Limited Proteolysis of Yeast Elongation Factor 3 SEQUENCE AND LOCATION OF THE SUBDOMAINS*

Elongation factor 3 (EF-3) is an ATPase essential for polypeptide chain synthesis in a variety of yeasts and fungi. We used limited proteolysis to study the organization of the subdomains of EF-3. Trypsinolysis of EF-3 at 30 °C resulted in the formation of three fragments with estimated molecular masses of 90, 70, and 50 kDa. Yeast ribosomes protected EF-3 and the large fragments from further degradation. ATP exposed a new tryptic cleavage site and stabilized the 70- and 50-kDa fragments. The conformation of EF-3 as measured by fluorescence spectroscopy did not change upon ATP binding. Poly(G) stimulated proteolysis and quenched the intrinsic fluorescence of EF-3. Using gel mobility shift, we demonstrated a direct interaction between EF-3 and tRNA. Neither tRNA nor rRNA altered the tryptic cleavage pattern. The proteolytic products were sequenced by mass spectrometric analysis. EF-3 is blocked NH 2 - terminally by an acetylated serine. The 90-, 70-, and 50-kDa fragments are also blocked NH 2 -terminally, con- firming their origin. The 50-kDa fragment (Ser 2 -Lys 443 ) is the most stable domain in EF-3 with no known function. glyc-erol). Aliquots containing 2 m g of EF-3 were removed at various times, and the digestions were terminated with a 5-fold excess (w/w) of soybean trypsin inhibitor. Samples were either assayed for the ATPase activity or diluted into SDS sample buffer, heated at 90 °C for 10 min, and subjected to electrophoresis. Fluorescence Measurements— All measurements were carried out at room temperature (22 °C) using an SLM 4800C spectrofluorometer equipped with xenon arc light source. Slits of 4-nm width were used for excitation and emission. The emission spectra of EF-3 (10 m g/ml in buffer A) were recorded by excitation at 294 nm in a rectangular quartz cuvette (inner diameter, 10 mm). The observed fluorescence intensities were corrected by subtracting the appropriate blanks and are expressed in arbitrary units. acid. Gradi-ent elutions from 0% B to 90% over 40 min at a flow rate of 3 m l/min provided separation of the peptides. The eluents were introduced into a Finnigan (San Jose, CA) LCQ mass spectrometer fitted with an ESI interface. The peptide samples were introduced without any splitting, and a mass spectrometer was used to scan from 360 to 2,000 Da. The mass of the peptides containing 3 or , 3 amino acid residues could not be detected accurately by this method. For peptides greater than 2 kDa, [M 1 2H]2 1 , [M 1 3 H]3 1 , and [M 1 4H]4 1 were monitored by the instrument.

Sequence homology searches with EF-3 from different fungal sources show significant conservation in the primary and secondary structure (10). EF-3 contains two bipartite nucleotide binding domains referred to as ATP-binding cassettes or nucleotide binding sequence (NBS) motifs. These sequence motifs are usually present in the membrane-associated transporter proteins (11). The carboxyl-terminal end of EF-3 is highly charged and contains blocks of lysine residues (7)(8)(9)(10). That the primary ribosome binding site of EF-3 resides at the COOHterminal end of the protein was reported in our earlier communication (12).
We and others have used limited proteolysis as a tool to define the domain structure of EF-3 (12)(13)(14). Our results indicated that the 116-kDa protein is organized into two distinct functional domains. The NH 2 -terminally derived 90-kDa domain is involved in ATP binding/hydrolysis, and the COOHterminally derived 30-kDa domain is essential for the ribosomedependent functions (12). In the present paper, we report the effects of ligands on the proteolytic cleavage and on the domain structure of EF-3.

MATERIALS AND METHODS
All materials used were obtained from standard sources as described in previous papers (12,15). Yeast 80S ribosomes and the subunits were prepared essentially as described (15). EF-3 was purified to homogeneity from an overexpressing yeast strain containing the plasmid-borne YEF3 gene according to Kambampati and Chakraburtty (16). SDSpolyacrylamide gel electrophoresis was performed using 10% gels as described (17). For Western blotting, proteins were transferred electrophoretically to polyvinylidene difluoride membrane and developed with a polyclonal antibody to EF-3 (18). 125 I-Protein A was used as a secondary antibody for the detection of anti-EF-3 cross-reacting proteins.
