Cloning and expression of cDNA encoding protein synthesis elongation factor-2 kinase.

A cDNA from rat skeletal muscle encoding calcium/calmodulin-dependent eukaryotic elongation factor-2 kinase (eEF-2K) has been cloned and sequenced, and the amino acid sequence of the protein has been deduced. The kinase is composed of 724 amino acids and has a predicted molecular mass of 81,499 Da. The cDNA was judged to be full-length, as the protein, expressed in rabbit reticulocyte lysate or wheat germ extract, migrated upon SDS-PAGE with the same apparent molecular weight as the purified kinase and possessed eEF-2K activity. eEF-2K contains all of the 12 catalytic subdomains present in the majority of protein kinases, but they are atypical and display only limited homology with other kinases. A putative calmodulin-binding domain is present C-terminal to the catalytic domain as is a putative pseudosubstrate sequence. Two antipeptide antibodies raised against sequences derived from a partial rabbit cDNA clone, cross-reacted with purified eEF-2K, and one also immunoprecipitated eEF-2K activity from cell extracts. Northern blot analysis demonstrated that eEF-2K mRNA is expressed in a number of different tissues and that it may exist in multiple forms.

A cDNA from rat skeletal muscle encoding calcium/ calmodulin-dependent eukaryotic elongation factor-2 kinase (eEF-2K) has been cloned and sequenced, and the amino acid sequence of the protein has been deduced. The kinase is composed of 724 amino acids and has a predicted molecular mass of 81,499 Da. The cDNA was judged to be full-length, as the protein, expressed in rabbit reticulocyte lysate or wheat germ extract, migrated upon SDS-PAGE with the same apparent molecular weight as the purified kinase and possessed eEF-2K activity. eEF-2K contains all of the 12 catalytic subdomains present in the majority of protein kinases, but they are atypical and display only limited homology with other kinases. A putative calmodulin-binding domain is present C-terminal to the catalytic domain as is a putative pseudosubstrate sequence. Two antipeptide antibodies raised against sequences derived from a partial rabbit cDNA clone, cross-reacted with purified eEF-2K, and one also immunoprecipitated eEF-2K activity from cell extracts. Northern blot analysis demonstrated that eEF-2K mRNA is expressed in a number of different tissues and that it may exist in multiple forms.
Elongation factor-2 kinase was initially identified as a calcium/calmodulin (Ca 2ϩ /CaM) 1 -dependent protein kinase (CaM-PK) that phosphorylated an abundant substrate of 100 kDa in mammalian cells (Palfrey, 1983;Nairn et al., 1985). It was termed CaM-PK III. Its substrate was subsequently shown to be eukaryotic elongation factor-2 (eEF-2) Ryazanov et al., 1988;Koizumi et al., 1989;Ryazanov 1987), one of two protein factors required for peptide chain elongation in higher eukaryotic cells (Proud, 1994). eEF-2 mediates the translocation step of peptide chain elongation, during which the ribosome moves by one codon's length relative to the mRNA and the peptidyl-tRNA migrates from the A-to the P-site in the ribosome. eEF-2 is a GTP-binding protein and is only active when bound to GTP (see Proud (1994)). In vitro, eEF-2K phosphorylates two adjacent threonine residues, Thr 56 and Thr 58 , in the N terminus of eEF-2 Price et al., 1991;Ovchinnikov et al., 1990), apparently in an ordered manner with phosphorylation of Thr 56 preceding, and being required for, phosphorylation of Thr 58 (Price et al., 1991;Redpath et al., 1993). Phosphorylation of Thr 56 alone (or of both) causes complete inactivation of eEF-2 (Redpath et al. (1993); see also Nairn and Palfrey (1987), Ryazanov et al. (1988), Ryazanov and Davydova (1989), and Redpath and Proud (1989)) by abolishing its ability to interact with the ribosome (Carlberg et al., 1990). eEF-2K has been tested against a variety of protein substrates, but none was phosphorylated at a significant rate compared with eEF-2 itself (Nairn et al., 1985). eEF-2 itself is not significantly phosphorylated by any other protein kinase.
