Transcriptome Analyses of a Salt-Tolerant Cytokinin-Deficient Mutant Reveal Differential Regulation of Salt Stress Response by Cytokinin Deficiency

Soil destruction by abiotic environmental conditions, such as high salinity, has resulted in dramatic losses of arable land, giving rise to the need of studying mechanisms of plant adaptation to salt stress aimed at creating salt-tolerant plants. Recently, it has been reported that cytokinins (CKs) regulate plant environmental stress responses through two-component systems. A decrease in endogenous CK levels could enhance salt and drought stress tolerance. Here, we have investigated the global transcriptional change caused by a reduction in endogenous CK content under both normal and salt stress conditions. Ten-day-old Arabidopsis thaliana wild-type (WT) and CK-deficient ipt1,3,5,7 plants were transferred to agar plates containing either 0 mM (control) or 200 mM NaCl and maintained at normal growth conditions for 24 h. Our experimental design allowed us to compare transcriptome changes under four conditions: WT-200 mM vs. WT-0 mM, ipt1,3,5,7-0 mM vs. WT-0 mM, ipt1,3,5,7-200 mM vs. ipt1,3,5,7-0 mM and ipt1,3,5,7-200 mM vs. WT-200 mM NaCl. Our results indicated that the expression of more than 10% of all of the annotated Arabidopsis genes was altered by CK deficiency under either normal or salt stress conditions when compared to WT. We found that upregulated expression of many genes encoding either regulatory proteins, such as NAC, DREB and ZFHD transcription factors and the calcium sensor SOS3, or functional proteins, such as late embryogenesis-abundant proteins, xyloglucan endo-transglycosylases, glycosyltransferases, glycoside hydrolases, defensins and glyoxalase I family proteins, may contribute to improved salt tolerance of CK-deficient plants. We also demonstrated that the downregulation of photosynthesis-related genes and the upregulation of several NAC genes may cause the altered morphological phenotype of CK-deficient plants. This study highlights the impact of CK regulation on the well-known stress-responsive signaling pathways, which regulate plant adaptation to high salinity as well as other environmental stresses.


Introduction
Salt stress is one of the major abiotic stresses experienced by plants worldwide, affecting approximately 7% of the world's total land area [1,2]. Mild salt stress primarily affects plant development, agronomy traits and agricultural productivity, but extremely high salinity stress can lead to plant death. In addition, climate change and declining water quality are of great concern because they contribute to land degradation by causing high salinity levels in soil. Thus, salt stress has been considered an increasingly serious problem underscoring the importance of developing salt-tolerant plants, through the use of genetic engineering, which are capable of surviving under saline conditions [2][3][4][5]. Salt stress negatively impacts photosynthesis, energy production, lipid metabolism, nutrient acquisition, the integrity of cellular membranes and the activity of various enzymes, thereby leading to a number of destructive processes, such as water deficit, hyperosmotic stress, secondary oxidative stress, homeostasis disruption and ionic toxicity [6][7][8][9].
To cope with salinity stress, plants employs various mechanisms, at both the whole plant and cellular levels, which are controlled by a variety of genes and signaling pathways and are expressed and activated at different times during the life of a plant [4,6,7]. With the availability of genomic sequences from various plant species and recent advances in microarray technologies, genes associated with high salinity tolerance have been identified on a large scale at a genome-wide level [2,10,11]. Together with other omic technologies, such as proteomics and metabolomics, transcriptomics has contributed significantly to the elucidation of stress responses [12][13][14].
