An in vivo Approach to tRNA Identity Citation Normanly, Jennifer (1989) An in vivo Approach to tRNA Identity. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/s9xt-6f20. https://resolver.caltech.edu/CaltechTHESIS:06032013-145407392 Abstract A leucine-inserting tRNA has been transformed into a serine-inserting tRNA by changing 12 nucleotides. Only 8 of the 12 changes are required to effect the conversion of the leucine tRNA to serine tRNA identity. The 8 essential changes reside in basepair 11-24 in the D stem, basepairs 3-70, 2-71 and nucleotides 72 and 73, all of the acceptor stem. Functional amber suppressor tRNA genes were generated for 14 species of tRNA in E. coli , and their amino acid specificities determined. The suppressors can be classified into three groups, based upon their specificities. Class I suppressors, tRNA Ala2 CUA , tRNA GlyU CUA , tRNA HisA CUA , tRNA Lys CUA , and tRNA ProH CUA , inserted the predicted amino acid. The Class II suppressors, tRNA GluA CUA , tRNA GlyT CUA , and tRNA Ile1 CUA were either partially or predominantly mischarged by the glutamine aminoacyl tRNA synthetase (AAS). The Class III suppressors, tRNA Arg CUA , tRNA AspM CUA , tRNA Ile2 CUA , tRNA Thr2 CUA , tRNA Met(m) CUA and tRNA Val CUA inserted predominantly lysine. Item Type: Thesis (Dissertation (Ph.D.)) Subject Keywords: Biology Degree Grantor: California Institute of Technology Division: Biology Major Option: Biology Awards: Lawrence L. and Audrey W. Ferguson Prize, 1989 Thesis Availability: Public (worldwide access) Research Advisor(s): Abelson, John N. Thesis Committee: Abelson, John N. (chair) Dervan, Peter B. Emr, Scott D. Richards, John H. Simon, Melvin I. Defense Date: 20 December 1988 Record Number: CaltechTHESIS:06032013-145407392 Persistent URL: https://resolver.caltech.edu/CaltechTHESIS:06032013-145407392 DOI: 10.7907/s9xt-6f20 Related URLs: URL URL Type Description https://doi.org/10.1146/annurev.bi.58.070189.005121 DOI Article adapted for Introduction. https://doi.org/10.1038/321213a0 DOI Article adapted for Chapter I. https://doi.org/10.1073/pnas.83.17.6548 DOI Article adapted for Chapter III. Default Usage Policy: No commercial reproduction, distribution, display or performance rights in this work are provided. ID Code: 7826 Collection: CaltechTHESIS Deposited By: Benjamin Perez Deposited On: 03 Jun 2013 22:23 Last Modified: 17 Sep 2021 20:53 Thesis Files Preview PDF - Final Version See Usage Policy. 31MB Repository Staff Only: item control page

AN IN VIVO APPROACH TO tRNA IDENTITY Thesis by Jennifer Normanly In Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy California Institute of Technology Pasadena, California 1989 (Submitted December 20, 1988) ii ACKNOWLEDGEMENTS I may never have contemplated pursuing a Ph.D. in Biology had it not been for the support and guidance of several people to whom I am grateful; Steve Wilson for suggesting that I go into science in the first place, and who provided me with constant encouragement and moral support, Frank Talamantes for giving me a shove in the right direction, Harry Noller for letting me loose in his lab, and Jeff Prince who redefines the word meticulous, and who set the standards for precision which I can only hope to live up to. I also wish to mention my parents, Ardis and Jerry, and siblings Brian and Colleen, who, while they where never quite sure exactly what I was doing, cheerfully supported me anyway, my aunt Nora who kept me well fed through college, and my grandmother Laura Naur who bankrolled me during most of my undergraduate days. My time spent as a graduate student has been enjoyable, made so by my various labmates and associates; my dear friends Eric Phizicky and Beth Grayhack, who took me under their wing, and Joey Zwass, a truely zany person, who provided many amusing diversions. Richard Ogden was a great deal of help to me early in my graduate career, as were Richard Schwartz and Peter Johnson. David Horowitz, who distracted me from my research countless times with stimmulating debate, is to be commended for his unflagging efforts to keep me grammatically correct. George Komatsoulis has been a very amiable labmate, always quick to be of help when asked. I have been enriched for having had the opportunity to collaborate with Jeffrey Miller, who not only added to my development as a scientist, but made sure that the dreary diet afforded by a graduate student stipend, was supplemented with more than one memorable meal. I have also enjoyed knowing fellow graduate student, Lynn Kleina, who slogged it out with me in the trenches of DHFR purification. Several members of the staff at Caltech, Nancy Gill, Patricia Bateman, Mary Marsh, Connie Katz and Cathy Bragg, have very graciously dealt with this hapless graduate student, ensuring smooth sailing for me in a beaurocratic sense. Finally, I am grateful to my advisor, John Abelson, who has enthusiastically supported and guided me in my research, while allowing me a great deal of indepedence. His intrepid approach to science is one I hope to emulate. iii ABSlRACf A leucine-inserting tRNA has been transformed into a serine-inserting tRNA by changing 12 nucleotides. Only 8 of the 12 changes are required to effect the conversion of the leucine tRNA to serine tRNA identity. The 8 essential changes reside in basepair 11-24 in the D stem, basepairs 3-70, 2-71 and nucleotides 72 and 73, all of the acceptor stem. t ' Functional amber suppressor tRNA genes were generated for 14 species of tRNA in E. coli, and their amino acid specificities determined. The suppressors can be classified into three groups, based upon their specificities. Class I suppressors, tRNAAla2cuA, tRNAGlyUCUA• tRNAHisACUA, tRNALysCUA, and tRNAProHCUA, inserted the predicted amino acid. The Class II suppressors, tRNAGluACUA , tRNAGlyTCUA, and tRNAllelCUA were either partially or predominantly mischarged by the glutamine aminoacyl tRNA synthetase (AAS). The Class III suppressors, tRNAAr&cuA, tRNAAspMcuA, tRNAlle2cuA, tRNAThr2CUA, tRNAMet(m)CUA and tRNAValCUA inserted predominantly lysine. iv TABLE OF CONfENTS Acknowledgements ii Abstract iii Introduction tRNA Identity 1 Chapter! Changing the identity of a tRNA 33 Chapter IT Eight base changes are sufficient to convert a leucine-inserting tRNA into 59 a serine-inserting tRNA Chapter ill Construction of two Escherichia coli 84 amber suppressor genes: tRNAPheCUA and tRNACysCUA Chapter IV Construction of Escherichia coli amber suppressor genes: Optimization of suppressor efficiency and determination of tRNA specificity 105 1 INTRODUCTION tRNA Identity Jennifer Normanly and John Abelson Department of Biology, California Institute of Technology, Pasadena CA 91125 [Annual Review of Biochemistry (1989) vol. 58, in press] 2 BACKGROUND In the translation of genetic information from nucleic acid to protein, transfer RNA plays a crucial intermediate role. Each codon is read by a unique tRNA that has been aminoacylated with the appropriate amino acid. Because of degeneracy in the genetic code, in general there is more than one tRNA for each amino acid (1). Although there are some exceptions to the rule, there is a single aminoacyl tRNA synthetase (AAS) for each amino acid. An AAS must correctly recognize and activate an amino acid and then join it to each of the cognate set of tRNAs. By virtue of the fact that they must function interchangeably on the ribosome during protein synthesis, all tRNAs have similar primary (2), secondary (3), and tertiary (4, 5) structures (Figure 1). Within these constraints each cognate set of tRNAs contains distinct elements that mediate correct recognition by the AAS. These elements comprise the identity of a tRNA. Identity elements can be both positive and negative. Positive elements are those features of the tRNA that the cognate AAS recognizes directly, and negative elements are those features that block the recognition by other AASs. Although tRNAs comprise a homogeneous set of molecules, AASs are remarkably diverse. The quaternary structures a, az, a... and az~ have all been observed (6, 7). AAS subunit sizes range from 300 to 900 amino acids. The details of the interaction between these diverse enzymes and their tRNA substrates will not be known until high resolution X-ray structures of tRNA-AAS complexes have been solved. Fortunately, there is the prospect that several such complexes will be solved in the near future (8, 9). Already, we have a remarkably detailed description of the amino acid and A TP binding sites for the B. stearothermophilus tyrosine AAS (10) and the E. coli methionine AAS (11). It is not our purpose to review the large body of information on the structure and properties of the AAS, or on the indirect efforts that have been employed to map their interactions with tRNA. These have been well reviewed before in these volumes (6, 7). Rather, we wish to describe the results of fresh and energetic approaches to the problem of tRNA identity, which have appeared in the last few years. 