Ion Binding Properties of the Dehydrin ERD14 Are Dependent upon Phosphorylation*

The ERD14 protein (early response to dehydration) is a member of the dehydrin family of proteins which accumulate in response to dehydration-related environmental stresses. Here we show the Arabidopsis dehydrin, ERD14, possesses ion binding properties. ERD14 is an in vitro substrate of casein kinase II; the phosphorylation resulting both in a shift in apparent molecular mass on SDS-PAGE gels and increased calcium binding activity. The phosphorylated protein bound significantly more calcium than the nonphosphorylated protein, with a dissociation constant of 120 μm and 2.86 mol of calcium bound per mol of protein. ERD14 is phosphorylated by extracts of cold-treated tissues, suggesting that the phosphorylation status of this protein might be modulated by cold-regulated kinases or phosphatases. Calcium binding properties of ERD14 purified from Arabidopsis extracts were comparable with phosphorylated Escherichia coli-expressed ERD14. Approximately 2 mol of phosphate were incorporated per mol of ERD14, indicating a minimum of two phosphorylation sites. Western blot analyses confirmed that threonine and serine are possible phosphorylation sites on ERD14. Utilizing matrix assisted laser desorption ionization-time of flight/mass spectrometry we identified five phosphorylated peptides that were present in both in vivo and in vitro phosphorylated ERD14. Our results suggest that the polyserine (S) domain is most likely the site of phosphorylation in ERD14 responsible for the activation of calcium binding.

Environmental stresses such as cold and drought have significant impact on plant growth and development; hence, agricultural productivity. Plants have evolved a wide variety of molecular responses to enable them to survive severe abiotic stresses (1,2). Among these responses is the alteration of expression of a family of genes that encodes dehydrins, a subfamily of the group II late embryogenesis abundant mRNAs (LEAs). Levels of dehydrins are increased in response to low temperature, drought, or osmotic stress or are abscissic acidinduced (3)(4)(5)(6)(7)(8). In Arabidopsis, dehydrins can be characterized as the acidic dehydrins, pI 4.6 -6.4 (COR47, ERD10, ERD14, 1 NP_195554, NP_195624, and X91920), which are generally highly enriched in glutamic acid, and the neutral/basic dehydrins, pI 7.6 -9.8 (RAB18, XERO1, and XERO2) which are enriched in glycine. Dehydrin proteins remain soluble after boiling, are extremely hydrophilic, and can be defined by the presence of at least one lysine-rich consensus sequence, the K domain (EKKGIMDKIKEKLPG), which is similar to class A2 amphipathic ␣-helical domains found in lipid-binding proteins (7,9). Some dehydrins meet the criteria for hydrophillins (10). Additionally, many dehydrins contain the S-domain, a Ser-rich sequence (3,(11)(12)(13). Although their biochemical and physiological roles are still unclear, it has been suggested that dehydrins may play a role in stabilizing proteins or membrane structures under environmental stresses through interactions with an amphipathic ␣-helix (9).
Calcium levels control a variety of plant developmental and signal transduction processes (14,15), and in particular, calcium signaling is requisite for plant responses to environmental signals (16 -19). A transient change in cytosolic calcium levels is one of the initial responses of some plants to adverse environmental conditions such as low temperature, drought, and salinity (16, 19 -23). Several studies show that the ability of plants to withstand calcium level changes is essential for them to survive different abiotic stresses (16,24,25). Typically, cytosolic free calcium is maintained at submicromolecular levels (ϳ200 nM) by homeostatic mechanisms involving a variety of calcium channels, pumps, and secondary transporters in a variety of cell organelles such as vacuole, endoplasmic reticulum, chloroplasts and mitochondria, and cell wall (14,15,26). Intracellular calcium homeostasis is also maintained in plant cells by a variety of calcium-binding proteins, such as calsequestrin (27,28), calnexin (29), and calreticulin (30). Calcium levels in the cytosol increases by several orders of magnitude upon signaling; however, sustained elevation of calcium can be toxic (31). It has been suggested that cold-induced [Ca 2ϩ ] cyt increases are caused by calcium leakage across membranes initiated at the plasma membrane and perpetuated by calcium release mainly from vacuole and endoplasmic reticulum (16,17,32).