Cytokinins (CKs) are master regulators of plant growth and development, and were recently shown to control plant adaptation to salt stress [15][16][17][18][19]. The most common naturally occurring isoprenoid CKs are N 6 -isopentenyladenine (iP), trans-zeatin (tZ) and cis-zeatin (cZ). Experimental evidence has demonstrated that in many plant species, including Arabidopsis thaliana, tZ is more abundant and more active than cZ, whereas in some other species, such as maize (Zea mays), cZ is an abundant and bioactive CK [20]. In Arabidopsis, CK biosynthesis is controlled by nine adenosine phosphate-isopentenyltransferases (IPTs), which are divided into two classes of IPTs: the ATP/ADP IPT class (IPT1, IPT3, IPT4,  IPT5, IPT6, IPT7 and IPT8), which are responsible for biosynthesis of iP and tZ CKs, and the transfer RNA IPT class (IPT2 and IPT9), which are responsible for biosynthesis of cZ CKs [21]. Loss-of-function studies using various combinations of Arabidopsis ipt mutants indicated that disruption of IPT1, IPT3, IPT5 and IPT7 resulted in significant reductions in CK levels in Arabidopsis seedlings. The ipt1,3,5,7 quadruple mutant, at 18-dayold, was reported to contain less than 20% of iP riboside 59phosphates, iP riboside, tZ riboside 59-phosphates, tZ riboside, tZ-7-N-glucoside, tZ-9-N-glucoside and tZ-O-glucoside and less than 50% of the free-base iP and tZ in comparison with the WT [21]. Recently, analysis of the CK-deficient ipt1,3,5,7 mutant under salt stress conditions demonstrated that this mutant possesses a strong salt-tolerant phenotype, suggesting the importance of CKs in regulating salt tolerance.
To identify genes involved in salt stress tolerance and regulated by CKs, as well as to elucidate molecular mechanisms associated with CK-regulated salt tolerance, we have conducted a genomewide transcriptional analysis to compare the transcriptomes of the wild-type (WT) and the CK-deficient ipt1,3,5,7 mutant under both normal and salt stress conditions using whole genome Arabidopsis microarrays. Our results suggest that reduction of bioactive CK levels induced changes in gene expression of many regulatory and functional genes, including transcription factors (TFs), calcium sensors, high affinity K + /Na + co-transporter proteins, heat shock proteins, late embryogenesis-abundant (LEA) proteins, xyloglucan endo-transglycosylases (XTR), glycosyltransferases (GTs), glycoside hydrolases (GHs), defensins and glyoxalase I family proteins, leading to improved salt and drought stress tolerance of the ipt1,3,5,7 mutant.

Plant materials and salt stress treatment
WT Arabidopsis (Columbia ecotype) and CK-deficient ipt1,3,5,7 (Columbia ecotype) seeds [21] were sowed on germination medium (GM) agar plates supplemented with 3% sucrose and kept for 4 days at 4uC. Plates were then transferred to 22uC and incubated for 10 days (22uC, 16 h light/8 h dark cycle, 60 mmol m 22 s 21 photon flux density). For microarray analyses, the 10-day-old plants grown on GM media were transferred onto 0.56 MS plates without sucrose, containing either 0 mM (control) or 200 mM NaCl and maintained for a period of 24 h. Three independent experiments were performed for each condition. The samples were collected as three biological replicates (10 plants/ replicate), frozen in liquid nitrogen and stored at 280uC until used for RNA extraction.

RNA purification and microarray analyses
Total RNA was extracted with TRIZOL Reagent according to the supplier's instructions (Invitrogen, Carlsbad, CA, USA). The microarray analyses, hybridization, data analyses and data mining were carried out as previously described [22]. Briefly, cDNAs were synthesized using 500 ng of total RNA and labeled with one color (Cy3) using a Quick Amp labeling kit (Agilent Technologies), followed by fragmentation and hybridization to the Arabidopsis Oligo 44K DNA microarray (Ver. 4.0, Agilent Technologies). Three biological replicates were performed for each treatment, making a total of 12 hybridizations. All arrays were scanned with a microarray scanner (G2505B, Agilent Technologies) and analyzed using GeneSpring Ver. 11 (Agilent Technologies). The Student's ttest (p-value) was used as a parametric test and the Benjamini and Hochberg False Discovery Rate (q-value) procedure was used to control the certainty level. Genes with at least a 2-fold difference in their expression levels and a q-value,0.05 were considered to be differentially expressed. The raw microarray data and the detailed protocol were deposited in the Gene Expression Omnibus database (http://www.ncbi.nlm.nih.gov/geo/browse/?view = series) with GEO ID: GSE32087.