3 Past and Present Approaches to Defining tRNA Identity. We will describe two new approaches to the determination oftRNA identity. Both have their roots in previous work but are now more powerful because of advances in gene synthesis technology. In the first, attempts are made to alter with the fewest changes, the identity of a tRNA while retaining complete biological function. Those changes responsible for the redirection of AAS recognition constitute identity eiements. We term this type of experiment an identity swap. In the second, tRNA variants are generated by altering the sequence of a tRNA, and their properties as a substrate for the AAS are determined in vitro. This approach has been greatly facilitated by the ability to synthesize tRNA in vitro with bacteriophage TI RNA polymerase. The identity swap has its origins in the study of amber suppressors. An amber suppressor is a mutant tRNA that can recognize and suppress the chain terminating effects of a UAG codon, by virtue of a change in the tRNA's anticodon (12, 13). Suppressors have greatly facilitated the application of genetics to the study of tRNA, and have been utilized extensively in bacteria (14), T4 bacteriophage (15, 16), and yeast (17). Such work has led to a formal proof of the secondary structure of tRNA (14) and to information about the mechanism oftRNA processing (18-20). Suppressors have also been used to address the tRNA identity problem. Attempts were made to alter the specificity of the E. coli tRNATyr suppressor, su+3, by selecting mutants of this tRNA that could suppress amber mutations at positions in which tyrosine was not an acceptable amino acid. This approach was taken independently by J. D. Smith and S. Brenner in Cambridge and by H. Ozeki and Y. Shimura in Kyoto nearly 20 years ago (21, 22). Both groups obtained the same results. The su+3 mutants that were isolated inserted glutamine instead of tyrosine into amber codons (23). Sequence analysis revealed changes in the fourth base from the 3' end, A72 _, G (22) and in the first and second base pairs of the acceptor stem 1-72 and 2-71 (21). This was an exciting result, but it did not prove to be universally applicable to the study of tRNA identity. Despite many attempts, su+3 mutants that inserted an amino acid other than glutamine or tyrosine were never found (23). It is likely that such transformations would have required more than one change in the tRNA sequence and therefore could not be obtained by genetic selection. This approach was abandoned for fifteen years. We have returned to this approach armed with an automated DNA synthesizer, which allows the rapid synthesis of altered suppressors de novo, as in the pioneering work of Khorana (24). The synthetic genes are assembled from oligonucleotides and inserted in 4 cassette form into a plasmid, which supplies a promoter, either the inducible lac promoter (25) or the constitutive lpp promoter (26, 27), and a transcription terminator. Usually a screen is employed, in which the suppression of an amber mutation in a gene requires the insertion of a specific amino acid to allow growth. The amino acid specificity of the suppressor is determined by sequencing the protein product of a suppressed amber mutant gene. For this purpose the E. coli dihydrofolate reductase (DHFR) gene with an amber mutation at codon ten is used. DHFR is easily purified by affinity chromatography on a methotrexate resin. X-ray crystallographic data (28-29) revealed that residue ten of DHFR lies on the surface of the protein, well removed from the active site, and it appears that any amino acid is acceptable at this position. If more than one amino acid is inserted by the suppressor, it can be readily detected by this method. Protein sequence data acquired in this manner reveal the outcome of the competition among all 20 AAS in the cell for the suppressor. The drawbacks to this method are twofold. First, the anticodon of the tRNA must be CUA to allow recognition of amber codons; so the contribution of the anticodon to tRNA identity cannot be assessed. As described below, the anticodon is a strong identity element for some tRNAs. Second, the effectiveness of the suppressor tRNA as a substrate for the AAS can only be measured by the efficiency of suppression. Multiple factors, such as the ability of the tRNA to be transcribed, processed, modified, and accepted by the ribosome, can affect the observed suppression efficiency, and in this method cannot be distinguished from the efficiency of aminoacylation. In the second method developed by Uhlenbeck (30), tRNA genes and their variants are synthesized and placed under the control of the bacteriophage 17 promoter so that the first nucleotide of the in vitro transcript is the mature 5' end of the tRNA. Alternatively, synthetic tRNAs with 5' precursor sequences can be processed correctly by incubation with the RNA component of RNAse P (31; L. Schulman, unpublished). A Bst Nl restriction endonuclease recognition site (5'CC·AGG) is encoded at the 3' end of the gene. Run-off transcription ofBst Nl cleaved DNA generates the correct CCA terminus. Prior to the availability of this methodology, the study of tRNA was hindered in that most variants were not processed or were unstable in vivo. This technique not only overcomes such obstacles but also allows the easy purification of the transcript. Reactions can be scaled up and milligrams of some transcripts have been produced for NMR studies (Hall et al, unpublished). These synthetic tRNAs do not contain the normal base modifications found in native tRNA (and are therefore excellent substrates for the study of any modifying enzyme). They are, nonetheless, good substrates for the AAS. Typically, the Kro values 5 for the substrates are one- to four-fold higher than for fully modified tRNA. V max values for the unmodified substrates are at most, two-fold lower, resulting in values of the specificity constant (k:cat!Krn) that are five- to seventeen-fold lower. In contrast, the kcat!Krn for a typical misacylation is orders of magnitude lower (6). These results indicate that the modified nucleotides are not likely to play a large role in recognition by the AAS, as their absence does not significantly diminish the efficiency of aminoacylation. In order to determine that the synthetic tRNAs are properly folded, Sampson and Uhlenbeck (32) have exploited the phenomenon of lead-mediated cleavage of tRNA. This reaction is mediated by a tightly bound Pb++ ion whose position is determined by the tertiary interactions between the D-loop and T-loop (33, 34). In native yeast tRNAPhe, cleavage takes place at residue 17. Unmodified tRNAPhe is ~l.eaved at the same position, with similar kinetics, indicating that the synthetic molecule has folded properly. The in vitro