The most recent version (V2.1) of a 12X assembly of the grapevine genome (V. vinifera) was downloaded from CRIBI (http://genomes.cribi.unipd.it/) for use in identifying grapevine CBL and CIPK proteins. Additionally, 10 CBL and 33 CIPK genes from rice (Oryza sativa) were downloaded from the rice genome database (http://rice.plantbiology.msu.edu//), and 10 CBL and 26 CIPK genes from Arabidopsis were downloaded from TAIR (https://www.arabidopsis.org/index.jsp) database. Grapevine CBL and CIPK genes were identified using the following two steps: (i) the well-characterized Arabidopsis 10 CBLs and 26 CIPKs were used as queries against the grapevine genome using BLASTP with e-values <1E-5. (ii) The identified grapevine CBL and CIPK genes were further confirmed by constructing a phylogenetic tree using the Neighbor-Joining (NJ) method of identified grapevine CBLs and CIPKs and the CBLs and CIPKs from Arabidopsis and rice. Only grapevine genes with an E-value <10⁻⁴, and an identy >55%, and that located in one the clades of derived from Arabidopsis and rice CBL and CIPK genes were regarded as VvCBL and VvCIPK genes, respectively. The genes considered as grapevine CBLs and CIPKs were then named in accordance with the gene nomenclature rules established by the grapevine scientific community (Grimplet et al., 2014). VvCBLs and VvCIPKs names were assigned based on their ortholog genes in Arabidopsis identified using bootstrap replicates of the Maximum Likelihood phylogenetic tree with values higher than 70. Only one-to-one orthologs were considered when allocating an Arabidopsis-like name to the Vitis gene. Otherwise, the names of the VvCBL and VvCIPK were assigned a number higher than the highest number used in the naming of Arabidopsis CBL and CIPK genes. The molecular weight (MW) and isoelectric point (pI) of each protein sequence were calculated using ExPASY (http://web.expasy.org/compute_pi/).

The protein sequence of all of the identified CBL and CIPK family members from grapevine, Arabidopsis, rice and poplar (Populus sp.) were aligned using Muscle. Phylogenetic trees were constructed using the Neighbor-Joining (NJ) method with the MEGA6.0 software program with 1,000 bootstrap replicates and the Poisson model. The classification of subfamilies of grapevine CBL and CIPK genes was consistent with the previous subfamilies reported in Arabidopsis and poplar (Yu et al., 2007; Zhang et al., 2008).

The coding DNA sequences (CDS) and genomic sequences of grapevine CBLs and CIPKs were retrieved from the CRIBI genomic database for grapevine. Gene structures of VvCBLs and VvCIPKs was analyzed using the Gene Structure Display Server (GSDS 2.0, http://gsds.cbi.pku.edu.cn/). Prediction of motifs were generated using the Multiple Em for Motif Elicitaton (MEME) program (http://meme-suite.org/tools/meme), with an optimum width of motifs ranging from 6 to 50, the maximum number of motifs set at 18, and default values for other parameters.

The chromosomal locations of grapevine CBL and CIPK genes were verified from the CRIBI database, and chromosomal images were drawn using MapInspect software. Tandem duplicated genes were defined as an array of two or more genes located on the same chromosome and found within a 100 kb genomic window (Zhu et al., 2016). Segmental duplication genes were identified as genes located on duplicated chromosomal blocks, and were determined using MCScanX software (http://chibba.pgml.uga.edu/mcscan2/) to detect gene duplication events using an E-value 10⁻⁵.

To better understand the patterns of macroevolution, the non-synonymous substitutions (Ka) and synonymous substitutions (Ks) rates and diverse time of CBLs and CIPKs from grapevine, rice, Arabidopsis, and poplar were extensively estimated. Firstly, the nucleotide coding sequences (CDSs) of duplicated VvCBL genes and VvCIPK genes were aligned using ClustalW 2.0 (Larkin et al., 2007), and then non-synonymous substitutions, synonymous substitutions, and the ratio between them (Ka/Ks) were calculated using MEGA 5.0 (Tamura et al., 2011). Secondly, the Ks value was used to calculate the time (Mya, million years ago) of duplication as T = Ks/2λ (λ = 6.5 × 10⁻⁹ for grape) (Cao et al., 2014). The collinearity relationship between grapevine and other species including rice, poplar, and Arabidopsis were retrieved from Plant Genome Duplication Database (PGDD, http://chibba.agtec.uga.edu/duplication/).

