A Calcium/Calmodulin-regulated Member of the Receptor-like Kinase Family Confers Cold Tolerance in Plants*

Cold is a limiting environmental factor that adversely affects plant growth and productivity. Calcium/calmodulin-mediated signaling is believed to play a pivotal role in plant response to cold stress, but its exact role is not clearly understood. Here, we report that CRLK1, a novel calcium/calmodulin-regulated receptor-like kinase, is crucial for cold tolerance in plants. CRLK1 has two calmodulin-binding sites with different affinities as follows: one located at residues 369–390 with a Kd of 25 nm, and the other located at residues 28–112 with a Kd of 160 nm. Calcium/calmodulin stimulated the kinase activity, but the addition of chlorpromazine, a calmodulin antagonist, blocked its stimulation. CRLK1 is mainly localized in the plasma membrane, and its expression is stimulated by cold and hydrogen peroxide treatments. Under normal growth conditions, there is no noticeable phenotypic difference between wild-type and crlk1 knock-out mutant plants. However, as compared with wild-type plants, the crlk1 knock-out mutants exhibited an increased sensitivity to chilling and freezing temperatures. Northern analysis showed that the induction of cold-responsive genes, including CBF1, RD29A, COR15a, and KIN1 in crlk1 mutants, is delayed as compared with wild-type plants. These results indicate that CRLK1 is a positive regulator of cold tolerance in plants. Furthermore, our results suggest that CRLK1 plays a role in bridging calcium/calmodulin signaling and cold signaling.

Cold, a major environmental factor, affects plant growth and reduces productivity (1). It has been well documented that numerous cold-induced physiological and biochemical responses exist, including accumulation of reactive oxygen species, changes in membrane lipid composition, and the accumulation of compatible osmolytes (1,2). Cold acclimation also induces the expression of certain COR genes such as RD29A, COR15a, and KIN1, which function to stabilize membranes against freezing-induced injury (1). The CBF/DREB1-or CRT/ DRE-binding factors, a family of closely related AP2/ERF transcription factors, play a major role in controlling the COR gene expression (3), and CBFs themselves are activated by cold through ICE1 (4,5).
Calcium, a universal second messenger, acts as a mediator of stimulus-response coupling in the regulation of plant growth, development, and responses to environmental stimuli (6 -10). A variety of stress signals, such as cold, high salt, and hydrogen peroxide, triggers rapid increases in Ca 2ϩ levels in plant cells (11)(12)(13). Intracellular free Ca 2ϩ changes are recognized by Ca 2ϩ receptors such as calcium-dependent protein kinases (10), calcineurin-B-like proteins (14), and calmodulin (CaM). 4 CaM is a ubiquitous Ca 2ϩ -binding protein in animals and plants (15,16). The presence of CaM-dependent protein phosphorylation in plants has been reported (17)(18)(19). It has been suggested that Ca 2ϩ /CaM-mediated signaling via the phosphorylation cascade is crucial for regulating COR gene expression and cold acclimation (20 -22). However, little is known about the identity of the CaM-regulated protein kinase(s) involved in cold signal transduction.
In the Arabidopsis genome, there are at least 600 receptorlike kinases (RLKs) representing nearly 2.5% of the annotated protein-coding genes. The RLKs resemble animal receptor kinases in that they have the general structure of an extracellular domain, a single transmembrane sequence, and an intracellular kinase domain. The function of most of the plant RLKs is not well understood (23). It has been reported that CaM interacts with Brassica oleracea S locus receptor kinase (24) and Arabidopsis AtCaMRLK1 (25), but CaM has no effect on the activity of these kinases. Earlier, we characterized a CaM upregulated cytoplasmic receptor-like kinase, but its function is not yet known (18). Here, we report the isolation of a plantspecific, plasma membrane-anchored Ca 2ϩ /CaM-regulated RLK (CRLK1) from an Arabidopsis cold-stressed cDNA library. Characterization of the crlk1 null mutants and complementation lines suggests that CRLK1 confers freezing tolerance by acting as a positive regulator of COR gene expressions.
vermiculite under 14-h light/10-h dark at 22°C with light intensity of 220 mol quanta m Ϫ2 s Ϫ1 if not otherwise specified.