approach is attractive in that one can quantitate the effects of individual nucleotide changes on the recognition of the tRNA by cognate and non-cognate AAS. The method does not allow assessment of the effect of competition between one AAS and the other 19 AASs. In the end, it will be desirable to apply both the in vivo and in vitro approaches to determine the identity elements of any tRNA. The Role of the Anticodon in tRNA Identity. Since the anticodon is directly correlated with the identity of any tRNA species, it is the most logical identity element. Indeed, one of the earliest recognition hypotheses stated that the "coding triplet (anticodon) plays an important, perhaps decisive role in the interaction with the synthetase" (35). Early evidence to support this hypothesis came from a combination of chemical modification and oligonucleotide interference experiments as well as from analysis of mutant tRNAs, and suggested that for a number of tRNAs, the anticodon can be implicated in recognition. Among them are E. coli tRNAGly, tRNALys. tRNAArg, tRNAGlu, tRNAMet(t), tRNATrp, yeast tRNAGly, tRNAVal, and beef pancreatic tRNATrp (reviewed in 36). The opposing notion, that the anticodon could not be a recognition element, was derived from the observation that the anticodon could be altered to create amber suppressors with no effect on amino acid specificity in at least four tRNAs. Additionally, translation of the codons (six each) for serine, leucine, and arginine requires a set of tRNAs with different nucleotides in two out of three positions for leucine and arginine, and in all three for serine. The idea that the anticodon is not an identity element seems to have held 6 sway, in some quarters at least, because a few textbooks still state this unequivocally. Nonetheless, it is now clear that the anticodon is a recognition element for tRNA identity in perhaps a majority of tRNAs. The strongest evidence in support of this view comes from L. Schulman's work on E. coli tRNAMet. Experiments conducted over a period of 15 years have demonstrated conclusively that the CAU anticodon of E. coli tRNAMet is an essential recognition element for the methionine AAS. Chemical modification experiments have shown that the anticodon nucleotides C34 and A35 cannot be modified without loss of methionine acceptor activity in vitro (37). The anticodon of tRNAMet was altered by the judicial insertion of oligoribonucleotides into the tRNA, which had been cleaved specifically by RNAse. These experiments revealed that substitution of~ with aU, A, or G lead to a decrease in the aminoacylation rate by four to five orders of magnitude (38). Altering the other two positions of the anticodon (A35 and U36) affect AAS recognition as well, decreasing the rate of aminoacylation one to four orders of magnitude to that of normal (39). Substitutions of any position in the anticodon with nucleotides containing functional groups in common with those of the wildtype nucleotides were tolerated more than if nucleotides, lacking similar functional groups were substituted The conclusion was that the synthetase must be interacting with specific functional groups in the anticodon (40). In addition, chemical crosslinking experiments demonstrated that C34 of the anticodon is uniquely crosslinked to Lys 465 in the methionine AAS (41). More recently, Schulman has conducted identity swap experiments in the in vitro system (42). In these experiments genes for both tRNATrp and tRNAVal of E . coli were synthesized with the methionine (CAU) anticodon, transcribed in vitro, and the kinetics of acylation by the methionine AAS and valine AAS were determined for the synthetic tRNAs (Table 1). tRNAVal (CAU) was aminoacylated by the methionine AAS with a VmaxfKrn similar to native tRNAMet. tRNATrp (CAU) was also acylated by the methione AAS, but the Vmax!Krn was down ten-fold. Conversely, tRNAMet with the valine anticodon (UAC) was acylated by the valine AAS with a Vmax!Km value 1/10 of that for normal tRNAVal. In comparison, the value for acylation of tRNAMet by the valine AAS was four orders of magnitude lower than that for tRNAVal. These results clearly show that the anticodon is a major, if not the sole, recognition element for tRNAMet and is important for tRNAVal identity as well. 7 Theories about the role of the anticodon in AAS recognition have generally been influenced by the existence and the properties of amber suppressors. The demonstration that suppressors for tRNATYT, tRNA5 er, tRNA0 ln, tRNALeu, were correctly charged, indicated that not all AASs recognize the anticodon nucleotides of their cognate tRNAs. A more complete picture of the effects of the amber anticodon on AAS recognition has been obtained from our efforts in collaboration with J. Miller and coworkers to construct a complete bank of 20 amber suppressors (27, 43 Normanly eta/, unpublished). These suppressors have been generated by synthesizing the genes of isoacceptors from the 20 species oftRNA in E. coli, each containing a CUA anticodon. As in the identity swap experiments described above, the tRNA genes are inserted in cassette form into a plasmid containing the constitutive lpp promoter and a terminator. The amino acid specificity of each suppressor was determined as described above by sequencing the suppressed product of the amber mutation at position ten in DHFR. Tables 2 and 3 give a summary progress report of this project. Table 2 lists the species of tRNA, which exist as amber alleles, derived either genetically or by synthetic means. Table 3 classifies the suppressors by their amino acid specificity. Seven of the synthetic suppressors insert the correct amino acid. Together with the four original suppressors, 11 of 20 tRNAs in E. coli can tolerate the amber anticodon without misacylating. In some cases suppressor tRNAs are mischarged by the glutamine AAS. The synthetically derived tRNAG1t1AcuA inserts 20% glutamine. From previous work (44) we know that tRNAMet(f)CUA is mischarged by the glutamine AAS in vitro, and the su+? allele of tRNATtp inserts a mixture of glutamine and tryptophan in vivo (45). The amber allele of one isoacceptor of tRNAGIY, tRNA01YUcuA. was correctly charged in vivo, another tRNAGlyTCUA• was misacylated by the glutamine AAS (46; Normanly eta/, unpublished). We were surprised to fmd that six of the suppressors generated in this work, tRNAArgCUA, tRNAAspMCUA, tRNAlle2cuA, tRNAMet(m)CUA, tRNAThr2CUA and tRNAVaiCUA• were misacylated by the lysine AAS. There are also examples of suppressors that are misacylated by both the glutamine and lysine AASs. Such is the case for tRNAilelcuA, as well as for some mutant suppressor tRNAs (46-48). Apparently, only the glutamine AAS and the lysine AAS rnisacylate amber suppressors. Each tends to misacylate a separate set Because the CUA anticodon is recognized by both the glutamine AAS and the lysine AAS, there must be negative elements that prevent acylation by these enzymes. Some suppressors have both elements, some only one, and others have neither. It is interesting to note that there are two lysine AAS genes in E. coli, one of which is preferentially expressed during heat shock (49). Preliminary 8 experiments designed to detect which of the enzymes is responsible for the observed misacylation indicated that both are. A Single Base Pair is a Major Identity Detenninantfor tRNAAID. Hou and Schimmel set out to isolate mutants of a tRNAAla amber suppressor, which could not be recognized by the alanine AAS using the in vivo approach (50). An impressive collection of synthetic mutants was generated with multiple changes throughout the tRNA. Only those mutants having in common a change in the G-U pair at position 3-70 failed to insert alanine. This fmding is consistent with two previous observations. First, earlier work by Murgola had indicated that a G-U pair at this position in a tRNALys missense