qRT-PCR
RNA samples from three biological replicates were used in the qRT-PCR validation of the microarray data. cDNA synthesis and qRT-PCR were performed according to previously described methods [23,24]. The primer pairs that were used in qRT-PCR reactions are listed in Table S1. UBQ10 was used as an internal control for expression analyses.

Determination of biological process categories of the upregulated and downregulated gene sets using gene ontology
The biological process categories of the up-and downregulated genes identified in four comparisons were determined as previously described [29,30]. The gene ontology (GO) terms assigned to each of the upregulated or downregulated genes were counted using the data set of Arabidopsis TAIR10 (ftp://ftp. arabidopsis.org/home/tair/Ontologies/Gene_Ontology/). The top 20 most abundant GO terms were subsequently used to classify the genes into biological process categories.

MapMan analysis
The list containing differentially expressed genes obtained from four comparisons was annotated and categorized into functional groups using MapMan version 3.5.1 (http://mapman.gabipd.org/ web/guest/mapman) according to the standard protocol [31,32].

Results and Discussion
Expression profiling of WT and CK-deficient ipt1,3,5,7 plants using microarrays under normal and salt stress conditions We have previously reported that the levels of free-base iP and tZ as well as their riboside and ribotide forms in 10-day-old Arabidopsis ipt1,3,5,7 seedlings were reduced to less than 25% of the WT levels [16]. Furthermore, using the system that was developed for salt-tolerance studies, we have shown that the 10-d-old ipt1,3,5,7 plants displayed a stronger salt-tolerant phenotype than WT [16,18]. Thus, to investigate the molecular changes brought about by CK deficiency and that contribute to enhanced salt stress tolerance, the same system was used for salt treatment. Ten-dayold WT and CK-deficient ipt1,3,5,7 plants were transferred onto plates containing 0 mM and 200 mM NaCl for 24 h. The collected samples were then subjected to microarray analyses using an Arabidopsis Oligo 44K DNA microarray (Ver. 4.0) for profiling the transcripts of WT and ipt1,3,5,7 plants in the absence and presence of 200 mM NaCl. The results of the microarray analyses are available in Table S2 or can be accessed through Gene Expression Omnibus (accession number: GSE32087).
This experimental design enabled us to investigate and compare the differentially expressed transcriptomes of WT and CKdeficient ipt1, 3 Tables S3, S4, S5, S6). Eight genes were selected to validate our microarray data by qRT-PCR. The results shown in Figure 2 supported the reliability of the microarray data.
All four differentially expressed gene sets were subsequently classified into biological process categories based on GO analysis. Figure 3 illustrated the distribution of the up-and downregulated gene sets in the 20 most abundant GO terms among which many abiotic stress-related categories, such as ''response to salt stress'', ''response to water deprivation'', ''response to ABA'', ''response to oxidative stress'' and ''response to cold'' were found. A significantly higher proportion of salt stress-responsive genes were detected in each of four upregulated gene sets ( Figure 3A) in comparison with the respective downregulated gene sets ( Figure 3B). Additionally, genes responsive to ABA and water stress were also found in higher abundance in all four upregulated gene sets than in the corresponding downregulated ones, which is consistent with the strong relationship observed among ABA, water stress and high salinity [33,34]. Moreover, we found that more stress-related genes, such as those classified into ''response to salt stress'', ''response to water deprivation'', ''response to ABA'' and ''response to cold'', are upregulated in comparison M-C/W-C or in comparison M-S/W-S than in comparison W-S/W-C ( Figure 3A); a phenomenon that may enable the CK-deficient plants to cope better with stresses as previously reported [16,35]. MapMan was also used to visualize the gene expression data to understand relations in terms of molecular functional categories among the four comparisons and find possible molecular mechanism(s) connecting salt stress response and CK deficiency at the genomic level as shown in Figure 4 and discussed in details in appropriate sections.
