Circadian Regulation of Alternative Splicing of Drought-Associated CIPK Genes in Dendrobium catenatum (Orchidaceae)

Dendrobium catenatum, an epiphytic and lithophytic species, suffers frequently from perennial shortage of water in the wild. The molecular mechanisms of this orchid’s tolerance to abiotic stress, especially drought, remain largely unknown. It is well-known that CBL-interacting protein kinase (CIPKs) proteins play important roles in plant developmental processes, signal transduction, and responses to abiotic stress. To study the CIPKs’ functions for D. catenatum, we first identified 24 CIPK genes from it. We divided them into three subgroups, with varying intron numbers and protein motifs, based on phylogeny analysis. Expression patterns of CIPK family genes in different tissues and in response to either drought or cold stresses suggested DcaCIPK11 may be associated with signal transduction and energy metabolism. DcaCIPK9, -14, and -16 are predicted to play critical roles during drought treatment specifically. Furthermore, transcript expression abundances of DcaCIPK16 showed polar opposites during day and night. Whether under drought treatment or not, DcaCIPK16 tended to emphatically express transcript1 during the day and transcript3 at night. This implied that expression of the transcripts might be regulated by circadian rhythm. qRT-PCR analysis also indicated that DcaCIPK3, -8, and -20 were strongly influenced by circadian rhythmicity. In contrast with previous studies, for the first time to our knowledge, our study revealed that the major CIPK gene transcript expressed was not always the same and was affected by the biological clock, providing a different perspective on alternative splicing preference.


Introduction
Drought stress-a major environmental factor-affects the geographical distribution of plants, restricts crop productivity in agriculture, and causes food security problems [1]. Dendrobium catenatum is an epiphyte that sticks to the surface of other plants or to naked rocks in the natural environment [2,3]. Because of minimal water interception in these habitats, its roots seriously lack available water and nutrients [4]. To survive the tough environment, D. catenatum has evolved succulent pseudobulbs and crassulacean acid metabolism (CAM), a photosynthetic metabolic pathway with high water use efficiency, in response to such a harsh natural environment [2,5,6]. As a highly drought-tolerant species, D. catenatum can serve as a model to study survival strategies of plants coping with drought. Studying the gene spectrum of a highly resistant plant that spans evolutionary timescales is the most direct and effective way to implement stress tolerance gene engineering.
The homologous proteins SnRK in green plants, SNF1 in yeast, and AMPK in mammals belong to the SNF1 protein kinase superfamily. Stress-signaling pathways related to yeast SNF1 and mammalian In total, 24 genes containing both a NAF domain and a kinase domain were identified as CIPK gene family members in the orchid. Twenty-two members inferred from the orchid genome annotation [2] and two new CIPK genes (Novl001335 and Novl001612) were newly annotated from novel genetic loci with transcript assembly. The number of amino acid residues of DcaCIPK primary proteins ranged from 401 to 502. The relative molecular weights of these CIPK kinase proteins varied from 45.15 to 57.14 KDa. All the proteins had high isoelectric points (pI > 7.0). Detailed information for these genes is listed in Table 1.  Phylogenetic analysis classified the CIPK family members into three subgroups (I, II, and III), whose genes contain nine, five, and ten members, respectively ( Figure 1A). To further study the features of the genes (from primary genome annotation), the composition and position of exons, introns, and conserved elements were analyzed. The DcaCIPK gene members were clearly divided into an intron-rich clade (>8 intron per gene) and an intron-poor clade [19,23]. All the intron-rich clade members belonged to subgroup I, and those intron-poor clade members related to subgroups II and III ( Figure 1B). In subgroups III and II, their genes have introns ranged from zero to four ( Figure 1C). In subgroup I, most members contained 13 introns, two members (DcaCIPK1, DcaCIPK21) contained 12 introns, and only one gene (DcaCIPK17) had 11 introns ( Figure 1C). Phylogenetic analysis classified the CIPK family members into three subgroups (I, II, and III), whose genes contain nine, five, and ten members, respectively ( Figure 1A). To further study the features of the genes (from primary genome annotation), the composition and position of exons, introns, and conserved elements were analyzed. The DcaCIPK gene members were clearly divided into an intron-rich clade (>8 intron per gene) and an intron-poor clade [19,23]. All the intron-rich clade members belonged to subgroup I, and those intron-poor clade members related to subgroups II and III ( Figure 1B). In subgroups III and II, their genes have introns ranged from zero to four ( Figure 1C). In subgroup I, most members contained 13 introns, two members (DcaCIPK1, DcaCIPK21) contained 12 introns, and only one gene (DcaCIPK17) had 11 introns ( Figure 1C).

