Mutagenesis of Bacterial Elongation Factor Tu at Lysine 136

Vol. 264, No . 14, Issue of May 15, pp. 8304-8309,1989 Printed in U.S. A.

THEJOURNALOF BIOLOGICAL CHEMISTRY 0 1989 by The American Society for Biochemistry andMolecular Biology, Inc.

Mutagenesis of Bacterial Elongation FactorTu at Lysine 136 A CONSERVED AMINOACID

IN GTP REGULATORY PROTEINS* (Received for publication, December 18, 1988)

Yu-Wen Hwang, Anthony Sanchez, and David L. Miller From the Molecular Biology Department, New York State Institute for Basic Research in Developmental Disabilities, Staten Island, New York 10314

We have studied the effects of specific amino acid tein-ligand interactions. EF-Tu consists of 393 amino acids replacements in EF-Tuupon the protein’s interactions and has amolecular weight of 43,000 (2). In Escherichia coli, with guaninenucleotides and elongation factor Ts(EF- EF-Tu is encoded by two distinct genes, tufA and tufB (3). TS). We found that alterations at the lysine residueof The corresponding gene products fromtufA and tufB are the Asn-Lys-Cys-Asp sequence, the guanine ring-bind- identical except for their carboxyl-terminal amino acid resiing sequence, differentially affect theprotein’s ability dues and both polypeptides are apparently fully functional to bind guanine nucleotides. Wild type EF-Tu (Lys- (3). EF-Tuis the specific target of the antibiotic aurodox ( N 136) binds GDP and GTP much more tightly than do methyl kirromycin) (4). Aurodox prevents the release of EFmany of the altered proteins. Replacing lysine by ar- Tu from ribosomes, which subsequently leads to immobilizaginine lowers theprotein’s affinity for GDP by about tion of EF-Tu on ribosomes and inhibition of protein synthe20-fold relative to the changeitsinaffinity forEF-Ts. Substitutions at residue 136by glutamine (K136Q)and sis (4). The allele sensitive to aurodox is dominant over the glutamic acid (K136E) further lower the protein rela- resistant one; therefore, phenotypic resistance to antibiotic tive affinity for GDP by factors of about 4 and 10, requires alteration in both tufA and tu@ genes (5). EF-Tu is a member of a family of guanine nucleotiderespectively. In contrast, replacement of the residue binding proteins, which includes translationalfactors (6), by isoleucine (K1361) eliminates guanine nucleotide binding as well as EF-Ts binding. Apparently, thedis- signal transduction proteins (7), and the ras gene family (8). tortion of this loop by substitution at residue 136 of a Comparison of the amino acid sequences of these functionally diverse proteins reveals a few regions with a remarkable bulky hydrophobic residue can hamper the binding for both substrates or disrupt the folding of the protein. degree of amino acid sequence similarity (9-11). These regions All altered proteins except EF-Tu(K136I) are able to have been ascribed to theguanine nucleotide-binding domain bind tRNAPhe;however, they require much higher con- (12-15). One of these highly conserved sequences, Asn-Lyscentrations of GTP than wild type EF-Tu. In minimal Cys-Asp, located at amino acid residue number 135-138 of media, Escherichia coli cells harboring plasmids en- EF-Tu, has been suggested to be the guanine ring-binding coding EF-Tu(K136E) or EF-Tu(K136Q) suffer sequence with the possibility that theside chains of Asn and growth retardation relative to cells bearing the same Asp would hydrogen bond to the 6-carbonyl and the 2-amino plasmid encoding wild type EF-Tu. Co-transformation substituents of the purine ring, respectively (12, 13). In a of these cells with a compatible plasmid bearing the previous study, the interaction between Asp-138 and the 2EF-Ts gene reverses this growthproblem. The growth amino group of guanine has been verified by demonstrating retardation effect of some of the altered proteins can that changing Asp-138 to Asn changes the base specificity of be explained by their sequestering EF-Ts. These re- EF-Tu from guanine to xanthine(16). Cysteine is nota sults indicate that EF-Ts is essential to the growth of conserved residue (9-11); presumably, its function in this E. coli and suggest a technique for studying EF-Ts mutants as well as for identifying other guanine nu- region is only to complete the loop structure that forms the purine binding pocket. On the other hand, Lys-136 is a highly cleotide exchange enzymes. conserved residue (9-11) but its role in the purine-binding domain is poorly defined. In this study, we have assessed the importance of the Lys-136 of EF-Tu by site-directed mutagenesis. Some of these altered proteinssequester the nucleoElongation factor Tu (EF-Tu)’ promotes the binding of tide exchange factor EF-Ts, a property that has interesting aminoacyl-tRNA (aa-tRNA) toribosomes during polypeptide consequences and possibilities. chain elongation. During the process, EF-Tu interacts sequentially with GTP, aa-tRNA, ribosomes, GDP, and EF-Ts MATERIALS A N D METHODS A N D RESULTS* (1).Because of its multiple functions, it provides a good model DISCUSSION for studying protein-protein, protein-nucleic acid, and proThe structural models of EF-Tu.GDP derived by x-ray * This work was supported by National Institutes of Health Grant crystallography place the guanine nucleotide in a pocket enGM30800 and the New York State Office of Mental Retardation and Developmental Disabilities. The costs of publication of this article closed by two loops containing residues 20-27 and 134-138 were defrayed in part by the payment of page charges. This article (12, 13). The former loop interacts with the pyrophosphate must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solelyto indicate this fact. The abbreviations used are: EF, elongation factor; aa-tRNA, aminoacyl-tRNA; DTT, dithiothreitol; SDS, sodium dodecyl sulfate; IPTG, isopropyl-1-thio-0-D-galactopyranoside; HPLC, high pressure liquid chromatography; bp, base pair; WT, wild type.

