2.1 Plant materials and growth conditions

The tea plant materials used in this study were all cultivated at the Tea Research Institute of the Chinese Academy of Agricultural Sciences, Hangzhou, China, including the cultivar ‘Longjing 43’ (LJ43), and three cold-resistant (#R1, #R2 and #R3) and three cold-susceptible (#S1, #S2 and #S3) tea plant accessions (Wang et al., 2022). The first leaves of six 3-year-old tea plant accessions (#R1, #R2, #R3, #S1, #S2 and #S3) at the top of the branch were collected on December 13 and January 17. For time-course expression analysis, 1-year-old potted tea plant seedlings of LJ43 were grown in the chamber at 12 h 22°C light/12 h 20°C dark, and were treated with 4°C for cold treatment. One bud with one leaf (1B1L) and mature leaf (ML) samples of tea plants were sampled at 0 h, 3 h, 6 h, 1 d, 3 d, 5 d and 7 d posttreatment, and after 2 days recovery. Three biological replicates were used for all experiments

The Arabidopsis thaliana ecotype Columbia-0 (Col-0) was used to generate transgenic lines. To construct CsCIPK11-overexpressing (OE) lines, the CsCIPK11 (Cha11g008040) ORF was amplified and cloned into the vector pCAMBIA S1300 containing a super promoter (Lee et al., 2007). The primer sequences used for gene amplification are listed in Supporting Information: Table S1. For the freezing treatment, Arabidopsis seedlings were grown under 10 h 22°C light/14 h 20°C dark before cold treatment, 21-day-old seedlings were subjected to a cold acclimation (CA) treatment at 4°C for 4 day (with normal light dark cycle as above), then exposed to −10°C for 8 h; 21-day-old seedlings grown at 22°C served as control. The aerial parts of the Arabidopsis before and after −10°C treatment was sampled for relative electrolyte leakage (REL), Fv/Fm, and antioxidant capacity measurements.

Transgenic Nicotiana benthamiana plants with a nuclear marker H2B-RFP were used for subcellular localization and bimolecular fluorescence complementation (BiFC) assays (Li, Zhao, et al., 2017). Wild-type tobacco N. benthamiana was used for luciferase complementation imaging (LCI) analysis and protein degradation analysis. Tobacco seedlings were grown under a 16/8 h light/dark cycle (22/20°C) for 28 days before injection.

2.2 Total RNA isolation and quantitative RT-PCR (qRT-PCR) analysis

Total RNA was extracted using an RNAprep Pure kit (Tiangen) following the manufacturer's instructions. Complementary DNA (cDNA) was produced using a PrimeScript™ RT reagent kit (TaKaRa). qRT-PCR was carried out using a Light Cycler 480 machine (Roch) with SYBR Green I Master Mix (Roche Basel). The relative expression levels of the genes were calculated to the CsPTB reference gene using the 2^−∆∆CT method (Hao et al., 2014, Livak & Schmittgen, 2001). The qRT-PCR primer sequences are listed in Supporting Information: Table S2.

2.3 Subcellular localization

To determine the subcellular localization of CsCIPK11 and tau class glutathione S-transferase 23 (CsGSTU23, Cha09g013270) in plants, the ORF of CsCIPK11 and CsGSTU23 without a stop code was fused with a green fluorescent protein (GFP). CsCIPK11-GFP and CsGSTU23-GFP fusion proteins were expressed in the leaves of N. benthamiana expressing the nuclear marker H2B-RFP, as previously described (Peng et al., 2022). A Zeiss LSM710 confocal laser scanning microscope (Zeiss) was used for imaging.