The connection between stress-induced increases in cytosolic calcium and the accumulation of dehydrins has suggested to us that a potential physiological role for dehydrins in cold stress management could be ion binding. It was recently determined by our laboratory that a celery protein, similar in sequence to dehydrins (VCaB45), binds calcium (33), and it has been reported that another dehydrin-like protein found in castor bean has metal binding capability (34). The present study is a characterization of a known Arabidopsis dehydrin, ERD14, its ion binding activity, and its regulation by phosphorylation.

EXPERIMENTAL PROCEDURES
Plant Materials and Extractions-Arabidopsis thaliana (Columbia ecotype) seedlings were grown with 16/8 h light/dark period at 20°C. Low temperature treatment was performed on 3-4-week-old plants for 2 days 4.7°C. Total protein extracts of Arabidopsis plants were obtained from cold-treated and control tissues (33). In some cases low density membranes were isolated from cold-treated A. thaliana in the presence of protease and phosphatase inhibitors as described in Heyen et al. (33). Triton X-100 (0.2% w/w) was used to release the contents of the membrane vesicles (33). Supernatants derived from the permeabilized membranes were heat-treated in an 85°C water bath for 20 min and quickly cooled in an ethanol bath (Ϫ50°C) for 5 min. The soluble fraction was then obtained after ultracentrifugation for 1 h at 100,000 ϫ g at 4°C. Anion exchange chromatography was performed on a Mono Q column (Amersham Biosciences) in 20 mM Tris-HCl, pH 8.2 at 4°C.
ERD14 Cloning and Expression-A -PRL2 expression library (obtained from Ohio State Plant Resource Center) was subcloned into the pHEP2 vector (50) using the PstI and XbaI sites. Bacterial colonies were screened on replicate lifts after induction with isopropyl-1-thio-␤-Dgalactopyranoside as reported previously (50), with the antibody raised against the celery dehydrin-like protein, VCaB45 (33), using a secondary antibody of anti-mouse/alkaline phosphatase. Blots were developed with the colorimetric substrate nitro blue tetrazolium/5-bromo-4chloro-3-indolyl phosphate. The fusion protein was expected to be ϳ5 kDa greater than ERD14 due to the inclusion of v-Ras sequences (50). However, the induced ERD14 was present as two bands. A minor band (of the expected size for the fusion protein) and an ϳ5-kDa smaller band (major band) were obtained. The major band was purified and sequenced by Edman degradation from the amino terminus. This revealed that the v-Ras portion of the fusion was missing, and additionally, the two amino-terminal amino acids (methionine-alanine) of ERD14 had been removed by proteolysis. The presence of the predicted aminoterminal tryptic peptide (lacking Met-Ala) was also confirmed by matrix assisted laser desorption ionization-time of flight/mass spectrometry (MALDI-TOF/MS). This product, ERD14 lacking the amino-terminal two residues, was used for all the studies described here.
Purification of ERD14 -Induced cells (1 mM isopropyl-1-thio-␤-Dgalactopyranoside) were harvested and resuspended in 2ϫ homogenizing buffer in the presence of protease inhibitors (33). Samples were sonicated for 5 min and heat-treated as described before. Lysates were then centrifuged at 4,700 ϫ g for 15 min at 4°C. Supernatants were treated with 0.2% protamine sulfate to precipitate DNA and ultracentrifuged for 1 h at 100,000 ϫ g at 4°C. The high speed supernatant was fractionated by anion exchange chromatography on a 40-ml packed bed of diethyl aminoethyl (DEAE)-Sepharose (Amersham Biosciences) (1 ml/min flow rate). ERD14 fractions were further concentrated on a Mono Q column (Amersham Biosciences).