Phylogenetic Analysis of CIPK in Plants
An NJ (neighbor-joining) phylogenetic tree was built to scrutinize the evolutionary relationships of CIPK family members. Eighty-three protein sequences, including those from Arabidopsis, rice, and D. catenatum, were divided into five groups (Groups I-V). Proteins from all nine intron-rich DcaCIPKs were clustered in Group IV. Including nine AtCIPKs and 11 OsCIPKs, Group IV was the largest group. Group I comprised seven DcaCIPKs, five AtCIPKs, and 13 OsCIPKs. Group II consisted of three DcaCIPKs, four AtCIPKs, and four OsCIPKs. Group III was the smallest group, containing one DcaCIPK, three AtCIPKs, and one OsCIPK. Finally, four DcaCIPKs, five AtCIPKs, and four OsCIPKs were within Group V ( Figure 3).

Phylogenetic Analysis of CIPK in Plants
An NJ (neighbor-joining) phylogenetic tree was built to scrutinize the evolutionary relationships of CIPK family members. Eighty-three protein sequences, including those from Arabidopsis, rice, and D. catenatum, were divided into five groups (Groups I-V). Proteins from all nine intron-rich DcaCIPKs were clustered in Group IV. Including nine AtCIPKs and 11 OsCIPKs, Group IV was the largest group. Group I comprised seven DcaCIPKs, five AtCIPKs, and 13 OsCIPKs. Group II consisted of three DcaCIPKs, four AtCIPKs, and four OsCIPKs. Group III was the smallest group, containing one DcaCIPK, three AtCIPKs, and one OsCIPK. Finally, four DcaCIPKs, five AtCIPKs, and four OsCIPKs were within Group V ( Figure 3).

Expression of D. catenatum CIPK Genes in Different Plant Tissues
To determine the roles of CIPK genes in D. catenatum, reassembled transcriptome data was used to illustrate the expression of DcaCIPKs in different tissues (root tip, root, flower bud, stem, and leaf; Figure 4; Table S1). Two primary clusters for expression patterns of DcaCIPK genes were found and both of them contain intron-poor and intron-rich genes ( Figure 4). Most DcaCIPK genes, except DcaCIPK13, were detected in these five tissues. Among these detected genes, DcaCIPK6, -15, and -18 were the most highly expressed genes in root tip and root; DcaCIPK1, -14, and -15 were highly expressed in stem; DcaCIPK3, -14, and -19 were highly expressed in leaf; and DcaCIPK6 and -18 were highly expressed in flower bud. These results suggested these specific highly expressed genes may have important roles in development or sensing of external signals in the corresponding tissues.

Expression of D. catenatum CIPK Genes in Different Plant Tissues
To determine the roles of CIPK genes in D. catenatum, reassembled transcriptome data was used to illustrate the expression of DcaCIPKs in different tissues (root tip, root, flower bud, stem, and leaf; Figure 4; Table S1). Two primary clusters for expression patterns of DcaCIPK genes were found and both of them contain intron-poor and intron-rich genes ( Figure 4). Most DcaCIPK genes, except DcaCIPK13, were detected in these five tissues. Among these detected genes, DcaCIPK6, -15, and -18 were the most highly expressed genes in root tip and root; DcaCIPK1, -14, and -15 were highly expressed in stem; DcaCIPK3, -14, and -19 were highly expressed in leaf; and DcaCIPK6 and -18 were highly expressed in flower bud. These results suggested these specific highly expressed genes may have important roles in development or sensing of external signals in the corresponding tissues.

Expression of D. catenatum CIPK Genes under Abiotic Stress
To figure out how CIPK gene expression changes in response to abiotic stress, we analyzed the expression patterns of CIPK members. Dataset A (see section Materials and methods) represents DcaCIPK expression profiles in D. catenatum leaves during the day (9:00 am) and night (9:00 pm) under drought treatment ( Figure 5A). Dataset B represents DcaCIPK expression profiles obtained under cold-and control-treatments ( Figure 5B). CIPK20 and -21 were significantly fluctuant under drought and cold treatment as compared with normal conditions, respectively.