suppressor caused the insertion of either alanine or glycine (51). Second, the G-U pair at this position is conserved in virtually all of the tRNAAla that have been sequenced. To test the significance of the G 3-U70 pair, Hou and Schimmel performed two identity swap experiments. tRNACys and tRNAPhe amber suppressors were synthesized, each containing the G-U pair at position 3-70. The altered cysteine suppressor inserted entirely alanine (fable 4). The altered phenylalanine suppressor inserted predominately alanine (63%) and some phenylalanine (37% ). This result is certainly one of the most dramatic identity shift experiments to date in that it implicated a single G-U pair as the major identity element for the alanine AAS. The ability of the G 3-U70 pair to direct alanine AAS recognition has been tested in other tRNAs with mixed results. Superimposing the G-U pair at position 3-70 of a lysine suppressor results in complete conversion to alanine identity (48). However, the same alteration to a glycine suppressor tRNA (GlyT) results in the insertion of predominantly glutamine, a small amount of glycine, and no detectable levels of alanine (46). It shoud be noted that the glycine amber suppressor with the normal G-C base pair at position 3-70 inserts predominantly glutamine (46; Normanly et al, unpublished). In the case of tRNAGiyT, simply superimposing the G-U pair at position 3-70 is not sufficient to counteract the identity determinants for the glutamine AAS. The glycine to alanine identity swap experiment might be more successful if the amber allele of another isoacceptor of tRNAGiy which is not misacylated, were used instead. McClain and Foss approached the problem of tRNAAla from a different angle (46). McClain has developed an algorithm (52, 53), which compares the sequences of a tRNA with its isoacceptors as well as isoacceptors of all 19 other tRNAs. This algorithm selects those positions in a tRNA, which are correlated with its isoacceptors and anticorrelated 9 with the other tRNA sequences. Subjecting tRNAAla to this type of analysis resulted in the prediction that four nucleotides were significant to tRNAAla in addition to the anticodon. These were the GrU70 pair, as well as G 20 and C 60 (Figure 2). The phenylalanine suppressor was altered to include these nucleotides, and a change was incorporated at position 51 to increase transcriptional efficiency (54). They also made changes at positions 16 and 17, which were designed to remove phenylalanine identity. In the DHFR assay, the resulting suppressor inserted 96% alanine and 4% lysine, indicating that the phenylalanine identity determinants had been removed. They examined the contribution of C60 to alanine identity, and found that altering this putative identity element in tRNAAla did not abolish alanine acceptance, nor did it give rise to acylation by the alanine AAS when superimposed upon tRNAPhe. In a similar fashion, the contribution of G20 to alanine identity was determined to be minimal. Alanine acceptance was not abolished when G 20 was altered in tRNAAla, nor did G 20 transplanted into tRNAPhe result in alanine insertion. It would appear that these elements of alanine identity predicted by the algorithm do not have a strong effect upon alanine AAS recognition. tRNAPhecontaining the G 3-U70 base-pair alone, inserted 24% alanine and 76% phenylalanine. This result was the inverse of that obtained by Hou and Schimmel (63% alanine and 37% phenylalanine). The reason for this discrepancy is not clear, because both groups used the same expression system and the same specificity assay; namely, suppression of an amber mutation in DHFR. In any case, the G 3-U70 pair is clearly the dominant determinant of Ala identity, and is the strongest contributor to alanine identity predicted by McClain's algorithm. The G-U pair as an identity element raises an interesting question. Does the alanine AAS recognize the functional groups of these bases via interactions in the deep groove of the A-helix or is it the perturbations to the helix stability and structure imposed by the G-U "wobble" pair that are recognized? To answer this question McClain has now constructed alanine suppressors that contain each of 16 possible base combinations at position 3-70 (48). All the variants had substantially lowered suppression efficiencies, and some of the base combinations yielded inactive suppressors. The four Watson-Crick base pairs at this position yielded suppressors that inserted lysine and glutamine. tRNAAla containing the mispairs G-A, C-A, and U-U at position 3-70 all inserted 90% alanine and, though less efficient, were similar to G-U at this position. Other variants, containing A-C and C-C, C-U or U-G at position 3-70 inserted -80% alanine, along with lysine and glutamine. The 10 mispairs, G-G, A-A, and A-G led to inactive suppressors, and the variant with a U-C mispair inserted predominantly glutamine. These results suggest that the alanine AAS is recognizing a structural perturbation imparted by the G-U pair and not specific bases. However, not all mispairs are acceptable; so the perturbation that is recognized must be sequence dependent. (It should be noted that NMR and X-ray crystallographic data reveal that a G-U base pair does not have a dramatic effect upon the overall structure of the acceptor stem helix.) Translocation of the G-U pair to position 4-69led to partial alanine insertion, but only if the 3-70 pair was A-U and not G-C. G-U at position 2-71 had no activity. More questions are raised by these observations than are answered. The results emphasize the fact that there is insufficient data on the sequence dependence of RNA structure. Among the structural variations that one can imagine is a bending of the helix or even a separation of the strands. The tRNALeu--+Ser Identity Swap. We have used the suppressor approach to define the elements of tRNASer identity (25). Serine was chosen for several reasons. There are six serine codons, AGU/C and UCN (where N denotes any nucleotide) recognized by at least four different tRNAs. One of the original amber suppressors (su+t) inserted serine; so in all, the serine AAS can recognize tRNAs with at least five anticodons. It is unlikely that there are any serine recognition elements in the anticodon although this assumption has not yet been tested in vitro. We set out to change the identity of a tRNALeu amber suppressor to serine. An examination of the six E . coli tRNASer sequences revealed fourteen nucleotides conserved in all tRNASer sequences but not found in the tRNALeus starting sequence. Additionally, the base pair at position 3-70, though not conserved (A-U or U-A) was found by McClain's statistical analysis to be correlated with serine and anticorrelated with other sequences (52, 53). This base pair was included as a potential element of serine AAS recognition (Figure 3a). The conserved nucleotides in the acceptor stem and in the D-stem and loop were included in the original trial construction. In addition, the size of the D-loop is highly conserved in the six tRNASer sequences. Changes were made in that region to conform to the serine D-loop size. The first trial construction contained 12 changes from the starting tRNALeu5 amber suppressor (Figure 3b). The resultant suppressor, tRNALeu-+Ser. was able to suppress an amber mutation in the f3-lactamase gene, which requires the insertion of serine to produce an active enzyme. However, the suppression efficiency of tRNAl.eu-+Ser, 1%, was quite low compared to the 60 to 100% efficiency of the starting tRNALeu suppressor. Using the 11 DHFR assay, the amino acid specificity of the tRNALeu~ser suppressor was determined to be serine. Following our initial successful attempt at the identity swap, we then set out to determine the minimum number of changes required to effect the transformation. This has now been accomplished and the