In addition, stress-responsive cis-elements have been reported for their essential roles in determining the stress-responsive expression patterns of genes, facilitating the stress-related functional classification [28][29][30]. Over the years, extensive promoter analyses have identified 12 stress-inducible cis-elements, which are important molecular switches involved in the transcriptional regulation of a dynamic network of gene activities controlling abiotic stress responses [25][26][27]. Therefore, to grasp the overall representation of stress-related genes among the upregulated genes identified in four comparisons we searched for the existence of these cis-elements in the 1-kb promoter region of each of upregulated genes identified in four comparisons. Data shown in Tables S3A, S4, S5, S6A indicate that a significant portion of upregulated gene sets are stress-responsive. It should be noticed that all 12 cis-elements used were identified as either drought-, cold-and/or ABA-inducible elements [25][26][27]. However, the results of the cis-element analysis are still useful because significant overlap has been reported for the gene sets upregulated by drought-, high-salinity-, cold-and ABA-inducible genes [33,34]. The percentages of stress-inducible genes might be higher if specific salt stress-inducible cis-motifs were also be included into the in silico analysis (Tables S3A, S4, S5, S6A). Unfortunately, the identity of salt stress-inducible cis-motifs is currently unknown.

Identification of salt stress-responsive genes in WT plants treated by the ''plate method''
Several groups have reported genome-wide expression analyses of salt stress response genes in Arabidopsis [33,36,37]. However, in these reports the methods used to treat the plants were different from ours. Plate-grown young seedlings were either transferred to a high saline solution (200-250 mM) and grown hydroponically for a short period of time (1-12 h) (solution method) [33,37], or soil-grown plants were irrigated with saline solution (150 mM) within 24 h (irrigation method) to generate salt-treated samples for investigation [36]. In another independent study, Krishnaswamy  2008) allowed Arabidopsis seeds to germinate and grow on 0.56 MS media containing 100 mM NaCl for 14 days (long-term treatment with mild concentration of NaCl). Here, we used the ''plate method'', which was developed for studying salt stress tolerance. This method, therefore, enabled us to identify molecular changes that were directly linked to a salt-tolerance phenotype. From the experiment comparing expression profiles of WT with and without 200 mM NaCl treatment for 24 h (comparison WS/ WC), we found that expression of 2268 and 3122 genes were increased and decreased, respectively, at least two-fold ( Figure 1B, Tables S3A and S3B). Although the treatment methods were different, the salt-inducible genes identified by our study significantly overlapped with those found by other groups [33,[36][37][38], indicating that all of these methods (soil-grown plants, plate-grown plants, treatment in solution or growth on agar plates containing salt) and different treatment time periods (shock short- term or long-term treatments) elicit similar responses at the transcriptional level. For instance, many high salinity-responsive genes known from the literature, such as several AP2-EREBP-type DREBs, RAP2, NACs, NCED2, NCED3, RD20, RD22, RD29B, PP2C and diverse LEA genes, were strongly induced (Tables S3A) [33,[36][37][38].
The differentially expressed genes could be classified into regulatory and functional categories. For the regulatory category, many upregulated genes were grouped into transcription factor (TF), protein degrading and protein modification groups. In the TF group, the AP2-EREBPs, such as DREB1A, DREB2A and DREB2B, zinc finger-type, MYB, WRKY and NAC TF encoding genes were present in high numbers. Heat shock TF and heat shock protein encoding genes were also found among the genes with increased transcript abundance. For the protein modification groups, we identified many salt-inducible genes encoding kinases, such as CIPKs and MAPKKKs, and PP2C proteins. The functional category contained many salt-regulated genes encoding LEA proteins, ABA metabolism-related proteins, osmoprotectant biosynthesis-related proteins, transporters and detoxification enzymes which were upregulated (Table S3A). These types of proteins are known to enable plants to adapt better to salt stress [6,39]. As for genes that were downregulated by salt stress, many photosynthesis-related genes were found (Table S3B). This finding is consistent with previously published results [33,[36][37][38].