ATPase Activity Assays-The nucleotide hydrolytic activity of EF-3 was measured under standard assay conditions (15). To measure the ribosome-stimulated ATPase activity of EF-3, 6.3 pmol of twice washed yeast ribosomes were included in the reaction. The amount of 32 P i released from [␥-32 P]ATP was measured according to our previously published protocol (15).
Tryptic Digestion of EF-3-EF-3 (0.6 mg/ml) in 0.1 M Tris/HCl, pH 8.0, was digested at 30°C with TPCK-trypsin at an EF-3/trypsin (w/w) ratio of 125:1. Reactions were carried out in buffer A (25 mM Tris/HCl, pH 7.5, 10 mM Mg(OAc) 2 , 50 mM NH 4 Cl, 1 mM dithiothreitol, 3% glycerol). Aliquots containing 2 g of EF-3 were removed at various times, and the digestions were terminated with a 5-fold excess (w/w) of soybean trypsin inhibitor. Samples were either assayed for the ATPase activity or diluted into SDS sample buffer, heated at 90°C for 10 min, and subjected to electrophoresis.
Fluorescence Measurements-All measurements were carried out at room temperature (22°C) using an SLM 4800C spectrofluorometer equipped with xenon arc light source. Slits of 4-nm width were used for excitation and emission. The emission spectra of EF-3 (10 g/ml in buffer A) were recorded by excitation at 294 nm in a rectangular quartz cuvette (inner diameter, 10 mm). The observed fluorescence intensities were corrected by subtracting the appropriate blanks and are expressed in arbitrary units. * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
** To whom correspondence should be addressed: Dept. of Biochemistry, Medical College of Wisconsin, 8701 Waterton Plank Rd., Milwaukee, WI 53226. Tel.: 414-456-8440; Fax: 414-456-6510; E-mail: chakra@ mcw.edu. 1 The abbreviations used are: EF, elongation factor; NBS, nucleotide Gel Shift Assay-The gel mobility shift assays were carried out according to the published protocol of Lohman et al. (19). EF-3 plus tRNA Phe was incubated in buffer A at 30°C for 10 min. Samples were analyzed by agarose-gel electrophoresis (1.5% agarose containing 0.5% ethidium bromide) in a low salt running buffer (20 mM Tris acetate, pH 7.8, 0.4 mM sodium acetate, 0.2 mM Na 2 EDTA). Samples of 30-l volume were mixed with 2 l of the gel loading buffer (0.04% bromphenol blue in 50% glycerol). Electrophoresis was conducted at 4°C at a constant voltage of 7 V/cm. The migration was monitored during the run with a hand-held UV lamp and was photographed in a UVP gel documentation system. NH 2 -terminal Analysis-Tryptic fragments of EF-3 were excised from the polyvinylidene difluoride membrane and subjected to sequence analysis on a pulsed liquid phase protein sequencer (Applied Biosystems/Perkin Elmer, model 477A). Phenylthiohydantoin-derivatives were chromatographed by an on-line HPLC unit, model 120A. Sequencing data were derived from the algorithm of the instrument and by manual evaluation of the chromatogram.
Mass Spectrometry-The excised peptides from Coomassie-stained SDS-polyacrylamide gels were reduced, denatured, and alkylated with iodoacetamide prior to digestion with trypsin. All cysteine residues were converted to carboxyamidomethylcysteine due to alkylation. Each peptide mixture was run through a capillary HPLC system using a Hewlett-Packard 1100 quaternary HPLC pump and an LC Packing Vydac C18 P3 300 A 300-m x 150-mm capillary column. Eluent A was 2% acetonitrile in water with 0.02% trifluoroacetic acid, and B was 2% acetonitrile in water with 0.018% trifluoroacetic acid. Gradient elutions from 0% B to 90% over 40 min at a flow rate of 3 l/min provided separation of the peptides. The eluents were introduced into a Finnigan (San Jose, CA) LCQ mass spectrometer fitted with an ESI interface. The peptide samples were introduced without any splitting, and a mass spectrometer was used to scan from 360 to 2,000 Da. The mass of the peptides containing 3 or Ͻ3 amino acid residues could not be detected accurately by this method. For peptides greater than 2 kDa, [Mϩ2H]2ϩ, [Mϩ 3 H]3ϩ, and [Mϩ4H]4ϩ were monitored by the instrument.