Portions of this paper (including “Materials and Methods” and “Results,” Tables 1-3, and Figs. 1-5) are presented in miniprint a t the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are included in the microfilm edition of the Journal that is available from Waverly Press.

8304

Mutagenesis of Lysine 136 of EF-Tu moiety; whereas the latter residues binding the guanine ring. A variety of observationssupportthese assignments. The amino acid sequences of the loops are highly conserved in all guanine nucleotide-binclingproteins, a finding which suggests that they are involved with GDP binding (9-11). Modification of Cys-137 of EF-Tu by N-ethylmaleimide inhibitsGDP binding (28). Replacement of Asp-138 with asparagine causes the protein to preferentially bind XDP rather than GDP (16). This finding establishes that Asp-138 hydrogen bonds to the exo-amino group of the guanine ring. Replacing any of the conserved residues, Gly-23, Lys-24, and Thr-25,causes inhibition of GDP binding, which confirms that .this loop is also essential for binding the n~cleotide.~ The role of Lys-136 is not well defined in the structural model. The side chain of the lysine residue i s clearly pointed to the general direction of the ribose moiety of the bound GDP and apparently is held fixed by certain interactions. It is unlikely that it directly interacts with the purine ring of the guanine nucleotide. One possibility is that it forms a salt bridge or hydrogen bond with an acidic residue or amide carbonyl to stabilize the loop or “lock in” thenucleotide (12). We have tried to assess the role of Lys-136 by site-directed mutagenesis. We avoided the problem of having to produce EF-Tu mutants in cells that require wild type EF-Tu by using the maxicell techniques, in which only plasmid-borne genes are expressed. The limitation of this approach is that the labeled proteins must be assayed by tracer techniques. Four assay procedures were employed to measure different properties of the altered EF-Tu molecules. The GDP-agarose column provided a qualitative measurement of a change in a mutant Tu’s GDP affinity relative to the wild type counterpart. Size exclusion chromatography provided an estimate of the protein’s ability to bind EF-Ts or GDP. The Phe-tRNA binding assay demonstrated whether the protein could interact with Phe-tRNA and GTP, and the aurodox-sensitivity test indicated whether the protein could function in vivo to bind aa-tRNA and ribosomes. These assay methods do not provide independent determinations of affinity constants. In particular, the size-exclusion chromatographic method gives a measure of GDP binding relative to EF-Tsbinding. Previous observations have shown that GDP oxidatively cleaved at the 2‘-3’-ribose bond binds relatively weakly to EF-Tu (29). The affinity resin prepared by reductive alkylation of the GDP-dialdehyde does not bind wild type EF-Tu quantitatively; therefore, altered EF-Tu molecules that bind GDP less tightly by a factor of 10 would not be expected to bind to GDP-agarose. We assume in this discussion that the alterations do not affect the binding energies of EF-Ts and GDP reciprocally. None of the EF-Tumolecules altered at residue 136 bound to GDP-agarose; however, three of these proteins formed complexes with EF-Ts that could be dissociated by higher concentrations of GDP. Our results indicate that Lys-136 is not absolutely required for guanine nucleotide binding. Replacement by arginine lowered the affinity constant by about a factorof 20. Eliminating the positive charge by substituting a negatively charged glutamic residue lowered the affinity for GDP by at least another factor of 10. Only substitution with isoleucine completely inactivated the protein. EF-Tusubstitutedat residue 136 with Arg(K136R), Gln(K136Q), or Glu(K136E) is able to bind Phe-tRNA in vitro; however, the process requires higher amounts of GTP. In theextreme case, the mutantK136E needs a GTP concentration in the millmolar range to promote significant complex Y. W. Hwang, P. G. McCabe, M. A. Innis, and D. L. Miller, submitted for publication.