results are shown in Figure 3c. The minimum number of base changes required is eight: residing in the first three base pairs of the acceptor stem, nucleotide 73 and the 11-24 base pair in the D-stem. The resulting suppressor containing these minimal changes is approximately 40% efficient The efficiency was regained when the unnecessary changes in the D-loop were discarded. Interestingly, we find that only an A-U is acceptable at position 3-70 even though some serine tRNAs contain U-A at this position. We speculate that the A-U is acting as a deterrant to the leucine AAS. We are • now using this information to conduct further identity swaps: Perhaps other serine identity elements exist that were already present in tRNALeus. By this iterative procedure we hope to arrive at a clear understanding of tRNASer identity. The E. coli serine AAS has been purified and crystallized by R. Leberman in Grenoble (55). The ultimate understanding of tRNASer recogniton will come when the manner in which these identity elements are recognized by the enzyme has been defined. tRNAPiu Recognition. As in the case of E. coli tRNAMet, the anticodon has been implicated in the recognition of yeast tRNAPhe by the yeast phenylalnine AAS. Early work by Zachau, in which the anticodon of yeast tRNAPhe was removed, resulted in a tRNA that was aminoacylated by the phenylalnine AAS at a level20% that of wildtype tRNAPhe (55a). Using recombinant RNA technology, Uhlenbeck and co-workers have demonstrated that substitution of any of the yeast tRNAPhe anticodon nucleotides diminishes aminoacylation (kcat/Km was decreased 3- to 10-fold) (56). In addition, transplanting the phenylalanine anticodon onto tRNATyr leads to aminoacylation of the hybrid molecule by the phenylalnine AAS (57). As the aminoacylation of the hybrid tRNA was fairly weak, they speculated that there must be phenylalanine recognition elements outside of the anticodon as well. Using the in vitro system, transcription of synthetic tRNA genes with 17 RNA polymerase, a large number of variants have been constructed and their properties as substrates have been tested (30, 58). This is by far the largest and most comprehensive set of tRNA variants ever constructed. At this point nearly all of the non-universal bases in tRNAPhe have been changed and the variants have been tested as substrates. Experiments 12 in which the nucleotides involved in tertiary interactions were altered, revealed that changes in these bases can be tolerated as long as the overall tertiary interactions are not disrupted. For example the 0 19- C56 tertiary Watson-Crick base pairs can be altered to C 19- 0 56 without affecting recognition by the phenylalnine AAS. Either of the single mutations, 0 19 ~ Cor C5 6 ~ 0 leads to a five-fold drop in kcatiK.rn, indicating that the tertiary structure of the tRNA must be maintained but that neither 0 19 or C56 is recognized directly. When OrC72 is altered to either 0-U or A-U the kcatiK.rn drops four- and seven-fold, respectively (58a). It is not clear yet whether 0 1- C 72 is important as a recognition element for the phenylalanine AAS, or as a structural feature. The results of this study led to the proposal that five nucleotides, all in single strand regions, are recognition elements for the phenylalanine AAS. They are the three anticodon bases, 0 34, A 35 , and A 36, 0 20 in the D-loop, and A73 the fourth base from the 3' end (Figure 4). Several tests of the model were undertaken. First, the E. coli tRNAPhe, a rather poor substrate for the yeast phenylalnine AAS, contained all of the putative identity elements except the one at position 20. Changing U 20 to 0 brought kcat/Km to within a factor of two for that of yeast tRNAPhe. Identity swaps based on the conserved bases were also carried outl, changing yeast tRNAMet, tRNAAr&, and tRNATyr to good stubstrates for the phenylalnine AAS. It appears that each of the five bases contributes independently to the specificity. Thus, a tRNA containing four of the five recognition elements has a kcatiK.rn one order of magnitude lower than wild type, and a substrate with only three recognition elements has a kcat!Km which is two orders of magnitude lower. Some years ago Dudock and collaborators reported the mischarging of a number of tRNAs by the phenylalanine AAS (59). With knowledge of the recognition elements and thus relative contributions to specificity, Sampson and Uhlenbeck were able to reconcile that data nicely. Of the yeast tRNAs that have been sequenced none have more than three of the recognition elements so that the in vivo specificity can be explained if one also assumes a contribution to overall specificity due to competition with the cognate AAS. The five recognition elements are located far apart from one another in the tRNA tertiary structure so the phenylalanine AAS must contact the entire surface of the tRNA molecule. I Changes were incorporated elsewhere as well, either to facilitate the runoff transcription reaction, or to introduce tRNAPhe structural elements where they differed from the starting tRNA. 13 McClain and Foss have used the suppressor approach to investigate E. coli tRNAPhe identity. From their results it appears that U20 and A73 are implicated in tRNAPhe identity. In addition, small effects are observed in changing the anticodon stem base pairs 27-43, and 28-42 (47). tRNAGln identity. It is pertinent to discuss what is known about the recognition of E. coli tRNAGln. In a short time we will know more about the interaction of this tRNA with its AAS than for any other system. At this point (October 1988) Steitz and Soli have generated tRNAGln..AAS co-crystals that refract to high resolution. Data has been collected and a map is being constructed Where do we expect the contacts will be made? First, as we have already mentioned, it is likely that the glutamine AAS recognizes the anticodon. The su+? allele of E. coli tRNATrp , which inserts both glutamine and tryptophan in vivo, has a single change in the anticodon from CCA to CUA implicating the U in the middle position (45). In vitro, the su+? tRNA is a better substrate for the glutamine AAS by about six orders of magnitude (60). Changing the anticodon of E . coli tRNAMet(t) from CAU to CUA leads to misacylation by the glutamine AAS in vitro, at levels similar to that seen for su+? tRNA (44 ). The su+3 allele of E. coli tRNATyr does not mischarge in vivo but elevated levels of the glutamine AAS can lead to mischarging in vitro (61). We also anticipate that the glutamine AAS will interact with the 3' end of the tRNA. Mutants of su+3 tRNATyr with alterations at the 3' end result in recognition by the glutamine AAS. One of these mutants, A 73~ G, is particularly effective (21-23). tRNAGln has a G at position 73. Rogers and Soli have also described mutational changes of the tRNA Ser in the acceptor base pairs 1-72 and 2-73, which convert this tRNA to glutamine specificity in the DHFR assay (62). Note that these are also positions recognized by the serine AAS so competition between the two enzymes was decreased by making these changes. The Variable Pocket- A Potential Site in tRNA for AAS Recognition. In discussing the structure of yeast tRNAPhe , Klug pointed out that the cluster of nucleotides formed by the interaction of the D-loop and T-loop forms a patch that arches out from the molecular surface of the molecule (63). The nucleotides involved in this patch are 16, 17, 20, 59, and 60 (Figure 1). The patch is variable in that these nucleotides are 14 not conserved among different tRNAs, and there can be insertions or deletions in the D-loop which could potentially change the configuration of this pocket Klug explicitly suggested that this variable pocket may "form part of a recognition system for different tRNAs, perhaps for sorting into classes for synthetase discrimination." We have already seen that 0 20 in yeast