Petri Dish Chilling and Freezing Tolerance Assays-Chilling sensitivity was tested as described previously (4). After 2 days of stratification at 4°C, the seeds were germinated at 22°C for 7 days on MS solid medium with 1% sucrose. Chilling stress was imposed by incubating the seedlings at 6 Ϯ 1°C with 50 Ϯ 2 mol quanta m Ϫ2 s Ϫ1 light. To measure the chlorophyll content, 2-week-old plants at 20 -22°C under 220 mol quanta m Ϫ2 s Ϫ1 light were transferred to a growth chamber at 4 Ϯ 1°C at the same light intensity. Chlorophyll extraction was performed as described previously (26). Freezing tolerance was essentially assayed as described previously (27). Seeds were sterilized and scattered on Petri dishes containing Gamborg basal salts solidified with 0.9% agar. The plates were kept at 4°C for 2 days and moved to 22°C under continuous light. Ten days after germination, the plates with seedlings were transferred to a chamber at Ϫ1°C without light for 16 h. Then the chamber was set to cool by decreasing the temperature 1°C per h. The plates were removed at desired temperatures, thawed at 4°C for 12 h in the dark, and then returned to the original growth conditions. After 2 days, the surviving plants were counted. Three repeats for each line were tested for the freezing tolerance and used for statistical analysis.
Cloning of CRLK1-The cold-treated Arabidopsis seedling cDNA expression library was screened using 35 S-CaM (potato PCM1 and PCM6) as described previously (18). The positive cDNA clones were sequenced on both strands, and CRLK1 was one of the positive clones. The full length of CRLK1 was cloned with the partial cDNA as a probe.
CaM Binding Assay-CaM binding overlay assay was carried out as described previously (18) using 35 S-labeled potato CaM PCM1. The synthetic peptides were prepared in the Laboratory of Bioanalysis and Biotechnology, Washington State University. Gel mobility assay was carried out as described previously (18) using bovine CaM (Sigma). The purified recombinant M369-440 and M28 -228 were used for CaM-binding affinity assay as described previously (18).
Transient Transformation Assays with GFP Fusion Constructs-The full-length CRKL1 or M28 -440 was translationally fused at the C terminus with GFP in the BamHI site of psmGFP. Plasmid DNA was introduced into the onion epidermal cells by particle bombardment. The onion peels were examined using argon laser (488 nm) for green fluorescence. Plasmolysis of the onion epidermal cells was induced upon addition of 0.8 M mannitol.
Antibody Preparation, Plant Protein Isolation, and Western Analysis-The peptide corresponding to amino acids 385-400 of CRLK1 was conjugated with keyhole limpet hemocyanin. Antibody was raised in rabbits by GenMed (San Francisco, CA). The soluble and membrane proteins were fractionated as described previously (18). Plasma membrane fraction was prepared from microsomes using two-phase partitioning experiments (28). Specifically, the microsomes were separated in a buffer containing 6.2% (w/w) dextran T500 (GE Healthcare), 6.2% (w/w) polyethylene glycol (Sigma), 11% (w/v) sucrose, 5 mM potassium phosphate buffer, pH 7.8, and 1 and 5 mM KCl in the first and second phase system, respectively. Membrane proteins in the upper and lower phase were pelleted by ultracentrifugation. The pellets were resuspended in resuspension medium (10% (v/v) glycerol, 0.5 mM dithiothreitol, 1 mM MOPS-KOH, pH 7, and 5 M leupeptin), immediately frozen, and kept at Ϫ80°C until used. The purity of the plasma membrane fraction and other membrane fractions was tested by Western blotting using antibody against plasma membrane marker H ϩ -ATPase (29) and tonoplast marker V-ATPase (Agrisera). Protein content was determined by Bradford reagent (Bio-Rad). Western blotting was carried out following the procedure suggested by the manufacturer (Roche Applied Science).
