Portal de Periódicos da CAPES

Investigation of Functional Aspects of the N-terminal Region of Elongation Factor Tu from Escherichia coli Using a Protein Engineering Approach

1998; Elsevier BV; Volume: 273; Issue: 8 Linguagem: Inglês

10.1074/jbc.273.8.4387

1083-351X

Martin Laurberg, Francisco Mansilla, Brian F.C. Clark, Charlotte R. Knudsen,

Enzyme Structure and Function

The function of the N-terminal region of elongation factor Tu is still unexplained. Until recently, it has not been visible in electron density maps from x-ray crystallography studies, but the presence of several well conserved basic residues suggest that this part of the molecule is of structural importance for the factor to function properly. In this study, two lysines at positions 4 and 9 were mutated separately to alanine or glutamate. The resulting four point mutants were expressed and purified using the pGEX system. The untagged products were characterized with regard to guanine-nucleotide interaction, intrinsic GTPase activity, and binding of aminoacyl-tRNA (aa-tRNA). The results show that Lys9 is especially strongly involved in the association with guanine nucleotides and the binding of aa-tRNA. Also Lys4 plays a role in the association of GDP and GTP and is also of some importance in aa-tRNA binding. Our results are discussed in structural terms with the conclusion that a complex network of interactions across the interface between domains 1 and 2 with Lys9 being a key residue seems to be important for the fine tuning of the dimensions of the cleft accommodating the acceptor end of aa-tRNA as well as delineating the structure of the effector region. The function of the N-terminal region of elongation factor Tu is still unexplained. Until recently, it has not been visible in electron density maps from x-ray crystallography studies, but the presence of several well conserved basic residues suggest that this part of the molecule is of structural importance for the factor to function properly. In this study, two lysines at positions 4 and 9 were mutated separately to alanine or glutamate. The resulting four point mutants were expressed and purified using the pGEX system. The untagged products were characterized with regard to guanine-nucleotide interaction, intrinsic GTPase activity, and binding of aminoacyl-tRNA (aa-tRNA). The results show that Lys9 is especially strongly involved in the association with guanine nucleotides and the binding of aa-tRNA. Also Lys4 plays a role in the association of GDP and GTP and is also of some importance in aa-tRNA binding. Our results are discussed in structural terms with the conclusion that a complex network of interactions across the interface between domains 1 and 2 with Lys9 being a key residue seems to be important for the fine tuning of the dimensions of the cleft accommodating the acceptor end of aa-tRNA as well as delineating the structure of the effector region. Elongation factor Tu (EF-Tu) 1The abbreviations used are: EF-Tu, elongation factor Tu; aa-tRNA, aminoacyl-tRNA; GDPNP, guanine 5′-(β,γ-imidotriphosphate). is engaged in the elongation step of protein biosynthesis with the main function of transporting aminoacylated tRNA (aa-tRNA) to the A site of the mRNA-programmed ribosome. EF-Tu is a G-binding protein capable of binding either GTP or GDP. EF-Tu must be in its GTP-bound form to bind aa-tRNA. When proper codon-anticodon interaction is established at the A site, EF-Tu hydrolyzes the bound GTP to GDP, whereby a conformational change is induced in EF-Tu. Thus EF-Tu looses its affinity for aa-tRNA and the ribosome and dissociates, whereas aa-tRNA remains bound, allowing the attached amino acid to be incorporated into the growing polypeptide chain. The elongation factor Ts is responsible for the reactivation of EF-Tu because it acts as a nucleotide exchange factor (for a review see Ref. 1Kjeldgaard M. Nyborg J. Clark B.F. FASEB J. 1996; 10: 1347-1368Crossref PubMed Scopus (232) Google Scholar). EF-Tu is well characterized both biochemically and structurally. During recent years the structures of the following forms of EF-Tu have been solved using x-ray crystallography: EF-Tu·GDP (2Kjeldgaard M. Nyborg J. J. Mol. Biol. 1992; 223: 721-742Crossref PubMed Scopus (246) Google Scholar, 3Polekhina G. Thirup S. Kjeldgaard M. Nissen P. Lippmann C. Nyborg J. Structure. 