Interaction of Plant Chimeric Calcium/Calmodulin-dependent Protein Kinase with a Homolog of Eukaryotic Elongation Factor-1α*

A chimeric Ca2+/calmodulin-dependent protein kinase (CCaMK) was previously cloned and characterized in this laboratory. To investigate the biological functions of CCaMK, the yeast two-hybrid system was used to isolate genes encoding proteins that interact with CCaMK. One of the cDNA clones obtained from the screening (LlEF-1α1) has high similarity with the eukaryotic elongation factor-1α (EF-1α). CCaMK phosphorylated LlEF-1α1 in a Ca2+/calmodulin-dependent manner. The phosphorylation site for CCaMK (Thr-257) was identified by site-directed mutagenesis. Interestingly, Thr-257 is located in the putative tRNA-binding region of LlEF-1α1. An isoform of Ca2+-dependent protein kinase (CDPK) phosphorylated multiple sites of LlEF-1α1 in a Ca2+-dependent but calmodulin-independent manner. Unlike CDPK, CCaMK phosphorylated only one site, and this site is different from CDPK phosphorylation sites. This suggests that the phosphorylation of EF-1α by these two kinases may have different functional significance. Although the phosphorylation of LlEF-1α1 by CCaMK is Ca2+/calmodulin-dependent, in vitro binding assays revealed that CCaMK binds to LlEF-1α1 in a Ca2+-independent manner. This was further substantiated by coimmunoprecipitation of CCaMK and EF-1α using the protein extract from lily anthers. Dissociation of CCaMK from EF-1α by Ca2+ and phosphorylation of EF-1α by CCaMK in a Ca2+/calmodulin-dependent manner suggests that these interactions may play a role in regulating the biological functions of EF-1α.

A chimeric Ca 2؉ /calmodulin-dependent protein kinase (CCaMK) was previously cloned and characterized in this laboratory. To investigate the biological functions of CCaMK, the yeast two-hybrid system was used to isolate genes encoding proteins that interact with CCaMK. One of the cDNA clones obtained from the screening (LlEF-1␣1) has high similarity with the eukaryotic elongation factor-1␣ (EF-1␣). CCaMK phosphorylated LlEF-1␣1 in a Ca 2؉ /calmodulin-dependent manner. The phosphorylation site for CCaMK (Thr-257) was identified by site-directed mutagenesis. Interestingly, Thr-257 is located in the putative tRNA-binding region of LlEF-1␣1. An isoform of Ca 2؉ -dependent protein kinase (CDPK) phosphorylated multiple sites of LlEF-1␣1 in a Ca 2؉ -dependent but calmodulin-independent manner. Unlike CDPK, CCaMK phosphorylated only one site, and this site is different from CDPK phosphorylation sites. This suggests that the phosphorylation of EF-1␣ by these two kinases may have different functional significance. Although the phosphorylation of LlEF-1␣1 by CCaMK is Ca 2؉ /calmodulin-dependent, in vitro binding assays revealed that CCaMK binds to LlEF-1␣1 in a Ca 2؉ -independent manner. This was further substantiated by coimmunoprecipitation of CCaMK and EF-1␣ using the protein extract from lily anthers. Dissociation of CCaMK from EF-1␣ by Ca 2؉ and phosphorylation of EF-1␣ by CCaMK in a Ca 2؉ /calmodulin-dependent manner suggests that these interactions may play a role in regulating the biological functions of EF-1␣.
Ca 2ϩ is a universal second messenger that regulates diverse developmental and physiological processes in plants (1). One of the major mechanisms decoding the change of intracellular Ca 2ϩ and transducing Ca 2ϩ signal is the action of Ca 2ϩ -modulated proteins. Calmodulin (CaM) 1 is a highly conserved and the most widely distributed Ca 2ϩ -binding protein (2). CaM is believed to be a primary receptor for intracellular Ca 2ϩ and functions as a regulatory element for its target proteins when activated by Ca 2ϩ . One group of CaM-modulated proteins is composed of Ca 2ϩ /CaM-dependent protein kinases.