Immunoprecipitations and Protein Phosphorylation Assays-Immunoprecipitation of CRLK1 from wild-type or crlk1 mutant plants was performed using the anti-CRLK1 antibody as described previously (6,18). Specifically, 1 g of rosette leaves was harvested, cut into ϳ5-mm 2 pieces, and soaked in ice-cold lysis buffer (10 mM HEPES-KOH, pH 7.4, 12.5% sucrose, 10 mM potassium acetate, 3 mM MgCl 2 , 1 mM CaCl 2 ) supplemented with 0.01% detergent Silwet-77 and 1% formaldehyde for 30 min at 4°C and subjected to two cycles of 5 min of vacuum treatment. The leaves were then washed two times for 20 min with 50 ml of lysis buffer supplemented with 0.3 M glycine at 4°C and stored at Ϫ80°C until further use. The total proteins or proteins from different fractions were prepared in lysis buffer supplemented with NaN 3 and 1ϫ Roche Applied Science complete protease inhibitors. Immunoprecipitation was carried out using anti-CRLK1 antibody. The immunoprecipitants were fractionated in SDS-PAGE and subjected to Western analysis. Calmodulin was detected using anti-calmodulin antibody. Phosphorylation of the recombinant proteins and phosphoamino acid analysis were carried out as described previously (30). One M purified CRLK1 and mutant proteins were used in the autophosphorylation and substrate phosphorylation assays. All the CPZ experiments were done in the presence of Ca 2ϩ /CaM, unless stated otherwise. The experiments were repeated at least three times.
Isolation of T-DNA Insertion Knock-out Plants-The crlk1 insertion mutant was identified from SalK_016240 by PCR using two gene-specific primer pairs, TGATGGGGTAAGCT-GCTTGCTA/CCCAAAAGAATTGAAAAACAAAATCA and a T-DNA left border primer (31). The amplified fragments were sequenced, and the T-DNA insertion site was determined. The mutant plants were further verified by Northern analysis using full-length cDNA as a probe and Western blotting analysis.
Genetic Complementation Analysis-Full length and deletions of the CRLK1 cDNA coding region were amplified by reverse transcription-PCR with gene-specific primers and fused with 35S promoter in pCAMBIA 1302 (Cambia, Canberra, Australia). These constructs were transformed into the crlk1 mutants by the floral dip method. Transgenic lines were selected on MS medium containing hygromycin. Homozygous transgenic lines were isolated using PCR. The expression levels of transgenes were studied using reverse transcription-PCR using the same primer pair (GCTGAGCTGGGATTT-GAGAG/GAGAGGCCAAAGTCAGCAAC). A representative plant with the comparable transgene expression level in each complementation group was selected for further studies. 35 S-labeled CaM screening, we isolated CRLK1 (At5g54590.2) from a cold-stressed Arabidopsis library. It encodes a 440-amino acid protein with a predicted molecular mass of ϳ49 kDa. Bioinformatic analysis indicated that CRLK1 contained all 11 subdomains of the catalytic domain of the kinase, a transmembrane domain but no extracellular domain, suggesting that it is a membrane-anchored protein kinase (supplemental Fig. S1). CRLK1 has high homology with another Arabidopsis gene (At5g15730; 50% identity) and other plant RLKs such as the rice gene Os08g04420 (74% similarity, 66.9% identity) and tobacco NPK15 (66.5% similarity, 55.9% identity).