RESULTS AND DISCUSSION
Treatment of EF-3 with TPCK-trypsin (8 g of trypsin/mg of EF-3) at 30°C generated three large and several small peptides (Fig. 1). The estimated molecular masses of the large peptides (designated as A, B 1 , and C) were 90, 70, and 50 kDa, respectively. A close examination of the intensities of the fragments generated during the tryptic time course revealed that fragment A was converted to fragment C through fragment B 1 . Fragment C remained stable to proteolysis up to 60 min (lane 8). The 90-kDa fragment A was stable to proteolysis at 0°C (12). The relative stability of the fragments B 1 and C from trypsinolysis suggests the possible existence of two distinct subdomain structures within fragment A. The low molecular weight peptides (Ͻ30 kDa) that appeared at 1 min (lane 2) were completely degraded by 10 min of digestion (lane 5). A similar proteolytic cleavage pattern for EF-3 was reported by Miyazaki et al. (13) and Ladror et al. (14). Data reported by Mizyaki et al.
showed the presence of an additional proteolytic fragment B 2 , which we observed only when EF-3 was digested in the presence of ATP (see Fig. 4A).
Tryptic cleavage destroyed both the intrinsic and 80S-stimulated ATPase activity of EF-3 ( Fig. 2). However, the inactivation kinetics were different for these two functions (see insets in Fig. 2). The ribosome-stimulated ATPase activity was destroyed quantitatively within 10 min of the reaction (Fig. 2B).
The digested products retained about 10% of the intrinsic ATPase activity ( Fig. 2A). Neither the intact EF-3 nor fragment A was visible at this time point (Fig. 1, lane 5). We attribute the 10% residual intrinsic ATPase activity to fragment B 1 . Further studies with the isolated peptides will be necessary to confirm the hydrolytic activity of the truncated fragment. The loss of more than 90% of the ribosome-stimulated function of EF-3 within the first few minutes of tryptic digestion (Fig. 2B) strongly suggests that the initial cleavage that generated fragments A and B 1 destroyed the primary ribosome binding site of the protein. This conclusion is in agreement with our previously published data (12).
In an attempt to analyze the effects of substrate binding on the domain structure of EF-3, tryptic digestions were carried out in the presence of 80S ribosomes, ATP, polynucleotides, rRNA, and tRNA (Figs. [3][4][5][6]. Data presented in Fig. 3 show the trypsinolysis of EF-3 in the presence and absence of ribosomes. These experiments were conducted exactly as reported in our previous publication (12) except in the present studies, the reactions were at 30°C instead of 0°C. The tryptic fragments of EF-3 were visualized by Western blotting to avoid interference by the ribosomal proteins. A comparative analysis of the immunoblot data presented in Fig. 3 with the Coomassiestained gel in Fig. 1 revealed some interesting differences. First of all, EF-3 was relatively more resistant to trypsinolysis in the presence of the ribosome even with 10-fold excess of enzyme (the ratio of EF-3 to trypsin was 12.5:1 instead of 125:1). This amount of trypsin was shown to have minimal effect on the structure and the conformation of the ribosome (12,21). Peptides B 1 and C1 were also protected from trypsinolysis as evidenced by the accumulation of these two fragments in the presence of the ribosome (Fig. 3). the protective effect of the ribosome on the carboxyl-terminally derived C1 peptide was reported in our earlier paper (12). The relative stability of the 70-kDa peptide from trypsin implicates the existence of a possible second ribosome binding site distinct from that identified near the carboxyl-terminal end (12).
The presence of a second ribosome binding site near the NH 2 -terminal end of EF-3 was suggested by Gontarek et al. (20). Alternatively, ribosomes may cause sufficient alterations in the EF-3 conformation, thus protecting the previously exposed trypsin-sensitive bonds. Reports in the literature support this hypothesis (21). Experiments are currently in progress to differentiate between these possibilities.
EF-3 interacts with ATP, tRNA, rRNA, and various synthetic oligonucleotides (15,16,20,(23)(24)(25). Data presented in Fig. 4A demonstrate the effect of ATP on the trypsinolysis and on the intrinsic fluorescence of EF-3. Comparative analysis of the rate of proteolysis of EF-3 in the absence and presence of ATP clearly demonstrates the protective effect of ATP on EF-3. The nucleotide also protected the A and B 1 peptides. A doublet band, designated as B 2 peptide, appeared with an electrophoretic mobility similar to that of B 1 peptide (Fig. 4A, lane 3). The 90-kDa A peptide was generated by cleavage of the most trypsin-sensitive bond in EF-3 within NBSII (12). This cleavage site was protected by ATP. A new trypsin-sensitive bond in NBSI was exposed in the presence of ATP, which resulted in the formation of fragment B 2 (for the amino acid sequence, see Fig. 7). Similar results were reported by Miyazaki et al. (13) for EF-3 and by Yike et al. (22) for the cystic fibrosis transmembrane conductance regulator protein CFTR. These results strongly suggest a change in the conformation of EF-3 upon ATP binding. However, there was no significant change in the intrinsic fluorescence of the protein in the presence of ATP (Fig.  4B). We conclude that ATP changed the conformation of EF-3 around its binding sites without causing a global change in the overall conformation of the protein. ATP showed no protective effect on the carboxyl-terminally derived C1, C2, and C3 peptides. It will be of interest to investigate the role of ATP in the activation of the ribosome-stimulated hydrolytic function of EF-3 similar to that observed for the GTP-binding proteins EF-1␣ (EF-Tu) and EF-2 (EF-G) (26).