1996; 4: 1141-1151Abstract Full Text Full Text PDF PubMed Scopus (206) Google Scholar, 4Abel K. Yoder M.D. Hilgenfeld R. Jurnak F. Structure. 1996; 4: 1153-1159Abstract Full Text Full Text PDF PubMed Scopus (182) Google Scholar), EF-Tu·GDPNP (5Berchtold H. Reshetnikova L. Reiser C.O. Schirmer N.K. Sprinzl M. Hilgenfeld R. Nature. 1993; 365: 126-132Crossref PubMed Scopus (513) Google Scholar, 6Kjeldgaard M. Nissen P. Thirup S. Nyborg J. Structure. 1993; 1: 35-50Abstract Full Text PDF PubMed Scopus (368) Google Scholar), EF-Tu·EF-Ts (7Kawashima T. Berthet Colominas C. Wulff M. Cusack S. Leberman R. Nature. 1996; 379: 511-518Crossref PubMed Scopus (279) Google Scholar), and the ternary complex, EF-Tu·GDPNP·aa-tRNA (8Nissen P. Kjeldgaard M. Thirup S. Polekhina G. Reshetnikova L. Clark B.F. Nyborg J. Science. 1995; 270: 1464-1472Crossref PubMed Scopus (802) Google Scholar). These structures show that the EF-Tu molecule is composed of three structural domains, of which domain 1 binds the guanine nucleotide. Both intra- and inter-domain rearrangements during the elongation cycle of EF-Tu are responsible for the allosteric behavior of the elongation factor. Despite the wealth of information about EF-Tu, the role of the N-terminal region of elongation factor Tu remains unknown. This region of EF-Tu, comprising the first 10 amino acids of the protein, contains several conserved basic residues. The lysines at positions 2 and 4 are semi-conserved among prokaryotes. Arginine 7 is more than 98% conserved among prokaryotes, whereas the eukaryotic counterpart of EF-Tu, EF-1α, has a 100% conserved lysine at this position. Finally, lysine 9 is fully conserved among all known EF-Tu and EF-1α species. Moreover, electron density maps of the EF-Tu·GDP and the EF-Tu·EF-Ts complexes have a very low resolution in this part of the crystal structures, indicating that the N-terminal region is a flexible part of the molecule. The conservation of these residues along with the ability of aa-tRNA to protect the residues against chemical modification (9Antonsson B. Leberman R. Eur. J. Biochem. 1984; 141: 483-487Crossref PubMed Scopus (16) Google Scholar) made them likely candidates to bind the negatively charged phosphate backbone of tRNA. This idea was supported by comparing the structures of the GDP and GTP forms of EF-Tu. This revealed that a positively charged cleft with the size of an RNA helix was created between domains 1 and 2 in the neighborhood of the N-terminal upon binding of GTP (10Sprinzl M. Trends Biochem. Sci. 1994; 19: 245-250Abstract Full Text PDF PubMed Scopus (84) Google Scholar). Now the structure of the ternary complex has been solved (Fig. 1) (8Nissen P. Kjeldgaard M. Thirup S. Polekhina G. Reshetnikova L. Clark B.F. Nyborg J. Science. 1995; 270: 1464-1472Crossref PubMed Scopus (802) Google Scholar, 11Nissen P. Kjeldgaard M. Thirup S. Clark B.F.C. Nyborg J. Biochimie. 1996; 78: 921-933Crossref PubMed Scopus (51) Google Scholar), and it confirms the accommodation of the acceptor stem of tRNA by the cleft. We decided to study the role of the basic residues of the N-terminal region in more detail using a protein engineering approach. The residues, Lys4 and Lys9, were changed into alanine or glutamate by site-directed mutagenesis. The resulting four recombinant proteins, K4E, K4A, K9A, and K9E, were expressed, purified, and characterized in various assays to increase our knowledge about this baffling part of the elongation factor. Site-directed mutagenesis was