The 2,000 bp upstream sequences of coding region of VvCBL and VvCIPK genes were downloaded from CRIBI (http://genomes.cribi.unipd.it/). The cis-regulatory elements were identified using PlantCARE (http://bioinformatics.psb.ugent.be/webtools/plantcare/html/) software (Wang et al., 2015).

“Pinot Noir” grape (V. vinifera, the sequenced genotype PN40024) plantlets were grown on MS medium in a tissue culture room with a photoperiod of 16 h light/8 h dark and a temperature of 23°C for 6 weeks.

Salinity and drought stress treatments were administered by irrigating the plants with either 200 mM NaCl or 10% polyethylene glyco 6000 (PEG), respectively. For the nutrient stress treatment (Xu et al., 2006), plants were irrigated with a modified 1/2 MS solution containing only 100 μM K⁺ (Xu et al., 2006). Low temperature and heat stress treatment were applied by putting the plants at 4°C (Drug storage box, HYC-360, Haier) and 42°C (Intelligent artificial climate box, RXZ-380C, Jiangnan Instrument Factory, Ningbo) with a photoperiod of 16 h light/8 h dark, respectively. Leaves in all the treatments were harvested at 0, 6, 12, and 24 h post treatment application. The collected leaves were immediately frozen in liquid nitrogen and stored at −70°C until further analysis. Three biological replicates (three independent plants) were used in all the stress treatments and the corresponding controls.

Total RNA was extracted from the sampled leaves using a previously described protocol (Gonzalez-Mendoza et al., 2008). The concentration and purity of RNA were determined by measuring the optical density (OD) ratio at 260 and 280 nm using a One Drop™ OD-1000 spectrophotometer (Thermo Fisher Scientific, USA). RNA integrity was evaluated by electrophoresis on a 1% agarose gel stained with ethidium bromide. A high capacity Prime Script™ RT reagent Kit with gDNA Eraser (Takara, Japan) was used to eliminate traces of genomic DNA and to synthesize first strand cDNA from the template RNA according to the manufacturer's instructions.

The specific expression of VvCBLs and VvCIPKs was examined by RT-qPCR using the SYBR Green method on a Bio-Rad CFX 96 PCR real-time thermo cycler. The oligo nucleotide primers for amplifying specific VvCBLs and VvCIPKs were designed using Primer Premier 5 (http://www.premierbiosoft.com/primerdesign/) software, and were based on the 3′-untranslated region (UTR) and the 3′ terminal sequence of the coding region (Wang G. et al., 2014). The housekeeping gene (actin-101-like, VIT_012s0178g00200) was used for normalization of the data (Wang M. et al., 2014; Liu et al., 2016). All primer sequences are listed in Table S5. The total volume of each RT-qPCR reaction was 20 μL which included 2 μL cDNA from each sample as a template, 10 μL SYBR Premix Ex TaqTM (Tli RNase Plus, Takara, Japan), 0.8 μL of each primer, and 6.4 μL ddH₂O. The PCR conditions consisted of denaturation step for 30 s at 95°C, followed by 40 cycles of 5 s at 95°C and 60°C for 20 s at the end. The melting curve from 65°C to 95°C for 15 s for each amplification was conducted immediately after the completion of the qRT-PCR to verify the specificity of each amplification. Three replicates were run for each treatment sample and the corresponding controls. Relative expression values (fold change) were calculated using the 2^−ΔΔCT method (ΔCT = CT target – CT reference; ΔΔCT = (CT target-CT reference) treatment -(CT target-CT reference)control) (Livak and Schmittgen, 2001). The relative expression of each of the examined genes was represented as values relative to the untreated controls. Mean values ± standard deviation (SD) were calculated from the three biological replicates.