CRLK1 Is a CaM-binding Protein-Using
To verify that CRLK1 is a CaMbinding protein, a bacterially expressed mutant construct (M28-440) lacking the putative transmembrane domain was tested in a CaM binding assay. As shown in Fig. 1A, CaM bound CRLK1 in the presence of CaCl 2 but not in the presence of EGTA, a Ca 2ϩ chelator, suggesting that CaM binding to CRLK1 is Ca 2ϩdependent. To map the CaM-binding site(s), several deletion mutants lacking different parts of the CRLK (Fig. 1A) were tested for CaM binding. These analyses showed that CRLK1 has two Ca 2ϩ -dependent CaM-binding sites, one located in 369 -390 amino acids, just downstream of the catalytic domain, and another in the N-terminal 28 -112 amino acids (Fig. 1A). The dissociation constant (K d ) of CaM for the Cand N-terminal CaM-binding domains was ϳ25 and 160 nM, respectively (Fig. 1B). Furthermore, a synthetic peptide corresponding to 369 -390 amino acids was used for gel mobility shift assay. This peptide bound CaM in the presence of CaCl 2 but not in the presence of EGTA (Fig. 1C). Similarly, a 20-mer peptide corresponding to the N-terminal 30 -49 amino acids was also able to bind CaM in a Ca 2ϩ -dependent manner (supplemental Fig. S2). However, about four times peptide:CaM molar ratio were needed for CaM to fully form the complex with the peptide, suggesting the low CaM affinity to this peptide. Furthermore, co-immunoprecipitation was carried out to pull down the proteins prepared from cold-treated wild-type plants. The interaction was fixed using a prior in vivo cross-linking strategy (6). CaM was detected in the immunoprecipitant by the anti-CaM antibody. However, there is no CaM detected in the preimmune precipitant. These results support that CaM interacts with CRLK1 in vivo (Fig. 1D).
Ca 2ϩ /CaM Regulates the Kinase Activity of CRLK1-To investigate the kinase activity of CRLK1, the recombinant deletion mutants were purified to homogeneity. Three proteins, M28 -440, M28 -392, and M28 -369, were shown to autophosphorylate in the presence of 10 mM Mg 2ϩ but not in the pres- . D, CaM is co-immunoprecipitated with CRLK1 in planta. The total proteins or proteins from different fractions were prepared from wild-type plants treated at 4°C for 4 h. Immunoprecipitation was carried out using anti-CRLK1 antibody. The immunoprecipitants were fractionated in SDS-PAGE and subjected to Western analysis. Calmodulin was detected using anti-calmodulin antibody. MARCH 5, 2010 • VOLUME 285 • NUMBER 10 ence of Ca 2ϩ alone (data not shown). To further study the role of Ca 2ϩ /CaM in regulating the kinase activity, all three proteins were subjected to autophosphorylation in the presence of CaM. The kinase activity of both M28 -440 and M28 -392 was enhanced with the increasing concentrations of bovine CaM. Fig. 2A shows that Ca 2ϩ /CaM stimulates the kinase activity of M28 -392 and calcium alone, or CaM alone did not show any measurable effect on the kinase activity of M28 -392 (supplemental Fig. S2B). Increasing concentrations of CaM antagonist, CPZ, abolished the Ca 2ϩ /CaM-stimulated kinase activity ( Fig.  2A). CPZ alone did not show obvious effect on kinase activity of M28 -292 (supplemental Fig. S2B). On the other hand, Ca 2ϩ / CaM showed no significant effect on kinase activity of M28 -369, which lacks the C-terminal CaM-binding site, under both low and high concentrations of CaM (Fig. 2, A and B). We further studied the effects of two plant CaM isoforms, PCM1 and PCM6, on the kinase activity. PCM1 enhanced the kinase activity of CRLK1 like bovine CaM; however, PCM6 did not show such a stimulation (Fig. 2C), suggesting that the kinase activity of CRLK1 is regulated by a specific CaM isoform(s). When M28 -392 was used to phosphorylate casein, in the presence of increasing concentrations of Ca 2ϩ /CaM, there was an increase in substrate phosphorylation. Addition of CPZ showed inhibition of Ca 2ϩ /CaM-stimulated activity. However, M28 -389 did not show the significant substrate phosphorylation changes by increasing CaM concentration (Fig. 2, D and E). These results suggest that the CaM-binding site (residues 369 -390) is responsible for the Ca 2ϩ /CaM stimulation of kinase activity, and the CaM-binding site in the N-terminal region has no pronounced effect on kinase activity. Phosphoamino acid analysis revealed that CRLK1 autophosphorylated at both serine and threonine residues (Fig. 2F), suggesting that CRLK1 is a serine/ threonine kinase.