carried out using the Sculptor™ in vitro mutagenesis system from Amersham Life Science, which is based on the Eckstein procedure (12Taylor J.W. Ott J. Eckstein F. Nucleic Acids Res. 1985; 13: 8765-8785Crossref PubMed Scopus (565) Google Scholar). The template used was M13mp11tufA containing the tufA gene fromEscherichia coli encoding EF-Tu positioned downstream of a stretch of nucleotides encoding the recognition site of the serine protease factor Xa (13Knudsen C.R. Clark B.F. Degn B. Wiborg O. Biochem. Int. 1992; 28: 353-362PubMed Google Scholar). The primers 5′-GTTCTAAAGAAGCATTTGAACG-3′, 5′-GTTCTAAAGAAGAATTTGAACG-3′, 5′-GAACGTACAGCACCGCACGTTA-3′, and 5′-GAACGTACAGAACCGCACGTTA-3′ gave rise to the mutants K4A, K4E, K9A, and K9E, respectively. Clones containing the mutations of interest were identified by Sanger sequencing of single-stranded DNA (14Sanger F. Nicklen S. Coulson A.R. Proc. Natl. Acad. Sci. U. S. A. 1977; 74: 5463-5467Crossref PubMed Scopus (52743) Google Scholar). BamHI fragments containing the mutated tufA genes and the nucleotides encoding the factor Xa recognition site were subcloned into the pGEX1 expression vector (15Smith D.B. Johnson K.S. Gene (Amst.). 1988; 67: 31-40Crossref PubMed Scopus (5047) Google Scholar). The continued presence of the mutations were verified by Sanger sequencing of double-stranded DNA (16Sambrook J. Fritsch E.F. Maniatis T. Molecular Cloning: A Laboratory Manual. 2nd Ed. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY1989: 13.3-13.6Google Scholar). The mutanttufA genes were expressed in the E. coli strain JM109 according to Ref. 13Knudsen C.R. Clark B.F. Degn B. Wiborg O. Biochem. Int. 1992; 28: 353-362PubMed Google Scholar with the exception that the temperature was 37 °C at the point of innoculation and later lowered to 28 °C upon induction. Protein purification was carried out as described (13Knudsen C.R. Clark B.F. Degn B. Wiborg O. Biochem. Int. 1992; 28: 353-362PubMed Google Scholar). The concentration of active protein capable of binding guanine nucleotides was determined as described by Miller and Weissbach (17Miller D.L. Weissbach H. Methods Enzymol. 1974; 30: 219-232Crossref PubMed Scopus (114) Google Scholar). The stability of the mutants were evaluated by following their irreversible heat denaturation. 0.5 μm EF-Tu was mixed with 10 μm[3H]GDP (1100 dpm/pmol) in binding buffer (50 mm Tris-Cl, pH 7.6, 4 °C, 100 mmNH4Cl, 50 mm KCl, 10 mmMgCl2, 1 mm dithiothreitol) and incubated for 90 min on ice. Hereafter, aliquots were incubated at different temperatures between 0 and 60 °C for 20 min. 75-μl samples were filtered through cellulose-acetate filters. The filters were washed three times with 1-ml aliquots of ice-cold wash buffer (10 mm Tris-Cl, pH 7.6, 4 °C, 10 mmMgCl2, 10 mm NH4Cl), and the amount of bound nucleotide was measured in a scintillation counter. The assay was also performed with GTP in which case EF-Tu was activated with [3H]GTP, phosphoenolpyruvate, and pyruvate kinase as described below. The apparent dissociation rate constants, k−1, for GDP and GTP were determined as described (18Knudsen C.R. Kjaersgard I.V. Wiborg O. Clark B.F. Eur. J. Biochem. 1995; 228: 176-183Crossref PubMed Scopus (25) Google Scholar) with the exception that the dissociation of [3H]GTP was followed by isotopic dilution with 100-fold excess of unlabeled GDP over [3H]GTP in GTPase buffer (50 mm Tris-Cl, pH 7.6, 4 °C, 60 mm NH4Cl, 10 mm MgCl2, 1 mm dithiothreitol). Nucleotide-free EF-Tu was prepared using the charcoal method essentially as described previously (19Knudsen C.R. Clark B.F.C. Protein Eng. 1995; 8: 1267-1273Crossref PubMed Scopus (22) Google Scholar). The exact concentration of active EF-Tu in the resulting supernatant was measured using the nucleotide binding assay. The efficiency of the charcoal method was confirmed using EF-Tu·[3H]GDP as a substrate. The apparent association rate constant, k+1, for the reaction between nucleotide-free EF-Tu and GDP/GTP was determined as follows. 