tRNAPhe is a recognition element for the phenylalanine AAS (58). In an analysis of the six E. coli tRNAArg sequences, McClain observed a strong correlation of A20 with arginine identity. Consequently, a phenylalnine to arginine identity swap was attempted (64). The A20 was inserted into the phenylalanine D-loop causing U20 to become U 20.1. This suppressor was not active. However, with the variable pocket model in mind, residues 16, 17, and 59 were altered in an attempt to create the variable pocket characteristic oftRNAArgACG-1· This suppressor inserts 75% arginine, 5% lysine and 6% tyrosine. It is a significantly better arginine suppressor than one constructed simply by changing the anticodon of tRNAArgACG-1 to CUA, which inserts 37% Arg and 55% lysine. Insertion of A20 and A59 alone had virtually the same effect. Taken together with the tRNAPhe results, this fmding certainly highlights residue 20 as an important synthetase recognition element. Given the exposed configuration of the variable pocket, other synthetases may well use different residues in this group as recognition elements. tRNA Specificity Defined by Nucleotide Modification. In general, tRNA modifications do not play a large role in tRNA identity. That this is so is most clearly demonstrated by the near normal substrate activity of unmodified transcripts. Indeed, even a tDNAPhe can be specifically aminoacylated (65). There is, however, a very interesting case in which a tRNA modification plays a crucial role not only in AAS recognition but in codon recognition as well. E. coli tRNAlle2 recognizes the codon AUA. It was therefore a surprise when it was discovered that the anticodon in the DNA sequence of the tRNAlle2 gene is CAT (66), the anticodon for methionine. The same sequence was found in the bacteriophage T4 tRNAile gene (67, 68), and in a spinach chloroplast tRNAlle gene (69). Recall that C 34 in tRNAMet is a crucial recognition element. Thus, we have the dilemma that this tRNAlle should not only miscode, but it should also mischarge. The dilemma was resolved by the discovery that the C in tRNAlle2 is modified to lysidine (Figure 5). This alteration affects recognition by the synthetase; the unmodified tRNA is recognized by the methionine AAS and the modified tRNA by the isoleucine AAS (66). 15 Lysidine also changes the ccx:ling capacity of the tRNA so that it recognizes AVA and not AUG. Why is this bizarre route taken, rather than simply making that anticodon UAU? Muramatsu et al, suggest that the evolution of the lysidine mechanism may have allowed a change in the genetic code converting AUA from methionine to isoleucine specificity. Once the mechanism was fixed it was retained except in some mitochondria where AUA has reverted to methionine specificity. Muramatsu et al, suggest that this was due to the loss of the lysidine biosynthetic mechanism (66). CONCLUSION Much progress has been made recently with the application of two powerful methods to the study of tRNA identity. The in vitro approach assesses the contribution of specific nucleotides to tRNA identity by examining the aminoacylation kinetics of specifically altered tRNAs, providing a very clear picture of recognition elements for a given tRNA. Adding to this approach, determination of the in vivo specificity of altered tRNAs provides clues into the physiological relevance of specific nucleotides, i.e., what effect they have upon the competition between all 20 AAS. The experiments reviewed here have revealed the following: a) tRNA identity is defined by a relatively small number of elements, and b) the anticodon appears to be an important identity element for a majority of tRNAs. Identity elements may vary in the extent of their evolutionary conservation. For example the G3 -U70 base pair so important for alanine identity is present in virtually every tRNAAla sequenced to date, over a wide phylogenetic spectrum. The same appears to hold true for the A20 found to be involved in arginine identity. The determinants of serine tRNA identity, however, are conserved only in bacterial genes. There are likely to be classes of tRNAs in regards to AAS recognition. On the simplest level, tRNAs can be classified according to the effect that an alteration in the anticodon has upon AAS recognition. Cognate recognition is either a) normal, or b) diminished/abolished. Within these subsets further divisions can be made. For example, both E. coli tRNAMet(m) and yeast tRNAPhe fall into the latter class. Yet for tRNAMet(m) the anticodon appears to be the sole identity element, while for yeast tRNAPhe two other nucleotides are important for complete phenylalanine acceptance. Similarly, E. coli tRNASer and tRNAAlafall into class one, i.e., the anticodon does not appear to play a role in cognate AAS recognition. While eight identity elements have been identified for tRNASer, a single basepair is primarily responsible for tRNAAla identity. 16 The phenomena oflysine and glutamine mischarging remain a mystery. The composition of the anticodon must in large part direct mischarging by these two AASs (e.g. aU at position 35 is a recurring theme in glutamine misacylation), but, as they do not misacylate universally, there must be identity determinants or deterrents outside of the anticodon that dictate which tRNAs can be readily misacylated. Both of these AASs may have rather loose specificity, and, subsequently, behave as "default" AAS for tRNAs that are not well charged by their cognate AAS. Sorting out identity determinants from deterrents will no doubt prove to be a difficult task; however, a systematic application of both the in vitro and in vivo approaches will go a long way towards solving the problem. Recent progress in the crystallography of tRNA-AAS complexes (8, 9) promises to give insight into the interaction between the two molecules, and will undoubtedly suggest new pathways to take in the study of the recognition problem. ACKNOWLEDGEMENTS We wish to thank Jeff Sampson, Olke Uhlenbeck, LaDonne Schulman, and Bill McClain for sharing their results prior to publication, and to David Horowitz, Olke Uhlenbeck and Jeff Sampson for critical reading of the manuscript The authors were supported by grants from Nlli (GM32637) and ONR (N00014-86-K-0755). REFERENCES 1. Crick, F. H. C. J. Mol. Bioi. 1966. 19:548-55 2. Sprinzl, M., Hartmann, T., Meissner, F., Moll, J., Vorderwiilbecke, T. 1987. 3. 4. 5. 6. 7. 8. Nucleic Acids Res. 15:r53-r188 Holley, R. W., Apgar, J., Everett, G. A., Madison, J. T., Marquisee, M., eta!, 1965. Science. 147:1462-65 Kim, S. H., Sussman, J. L., Suddath, F. L., Quigley, G. J., McPherson, A. et al. 1974. Proc. Nat/. Acad. Sci. USA 71:4970-74 Klug, A., Robertus, J. D., Ladner, J. E., Brown, R. S., Finch, J. T. 1974. Proc. Nat/. Acad. Sci. USA 71 :3711-15 Schimmel, P. R., and Soli, D. 1979 Ann. Rev. Biochem. 48:601-48. Schimmel, P.R. 1987. Ann. Rev. Biochem. 56:125-58. Podjarny, A., Rees, B., Thierry, J .C., Cavarelli, J., Jesior J. C. et al, 1987. J. Biomol. Struct. Dynam. 5:187-98. 17 9. 10. Perona, J. J., Swanson, R., Steitz, T. A., Soli, D. 1988. J. Mol. Bioi. in press. Fersht, A. R., Shi, J. P., Knill-Jones, J., Lowe, D. M., Wilkinson, A. J., eta/, 1985. Nature. 314:235-38 11. Zelwer, C., Risler, J. L., Brunie, S. 1982. J. Mol. Bioi. 155:63-81 12. Goodman, H . M., Abelson, J., Landy, A., Brenner, S., Smith, J. D. 1968. Nature. 217:1019-24 13 . Steege, D. A., Soil, D. 1979. in Biological Regulation and Development ed. R. F. Goldberger, 1:433-85. New York, Plenum Press 14. Smith J. D. in Nonsense Mutations and tRNA Suppressors. 1979. eds J. 15. 16. E. Celis, J.D. Smith, pp.109-25. London, Academic Press 349 pp. Wilson, J. H., Abelson, J. N. 1972. J . Mol. Bioi. 69:57-73 ' McClain, W. H., Wilson, J. H., Seidman, J. G. ' 1988. J. Mol. Bioi. 203: in press. . 