The functional dependence of the yeast ribosomes on EF-3 strongly suggests that the protein may interact directly with rRNA. In the present studies, we investigated the effect of synthetic oligonucleotides and ribosomal RNA on the proteolytic cleavage pattern and on the intrinsic fluorescence of EF-3. Of the three homopolynucleotides tested, only poly(G) showed the most significant effect. The effect was visible when the trypsinolysis was carried out at 30°C and also at 0°C (Fig. 5,  A and B). In the presence of poly(G), no detectable amount of the full-length EF-3 or fragment A was visible at the end of the 10-min digestion time (Fig. 5A). The cleavage pattern in the absence of oligonucleotides was very similar to those with poly(U) and poly(C). Fragment C (50 kDa) was formed only when EF-3 was digested at 30°C (Fig. 5A, EF-3 alone). However, in the presence of poly(G), fragment C was formed even when the reactions were carried out at 0°C (Fig. 5B, ϩ poly(G)). The binding studies as measured by fluorescence quenching (Fig. 5C, filled circles) reflected a similar effect of poly(G). We reported previously a high affinity binding of EF-3 to a guanosine-rich sequence in RNA (23,24). An altered proteolytic cleavage pattern and fluorescence quenching by poly(G) suggest a possible conformation change of the protein upon binding to a guanosine-rich sequence in rRNA. However, neither 18S nor 26S rRNA separately or as a mixture showed any detectable effect on the proteolysis of EF-3 (data not shown).
The primary function of EF-3 is to remove deacylated tRNA from the ribosomal exit site (E-site) and stimulate binding of the ternary complex to the A-site (6,25). The deduced amino acid sequence of EF-3 revealed the presence of a putative aminoacyl-tRNA synthetase sequence motif (10). Based on this information, we examined for a possible interaction between EF-3 and tRNA by gel mobility shift assays. Data presented in Fig. 6A demonstrate that indeed, the mobility of tRNA was retarded upon interaction with EF-3 (lanes 2-6). A nonspecific protein, bovine serum albumin, had no effect (lane 7). It should be noted that the complex was not detected when electrophoresis was continued for longer than 10 -15 min. The disappearance of the EF-3⅐tRNA complex does not represent degradation of either tRNA or the protein because these macromolecules remained intact at the end of the experiment (data not shown). However, tRNA did not change the tryptic cleavage pattern of the protein (Fig. 6C). The labile nature of the EF-3⅐tRNA complex and the inability of tRNA to protect it from proteolysis may be a reflection of a weak interaction between these two macromolecules. To analyze further the domain structure of EF-3, several of the tryptic fragments were sequenced by mass spectrometry. Coomassie-stained bands from the gel (Fig. 1) were excised and subjected to complete tryptic digestion. The peptides were sequenced using a mass spectrometer as described under "Materials and Methods." The peptides identified in each fragment are listed in Tables I-III. As noted in the earlier section, trypsinolysis of EF-3 at 0°C resulted in the cleavage of a single peptide bond generating a large 90-kDa fragment (N peptide) and a small 30-kDa fragment (C1 peptide). The N peptide is equivalent to fragment A described in the current studies (Fig. 1). The C1 peptide originated from the carboxyl-terminal end and contains the amino acid residues Gln 775 -Phe 1044 (12). Attempts to sequence fragments A, B 1 , and C were unsuccessful because all three were amino-terminally blocked. We conclude that these fragments are from the NH 2 -terminal end. The 90-kDa fragment contains residues Ser 2 -Arg 774 . This domain of EF-3 was stable to further proteolysis at 0°C and retained the intrinsic ATPase activity (12). Our results indicate that proteolysis at 30°C exposed additional trypsin-sensitive bonds in the 90-kDa domain, resulting in the formation of a stable 50-kDa fragment through the intermediate formation of the 70-kDa fragment (Fig. 1).