0.016 μm nucleotide-free EF-Tu was mixed with 0.04–0.08 μm [3H]GDP (5100 dpm/pmol) or 0.12–0.4 μm [3H]GTP (8650 dpm/pmol) in GTPase buffer. Samples containing 1.6 pmol EF-Tu were withdrawn, spotted on cellulose-acetate filters (which were washed with 3 ml of ice-cold wash buffer), and counted. Second-order kinetics was assumed in the following data processing. The [3H]GTP was pretreated with phosphoenolpyruvate and pyruvate kinase prior to the assay to rephosphorylate traces of spontaneously hydrolyzed GTP. The tRNA was charged as follows: 40 μm yeast tRNAPhe, 5 mm ATP, and 200 μm [14C]Phe (120 dpm/pmol) in charging buffer (0.1 m Tris-Cl, pH 7.5, 50 mmNH4Cl, 12 mm MgCl2, 2 mm ATP, 0.24 mm CTP, 2.8 mmβ-mercaptoethanol) was incubated with an adequate amount (empirically determined) of crude synthetase extract from bakers' yeast (20von der Haar F. Methods Enzymol. 1979; 59: 257-267Crossref PubMed Scopus (36) Google Scholar) at 37 °C for 13 min. A charging degree exceeding 50% is suitable for subsequent assays. The RNaseA protection assay was performed by mixing 0.5 mmphosphoenolpyruvate, 0.1 mg/ml pyruvate kinase, 10 μmGTP, 2 μm EF-Tu·GDP in buffer F (50 mmTris-Cl, pH 7.0, 100 mm KCl, 40 mmNH4Cl, 7 mm MgCl2, 2 mmdithiothreitol, 0.1 mm EDTA) and incubating the mixture at 4 °C for 1 h. An equal volume of [14C]Phe-tRNAPhe (0.6 μm) in buffer F was added, and the mixture was incubated at 30 °C for 10 min allowing the ternary complex, EF-Tu·GTP·Phe-tRNAPhe, to form. RNaseA was added to a concentration of 10 μg/μl, and the ternary complex degradation was followed on ice by successive withdrawal of samples containing 25 pmol of EF-Tu into ice-cold 10% trichloroacetic acid. After 10 min, samples were filtered through nitrocellulose filters, which were then washed with 3 ml of 10% trichloroacetic acid and counted. First-order kinetics was assumed, and the apparent dissociation rate constant,k−1, was determined as the slope of a plot of −ln(Ct/C0) versusreaction time. Ct and C0denote the concentration of ternary complex at time t and time 0, respectively. The nonenzymatic hydrolysis assay was carried out essentially as described by Andersen and Wiborg (21Andersen C. Wiborg O. Eur. J. Biochem. 1994; 220: 739-744Crossref PubMed Scopus (24) Google Scholar). The half-life,t½, of the aminoacyl bond was determined as ln(2)/slope of a representation of −ln(Ct/C0) versustime. The intrinsic GTPase activity was examined by determination of Km and kcat for the one-round reaction as described by Parmeggiani and Sander (22Parmeggiani A. Sander G. Mol. Cell. Biochem. 1981; 35: 129-158Crossref PubMed Scopus (95) Google Scholar). Prior to the assay, EF-Tu·GDP was converted to EF-Tu·GTP by incubation with phosphoenolpyruvate and pyruvate kinase. The assay was carried out at 30 °C. The concentration of EF-Tu was fixed at 0.3 μm, and the concentraion of [γ-32P]GTP (500–1500 dpm/pmol) was varied within the range of 1–25 μm. The kinetic parameters Km and kcat were determined from the resulting data set presented in a Hanes plot. Michaelis-Menten kinetics are assumed. The yields of the different purified EF-Tu species were 1 ± 0.2 mg/g wet cell paste. The amount of active protein was 60–80% of the total protein concentration estimated after the Bradford procedure (23Sedmark J.J. Grossberg S.E. Anal. Biochem. 