eEF-2K has been purified to homogeneity from rabbit reticulocytes and rat pancreas (Redpath and Proud, 1993a;Mitsui et al., 1993). It is a monomer of about 100 kDa as judged by SDS-PAGE and binds CaM tightly. The kinase is normally entirely dependent on Ca 2ϩ /CaM for activity. The enzyme undergoes Ca 2ϩ /CaM-dependent intramolecular autophosphorylation, and this results in it becoming partially Ca 2ϩ /CaMindependent (Redpath and Proud, 1993a;Mitsui et al., 1993), a property shared with the type II (multifunctional) CaM-PK (Hanson and Schulman, 1992a). Autophosphorylation occurs on multiple sites involving both Ser and Thr residues. eEF-2K is also subject to phosphorylation, at least in vitro, by cAMPdependent protein kinase (Redpath and Proud, 1993b;Mitsui et al., 1993). This also renders the enzyme partially Ca 2ϩ /CaMindependent (Redpath and Proud, 1993b). Thus, eEF-2K can potentially be activated by two different intracellular messenger systems, calcium ions and cAMP, and therefore has the potential to mediate the effects of either on the translational machinery of the cell, resulting in each case in inhibition of peptide chain elongation (see Redpath and Proud (1994) for a recent review). eEF-2K is likely to play a key role in the control of peptide chain elongation by hormones and growth factors. Indeed, it has been shown that agents that raise intracellular calcium ion concentrations lead to increased phosphorylation of eEF-2 (Hincke and Nairn, 1992;Demolle et al., 1990;Mackie et al., 1989).
We now report the cloning and expression of a cDNA encoding rat eEF-2K. Analysis reveals that it is a novel protein kinase, only rather distantly related to other CaM-PKs.

EXPERIMENTAL PROCEDURES
Materials-Chemicals, biochemicals, and other items were obtained as described previously (Redpath and Proud, 1993a). All radioisotopes were from DuPont NEN except [␥-32 P]ATP, which was from Amersham (Buckinghamshire, United Kingdom) as were the Multiprime DNAlabeling kit and ECL immunodetection kit. The Sequenase version 2.0 DNA sequencing kit was obtained from Upstate Biotechnology, Inc./Life Technologies, Inc. cDNA libraries were obtained from Clontech and Stratagene. Restriction enzymes and general molecular biology reagents were from Boehringer Mannheim. Nitrocellulose-coated nylon membranes for library screening were obtained from Sartorius. T3/T7coupled TnT expression kits were obtained from Promega, and the pCRII TA cloning kit was from Invitrogen.
Sequencing of Peptides Derived from eEF-2K-Rabbit reticulocyte eEF-2K (0.1 mg) (low molecular mass form (95 kDa)) (Redpath and Proud, 1993a) was subjected to electrophoresis on a 15% SDS-polyacrylamide gel, and the gel chips containing the kinase were excised. CNBr digestion of the gel chips, Tricine gel electrophoresis, and Western blotting were carried out as described (Price et al., 1994). The membrane was stained and destained, and regions containing stained peptides were excised and subjected to peptide sequencing by solid phase automated Edman degradation using an Applied Biosystems 477A sequencer.
Cloning of eEF-2K cDNA-From the obtained peptide sequences, various oligonucleotides were designed for use in PCR. Two of them had the sequences ATGTGGAT(T/C/A)AT(T/C/A)GA(A/G)(C/T)T(T/C/A/G)-AC (sense encoding MWIIEL(T)) and GT(T/C)CT(G/A)GT(G/A)(G/A)A-(T/C/A/G)CT(G/A)TTGGT (antisense corresponding QDHLDN(Q)); the parentheses indicate that only the first two bases of the triplets that code for these amino acids were used). PCR conditions were 1 min at 94°C, 1 min at 55°C, and 1.5 min at 72°C for 35 cycles followed by 72°C for 10 min. PCR products were gel-purified before cloning into pUC18 or pCRII. cDNA libraries were screened according to the manufacturer's instructions. Membranes were probed with cDNA (labeled with [␣-32 P]dCTP by random priming) at 42°C overnight in 50% formamide, 5 ϫ SSPE, 5 ϫ Denhardt's solution, 0.1% SDS, and 0.1 mg⅐ml Ϫ1 sheared salmon sperm DNA. 1 ϫ SSPE is 0.5 mM EDTA, 0.15 M NaCl, 10 mM NaH 2 PO 4 . 1 ϫ Denhardt's solution is 0.1% bovine serum albumin, 0.1% Ficoll 400, 0.15 polyvinylpyrrolidine. After incubation, the membranes were washed twice with 2 ϫ SSC, 0.05% SDS at 55°C. Positive clones were purified by two further rounds of screening. The membranes from the tertiary screen were washed twice with 2 ϫ SSC, 0.05% SDS at room temperature and twice with 1 ϫ SSC, 0.1% SDS at 55°C. Insert cDNAs were excised into pBluescript II SK (Stratagene) as per the manufacturer's instructions. The largest of the cDNA clones were chosen for DNA sequencing. Manual sequencing in both directions was carried out using the Sanger dideoxy termination method using a Sequenase version 2.0 kit, and parts of the sequence were verified by automated sequencing. Sequence alignments were made using the LA-SERGENE suite of programs (from DNASTAR).