CRLK1 Is Primarily a Plasma
Membrane-associated Protein-The existence of a transmembrane domain in the CRLK1 N terminus indicates it is a membrane protein.
To verify this, full-length CRLK1: GFP fusion and M28 -440:GFP fusion, which lacked the putative transmembrane domain, were transiently expressed in onion inner epidermal cells. The CRLK1:GFP predominantly localized on the periphery of the cell (Fig. 3A, upper  panel). However, M28 -440:GFP was detected throughout the cytoplasm and nucleus (Fig. 3B, upper  panel). To further differentiate the location of CRLK1, the cells were plasmolyzed using 0.8 M mannitol. The green fluorescence for CRLK1: GFP was not in the cell wall but was mainly found in the periphery of the protoplast (Fig. 3A, lower panel). In addition, microsome preparation from wild-type (WT) plants was subjected to two-phase separation (28). The Western blot analysis using anti-CRLK1 antibody revealed that CRLK1 is located in the microsomes but not in the soluble fraction (Fig. 3C), confirming that CRLK1 is a membrane-associated protein. Furthermore, after the two-phase separation, CRLK1 was found predominantly in the plasma membrane-enriched upper fraction (Fig. 3C), a distribution pattern similar to that of a plasma membrane marker, proton-ATPase AHA2 (29,32,33). No CRLK1 enrichment was observed in the lower fraction where non-plasma membrane proteins such as tonoplast V-ATPase were enriched (Fig. 3C). Thus CRLK1 is a membrane protein predominantly associated with the plasma membrane.
CRLK1 Is a Positive Regulator of Cold Tolerance-The expression pattern of CRLK1 was further investigated by Northern analysis. One band around 2 kb showed a similar expression pattern in all tested tissues except in siliques (Fig. 3D). However, CRLK1 protein was differentially expressed in these tissues, as revealed by Western analysis (Fig. 3E). Highest expression levels of the CRLK1 protein were detected in 7-day-old seedlings. The CRLK1 protein was also detected in roots, leaves, flowers, and siliques. Because CRLK1 was isolated from a cold-treated cDNA library, the effect of low temperature on CRLK1 expression was further studied. Although there was no major change in mRNA level (Fig. 3F), the protein level of CRLK1 rapidly increased after cold treatment (Fig. 3G). After 30 min of cold treatment, the CRLK1 protein level began to increase, reached a maximum level at 3 h, then started to decrease after 6 h. Because cold stress induces the accumulation of H 2 O 2 , we examined its effect on the expression of CRLK1. H 2 O 2 treatment resulted in an increase in CRLK1 level (Fig. 3H), suggesting that it is involved in cold-related oxidative stress signal transduction pathway.