Western Blotting-Antibodies raised against phosphoserine and phosphothreonine were obtained from Zymed Laboratories Inc. and Cell Signaling Technology, respectively. Anti-dehydrin antiserum was kindly supplied by Dr. Timothy J. Close (3). Samples were separated by 10% SDS-PAGE and electroblotted onto nitrocellulose membranes (Protran, Mid-West Scientific). For anti-dehydrin antibody, membranes were blocked with 5% nonfat dry milk in phosphate-buffered saline. When anti-phosphoserine and anti-phosphothreonine were used, membranes were blocked with 5% Tween 20 and 5% nonfat dry milk, 0.1% Tween 20, respectively, in Tris-buffered saline. Anti-rabbit and antimouse immunoglobulin G (goat) conjugated to horseradish peroxidase (Sigma) were used as secondary antibodies. Antibody detection procedures were with SuperSignal West Pico reagent (Pierce).
ERD14 in Vitro Phosphorylation and Dephosphorylation-Purified E. coli-expressed ERD14 (ϳ7.5 g) was incubated with 10 units of casein kinase II in 1ϫ kinase buffer (BIOMOL Research Laboratories, Inc.) and 1 mM ATP at 30°C in a total of 10 l. In some cases, 1 mM Staurosporine (BIOMOL Research Laboratories, Inc.) was added after 30 min to inhibit protein kinase activity. Dephosphorylation was performed with shrimp alkaline phosphatase (Roche Applied Science) (33). Reactions were stopped by the addition of SDS-PAGE sample buffer and were boiled for 5 min. For the 32 P incorporation experiments, purified ERD14 was phosphorylated in the presence of 0.8 mM ATP (0.17 Ci of [␥-32 P]ATP). 32 P incorporation was visualized with a Phos-phorImager (STORM 840, Molecular Dynamics). In some cases, reactions were terminated with 25% trichloroacetic acid, and precipitates were collected on GFA glass fiber filters (Mid-West Scientific) and washed with 5% trichloroacetic acid and then 100% acetone. In such cases, the incorporated 32 P was determined by liquid scintillation counting.
Phosphorylation of ERD14 by A. thaliana Protein Extract-Extracts from whole A. thaliana plants treated for 2 days at 4 or 20°C were incubated with or without the bacterial-purified ERD14 (final concentration of 95 ng/l), 0.1 mM ATP containing [␥-32 P]ATP (1 Ci) in 50 mM Hepes, pH 7.2, 0.5 mM CaCl 2 , 10 mM MgCl 2 , and 10 mM NaF at 30°C. Reactions were stopped by adding SDS-PAGE loading buffer and boiling for 5 min.
MALDI-TOF Mass Spectrometry-Protein samples (purified from E. coli) were mixed with an equal volume of 8 M urea and incubated for 30 min at 55°. To reduce the urea concentration to 1 M, 10 mM ammonium bicarbonate buffer, pH 8.5, was added. Protein was then digested overnight with shaking at 37°C with 1 g of freshly prepared sequencing grade trypsin (Promega, V511A) dissolved in 50 mM acetic acid. A C18 Zip-Tip (Millipore) was used to remove urea and desalt the digestion solution. ERD14 purified from Arabidopsis was separated on 10% SDS-PAGE gels and Coomassie-stained. Protein bands corresponding to ERD14 were excised and manually processed. Samples were destained with 50% acetonitrile in water for 15 min at 20°C, 50% acetonitrile in 50 mM ammonium bicarbonate for 15 min at 20°C and 100% acetonitrile for 10 min at 20°C. After removing the destain solution, 10 mM dithiothreitol in 100 mM ammonium bicarbonate was added to cover the gel pieces and incubated for 45 min at 50°C. For alkylation, dithiothreitol was replaced with 55 mM iodoacetamide in 50 mM ammonium bicarbonate, and the samples were incubated for 30 min in the dark at 20°C. Gel pieces were rinsed twice with 50 mM ammonium bicarbonate for 15 min at 20°C. Protein bands were digested overnight at 37°C with 150 ng of freshly prepared sequencing grade trypsin. Equal volumes of water and 100 mM ammonium bicarbonate were added to the digestion solution to give a final trypsin solution of 1 ng/l. After a C18 Zip-Tip (Millipore) purification, the protein solution was mixed with an equal volume of ␣-cyano-4-hydroxycinnamic acid matrix (Sigma) and analyzed by MALDI-TOF/MS (MALDI LR™, Micromass, UK, located at the Indiana University, Purdue University Indianapolis Proteomic Core Facility). The mass of peptides obtained from each band were compared with the computer-generated list of deduced tryptic fragments to identify the 80-Da shift corresponding to phosphorylation using the FindMod site (ca.expasy.org/tools/findmod) for single phosphorylation sites and a Command Prompt program written to find multiple phosphorylations on single peptide fragments. Genomic Solutions™ (65.219.84.5/index.html) web site was used to search for protein identification (ProFound™) and to identify the theoretical phosphorylation sites on ERD14 (Protein Association Work Station, PAWS™).