CK deficiency alters expression of a subset of stressresponsive genes under normal conditions
Next, because CK-deficiency enhanced the tolerance of the CK-deficient ipt1,3,5,7 mutant to high salinity compared with the WT [16], we examined differences in gene expression between the ipt1,3,5,7 mutant and WT plants in the absence of NaCl in the M-C/W-C comparison. We identified a total of 3144 genes that responded to CK deficiency, representing more than 10% of the total number of genes in the Arabidopsis genome, of which 1625 genes were upregulated and 1519 genes were downregulated ( Figure 1B, Tables S4A and S4B). Both the down-and upregulated gene sets identified in comparisons W-S/W-C and M-C/W-C were subjected to Venn diagram analysis to identify overlapping genes between the two comparisons. As shown in Figure 5, many genes that were upregulated or downregulated in the untreated ipt1,3,5,7 mutant were also upregulated or downregulated, respectively, in the salt-treated WT, indicating significant overlap of gene expression profiles between the untreated ipt1,3,5,7 mutant and the salt-treated WT plants (comparisons W-S/W-C and M- C/W-C). These data might explain the better adaptation of the CK-deficient plants to salt stress. In addition, expression of the IPT1, 3, 5 and 7 genes was significantly decreased in the ipt1,3,5,7 mutant background, further demonstrating the reliability of our microarray analyses (Table S4B).
A number of genes encoding regulatory proteins, such as AP2-EREBP-, bZIP-, NAC-, WRKY-and ZFHD-type TFs, including the well-known DREB2A, ANAC021/NAC1, ANAC092/At-NAC2, ANAC055/AtNAC3, ANAC072/RD26 and ZFHD1 TFs, and protein kinases, including CIPKs and CDPKs, were upregulated in the ipt1,3,5,7 mutant (Table S4A). The ANAC092, ANAC055, ANAC072 and ZFHD1 genes were induced by ABA and salt stress, but DREB2A has been shown to be induced by salt stress and not by ABA [40][41][42][43], suggesting that both the ABA-dependent and ABA-independent TFs are regulated by CKs. It was reported that overexpression of ANAC021 and ANAC092 in Arabidopsis promoted lateral root development, which is in agreement with enhanced lateral root development observed in the ipt1,3,5,7 mutant [21,41,44]. On the other hand, Arabidopsis transgenic plants overexpressing the active form of DREB2A exhibited enhanced salt tolerance [45], whereas those overexpressing either ANAC055, ANAC072 or ZFHD1 enhanced drought tolerance [40,42]. However, there is no information regarding the saltresponsive phenotype of the ANAC055, ANAC072 or ZFHD1 overexpressing plants. In good accordance with these reports, the CK-deficient ipt1,3,5,7 mutant also exhibited improved drought tolerance [16].  Many genes encoding functional proteins, such as glyoxalase I proteins, XTRs, GTs, GHs and defensins (PDF1.2), were found among the upregulated genes in the ipt1,3,5,7 mutant in comparison with WT. As for glyoxalase I proteins, overexpression of Brassica juncea glyoxalase I and Oryza sativa glyoxalase II encoding genes, either alone or together, conferred improved salinity tolerance to tobacco (Nicotiana tabacum) [46]. In Arabidopsis, glyoxalase I proteins encoding genes were reported as potential downstream targets of drought and salt stress-responsive NAC TFs, such as ANAC019, ANAC055 and ANAC072 [40,47]. It was suggested that glyoxalase enzymes are involved in the glutathionebased detoxification of methylglyoxal, a toxic aldehyde formed primarily as a byproduct of carbohydrate and lipid metabolism, thereby enhancing salt stress tolerance [48,49]. XTR, GT and GH enzymes are known to play important roles in the modification of the xyloglucan cross-links that control the extensibility and strength of the plant cell wall [50][51][52]. Defensins, in addition to being implicated in mediating plant responses to pathogens, are also believed to play a role in salt stress tolerance because their transcript abundance is much higher in the Arabidopsis relative halophyte salt cress (Thellungiella halophila) than in Arabidopsis [53]. Moreover, induced expression of the defensin PDF1.2 gene was reported to result in enhanced root development, consistent with the better enhanced root development observed with the ipt1,3,5,7 mutant [54].