Assay of eEF-2K-eEF-2K was assayed in buffer containing 50 mM Mops (pH 7.2), 11 mM MgCl 2 , 0.2 mM ATP, 100 C⅐ml Ϫ1 [␥-32 P]ATP, 5 mM dithiothreitol, 10% glycerol, 0.5 mM benzamidine, 1 g⅐ml Ϫ1 each of pepstatin, antipain, and leupeptin, 0.4 mM EGTA, 2 mM HEDTA, and 150 pmol of eEF-2 in a final volume of 30 l for 30 min at 30°C. CaCl 2 (0.67 mM) and CaM (2 g⅐ml Ϫ1 ) were added as required. Samples were then analyzed by SDS-PAGE and autoradiography. Immunoprecipitation of eEF-2K was performed using 5 mg of protein A-Sepharose and a 1:25 dilution of antiserum in a final volume of 0.2 ml. The mixture was tumbled at 4°C for 2 h followed by extensive washing with buffer (50 mM Hepes, pH 7.5, 1 mM sodium orthovanadate, 10 mM sodium fluoride, 30 mM sodium pyrophosphate, 2.5 mM benzamidine, 1 g⅐ml Ϫ1 each of pepstatin, antipain, and leupeptin, 10 mM EDTA, and 1% Triton X-100) and then with a 1:10 dilution of the buffer before resuspension of the beads in 1:10 diluted buffer. The resuspended beads were used in the assay for eEF-2K as above.
In Vitro Expression of eEF-2K cDNA-cDNA in pBluescript was expressed in reticulocyte lysate and wheat germ extract in the presence of [ 35 S]methionine/cysteine using the T3/T7-coupled TnT systems according to the manufacturer's instructions.
Preparation of Antipeptide Antibodies-Two peptides, CLRLSENS-GDENMSDVT (P1) and CDLSPREQDAVNQNTKL (P2), were synthesized and linked to keyhole limpet hemocyanin (van Regenmortel et al., 1988). Antibodies were raised in New Zealand White rabbits. The peptides were sequences that were deduced from the sequence of the 663-bp PCR product, amplified from rabbit cDNA (see "Results and Discussion").
Northern Blot Analysis-Poly(A) ϩ RNA was prepared from 0.5 mg of total RNA from various rat tissues as described (Craddock et al., 1995). This was separated on a formaldehyde-containing gel and blotted onto Hybond-N membrane. eEF-2K mRNA was detected using a probe derived from an EcoRI restriction product from eEF-2K cDNA corresponding to bases 589-2681. Blots were incubated overnight at 42°C in 50% formamide, 5 ϫ SSPE, 5 ϫ Denhardt's solution, 0.5% SDS, and 0.2 mg⅐ml Ϫ1 sheared salmon sperm DNA followed by washing with 2 ϫ SSC, 0.01% SDS three times at room temperature and with 1 ϫ SSC, 0.1% SDS three times at 65°C. The blot was stripped by washing extensively in 0.01 ϫ SSC, 1% SDS at 90°C and then reprobed for ␤-actin.
Labeling of Proteins with [ 14 C]Fluorosulfonylbenzoyladenosine-Labeling was performed in 20-l incubations containing 5 mM MgCl 2 , 50 M CaCl 2 , 2 g⅐ml Ϫ1 CaM, 0.04 Ci of FSBA. Where required, 200 M ATP was also added. As FSBA binds at the ATP-binding site, ATP should act as a competitive inhibitor of labeling. Incubations were for 25 min at 30°C.