RESULTS
In Vitro Phosphorylation and Calcium Binding of ERD14 -ERD14 was cloned from an Arabidopsis expression library using antibody raised against celery VCaB45, a dehydrin-like protein (33). The apparent molecular mass of the purified E. coli-expressed ERD14, as estimated by 10% SDS-PAGE, was 37 kDa (See Figs. 1, 2, and 3B). The molecular mass of ERD14, deduced from the amino acid sequence, is 20.9 kDa. The mass of the protein used in these studies is predicted to be 20.6 kDa (lacks the amino-terminal two amino acids). The anomalous migration resulting in the overestimation of mass by SDS-PAGE is commonly found in other dehydrins (5,33,(35)(36)(37)(38).
Our previous work with a dehydrin-like protein (33) and the presence of potential casein kinase II phosphorylation sites on ERD14 and other dehydrins suggested that ERD14 might be a substrate for CKII. Indeed, treatment of the purified E. coliexpressed ERD14 with CKII resulted in a shift in the apparent molecular mass on SDS-PAGE gels (Figs. 1A and 2), with an apparent increase of ϳ5 kDa. The shifted polypeptides were confirmed to be ERD14 with a dehydrin-specific antibody, raised against the conserved K-domain of dehydrins (Fig. 1A). The ERD14 gel shift was nearly completed after a 30-min phosphorylation (Fig. 1A). Dephosphorylation of CKII-phos-phorylated ERD14 with shrimp alkaline phosphatase (SAP) resulted in a quantitative return to the smaller molecular mass (Fig. 2). We have utilized this gel shifting phenomena to routinely assess the phosphorylation status of ERD14, as demonstrated for other phosphorylated proteins (33,39,40). To quantitate ERD14 phosphorylation, ERD14 was phosphorylated with CKII in the presence of [␥-32 P]ATP. A rapid increase in phosphorylation of ERD14 was observed in the first 30 min of CKII treatment and approached a plateau after 1 h. Phosphorylation of ERD14 correlated temporally with the gel shifting of ERD14 on SDS-PAGE gels (Fig. 1B).
To determine whether the E. coli-expressed ERD14 bound calcium and whether this activity was regulated by phosphorylation, ERD14 was in vitro phosphorylated with CKII and separated on SDS-PAGE. Calcium binding activity was then assessed by calcium ( 45 CaCl 2 ) ligand blots and visualized by phosphorimaging. Fig. 1C demonstrates that E. coli-expressed ERD14 binds calcium only when phosphorylated. Quantitative analysis of the calcium ligand blot showed an increase in ERD14 calcium binding with increased phosphorylation and that ERD14 calcium binding activity approached saturation after 1 h (Fig. 1C). Calcium binding, as measured by a calciumligand blot (Fig. 1C), correlated well with the gel shifting of ERD14 on SDS-PAGE gels (Fig. 1A) and with the extent of phosphorylation (Fig. 1B).
