Four-year-old tree peony plants (P. suffruticosa ‘Lu He Hong’) were obtained from the Tree Peony Research Institute of Qingdao Agricultural University, Qingdao, China, and planted in pots with a diameter of 38 cm and a height of 32 cm on Oct 15, 2019. The plants were moved to a 0–4°C dark refrigerating chamber on Nov 12 when the buds entered dormancy and underwent variable days of chilling (DC) (0, 7, 14, 21, and 28 DC). The chilled buds were collected for transmission electron microscopy (TEM) observation and gene expression analysis. The morphological characteristics of the chilled plants were evaluated to assess the dormancy status after being transferred to a greenhouse (18-22°C, 16-h-light/8-h-dark cycle) as described previously (Xin et al., 2019). In the experiment, 0-7 DC was a chilling perception period with less than 10% bud burst, 14-21 DC was a transition to dormancy release, 21 DC was completely dormancy release with all apical buds burst in greenhouse, and 28 DC was an ecodormancy state.

After this, 0.6 mM exogenous GA₃ and GA₄ were applied respectively, and the samples treated with sterile distilled water were used as a control group. Apical buds were harvested after 48 h for reverse transcription quantitative PCR (RT-qPCR) analysis. Three biological repeats were set for each treatment, and no less than nine plants were used per replicate.

The calcium-blocking treatments were performed by spraying the calcium channel blocker lanthanum trichloride (LaCl₃) or the calcium chelator ethylene glycol-bis (β-aminoethylether)-N, N, N´, N´-tetraacetic acid (EGTA) to buds of tree peony on Nov 11. Then the recovery treatments were executed by application of Ca²⁺ solution after the removal of LaCl₃ or EGTA solution by ddH₂O washing. Nine tree peony plants were included in each treatment and set up in triplicate.

For the blocking treatments, the buds of tree peony were sprayed with 10 mM and 30 mM EGTA or LaCl₃, respectively, and then washed with double-distilled water (ddH₂O) after 12 h. For the recovery treatments, the buds treated with LaCl₃ or EGTA were sprayed with 10 mM or 30 mM CaCl₂ solutions after being washed with ddH₂O. After 12 h, all the treated plants were moved to a refrigeration room (0-4°C) for 21 d, and then transferred to the greenhouse (18-22°C, 16 h light and 8 h darkness). The percentages of bud break were monitored daily to assess the effect of the treatments on dormancy release. Tree peonies sprayed with H₂O alone served as the control.

In the blocking treatments, the tree peony buds were sprayed with 30 mM LaCl₃ or EGTA and washed with ddH₂O as described above, and then 0.6 mM GA₃ was applied. For the recovery treatments, 10 mM or 30 mM CaCl₂ was sprayed after treated with LaCl₃ or EGTA, along with GA₃. Subsequently, all the GA-related plants were immediately moved to the greenhouse, and the bud break rate was monitored. The plants treated with 0.6 mM GA₃ alone served as the control. Three replicates were performed as described above.

The differences of morphological data were analyzed using Duncan’s multiple range tests at a significance level of 0.05 using SPSS 13.0 for Windows (SPSS, USA).

Buds were collected at 11:00 a.m. from plants treated with different chilling days (0, 7, 14, 21, and 28 DC). The leaf primordia from the same location in different samples were cut into a cube with a volume of approximately 1 mm³ and prepared for calcium localization, as described by Jian et al. (1997) with minor modifications. The samples were immersed in a fixative solution containing potassium pyroantimonate, glutaraldehyde, and paraformaldehyde in potassium phosphate buffer, post-fixed by osmium tetroxide, dehydrated in a graded series of acetone, and embedded in Embed_812 (EMS, New Jersey, USA). The embedded samples were then sectioned with an Ultra microtome, and the sections were 60-70 nm thick.

After uranyl acetate double staining, transmission electron microscopy images about cell structure were acquired using a HITACHI 7700 microscope (HT7700). In order to verify the location of calcium, an additional chelation of calcium ion with EGTA treatment was performed (Wick and Hepler, 1982). The grids were immersed in a 200 mg/L EGTA solution, incubated at 60°C for 1 h, and subsequently rinsed briefly in distilled water. After EGTA treatment, the samples were restained with uranyl acetate and reexamined using TEM.

Total RNA was extracted using the TRIZOL reagent (Qiagen) following the manufacturer’s instructions, and RNA integrity was checked by an Agilent Bioanalyzer 2100 (Agilent technologies, US). Chromosomal DNA was removed with RNase-free DNase (Fermentas, USA). First-strand cDNA was synthesized from 2 μg of total RNA using the PrimerScript™ RT reagent Kit (Takara, Dalian, China) according to the manual.