Characterization of the Low and High Molecular Weight
Forms of eEF-2K-As previously reported (Redpath and Proud, 1993a;Mitsui et al., 1993), purification of eEF-2K often results in the co-isolation of two polypeptides, differing in apparent molecular mass by about 10 kDa. We previously concluded that the two forms are likely to be closely related to each other since they both possess eEF-2K activity and become phosphorylated in the presence of ATP and Ca 2ϩ /CaM (Redpath and Proud, 1993a;Mitsui et al., 1993). Further evidence presented here provides compelling evidence for this and indicates that the low M r form is derived from the high M r form. First, both forms are labeled by the ATP analogue [ 14 C]FSBA ( Fig. 1), suggesting that both possess an ATP-binding site. Furthermore, no other proteins are labeled in these highly purified kinase preparations. This demonstrates that the 100-kDa proteins do not associate or co-purify with another protein that binds ATP and might thus possess eEF-2 kinase activity. Second, antisera raised against two peptide sequences deduced from a partial eEF-2K cDNA clone (see below) cross-react with both forms of the kinase (Fig. 2a). Furthermore, Fig. 2b also shows that one of the antipeptide antibodies prepared (anti-P1) immunoprecipitated eEF-2K activity from cell extracts. Finally, as described below, our cloning strategy, starting from amino acid sequence data for the 95-kDa species, led us to isolate cDNA clones encoding a polypeptide of identical size to the larger species, and this clone contained all of the amino acid sequences obtained from peptides derived from the 95-kDa protein. Although it cannot be completely ruled out that the two forms of eEF-2 kinase could result from two alleles or two closely related genes, it seems highly likely that the 95-kDa protein is produced by proteolytic nicking of the larger species, particularly considering that different preparations of kinase differ in the relative amounts of the two forms of the kinase (see Redpath and Proud (1993)). It seems unlikely that the two forms result from alternative initiation sites, since expression of the cDNA in rabbit reticulocyte lysate (see below) results in the synthesis of only the higher molecular weight form of the kinase. In this context it should be noted that while the larger polypeptide has a blocked N terminus, the smaller one's amino terminus was not blocked (data not shown), suggesting that the proteolytic modification includes removal of N-terminal peptides.
Peptide Sequence Data for eEF-2K-Digestion of the 95-kDa species of eEF-2K (Redpath and Proud, 1993a) with CNBr followed by Tricine gel electrophoresis of the resulting peptides yielded eight distinct stained peptides ranging in size from 20 to 2.5 kDa (data not shown). Amino acid sequence analysis of the individual peptides gave unambiguous sequences of four peptides ( Table I). The other bands from the gel either contained more than one peptide or appeared to be partial digestion products.
Isolation of cDNAs Encoding eEF-2K-The peptide sequences obtained were used to design redundant oligonucleotide primers. Use of two of the primers whose sequences are given under "Experimental Procedures" in the PCR amplified a single product (663 bp) when rabbit reticulocyte cDNA was used as the template. This was sequenced and shown to contain sequences corresponding to peptide B as well as peptides C and A, from which the PCR primers were derived. The 663-bp cDNA was used to screen an oligo(dT)-primed rabbit liver cDNA library in gt11 (Clontech). One positive clone was obtained from this screen with an insert size of approximately 1.3 kb. Sequencing of this showed that it contained the entire sequence of the 663-bp cDNA apart from its 5Ј-terminal 76 nucleotides. Screening of a rat liver cDNA library (Stratagene) and a rabbit reticulocyte cDNA library (provided by J.-J. Chen, Massachusetts Institute of Technology) with the 663-bp cDNA did not result in the identification of any positive clones. The 1.3-kb cDNA was then used as a probe to screen a rat skeletal muscle cDNA library in Unizap (Stratagene). Approximately 53 positive plaques were obtained from a screen of about 1.5 ϫ 10 6 plaque-forming units. The cDNA and deduced amino acid sequences of the largest clone are shown in Fig. 3A. The clone consists of 4800 nucleotides having a 290-nucleotide 5Ј-untranslated region (UTR), a 2175-nucleotide coding region (including the termination codon), encoding 724 amino acids, and a 2335-nucleotide 3Ј-UTR. The encoded eEF-2K has a predicted molecular mass of 81,499 Da and an isoelectric point of 5.0. The predicted molecular weight is somewhat lower than the molecular weight of the kinase as judged by SDS-PAGE (100 kDa) (Redpath and Proud, 1993a). Apparent anomalies of this kind, however, are not uncommon. For example, the eukaryotic initiation factor-2␣ kinase (heme-controlled repressor) migrates as a 90-kDa protein upon SDS-PAGE, but it has a predicted molecular mass of only 70 kDa (Chen et al., 1991). It is possible that the highly acidic N-terminal region of eEF-2K is responsible for its anomalous behavior upon SDS-PAGE; 31% of the first 35 residues are either glutamate or aspartate.
All four peptides obtained by sequencing could be identified in the cloned rat sequence, although peptide A was only a poor match that is likely to reflect differences between the sequences of the rat and rabbit proteins.