2.4 Measurement of relative electrolyte leakage, Fv/Fm, antioxidant capacity, and GST enzymatic activity

The REL of tea plants and Arabidopsis was determined as previously described (Wang et al., 2019; Yao et al., 2020). Briefly, the 15 leaf discs of tea plant and aerial parts of Arabidopsis were collected, respectively, and placed into a 50 mL tube with 15 mL deionized water. The tubes were shaken at 200 rpm under 25°C for 2 h. The conductivity of the solutions before and after boiling was measured. Before Fv/Fm measurements, the plants were placed in the dark for 20 min, and the Fv/Fm values of the leaves were conducted on a FluorCam 7 instrument (Photon Systems Instruments). The clearance rate of free radicals and total antioxidant capacity were determined using DPPH and FRAP kits (Comin), according to the manufacturer's protocol. To measure the GST enzymatic activity in vivo, 35S-CsGSTU23 was transiently co-expressed with 35S-CsCIPK11 and the control vector in the leaves of N. benthamiana plants, respectively. After 3 days, the leaves were sampled, and GST enzymatic activity was measured using a GST activity test kit (Comin) following the manufacturer's instructions. To measure the effect of the mutation of site 37 on GST enzymatic activity, the Serine37 of CsGSTU23 was mutated to Alanine37. Proteins were expressed in Escherichia coli BL21(DE3) cells and purified using Ni-NTA Agarose Resin (Thermo Fisher Scientific). Then, the same amount of CsGSTU23 (WT; Serine37) and CsGSTU23 (37A, Alanine37) proteins were used for GST activity measurement.

2.5 CsCIPK11 gene suppression in tea plants

Antisense oligonucleotides (AsODN) and sense oligonucleotides (sODN) sequences were designed using Soligo software by inputting the CsCIPK11 cDNA sequence (Ding & Lawrence, 2003). The sequences are listed in Supporting Information: Table S3. The first leaves of young shoots were collected from ‘LJ43’ tea plants for sODN and AsODN treatments. The leaves were soaked in 1 mL of 20 μM AsODN or sODN solution and cultivated in a climatic chamber under a 12/12 h light/dark cycle (22/20°C) as previously described (Li et al., 2022; Zhao et al., 2023). After 24 h of incubation, the leaves were divided into two groups: one group was sampled immediately, and the other group was sampled after being subjected to −2°C for 2 h. In each group, 10 leaves were used as 10 replicates for Fv/Fm analysis, three leaves were used as three replicates for REL measurement, and 18 leaves were divided into three replicates and were used for total antioxidant capacity, free radical clearance rate, and expression analysis. The phenotype was recorded under recovery for 2 h at room temperature after −2°C treatment.

2.6 Yeast two-hybrid (Y2H) assay

Coding DNA sequences or truncated fragments of CsGSTU23 and CsCIPK11 were conducted into the prey vector pGADT7 and bait vector pGBKT7, respectively. The recombinant pGADT7 and vectors pGBKT7 were co-transformed into chemically competent AH109 yeast cells in specified combinations, following the manufacturer's protocol (Weidi). Positive transformants grown on SD-Trp/-Leu (DDO) medium were transferred to a screening medium (SD-Trp/-Leu/-His/-Ade, QDO) to confirm possible interactions.

2.7 BiFC and LCI assays

For the BiFC and LCI assays, Agrobacterium tumefaciens GV3101 strain was transformed with binary constructs, nYFP/cYFP and nLuc/cLuc were transiently expressed in the leaves of N. benthamiana plants expressing the nuclear marker H2B-RFP and wild-type N. benthamiana, respectively. After incubation for 3 days, BiFC assays were conducted using the Zeiss LSM710 confocal laser scanning microscope (Zeiss). The LCI assay was conducted on a chemiluminescence image analysis system (Tanon).

2.8 GST pull-down assay

GST or CsCIPK11-GST proteins were expressed in Escherichia coli BL21(DE3) cells and purified using Pierce Glutathione Agarose (Thermo Fisher Scientific). For the pull-down assays, 10 μg of GST or CsCIPK11-GST proteins and 10 μL glutathione agarose beads were incubated in 500 μL pull-down buffer at 4°C for 1.5 h with gentle shaking. Then, the agarose beads were collected and further incubated with CsGSTU23-His proteins at 4°C for 1.5 h with gentle shaking. Proteins bound to the agarose beads were released by boiling in sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) loading buffer. The proteins were detected via immunoblotting using an anti-His antibody (Abmart).