Differential expressed calcium-sensor-encoding genes were screened from the tree peony transcriptional database of chilling duration (accession to the GEO data: GSE4004) and GA- induced dormancy release (accession to the SRA data: PRJNA720276) (Gai et al., 2013; Zhang et al., 2021).

The expression patterns of target genes during dormancy release and chemical substance treatment were analyzed using RT-qPCR. The PCR reactions were performed in 25 μL volume, containing 12.5 μL of 2× SYBR Green Master mix (Takara), 0.75 μL of each primer, 9 μL of ddH₂O, and 2 μL of 10× diluted cDNA template. The PCR reactions were conducted in a Roche LightCycler^® 480 (Roche, Germany) using the following program: 95°C for 2 min, followed by 45 cycles of 95°C for 5 s, 57°C for 30 s, and 72°C for 30 s. Actin was used as a reference gene to normalize the RT-qPCR results. The reactions were performed in triplicate. The relative expression of each gene was quantified using the 2^-ΔΔCt method as described by Livak and Schmittgen (2001). Significance was tested using SPSS 13.0 for Windows (SPSS, USA). The primers used for qPCR are listed in Supplementary Table S1 .

Most of the control buds chilled for 21 and 28 days burst after two weeks in the greenhouse and eventually blossomed, while only some of the buds chilled for 0-14 days burst in two weeks. The state after 21 d indicated a dormancy release stage, and that of 28 d indicated ecodormancy status.

The calcium blocking treatments EGTA and LaCl₃ delayed first bud burst and reduced bud break rates, with the most inhibitory effect obtained with 30 mM EGTA treatment application before 18 d. On average, the buds treated with 30 mM EGTA required an additional day to reach maximum burst. The control buds were on full flushing 18 d after being transferred to the greenhouse, while the bud break rates of 10 mM EGTA-, 30 mM EGTA-, and 30 mM LaCl₃-treatments were 78.43%, 71.70%, and 71.93%, respectively ( Figure 1A ). Despite the delays, the buds of all treatments eventually burst.

Recovery treatments with both 10 mM and 30 mM Ca²⁺ solution promoted bud burst before 19 d, reducing 1.2 d compared to the buds in the 30 mM EGTA treatment ( Figure 1B ). Comparatively, the recovery effect of 30 mM Ca²⁺ was superior to that of the 10 mM Ca²⁺ solution. At 18 d, the bud break rate of the 10 mM Ca²⁺-treated group was 84.91%, while that of the 30 mM Ca²⁺- group was almost full flushing, nearly identical to that of the H₂O control from 14 d to 20 d after being moved to the greenhouse, with the only exception being a 1.1 d delay at beginning of bud burst compared to the control ( Figure 1C ).

Bud break first occurred after 4 d of GA₃ feeding, which was about 9 d earlier than that with chilling treatment alone ( Figures 1 , 2 ). The buds treated with EGTA and LaCl₃ showed delayed bud burst and declined breaking percentages for up to 9 days in the greenhouse, and the buds of all treatments eventually burst. After 9 days in the greenhouse, the mock was full flushing and the breaking rate of 10 mM EGTA was 93.01% with no significant difference to that of the mock, while that of 30 mM EGTA dramatically decreased with a percentage of 83.54%. The results indicated that application of EGTA delayed bud break in a concentration-dependent manner ( Figure 2A ). Unlike the chilling groups, the inhibitory effect of 10 mM LaCl₃ was almost the same as that of the 30 mM LaCl₃ and the breaking percentages of 10 mM LaCl₃ and 30 mM LaCl₃ treatments were lower than that of the control ( Figure 2B ).

The inhibitory effect of calcium blockage was eliminated after 5 to 10 days in the greenhouse when 30 mM Ca²⁺ was supplied following EGTA or LaCl₃ treatment, as no significant differences were observed between the two groups during this period. However, the buds in the calcium recovery group burst one day later than those in the mock.