High-throughput microarray data from published research (Fasoli et al., 2012) were downloaded from the Gene Expression Omnibus (GEO, https://www.ncbi.nlm.nih.gov/geo/) database at the National Center for Biotechnology Information (NCBI) to characterize the general spatial and temporal expression patterns of VvCBL and VvCIPK genes during development. In the analysis of the microarray data, 54 samples from green and woody tissues and organs of grapevine at different developmental stages (GSE36128) were examined, and the log10 value of gene expression was used. The derived heatmaps were drawn using Multi experiment viewer 4.9 (MeV, http://www.tm4.org/mev.html) software.

The Pearson correlation coefficient (PCC) value was calculated between each pair of VvCBLs and VvCIPK using gene expression values from the high-throughput transcriptome data and qRT-PCR data using Statistical Product and Service Solutions (SPSS v20.0) software. Co-expressed gene pairs were filtered with a PCC cut-off of 0.7 (Liu et al., 2016).

For the protein interaction networks, high confidence, experimental data of interactive CBL and CIPK proteins in Arabidopsis were constructed by using STRING (http://string-db.org/) using an option value > 0.7. The homolog proteins of the determined interactive Arabidopsis proteins were identified in grapevine by reciprocal best BLASTP analysis.

All results are presented as means ±SD of three biological replicates where each biological replicate consisted of three technical replicates. Significant statistical differences between treatments were determined by a Student's test (P < 0.05) using SPSS.

Based on the BLAST query and subsequent phylogenetic tree construction (Figure S1, Table S1), a total of eight non-redundant CBLs and 20 CIPKs were identified from the latest version (V2.1) of grapevine genome annotation, respectively (Table 1). Among them, the sequences of VvCBL12 was manually corrected due to its ambiguous annotation from V2.1 genome annotation (see the notes of Table 1). The grapevine CBLs and CIPKs were then named according to the nomenclature rules recently established by the grapevine scientific community (Grimplet et al., 2014). The detailed information of the identified VvCBLs and VvCIPKs are listed in Table 1.

Furthermore, the multiple amino acids sequence alignments of VvCBLs along with AtCBLs were shown in Figure S2A, indicating that general structures of VvCBLs protein are highly conserved, with all VvCBLs having four EF hands similar to those of EF-hand motifs in AtCBLs. All CBL proteins have invariant spacing between each EF-hand motif (Figure S2A). A total of 23 amino acids lie between EF1 and EF2, 25 amino acids between EF2 and EF3, and 32 amino acids betweenEF3 and EF4. Notably, a recently identified PFPF motif, which is used for phosphorylation of CBLs and CIPKs (Kanwar et al., 2014), was also identified, and which is highly conserved in all eight VvCBLs (Figure S2A). In addition, three VvCBLs, including VvCBL13, VvCBL4, and VvCBL5, were found to harbor a myristoylation site in the N-terminal sequence (Figure S2A).

Similarly, the multiple sequence alignment of VvCIPK proteins shown that all the VvCIPKs contain an N-terminal catalytic kinase domain and a C-terminal regulatory domain which were necessary and very important for protein-protein interaction, except VvCIPK35, which had the high homology with other Arabidopsis CIPKs but seems to be missing both the C-terminal regulatory domain and the NAF domain. In addition, the other VvCIPKs also possess an activation domain in the N-terminal sequence as well as a NAF domain in the C-terminal sequence simultaneously.

In order to investigate the evolutionary relationships between grapevine CBL and CIPK proteins and other species, a Neighbor-Joining phylogenetic tree was constructed using the full amino acid sequences of CBL and CIPK family proteins from grapevine, Arabidopsis, rice, and poplar (Populus sp.). The CBLs and CIPKs were clustered into four (Figure 1A) and five (Figure 1B) subfamilies, respectively. As shown in Figure 1A, VvCBL10a and VvCBL10b were clustered in group I and were identified as orthologous gene pairs with OsCBL9. However, VvCBL11 and VvCBL12 formed as group II, and VvCBL13 formed as Group III, which seemed to be more close to poplar CBLs. In group IV, VvCBL5 and VvCBL8 were homologous to AtCBL5 and AtCBL8, respectively, while the VvCBL4 seems to be more close to popular CBLs. The variant phylogenetic relationships of VvCBLs indicated that the CBL genes have different evolutionary rate during grapevine evolution.