17. Kurjan, J., Hall, B. D., Gillam, S., Smith, M. 1980. Cell. 20:701-9 18. Altman, S. in Nonsense Mutations and tRNA Suppressors 1979. eds J. E. Celis, J.D. Smith, pp. 173-89. London, Academic Press 349 pp. 19. Willis, 1., Hottinger, H., Pearson, D., Chisholm, V., Leupold, U. et a/1984 EMBO 3:1573-80 20. Nishikura, K., Kurjan, J., Hall, B. D., DeRobertis, E. M. l982. EMBO. 1:263-68 21. Hooper, M. L., Russell, R. L, Smith, J. D.l972. Febs Lett. 22:149-56 22. Shimura, Y., Aono, H., Ozeki, H., Sarabhai, A., Lamfrom, H., Abelson, J. 1972. Febs Lett. 22:144-48 23. Ghysen, A., Celis, J. E. 1974. J Mol. Bioi. 83:333-51 24. Khorana, H. G. 1979. Science 203:614-25 25. Normanly, J., Ogden, R.C., Horvath, S.J., Abelson, J. 1986. Nature. 321:213-19. 26. Masson, J.-M., Miller, J.H. 1986. Gene. 47:179-83. 27. Normanly, J., Masson, J. -M., Kleina, L. G ., Abelson, J ., Miller, J. H. 1986. 28. 29. Proc. Nat/. Acad. Sci. USA. 83:6548-52 Bolin, J. T., Filman, D. J., Matthews, D. A., Hamlin, R. C., Kraut, J. 1982. J . Bioi. Chern. 257:13650-662 Filman, D. J., Bolin, J. T., Matthews, D. A., Kraut, J. 1982. J. Bioi. Chern. 257:13663-672 18 30. Sampson, J. R., Uhlenbeck, 0. C. 1988. Proc. Nat/. Acad. Sci. USA. 31. 85:1033-37 Samuelsson, T., Boren, T., Johansen, T. -1., Lustig, F. 1988. J. Bioi. Chern. 32. 263:13692-9 Sampson, J. R., Sullivan, F. X., Behlen, L. S., DiRenzo, A. B, Uhlenbeck, 0. C.l987. Cold Spring Harbor Symp. Quant. Bioi. 52:267-79 33. Brown, R. S., Dewan, J. C., Klug, A. 1985. Biochemistry. 24:4785-801 34. Dirheimer, G., Ebel, J. -P., Bonnet, J., Gangloff, J., Keith, G., eta/. 1972. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. Biochimie. 54:127-44 Engelhardt, W. A, Kisselev, L. L.l966. in Current Aspects of Biochemical Energetics. eds. N. O.Kaplan, E. O.Kennedy, pp.213-15 New York, Academic Press Kisselev, L. L. 1985. in Progress in Nucleic Acids Research and Molecular Biology. eds. W. E. Cohn, K. Moldave, v32:237-66. Florida, Academic Press. Schulman, L. H. in Transfer RNA: Structure, Properties andRecognition.1979. eds. P.R. Schimmel, D. Soli, and J. N. Abelson.pp.311-24. New York, Cold Spring Harbor Laboratory Press Schulman, L. H., Pelka, H., Susani, M. 1983. Nucleic Acids Res. 11:1439-55 Schulman, L. H., Pelka, H. 1983. Proc. Nat/. Acad. Sci. USA. 80:6755-59 Schulman, L. H., Pelka, H.l984. Fed. Proc. 43:2977-80 Valenzuela, D., Leon 0., Schulman, L. H.l984. BBRC. 119:677-84 Schulman, L. H., Pelka, H. 1988. Science in press Miller, J. H., Normanly, J., Masson, J. -M., Kleina, L. G., Abelson, J. 1987. in Integration and Control ofMetabolic Processes: Pure and Applied Aspects.pp. 571-81 . Cambridge University Press, England Schulman, L. H. Pelka, H. 1985. Biochemistry. 24:7309-14 Celis, J. E., Coulondre, C. Miller, J. H. 1976 J Mol. Bioi. 104:729-34 McClain, W. H., Foss, K.l988. Science, 240:793-96. McClain, W. H., Foss, K. 1988. J. Mol. Bioi. 202:697-709 McClain, W. H., Foss, K., Chen, Y. -M., Schneider, J. 1988. Science. in press Hirshfield, I. N. Bloch, P. L., Van Bogelen, R. A., Neihardt, F. C. 1981. J . Bacterial. 146:345-51 Hou, Y.-M., Schimmel, P. R.l988. Nature. 333:140-45. Prather, N. E., Murgola, E. J. Mims, B. H.l984 J. Mol. Bio/.172:177-84 Atilgan, T., Nicholas, H. B., McClain, W. H. 1986. Nucleic Acids Res. 14:375-80 19 53. 54. 55. 55a. 56. 57. 58. 58a. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. McClain, W. H., Nicholas, H. B.l987. J. Mol. Bio/.194:635-42 McClain, W. H., Guenier-Tak:ada, C., Altman, S.l987. Science. 238:527-30 Leberman, R., Berthet-Colominas, C., Cusack, S., Hanlein, M. 1987. J. Mol. Bioi. 193:423-25 Thiebe, R., Harbers, K., Zachau, H. G. 1972. Eur. J . Biochem. 26:144-52 Bruce, A. G., and Uhlenbeck, O.C. 1982. Biochemistry. 21:3921-27 Bare, L., and Uhlenbeck, O.C. 1985. Biochemistry. 24:2354-60 Sampson, J. B., DiRenzo, A., Behlen, L., Uhlenbeck, 0. C. 1988. Science. in press Uhlenbeck, 0. C., Wu, H. -N., Sampson, J. R.l987. in Molecular Biology of RNA, New Perspectives. eds. M. Inouye, B. S. Dudock, pp. 285-94. Academic Press Roe, B., Sirover, M., Dudock, B. 1973. Biochemistry. 12:4146-54 Yarus, M., Knowlton, R., Soli, L. 1977. in Nucleic acid-protein recognition, pp. 391-408 New York, Academic Press Hoben, P., Uemura, H., Yamao, F., Cheung, A., Swanson, R., et al, 1984. Fed. Proc. 43:2972-76 Rogers, M. J., and Soli, D. 1988. Proc. Nat/. Acad. Sci. USA. in press Ladner, J. E., Jack, A., Robertus, J. D., Brown, R. S., Rhodes, D., et a/1975. Proc. Nat/. Acad. Sci. USA. 72:4414-18 McClain, W. H., Foss, K.l988. Science. 241 :1804-7 Khan, A. S., Roe, B. 1988. Science. 241:74-9 Muramatsu, T., Nishikawa, K., Nemoto, F., Kuchino Y., Nishimura, S. et al. 1988. Nature. in press. Fukada, K., Abelson, J. 1980. J . Mol. Bio/. 139:377-91 Mazzara, G. P., Plunkett, G., III, McClain, W. H. 1981. Proc. Nat/. Acad. Sci. USA. 78:889-92 Kashdan, M.A., Dudock, B. S. l982. J . Bioi. Chem. 257:11191-94 Thorbjarnardottir, S., Dingermann, T., Rafnar, T., Andresson, 0 . S., Soli, D. et al, 1985. J. Bacteriol. 161:219-22 Murgola, E. J.l985. Ann. Rev. Genetics.l9:57-80 Murgola, E. J., Hijazi, K. A. 1983. Mol. Gen. Genet. 191:132-37 Murgola, E. J., Prather, N. E., Pagel, F. T., Mims, B. H., Hijazi, K. A. 1984. Mol. Gen. Genet. 193:76-81 20 Table 1. Kinetic parameters for aminoacylation of tRNAs with E.coli methionine and valineAAS. Methionine AAS Relative V!Km tRNA Anticodon Met(f)* CAU 1.3 7xlo5 Met(m)* CAU 3.7 Met(m) CAU 1.7 2x106 9x105 Val 1.3 2x1Q-6 7x105 Val CAU UAC Trp Trp CAU CCA 0.1 2xl0-6 5x104 1.0 1.0 ValineAAS Val* UAC 4.0 5xl05 Val UAC 1.5 2x105 Met(m) UAC 2x104 Met(m) CAU 0.16 8x1Q-6 *Native tRNA Data from (42). 1.0 21 Table 2. Amber suppressor tRNA genes in E. coli*. Derived by genetic means Derived by synthetic means Ser (su+l)a Gln (su+2)a A1a2 Arg Tiel lle2 Tyr (su+3)a Leu (su+6)b Trp (Su+'7)c AspM Asn# Cysf Leu58 Lys Met(m) GlyUd GlyU GlyT GluA HisA Phef Gly'P *The specificity of these tRNAs is not indicated. #Not a functional suppressor, specificity unknown. From (Normanly et al, unpublished) unless otherwise indicated. a(13) b(70) C(45) d(71) e(72, 73) f(27) g(25) Pro Thr2 Val 22 Table 3. tRNAs classified by the effect of a CUA upon AAS recognition. Class I Class II Class III No affect upon charging Glutamine AAS mischarging Lysine AAS mischarging A1a2 GlyU Cys Phe ProH HisA Lys Ser (su+l) Gin (su~) Tyr (su+3) Leu (su+6;also synthetic Leu5) *Demonstrated in vitro only (44) llel GlyT Met(f)* GluA Trp (su+'?) lle2 Arg Met(m) AspM Thr2 Val 23 Table 4. Conversion to tRNAAla identity. Starting tRNA identity Nucleotide changes tRNA identity (%amino acid inserted) (E.colil Cys (C3-G7o)-+ (G-U) . Ala,tooa ' Gly C7o-+ u Phe (C3-G70)-+ (G-U) Gln,95;Gly,5b Ala,63;Phe,37a Phe,7 6;Ala,24h Phe (C3-G70)-+ (G-U); U 16-+ C; Ala,94;Lys,6h C 17-+ U; U 20-+ G; U6o-+ C; Ust-+ C Lys a (50) b (46) c (48) C7o-+ U Ala,95c 24 Table 5. Aminoacylation kinetics ofwildtype and mutant tRNAPhe transcripts. tRNA Nucleotide alterations Relative kcat/Km yeast Phe wildtype 1.0 E. coli Phe wildtype .04 E. coli Phe U2o~G .52 yeast Met~ Phe A 20--+G; C:M--+G; U 36--+A; .68 (G49-<:6s)--+(<:-G); As9--+U; yeast Arg~ Phe <:2o--+G; U34--+G; <:35--+A; U36--+A; .64 <:59--+U; <:73--+A;U--+17* yeast Tyr~ Phe (U2o•U20-l•U2o-2)--+G; A13--+<:; A22--+G; (<:12-023)--+ (U-A); (<:1-Gn)--+(G-<:); U3s--+A; A46--+G; (U2-A71 )--+(<:-G); (<:3-070)--+(G-<:) *Represents an insertion Data from (58) 1.5 25 FIGURE LEGENDS Figure 1. Secondary and tertiary structure of tRNA. In the cloverleaf structure (left) invariant nucleotides that are involved in stabilizing tertiary interactions are indicated; the lines indicate tertiary interactions between bases. Dots indicate base pairing. (Right) The L-model for tRNA tertiary structure. (4, 5) Figure 2. Cloverleaf secondary structure of tRNAAla. The G-U basepair at position 3-70, along with bases 20 and 60 (indicated by open circles) were predicted to be important alanine identity elements by statistical analysis (46). Figure 3a. Composite E. coli tRNASe:r. Bases highly conserved in serine tRNAs and not found in tRNALeuscuA are indicated. Figure 3b. Mutations introduced into tRNALeuscuA to effect the conversion from leucine to serine identity result in the mutant tRNALeu-+Ser. Figure 3c. The minimum number of nucleotides required to effect the leucine to serine conversion (25; Normanly & Abelson, unpublished). Figure 4. Nucleotides postulated to be important for yeast tRNAPhe identity (58). Figure 5. Chemical structure of Lysidine. - ~ :n ~ \. - 67 66 T stem • • AA 47 G ~ ~~ u 4950~1 ~2 53 o48 57 56 61)646362 61 • • • • • ceoTeelOOp Acceptor stem • • • • • loop I 36 37 40 Anticodon 39 stem 38 41 42 Anticodon I 34 JJU 35 • • 30 31 32 • • 28 29 ----------~27~~·~4~3 23 24 25 \ ----..6 ~5 Variable 1 r 1 2 3 4 73 72 71 70 69 68 74 c 7~C 76A stem Anticodon Oloop T loop T stem r~ N 0\ 27 I 3G U 70 Figure 2 60 28 G 73 I C 72 2 G C 71 3 AU UA 70 (~~ ~~9 20-1 D-A...