8305

formation. In addition, the K136R and K136Q mutants seem to be able to bind aa-tRNAto ribosomes, because the aurodoxresistant host became aurodox-sensitive when either of these proteins was expressed in the cells. Although mutant K136E does not confer the aurodox sensitivity to the aurodox-resistant host, the growth of the hostharboring pTAg(K136E)was inhibited by aurodox. This phenomenon can be attributed to the drastic reduction in theaffinity of the mutantfor guanine nucleotides and the intracellular GTP concentration being only enough to partially sustain the function of mutant EFTu(K136E). Based on our observations, we do not understand why Lys-136 is conserved in all known GTP-regulatory proteins. The intracellular concentration of GTP hasbeen determined to be in the approximations of 0.3-1.0 mM for rapidly growing cells (30, 31). It does not seem that the minor rate advantage conferred by the lysine residue would permit its selection over arginine, sincethe endogenous guanine nucleotide concentration should render EF-Tu(K136R) nearly as active as itsLys-containing counterpart. The analogous mutations in theras p21 protein were found to similarly affect the protein’s affinity for guanine nucleotides. The mutation K117Q (equivalent to EF-Tu K136Q) diminished guanine nucleotide binding of the ras p21 protein by severalfold (32); while the mutant K117E (equivalent to EF-Tu K136E) displayed a 5000-fold reduction in affinity for guanine nucleotides (33). However, both mutants appear to maintain theirtransforming capability. Overproduction of either K136E or K136Q EF-Tu induces growth retardation in the host. The growth problem can be eliminated by supplying extra copies of the tsf gene. Our interpretation of this result is that EF-Tu mutantssequester EF-Ts and cause growth inhibition. The finding that the K136E and K136Q mutants existed primarily as EF-Ts-EFTu complexes further support our conclusions. The reason for the sequestering of EF-Ts must bethat each mutation has little or no effect on the ability of EF-Tu to bind EF-Ts but reduces the protein’s affinity for guanine nucleotides. In the cycle of reactions involved in binding aa-tRNA toribosomes, EF-Ts catalyzes the exchange of guanine nucleotide bound to EF-Tu (1). It is known from kinetic analysis and binding studiesthatthe nucleotide-exchange reaction proceeds through an EF-Ts.EF-Tu complex (1, 34). Apparently, the endogenous guanine nucleotide concentration cannot overcome the reduction inguanine nucleotide affinity of the K136E and K136Q mutants and therefore EF-Ts remains trapped as the EF-Ts.EF-Tu(mutant)complex, which is unable to be recycled under physiological conditions. Calculation of the uncatalyzed dissociation rate of GDP from EF-Tu indicates that thisprocess is too slow to support the observed rate of protein synthesis (1);however, the importance of EFTS to the growth of E. coli has been conjectural. Our observations support the conclusion that EF-Ts, a protein whose only known function is to catalyze EF-Tu-guanine nucleotide exchange, is an essential elementfor growth. The intracellular concentration of EF-Ts is positively regulated by growth rate (35); therefore, it is reasonable that growth inhibition conferred by mutants would be more prominent on medium K than Y T medium (a richer medium) because of the smaller intracellular EF-Ts pool. EF-Tu K136E and K136Q can be classified as dominant negative mutants (36). They interferewith normal cell growth by interrupting the recycling of EF-Ts, the guanine nucleotide-exchange enzyme of EF-Tu. Because of many conserved features shared by guanine nucleotide-binding proteins, our finding should be generally applicable to other proteinsin the family. Theoretically, a mutant with a significant reduction

Mutagenesis of L3{sine 136 of EF-Tu

8306

in its affinity for guanine nucleotides that still maintainedits ability to interact with guanine nucleotide exchange factor should be a dominant negative regulator. Interestingly, overproduction of YPT mutant protein N1211 (corresponding to residue 135 of EF-Tu) is dominantly lethal to host Saccharomyces cereuisiae (37). Because the Asn and Lys residues are adjacent, it is possible that YPT toxicity involves a mechanism similar to the one we proposed for the EF-Ts/EF-Tu cycle. This type of dominant negative mutant might prove to be quite general in GTP-regulatory proteins and one might be able to utilize itto identify otherputative nucleotide exchange enzymes for GTP-regulatory proteins. The natureof the EF-Ts-binding siteon EF-Tu has not yet been defined. The kinetic mechanism requires a EF-Ts-EFTu. GDP complex with separate but interacting binding sites for EF-Ts and GDP (1, 34). Several observations suggested that the sites should lie near each other since both GDP and EF-Ts binding prevent modification of EF-Tu atCys-137 by N-ethylmaleimide (28). In this study we found that some amino acid replacements at residue 136 do not affect EF-Ts binding, from which we conclude that this residue is not a site of EF-Ts binding. The finding that substitution of isoleucine at this site prevents binding of both GDP and EF-Ts may indicate that EF-Tsrecognizes someother featureof the GDP-binding site that is distorted by this hydrophobic residue. The discovery of an EF-Ts-sequestering EF-Tu mutant will allow one to probe the EF-Tsbinding site of EF-Tu. Acknowledgments-We thank Drs. Leendert Bosch for strains LBE2012 and LBE2050, James Friesen for plasmid pDB9, Paul Schimmel for plasmid pMT101, and Hoffmann-LaRoche Inc. for aurodox. We also thank Drs. Richard Kascsak and Marshall Elzinga for their critical reading of this manuscript. REFERENCES 1. Miller, D. L., and Weisshach, H. (1977) in Molecular Mechanism of Protein Biosynthesis (Weissbach, H., and Pestka, S., eds) pp. 323-373, Academic Press, Orlando, FL 2. Arai, K., Clark, B. F. C., Duffy, L., Jones, M. D., Kaziro, Y., Laursen, R. A., L'Italien, J., Miller, D. L., Nagarkatti, S., Nakamura, S., Nielsen, K. M., Petersen, P. E., Takahashi, K., and Wade, M. (1980) Proc. Natl. Acad. Sci. U. S. A. 77,13261330 3. Bosch, L., Karrl, B., Van Der Meide, P. H., Duistenvinkel, F. J., and Van Noort, J. M. (1983) Prog. Nucleic Acids Res. Mol. Biol. 30,91-126 4. Parmegaiani, A., and Swart, G. W. M. (1985) Annu. Rev. Microbid. 39, 557-577 5. Van De Klundert. J. A. M.. Van Der Meide, P. H.. Van De Putte, P., andBosch,'L. (1978) Proc. Natl. Acad. Sci. U. S. A . 75, 4470-4473 6. Kaziro, Y. (1978) Biochim. Biophys. Acta 505,95-127 7. Gilman, A. G. (1987) Annu. Reu. Biochern. 56,615-649 ~