2.9 Phosphorylation assays and LC-MS/MS analysis

For the phosphorylation assay, 5 μg CsCIPK11-GST and 5 μg CsGSTU23-His proteins purified from E. coli were incubated in a kinase reaction buffer (50 mM HEPES, 2 mM DTT, 10 mM MgCl₂, 4.6 mM CaCl₂, 5 mM EGTA and 50 μM ATP) at 25°C for 30 min. ADP production from reactions buffer was determined by an ADP-Glo assay kit (Promega). Briefly, the ADP-Glo reagent was used to stop the kinase reaction, and incubated at 25°C for 50 min. Then, the kinase detection reagent was added and incubated at 25°C for another 50 min. Luminescence was measured using a GloMax 20/20 system (Promega).

Protein phosphorylation levels were detected via immunoblotting, including a dephosphorylation reaction with the FastAP™ alkaline phosphatase (Thermo Fisher Scientific), phosphorylation reaction in vitro, incubation with the corresponding antibodies (anti-phospho-serine (anti-pSer) (Sigma) and anti-phospho-threonine (anti-pThr) (CST), and detection following protein separation using SDS-PAGE.

For the phosphorylation assay in vivo, 35S-CsCIPK11-GFP and empty vector were co-expressed with 35S-CsGSTU23-HA in N. benthamiana leaves for 3 days, respectively, and then the total proteins of N. benthamiana leaves were extracted and incubated with 10 µL of anti-HA agarose beads. After 4 h of gentle shaking at room temperature, the beads were washed three times with washing buffer. The immunoprecipitated proteins were then released from the agarose beads by boiling them in SDS-PAGE loading buffer. CsGSTU23-HA proteins were separated and collected via SDS-PAGE for further digestion with trypsin for 15 h. The phosphopeptides were analyzed using a timsTOF Pro2 system (Bruker Daltonics). Raw data were processed by PEAKS Studio version 10.6 (Bioinformatics Solutions Inc.). The species information in the database is Camellia sinensis (L.) O. Ktze. Database search parameters fragment ion mass tolerance error was 0.05 Da, parent ion mass tolerance error was 20ppm. Peptide card value was −10lgP ≥ 15.

2.10 Protein degradation measurement

To measure protein degradation in vivo, the 35S-CsGSTU23-Luc vector transformed into the A. tumefaciens GV3101-p19 strain was transiently co-expressed with 35S-CsCIPK11-GFP and 35S-GFP in the leaves of N. benthamiana plants. After incubation for 3 days, the protein level of CsGSTU23-Luc was measured using a dual-luciferase reporter assay system (Promega) on a Synergy H1 instrument (BioTek). Images were recorded using a chemiluminescence image analysis system (Tanon).

For cell-free degradation, CsGSTU23-His protein was expressed and purified from E. coli. The total proteins of the wild-type tobacco expressing 35S-GFP and 35S-CsCIPK11-GFP, respectively, were extracted. The protein concentration was determined by the BCA assay reagent (Kangwei). Each reaction mix contained 100 ng of CsGSTU23-His and 500 μg of total protein from tobacco. For the proteasome inhibitor experiments, 20 μM MG132 was added. The reaction mixes were incubated at 22°C and were sampled at 0, 0.5, 1, 3, 6, 12 and 24 h. The reactions were stopped by SDS-PAGE loading buffer and boiling at 98°C for 10 min.