Extensive investigation of protein phosphorylation and dephosphorylation has revealed that protein kinases play key roles in various signal transduction pathways leading to cellular and physiological responses (3). Ca 2ϩ /CaM-dependent protein phosphorylation has been implicated in regulating a broad array of biological functions and is believed to play a pivotal role in amplifying and diversifying the action of Ca 2ϩ /CaMmediated signals in animals. A number of different types of Ca 2ϩ /CaM-dependent protein kinases, including CaM kinases (CaMKs) I-IV, phosphorylase kinase, and myosin light chain kinase, have been cloned and demonstrated to regulate various cellular processes (4,5). Some indirect evidence for the existence of Ca 2ϩ /CaM-dependent protein phosphorylation has been reported in plants (1). However, until recently, no direct and convincing evidence has been presented. In recent years, three Ca 2ϩ /CaM-dependent protein kinase genes having some features similar to the mammalian multifunctional Ca 2ϩ /CaM-dependent protein kinase (CaMKII) were cloned from plants (6 -8). The apple Ca 2ϩ /CaM-dependent protein kinase was isolated using an interaction screening method with labeled CaM as a ligand probe (6). CCaMK from lily and MCK1 from maize were cloned using the polymerase chain reaction screening method with oligonucleotide primers designed based on the conserved regions of mammalian CaMKs (7,8). However, the sequence similarity between lily CCaMK, maize MCK1, and apple Ca 2ϩ / CaM-dependent protein kinase is only observed in their conserved kinase domains. No obvious sequence similarity exists outside their kinase domains. Although these three kinases were directly or indirectly demonstrated to bind CaM in a Ca 2ϩ -dependent manner, only biochemical properties for CCaMK have been reported to show that its kinase activity is regulated by Ca 2ϩ /CaM (9,10).
CCaMK has a unique structural feature characterized by the presence of a kinase domain, a CaM-binding domain, and a neural visinin-like Ca 2ϩ -binding domain within a single polypeptide (7). Its kinase domain and CaM-binding domain are highly similar to those of mammalian CaMKII. The sequence of its C-terminal domain does not show significant homology to any known protein kinase. This domain has high similarity with a family of neural visinin-like Ca 2ϩ -binding proteins (7), and its Ca 2ϩ -binding property was confirmed by biochemical characterization (9). In plants, a predominant and widely expressed form of protein kinase is the Ca 2ϩ -dependent protein kinase or CaM-like domain protein kinase (CDPK) (11,12). CDPK contains a kinase domain similar to that of mammalian CaMKII, a junction domain, and a CaM-like domain (11). Although both CCaMK and CDPK contain a kinase domain and a Ca 2ϩ -binding domain and have similarity in their overall structures, they distinctly differ in terms of their se-quences as well as regulation of their kinase activities. There is no striking sequence similarity shared between them except their kinase domains. Unlike the CaM-like domain of CDPK, the Ca 2ϩ -binding domain of CCaMK contains three EF-hand Ca 2ϩ -binding sites and has higher similarity to visinin-like proteins than CaM (7). Unlike CCaMK, the full-length CDPK does not bind to CaM, and its kinase activity is Ca 2ϩ -dependent but does not require exogenous CaM (13). Another difference between these kinases is that CDPK is encoded by a large family of genes and is ubiquitously distributed among plant tissues, while CCaMK is a single copy gene in lily and its expression is developmentally and spatially regulated (7).
Study of protein-protein interactions between a kinase and its target proteins is crucial to illustrate intracellular signal transduction that eventually leads to physiological responses. Isolation and characterization of interacting proteins/substrates of CCaMK is a major step in understanding its in vivo regulation, cellular compartmentalization, and biological functions. In this study, the yeast two-hybrid interaction cloning system was used to isolate cDNAs encoding proteins that interact with CCaMK. Here we report that one of the cDNA clones (LlEF-1␣1) isolated from the screening is homologous to the eukaryotic elongation factor-1␣ (EF-1␣), and CCaMK phosphorylates LlEF-1␣1 in a Ca 2ϩ /CaM-dependent manner, but their binding does not require Ca 2ϩ .

EXPERIMENTAL PROCEDURES
Plant Material-Lily (Lilium longiflorum Thunb cv. Nellie White) plants were grown in the greenhouse. Various organs and anthers of different stages were collected and immediately frozen in liquid nitrogen for isolation of RNA and proteins.