Data presented in Table I show the bulk of the peptides identified in the 70-kDa fragment. Tandem mass spectrometry sequencing confirmed the presence of an NH 2 -terminally blocked octapeptide ( 2 SDSQQSIK 9 ) containing an acetylated serine. Analysis of the total tryptic digest of the 70-kDa fragment identified the presence of the peptide Phe 650 -Lys 660 , but peptides Val 669 -Arg 696 and Ile 697 -Lys 707 were absent. Peptides Thr 661 -Lys 662 and Gln 663 -Lys 664 were too small for the detection by LCQ mass spectrometric analysis. We failed to identify the peptide Ala 665 -Lys 668 (430 Da) in the total digest of the full-length EF-3 and also in the 70-kDa fragment. This peptide may or may not be present in the 70-kDa fragment. Based on these analyses, we concluded that the 70-kDa fragment of EF-3   began at residue 2 and ended somewhere between the residues 660 and 668 (Fig. 7). The 70-kDa fragment contains one complete nucleotide-binding cassette (NBSI). ATP protected this fragment from further degradation (Fig. 4). We hypothesize that the 70-kDa fragment forms the core ATP binding subdomain within the NH 2 -terminally derived 90-kDa domain.
The 50-kDa fragment gave a mass of 48.4 kDa in the LCQ mass spectrometer (data not shown). The last peptide confirmed in the total tryptic digest of the 50 kDa fragment was Ala 411 -Lys 443 . Peptide Ile 444 -Lys 448 and those COOH-terminal to this peptide were absent in the 50-kDa fragment. Thus, the 50-kDa fragment spans from Ser 2 to Lys 443 . The calculated molecular mass of 48.38 kDa is in good agreement with the size of this fragment.
Tables II and III list the identified peptides in the COOHterminally derived 21-and 15-kDa fragments. Peptides Glu 767 -Arg 771 , Gln 775 -Lys 786 , Ile 787 -Lys 789 , and Met 971 -Arg 987 were absent in the total tryptic digest of the 21-kDa fragment. The first and the last identified peptides were Ile 790 -Arg 795 and Asn 958 -Lys 967 , respectively (Table II). Thus, the 21-kDa fragment spans from residue 790 through 967/970. The 15-kDa fragment starts at residue 843 and ends at 970 (Table III). The most trypsin-sensitive cleavage sites of EF-3 are indicated by vertical arrows in Fig. 7. EF-3 contains 126 possible trypsin cleavage sites. Only a few of these sites were accessible to trypsin under the experimental conditions used in the current studies. Results presented in this paper and elsewhere (12,14) strongly suggest the existence of at least two organized functional domains in EF-3. The protein also contains two trypsin-resistant cores (B 1 and C) near the NH 2 -terminal end. Our results demonstrate that the lysine and arginine residues in NBSI (residues 463-575) are relatively inaccessible to trypsin compared with those in NBSII (residues 701-928). The most trypsin-sensitive bond of EF-3 (Arg 774 -Gln 775 ) is located within NBSII (between the purine binding sequence and the phosphate binding loop, see Fig. 7). The sequence around this cleavage site is exposed and may function as a hinge between the stable amino-terminal domain and the flexible carboxyl-terminal domain. The phosphate binding loop of the second nucleotide binding domain along with the putative aminoacyl-tRNA synthetase homology region (residues 820 -865) and the positively charged polylysine blocks (residues 1009 -1031) form a loosely structured carboxyl-terminal domain that contains the primary ribosome binding site of the protein (12). The protective effect of yeast ribosomes on the trypsinolysis of EF-3 suggests that the cleavage sites Arg 842 , Arg 970 , and the lysine residues located near the carboxyl-terminal end are protected by the ribosome. Fragment C (residues 2-443) is extremely resistant to trypsin (12,14). The functional significance of this stable domain near the NH 2terminal end of EF-3 remains undefined. Yeast strains expressing the truncated form of EF-3 (lacking Glu 12 -Ala 411 ) were shown to be nonviable (27). 2 Further work with the purified EF-3 fragments is needed to fully understand the structural organization of EF-3. Attempts to crystallize the full-length protein were unsuccessful (Kambampati & Chakraburtty, unpublished data). The loosely structured carboxyl-terminal end may have interfered the crystal formation. With the identification of several functional subdomains, it should be feasible to determine the structures of the individual domains by x-ray crystallography.