Calcium/CaM/CRLK1 Regulates Plant Cold Tolerance
To gain insight into the functional significance of CRLK1, we isolated a crlk1 T-DNA insertion mutant. Northern analysis indicated that an ϳ2-kb band corresponding to CRLK1 size was present in WT and heterozygous plants but not in homozygous plants. After a longer exposure, there was one weak band with a smaller size that appeared in homozygous plants (Fig. 4A). However, no CRLK1 proteins were detected in the crlk1 plants, whereas CRLK1 proteins were well detected in WT and heterozygous plants (Fig. 4B). Together the results indicate that crlk1 is a recessive null mutant at the protein level. The homozygous crlk1 and WT plants had no obvious phenotypic difference at 22°C (Fig. 4, C and D). Plant response to chilling was tested at 6°C, and chilling injury became apparent in the mutant plants after 3 weeks of cold treatment (supplemental Fig. S3D). Prolonged cold treatment up to 7 weeks produced significant phenotypic differences between the crlk1 and WT plants (Fig. 4, E and F). These differences included inhibited root growth, decreased number of lateral roots, and decreased shoot growth. In addition, the mutant leaves also showed yel-lowing, a sign of early senescence and the severity of chilling damage (Fig. 4, E and F). The F1 heterozygous plants obtained by backcrossing with WT plants showed similar chilling tolerance as WT (Fig. 4G). About 25% of the F2 population was sensitive to chilling stress, indicating that the crlk1 mutant has a single recessive mutation in crlk1 gene locus. The total chlorophyll content was tested to study the rate of senescence (Fig. 4H) (Fig. 4I). At Ϫ14°C, about 50% of WT plants were still alive, whereas only 20% of the crlk1 plants survived (Fig. 4, I and J), which was comparable with hos9 and hos10 plants (26 and 28% survival rate, respectively) under the same condition. Most of the surviving crlk1 mutant seedlings also appeared yellow in color as compared with the WT. Without cold acclimation, both WT and crlk1 seedlings showed a similar LT 50 of ϳ5.8°C (data not shown). This suggests that CRLK1 is required for cold acclimation.
We further prepared constructs of 35S::CRLK1 and ectopically expressed CRLK1 in crlk1 mutants (supplemental Fig.  S3A). Four independently complemented plants (crlk1/CRLK1) showed a higher level of both CRLK1 mRNA and protein as compared with WT plants. There was no obvious change in CRLK1 expression in the control plants transformed with the empty vector. Under normal conditions, no significant difference in the phenotypes of the complemented lines and crlk1 plants was observed. However, after exposing to long term low temperature (6°C), crlk1/CRLK1 plants rescued the cold sensitivity of crlk1 mutants and showed growth similar to WT plants. Furthermore, after cold acclimation, the freezing tolerance of crlk1/CRLK1 was enhanced as compared with crlk1 plants (Fig. 4, I and K). The LT 50 value for crlk1/CRLK1 plants is around Ϫ14.3°C, which was comparable with WT plants, sug- The full length of CRLK1 was used as a probe. 18 S RNA shows equal loading of total RNA. E, differential expression of CRLK1 protein in tissues by Western blotting using the anti-CRLK1 antibody. The microsomal fraction (5 g) was used for Western analysis. F and G, expression of CRLK1 mRNA and protein treated by cold stress. H, expression of CRLK1 protein is stimulated by H 2 O 2 . Three-week-old plants were subjected to cold treatment (4°C) or sprayed with 10 mM H 2 O 2 and harvested after various periods as indicated. The total protein (20 g) was used for Western analysis. The Coomassie-stained ribulose-bisphosphate carboxylase/oxygenase (Rubisco) bands show equal loading of protein samples. MARCH 5, 2010 • VOLUME 285 • NUMBER 10 gesting that the level of CRLK1 in plants is correlated to freezing tolerance.

Calcium/CaM/CRLK1 Regulates Plant Cold Tolerance
To understand the functional significance of CaM binding to CRLK1 in planta, we introduced the truncated CRLK1 (crlk1/ ⌬C2) lacking the C-terminal CaM-binding site into crlk1 plants. The crlk1/⌬C1, a truncated CRLK1 missing the C terminus beyond the CaM-binding region, was used as a control. Furthermore, crlk1/⌬T, a truncated CRLK1 without the N-terminal transmembrane domain, was introduced into crlk1 plants to study the function of membrane localization (supplemental Fig. S3A). Complementation plants carrying WT and mutated versions of CRLK1 with a comparable expression level of transgenes (supplemental Fig. S3B) were used for further studies. Without cold acclimation, no difference in freezing tolerance between the full-length and the deletion mutant lines was observed. However, after cold acclimation, crlk1/⌬C2 and crlk1/ ⌬T plants showed a higher sensitivity to freezing as compared with crlk1 plants (Fig. 4, I and K, and supplemental Fig. S3C). The LT 50 value for these lines was comparable with crlk1 plants, indicating that the C-terminal CaM-binding site and N-terminal transmembrane domain are crucial for CRLK1 function in freezing tolerance.