2.11 Transient overexpression of CsGSTU23

CsGSTU23 was transiently overexpressed in the leaves of 1-year-old tea plant seedlings of #S3 accession according to a previously described method (Chen et al., 2023). One-year-old tea plant seedlings of #S3 were obtained by asexually cutting propagation. Thirty tea plants whose leaves expressed with 35S-CsGSTU23-GFP and 30 tea plants whose leaves expressed with 35S-empty vector (EV-GFP) were grown in a growth chamber under a 12/12 h light/dark cycle (22/20°C) for 2 days, then they were divided into two groups: one group was sampled immediately, and the other group was sampled after being subjected to −2°C for 4 h. In each group, three tea plants were used for phenotypic analysis, the second leaves of three tea plants were used for REL measurement, the second leaves of nine tea plants were divided into three replicates and were used for total antioxidant capacity, free radical clearance rate, and expression analysis, and the third leaves of 10 tea plants were used for Fv/Fm analysis. The phenotype was recorded under recovery for 6 h at room temperature after −2°C treatment.

2.12 Statistical analysis

All experimental data with more than three replicates were analyzed using IBM SPSS Statistics 21 software. Student's t-test and one-way analysis of variance (ANOVA) were used to calculate significant differences between pairs.

3.1 Expression pattern of CsCIPK11 and subcellular localization

CsCIPK11 is an ortholog of AtCIPK11 (63.1% identity). qRT-PCR was performed to determine the transcript abundance of CsCIPK11 in response to low temperatures. The expression levels of CsCIPK11 in one bud with one leaf (1B1L) and mature leaf (ML) of tea plants was significantly induced when subjected to 4°C within 3 h and 6 h, respectively, and reached peak expression at 5 days (9.23-fold) and 7 days (3.98-fold) posttreatment, respectively (Figure 1a). In addition, the induction of CsCIPK11 expression under 4°C treatment in 1B1L was higher at 1 day after treatment than that in ML.

To investigate the subcellular localization of CsCIPK11, the 35 S CaMV-GFP construct of CsCIPK11 and the control vector (35 S CaMV-GFP) were transiently expressed in transgenic N. benthamiana leaves expressing H2B-RFP as a nuclear marker. The results showed that the CsCIPK11-GFP fusion protein was distributed throughout the cell, nucleus, and cell membrane, consistent with the expression of the 35S CaMV-GFP (Figure 1b).

3.2 Overexpression of CsCIPK11 improved cold tolerance in Arabidopsis

To test if CsCIPK11 plays a role in cold tolerance, transgenic Arabidopsis plants over-expressing CsCIPK11 were generated. Three independent OE lines (#1, #2 and #3) were selected for the cold tolerance assay (Supporting Information: Figure S1). No morphological differences were observed between the Col-0 and CsCIPK11-OE plants under normal conditions. After subjecting to −10°C for 8 h, CsCIPK11-OE plants maintained higher levels of chlorophyll fluorescence, with Fv/Fm values significantly higher than Col-0 plants (Figure 2a). The REL in the CsCIPK11-OE lines was significantly lower than those in Col-0 after −10°C treatment (Figure 2b). The free radical clearance rate and total antioxidant capacity in the CsCIPK11-OE lines were significantly higher than those in Col-0 after −10°C treatment (Figure 2c,d). Taken together, these results indicate that CsCIPK11-OE plants were more tolerant to low temperatures than Col-0 plants and that CsCIPK11 played a positive role in cold tolerance in transgenic Arabidopsis.

3.3 Transient knockdown of CsCIPK11 expression decreased cold tolerance in tea plants

To verify the role of CsCIPK11 in the cold response of tea plants, we suppressed CsCIPK11 expression in young tea plant leaves by soaking them in an antisense oligonucleotides (AsODN) solution and young leaves cultivating with sense oligonucleotides (sODN) served as control group. After incubation for 24 h, the relative expression level of CsCIPK11 in AsODN-CsCIPK11 leaves was significantly reduced by 26.82% and 60.22% compared to sODN before and after −2°C treatment, respectively (Figure 3a). The phenotypes were recorded after 2 h of recovery at 22°C. There were no differences observed in the young leaves of tea plants at 22°C; however, low-temperature injury was more severe in the CsCIPK11-silenced young leaves than in the controls after −2°C treatment (Figure 3b), as judged by a significantly lower chlorophyll fluorescence values (Figure 3c). Meanwhile, AsODN-CsCIPK11 leaves exhibited significantly lower Fv/Fm values (Figure 3d) and higher REL values (Figure 3e), as well as a reduced total antioxidant capacity (Figure 3f), and free radical clearance rate (Figure 3g) than the control under −2°C treatment. Taken together, these results indicated that CsCIPK11 positively modulates cold tolerance in tea plants.