Expression of D. catenatum CIPK Genes under Abiotic Stress
To figure out how CIPK gene expression changes in response to abiotic stress, we analyzed the expression patterns of CIPK members. Dataset A (see section Materials and methods) represents DcaCIPK expression profiles in D. catenatum leaves during the day (9:00 am) and night (9:00 pm) under drought treatment ( Figure 5A). Dataset B represents DcaCIPK expression profiles obtained under coldand control-treatments ( Figure 5B). CIPK20 and -21 were significantly fluctuant under drought and cold treatment as compared with normal conditions, respectively.
Interestingly, we found transcripts in DcaCIPK1 and DcaCIPK3 had specific expression patterns: all transcripts in the former gene showed a high abundance in nights but extremely low levels in days ( Figure 6B), and the two main transcripts (tanscript1 and transcript3) in the latter gene had a mutually day-and-night reversed expression ( Figure 6B). Hence, we analyzed the protein domain of their main expressed transcripts on Pfam according to the hidden Markov model (HMM) profile ( Figure 7). Transcript1, 3, and 4 are DcaCIPK1′s major isoforms. Besides including the two basic protein domains-pkinase and NAF-KA1 domain is found in transcript1 and 3. Moreover, KA1 is closely adjacent to NAF domain in transcript1 compared with that in transcript3. As for DcaCIPK3, transcript2 and transcript5 are the main isoforms of alternative splicing. However, the pkinase domain in transcript2 is incomplete.
Interestingly, we found transcripts in DcaCIPK1 and DcaCIPK3 had specific expression patterns: all transcripts in the former gene showed a high abundance in nights but extremely low levels in days ( Figure 6B), and the two main transcripts (tanscript1 and transcript3) in the latter gene had a mutually day-and-night reversed expression ( Figure 6B). Hence, we analyzed the protein domain of their main expressed transcripts on Pfam according to the hidden Markov model (HMM) profile ( Figure 7). Transcript1, 3, and 4 are DcaCIPK1 s major isoforms. Besides including the two basic protein domains-pkinase and NAF-KA1 domain is found in transcript1 and 3. Moreover, KA1 is closely adjacent to NAF domain in transcript1 compared with that in transcript3. As for DcaCIPK3, transcript2 and transcript5 are the main isoforms of alternative splicing. However, the pkinase domain in transcript2 is incomplete.  Prediction of cis-acting regulatory elements indicated that these genes all had elements that are involved with light responsiveness. A circadian element was found in DcaCIPK2, -3, -8, and -12. The MBS element, an MYB binding site involved in drought-inducibility, was found in DcaCIPK5, -12, -15, -16, and -19. DcaCIPK15 was found to have a low-temperature responsiveness element. In addition, many of these genes contained plant hormone responsiveness elements such as an ABA-responsive element, a methyl jasmonate inducible element, a salicylic acid-inducible element, an ethylene-responsive element, a gibberellin-responsive element, and an auxin-responsive element (Table S5). Prediction of cis-acting regulatory elements indicated that these genes all had elements that are involved with light responsiveness. A circadian element was found in DcaCIPK2, -3, -8, and -12. The MBS element, an MYB binding site involved in drought-inducibility, was found in DcaCIPK5, -12, -15, -16, and -19. DcaCIPK15 was found to have a low-temperature responsiveness element. In addition, many of these genes contained plant hormone responsiveness elements such as an ABA-responsive element, a methyl jasmonate inducible element, a salicylic acid-inducible element, an ethylene-responsive element, a gibberellin-responsive element, and an auxin-responsive element (Table S5).
Under drought stress, the expression level of DcaCIPK3 was upregulated, but the major functional transcript was not the same between day and night. During daytime, transcript5 was the major transcript detected. In contrast, transcript2 was the major transcript at nighttime ( Figure 6B, Figure 9C). Promoter region analysis showed DcaCIPK3 has two circadian rhythm binding sites located at 54 and 84 bp (marked as green boxes in Figure 8).  Under drought stress, the expression level of DcaCIPK3 was upregulated, but the major functional transcript was not the same between day and night. During daytime, transcript5 was the major transcript detected. In contrast, transcript2 was the major transcript at nighttime ( Figure 6B, Figure 9C). Promoter region analysis showed DcaCIPK3 has two circadian rhythm binding sites located at 54 and 84 bp (marked as green boxes in Figure 8). Prediction of cis-acting regulatory elements indicated that these genes all had elements that are involved with light responsiveness. A circadian element was found in DcaCIPK2, -3, -8, and -12. The MBS element, an MYB binding site involved in drought-inducibility, was found in DcaCIPK5, -12, -15, -16, and -19. DcaCIPK15 was found to have a low-temperature responsiveness element. In addition, many of these genes contained plant hormone responsiveness elements such as an ABA-responsive element, a methyl jasmonate inducible element, a salicylic acid-inducible element, an ethylene-responsive element, a gibberellin-responsive element, and an auxin-responsive element (Table S5).
Under drought stress, the expression level of DcaCIPK3 was upregulated, but the major functional transcript was not the same between day and night. During daytime, transcript5 was the major transcript detected. In contrast, transcript2 was the major transcript at nighttime ( Figure 6B, Figure 9C). Promoter region analysis showed DcaCIPK3 has two circadian rhythm binding sites located at 54 and 84 bp (marked as green boxes in Figure 8). Figure 8. Analysis of the important cis-regulatory elements in the sequence 1 kb upstream of DcaCIPK3. AT1-motif, part of a light responsive module; GCN4 motif, cis-regulatory element involved in endosperm expression; circadian, cis-acting regulatory element involved in circadian control; TGACG-motif, cis-acting regulatory element involved in MeJA responsiveness; Box 4, part of a conserved DNA module involved in light responsiveness; TGA-element, auxin-responsive element; GATA-motif/I-box, part of a light-responsive element; box S, wounding and pathogen responsiveness; G-box, cis-acting regulatory element involved in light responsiveness; GT1-motif, light-responsive element; Box I, light-responsive element; TCT-motif, part of a light-responsive element; Sp1, light-responsive element; ARE, cis-acting regulatory element essential for anaerobic Figure 8. Analysis of the important cis-regulatory elements in the sequence 1 kb upstream of DcaCIPK3. AT1-motif, part of a light responsive module; GCN4 motif, cis-regulatory element involved in endosperm expression; circadian, cis-acting regulatory element involved in circadian control; TGACG-motif, cis-acting regulatory element involved in MeJA responsiveness; Box 4, part of a conserved DNA module involved in light responsiveness; TGA-element, auxin-responsive element; GATA-motif/I-box, part of a light-responsive element; box S, wounding and pathogen responsiveness; G-box, cis-acting regulatory element involved in light responsiveness; GT1-motif, light-responsive element; Box I, light-responsive element; TCT-motif, part of a light-responsive element; Sp1, light-responsive element; ARE, cis-acting regulatory element essential for anaerobic induction; CAT-box, cis-acting regulatory element related to meristem expression; GA-motif, part of a light-responsive element.