CRLK1 Regulates COR Gene Expression-To investigate whether CRLK1 regulates cold acclimation by the CBF-dependent pathway, we studied the gene expression of CBF and COR genes RD29A, KIN1, and COR15a in the crlk1 null mutants and CRLK1 transgenic lines. The CBF family is the key cold-responsive genes controlling the expression of downstream COR genes (3). We used the full length of CBF1 cDNA as probe, which may also hybridize with CBF2 and CBF3 because CBF1, -2, and -3 share high homology in nucleotide sequences. Northern analysis revealed differences in the expression kinetics of CBF1 between WT and crlk1 plants in response to cold treatment (Fig.  5A). Significant cold-induced CBF1 expression was observed in WT after 0.25 h of cold treatment. However, the induction of CBF1 in crlk1 plants appeared at 1 h after cold treatment. We also studied the expression of COR genes RD29A, KIN1, and COR15a in the crlk1 mutants and CRLK1 complemented lines. In comparison with wild-type plants, the cold induction of RD29A, KIN1, and COR15a was delayed by more than 4 h in the crlk1 mutants, and the induction levels were also decreased for KIN1 and COR15a (Fig. 5B). In the complementation line with the full-length CRLK1(crlk1/CRLK1), cold induction of COR genes was also observed. Even under normal conditions, expression of COR genes was detected in these plants. Furthermore, the complementation line crlk1/⌬C2 carrying a truncated CRLK1 without the C-terminal CaM-binding FIGURE 4. crlk1 knock-out mutants are sensitive to low temperature. A, expression of CRLK1 mRNA in crlk1 mutants. The total RNAs (10 g) from wild-type (ϩ/ϩ), crlk1 heterozygous (ϩ/Ϫ), and homozygous (Ϫ/Ϫ) seedlings were subjected to Northern analysis using the full-length CRLK1 as a probe. 18 S RNA shows equal loading of total RNA. B, expression of CRLK1 protein was abolished in homozygous crlk1 mutants. The total protein (20 g) isolated from 3-week-old plants were used for Western analysis. Ribulose-bisphosphate carboxylase/oxygenase (Rubisco). C and D, crlk1 mutants had no obvious phenotypic difference as compared with wild-type plants at 22°C. E-G, crlk1 homozygous plants (Ϫ/Ϫ) were more sensitive to chilling. The photos were taken after 7 weeks at 6°C. Bar, 10 mm. H, chlorophyll content in WT and mutant plants. I, freezing survival rates following cold acclimation. The full-length and truncated forms of CRLK1 cDNA were inserted downstream of cauliflower mosaic virus 35S of pCAMBIA 1302. ⌬C1 lacks the C-terminal 48 amino acids beyond the CaMbinding region; ⌬C2 lacks the C-terminal 71 amino acids including the CaM-binding domain; ⌬T lacks the N-terminal transmembrane domain. The average survival rates from three repeats of WT and complementation lines were used. J, crlk1 mutants are less tolerant to freezing than WT plants. Ten-day-old seedlings grown at 22°C were incubated for 4 days at 4°C before freezing treatment at Ϫ14°C. The picture was taken 12 days after the freezing treatment. K, freezing tolerance test showing that complementation with truncated CRLK1 lacking the C-terminal CaM-binding site (crlk1/⌬C2) cannot rescue the freezing sensitivity.
site showed delayed expression of COR genes similar to crlk1 plants. This is in agreement with the results of the freezing tolerance tests. These results suggest that CRLK1 acts as a positive regulator of CBF1, RD29A, KIN1, and COR15.