Cloning of LlEF-1␣1 Using the Yeast Two-hybrid System-The twohybrid cDNA cloning kit was purchased from Stratagene, and the screening procedure was followed as recommended by the manufacturer. The mRNA was isolated from immature lily anthers (flower bud size 0.5-2.5 cm) using standard protocols (14). A lily immature anther cDNA library was constructed in HybriZAP two-hybrid vector and expressed as fusion proteins with the activation domain of GAL4 in pAD-GAL4. The full coding region of lily CCaMK was cloned in frame into pBD-GAL4 and expressed as a fusion protein with the binding domain of GAL4. The yeast strain YRG-2 carrying the pBD-GAL4/ CCaMK plasmid was transformed with the plasmids excised from the library. Approximately 5 ϫ 10 5 transformants were screened for growth on medium lacking leucine, tryptophan, and histidine. Growing yeast colonies were screened for expression of ␤-galactosidase using the chromogenic substrate, 5-bromo-4-chloro-3-indolyl-␤-D-galactopyranoside. The positive clones from the His ϩ and LacZ ϩ colonies were retransformed into YRG-2 alone, with pBD-GAL4/CCaMK, or with the negative control plasmids (pGD-GAL4, p53, and pLamin C). Only clones that did not activate lacZ alone and in the presence of the negative control plasmids were characterized further. In order to clone the cDNA of LlEF-1␣1 containing the complete coding region, the partial clone obtained from the screening was used as a probe to screen a cDNA library derived from developing anthers of lily (7).
Preparation of the Anti-CCaMK Antibody and Western Analysis-A rabbit polyclonal antibody was raised against the visinin-like domain (amino acids 358 -520) of lily CCaMK. The anti-CCaMK antibody was further purified from antiserum by blot affinity purification essentially as described (15). Proteins (60 g) extracted from different tissues of lily and anthers at different stages were separated on a 10% SDS-polyacrylamide gel and transferred onto polyvinylidene difluoride membrane (Millipore Corp.). The membrane was immunoblotted with either the affinity-purified anti-CCaMK antibody or a polyclonal anti-EF-1␣ antibody, followed by incubation with the goat anti-rabbit IgG conjugated with horseradish peroxidase (Bio-Rad). The immunoblot was developed using an enhanced chemiluminescent substrate (Pierce) and exposed to autoradiography film.
Plasmid Construction and Site-directed Mutagenesis-The original cDNA clone isolated from the yeast two-hybrid screening encodes the C-terminal portion (amino acids 204 -447) of LlEF-1␣1. This cDNA clone was amplified, and BamHI and EcoRI cloning sites were introduced by polymerase chain reaction. The amplified fragment was subcloned in frame into the expression vector pGEX-3X (Amersham Pharmacia Biotech), resulting in pmC. The cDNA fragment with engineered BamHI and EcoRI restriction sites coding for the N-terminal portion (amino acids 1-221) of LlEF-1␣1 was amplified and inserted in frame into pGEX-3X to generate pmN. The full coding region of CRPK2, a Ca 2ϩ -dependent protein kinase gene from corn root tips (16), was amplified, and NdeI and EcoRI restriction sites were introduced and subcloned in frame into pET-14b (Novagen) to generate the CRPK2 expression construct pCRPK2. All of these constructs were sequenced to confirm that no mutation was introduced. The site-directed mutagenesis was performed using a kit purchased from Stratagene. The oligonucleotide primers used for generating the mutant pmC 257A (Thr was mutated into Ala at the amino acid residue 257 in pmC) were 5Ј-CAGTCGGCCGTGTGGAGGCTGGTATTGTGAAG-3Ј and 5Ј-CTTCA-CAATACCAGCCTCCACACGGCCGACTG-3Ј. The sequence of the mutant was confirmed by DNA sequencing, and no additional mutation was introduced.