It has been established that CKs also affect plant disease resistance [55]. In the CK-deficient mutant we detected 35 and 10 disease resistance-related genes with increased and diminished expression levels, respectively, as compared with WT (Tables S4A  and S4B). Almost all of the up-and downregulated disease resistance-related genes encode CC-NBS-LRR or TIR class proteins, which mediate resistance to various plant pathogens [56][57][58]. Many NBS-LRR-encoding genes are upregulated in response to pathogen attack [56]. Therefore, it would be interesting to examine the relationship between the functions of CKs, CK signaling and NBS-LRR proteins in pathogen responses. Furthermore, among the genes downregulated by CK deficiency, MapMan analysis has identified many photosynthesis related genes ( Figure 4, Table S4B). Consistent with our data, genes related to photosynthesis were found to be upregulated in Arabidopsis plants treated with CKs [59]. All together, these results suggest that reduced photosynthesis may cause the retardation of shoot growth observed with CK-deficient plants which is referred to as CK deficiency syndrome [21,60]. The restraint of shoot growth has been regarded as an advantageous growth adjustment in response to environmental cues, including salt stress, and is known to be an important mechanism of plant survival [61][62][63].
Modulation of salt stress-responsive genes in the CKdeficient ipt1,3,5,7 mutant However, many overlapping genes exhibited differential expression levels in the degree of modulation of transcript abundance. Among six DREB-type TF encoding genes, expression of DREB1A, DREB1B, DREB1C, DREB1D and DREB2B was increased by 3.7-, 60.6-, 14-, 31.8-and 18.5-fold, respectively, in salt-treated WT seedlings, but by 7.1-, 122.4-, 22.1-, 89-and 24.6-fold in salt-treated ipt1,3,5,7 seedlings compared with respective untreated controls ( Table 1, comparisons M-S/M-C and W-S/W-C). Similarly, among 17 overlapping salt-inducible NAC genes, the majority of them (11 genes), including ANAC013, ANAC019, ANAC036, ANAC044, ANAC055, ANAC071, ANAC079, ANAC081, ANAC085, ANAC090 and ANAC102 were more highly induced in the ipt1,3,5,7 mutant than in the WT under salt stress. On the other hand, only two genes, ANAC061 and ANAC072, exhibited smaller inductions in comparison M-S/M-C than in comparison W-S/W-C. All of the overlapping upregulated ABA-responsive LEA protein encoding genes also showed higher increases in transcript abundance in salt-treated ipt1,3,5,7 than in salt-treated WT compared with respective untreated controls ( Table 1, comparisons M-S/M-C and W-S/W-C). It should be noted that CK deficiency does not have a significant effect on the expression of the DREB, NAC and LEA genes, except for ANAC055, ANAC072, ANAC079 and DREB2A (Table 1, comparison M-C/W-C), suggesting that transcript abundance of these genes was altered in salt-treated ipt1,3,5,7 plants in a salt-dependent rather than CKdependent manner.