To analyze the extent of phosphorylation of the E. coli-expressed ERD14, CKII-phosphorylated ERD14 (with [␥-32 P] ATP) was precipitated with trichloroacetic acid and collected on glass fiber filters. Approximately two mol of phosphate were incorporated per mol of ERD14 after 3 h of phosphorylation (Fig. 3A), indicating at least 2 possible phosphorylation sites on ERD14. No further 32 P was incorporated into ERD14 for up to 5 h of phosphorylation. In some experiments fresh CKII was added after 3 h, 2 with no further incorporation observed. Western blots probed with antibodies raised against phosphoserine and phosphothreonine indicated that both threonine and serine were possible phosphorylation sites on ERD14 (Fig. 3B). Because phosphothreonine was detected only after 3 h of phosphorylation it is unlikely the phosphorylation requisite for the gel shifting and calcium binding activation was due to phosphorylation on a threonine. Western blots probed with anti-   phosphoserine showed a subtle increase in serine phosphorylation over time (15-120 min). However, the anti-phosphoserine antibodies used were not specific for serine phosphorylation (Sigma, catalog number P3430, and Zymed Laboratories Inc., catalog number 61-8100) because they reacted with non-phosphorylated ERD14.
Because dehydrins seem to lack significant secondary structure and have been considered to be intrinsically unstructured proteins (41), we thought it was possible that activation of calcium binding could be due to a rapid phosphorylation but require a slower induced structural change. To test this hypothesis ERD14 was treated for 30 min with CKII, further phosphorylation was prevented by adding staurosporine, and the incubation was continued. No further increase in calcium binding occurred with further incubation in the presence of staurosporine (Fig. 4). These results suggest that gel shifting requires an enzymatically active kinase and that calcium binding activity of the purified ERD14 does not require slow phosphorylation-dependent conformational changes.
The quantitative effect of the phosphorylation on ERD14 calcium binding activity is illustrated by equilibrium dialysis experiments, which show saturable binding of calcium to native ERD14 (see Fig. 5). Scatchard plot analysis (Fig. 5B) indi-cated that CKII-phosphorylated ERD14 binds ϳ3 mol of calcium per mol of ERD14 with relatively low affinity, having an apparent K d for calcium of 120 M.
We examined inhibition of calcium binding by various ions (K ϩ , Mg 2ϩ , Zn 2ϩ , Mn 2ϩ , Fe 2ϩ , Fe 3ϩ , and La 3ϩ ) at 0.25 mM final concentration (Fig. 6). K ϩ and the trivalent Fe 3ϩ had little affect on calcium binding to ERD14. In contrast, the divalent cations Mg 2ϩ , Zn 2ϩ , Mn 2ϩ , or Fe 2ϩ did significantly reduce ERD14 calcium binding activity. In the presence of the tri-  valent cation, lanthanum, calcium binding to ERD14 was strongly inhibited.
In Vivo Phosphorylation and Calcium Binding of ERD14 -The accumulation of dehydrins in response to low temperature has been previously reported (5,8,33,35). Here we wanted to determine whether the dehydrin ERD14 phosphorylation could be modulated by a cold-regulated kinase. E. coli-expressed ERD14 was phosphorylated in the presence of [␥-32 P]ATP with extracts from 4-or 20°C-treated Arabidopsis plants. In the absence of added ERD14, no incorporation of 32 P into extract proteins was observed. In the presence of ERD14, phosphorylation was detected at the expected mass of ERD14. A subtle increase of kinase activity toward ERD14 in cold-treated Arabidopsis extracts was observed (Fig. 7, A and B).
To determine whether the calcium binding activity of ERD14 derived from Arabidopsis extracts is comparable with E. coliexpressed ERD14, we isolated ERD14 from 4-week-old coldtreated Arabidopsis plants. ERD14 was purified from low density membranes which were subsequently permeabilized with 0.2% w/w Triton X-100 followed by heat treatment and anion exchange chromatography. Peak fractions were analyzed by SDS-PAGE and Western blotting (Fig. 8A). A band corresponding to ERD14 was excised and analyzed by MALDI-TOF/MS. ERD14 identification was confirmed by comparison of the observed peptides to the SWISS-PROT nr data base (ProFound™ web site in Genomic Solution™, 65.219.84.5/index.html). The best match was found to be the Arabidopsis dehydrin ERD14, with a Z-value of 1.86, (with 44% coverage) and a probability of 1.0e ϩ000 . Similar to E. coli-expressed ERD14, treatment of Arabidopsis ERD14 with SAP resulted in a shift to a smaller molecular mass on SDS-PAGE gels and greatly reduced the ability of this protein to bind calcium (Fig. 8B). These results indicated that the Arabidopsis ERD14 is in vivo phosphorylated in cold-treated plants and its calcium binding activity is phosphorylation-dependent.