In regards to the 20 VvCIPKs, group A included VvCIPK3, -9, -21, -24, -39, -40, -41, group B included only VvCIPK29, group C included VvCIPK30, -31,-32, -33, -34, -35, -36, -37, -38, group D included VvCIPK27 and VvCIPK28, and group E included only VvCIPK12. Careful examination of the phylogenetic relationships of CIPKs between grapevine and other two species (Arabidopsis and poplar) showed that phylogenetic relationships of VvCIPKs were more close to poplar comparing that to Arabidopsis, indicating that the evolutionary rate of grapevine CIPK gene family is faster than what they supposed to be.

Furthermore, Closely-related orthologous pairs of CBLs and CIPKs were identified between grapevine and Arabidopsis based on high bootstrap value. Such as VvCBL8 and AtCBL8, VvCBL10a, -10b and ATCBL10, VvCBL4 and AtCBL4, VvCBL5 and AtCBL5, VvCIPK3 and AtCIPK3, VvCIPK9 and AtCIPK9, VvCIPK12 and AtCIPK12, VvCIPK21 and AtCIPK21, VvCIPK24 and AtCIPK24. All bootstrap values are higher than 70 and ranged from 70 to 100, suggesting that an ancestral set of CIPK and CBL genes existed prior to the divergence of grapevine and Arabidopsis.

Intron/exon organization and conserved motifs were analyzed in order to further investigate the structural features of grapevine CBLs and CIPKs. As illustrated in Figure 2C, VvCBLs genes in group I have the greatest number of introns with eight, genes in group III and group IV have seven introns, as does VvCBL11 in group II. VvCBL12 in group II has six introns. VvCIPK family members clustered into an intron-rich clade (> 8 introns per gene) and an intron-poor clade (<3 introns per gene) (Chen et al., 2011; Ye et al., 2013). The intron-rich VvCIPK genes clustered in subgroup A, while the intron-poor genes are present in all four of the other subgroups B, C, D, and E (Figure 2D). Within the intron-poor clades, only VvCIPK36 in subgroup C has two introns, while VvCIPK35 and VvCIPK33 in subgroup C, and VvCIPK27 in subgroup D, have one intron. The remaining CIPK genes have no intron.

Furthermore, the conserved motifs of VvCBLs (Figure 2A) and VvCIPKs (Figure 2B) were determined by using MEME software. Twelve conserved motifs were detected in VvCBL proteins and the detailed information about each motif is presented in Figure S3A. All of the grapevine CBL proteins contain motif 1 to motif 5, which were also annotated as the four EF-hand domain and PFPF motif, respectively. However, motif 6 is only present in subgroup I, and both of motif 7 and 9 are only present in subgroup II, suggesting that those motifs might play important role specific to corresponding subgroups.

A total of 18 motifs were identified in VvCIPK proteins (Figure 2B) and amino acid sequences of each motif are presented in Figure S3B. Motif 8 is the functional NAF domain of CIPK proteins and is widely distributed in all of the VvCIPK proteins, except VvCIPK35, which only has motifs 3 and 10 within the N-terminal catalytic kinase domain. VvCIPK40 appears to have lost motif 5, which also annotated as an ATP-binding domain. Intron-rich VvCIPK proteins possess motif 17, except VvCIPK21 and VvCIPK40 in group A. CIPK proteins in group C all possess motif 11, except VvCIPK30, VvCIPK31, and VvCIPK35.