8. Barhacid, M. (1987) Annu. Rev. Biochem. 56, 779-827 9. Halliday, K. R. (1984) J . Cyclic Nucleotide Protein Phosphorylation Res. 9,435-448 10. Leberman, R., and Egner, U.(1984) EMBO J 3,339-341 11. Dever, T. E., Glynias, M. J., and Merrick, W.C. (1987) Proc. Natl. Acad. Sci. U. S. A. 84, 1814-1818 12. La Cour, T. F. M., Nyhorg, J., Thirup, S., and Clark, B. F. C. (1985) EMBO J 4,2385-2388 13. Jurnak, F. (1985) Science 230,32-36 14. McCormick, F., Clark, B. F. C., La Cour, T. F. M., Kjeldgaard, M., Norskov-Lauritsen, L., and Nyborg, J. (1985) Science 230, 78-82 15. De Vos, A. E., Tong, L., Milburn, M. V., Matias, P. M., Jancarik, J., Noguchi, S., Nishimura, S., Miura, K., Ohtsuka, E., and Kim, S.-H. (1988) Science 239,888-893 16. Hwang, Y.-W., and Miller, D. L. (1987) J. Biol. Chem. 262, 13081-13085 17. Miller, J. H. (1972) Experiments in MolecularGenetics,Cold Spring Harbor Laboratory, Cold Spring Harbor, NY 18. Zoller, M. J., and Smith, M. (1983) Methods Enzymol. 100,468500 19. Mead, D.A., Szczesna-Skprupa, E., and Kemper, B. (1986) Protein Eng. 1,67-74 20. Leive, L. (1974) Ann. N. Y. Acad. Sci. 235, 109-129 21. Miller, D. L., and Weissbach, H. (1974) Methods Enzymol. 30, 219-232 22. Meacock, P. A., and Cohen, S. N. (1979) Mol. Gen. Genet. 174, 135-147 23. Jasin, M., and Schimmel, P. (1984) J. Bacteriol. 159, 783-786 24. Balbas, P., Soberon, X., Bolivar, F., and Rodriguez, R. L. (1988) in Vectors, a Survey of Molecular Cloning Vectors and Their Uses (Rodriguez, R. L., and Denhardt, D. T. eds) Butterworth Publishers, Woburn, MA 25. An, G., Bendiak, D. S., Mamelak, L. A., and Friesen, J. D. (1981) Nucleic Acids Res.9, 4163-4172 26. Sancar, A., Hack, A. M., and Rupp, W. D. (1979) J. Bacteriol. 137,692-693 27. Zengel, J. M., and Lindahl, L.(1982) Mol. Gen. Genet. 185, 487492 28. Miller, D.L., Hachmann, J., and Weissbach, H. (1971) Arch. Biochem. Biophys. 144, 115-121 29. Bodley, J . W., and Gordon, J. (1974) BiochemistTy13,3401-3405 30. Friesen, J. D., Fiil, N. P., von Meyenburg, K. (1975) J . Biol. Chem. 250,304-309 31. Villadsen, I. S., and Michelsen, 0.(1977) J . Bacteriol. 130, 136143 32. Clanton, D. J., Hattori, S., and Shih, T. Y. (1986) Proc. Natl. Acad. Sci. U. S. A. 8 3 , 5076-5080 33. Der, J. C., Pan, B.-T., and Cooper, G. M. (1986) Mol. Cell. Biol. 6,3291-3294 34. Hwang, Y. W., and Miller, D. L. (1985) J . Biol. Chem. 260, 11498-11502 35. Pedersen, S., Bloch, P., Reeh, S., and Neidhardt, F. (1978) Cell 14,179-190 36. Herskowitz, I. (1987) Nature 329,219-222 37. Schmitt, H. D., Puzicha, M., and Gallwitz, D.(1988)Cell 53, 635-647 38. Yanisch-Perron, C., Vieira, J., and Messing, J. (1985) Gene (Arnst.) 33,103-109

8307

Mutagenesis of Lysine 136 of EF-Tu Supplementary Material To Mutagenesis ofBacterial Elongation Factor Tuat Lysine-136, a Conserved Amino Acid in GTP-Regulatory Proteins Yu-Wen Hwang, Anthony Sanchez andDavid L. Miller Materials and Methods

.

.

Bacterial strains used and constructed in this study are listed in Table 1 with their relevant genotypes. PI phage transduction was carried out according to Miller (17) but the transduced cells were plated directly without using top agar. YT medium wntains Sg Bacto-Tryptone, 5g yeast extract and 5g NaCl per liter of medium. K medium is composed of 1% casamino acids and 0.5% glucose in M9 medium base (17). A l l experiments involving antibiotics were performed using YT medium except when otherwise indicated. Ampicillin was supplemented at 20 uglml, chloramphenicol at 20 ug/ml,streptomycin at 100 ug/mlandaurodoxat 100 ug/ml. Oligonucleotldeswere synthesized by the beta-qanophosphoramidite chemistry in an Applied Biosystem automatic synthesizer model 380B. Each oligonucleotide was partially purified by filtration through a lOcm Sephadex G-25 w l u m n before use.

.

-

Mutant Cbnstructions. Site-specific mutanls were wnstructed by a modification of the procedure of Zoller and Smith (18) as described previously (16). Single stranded DNA of phage M13TA18, alufA-containing MI3 clone (16). was usedasthethetemplate. Mutagen~c primers were: used K136E (TCGCAlTCGlTCAGGAA), K1361 (TCGCATATGTTCAGGAA). K136Q (CGCAlTGGTTCAGGAAC) and K136R (TCGCATQGTKAGGAA).The mismatched position in each oligonucleotide is underlined. Subsequently, each &A mutant was subcloned into plasmid pTzl8U (19) by insertion of the M13 phage EcoRI-Hindlll restriction fragment into the E w R I and Hind111 sites of pTZlXU.