1977; 79: 544-552Crossref PubMed Scopus (2506) Google Scholar). The temperature at which the amount of active EF-Tu·GDP is reduced to 50%, φ, is 53 °C for wild-type EF-Tu and all of the mutants (data not shown). φ for EF-Tu·GTP is 45 °C for the wild type and the mutant K4E and 40 °C for the mutants K4A, K9A, and K9E. The following assays were therefore performed at 30 °C or below. The apparent dissociation rate constants, k−1, for the EF-Tu·GDP complexes of the mutants do not differ significantly from those of the wild-type EF-Tu complex. Slight increases are observed fork−1(GTP) with the largest deviance observed for the mutant K9E (Table I).Table IDissociation and association rate constants determined at 0 °CGDPGTPk−1 (×103)k+1(×10−4)Kdk−1(×103)k+1(×10−4)Kds−1M−1s−1nMs−1M−1s−1nMWild-type EF-Tu0.38 ± 0.02630 ± 1.51.33.8 ± 0.878.4 ± 0.05745K4A EF-Tu0.33 ± 0.008715 ± 0.712.29.2 ± 2.56.2 ± 1.1148K4E EF-Tu0.40 ± 0.04125 ± 7.01.66.4 ± 0.916.1 ± 4.9105K9A EF-Tu0.35 ± 0.01211 ± 0.0793.27.6 ± 1.63.2 ± 0.51238K9E EF-Tu0.33 ± 0.0120.86 ± 0.143815 ± 2.30.44 ± 0.0263400 Open table in a new tab The apparent association rate constants, k+1, of the binary complexes with both GDP and GTP are lowered for all of the mutants, but the decreases are most pronounced for the Lys9mutants (Table I). The introduction of a glutamate causes the largest effect at position 9, whereas the identity of the introduced amino acid is of no importance at position 4. The dissociation constants, Kd, summarizing the overall effect of the mutations on the nucleotide affinities were calculated as k−1/k+1(Table I). The most dramatic effect is observed for the mutant K9E, for which the affinity for GDP and GTP has been reduced 30- and 75-fold, respectively. The apparent dissociation rate constant,k−1, for dissociation of the ternary complex was determined using the RNase digestion assay, in which the protective effect of EF-Tu on the aminoacyl bond of aa-tRNA was measured (Fig. 2 and Table II). The mutant K9E offers no protection of the aa-tRNA at all, whereas K4A protects with the same efficiency as wild-type EF-Tu. The two other mutants, K4E and K9A, fall in between these two extremities.Table IIParameters characterizing the interaction between EF-Tu and Phe-tRNAPheApparent k−1 (× 103)Half-life of amino acyl bonds−1minWild-type EF-Tu4.6 ± 0.652240 ± 300K4A EF-Tu4.4 ± 0.38900 ± 36K4E EF-Tu6.3 ± 0.78540 ± 26K9A EF-Tu7.9 ± 0.66410 ± 7.8K9E EF-Tu25 ± 3.853 ± 2.0Unprotected aa-tRNA32 ± 4.232.4 ± 2.5 Open table in a new tab The half-lives for the spontaneous hydrolysis of the aminoacyl bond are calculated from Fig. 3 and given in Table II. The results resemble those of the RNaseA protection assay, but in this case the mutants K4E and K9A resemble the wild type more than in the RNaseA protection assay. This is seen most clearly in Figs. 2 and3, which more readily enable an evaluation of the results within the frame of the limits of each assay. The span of the nonenzymatic hydrolysis assay is seen to be wider (the ratio between the wild type and the blank is 70) than that of the RNaseA protection assay (the corresponding ratio is only 8). A 2-fold difference in the RNaseA protection assay, as observed for the mutant K9A, is thus comparable with a 18-fold difference in the nonenzymatic hydrolysis assay. The parameters characterizing the intrinsic GTPase activity, Km and kcat, were determined from Hanes plots. The results are summarized in Table III. The Michaelis-Menten constants, Km, are slightly increased for the mutants K4A and K4E and further increased for mutants K9E and K9A, indicating that the stability of the substrate complexes, EF-Tu·GTP, are lowered for the mutants. Minor increases are also observed for the rate constants of product formation,kcat. The kcat/Km value, the apparent second-order rate constant of substrate complex formation, is not affected by the mutations. Overall, the observed kinetic parameters of the intrinsic GTPase activity for the mutants are not significantly changed compared with the wild type.Table IIIKinetic parameters, Km and kcat, characterizing the intrinsic GTPase activities of the different EF-Tu species at 30 °CKmkcat (× 106)kcat (× 106)/KmμMs−1μM−1 · s−1Wild-type EF-Tu3.9 ± 0.0711350 ± 330346K4A EF-Tu6.8 ± 0.143020 ± 710444K4E EF-Tu4.8 ± 1.61870 ± 230390K9A EF-Tu9.0 ± 1.32590 ± 560288K9E EF-Tu9.1 ± 2.62230 ± 410245 Open table in a new tab We have investigated the function of the N-terminal region of