Effect of Time-of-Day and Drought Stress on Selected Genes mRNA Expression
We selected six genes (DcaCIPK1, -3, -7, -12, -15, and -16) to conduct qRT-PCR validation of their mRNA expression (Figure 9). Time-of-day affected the expression of four selected genes (DcaCIPK1, -3, -7, and -15) under drought and control conditions. The expression of DcaCIPK1, -7, and -15 exhibited a strong circadian rhythmicity and peaked during the active phase both under drought and control. DcaCIPK1 significantly elevated during dark phase, while DcaCIPK7 and 15 were active in day time. The expression levels under drought were lower than that under control both for DcaCIPK7 and -15. DcaCIPK3 had no obviously difference in day and night, but the gene expression level under drought was higher than that under control. qRT-PCR assays also validated the predicted expression patterns of DcaCIPK12 and DcaCIPK16 under drought treatment, respectively. The statistics result shows differences of expression level between drought and control is significant. The average expression level of the two genes under drought stress is higher than that under control, respectively.

Discussion
How to sense and conduct external stress signals is a basic biological competence in plants. Osmotic stress is the major signal caused by drought stress [10]. As a major calcium sensor protein, Light phase began at Zeitgeber Time 6 (ZT6) and dark phase (shaded areas) began at ZT18. Data are presented as mean ± SEM. Experiments were performed 4 times. * indicates differences the effect of base material water content is significant. Different letters above the column indicate statistically significant differences among sampling time treatment (p < 0.05). (C) Relative abundance of drought treatment on transcripts (2 and 5) of DcaCIPK3 expression. DD represents drought treatment in daytime; DN represents drought treatment in nighttime; MD represents moist treatment in daytime; MN represents moist treatment in nighttime. "2" and "5" after "DD, MD, MN, and MN" represent transcript2 and 5, respectively.