Expression and Purification of CCaMK, CRPK2, and LlEF-1␣1 Deletion Mutants-CCaMK was expressed and purified as described previously (9). For expression and purification of CRPK2, Escherichia coli BL21 (DE3) cells transformed with pCRPK2 were grown at 37°C in M9 minimal medium supplemented with 0.2% casein enzymatic hydrolysate, 0.4% glucose, 10 mM magnesium sulfate, and 100 mg/liter ampicillin. The expression of CRPK2 was induced for 3 h by adding 0.5 mM isopropyl ␤-D-thiogalactoside after A 600 reached 0.6 units. CRPK2 was purified according to the protocol provided by the manufacturer (Novagen). The glutathione S-transferase (GST) fusion proteins (mN, mC, and mC 257A ) were purified on glutathione-Sepharose 4B columns using standard procedures, except that the columns were washed extensively with phosphate buffer containing 1 mM DTT, 0.1% Tween 20, 2 mM ATP, 10 mM MgSO 4 , and 1.2 M NaCl to further eliminate contamination. Eluted proteins were thoroughly dialyzed with Centricon-30 against Tris-HCl (pH 7.5) buffer containing 1 mM DTT and 10% ethylene glycol. Protein concentrations were determined by Bradford's method or by SDS-PAGE using bovine serum albumin as a standard.
Protein Kinase Assay-Phosphorylation assays were carried out as previously reported (9). The indicated amounts of CCaMK, CRPK2, and LlEF-1␣1 deletion or site-directed mutants were added into the reaction mixtures. Proteins were resolved by SDS-PAGE and stained with Coomassie Brilliant Blue. The gels were dried and subjected to autoradiography. 32 P incorporation was determined by counting the excised protein bands in a liquid scintillation counter.
In Vitro Binding Assay-In vitro binding experiments were performed as described by Watanabe et al. (17). The assays were carried out in 500 l of 40 mM Tris-HCl (pH 7.5) containing 1 mM DTT and 30 l of glutathione-Sepharose 4B. CCaMK and the deletion mutant of LlEF-1␣1 (mN or mC) were added into the mixture in the presence of 0.5 mM CaCl 2 , 0.5 mM CaCl 2 plus 1 M CaM, or 2.5 mM EGTA. The reaction was incubated at 4°C for 1 h. The Sepharose beads were collected and washed three times in 1 ml of 40 mM Tris-HCl buffer containing 300 mM NaCl and either 0.5 mM CaCl 2 or 2.5 mM EGTA. After the final wash, the beads were resuspended in 50 l of 2.5ϫ SDS-PAGE sample buffer and boiled for 5 min. Eluted proteins were subjected to SDS-PAGE and visualized by staining with Coomassie Brilliant Blue or transferred onto polyvinylidene difluoride membrane for immunoblot analysis.
Immunoprecipitation-Lily anthers (flower bud size 0.9 -2.0 cm) were powdered and homogenized in the extraction buffer containing 1 mM phenylmethylsulfonyl fluoride, 5 g/ml antipain, 5 g/ml aprotinin, 5 g/ml leupeptin, and 5 g/ml pepstatin in the presence of either 1 mM CaCl 2 or 2.5 mM EGTA. The homogenate was centrifuged at 21,000 ϫ g for 30 min at 4°C. The supernatant was used as the crude protein extract. 60 l of protein G-agarose beads (Life Technologies, Inc.) were incubated with 4 g of the anti-CCaMK antibody in the extraction buffer for 4 h at 4°C with gentle agitation. The beads were collected and washed three times in 1 ml of the extraction buffer. An aliquot of the crude extract containing 800 g of protein was added to the conjugate of the anti-CCaMK antibody and protein G-agarose beads and incubated with gentle agitation for 4 h at 4°C. The immunoprecipitate was washed three times with 1 ml of the extraction buffer in the presence of either 1 mM CaCl 2 or 2.5 mM EGTA and subjected to immunoblot analysis with a mouse monoclonal antibody against urchin EF-1␣. Control experiments were carried out by incubation of protein G-agarose beads with either the anti-CCaMK antibody alone or the crude protein extract alone.

RESULTS
Cloning of LlEF-1␣1 cDNA-The yeast two-hybrid interaction cloning method was used to isolate genes encoding interacting proteins and/or substrates of CCaMK. Since lily CCaMK is expressed in a stage-specific manner during anther development (7), a cDNA library was constructed using mRNA isolated from lily anthers at the corresponding stages and expressed as fusion proteins with the GAL4 activation domain. The fusion protein of the GAL4 DNA-binding domain/CCaMK was expressed in pBD-GAL4 and utilized as the bait protein to screen the cDNA library. Among the positive clones identified from several rounds of screening, sequence comparison showed that one of these clones has high homology with the eukaryotic elongation factor-1␣. The clone is 1016 base pairs long and encodes the C-terminal portion (amino acids 204 -447) of LlEF-1␣1 ( Fig. 1). This cDNA clone was used as a probe to obtain its clone containing the full coding region by screening a cDNA library from which CCaMK was originally isolated (7). The clone coding for the full-length polypeptide was designated as LlEF-1␣1 (Fig. 1).
