated through Student’s t-test at p-values <0.05 and <0.01 using JMP software (version 11.0; SAS Institute Inc., Cary, NC, USA).

To investigate the molecular mechanisms of bud dormancy break in tree peony, a 783-bp transcript encoding a putative RING-H2 finger protein, designated PsATL33, was identified from transcriptome data of chilling-treated tree peony buds (Yuan et al., 2024). Sequence analysis revealed that its cDNA contains a complete coding region of 483 bp, encoding a polypeptide of 161 amino acids ( Supplementary Figure S1 ). PsATL33 was phylogenetically close to AtATL33 from Arabidopsis thaliana, PvRNF167 from Pistacia vera, and other ATL33s from Vitis vinifera, Glycine max, Medicago truncatula, and Camellia sinensis ( Figure 1A ). The C-terminuses of these amino acid sequences shared a conserved RING-H2 domain. The characteristic sequence of this domain is Cys-2X-Cys-14–15X-Cys-1X-His-2X-His-2X-Cys-10X-Cys-2X-Cys (X represents any amino acids except Cys and His), which belongs to the C3H2C3-type RING-H2 domain ( Figures 1B, C ). To study the subcellular localization of PsATL33, the fusion protein PsATL33-GFP was transiently expressed in tobacco leaves. DAPI staining and co-expression with PIP2A-mCherry were used as nucleus and plasma membrane markers, respectively. The results indicate that PsATL33 is localized into the nucleus and plasma membrane ( Figure 1D ).

To understand expression patterns of PsATL33, RT-qPCR analysis was performed using tree peony buds. It was found that expression levels of PsATL33 initially increased during chilling-induced bud dormancy release, followed by a slight decrease in the last 10 days ( Figure 2A ). Transcription of PsATL33 increased significantly after GA₃ treatment but decreased after ABA treatment ( Figure 2B ). The effects of several stress-associated hormones and abiotic stresses on PsATL33’s expression were also examined. ETH and MeJA treatments resulted in a significant increase in the expression of PsATL33, but there was no significant change after SA treatment ( Figure 2C ). In addition, transcript levels of PsATL33 increased following treatments with drought, salinity, and freezing rather than heat ( Figure 2D ). These data indicate that PsATL33 is responsive to dormancy- and stress-related hormonal or environmental signals.

To study the function of PsATL33 in bud dormancy, we carried out a heterologous transformation experiment in petunia (P. hybrida). Compared with WT plants, PsATL33-overexpressing transgenic plants exhibited accelerated seed germination ( Figure 3A ). RT-qPCR results confirmed the constitutive expression of PsATL33 in transgenic petunia lines ( Figure 3B ). A shorter seed germination time was found in transgenic lines overexpressing PsATL33 than that in the WT line ( Supplementary Figure S2A ). Overexpression of PsATL33 resulted in increased petunia plant height and internode length in the following growth periods ( Figures 3C, D ; Supplementary Figure S2B ). Its overexpression also led to enlarged leaf size in petunia ( Supplementary Figure S3 ). In addition, PsATL33-overexpressing transgenic plants displayed promoted growth of axillary buds ( Figure 3E ). The number and length of axillary buds in transgenic petunia lines were significantly higher than those in WT control ( Figures 3F, G ). Increased production of bioactive GAs, GA₁ and GA₃, was found in PsATL33-overexpressing transgenic lines ( Figure 3H ). These findings suggest that overexpression of PsATL33 promotes the dormancy release of petunia seeds and axillary buds.

To further verify the function of PsATL33 in bud dormancy, we performed a TRV-based VIGS experiment. Compared with empty vector control, tree peony plants inoculated with TRV-PsATL33 showed delayed bud break and growth at 2 and 3 weeks after inoculation ( Figure 4A ). RT-qPCR results showed that expression levels of PsATL33 in the buds inoculated with TRV-PsATL33 were significantly reduced in comparison to the control, suggesting a successful silencing of PsATL33 in the buds ( Figure 4B ). Consistent with the phenotypes, the PsATL33-silenced buds showed significantly reduced bud break rate and plant height ( Figures 4C, D ). Transcript levels of two bud dormancy break-associated genes, EARLY BUD-BREAK 1 (PsEBB1) (Zhang et al., 2021) and CARBOXYLESTERASE 15 (PsCXE15) (Huang et al., 2008), were dramatically lower in PsATL33-silenced buds than that in empty vector control ( Figure 4E ), indicating that silencing of PsATL33 inhibits tree peony bud dormancy release.

Given the essential roles of GAs and ABA in the regulation of bud dormancy, we detected their contents in tree peony buds upon VIGS assay. In comparison to empty vector control, the accumulation of bioactive GAs, GA₁ and GA₃, in TRV-PsATL33-infected buds was significantly reduced at different weeks after infiltration, and notably, the decrease of GA₃ accumulation was more significant than that of GA₁ ( Figure 5A ). However, no significant alteration in ABA accumulation was found in the buds infiltrated with TRV-PsATL33 ( Figure 5B ).

To further study the role of PsATL33 in mediating the GA pathway, we analyzed the transcription of some genes related to GA biosynthesis and signal transduction in the buds. RT-qPCR analysis showed that expression levels of GA biosynthetic genes GA20-OXIDASE 1 (PsGA20ox1), GA3-OXIDASE 1 (PsGA3ox1), and PsGA3ox3 decreased significantly in PsATL33-silenced buds compared to the control, whereas transcript abundances of GA catabolic gene GA2-OXIDASE 1 (PsGA2ox1) and GA signal transduction repressor GA-INSENSITIVE 1A (PsGAI1A) increased ( Figure 6 ). No significant difference in transcript levels of other GA pathway-related genes, ENT-KAURENE OXIDASE (PsKO), ENT-KAURENOIC ACID OXIDASE 1 (PsKAO1), GA-INSENSITIVE DWARF 1A (PsGID1A), and RGA-LIKE 1 (PsRGL1), was observed in the buds infected with empty vector and TRV-PsATL33 ( Figure 6 ). These results demonstrate that PsATL33 affects bud dormancy release by modulating GA production.