Dormancy release was accompanied by variations of ultrastructure and calcium distribution, gradual disappearance of starch grain, and the reopening of the transport channel. Mesophyll parenchyma cells were relatively small in size and had comparatively large nuclei centrally located in the cells, as well as several small vacuoles at the early dormant period ( Figures 3A, B ). Mitochondria, plastids, endoplasmic reticulum, and ribosomes presented in the cytoplasm, and the plastids were not well developed. Upon dormancy release (21 and 28 DC), the cells enlarged obviously compared to those chilled for 0 d. During the early dormant period, abundant starch grains accumulated in the apical bud cells and gradually diminished with prolonged chilling. In the deep dormant state, the plasmodesmata in the cell walls were blurred, and the entrances of the plasmodesmata appeared to fuse with each other, forming a continuous membrane ( Figures 3A–C ). At 21 and 28 DC, the plasmodesmata were distinct in the walls of adjacent cells ( Figures 3D, E ).

Electron microscopic observations revealed that the sample sections immobilized by the fixative solution containing potassium antimonite presented electron-dense deposits in the cells ( Figures 3A–E ). The calcium antimonite precipitates were clearly localized in the intercellular spaces, cell walls, and cytosols. After EGTA chelation, most granules disappeared, accompanied by legible holes presented in the EGTA-treated section ( Figure 3F ), where the Ca²⁺ antimonate precipitate might be before EGTA chelation. This suggests that the electron-dense deposits were calcium precipitates and were indicative of subcellular Ca²⁺ localization in tree peony buds.

Prolonged chilling triggered a Ca²⁺ influx to the cytosol and nucleus from the cell wall and intercellular space. Ca²⁺ deposits were prominently observed in the intercellular spaces throughout the whole chilling duration period, but were notably scarce in the vacuoles, possibly owing to the inadequate development of vacuoles for storing Ca²⁺. At 0 and 7 DC, the most intense Ca²⁺ signals were localized in the intercellular spaces and cell walls, and some granules existed in the nucleus, cytosol, plastid, and other organelles ( Figures 3A, B ). Along with chilling accumulation, the Ca²⁺ deposits in intercellular space and cell wall gradually decreased, while there was an increase in the nucleus, cytosol, and endoplasmic reticulum. At 14 DC, most of the calcium precipitates in the cell wall had disappeared, and dense Ca²⁺ granules were observed congregating along the plasmalemma ( Figure 3C ), suggesting a Ca²⁺ influx from intercellular to intracellular regions. At 21 and 28 DC, calcium signals in the cell wall almost completely vanished, and only a few residual signals remained in the intercellular spaces ( Figures 3D, E ).

Putative calcium sensor encoding genes involved in chilling- and GA- induced dormancy release were identified from the previous transcriptional database. One CaM homolog (Transcript_34059) and four CDPK homologs (transcript_12835, 13256, 16356, and 3925) were significantly differentially expressed both during chilling duration and GA-exposure according to the RNA-seq data (Gai et al., 2013; Zhang et al., 2021). These were chosen as candidates for RT-qPCR analysis. Transcript_34059 (PsCaM), transcript_12835 (PsCDPK32), and transcript_3925 (PsCDPK26;2) were induced at 14 d of chilling treatment, while transcript_16356 (PsCDPKS5) and transcript_12835 (PsCDPK32) peaked at 21 d of chilling, and transcript_13256 (PsCDPK26;1) was not up-regulated until 28 DC. The calcium sensor encoding genes significantly declined after reaching a peak, except for transcript_13256 ( Figure 4 ). The changes in expression patterns suggested that transcript_34059, 12835, 3925, and 16356 might participate in chilling-induced dormancy release, while transcript_13256 was not associated with dormancy release.

Subsequently, the expression patterns of the calcium sensor encoding genes were detected after GA₃ and GA₄ feedings. All the candidate calcium sensor encoding genes were significantly upregulated by GA₃ treatment, whereas GA₄ treatment only enhanced the expressions of transcript_34059, 12835, and 3925. Notably, transcript_3925 was the most dramatically activated gene by GA feedings. In most cases, GA₃ treatment led to higher levels of gene expression than GA₄, with the exception of transcript_3925 ( Figure 4B ). The changes in expression patterns after GA₃ and GA₄ feeding suggested that transcript_34059 (PsCaM), 12835 (PsCDPK32), 3925 (PsCDPK26;2), and 16356 (PsCDPKS5) were involved in chilling- and GA-induced dormancy release in tree peony.