Expression of CCaMK and EF-1␣-Since the visinin-like domain is unique to CCaMK (7), the fusion protein of the visininlike domain was expressed and purified from E. coli for producing a polyclonal antibody to analyze the expression pattern of CCaMK. CCaMK is specifically expressed only in root tips and anthers, and its expression is undetectable among other organs examined (Fig. 3). The expression of CCaMK is developmentally regulated in anthers. At the early stages, the protein level is very low. CCaMK expression reaches the highest level at the stage when the flower bud size is between 0.9 and 2.0 cm, coinciding with microsporogenesis (18). Its expression then decreases until it becomes undetectable at the later stages of anther development. On the other hand, EF-1␣ is a ubiquitous protein and expressed in all of the organs examined. However, its expression level varies dramatically. EF-1␣ is highly expressed in root tips and anthers at the early stages of development, coinciding with the expression pattern of CCaMK. It appears that both CCaMK and EF-1␣ are highly expressed in tissues with high mitotic and meiotic activities.
CCaMK binds CaM in a Ca 2ϩ -dependent manner (7,9). Endogenous CCaMK was obtained from lily anthers by passing the protein extract through a calmodulin-Sepharose column. Interestingly, EF-1␣ was copurified with CCaMK on the calmodulin-Sepharose column (Fig. 4). However, this result does not demonstrate a direct interaction between CCaMK and EF-1␣, since EF-1␣ was reported to bind to CaM in the presence of Ca 2ϩ (19,20).
Ca 2ϩ /CaM-dependent Phosphorylation of LlEF-1␣1 by CCaMK-The fact that LlEF-1␣1 was cloned using the yeast two-hybrid system indicates that it interacts with CCaMK. There may be different ways in which CCaMK interacts with LlEF-1␣1: (a) LlEF-1␣1 may physically bind to CCaMK; (b) it may be one of the substrates of CCaMK; and (c) it may serve as an effector of CCaMK. To determine the nature of interaction between CCaMK and LlEF-1␣1, we first examined whether CCaMK can phosphorylate LlEF-1␣1 because of the identity of CCaMK as a protein kinase. The N-terminal and C-terminal regions of LlEF-1␣1 were expressed and purified as GST fusion proteins (mN and mC, Fig. 5A). In the presence of EGTA or Ca 2ϩ , CCaMK phosphorylated the mC at the basal level (Fig.  5B). The phosphorylation level was stimulated up to around 25-fold by adding Ca 2ϩ and CaM, indicating that CCaMK phosphorylates the mC in a Ca 2ϩ /CaM-dependent manner. The stimulation of the kinase activity by Ca 2ϩ /CaM is comparable with previous results where other substrates were used (9). The mN was not phosphorylated by CCaMK even in the presence of Ca 2ϩ and CaM (Fig. 5C). The mC was phosphorylated by CCaMK, which agrees with the result of the yeast twohybrid screening, since the original cDNA isolated from the screening codes for the C-terminal region (204 -447) of LlEF-1␣1.
Identification of the Phosphorylation Site for CCaMK in LlEF-1␣1-CCaMK is similar to the mammalian multifunctional CaMKII in terms of the sequences in their kinase and CaM-binding domains and the regulation of their substrate phosphorylation (7,9). The phosphorylation sites of CCaMK in LlEF-1␣1 were predicted based on the consensus sequence of the minimal motif, RXX(S/T), which is present in majority of the phosphorylation site sequences of the CaMKII protein substrates (5, 21). There are only two putative phosphorylation sites in LlEF-1␣1 (Fig. 1). One is Thr-72 in the N terminus.