DISCUSSION
Our results document that CRLK1 is a calcium/CaM-regulated RLK anchored in plasma membrane. Calcium/CaM binding to CRLK1 regulates its kinase activity and plays a critical role in cold acclimation in planta. It has been well documented that the cold-induced changes in intracellular Ca 2ϩ level is critical for plants to fully acclimate and for maximal cold induction of some COR genes (11). It has been suggested that calcium/ CaM-mediated signaling is involved in the cold induction of COR genes because CaM genes are themselves induced by low temperature (34), and CaM inhibitors have been shown to inhibit cold-induced COR expression (20). Our results support that CaM acts as a positive regulator during cold acclimation. However, overexpression of CaM3 in Arabidopsis resulted in the inhibition of the COR expression, suggesting that this specific CaM gene may function as a negative regulator (21). There are 11 CaM genes in Arabidopsis, some of which show spatial and temporal regulation in the cell (15). Hence, it cannot be generalized that all CaMs show similar response to cold.
There are two CaM-binding sites in CRLK1. Although no significant role was detected for CaM binding to the N-terminal site on protein phosphorylation, we cannot exclude other roles for this binding. The mammalian epidermal growth factor receptor is a member of the CaMbinding receptor tyrosine kinase family with a CaM-binding site near the transmembrane domain. CaM regulates intracellular trafficking of epidermal growth factor receptor and mitogen-activated protein kinase (MAPK) signaling (35). It will be interesting to further dissect the function of the N-terminal CaM binding.
CRLK1 is localized in the plasma membrane, which is required for its function in planta because the truncated form of CRLK1 lacking the transmembrane domain did not rescue the phenotype of crlk1 null mutants. In the Arabidopsis genome, out of 600 RLK genes, more than 200 RLKs have no receptor configuration, and very few of them have known functions (36). Tomato Pto protein kinase, which confers resistance to the pathogen Pseudomonas syringae, is a cytoplasmic RLK without an extracellular domain and transmembrane domain. Pto is activated by avrPto, a small peptide secreted by P. syringae (37). Thus, even though CRLK1 has no receptor configuration, it could function directly in transducing an extracellular signal into cells in a manner similar to Pto, as well as playing a role in transducing the Ca 2ϩ /CaM-mediated signal triggered by stress. Alternatively, CRLK1 may form a dimer or oligomer with an RLK(s) with receptor configuration to recognize an extracellular ligand(s) or interact with other membrane protein(s) similar to TAK1. TAK1, a thylakoid membrane protein kinase without a receptor configuration, is required for light harvesting by phosphorylating the light-harvesting complex proteins in the thylakoid membrane (38). Isolation of CRLK1-interacting proteins/substrates will help in further dissecting this signal pathway. Accumulating evidence indicates that protein phosphorylation is involved in the pathway connecting the cold-triggered calcium changes and cold acclimation (22,39). However, the protein kinase(s) responsible for inducing cold-regulated genes and activating freezing tolerance is elusive. Our results suggest that calcium/CaM/CRLK1-regulated phosphorylation can be an important signal node in the cold signal transduction pathway.
Identification of CRLK1 as a Ca 2ϩ /CaM-regulated protein kinase lays the groundwork for improving plant stress tolerance. In recent years, we have been able to define the structure/ function relationships of Ca 2ϩ /CaM-binding proteins by manipulating their Ca 2ϩ and/or CaM-binding sites (40). For example, altered Ca 2ϩ /CaM binding to DWF1 documented the possibility of producing size-engineered plants (6), although altered Ca 2ϩ and Ca 2ϩ /CaM binding to chimeric calcium/calmodulin-dependent protein kinase produces spon- taneous nodulation (7). Recently, we characterized the CaMbinding transcription factor AtSR1, which doses the salicylic acid level and regulates plant immunity (41). Redesigning CRLK1 orthologs in crop plants by manipulating their CaMbinding sites will help to enhance their ability to respond to the stress signal and to engineer stress-tolerant crops.