To elucidate the potential role of calcium in modulating dormancy release, we examined the expression patterns of genes associated with dormancy release following calcium treatments, including PsEBB1 (Zhang et al., 2023b), PsEBB3, PsSVP, PsRGL (Gao et al., 2023), PsGA20ox, PsCYCD (Gai et al., 2013), PsBG6, and PsBG9 (Gao et al., 2021). In the chilling group, PsEBB1, PsEBB3, PsGA20ox, PsCYCD, and PsBG6 displayed similar expression patterns, being inhibited in the calcium blocking subgroup and reawakened in the calcium recovery subgroups. Conversely, the expressions of PsSVP and PsRGL, negative regulators of dormancy release, increased through calcium blocking but decreased with replenishment of CaCl₂ ( Figure 5 ). PsBG9 exhibited minimal response to calcium fluctuation in the chilling group.

In the GA treatment group, PsEBB1, PsEBB3, and PsCYCD were also inhibited by calcium blocking and recovered by supplement of CaCl₂, while PsSVP displayed the opposite trend. PsGA20ox was suppressed by calcium blocking but could not be recovered by calcium supplement, whereas PsRGL was relatively stable compared to the mock. Different to the chilling group, PsBG9 was down-regulated by calcium blocking, but was sharply and significantly up-regulated by CaCl₂ recovery, while PsBG6 was unaffected by LaCl₃ and slightly increased with CaCl₂ treatment ( Figure 5B ). The differing responses of PsBGs to calcium treatment between the two groups might be associated with their sensitivity to chilling and GAs.

Both dormancy induction and release undergo obvious ultrastructural changes to accommodate changes in growth signals. During the development of dormancy, several ultrastructural changes take place in the Shoot Apical Meristem (SAM) cells of the bud. In Populus, the number of starch granules increases, and the plasmodesmata are gradually blocked along with SD exposure (Jian et al., 1997; Rinne et al., 2001; Ruonala et al., 2008). However, there is limited documentation of the ultrastructural variations in buds related to dormancy release. After exposure to prolonged chilling, the number of starch grains gradually reduced and became almost undetectable by the end of dormancy in tree peony buds. The result coincided with the activation of starch hydrolysis and carbohydrate metabolism in the dormancy release process of tree peony (Zhang et al., 2018; 2020a), which will provide sufficient substance and energy supplements for bud dormancy release and subsequential bud burst. Simultaneously, the constriction and blockage of plasmodesmata were relieved at the dormancy release stage ( Figures 3D, E ), indicating a reopening of the transport corridor to transmit growth signal and nutrient substance. Recently, we identified that PsBG6 and PsBG9 are induced by chilling and exogenous GAs to break down callose deposited on the PD, respectively (Gao et al., 2021), which might contribute to the process. By comparison, dormancy release and dormancy induction seemed to undergo opposite ultrastructural changes when the number of starch granules and the status of plasmodesmata were concerned.

The variations of subcellular Ca²⁺ distribution indicated that prolonged chilling triggered a Ca²⁺ influx to the cytosol and nucleus, which contributed to dormancy release in tree peony. The relationship between cellular Ca²⁺ distribution and dormancy induction has been extensively reported (DeHayes et al., 1997; Jian et al., 1997; 2004), while little is known about the dormancy release process. Our results presented the dynamic changes of calcium distribution for the first time during the chilling-induced dormancy release process. Initially, numerous Ca²⁺ deposits were found in the intercellular spaces and cell wall of the tree peony bud, which is consistent with findings in poplar SAM cells after 77 d SD exposure (Jian et al., 1997). Additionally, the genes related to calcium sensors, CaM and CDPK, were up-regulated during this period. The results suggested an increased influx of Ca²⁺ translocated from the intercellular space into the inner side of the cell, resulting in an increase of [Ca²⁺]_cyt. Under long day conditions, cellular calcium is mainly localized in vacuoles, intercellular spaces, and plastids in poplar. With prolonged SD exposure, there is an increased presence of Ca²⁺ deposits in the cytosol and nuclei, while numerous Ca²⁺ precipitates reappear in the cell walls and intercellular spaces at a deep dormant status (Jian et al., 1997). Together, the process of dormancy release exhibited a similar variation in calcium redistribution as observed during dormancy induction. Lautner and Fromm (2010) documented a marked increase of [Ca²⁺] in meristem after dormancy break in trees, but we speculated that it was the increase of [Ca²⁺]_cyt, rather than the tissue calcium content, associated with dormancy release in deciduous plants.

Additionally, the influx of Ca²⁺ suggests the need for investigation and function analysis of the candidate Ca²⁺ influx channels and efflux transporters during chilling duration, which will be benefit the understanding of the activation and modulation of calcium signals during the dormancy release process.