To understand the functional divergence of grapevine CBL and CIPK genes, the chromosome distribution and gene duplication events were surveyed (Figure 3). The locations of each VvCBL and VvCIPK gene were given a diagrammatic representation based on the latest version (V2.1) of grapevine genome annotation. The results showed that the distribution of VvCBL and VvCIPK genes in chromosomes is not even. 7 VvCBL genes are distributed on 4 grapevine chromosomes (VvCBL12 which locates to chromosome Unknown is not shown). Among them, three VvCBL genes is situated on chromosome 2, chromosome 4 has two VvCBL genes, and other two located each in chromosomes 16 and 19, respectively (Figure 3, Table 1). The 20 VvCIPK genes were mapped to 11 chromosomes (Figure 3). Chromosomes 4, 13, 15, 16, 18 have one VvCIPK gene, chromosomes 5, 8, and 9 have two VvCIPK genes, and chromosomes 6, 10, and 11 have three VvCIPKs. No VvCIPK genes are located on grapevine chromosomes 1, 3, 7, 12, 14, and 17 (Figure 3, Table 1). Furthermore, the gene duplication events were analyzed by using MCScanX program and mapped into chromosome location. As shown in Figure 3, two tandem duplication events were detected between 4 VvCBL genes, accounting for 50% of VvCBL gene family. The gene pairs include VvCBL5/VvCBL8 (which belong to subgroup IV) and VvCBL10a/VvCBL10b (which belong to subgroup I). In contrast, no segmental duplications of VvCBL genes were detected. Four tandem duplication events were detected between 8 out of 20 (40%) grapevine CIPK genes including VvCIPK38/VvCIPK28, VvCIPK34/VvCIPK27, VvCIPK33/VvCIPK30, and VvCIPK36/VvCIPK12. Additionally, four segmental duplication events were detected between 5 out of 20 (25%) VvCIPK genes. The gene pairs include VvCIPK38/VvCIPK34, VvCIPK38/VvCIPK37, VvCIPK37/VvCIPK34 of subgroup C, and VvCIPK28/VvCIPK27 of subgroup D, and among them, 4 VvCIPK genes including VvCIPK38/VvCIPK28, VvCIPK34/VvCIPK27 have already shown to be derived from tandem duplication events. Interestingly, all gene duplication events of VvCIPKs were only detected in intron-poor clade (Figure 2).

In addition, to better understand the duplication and functional divergence of VvCBL and VvCIPK genes during their evolution course, the Ka, Ks, and Ka/Ks ratio were calculated. As shown in Figure 4, Table S2, the Ka/Ks ratio of two VvCBL tandem duplicated gene pairs were 1.02 and 1.26 with an average of 1.14, and these tandem duplication events of VvCBL genes therefore were likely happened around 11 to 17 averaged about 14 Mya (Figure 4, Table S2). The Ka/Ks ratio of one of VvCIPK tandem duplicated gene pair was 0.26 and which of four VvCIPK segmental duplicated gene pairs were ranged from 0.11 to 0.14 with an average of 0.26. Tandem duplication events of VvCIPK genes therefore might happen around 17 Mya, and segmental duplication events of VvCIPK genes might happen around 8 to 14 averaged about 11 Mya (Figure 4, Table S2). However, No VvCBL orthologous gene pairs with Arabidopsis, rice and poplar were identified. On the contrary, 7 CIPK orthologous gene pairs existed between grapevine and poplar with an average Ka/Ks ratio of 0.11, 2 CIPK orthologous gene pairs existed between grapevine and Arabidopsis with an average Ka/Ks ratio of 0.06, 4 CIPK orthologous gene pairs existed between grapevine and rice with an average Ka/Ks ratio of 0.14 (Table S3). The data indicated that CIPK genes among these species were under strong purifying selection.