CK deficiency affects the expression of known stressrelated genes in response to high salinity Because CKs mediate salt stress responses, it is also desirable to identify genes whose expression was altered by CK deficiency during salt stress. Therefore, comparison M-S/W-S was designed for this purpose ( Figure 1A). We identified a total of 3276 genes that were significantly (q,0.05) differentially expressed by at least two-fold under salt stress, among which 2354 genes were upregulated and 922 were downregulated ( Figure 1B These results might suggest that CKs, rather than high salinity stress, altered the expression profile of salt-treated ipt1,3,5,7 when compared with salt-treated WT in comparison M-S/W-S. As for regulatory genes, among the major stress-related TFs, all of the salt-inducible NAC and ZFHD genes, such as ANAC055, ANAC072 and ZFHD1, which confer stress tolerance when overexpressed (as previously discussed), were also induced in salt-treated ipt1,3,5,7 when compared with salt-treated WT. Moreover, more salt-inducible NAC genes were identified in comparison M-S/W-S than comparison M-C/W-C (Table S7). Additionally, we found that the DREB1B and DREB1D were induced in comparison M-S/W-S but not in comparison M-C/W-C (Table S7). Expression of DREB1B is induced by ABA, salt and cold stresses, while DREB1D is induced by ABA, salt and drought stresses [64,65]. Overexpression of DREB1B and DREB1D was shown to enhance salt tolerance of transgenic rice plants and drought tolerance of transgenic Arabidopsis plants, respectively [66,67], suggesting that combinatory effects of various types of TFs may contribute to improved salt and drought tolerance of the CKdeficient ipt1,3,5,7 plants. Furthermore, CK deficiency may also induce the SOS pathway in response to salt stress because the SOS3 gene, which encodes an EF-hand Ca 2+ -binding protein that functions as a calcium sensor for salt tolerance, was upregulated 2.36 fold in comparison M-S/W-S (Table S7). Recently, Yang et al. (2009) reported that overexpression of SOS3 resulted in enhanced salt tolerance of Arabidopsis transgenic plants [68]. In addition, we observed that the expression of HKT1;1, encoding the well-known high affinity K + /Na + co-transporter that works in coordination with SOS proteins to control Na + and K + homeostasis, was significantly increased in the ipt1,3,5,7 mutant under both normal and salt stress conditions (Figure 2). Consistent with our results, treatment of WT with exogenous CKs decreased transcript abundance of HKT1;1 [69]. The increased transcript abundance of HKT1;1 might also contribute to salt tolerance of CK-deficient plants because (i) expression of HKT1;1 was also upregulated significantly in roots of the salt-tolerant arr1,12 mutant, (ii) overexpression of HKT1;1, specifically in the mature root stele, enhanced salt tolerance in Arabidopsis transgenic plants and (iii) loss-of-function of HKT1;1 rendered mutant plants hypersensitive to salt stress [69][70][71].
Regarding functional proteins, many genes encoding glyoxalase I proteins, XTRs, GTs, GHs and defensins were also upregulated in comparison M-S/W-S, as found in comparison M-C/W-C. The functions of these proteins have been discussed previously. Interestingly, 12 LEA protein encoding genes were found to be upregulated in comparison M-S/W-S, whereas only two genes   (At1g54890 and XERO1/At3g50980) were upregulated in comparison M-C/W-C. Ten of these 12 genes were also induced by salt stress (Table S7). Several lines of evidence have demonstrated that transgenic plants overexpressing LEA encoding genes display enhanced salt-tolerant phenotypes [72,73], although their precise functional roles remain to be determined.