In tree peony, LaCl₃ and EGTA feedings dramatically delayed bud break in both chilling- and GA-induced dormancy release, while exogenous application of CaCl₂ mitigated the retardation, which was similar to findings in grape (Pang et al., 2007). Additionally, the genes encoding calcium sensors, such as CaM (transcript_34059) and CDPK (transcript_34059, 39255 and 16356), were significantly induced prior to bud dormancy release, and they were further up-regulated by exogenous GA feeding. The up-regulation of calcium-sensor-associated genes confirmed that an increase of [Ca²⁺]_cyt and an activation of calcium signaling were triggered by prolonged chilling and exogenous GAs.

Recent research has indicated the involvement of calcium in bud dormancy regulation. Firstly, calcium contents undergo alterations during the dormancy induction and release processes (Lautner and Fromm, 2010; Furukawa et al., 2012). Secondly, the calcium signaling pathway has been identified as enriched in the differentially expressed genes between dormancy and dormancy release phases (Anderson et al., 2005; Zhang et al., 2021). Most importantly, exogenous application of calcium and calcium blocking reagents influence dormancy status (Weis et al., 1999; Pang et al., 2007). Collectively, these findings lead us to speculate that activation of calcium signaling promoted bud dormancy release in tree peony.

Calcium blocking delayed bud break, while calcium recovery treatments promoted bud break in chilling- and GA-induced dormancy release in tree peony. The results help to elucidate the mechanism by which calcium regulates dormancy release through the GA and ABA pathway. GA20OX, a key GA biosynthesis gene, contributes to bioactive GA synthesis and dormancy release (Gai et al., 2013; Zhang et al., 2020a), and RGL1 is a negative regulator of the GA pathway, suppressing dormancy release in tree peony (Gao et al., 2023). EBB1, SVP-like (SVL), EBB3, and CYCD3 represent a network of ABA pathway to modulate dormancy release in hybrid poplars (Singh et al., 2018; Azeez et al., 2021). As shown, treatments with calcium blockers depressed the expressions of PsEBB1, PsEBB3, PsGA20OX, and PsCYCD, while promoting PsSVP and PsRGL1 transcripts during dormancy release induced by chilling and GA. Conversely, replenishing calcium either increased or decreased their expressions. The results indicated that calcium blocking to reduce calcium content might suppress bioactive GA synthesis, and promote ABA biosynthesis and signaling, while calcium recovery to free calcium compensated for the imperfection.

Calcium signaling modulating GA and ABA pathways has been documented in several plants and physiological processes (Kim et al., 2010; Ho et al., 2013; Zhang et al., 2023a). In soybean, low concentrations of CaCl₂ inhibit GA biosynthesis and impede radical protrusion, whereas high concentrations of CaCl₂ suppress ABA biosynthesis (Wang et al., 2021). The external application of calcium nitrate induces gibberellin biosynthesis and signal transduction, thereby promoting stem elongation of Dendrobium officinale (Du et al., 2023). Taken together, we assumed that calcium might facilitate dormancy release by modulating the GA and ABA signaling pathways, namely activating GA biosynthesis and signaling, and depressing ABA biosynthesis and signaling.

Additionally, PsBG6 and PsBG9 were also significantly up-regulated by calcium recovery in chilling- or GA- induced dormancy process. The results hinted that fluctuation in free [Ca²⁺] modulates the reopening of transport channels in tree peony. Further study to investigate how calcium triggers GA biosynthesis and the expression of PsBG6 and PsBG9 will be helpful in understanding the mechanism of calcium facilitating dormancy release in tree peony.

The mechanism of calcium modulating dormancy release has been poorly described up to now. We screened several PsCDPKs involved in the regulation of dormancy release. In wheat, TaCDPK30 interacts with TabZIP60 to regulate salt tolerance via modulating ABA synthesis (Zhang et al., 2023a). We speculated that PsCDPKs might function through the phosphorylation of downstream proteins to convey the calcium signal. Further investigations on PsCDPK-interacting proteins will be helpful in understanding the transduction of calcium signals during chilling- and GA-induced dormancy in deciduous plants.

In summary, we confirmed the positive role of calcium in bud dormancy release in tree peony. Prolonged chilling resulted in a calcium influx from the intercellular space and cell wall to the cytosol and nucleus, which was perceived by calcium sensors such as CaM and CDPK, leading to the activation of calcium signaling. Subsequently, biosynthesis of bioactive GAs was stimulated, while the ABA pathway was inhibited. Ultimately, dormancy was broken along with the activation of the cell cycle. These results presented a potential mechanism through which calcium facilitates dormancy release, offering beneficial insights to tree peony production.