.plasmid pTS32, plasmid pMTlOl containing a lac promotercontrolled rsf (EF-Ts) gene, was wnstructed by the procedures described below (Figure 1). pMTlOl was derived from pPM103 (22) and contained a temperature sensitive repliwn (23). This plasmid is wmpatible with the ColEl wntaining plasmid (24). The 1.4 Kbp HindlII-PstI fragment, wntaining the entire d gene, t h e ~ ~ s B - intergenic rsf region and a b u t 18 hp of the3'end of the of rpsB gene,was excised from pDB9 (25) and insened into the Hind111 and Pstl sites of pUC18 to yield pTSR1. In pTSR1, the rsf gene is in reverse orientation with respect to the lac promoter. The 1.3 Kbp Sal1 and partial EcoRl fragment

t

The procedure used for maxicell expression was asdescribed Maxicell .(16) except that bacterial strain MV1190 was used instead of strain SR58. The presence of k l q in strain MV1190 minimizes thegrowthproblem of somemutants(see Results). However. strain MV1190 is considerably more resistant to UV than strain SR58; a2-3 times higher dose of UV exposure is required to inactivate gene expression from thehost genome. e Bindine-. The EF-Ts binding assay involved the addition of excess purified EF-Ts to the testiig lysate and followed by elution with a buffer (50 mM Tris-HCI. pH8.0, IO mM MgCl and 1 mM DTT) fromaPharmaciaHR-12 Superose gel filtration column attached to'a HPLC. The flow rate was set at 0.5 ml/min. and one fraction per min was collected. Under these conditions, the retentiontimes were 26 min for EF-Ts.EF-Tu wmplex and 28 mm for EF-Tu. Complex formation behueen EF-Tu and EF-Ts was identified by the radioactive peak shifting from fraction 28 to 26. The EF-Ts EF-Tu complex-guaninenucleotide dissociation assay was performed by a huostep procedure exactly as described (16). resistant -ox Sensitivin A s u Strain HW123, a strain containing an aurodox 3 on the strainJM105 background, was used as the u f A gene ( M A R ) and a deleted~ 8 gene host for testing the aurodox sensitwity of residue number 136 mutants. The construction of strain HW123 was outlined below. The P1 phage prepared on the strain LBE2012 ( 5 ) was used to transduce strain LBEZOSO (3) to aurodox resistance at 32OC. The aurodox resistant cells were then stabilized by plating on aurodox containing plates at 42 OC. The sunvors selected at elevated temperature and In the presence of aurodox should contain deletions extending from phageMu into part or all of the t u!B gene. The strain thatis able togrow on aurodox at 42 "C was designated strain HW122. Next. the P1 phage obtained on strain JM103 was used to transduce strain LBE2012 t o streptomycin resistance. About 90% of streptomycin resistant transductants becane aurodox sensitive. This is consistent with the previous reportaboutthecotransductionfrequency of Q,$L andtufAgenes (5). A streptomycin resistanf cell that still remainedaurodox resistant was designatedstrain HWl2O and used as the donor to transduce strain JM105 to streptomycin resistance by P I phage. Since the Q,$L locus is tightly associated with tllfA, a very high proportion of streptomycin resistant cells resulting from the transduction should receive the IvfA locus from strain HWlZO (originally from strain LBE2012). Several of the streptomycin rFsistant cells were p l e d and designated strain HW121. Finally, strain HW121 was transduced to aurodox resistance by PI phage prepared on the strain HW122. The resulting aurodox strain

Figure 1. Construction of v e n o n containing lac promoter-wntrolled rsf gene.The procedures for vector wnstruction were described in Materials and Methods. The figures were not drawn tothescaleandthe arrows indicatethedirection of transcription or translation. ~ G P , promoter; WBP I Q ~ Bpromoter; h p R , ampicillin resistance gene; CamR, chlorampheniwl resistance pen;; TetR, tetracycline resistance gene. The restriction enzyme sites are, C, Clal;H. Hindlll; N, Narl; P. Pstl; R, EcoRI;S , Sal1 and Sm, Smal.

Table 1. Bacterial Strains Used in This Study

Relevant Reference Strain Genotypes or Source

which contains the rsf gene was lhen isolated from pTSRl and cloned into the EcoRl and Sal1 sites of M13mp18. This construct (M13TS) places the M g e n e in the right orientation with respect to the lac promoter and removes the 3' end of the rprB gene and part of the intergenic region. Subsequently, pTs32 was obtained by cloning theNarl-Sal1 fragment from M13TS, which includes the lac promoter and downstream rsf gene into the Sal1 and Clal (partially digested) sites of pMT101. The intergenic region between @B and does not possess promoter activiry (25); therefore. the expression of rsf is wntrolled by the k promoter in pTS32. Because part of the tetracycline resistance gene was remowd during the wnstruction, pTS32 should only confer chloramphenicol resistanceto the host,

a

JMI03 JM105 SRSM MV1190 LBE2012 LBEZOSO HW120 HWl2l HW122 HW123

that possessed the tufA locus fromstrain LBE2012 adeleted tu!B gene from strain HW122 andthe g e g fromstrain JM105 was designatedstrainHWI23. The aurodox sensitivity assay was carried out by transforming strain HW123 to ampicillin resistance with testing plasmids and then scoring the transformants for aurodox sensitivity on YT plates containing 100 ug/ml of aurodox. EDTA at 2 mM was included in the plates to prevent resistant cells from growing due totranspart failure (20).

. . w.