To gain more insight into the regulatory mechanism of VvCBL and VvCIPK genes, The cis-acting elements in 2,000 bp upstream sequences of coding region of VvCBL and VvCIPK genes were surveyed by PlantCARE software. Besides the basic TATA and CAAT boxes and other uncertainty function cis-acting, 80 cis-acting element were detected, and among them, 65 well-function annotated cis-acting were arbitrarily divided into four main kinds of cis-acting elements based on their biological functions (Table S4). The first kind was light responsive related elements such as AE-box, rbcS-CMA7a, G-box, SP1, and Box I (Table S4). The Second type was hormone responsive elements, such as GARE-motif, CGTCA-motif, TGA-element, TCA-element, and TATC-box (Table S4). The third type was environment stress related elements, as shown in Table S4, all VvCBL and VvCIPK genes were found to contain HSE, a cis-acting element involved in heat response except VvCBL12 and VvCIPK31. VvCBL4, VvCBL13, VvCBL5, VvCIPK38, VvCIPK37, VvCIPK34, VvCIPK30, VvCIPK31, VvCIPK32, VvCIPK33, VvCIPK12, VvCIPK29, VvCIPK27, VvCIPK9, VvCIPK3, VvCIPK40 contained MBS, which is MYB binding site involved in drought-deducibility. VvCBL5, VvCIPK32, VvCIPK12, VvCIPK41, VvCIPK40 contained LTR, which is a cis-acting element involved in low temperature responsiveness. WUN-motif, Box-W1, ARE, GC-motif, TC-rich repeats were also detected in VvCBL and VvCIPK genes. The last kind was plant development cis-acting elements, such as Skn-1_motif, which is a cis-acting regulatory element required for endosperm expression, it was detected in all VvCBL and VvCIPK genes except VvCIPK21 (Table S4), and As-2-box, which involved in both shoot specific expression and light responsive (Table S4), and it was detected in VvCBL13, VvCIPK34, VvCIPK41, and VvCIPK21.

Previous studies have demonstrated the important roles that CBL and CIPK genes play in plant response to abiotic stresses. Therefore, qRT-PCR analysis was conducted to characterize the expression profiles of grapevine CBL and CIPK genes when plants were subjected to various stress conditions. Transcript levels of all VvCBL and VvCIPK were characterized in 6-week-old grapevine leaves from plants that were subjected to various stress treatments, including ion stress including salt (NaCl; 200 mM) and low potassium (LK; 100 μM), osmotic stress (PEG; 10%), and low (4°C) and high (42°C) temperature stresses. As shown in Figure 5A and Figure S4, transcript levels of all VvCBL and VvCIPK genes were altered in response to the various stress treatments, and Figure 6A also show the more than three-fold changes up/down regulated genes. VvCBL10a and VvCBL10b of subgroup I were derived from tandem duplication, they were up-regulated by salt, PEG and cold stress, and were down-regulated by heat stress, however, VvCBL10a was down-regulated and VvCBL10b was up-regulated by LK stress (Figures 2–4). VvCBL11 and VvCBL12 of subgroup II have same expression profile, they were up-regulated by salt, PEG and cold stress, and were down-regulated by LK and heat stress (Figures 2, 4). VvCBL5 and VvCBL8 of subgroup IV were derived from tandem duplication, they were up-regulated by salt, PEG and LK stress, but VvCBL5 was down-regulated and VvCBL8 was up-regulated by temperature stress (Figures 2–4). Besides, VvCBL8 was highly induced in response to most of the stress treatments (Figure 4).

VvCIPK24, VvCIPK39, and VvCIPK3 were up-regulated by almost all stress treatments which were intron-rich clade of subgroup A (Figures 2, 4). In contrast, VvCIPK34, VvCIPK35, and VvCIPK36 of subgroup C were down-regulated in response to almost all of the stress treatments (Figures 2–4). VvCIPK12 which was derived from tandem duplication with VvCIPK36, was highly up-regulated by high temperature stress (Figures 3, 4). VvCIPK27 which was derived from tandem duplication with VvCIPK34, were down-regulated by salt, PEG, and LK stress but up-regulated by cold and heat stress (Figures 3, 4). The expression profile of VvCIPK28 is same to VvCIPK27 expect the cold stress, as was derived from segmental duplication with VvCIPK27 (Figures 3, 4). VvCIPK38 which was derived from tandem duplication with VvCIPK28, was highly up-regulated in response to temperature stress (Figures 3, 4). But VvCIPK37 which was derived from segmental duplication with VvCIPK38, was down-regulated by temperature, especially heat stress (Figures 3, 4). Besides, VvCIPK33 and VvCIPK30, which were derived from tandem duplication, have same expression profile, were down-regulated by salt, PEG and LK stress and up-regulated by temperature stress (Figures 3, 4).