The sequence around the initiation codon (nucleotides 291-293) fits very well with the consensus sequence (GCCGCC-(A/G)CCAUGG) (Kozak, 1989), the most important nucleotides FIG. 2. Cross-reaction of antipeptide antibodies with eEF-2K. a, purified eEF-2K (0.5 g) was subjected to SDS-PAGE and transferred to membranes, which were then probed with antipeptide 1 antiserum (panel A) or antipeptide 2 antiserum (panel B) at 1:1000 dilution. Both the low (lane 1) and the high (lane 2) molecular weight forms of the kinase were used (Redpath and Proud, 1993a). Membranes probed with nonimmune serum or with antiserum raised against a peptide from p70 S6k gave no signal (not shown). The anti-p70 S6k antibody was used as a control for an antiserum that had been raised against a peptidekeyhole limpet hemeocyanin complex. b, antibodies were tested for their ability to immunoprecipitate eEF-2K activity from extracts prepared from serum-starved CHO.T cells (Welsh et al., 1994). Cell extract (310 g) was immunoprecipitated with antiserum against peptide 1 (lane 1), peptide 2 (lane 2), nonimmune serum (lane 3), or antiserum against p70 S6k (lane 4) as described under "Experimental Procedures." The immunoprecipitate was resuspended with 50 l of buffer, and 9 l were used to phosphorylate eEF-2 in a 5-min incubation that contained Ca 2ϩ /CaM. The position of eEF-2 is indicated. The labeled protein that migrated slightly more slowly than eEF-2 is likely to be eEF-2K. The figure is an autoradiograph.

TABLE I Partial sequence of CNBr fragments of eEF-2K
Sequences were obtained from rabbit reticulocyte eEF-2K as described under ''Experimental Procedures. X indicates an unidentified amino acid. Single letter codes are used. The parentheses indicate that methionine is assumed to be immediately N-terminal to the sequenced peptides since they were derived by CNBr cleavage. being the guanosine at ϩ4 and the purine at Ϫ3 (shown in boldface type). In-frame termination codons are present upstream of the initiation codon, suggesting that this is indeed the authentic start site.
Features of the Sequence of eEF-2K-The domain structure of eEF-2K is depicted in Fig. 3B. The region including residues 593-609 (KTKGFDYLLKAAEAGDR) in eEF-2K may represent part of the CaM-binding domain of the protein (Fig. 3A). Such domains are short stretches that are variable in sequence but contain a number of basic and hydrophobic amino acids that are predicted to form ␣-helices (see O'Neil and DeGrado (1990)). This region of eEF-2K is indeed predicted to have ␣-helical structure by the methods of Chou-Fasman, Deleage and Roux, and Garnier-Robson (using the LASERGENE programs, not shown). Fig. 3C shows a right-handed ␣-helical wheel model of residues 595-612. The residues form a hydrophobic face and a basic face, consistent with the requirements of a CaM-binding domain helices (O'Neil and DeGrado, 1990).
Some protein kinases contain so-called autoinhibitory domains, which function to maintain the kinase in an inactive state in the absence of an activator, presumably by acting as a pseudosubstrate that interacts with and occludes the catalytic FIG. 3. Sequence of rat skeletal muscle eEF-2K cDNA and deduced amino acid sequence. A, the underlined sequences indicate the four peptides obtained by peptide sequencing. The discrepancies between the peptide sequences (from the rabbit protein) and the sequence deduced from the rat cDNA are likely to be due to species differences. The numbers on the left refer to the nucleotide sequence, whereas those on the right refer to the amino acid sequence. The star denotes the termination codon. Also indicated is the putative calmodulin-binding domain (double underlined). The putative pseudosubstrate sequence is located between residues 609 and 626 and is boxed. B, the domain structure of eEF-2K is shown. RD indicates the regulatory domain, which contains the putative CaMbinding and autoinhibitory domains. The figure is drawn to scale. C, residues 595-612 of eEF-2K are presented as a righthanded ␣-helical wheel. The hydrophobic and basic faces are indicated. D, the putative pseudosubstrate sequence in eEF-2K is compared with the region containing the phosphorylation sites in eEF-2. The main phosphorylation site in eEF-2 is Thr 56 . The solid lines indicate sequence identity, whereas the dashed lines indicate residues of similar character. site, thus excluding the substrate. Examples of such kinases include CaM-PKs I (Yokokura et al., 1995), II , and IV (Tokumitsu et al., 1994), myosin light chain kinase (Kreuger, et al., 1995) and p70 S6k (see Banerjee et al. (1990)). In the case of the CaM-PKs, the binding of Ca 2ϩ /CaM activates the kinase by removing the autoinhibitory domain from the active site. In these protein kinases, the sequence within the autoinhibitory domain is often similar to that surrounding the phosphorylation site in the substrate. Residues 609 -626 in eEF-2K are rather similar to the region surrounding the main phosphorylation site in eEF-2 (Thr 56 ). Alignment of residues 609 -626 of eEF-2K with residues 41-58 of eEF-2 (Fig. 3D) shows that these regions have 28% identity and 67% homology. The equivalent of the main phosphorylated residue in eEF-2 (Thr 56 ) is Asn 624 in eEF-2K.