elongation factor Tu by production and characterization of the four point mutants, K4A, K4E, K9A, and K9E. The recombinant proteins behave like the wild-type protein with respect to temperature stability in the EF-Tu·GDP conformation, whereas slight destabilizations are seen for mutants K4A, K9A, and K9E in the EF-Tu·GTP conformation. Overall, the structures of the mutants can be considered to be maintained in both conformations excluding any long range effects caused by the substitutions. This is supported by the finding that the GTPase activity and k−1(GDP) are not changed significantly by any of the mutations. The enzymatic hydrolysis protection assay and the nonenzymatic hydrolysis protection assay revealed that aa-tRNA binding is severely impaired for K9E. Also the mutants K4E and K9A have a reduced ability to protect aa-tRNA. K4A, on the other hand, does not show a noteworthy change compared with the wild type in its capacity to protect the aminoacyl bond. Thus it appears that both Lys4 and Lys9 are involved in the binding of aa-tRNA, although Lys9 seems to play the more important role. The crystal structure of the ternary complex, EF-Tu·GTP·aa-tRNA, composed of EF-Tu from Thermus aquaticus and Phe-tRNAPhe from yeast has been solved recently (8Nissen P. Kjeldgaard M. Thirup S. Polekhina G. Reshetnikova L. Clark B.F. Nyborg J. Science. 1995; 270: 1464-1472Crossref PubMed Scopus (802) Google Scholar, 11Nissen P. Kjeldgaard M. Thirup S. Clark B.F.C. Nyborg J. Biochimie. 1996; 78: 921-933Crossref PubMed Scopus (51) Google Scholar), allowing a structural evaluation of our biochemical data. In the following, the corresponding E. coli EF-Tu residues will be given in parentheses when different. The structure of the ternary complex reveals that none of the basic residues of the N-terminal region are in direct contact with the aa-tRNA (Fig. 1). Glu4 (Lys4) is solvent exposed, whereas the side chain of Lys9 is part of a tight network of interactions (Fig. 4). Lys9 forms a salt bridge to Glu71(Asp70) and a hydrogen bond to the main chain oxygen of Thr72 (Thr71). Glu71 and Thr72 are situated centrally in domain 1 in the loop connecting β-strands b and c. The side chain of Thr72 is associated with helix F via the side chain of Asp207(Asp196), which is furthermore in contact with the side chain of Tyr77 (Tyr76). In another direction, the Thr72 main chain nitrogen forms a hydrogen bond with Asn39 at the beginning of the effector region (residues 39–65 (38–64)). The amino acids 1–10 constitute a coiled structure protruding from domain 1 and contacting the surface of domain 2. The major binding to domain 2 takes place via a double salt bridge formed between the side chains of Arg7 and Asp284(Glu272). In this way, the effector region is connected to domain 2 via the C-terminal of β-strand b and the extreme N-terminal region. In another study (24Mansilla F. Knudsen C.R. Laurberg M. Clark B.F.C. Protein Eng. 1997; 10: 927-934Crossref PubMed Scopus (10) Google Scholar), we have shown that Arg7 is important for the stabilization of the GTP conformation of EF-Tu and essential for the concomitant binding of aa-tRNA. Lys9 is not essential but is important for fine tuning the dimensions of the cleft into which the aminoacyl end of aa-tRNA is docked via the intricate network of salt bridges and hydrogen bonds described above. Also Phe5 may play a role in stabilizing the interaction between domains 1 and 2. This phenylalanine points inward and establishes a hydrophobic cluster with His76(His75), Met272 (Met260), His273 (Phe261), and Lys275(Lys263). Of these, the last three residues are situated in domain 2, whereas the first is part of domain 1. In addition, Lys275 (Lys263) makes a salt bridge with Glu69 (Glu68), thereby enhancing the interaction further. All of the mentioned residues are well conserved, emphasizing the importance of the connection between the two domains established via the