Discussion
How to sense and conduct external stress signals is a basic biological competence in plants. Osmotic stress is the major signal caused by drought stress [10]. As a major calcium sensor protein, CIPK functioning with CBL could transfer the stress signal downstream. The CIPK family has been analyzed in many C3 and C4 plants [18,19], but comprehensive information about CAM plants is limited. D. catenatum, a facultative CAM species, has a high level of drought resistance. Previous researches related to this species have mostly focused on polysaccharide hydrolysis and synthesis [24,25], while genes involved in signal transduction have been rarely studied.

Phylogenetics and Structure of DcaCIPK Genes
The phylogenetic analyses of gene structure and evolution provided a theoretical basis for the functional annotation of DcaCIPK genes. The intron-rich CIPK members from Arabidopsis, rice, and Dendrobium catenatum were all clustered in Group IV, whereas the genes in the other four groups were all intron-poor members (Figure 3) [18,20,26]. The closely related orthologous CIPKs among different species suggest that an ancestral series of CIPK genes existed before species divergence. The bipolarity of intron numbers in D. catenatum genes was also observed in those of other species such as Arabidopsis, maize, and rice; these cases suggest that intron loss and gain events may have happened before the monocot-eudicot divergence [18,19,21].
Previous reports have demonstrated the critical role of CIPK genes under diverse stress, which drove us to collect and analyze the expression profiles of these genes in different tissues and under various abiotic stresses from public transcriptome data. The analysis of global gene expression patterns indicated that the most drought-and cold-inducible genes belonged to the intron-poor cluster ( Figure 5). This finding was consistent with the analysis of the CIPK family in soybean [21]. Previous research has indicated that the intron-poor CIPK group evolved much later, very likely derived by discarding introns of intron-rich members [21,27]. Therefore, the loss of introns may be an adaption to environmental stress.

DcaCIPK Gene Expression and Functions in Drought Stress
Our results revealed that DcaCIPK3, -12, and -16 can be significantly induced during both the day and the night under drought treatment ( Figures 5, 6 and 9B). DcaCIPK16 is orthologous to AtCIPK16 (Figure 3), which can phosphorylate AKT1 to regulate the absorption of potassium by interacting with CBL1/CBL9 [28]. DcaCIPK16 was thus predicted to have a similar function to AtCIPK16 in dealing with the iron stress resulting from drought. DcaCIPK3, as an orthologous gene of AtCIPK3, may play a critical role in ABA signaling and multiple stress responses as shown by studies on AtCIPK3 [29]. AtCIPK3 can be rapidly activated by salt, drought, cold, and osmotic stress treatments. Both drought and salt have a common ABA-dependent pathway. Disruption of AtCIPK3 altered the expression patterns of stress-responsive genes (RD29A and KIN1/KIN2) triggered by cold, high salt, and ABA. However, the disruption of AtCIPK3 did not affect the expression of genes (RD29A and KIN1/KIN2) induced by drought, suggesting that AtCIPK3 acts in a distinct way in response to drought stress. On account that cold-induced gene expression has been shown to be independent of ABA production, these results indicated AtCIPK3 represents a cross-talk node between the ABA-dependent and ABA-independent pathways in stress responses [29,30]. The precise role of DcaCIPK12 cannot be elucidated from evolutionary relationships because the function of neighboring AtCIPKs (AtCIPK12/19) is still unclear. These genes were upregulated during both the day and the night under water deficit stresses and therefore are forecast to play a critical role in drought resistance.
DcaCIPK14 was significantly upregulated under two types of abiotic stress, which indicated this gene presumably had a functional overlap with responses to both water deficit and low-temperature stress ( Figure 5). Moreover, DcaCIPK14 was abundantly expressed in pseudobulbs and leaves, which are rich in polysaccharides (Figure 4). Plentiful polysaccharides could supply a large amount of bound water and also hydrolyze into soluble monosaccharides in case of severe abiotic stress [31][32][33]. Previous studies have indicated that the CIPK gene may have a link with the sucrose content of plants. For example, ScCIPK8, a sugarcane CIPK gene, had a negative correlation with sucrose content of leaves [34,35]. Light-interruption stress, which inhibits the synthesis of sucrose, reduces the expression of AtCIPK14 expression; after feeding sucrose, however, the expression of AtCIPK14 goes up [36]. As an orthologous gene of AtCIPK14, DcaCIPK14 may act in connecting and regulating stress signal induction and energy metabolism.