Examination of the amino acid sequence of eEF-2K reveals a surprisingly low homology with other serine/threonine protein kinases. The vast majority of kinases whose sequences are known possess 12 catalytic subdomains that contain a number of highly conserved residues, most of which are involved in binding Mg 2ϩ /ATP (see Hanks and Hunter (1995)). Fig. 4 shows the conserved consensus sequence of the catalytic domain of cAMP-dependent protein kinase (PKA) compared with the subdomains of eEF-2K. PKA fits well with the consensus sequence, as do the other CaM-PKs (Hanks and Hunter, 1995).
In subdomain I, eEF-2K contains the GXGXXG motif, which is conserved in ATP-binding proteins but lacks the nearly invariant valine two residues downstream of this sequence, which contributes to a hydrophobic pocket. Serine 53 in PKA makes a hydrogen bond with ATP ␤-phosphate oxygens. The equivalent in eEF-2K (Asn 298 ) should also be able to make a hydrogen bond. Although eEF-2K has a usually conserved hydrophobic residue (Phe) immediately before the first glycine, it does not have the usually conserved hydrophobic residue seven residues before the first glycine (this is Thr in eEF-2K). The next lysine in the eEF-2K sequence is at position 340 and is likely to be the invariant lysine in subdomain II. This residue appears to be essential for the binding of ATP, as it interacts with the ␣and ␤-phosphates (Hanks and Hunter, 1995). Alanine 70 (in PKA) contributes to the hydrophobic adenine ring pocket, but this is replaced in eEF-2K by Ser 338 . Subdomain V is important in linking the small and large lobes of the catalytic domain in PKA. Residues important in this appear to be Glu 121 , Val 123 , and Glu 126 . These are generally poorly conserved in eEF-2K. However, three residues that contribute to the hydrophobic pocket in PKA (Met 120 , Tyr 122 , and Val 123 ) are conservatively replaced by Pro 412 , Phe 414 , and Gly 415 in eEF-2K.
The most notable difference between the sequences of eEF-2K and PKA occurs in subdomain VIb. The conserved aspartate residue (Asp 166 in PKA, Asp 433 in eEF-2K) is vital in FIG. 4. Alignment of eEF-2K and PKA catalytic subdomains. The catalytic subdomains of eEF-2K and human PKA-C␣ were aligned by eye. The conserved consensus sequence as given in Hanks and Hunter (1995) is shown in the top line. Uppercase letters in the consensus indicate invariant residues, lowercase letters are nearly invariant residues, E indicates conserved nonpolar amino acids, ϩ represents conserved small residues with near neutral polarity, and * indicates conserved polar residues. The catalytic subdomains are indicated by roman numerals. Boldface type indicates those residues that are deemed to fit the consensus sequence.

FIG. 3-continued
kinase activity, since this residue seems to accept a proton from the attacking hydroxyl group. The near-invariant lysine (Lys 168 in PKA) appears to be involved in neutralizing ␥-phosphate negative charge. This is replaced by Glu 435 in eEF-2K, which obviously cannot carry out this function. Furthermore, eEF-2K does not have the highly conserved asparagine in this subdomain (Asn 171 in PKA). Subdomain VII in most protein kinases contains a highly conserved DFG sequence, which is DSG in eEF-2K. This means that subdomains VIb and VII in eEF-2K are relatively close together in the primary sequence, the more usual spacing being at least 19 residues. Asp 184 and Gly 186 (in PKA) form hydrogen bonds to stabilize a loop, and Asp 186 also chelates Mg 2ϩ that bridges the ␤and ␥-phosphates.
The highly conserved motif in subdomain VIII is PE, almost always APE or SPE. In eEF-2K this is DPE. Glu 208 (in PKA) forms an ion pair with the conserved arginine in subdomain XI, stabilizing the large lobe. Subdomain IX contains a highly conserved DXXXXG motif, within which there are almost always several hydrophobic residues. This subdomain in PKA may play a major role in peptide substrate recognition and therefore may not be expected to show a high degree of conservation with eEF-2K. This motif, however, is present in eEF-2K, although it lacks the conserved highly hydrophobic nature of this motif. This sequence in eEF-2K is, however, rather similar to the corresponding sequence in casein kinase-1␣ (DMESLG).