N-terminal region. K4E and K9A are impaired more strongly with respect to their ability to protect against hydrolysis by RNaseA than with respect to their ability to protect against nonenzymatic hydrolysis (compare Figs. 2 and 3). On the other hand, K9E, is affected almost equally in both assays. The K9E mutation is dramatic in terms of charge, showing that when the connection between domain 2 and the effector region is not only interrupted but kept apart by force (repulsion), EF-Tu becomes unable to protect aa-tRNA. Less drastic mutations like K4E and K9A affect the protection pattern in a more delicate manner, showing that the roles of these residues are to adjust the binding of aa-tRNA rather than a direct involvement in the binding of aa-tRNA as in the case of Arg7. The 3′-CCA-Phe end is docked into the cleft between domains 1 and 2. The amino acid is positioned in a pocket formed by the side chains of Phe229 (Phe218), Asp227 (Asp216), Glu226(Glu215), and Thr239 (Thr228). The amino group can form a hydrogen bond to the main chain CO of Asn285 (Asn272) and the main chain NH of His273 (Phe261), whereas the carbonyl group of the aminoester can form a hydrogen bond with the main chain NH of Arg274 (Arg262), the side chain of which interacts with the phosphate of A76 of the tRNA. Thus the amino group of the aminoacyl bond interacts with EF-Tu at the residue neighboring the interaction partner of Arg7. The phosphates of C74 and C75, which are the positions attacked by RNaseA, are in ionic contact with Lys52(Asn51) in the effector region. In addition, Asn64 (Asn63) of the effector region is in contact with C74 via a hydrogen bond to the O4′. All of the mutants except K9E protect the aminoacyl bond to some extent. Only when a direct repulsion is introduced, does the aa-tRNA-binding site collapse. Nevertheless, the mutations K4E and K9A also affect the binding of the tRNA, in particular the part of tRNA attacked by RNaseA. This phenomenon could well be caused by a misfolding of the effector regions of the mutants. Lys2 and Lys4 have been found to be protected against chemical modification by ethyl acetimidate when EF-Tu·GTP is in complex with aminoacyl-tRNA (9Antonsson B. Leberman R. Eur. J. Biochem. 1984; 141: 483-487Crossref PubMed Scopus (16) Google Scholar). This is not in accord with either the structure of the ternary complex (8Nissen P. Kjeldgaard M. Thirup S. Polekhina G. Reshetnikova L. Clark B.F. Nyborg J. Science. 1995; 270: 1464-1472Crossref PubMed Scopus (802) Google Scholar) or our results (Ref. 24Mansilla F. Knudsen C.R. Laurberg M. Clark B.F.C. Protein Eng. 1997; 10: 927-934Crossref PubMed Scopus (10) Google Scholar and this study). These lysines are located in an environment, which changes dramatically depending upon the functional state of EF-Tu. Apparently, it is these conformational changes rather than a shielding by aa-tRNA that is reflected in the modification study, illustrating the danger of interpreting chemical reactivity studies without structural information. The Lys9 mutants are affected with respect to association of guanine nucleotides. A glutamate at position 9 strongly decreasesk+1 of both GDP and GTP. A smaller effect is obtained upon introduction of an alanine. These results indicate that Lys9 either plays a role in the stabilization of the nucleotide-free form of EF-Tu or is involved in part of the structural rearrangements taking place upon nucleotide binding. The stabilizing role of Lys9 is apparently dependent upon an electrostatic interaction. The structures of the N-terminal regions of EF-Tu in complex with either GDPNP alone or GDPNP·Phe-tRNA are equivalent within the limitations of the method of x-ray crystallography. All of the interactions described for the ternary complex apply to the active form as well. Therefore, the effect observed on association with GTP is likely to be caused by the disruption of the salt bridge between Lys9 