Since the mN containing Thr-72 was not phosphorylated by CCaMK (Fig. 5C), obviously this site is not the phosphorylation site for CCaMK. The other phosphorylation site is Thr-257 in the putative tRNA-binding region ( Figs. 1 and 2). Therefore, Thr-257 in the mC was mutated into Ala to investigate whether Thr-257 is a real phosphorylation site for CCaMK. The result revealed that the mC 257A was no longer phosphorylated by CCaMK (Fig. 6), indicating that Thr-257 is indeed the phosphorylation site for CCaMK. Taken together, it is clear that Thr-257 is the only phosphorylation site for CCaMK, which is located in the putative tRNA-binding domain of LlEF-1␣1.
Ca 2ϩ -dependent and CaM-independent Phosphorylation of LlEF-1␣1 by CRPK2-In plants, it has been reported that EF-1␣ can be phosphorylated by CDPK (22). To investigate whether LlEF-1␣1 can also be phosphorylated by CDPK, one of CDPK isoforms (CRPK2) isolated from corn root tips was expressed and purified from E. coli. Although CCaMK and CRPK2 show overall structural resemblance (Fig. 7A), there is no significant sequence similarity outside of their kinase domains. CRPK2 is a typical form of CDPK and has extensive and significant homology to other CDPKs (16). The amino acid sequence of CRPK2 shares 75% similarity and 65% identity with the prototype of CDPK, soybean CDPK␣, along their three functional domains (kinase domain, junction domain, and CaM-like domain) (11).
Like other CDPKs, CRPK2 showed Ca 2ϩ -dependent and CaM-independent kinase activity (Fig. 7B). There was no kinase activity in the absence of Ca 2ϩ , and its kinase activity was not stimulated by CaM. CRPK2 phosphorylated both the mN and the mC in a Ca 2ϩ -dependent and CaM-independent manner. In fact, CaM slightly decreased the phosphorylation of the mN by CRPK2. The reason may be that CaM masked some of CRPK2 phosphorylation sites by binding to the mN, since CaM binds to EF-1␣ in the presence of Ca 2ϩ (Refs. 19 and 20; Fig. 4).
In an attempt to determine whether CCaMK and CRPK2 phosphorylate the same site, the mC 257A , which was not phosphorylated by CCaMK (Fig. 6), was tested for CRPK2 phosphorylation. Fig. 7C showed that CRPK2 phosphorylated both the mC and the mC 257A to a similar extent, suggesting that Thr-257 may not be the phosphorylation site for CRPK2. CDPKs were reported to preferentially phosphorylate serine and threonine residues in the four-residue motifs of both RXX(S/T) and KXX(S/T) (12). Besides the two RXXT motifs, there are also a number of KXX(S/T) motifs in LlEF-1␣1 (Fig. 1). The mN and the mC cover the full-length LlEF-1␣1 and share a stretch of overlapping sequence (amino acids 204 -221; Fig. 5A). Since CRPK2 phosphorylated both the mN and the mC (Fig. 7B), we studied whether CRPK2 phosphorylated only sites in this overlapping region. There is indeed a threonine (Thr-215) residue located in the motif KXXT in this region. Thr-215 was mutated to Ala in both the mN and the mC. However, both of the site-directed mutants were still phosphorylated by CRPK2 (data not shown). These results indicate that CRPK2 phosphorylates multiple sites of LlEF-1␣1 in a Ca 2ϩ -dependent and CaM-independent manner, and its phosphorylation sites are different from that for CCaMK.
Binding of CCaMK to LlEF-1␣1-To test whether there is direct association between CCaMK and LlEF-1␣1, in vitro binding assays were carried out, and the results are shown in Fig. 8. CCaMK did not bind to glutathione-Sepharose beads when incubated with GST in the presence of EGTA, Ca 2ϩ , or Ca 2ϩ plus CaM, suggesting that CCaMK did not bind to glutathione-Sepharose beads and GST. The protein that migrated at a similar molecular mass as CCaMK (ϳ58 kDa) was only observed when CCaMK was incubated with the mN or the mC in the absence of Ca 2ϩ . In the presence of Ca 2ϩ or Ca 2ϩ plus CaM, this protein band disappeared, indicating that its association with the mN and the mC was disrupted. To examine whether the protein bands were CCaMK, a similar experiment was carried out, and the proteins were blotted onto polyvinylidene difluoride membrane for an immunoblot assay with the anti-CCaMK antibody. The result showed that the proteins immunologically reacted with the anti-CCaMK antibody and were indeed CCaMK. These results suggest that CCaMK binds to LlEF-1␣1 in a Ca 2ϩ -independent manner, and Ca 2ϩ disrupts their direct and physical interaction.