Effect of CK deficiency and/or salt stress on CK-related genes
Because this study investigated the effect of CK deficiency on the expression of downstream genes under normal and salt stress conditions, it was desirable to examine the transcription of CKrelated genes in detail. Therefore, we extracted the expression data for all of the genes that are known to be involved in CK metabolism (IPT, CKX, CYP735A and LOG genes), CK translocation (PUP genes) and CK signaling (AHK, AHP, ARR and APRR genes) [20,74,75]. The expression patterns of all of these genes were analyzed in all four comparisons and displayed in a heat map ( Figure 6). We observed that the number of downregulation events (65) was more than three times that of upregulation events (21) (Table S8). In comparison W-S/W-C, among 92 genes analyzed, 25 genes were downregulated, of which a large portion included CK metabolism-related genes, while only three genes were upregulated, demonstrating a suppression of CK metabolism, which is consistent with a reduction in CK content under salt stress that was recently reported (Table S8) [16]. We also found that expression of the majority of CKX gene family members was decreased in all four comparisons ( Figure 6, Table S8), which indicates that reduction of CK content caused by either salt stress or mutations in ipt genes diminishes expression of CKX genes. On the other hand, it has been reported that expression of CKX genes is induced by exogenous treatments of CKs [76,77]. Taken together, these results suggest that the expression of CKX genes is CK-dependent.
In addition, among genes related to CK signaling, five genes were downregulated (one AHP, three type-A ARR and one type-B ARR genes), whereas only one gene, AHP4, was upregulated in the WT by salt treatment, indicating that CK signaling might be repressed by salt stress (Table S8, comparison W-S/W-C). These data suggest that a decrease in CK metabolism by salt stress will subsequently lead to repression of CK signaling. This provides a mechanism enabling plant adaptation to high salinity conditions. Indeed, it has been reported that suppression of CK signaling by mutations in genes encoding CK receptor kinases or type-B ARRs enhanced salt tolerance [18,69]. It is worth mentioning that among 10 type-A ARR genes, more than half were downregulated in comparison M-C/W-C (five genes) and comparison M-S/W-S (six genes), but only one type-A ARR gene was downregulated in comparison M-S/M-C (Table S8), indicating that salt stressresponsive expression of type-A ARR genes is CK-dependent.
Among 18 PUP genes, which were reported to be involved in CK transport [20,78], nine, two and four genes were downregulated in comparisons W-S/W-C, M-C/W-C and M-S/M-C, respectively, while none of the PUP genes had a diminished expression level in comparison M-S/W-S. At the same time, we detected a total of one, two, one and one upregulated genes in comparisons W-S/W-C, M-C/W-C, M-S/M-C and M-S/W-S, Figure 6. Heat map representation for the regulation of CKrelated genes by CK deficiency under normal and salt stress conditions. Changes in expression levels are shown by color scales with saturation at five-fold changes. Blue and red color gradients indicate an increase or decrease in transcript abundance, respectively. doi:10.1371/journal.pone.0032124.g006 respectively ( Figure 6, Table S8). Therefore, the transcription of the PUP genes depends more on salt stress than on CK levels. Brenner et al. (2005) reported that among 15 PUP genes analyzed, only expression of the PUP4 gene was lower in WT plants treated with CKs, which further supports the CK-independent expression of the majority of the PUP gene family [59].

Conclusions
When compared to WT under both normal and salt stress conditions, our genome-wide microarray analyses have detected transcriptional changes of more than 10% of all annotated Arabidopsis genes in the CK-deficient ipt1,3,5,7 mutant. These genes encode proteins with a wide-range of functions, demonstrating the importance of CKs in the regulation of various biochemical, physiological and developmental processes. Downregulation of photosynthesis-related genes may cause shoot growth retardation, while upregulation of NAC TF encoding genes may play a role in enhanced lateral root development of the CKdeficient plants. The data gained through this study also suggest that CKs influence the established transcriptional networks, such as the DREB, NAC and ZFHD pathways, as well as ion signaling pathways, such as the SOS pathway. Ultimately, the impact of CKs on these regulatory networks results in optimized adaptation of sessile plants to adverse conditions. Overall, the results provided by this study may explain, at least in part, the phenotypic differences in plant growth and development between the WT and CK-deficient plants, as well as the enhanced tolerance to salt stress conferred by CK deficiency.

Supporting Information
Table S1 Primers used for qRT-PCR.