This experiment was performed as previously described (16). GDP-agarose was purchased from Sigma Chemical Co. and had a GDP content of 4.1 umole/ml of gel. . . . Ternary corn lex formation was assayed by the n i t r o c e n m i k d below. &]-labeled EF-TU'Sfrom wild type and mutant maxicell lysales were partially purified as the EF-Ts.EF-Tu complex by gel filtration before use. Because it lacked the ability to bind EF-Ts. mutant K1361 was purifed as EF-Tu andthensupplemented with excess amountc ofEF.Ts. T h e EF-TiEF-Tu complex was incubated in 1 0 0 UL ofreaction mixture containing SO mM Tris-HCI. pH 7.5 10 mM MgCl 50 mM NH,CI. 1 mM D T , 4.5 mM phosphoenolpyruvate, 1 ug pymvate'kinase and thf indicared amounts of GTP at 37 OC for 5 min. Then, 140 pmoles of Phe-tRNA were added and the mixture was incubated on ice for an additional 2 min before passing through Millipore type HA filter. A set of controls lacking the Phe-tRNA was similarly treated. The fraction of EF-Tu reacting wulth Phe-tRNA was calculated from 1-(CPM(ttRNA]/CPM(. tRNA]).where C P M [ t t R N A ] is thefilter-bound radtoactivity in the mixture containing Phe-rRNA and CPM(-tRNA]is that of control mixture without Phe-tRNA.

Results Four single point mutations at position 136 of EF-Tu were constructed in the rvfA gene by oligonucleotide site-directed mutagenesis as described in Materials and Methods. Each of the UfA mutants was then subcloned from MI3 phage into the E w R l and Hind111 sites of plasmid vecior pTZlSU (19) to facilitate iriyiy~expression of mutant proteins by the maxicell protocol (26). In addition. this wnstruct also places the lufA gene under wntrol of the lac promoter. Maxicell strain SR58 was transformed with mutant-harboring plasmids. Transformantswereselectedon YT platescontaining ampicillin. ThetransformedSR58 cells were then recultured i n K medium for maxicell processing as described previously (16). The presence of plasmids containing either wild type (pTA9) (Figure 2). K1361 (data not shown) or K136R (data not shown) rufA gene in strain SRSX does not confer a detectable growth disadvantage to the host; however. strain SRSX harboring mutant K136E or K136Q in a similar plasmid construct suffers growth retardation. The growth problem is much more severe with plasmid pTA9(K136E) than with pTAAP(K136Q). The growth retardation is medium dependent; the differenceis slight in YT medium, but it is prominent in K medium although the retardation is not permanent (Figure 2). Furthermore. this growth problem can be largely eliminated by the presence of b l q . a strong lac repressor. (streaks 3 and 7 of Figure 2) or by cutting out the EcoRl and MstlI fragment of the EF-G geneF e n d (data not

Mutagenesis of Lysine 136 of EF-Tu

8308 YT

+ AMP

K t AMP

Figure 2. Growth study ofmutant EF-Tu harboring cells. Cells harboringtheindicated at .~ 32 OC ." for 30 ~. IYT 2nd I ."S IK olasmidk) were streaked on olates and incubated.~ . nlntri ____, plate) hours. Both plates wniain 20 ug/ml ampicillin. 1 pTZ18U/SR58' 2 pTA9/SW&'ipTA9(K136Q)/MV1190: 4, pTA9(K136Q)/SR58; 5, p?A9(K136Q)/Sd5{ t pMTIO1: 6: Same as 5 but with pTS32: 7. pTA9(K136E)/MV1190; 8, pTA(K136E)/SR58; 9 ~ olasmid oTZ18u' pTA9(K136E)/SR58 + PMTIOI; 10. same as 9 but with oTS32. D T A is containing wild type u ~ i Agene. Plasmids pTA9(K136Q) and 'pTN(Ki36E) a;e pTA9 containingthe indicated mutations.pMTlOl is apTZI8Ucompatible plasmid used for constructlng pTS32 (see Figure I). ~~

"

__

0

1

242628

"

" ~ " " ~ " " 30 32242628303224 26 2 8 3 0 3 2 FRACTION NUMBER

600

shown), which might wntain promoter activities (27). Clearly, overproduction of EFTu(K136E) o r EF-Tu(K136Q) is harmful to the cells. Thisobservation suggested that mutants EF-Tu(K136E) and EF-Tu(K136Q) might sequester or inactivate a certain essential element of cell. I n order to circumvent thegrowth problem imposed by mutations K136E and K136Q. a klqcontaining =A strain MV1190 was used instead of strain SR58 as the maxicell host. TheEF-Tuproteinswerelabeled with [35S]-methionine andreleased by lysozyme and osmotic shock. Polyacrylamide gel electrophoresis of the lysates revealed 43 KDa EF-Tu and 31 KDa beta-lactamase (encoded by the ampicillin resistance gene) as the two major labeled bands produced by wild type as well as mutant U$A genes (Figure 3). The maxicell lysates were then subjected to U assays.