Association between CCaMK and EF-1␣ in Lily Anthers-All of the in vitro binding assays mentioned above were carried out using the N-terminal and C-terminal deletion mutants of LlEF-1␣1. To examine whether CCaMK binds to intact EF-1␣ in lily anthers, we performed immunoprecipitation assays using the protein extract from lily anthers. The affinity-purified anti-CCaMK antibody was used for immunoprecipitation to investigate whether a protein complex containing CCaMK and EF-1␣ exists in lily anthers. The result showed that EF-1␣ was coimmunoprecipitated with CCaMK in the presence of EGTA in the extraction buffer (Fig. 9). In the presence of Ca 2ϩ , no EF-1␣ was detected in the immunoprecipitates. When the anti-CCaMK antibody alone or the protein extract alone was incu- bated with protein G-agarose beads in the control assays, there was no signal detected by the monoclonal anti-EF-1␣ antibody. This experiment demonstrates that CCaMK binds to EF-1␣ in lily anthers in a Ca 2ϩ -independent manner. DISCUSSION In this study, we identified a CCaMK-interacting protein using the yeast two-hybrid interaction screening method. Sequence comparison revealed that this protein (LlEF-1␣1) is a homolog of EF-1␣. It is well known that the eukaryotic EF-1␣ is an essential component of polypeptide synthesis complex and promotes the GTP-dependent binding of aminoacyl-tRNA to the A-site on ribosomes during the elongation phase of translation in the cytoplasm (23). In eukaryotes, EF-1␣ is a ubiquitous and highly conserved protein. It is one of the most abundant proteins and is normally present in excessive molar ratios compared with other essential components of the translational machinery. Growing evidence indicates that EF-1␣ is a multifunctional protein and that it is involved in other cellular processes besides protein synthesis (23)(24)(25). An EF-1␣ from tobacco cells was reported to have similar properties to vitronectin, an adhesion protein in animals (26). In carrot cells, an EF-1␣ homolog was identified as an activator of a plasma membrane phosphatidylinositol 4-kinase (22). EF-1␣ has been demonstrated to bind to actin filaments and microtubules and affect their dynamics both in vitro and in vivo (19,24,(27)(28)(29)(30). These lines of evidence indicate that EF-1␣ possesses other biological functions beyond its activity in protein synthesis.
In both plants and protozoan parasites, the interaction between CaM and EF-1␣ has been demonstrated, and their association depends on the presence of Ca 2ϩ (19,20). Since CaM is an important receptor for intracellular Ca 2ϩ in regulating cytoskeletal functions (2), the finding of the Ca 2ϩ -dependent association of CaM and EF-1␣ led to speculation that this might be one of the mechanisms mediating the dynamics and functions of cytoskeleton by Ca 2ϩ /CaM. The significance of interaction between Ca 2ϩ /CaM and EF-1␣ has been implicated in several studies. It has been reported that the ability of EF-1␣ to bundle microtubules and F-actin was regulated by CaM in the addition of Ca 2ϩ (24,31,32). Considering the multipotentiality of EF-1␣, the connection between Ca 2ϩ , CaM, and EF-1␣ may involve not only the dynamics of cytoskeleton but also other cellular and biochemical processes. EF-1␣ undergoes three different post-translational modifications, i.e. methylation, formation of glycerylphosphoryl-ethanolamine, and phosphorylation (23,33). However, the effects of these post-translational modifications on its translational activity remain unclear. Protein phosphorylation is one of the major mechanisms in regulating cellular and biochemical events. It is also highly involved in protein synthesis through phosphorylation of various components of the translational machinery including EF-1␣ (33). EF-1␣ was reported to be phosphorylated by protein kinase C, S6 kinase, and plant CDPK (22,34,35). Phosphorylation of EF-1␣ would facilitate its interaction with aminoacyl-tRNA and decrease its binding to ribosomes, and dephosphorylation is required for the transfer of aminoacyl-tRNA to 80 S ribosomes (33). Phosphorylation of EF-1 by multipotential S6 kinase resulted in stimulation of EF-1 activity (35). However, an in vitro assay indicated that phosphorylation of isolated EF-1␣ by protein kinase C did not affect its function as an elongation factor (34). Instead of its effect on protein synthesis, phosphorylation of a carrot EF-1␣ was shown to be pivotal to its function as an activator of phosphatidylinositol 4-kinase (22). These studies indicate that phosphorylation of EF-1␣ may not necessarily regulate its activity only in protein synthesis and may be involved in mediating its other functions also.