Analysis of Phosphorylation Sites in ERD14 -E. coli expressed ERD14 is phosphorylated by CKII at a minimum of two phosphorylation sites (ϳ1.6 mol of phosphate incorporated per mol of ERD14) (Fig. 3A). Furthermore, based on the kinetics of phosphorylation (Figs. 1, A and B, and 3A), it appeared that one site was phosphorylated rapidly, whereas a second was phosphorylated more slowly. Purified E. coli expressed ERD14 was phosphorylated with casein kinase II for 0, 30, and 180 min and then analyzed by MALDI-TOF/MS. Mass spectrometric analysis consistently identified one phosphorylated peptide (peptide IV) from the in vitro phosphorylated ERD14 (Fig. 9), although several others were observed in more than one experiment. We compared phospho-peptides from the in vitro phosphorylated ERD14 to that of the in vivo phosphorylated ERD14 derived from cold-treated Arabidopsis plants. Because both the in vivo and in vitro phosphorylated ERD14 proteins showed alterations in both gel shifting and calcium binding, we reasoned that the common phosphorylated peptides would be likely candidates for containing the phospho-peptide responsible for calcium binding. Four phosphorylated peptides (II, IV, V, and VII) were consistently identified (Fig. 9) on the in vivo phosphorylated Arabidopsis ERD14 and were found consistently in common to both in vitro and in vivo phosphorylated ERD14. Phosphorylation of peptide VII was deduced not to be important for calcium binding activity as the phosphate was not removed from that site by SAP treatment even though this treatment abrogated calcium binding (Fig. 8). The remaining peptides were dephosphorylated after treatment of Arabidopsis ERD14 with SAP. Of these only three phospho-peptides (II, IV, and V) appeared to temporally correlate with the kinetics of the gel shifts and calcium binding; that is, their presence was consistently observed after only 30 min of phosphorylation. Because of the lack of a significant amount of phosphothreonine detectable at short times (see Fig. 3B), we concluded that the peptide V was not a likely candidate responsible for the shifting and calcium binding activation. Of the two remaining phosphopeptides only peptide IV, containing the S-domain, was identified in every experiment; its presence consistently correlated with calcium binding activity. The ERD14 S-domain (peptide IV) has four potential phosphorylation sites for CKII. Analyses at 30 min of phosphorylation of E. coli-expressed ERD14 with CKII indicated 2 phosphorylated serines in the S-domain and 3 phosphorylation sites after 3 h of phosphorylation. MALDI-TOF analyses of purified in vivo phosphorylated Arabidopsis ERD14 also revealed up to three phosphorylated serines in the S-domain. DISCUSSION ERD14 was previously identified as a cDNA encoding a member of the dehydrin family of proteins (42,43). Although the expression patterns and transcriptional regulation of dehydrins have been well characterized (5,6,8,(42)(43)(44), the biochemical function of these proteins and their physiological role in plants during stress conditions are not fully understood. Here, we have shown that ERD14 possesses ion binding activity that is dependent on its phosphorylation status. Interaction of dehydrins on immobilized metal-ion affinity chromatography columns (38) was consistent with the postulate that dehydrins might interact with ions (45). Recently a phloem-localized dehydrin (iron transport protein) from castor bean was shown to bind Fe 3ϩ and was postulated to play a role in iron transport (34). In that study, it was shown that the iron transport protein preferentially binds Fe 3ϩ over Fe 2ϩ and complexes with Cu 2ϩ , Zn 2ϩ , and Mn 2ϩ in vitro. In another case, a vacuole-localized celery dehydrin (VCaB45) was shown to bind Ca 2ϩ (33) and was postulated to function as a calcium buffer or as a calciumdependent chaperone. In both cases proteins were isolated from plant tissues in an active ion binding form (likely phosphorylated). In this study we advance this understanding by showing that the Arabidopsis dehydrin ERD14 is capable of binding a variety of ions and that ion binding (calcium in particular) is dependent upon the ERD14 phosphorylation state. Arabidopsis ERD14 is phosphorylated in vivo. Furthermore, we have localized the activating phosphorylation site to be contained within the S-domain, a common feature of most dehydrins.