The ability of mutantEF-Tu to bind EF-Ts was examined by gel filtration chromatography. Under the conditions employed,EF-Tu(WT) eluted in fraction number 28 (Figure 4A-I) and beta-lactamase emerged two fractions later in fraction 30 (Figure 4A-11). Since the amount of EF-TUin the wild type lysate is much higher than beta-lactamase (Lane 1 of Figure 3). the beta-lactamase could be observed in the chromatogram only if the EF-Tu peak had shifted to an earlier fraction (fraction number 26) by forming the EF-Tu EF-7s complex (Figure 4A-11). Lysates containingmutant K1361 (data not shown) and K136R (Figure 5) exhibited elution profiles similar to wild type EF-Tu; however, mutants K136E and K136Q showed differentchromatographic profiles. Instead of elutingat fraction number 28, mutant EF-Tu(K136Q) peaked at fraction number 27 (Figure 4C-I); whereas, mutantEF-Tu(K136E)appearedatfractionnumber 26 (Figure 48-1). In addition,the chromatogramobtainedfrommutant K136E crude I r a t e showed a prominentbetalactamase peak which could be readily understood by examining the ratio of EF-Tuto betalactamase in Figure 3. Addition of excess EF-Ts to the crude maxicell lysates shifted the EF-Tu peak of mutants K136R (Figure 5) and K136Q (Figure 4C-II) to fraction number 26, which corresponded to the EF-Ts.EF-Tu complex. But thesame treatment failed to change the chromatographic profile for mutants K1361 (data not shown) and K136E (Figure 48-11), This observation demonstrates that mutantKt361 is unable to bind EF-Ts.

o

2426283032

l1

h 24

d

-2628

30 3 2 24 2628

30 3 2

FRACTION NUMBER

500 6 o l "

200

100

1

2

3

4

5

I 1 242628

4 5 q

.

A+. ui"

I

Fi re 3 Polyacrylamide gel electrophoresis of maxicell lysates. Proteins were labeled with [3#-me;hionine by the maxicell protocol as described in Materials and Methods. Labeled proteins were electrophoresed in a 10% polyacrylamide-SDS gel and visualized by autoradiography. EF-Tu has a molecular weight of 43 KDa. The band near the 30 KDa marker is beta-lactamase e n d e d by the ampicillin resistance gene. Lane I, WT; lane 2, K1361; lane 3, K136R. lane 4. K136Q and lane 5. K136E.

30 32242628

30 3 2 24 26 2 8 30 3 2

FRACTION NUMBER

Figure 4. Gel filtration chromatography of EF-Tu species. The assays were performed as described in Materials and Methods. A-I, EF-Tu(WT) alone; A-II, E F - T u ( W ) plus EF-Ts; A-Ill,sameas A-Il but eluted with 20 uM GDP. B-I, EF-Tu(KI36E)alone; B-11, EFTu(K136E) plus EF-Ts; 8-111. same as 6-11 but eluted with 1 mM GDP. C-I. EF-Tu(K136Q) alone; C-11, EF-Tu(K136Q) plus EF-TS; C-Ill. same as C-II but eluted with 200 uM GDP. For convience of presentation, the first 23 fractions, which contained no radioactivity, were omitted from the charts.

We then tested whether the elution ofEF-Tu(K136E) or EF-Tu(K136Q) at the earlier fraction might be due to their existing as EF-Ts.EF-Tu complexes in the cell lysates. This was accomplished by eluting the crude maxicell lysate in the presence of a concentration of GDP sufficient to dissociate the EF-Ts.EF-Tu complex. shown in Figure 46-Ill.the incorporation of 1 mM GDP in the elution buffer shifted the peak of EF-Tu(K136E) from fraction number 26 to 28. Mutant EF-Tu(K136Q) exhibited similar behavior but with a These results indicate thatmutantEFlower concentration of GDP (Figure 4C-Ill). Tu(K136E) and EF-Tu(K136Q) existed as the EF-Ts.EF-Tu complexes in crude maxicell 50% EF-Ts.EF-Tu lysates. Subsequently, we measuredthe GDP concentrationfor dissociation (peak at fraction 27 and with equal amounts distributed in fractions 26 and 28) for all the mutants that were able to bind EF-Ts. The results are summarized in Table 2 and the chromatogram of one of the mutants (K136R) is shown in Figure 5. Even the consewed change (Lys to A r g ) significantly reduced EF-Tu's affinity forguanine nucleotides. It is likely thatthe reduction in affinityforGDP is responsible for EF-Tu(K136E) and EFTu(K136Q) existing predominantly as EF-Ts,EF-Tu complexes. An increase in EF-Tu's affinity for EF-TS might appearasadecrease in EF-Tu's affinity for GDP in the complex dissociation assay: theGDPconcentration required to dissociate the EF-Ts.EF-Tu complex is not a true measurement of the protein's affinity for GDP. We attempted to directly estimate each protein's affinity for GDP by measuring the ability of each mutant EF-Tu to bind to a GDP-agarose column; however, we were unable to detect that any bound to the affinity column. The GDP concentration producing 50% complex dissociation for mutant EF-Tu(V20G) is about 5 uM and this mutation reduces the protein's affinity for GDP-agarose to about70%of wild type (Hwang, Y.W. et.al. submitted). TheGDP concentrationsfor 50% complex dissociation for the mutations reponed here are all far greater than 5 uM (Table 2); therefore, it is likely that none of the residue 136 mutants would bind to GDP-agarose under the conditionsemployed. was assessed by introducing a The capability of eachmutant to function plasmid containing a UfA mutation into the aurodox resistant strain and then testing for its ability to confer aurodox sensitivity to the host strain. The mode of action of aurodox requires EF-Tu to bind GTP. aa-tRNA and antibiotic (4). This quaternary complex then binds irreversibly to ribosomes and inactivates translation (4). Aurodox sensitivity is