Circadian Rhythm and Drought Stress Both Influence Alternative Splicing of CIPK Members
The abundance analysis for DcaCIPK genes' alternative splicing isoforms indicated that circadian rhythm influences expression levels of transcripts. DcaCIPK3 may play a significant role in drought response similar to the orthologous gene AtCIPK3. The overall abundance of DcaCIPK3 transcripts and the time-of-day qRT-PCR results also supported the important role of this gene in drought stress. AtCIPK3 has the highest number of splice variants of CIPK members in A. thaliana, and two of the transcripts (AtCIPK3.1 and AtCIPK3.4) were more highly induced under ABA and drought treatment [37]. Similarly, preference of specific transcripts was also found in D. catenatum. However, it seemed this phenomenon was more likely raised by circadian rhythm in our experiment. Under drought treatment, the expression levels of different transcripts suggested that the most highly expressed transcript variant of DcaCIPK3 tended to be transcript1 during the daytime and transcript3 at nighttime. The moist treatment had a negative effect on the sum expression abundance of various transcripts, but did not change the routine preference of transcript1 during the day and transcript3 at night, respectively. The analysis of transcript isoforms and the prediction of cis-acting regulatory elements of DcaCIPK3 indicated that circadian rhythm may influence the preference of alternative splicing ( Figure 6). DcaCIPK3 may use different splicing ways during day and night to cope with drought stress. DcaCIPK3 represents a critical role for improving tolerance of D. catenatum to drought and deserves further analysis to reveal its functional roles in drought response and the underlying molecular mechanisms for the preference of splicing, which is currently underway.
Diurnal variation in gene expression levels might be ascribed to the impact of biological clocks. If the sampling time had been fixed during the day, the association with drought tolerance of genes such as DcaCIPK1, -7, and -15 ( Figures 6 and 9 and Table S4) would have been missed, because its expression level was nearly negligible. Therefore, the choice of sampling time should be taken into consideration in terms of the influence of the bioclock.
Alternative splicing increases gene functional diversity by regulating alternative exon recognition [38]. Our analysis indicated that this combinatorial control mechanism greatly influenced the protein domain in main expressed transcript (Figure 7). Different domain combinations may undertake different duties. Incomplete pkinase domain in transcript2 of DcaCIPK3 may cause losing function of phosphorylation which eventually prevents signal downward transduction in night time.
As a facultative CAM plant, the specific photosynthetic pathway may also have a great influence on the induction of DcaCIPK splice variants. How DcaCIPK genes are influenced by the biological clock and how drought stress affects the abundance of transcripts should be further studied.

Genome-Wide Identification of the CIPK Gene Family in D. catenatum
The D. catenatum genome (assembly ASM160598v1) [2] and its annotation (general feature format, GTF) were downloaded from the National Center for Biotechnology Information (NCBI) to extract protein and CDS sequences for all genes. To identify CIPK proteins, the hidden Markov model (HMM) profile of NAF with the signature domains (PF03822) was downloaded from Pfam [39] and used to search by HMMER 3.0 [40]. Each DcaCIPK protein sequence was examined for the presence of the serine/threonine protein kinase domain in the N-terminus to be considered as a member of the D. catenatum CIPK family. The putative CIPK family members were further reviewed by Pfam and SMART software [41]. Molecular weight (MW) [42] and isoelectric point (pI) [43,44] of CIPK members were calculated by "Compute pI/Mw function"of ExPASy [45].

Multiple Protein Sequence Alignment and Phylogenetic Tree Construction
The protein sequences of all DcaCIPK family members were aligned using Muscle [46] with default parameters, and a phylogenetic tree was constructed using neighbor-joining (NJ) [47] method built-in MEGA7.0 [48]. The bootstrap values for phylogenetic trees were based on 1000 replicates. We downloaded amino acid sequences for 26 AtCIPK [14,17] and 33 OsCIPK [20] proteins from NCBI and pooled them with sequences of 24 DcaCIPK proteins. An NJ phylogenetic tree for protein sequences from D. catenatum, Arabidopsis thaliana, and Oryza sativa was constructed by using the same procedure described above, and pairwise deletion and the Poisson model were introduced.