Studies have also reported the participation of CBL and CIPK genes in different development processes (Tripathi et al., 2009; Kanwar et al., 2014). Therefore, the expression pattern of CBL and CIPK genes in different grapevine tissues and organs was examined in order to obtain more insight into their role in plant growth and development. Figure 5B presents a heat map of the expression pattern of VvCBL and VvCIPK genes based on microarray data from 54 grapevine samples (Fasoli et al., 2012). Specific information about the different samples is provided in Table S6. The genes were ordered based on a hierarchical clustering analysis. As illustrated in Figure 5B, VvCIPK39, VvCIPK37, and VvCIPK3 have a high level of expression throughout the various developmental stages of different grapevine organs and tissues. In contrast, VvCBL8, VvCIPK35, VvCBL5 exhibited low levels of expression in the microarray data. These data collectively illustrate that grapevine CBL and CIPK respond to developmental stimuli.

Figure 6A illustrates the networks of high confidence of interactive Arabidopsis CBL and CIPK proteins using STRING software, and identifies their VvCBL and VvCIPK homolog proteins by reciprocal best BLASTP analysis. The expression profiles of VvCBL and VvCIPK in response to abiotic stresses and the gene pairs with PCC value ≥0.7 were also intuitively added to the figure. The different interaction networks of CBL and CIPK proteins suggest that divergent mechanisms involving specific CBL/CIPK proteins may be present in Arabidopsis and grape.

The Pearson's correlation coefficient (PCC) among VvCBL and VvCIPK genes was calculated based on the qRT-PCR data in order to further characterize the co-expression relationship between VvCBLs and VvCIPKs. As shown in Table S7, the PCC value of 20 gene pairs was more than 0.7 at the 0.01 level (2-tailed), and all these pairs exhibited a significant positive correlation. Importantly, the VvCBL13 and VvCIPK30 pair exhibited both a strong PCC value and a similar expression response in the qRT-PCR data as both were significantly up-regulated in response to both low and high temperature stress (Figure 6A, Table S7). VvCBL5 and VvCIPK29 pair also had both a strong PCC value and a similar expression pattern in the qRT-PCR data. They were both significantly up-regulated in response to ion (salt and low K⁺) and osmotic (PEG) stress and significantly down-regulated during some of the time points of temperature stress (Figure 6A, Table S7). Besides, as shown in Figure 6A, 11 strong correlated gene pairs (55%) were up-regulated by low temperature stress, including VvCBL4/VvCBL8, VvCBL10b/VvCBL8, VvCBL10b/VvCBL11, VvCBL13/VvCIPK30, VvCBL10a/VvCBL10b, VvCBL4/VvCBL11, VvCBL10a/VvCBL4, VvCBL10b/VvCBL4, VvCBL10a/VvCBL11. Most of them were VvCBL genes, so the expression profile of VvCBLs in response to low temperature were more intuitively drafted as Figure 6B individually, we found that all VvCBLs were highly up-regulated by cold stress at 24 h except VvCBL5.

PCC and co-expression network analysis were also carried out using the VvCBL and VvCIPK microarray data. Results indicated that 56 pairs of 28 VvCBLs and VvCIPKs exhibited a significant correlation. The PCC values of 42 pairs were ≥0.35 and exhibited a significant positive correlation and the PCC values of 14 pairs were ≤ -0.35 and exhibited a significant negative correlation. Among the significant correlations, the PCC values of four gene pairs were ≥0.7 and those of four gene pairs were ≤ -0.7. PCC of VvCIPK32/VvCIPK38, VvCBL12/VvCIPK32, VvCBL10a/VvCIPK32, and VvCBL12/VvCBL10a exhibited a positive correlation each with a PCC ≥0.7.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.