The ICE (inducer of CBF expression)—C repeat binding factor (CBF)—COR (cold-responsive genes) pathway plays a key role in cold stress response in plants (Shi et al., 2018). To explore the molecular mechanism of CsCIPK11 in regulating cold tolerance, the expression levels of genes involved in the ICE—CBF—COR pathway were determined. The expression levels of these genes showed no difference between AsODN-CsCIPK11 and sODN-CsCIPK11 under 22°C conditions. After −2°C treatment, the expression levels of CsICE, CsCBF1, CsCBF2, CsCBF3, CsCBF4, CsCOR47, CsCOR413 and CsCOR414 were significantly reduced in AsODN-CsCIPK11 compared with those in sODN-CsCIPK11 (Figure 3h). This suggests that CsCIPK11 participates in the regulation of the ICE—CBF—COR pathway.

3.4 CsCIPK11 interacted with the CsGSTU23 protein

To explore the molecular mechanism by which CsCIPK11 regulates the cold response in tea plants, yeast two-hybrid screening was performed with a tea plant cDNA library using CsCIPK11 as bait. Interestingly, CsGSTU23, which displays low expression in cold-sensitive tea plant cultivars (Wang et al., 2019), was identified as a candidate that interacted with CsCIPK11. To confirm the interaction of involving CsCIPK11-CsGSTU23, Y2H, LCI, BiFC and pull-down assays were performed. All yeast cells grew normally in the DDO medium. Yeast cells containing a combination of the CsCIPK11-BD and CsGSTU23-AD constructs grew effectively on QDO/X medium (Figure 4a). Pull-down assays showed that the CsGSTU23-His fusion protein bound to GST-CsCIPK11 but not to the negative control (GST) (Figure 4b). The BiFC assay showed that the yellow fluorescence (YFP) of CsCIPK11-CsGSTU23 complex was mainly present in the cell nucleus and cell membrane (Figure 4c). The LCI results showed that luminescence signals appeared only in the leaf regions co-expressing CsCIPK11-nLuc and CsGSTU23-cLuc (Figure 4d). These results demonstrate that CsCIPK11 interacts with CsGSTU23 in vitro and in vivo.

Bioinformatic analysis predicted that CsCIPK11 possesses an N-kinase domain and a C-regulatory domain and that CsGSTU23 contained a GSH-binding site at the N-terminus and a polypeptide binding site at the C-terminus (Figure 4e). Further LCI assays of the truncated sequence revealed that the kinase domain of CsCIPK11 and the N-GSH-binding domain of CsGSTU23 are necessary for the CsCIPK11-CsGSTU23 interaction (Figure 4f).

3.5 CsCIPK11 phosphorylated and stabilized the CsGSTU23 protein

As the N-kinase domain of CsCIPK11 interacts with the N-GSH-binding domain of CsGSTU23, we speculated that CsGSTU23 is a potential substrate for the phosphorylation of CsCIPK11 kinase. We analyzed the potential phosphorylation site of the CsGSTU23 N-terminus, with four Ser sites, one Thr site and one Tyr (Tyrosine) can potentially be phosphorylated (Figure 5a). Due to CIPK was S/T protein kinases, immunoblotting with anti-pSer and anti-pThr antibodies demonstrated that the GST-CsCIPK11 fusion recombinant protein could phosphorylate the CsGSTU23 protein at a Ser site, but not at the Thr site, of CsGSTU23 (Figure 5b). In vitro phosphorylation was measured using a luminescence-based kinase assay. The amount of ADP formed by ATP during the phosphorylation reaction was measured using a luminescence assay system (Zhang, Liu, et al., 2019). When CsCIPK11 was incubated with CsGSTU23, the production of ADP was significantly higher than that with CsCIPK11 and CsGSTU23 alone, indicating that CsGSTU23 was a substrate for CsCIPK11 kinase (Figure 5c).