Mutagenesis of Lysine 136 of EF-Tu E

0

C

0

A

c

I-

h 24

20

8309

We also examined the interaction of mutant EF-TUwith Phe-tRNA. At 10 uM GTP, a concentration normally employed for the Phe-tRNA binding assay, mutants K136E and K1361 did not react with Phe-tRNA whereas mutants K136Q and K136R formed barely detectableamountsofternary complex. In contrast, wild type EF-TU reacted with PhetRNA extensively under these conditions (Table 3). The inability ofthealtered EF-Tu molecules lo bind Phe-tRNA under these wnditions might be due IO their reduced affinities for guanine nucleotides (Table 2); therefore. we measured ternary complex formation at higher concentrations of GTP. As shown in Table 3, at higher concentration of GTP mutants K1360, KIMR and even K136E were able to form ternary complexes with PhetRNA. In addition, the results show that the concentration of GTP required for effective ternary wmplex formation is proportional to the mutant's affinity for GDP (Tables 2 and 3). For example, mutant K136R has higher affinity for GDP than K l M Q and it formed ternary complex more effcently thanK136Q at a given GTP concentration. Nevertheless, The GTP concentrations needed for mutants toform comparable amounts of ternary wmplex as wild type were much higher than the 10 uM GTP required by wild type EF-Tu. Again. mutant K1361 was found to be completely devoid of Phe-tRNAbinding activity.

28 50

FRACTION NUYMER

Figure 5 . Interaction of EF-Tu(KI36R) with EF-Ts and GDP. The assays were performed by a two-step double-labelling procedure as previously described (16). Mutant K136R lysate (closed circles) was prepared by labeling maxicells with ["S]-methionine (see Materials and Methods). ['H]-EF-Tu(WT) (open circles) was incorporated in the assays and used as an internal wntrol. A no GDP; 8-E.1.5, 20, and 40 uM GDP. respectively.

Table 3. Phe-tRNA Binding Activities of WT and Residue 136 Mutants

[GTP](uM)

WT

Ternaty Complex Formed (%) K136E K136I K136Q

Table 2. Properties of W and Residue 136 Mutants 10

30 Mutant

[GDP],,,' (uM)

Aurodox Sensitivity

GDP-agarose Binding

100

200 500 loo0

w K136E K1361 K136Q K136R

1.O 200

N/A'* Bo

20

Yes, No No Yes Yes

6 nd nd nd nd nd

6

K136R

0 nd

0

4

14

nd

10

24 57

0

2

6 16

0

36 43

1

56

24

ndnd

nd

nd nd

+

+ GDP concentration for 50% EF-TsEF-Tu cpmplex dissociation. * Growth of the host c e l l was inhibited by aurodox. This number can not be determined bacause mutant K1361 is unable to bind EF-Ts.

..

dominant over theresistant trait (5); therefore, converting the antibiotic resistant phenotype to a sensitive one can be uwd as an indicator for a fully functional EF-Tu. Aurodox resistant strain LBE2012 harboring either the K136E or the K136Q mutant gave inmnsistent assay results: therefore, we employed the same strategy as the one used in the maxicell expression procedure by WnstNctinga Mq-wntaining aurodox-resistant strain. Strain HW123, a strain containing the aurodox resistant tufA gene from strain LBE2012 and a deleted tufB gene in strain JM105 background. was WnstNcted (see Materials and Methods) and used as the host strain for testing aurodox sensitivity. The results of the aurodox experiment are summarized in Table 2. Mutants KlMO and K136R were able to mnvert the aurodoxresistant phenotype of strain HW123 to a sensitive one; however. mutant K136I was unable t o d o so. The result of the mutant K1361 experiment is consistent with the indtrp obxlvation; apparently, the substitution of isoleucine for lysine-136 is detrimental to the proper function of EF-TU. Mutant K136E was also unable to convert strain HW123 to the sensitive phenotype; nevertheless, it inhibited growth of host in the presence of aurodox.

nd. not determined

Since EF-TU from each mutant that caused a growth problem (KIME and K136Q) existed primarily as the EF-Ts.EF-Tu complex in crude maxicell lysates, it is rearonable to speculate that EF-Ts is the sequestered essential element. If EF-Ts is the sequestered factor, one should be able to obviate the growth retardation by supplementing to e l l s with e x e s amounts of EF-Ts. This h p t h e s i s was tested by introducing plasmid pTS32 (see Figure I), a plasmid containing a plZ18U wmpatible replicon and a lac promoterantrolled trf gene, intostrain SRS8 containing plasmid pTAg(K136E) or pTA9(Kl36Q). Strain SR58 cells harboring two plasmids were then rreened for the ability of EF-Ts expression to relieve the growth problem. Cells resistant to chloramphenicol (encoded by pTS32) and ampicillin (enwded by pTZ18U) were selected and screened for growth on YT and K medium plates. As shown in Figure 2, the presence of plasmid pTS32is able toeliminatethe growth problem conferred by either mutant EF-Tu(K136E) (streak 10 of Figure 2) or EFTu(K136Q) (streak 6 of Figure 2). Ontheotherhand,theparent vector for pTS32 construction was found to be ineffective in curing the growth problem (streak S and 9 of Figure 2). IFTG was not required in the process; presumably, endogenous lac repressors have been titrated out by plasmid-borne Lar operators. Since pTS32 wntains a temperature sensitive replicon, we also performed the same experiment atthe non-permissive temperature. The advantage that pTS32 confers on the K136E or the K136Q harboring strain SRS6 was found to be diminished; but, not completely lost (data not shorn). These observations demonstrate that EF-Ts is the essential factor that becomes limiting in alls harboringeither K136E or K136Q mutant and leadsto growth retardation.