and Glu71 (Asp70). In contrast, k+1(GDP) of K4E is equal to the wild-type value, whereas that of K4A is halved, suggesting that the residue is exposed to the solvent and that the hydrophilic character of the side chain is more important than the actual property of the charge. In T. aquaticus, this residue is a glutamate supporting this assumption. The mutations to alanine or glutamate at position 4 do not affect association of the EF-Tu·GTP complex. Our data suggest that Lys9 is involved in a late rearrangement step taking place upon binding of a nucleotide. If the N-terminal region were involved in an earlier stage of nucleotide binding, then the stability and the GTPase activity of the mutants would probably have been impaired, too. The structural interpretation of our results on nucleotide interaction is unfortunately incomplete due to the absence of the N-terminal region in the structures of EF-Tu in complex with GDP and EF-Ts. The connection between domains 1 and 2 would be impossible in the GDP form, whereas the contact between the N-terminal region and the effector region via Glu71 (Asp70) might be present. It is, however, difficult to interpret exactly how the connection to the effector region is formed in E. coli EF-Tu·GDP as the only structure available of intact EF-Tu in its inactive form is of EF-Tu from T. aquaticus. The EF-Tu molecules from these two sources have no homology in the beginning of their effector regions. The structure of the effector region is strongly dependent upon the identity of the bound nucleotide (3Polekhina G. Thirup S. Kjeldgaard M. Nissen P. Lippmann C. Nyborg J. Structure. 1996; 4: 1141-1151Abstract Full Text Full Text PDF PubMed Scopus (206) Google Scholar). The effector region and β-strand b describe a curve starting at Asp39(Thr39) and ending at Thr72(Thr71). The curve is “held together” by a hydrogen bond formed between its two extremities. The effects of the Lys9 mutants on nucleotide binding could be caused by an inability of the effector region to adjust its structure properly when its anchorage to the N-terminal is lost. The effects of the Lys4 mutations on GDP/GTP binding are very small. However, it seems likely that the function of the residues 1–4 is to ensure that the residues 5–9 are in the correct conformation for entering into the network of interactions described above. This conclusion is supported by a study of mutants of the solvent-exposed Lys2 (24Mansilla F. Knudsen C.R. Laurberg M. Clark B.F.C. Protein Eng. 1997; 10: 927-934Crossref PubMed Scopus (10) Google Scholar), which are affected in their formation of the GTP complex, but to a lower extent than the similar Lys4 mutations. In summary, we draw the following conclusions on the basis of the results presented herein. Lys4 plays a role in the association of GDP and GTP probably by influencing the conformations of Arg7 and Lys9. The residue is also of some importance in aa-tRNA binding. A connection from the effector region of domain 1 to domain 2 is established through Arg7 and Lys9 of the N-terminal region, which thereby acts as a molecular “hasp” locking domains 1 and 2 together probably in a late step of nucleotide association. The presence of the hasp is important for the binding of guanine nucleotides and aa-tRNA. The complex network of interactions across the interface between domains 1 and 2 with Lys9 being a key residue thus seems to be important for the fine tuning of the dimensions of the cleft accommodating the acceptor end of aa-tRNA as well as delineating the structure of the effector region. This work illustrates satisfyingly the interrelation of structural and biochemical studies for the exploration of the biological activity of a macromolecule. We thank Karen Margrethe Nielsen and Gitte Hartvigsen for invaluable technical help and Drs. Poul Nissen and Jens Nyborg for fruitful discussions. Figs. 1 and 4 were kindly provided by Dr. Poul Nissen.