We further examined whether CsCIPK11 influenced the enzymatic activity and stability of CsGSTU23 protein. The GST enzymatic activity of N. benthamiana leaves co-expressed with 35S-CsCIPK11 and 35S-CsGSTU23 was significantly higher than that of leaves co-expressed with the empty vector and 35S-CsGSTU23, indicating that CsCIPK11 contributed to increased GST activity in vivo (Figure 5d). Protein degradation assays were also performed. Briefly, the 35S-CsGSTU23-Luc protein was co-expressed with equal amounts of 35S-CsCIPK11-GFP and 35S-GFP control proteins in N. benthamiana leaves. The luciferase assay showed that CsGSTU23-Luc protein levels were higher when co-expressed with CsCIPK11-GFP than when co-expressed with the control GFP protein (Figure 5e). Cell-free degradation also showed that CsGSTU23-His protein was more stable in protein extracts of 35S-CsCIPK11 tobacco than in those of the empty control (Figure 5f). These results indicated that the CsCIPK11-mediated interaction of CsGSTU23 increased the stability of CsGSTU23.

To identify the CsCIPK11-mediated phosphorylation sites of CsGSTU23, phosphorylated CsGSTU23 was subjected to LC-MS/MS analysis. One CsGSTU23 phosphopeptide incorporating the Ser37 site was found to be phosphorylated by CsCIPK11 (Figure 5g). Mutations at Ser37 site from Serine (S) to Alanine (A) resulted in a significant reduction in GST activity of CsGSTU23 (Figure 5h), indicating that the phosphorylation of Ser37 is critical for its enzymatic activity.

3.6 The transient overexpression of CsGSTU23 increased cold resistance in tea plants

To investigate the cold-responsive pattern of CsGSTU23 expression in tea plants, qRT-PCR analysis revealed that transcript abundance of CsGSTU23 was significantly induced by 4°C low temperatures in 1B1L and ML, and its induction in ML was earlier and higher than in 1B1L (Figure 6a). Meanwhile, the expression level of CsGSTU23 in cold-resistant tea plant accessions was notably higher than in the cold-susceptible accessions (Wang et al., 2022) (Figure 6b). These results suggest that CsGSTU23 plays an important role in cold tolerance in tea plants. The subcellular localization of CsGSTU23 was observed in the nucleus and cell membrane, similar to that of CsCIPK11 (Figure 6c).

To further characterize the biological function of CsGSTU23 in tea plants, we designed experiments to suppress and increase its expression. Despite several attempts we did not obtain an effective AsODN to knock down CsGSTU23 in tea plants. Transient overexpression of CsGSTU23 in tea leaves was performed as described previously (Chen et al., 2023). qRT-PCR analysis revealed that the expression of CsGSTU23 in OE leaves was significantly overexpressed (44.8 times higher) than in EV control leaves (Figure 6d). Western blot analysis revealed the overexpression of CsGSTU23-GFP protein with a molecular weight of 60 kDa in 35 S CaMV-CsGSTU23 leaves, and the overexpression of EV-GFP protein with a molecular weight of 26 kDa in EV leaves (Figure 6e). After 4 h of treatment at −2°C, EV plants were more wilted and showed severe frost damage compared to OE plants (Figure 6f). Consistent with this phenotype, chlorophyll fluorescence imaging, Fv/Fm values (Figure 6g), total antioxidant capacity, and the free radical clearance rate (Figure 6h) were all significantly improved in OE leaves. In contrast, the REL value was significantly lower in OE leaves than in EV leaves (Figure 6h). These results indicate that the transient overexpression of CsGSTU23 leads to substantially decreased cold susceptibility in tea plants.