Of all the ions tested for interaction with ERD14, it seems most likely that the binding of calcium and magnesium to ERD14 are the most physiologically relevant. Intraorganellar levels of calcium and magnesium may exceed 1 mM, and cytoplasmic levels of these ions reach micromolar concentrations under a variety of conditions. Thus, the observed K d of 120 M for calcium suggests the possibility for reversible binding at near physiological conditions. ERD14 binds about 3 mol of calcium per mol, significantly less than the ϳ25 mol bound to VCaB45 (33). No information is available for the number of Fe 3ϩ binding sites for the dehydrin-like iron transport protein.
Unlike iron transport protein, the inhibition of calcium binding suggests that ERD14 may bind Fe 2ϩ preferentially over Fe 3ϩ . The K d for the highest affinity calcium binding sites for ERD14 (120 M) and VCaB45 (200 M) are similar. Known calciumbinding proteins such as calreticulin, calsequestrin, calnexin, and calmegin use the negatively charged acidic amino acid rich-region at the carboxyl-terminal (C-domain) quarter of the protein to bind calcium at high capacity and low affinity (0.3-2 mM) (46). Calreticulin also binds calcium with high affinity in a proline-rich domain (ϳ1 M) (46). Generally, calcium binding domains include the presence of glycine or proline and the presence of acidic or hydroxylated residues. ERD14, an acidic protein with a pI of 5.28, which is highly enriched in glutamic acid (constitutes 21% of the protein), has several regions that meet these general calcium binding characteristics.
We demonstrated that the dehydrin ERD14 is phosphorylated in planta and can be phosphorylated in vitro with casein kinase II. The phosphorylation state of the S-domain likely regulates ion binding. Like other dehydrins, ERD14 has many potential phosphorylation sites. Both within and outside the S-domain there are multiple consensus phosphorylation sites for the protein kinase CKII. When E. coli-expressed ERD14 is phosphorylated with CKII, phosphorylation approaches saturation at 2 mol of phosphate per mol of ERD14. Mass spectrometry revealed several phosphorylated peptides in common to the purified Arabidopsis ERD14 and CKII-phosphorylated E. coli-expressed ERD14. Only one phospho-peptide, SDSSSS-SSSEEEGSDGEK, was found consistently by MALDI and was always correlated with the characteristics of the gel shifts and calcium binding. It is intriguing that this site and the regions immediately surrounding it are highly conserved in a number of Arabidopsis dehydrins. Because the phosphothreonine-specific antibodies revealed no phosphothreonine that correlated with the ability of ERD14 to bind calcium, we have concluded that the S-domain contains the activating phosphorylation site. A similar site has been shown to be phosphorylated in maize dehydrin, Rab17 (11)(12)(13); in this case it was concluded to be important for nuclear targeting or membrane association. The commonality between in vivo and in vitro phosphorylated peptides on ERD14 suggests that CKII may be responsible for in vivo phosphorylation of ERD14. We observed that the activity of an endogenous kinase capable of phosphorylating ERD14 is increased after cold stress (Fig. 7), suggesting that this might be a cold-modulated kinase, perhaps CKII. It is unlikely that Purified E. coli-expressed ERD14 was phosphorylated with casein kinase II for the indicated times and tryptic-digested. ERD14, isolated from 4-week-old Arabidopsis seedlings treated for 2 days at 4°C was incubated for 20 min at 37°C in the presence or absence of SAP. After SDS-PAGE, ERD14 bands were stained, excised, destained, reduced, alkylated, trypsin-digested, and extracted as described under "Experimental Procedures." ERD14-digested solutions were mixed with equal volumes of ␣-cyano-4-hydroxycinnamic acid matrix and analyzed by MALDI-TOF/MS. ProFound™ and PAWS™ in the Genomic Solutions™ (165.219.84.5/index.html) web site were used for protein identification and to identify the theoretical phosphorylation sites on ERD14 with four tryptic digestion missed cleavages, respectively. Panel A, ERD14 amino acid sequence. Bold letters indicate the predicted phosphorylation sites for protein casein kinase II, determined by