In summary, although the sequence of the catalytic domain of eEF-2K shows surprisingly low conservation with the consensus sequence and the other CaM-PKs, almost all of the essential residues are present in eEF-2K, with the exception of the highly conserved lysine and asparagine residues in subdomain VIb. Molecular modeling will be required to determine how the three-dimensional structure of the catalytic domain of eEF-2K compares with those of the other protein kinases.
We have previously shown that rabbit eEF-2K is a substrate for PKA (Redpath and Proud, 1993b). PKA generally phosphorylates Ser/Thr residues with basic residues N-terminal to the target residue, usually RRX(S/T) (Hanks and Hunter, 1995). There are no perfect matches to this consensus in rat eEF-2K, but there are several approximate fits, i.e. KKLS (residues 149 -152), . There are also a number of Ser/Thr-Pro motifs, which are potential phosphorylation sites for "proline-directed" kinases such as mitogen-activated protein kinase. Further work is required to determine whether eEF-2K can indeed be phosphorylated by such enzymes and whether this modulates its activity. The availability of the primary sequence of eEF-2K will be a very valuable aid in the identification of the sites in eEF-2K phosphorylated by PKA, by autophosphorylation and, where relevant, other protein kinases.
Expression of cDNA Encoding eEF-2K-The cDNA was expressed by coupled transcription/translation in the reticulocyte lysate system in the presence of 35 S-labeled amino acids. A single radiolabeled band of about 100 kDa, which was present only in the lysate supplemented with cDNA, was observed (Fig.  5). This corresponds closely to the molecular weight of the purified kinase from rabbit reticulocytes (Redpath and Proud, 1993a;Mitsui et al., 1993). Attempts were made to detect an increase in eEF-2K activity associated with expression of this protein relative to the endogenous activity present in the lysate. However, no detectable increase was apparent above the endogenous level, which is already quite substantial in reticulocyte lysates.
We therefore turned to the wheat germ system, which is reported to be devoid of endogenous eEF-2K (Smailov et al., 1993). Transcription/translation in the presence of 35 S-labeled FIG. 5. Expression of eEF-2K cDNA in the reticulocyte lysate. eEF-2K cDNA in pBluescript (1 g) was expressed in reticulocyte lysate as described under "Experimental Procedures." In a control incubation, cDNA was omitted. Samples (5 l) from the incubations were removed at the indicated times and subjected to SDS-PAGE. The gel was then autoradiographed overnight. The positions of molecular weight markers are shown.
FIG. 6. Expression of active eEF-2K in wheat germ extract. A, wheat germ lysate was incubated with or without eEF-2K cDNA (1 g) for 90 min as indicated. eEF-2K was then immunoprecipitated using antipeptide 1 antiserum. eEF-2K activity in the immunoprecipitates was then assayed in a 30-min incubation in the presence of Ca 2ϩ /CaM and [␥-32 P]ATP. The position of 32 P-labeled eEF-2 is indicated. B, eEF-2K expressed in wheat germ extract and immunoprecipitated as in A was assayed in the presence and absence of eEF-2, Ca 2ϩ , and CaM as indicated in the figure.
methionine/cysteine gave low levels of a labeled band at 100 kDa, but the efficiency of expression was much lower than in the reticulocyte lysate system. Therefore, only a low level of eEF-2K activity was detected in samples that contained the eEF-2K cDNA, which was absent from the minus cDNA control (not shown). In order to eliminate the interfering background phosphorylation due to endogenous proteins in the wheat germ lysate, we made use of anti-eEF-2K peptide antibodies. The antiserum raised against peptide 1 immunoprecipitated eEF-2K activity from the wheat germ system supplemented with the cDNA but not from an appropriate control lacking the eEF-2K cDNA (Fig. 6A). Furthermore, the kinase expressed in wheat germ extract was fully dependent on Ca 2ϩ /CaM for activity (Fig. 6B). These data demonstrate conclusively that the cDNA does indeed encode eEF-2K.
Homology of eEF-2K with Other Proteins-A search of a protein sequence data base (SwissProt) revealed no significant homology of eEF-2K with any other mammalian protein. The most highly homologous sequence (37% identity) was from Caenorhabditis elegans (Fig. 7). This sequence is a putative open reading frame in the F42A10.4 gene sequence (cosmid accession number U10414) and was obtained from the C. elegans genome-sequencing project (see Wilson et al. (1994)), and it is reported as having highest homology with ATP-binding transport proteins. Its relationship with eEF-2K is unknown, but it is possible that it is the C. elegans homologue of eEF-2K. However, the fact that the region of highest homology occurs outside the catalytic domain of eEF-2K may argue against this. The only other protein to show any significant homology with eEF-2K was Dictyostelium myosin heavy chain kinase (Gen-Bank TM accession number P42527) (not shown).