Exon-Intron Structure Analysis and Identification of Conserved Motifs
Visualization of gene features was performed using the gene structure display server by mapping the sequences coding for amino acids in proteins (CDS) to genomic sequences [49]. Motifs were discovered by MEME [50]. The parameters were set as follows; number of repetitions, any; maximum number of motifs, 30; optimum motif width, between 6 and 200 residues.

Transcriptome Analysis of D. catenatum CIPK Gene Expression
The expression patterns of CIPK gene family members in D. catenatum under different stress treatments (drought [51], Dataset A; low temperature [31], Dataset B) and in different tissues (Dataset C) [52] were analyzed ( Table 2). Transcriptome data for drought stress, cold treatment, and specific tissues were downloaded from the Sequence Read Archive (SRA) database at the NCBI. "n.a." indicates that the current item is not available.
The raw data was stripped of adapters and low-quality reads (and bases) and then rRNA and virus data were filtered out, which were performed by using software Fastq_clean [53] with the settings described in our previous work [51]. The clean reads were aligned to the D. catenatum genome using Hisat2 [54,55] with options: -dta and -no-unal. The aligned outputs were converted from SAM to BAM format by using samtools [56]. The "htseq-count" function of Python package Htseq [57] was used to calculate the read-count of genes with default parameters. Expression differences among gene family members were further analyzed by using the R package DESeq2 with relative log expression (RLE) normalization and an adjusted p-value cutoff 0.05 [58]. HemI 1.0 [59] was used to illustrate heatmaps with settings: average linkage for clustering and Pearson distance for similarity metric.

Alternative Splicing Analysis and Cis-Acting Regulatory Element Prediction
As the protocol suggested by Pertea et al. [55], the BAM files described above were used as inputs for StringTie [60] to assemble and quantify transcripts of each gene. The visualization of transcripts, including structure and expression levels, was achieved using R [61] package Ballgown [62]. Prediction of cis-acting regulatory elements in promotor sequences was performed by using PlantCARE [63]. The sketch of the domains in main expressed transcripts was obtained at Pfam by inputting transcripts amino acid sequences [39]. 4.6. Plant Material, Growth Conditions, and Experimental Treatments D. catenatum plants were grown in a 5-cm diameter plot in an artificial climate chamber maintained under 12 h light (light intensity~100 µmol·m −2 s −1 , 28 • C, 60% relative humidity)/12 h dark (22 • C, 70% relative humidity) cycles. Irrigation was withheld from the drought group until the volumetric water content of the base material dropped to 0%. The volumetric water content of base material was kept at between 30% and 35% in control group and~0% in drought group. The topmost mature leaves, generally the fourth and fifth leaves, were harvested with a two-hour interval in a day (8:00 am-24:00 pm-6:00 am) and immediately frozen in liquid nitrogen. Moreover, we also sample specifically at 9:00 AM at day and 9:00 PM at night. These samples were for Quantitative Real-Time PCR (qRT-PCR) assessments. Each experimental treatment was performed in four replicates.

RNA Extraction
RNA was isolated from ground tissue using the RNAprep Pure Plant Kit (Polysaccharides & Polyphenolics-rich, TIANGEN Co. LTD, Beijing, China) according to the manufacturer's protocols. RNA integrity was analyzed on a 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA).

Quantitative Real-Time PCR
Primer pairs (Table S6) were designed by Primer 3. qRT-PCR experiments were performed using Roche LightCycler ® 480 II system (LC480, Basel, Switzerland) with TB Green Premix Ex Taq (Takara Bio. Inc., Kusatsu, Japan). Each experimental treatment was performed in four replicates. The relative expression level was calculated as 2-∆∆Ct and normalized using 18s as internal standard.

Statistical Analysis
Data are presented as means ± SEM. To determine the effect of the volumetric water content of the base material (drought and control) and the harvest time (day and night) on the expression level of DcaCIPK transcripts, a two-way ANOVA was performed using Statistical Analysis System 9.4 (SAS) and then followed by a Duncan's test (p < 0.05) to determine if there was a statistical difference between the mean expression levels of different isoforms under different treatments [64]. The time-of-day effect was evaluated by ANOVA one-way for drought and control situations followed by LSD post hoc analysis. p < 0.05 was considered significant.