PhosphoBase Version 2.0 (www.cbs.dtu.dk/databases/PhosphoBase). Lines and roman numerals show phospho-peptide positions and designated peptide numbers. K-domains are indicated by dotted underlines. Panel B, phosphorylation sites in E. coli-expressed and Arabidopsis ERD14. Results are from three experiments, a, b, and c. Fragments obtained from each reaction were compared with the computer-generated list to identify the 80-, 160-, 240-, and 320-Da shift corresponding to phosphorylation sites 1, 2, 3, and 4, respectively. The sequences shown are the minimum sized tryptic fragments (except peptide IV); however, alternative peptides (up to four missed tryptic cleavages) were examined for possible phosphorylation modifications and included in the analysis. Panel C, the ion spectra map depicting the monoisotopic profile of the phospho-peptides (II, IV, V, and VII) found in the ERD14 isolated from Arabidopsis plants (not treated with SAP) acquired by MALDI-TOF/MS. Peptide II (mass/z was 1584.72) has no missed cleavages and one phosphorylation; peptide IV (mass/z was 2186.77) has one missed cleavage (indicated by R) and three phosphorylations; peptide V (mass/z was 982.48) has no missed cleavage and one phosphorylation; and peptide VII (mass/z was 995.4) has no missed cleavage and two phosphorylations. In parentheses are the peptide numbers on ERD14 amino acid sequence (from panel A). x axis is the mass-to-charge value of the ion (m/z), and y axis represents peak relative intensity. this increase reflects an alteration in phosphatase activity, because phosphatase inhibitors were present in the assay. Accompanying phosphorylation was a shift in apparent mass. Whether under normal growth conditions or under cold stress conditions, Arabidopsis ERD14 was always present in the phosphorylated state and always had calcium binding activity (data not shown). This suggests that the phosphorylation is constitutive and might not be used as a means to modulate ion binding activity under these conditions.
The physiological function of the cold-induced dehydrins remains unknown. It was recently shown (9) that an acidic dehydrin, DHN1, binds preferentially to acidic phospholipids. In addition, phosphorylation of this same protein appears to be requisite for targeting to nucleus (13). These latter results are consistent with a possible role for dehydrins in stabilizing membranes in stress conditions (7,47). It was of interest to determine whether calcium or phosphorylation of the dehydrins had an impact on lipid interactions. A logical extension from our work was to determine whether membrane association of dehydrins might be dependent on calcium. Although low concentrations of Triton X-100 readily remove ERD14 from membranes (a routine purification step used in this paper) we have been unable to remove ERD14 from the membranes with EGTA (data not shown). This is consistent with the conclusion drawn by Koag et al. (9) that dehydrins may interact with membranes through a lipid binding class A2 ␣-helix structure. Ion binding is likely not requisite for membrane interaction, and we speculate that the calcium (or ion) binding has an independent function. We have proposed that an alternative function of the dehydrins might be as calcium-dependent protein or carbohydrate chaperones, a function similar to that performed by calreticulin and calnexin (48,49). The similarity in structure of the chaperones and ERD14, number of calcium binding sites, and affinity make this postulate attractive. This hypothesis has not yet been tested.
This study provides evidence that ion binding, most likely to be calcium, is one of the major biochemical functions of the Arabidopsis ERD14 and perhaps of the other acidic dehydrins. Furthermore, the ability of the acidic dehydrins, in particular ERD14, to bind calcium is regulated by their phosphorylation state. On the basis of these results, we hypothesize that ERD14 and perhaps other dehydrins could act as either calcium buffers, conferred by their high calcium binding capacity, or have calcium-dependent chaperone-like activity similar to the function of calreticulin and calnexin. However, further studies are under way to evaluate which of the models is the likely biochemical function of ERD14.