Northern Blot Analysis of eEF-2K mRNA-Probing of poly(A) ϩ RNA for eEF-2K revealed the presence of three or four labeled species that were all expressed in the tissues tested (Fig. 8). The most abundant band ran at about 2.5-3 kb, while the larger species were 4 -5 kb. This suggests that eEF-2K mRNA may exist mainly in a form considerably smaller than the cDNA cloned here. Alternatively, the long 3Ј-UTR may have been an artifact produced during the construction of the cDNA library such as the fusion of unrelated sequence to the 3Ј-UTR of eEF-2K. However, a data base search for homology to the 3Ј-UTR of eEF-2K did not reveal any identity with other sequences. In addition, PCR was used to show whether a product corresponding to the size of the 3Ј-UTR could be amplified from rat skeletal muscle cDNA. Nonredundant primers complimentary to the 3Ј end of the coding region and the 3Ј end of the 3Ј-UTR were used, and an approximately 2.4-kb cDNA was amplified (not shown). This demonstrates that eEF-2K mRNA species with very long 3Ј-UTRs do exist. Furthermore, an ap-proximately 2.6-kb cDNA clone was isolated, the 3Ј-UTR of which was only 113 bases, ended in a poly(A) tail, and was identical to the first 113 bases of the 3Ј-UTR of the clone shown in Fig. 3A. It is therefore very likely that the long 3Ј-UTR of the clone shown here results from incomplete processing and that the completely processed mRNA is about 2.6 kb, consistent with the results obtained from Northern analysis. The 2.6-kb clone also had an identical 5Ј-UTR to the clone shown here, indicating that the 5Ј-UTR is authentic and not artifactually produced during construction of the library. Finally, it should be noted that the largest clone obtained was chosen for sequence analysis. Of seven clones isolated, only two were larger than 2.6 kb, indicating that eEF-2K mRNA transcripts larger than 2.6 kb are a minor species, again consistent with the Northern analysis shown here. eEF-2K mRNA is expressed at high levels in skeletal muscle in comparison with the other tissues tested, and the relative levels of mRNA are in general agreement with the relative amount of eEF-2K activity that was measured in various rat tissues by Nairn et al. (1985). The relative eEF-2K activity in skeletal muscle, liver, heart, and cerebellum was found to be 1:0.2:0.1:0.08, respectively (Nairn et al., 1985).
General Discussion-The data presented here detail the first cloning of cDNA encoding eEF-2K and deduction of the amino acid sequence. Several pieces of evidence prove that the cloned cDNA does indeed encode eEF-2K. Most compellingly, the expressed protein has the same molecular weight as the purified kinase and possesses Ca 2ϩ /CaM-dependent eEF-2K activity. Furthermore, all of the peptide sequences obtained by sequencing of purified kinase were found in the protein sequence de- duced from the cDNA, and antibodies raised against peptide sequences within the kinase cross-react with purified eEF-2K and immunoprecipitate eEF-2K activity from cell extracts.
Perhaps the most surprising finding resulting from the determination of the primary sequence of eEF-2K is the relatively low homology of the catalytic domains with other serine/threonine protein kinases. There are other proteins that possess protein kinase activity but which have kinase domains completely unrelated to any eukaryotic or prokaryotic kinase (e.g. Bcr (Maru and Witte, 1991) and "A6" kinase (Beeler et al., 1994)), and compared with them eEF-2K is much closer to the norm. Nonetheless, it is surprising that the catalytic subdomains of eEF-2K do not more closely resemble those of the other protein kinases, particularly the CaM-PKs. Cloning of eEF-2K from other species will identify the relationship of eEF-2K from different species and which residues are highly conserved and, by implication, important for the function of eEF-2K.
The identification of regions of eEF-2K as CaM-binding and autoinhibitory domains as discussed above is at this stage speculative. Where they have been identified, the CaM-binding and autoinhibitory domains in the other CaM-PKs lie close to each other. In CaM-PKs I, II, and IV and in myosin light chain kinase, the autoinhibitory domain is proposed to be N-terminal to, but overlapping with, the CaM-binding domain (Kreuger et al., 1995;Tokumitsu et al., 1994;Hanson and Schulman, 1992b;Yokokura et al., 1995). Mutagenesis studies should help to identify which residues are important in the CaM binding and autoinhibition of eEF-2K. These studies are currently under way.