In this report, we demonstrated that a homolog of EF-1␣ was phosphorylated by a plant Ca 2ϩ /CaM-dependent protein kinase in a Ca 2ϩ /CaM-dependent manner. The only phosphorylation site for CCaMK is located in the putative tRNA-binding region. This threonine residue is conserved among all EF-1␣ sequences available in eukaryotes. The functional significance of the phosphorylation and the localization of the phosphorylation site in the tRNA-binding region are currently not known. It is unclear whether the same site can be phosphorylated by the mammalian CaMKII or other protein kinases. We also showed that LlEF-1␣1 was phosphorylated by a CDPK in a Ca 2ϩ -dependent but CaM-independent manner, which is consistent with the finding of a previous study (22). The CDPK phosphorylates multiple sites in LlEF-1␣1, and these sites differ from the phosphorylation site of CCaMK. This suggests that these two kinases have some structural similarities, yet they differ in their regulation of kinase activity. Phosphorylation of EF-1␣ by CDPK was reported to be crucial for its function as an activator of phosphatidylinositol 4-kinase (22). Since CCaMK and CDPK phosphorylate different sites in EF-1␣, it is reasonable to speculate that their phosphorylation may regulate different aspects of its biological functions.
In vitro binding experiments revealed that CCaMK binds to EF-1␣ in a Ca 2ϩ -independent manner, and the addition of Ca 2ϩ disrupts their association. In eukaryotic cells, a large family of protein kinases regulate a variety of cellular processes. How these kinases coordinate their functions is a very challenging and fundamental area in signal transduction. In recent years, progress in this area has revealed that one of the mechanisms for accomplishing the coordination is regulation of their subcellular localization through association with their anchoring proteins (36 -38). Association between CCaMK and EF-1␣ may provide a mechanism in the regulation of the kinase activity of CCaMK by localizing it into specific cellular compartments. This hypothesis is supported by the fact that EF-1␣ is associated with many proteins, including the components of the translational machinery, microtubules, and actin filaments (19,23,27,28,39,40). Both CCaMK and EF-1␣ are highly expressed in similar tissues in lily (Fig. 3). It is very intriguing that targeting CCaMK into cytoskeleton or translational machinery through EF-1␣ may be one of the mechanisms to localize CCaMK into the neighborhood of its substrates. The results revealed that CCaMK phosphorylates EF-1␣ in a Ca 2ϩ /CaM-dependent manner but binds to it in a Ca 2ϩ -independent manner. We propose that CCaMK may be localized in cells through direct association with EF-1␣ in a Ca 2ϩ -independent manner. The transient kinase/substrate interaction of CCaMK with EF-1␣ requires Ca 2ϩ /CaM, while EF-1␣ serves as a substrate for CCaMK. It can be envisioned that CCaMK is compartmentalized into the subcellular domains, where its substrates exist, by association with EF-1␣ at the resting level of Ca 2ϩ in cells. At this stage in the process, CCaMK is inactive. When Ca 2ϩ concentration increases in cells due to external or internal signals, CCaMK dissociates from EF-1␣ and subsequently, upon binding Ca 2ϩ /CaM, becomes active to phosphorylate its substrates such as EF-1␣ or proteins near EF-1␣. To some extent, this possible mechanism is analogous to the regulation of another second messenger-regulated protein kinase, cAMP-dependent protein kinase, where activation of the kinase is achieved through the release of the catalytic subunits from the inactive holoenzyme by cAMP (3). Various binding proteins for several protein kinases have been identified. Their interactions are mediated by either kinase domains or other regions and are subject to different types of regulation (41)(42)(43)(44)(45). It remains to be investigated how CCaMK mediates its direct association with EF-1␣ and how Ca 2ϩ disrupts their binding. Future studies will aim at illustrating the functional significance of the phosphorylation of EF-1␣ by CCaMK and its possible roles in tRNA binding, protein synthesis, and/or other cellular processes.