--~ C48~ 21 27C G 32J Figure 3a 29 Figure 3b 30 G 73 I 2G 3A c 72 c 71 u 70 Figure 3c 31 A 73 34 35 36 Figure 4 32 I 0 0 0 + .M :C:c o-z C\J :c 0 C\J· :c 0 C\J :c (.) C\J :J: (.) ::r: z C\J :c z~ z +Z :c 0 ::r: ::r: :c 0 Figure 5 33 CHAPTER I Changing the identity of a tRNA Jennifer Normanly, Richard C. Ogden I, Suzanna J. Horvath and John Abelson Division of Biology, California Institute of Technology, Pasadena, California 91125, USA IAgouron Institute, 505 Coast Boulevard South, La Jolla, California 92037 USA [Nature (1986) 321 :213-219] 34 ABSTRACT A leucine transfer RNA has been transformed into a serine transfer RNA by changing 12 nucleotides. This result indicates that a limited set of residues determine tRNA identity. INTRODUCTION The rules for specificity in protein synthesis are to be found in the specific interactions between transfer RNA molecules and aminoacyl tRNA synthetases (AAS) (1). There exists a single AAS for each of the 20 amino acids, but because of the degeneracy of the genetic code and the limited range of allowed codon-anticoOOQ interactions (2), there are generally multiple tRNAs specifying each amino acid. The accurate transfer of information from messenger RNA to protein therefore relies on each AAS recognizing and acylating its cognate set of tRNAs while failing to acylate all others. All tRNAs so far sequenced (and the number is greater than 300) (3), can be folded into the familiar cloverleaf structure (4) and share a set of absolutely conserved nucletotides. The X-ray crystal structures of several tRNAs (5-8) show that these constant nucleotides are involved in stabilizing the tertiary structure of the tRNA molecule, therefore, by extrapolation, the tertiary structures in the cell are very similar. In some way, within the variations allowed by this framework, there exists a set of distinguishing features that constitute amino-acid specificity, that is, the identity of each tRNA. By contrast, aminoacyl tRNA synthetases are quite variable in structure; they range in size from 300 to 1,000 amino-acid residues and are found in various quaternary forms (9). Despite the variability of these enzymes, a unifying model has been proposed for the interaction between the tRNA molecules and AAS. Summarizing diverse chemical and physical studies involving several cognate complexes (10) (see also refs 9, 11, 12), Rich and Schimmel proposed that all AAS interact with tRNA along the inside diagonal of the L structure (5,6); the regions of contact inferred from these studies extend from the acceptor stem through the D stem to the anticodon loop. While the range of AAS contacts may be similar for all tRNA substrates, the rules for recognition lie in the detailed interactions and these must, of course, be idiosyncratic. For example, while the anticodon is clearly an important recognition element for the Escherichia coli tRNAfMet synthetase (13-15), this seems less likely to be the case for the serine and leucine synthetases because there are six 35 codons specifying each of their respective amino acids, requiring a set of isoacceptor tRNAs with different anticodons. A genetic approach to understanding AAS-tRNA recognition was taken over 15 years ago (16-19). In those experiments, a genetic selection was devised to isolate mutant tRNAs having altered specificity. In the initial experiments, the E. coli amber suppressor gene supF was used. This suppressor is a mutant allele of one of three tRNATyr genes in E. coli and can recognize the amber codon UAG because it has an altered anticodon, CUA (20) (for a review of nonsense suppression, see ref. 21). Among the amber mutants in bacterial and phage genes, there are some that can be suppressed by supD (serine) and supE (glutamine) but not by supF. Selection of supF mutants that could suppress this class of amber resulted in tRNATyr alleles with altered specificity. The mutations occurred in the first and second base pairs of the acceptor stem and in the fourth nucleotide from the 3' end of the tRNA ( 16,17,22-25). In all cases, the amino acid inserted by the mutant tRNAs was glutamine. This interesting result was somewhat obscured later by the fmding that amber suppressor alleles of the tRNATrp gene also insert glutamine (26). One interpretation of these results is that the glutamine synthetase is indiscriminate and charges substrates that are not recognized by other synthetases. Other alterations in tRNATyr specificity were never observed, perhaps because too many changes are required. The approach remains an attractive one, however, because the alteration of identity with retention of tRNA function, by definition, reveals elements that determine specificity. The advances in oligonucleotide synthesis allowing rapid gene synthesis (27) have prompted a return to the genetic approach to studying AAS-tRNA interaction. If multiple changes are required to alter the specificity of a tRNA, they cannot be selected but they can be constructed. This is the approach we have taken here. METHODS Strains, Media, and Enzyme Assays. JMlOlis (F' traD36, proAB, lacNZml51t1(iac proB) supE thi). XAC-A16 is (F'laclam-Z!argEam, nalA, rif, t1(lac proB) metB, ara). The strains carrying supD, supE, supF, and supP are isogenic; (lac, proB, metB, argEam, rif, nalA, ara). MC1061/9100.1 is F'(lacN'2YA tn5/ lac, galU, galK, rpl t1(ara ABIOC-leu7679 )hsr-hsm+). L broth was made from lOg!l Bactotryptone, 5g!l yeast extract, 8g!l NaCl. L plates contained 15g!l Bacto agar as well. 13-lactamase activity was 36 assayed as described by Dalbadie-McFarland eta/. (51) using ampicillin as a substrate. ~-galactosidase activity was assayed as described by Miller (44). Gene Synthesis. Deoxyoligonucleotides were synthesized on an Applied Biosystems 380A DNA synthesizer. Three to five A 260 units of oligonucleotide were electrophoresed through a 20% acrylamidenM urea gel, visualized by ultraviolet shadowing, excised and eluted in 0.3M NaCl, lOmM Tris-Hel pH 7.4, lmM EDTA, 1% phenol at 37oe overnight. The DNA was precipitated with 100% ethanol, washed with 70% ethanol, dried and resuspended in dH20. Phosphorylations were carried out on 2ug of each oligonucleotide in 70mM Tris-Hel pH 7.6, lOmM MgCI2, 5mM dithiothreitol (DTT), lOOuM ATP, lOOuei 32P-ATP, with four units of polynucleotide kinase for 1 hat 37oe, followed by phenoVchloroform extraction and ethanol precipitation. 80 pmol of each phosphorylated oligonucleotide were incubated together in lOOmM Nael at sooe for 5 min, then allowed to cool to room temperature over a period of 3 hr. The mixture was made 50mM in Nael and combined with digested vector in a 10:1 insert to vector (mass) ratio. The vector in this case was M13mp8 containing a synthetic rrnC terminator (40) inserted into the Pst I-Hind ill sites. Use of this transcription terminator was suggested by the work of Y arus et a/.(60). Ligation was carried out in 50mM Tris-Hel, pH 7 .6, lOmM Mgel2, 20mM DTT, lmM ATP, 50ug/ml bovine serum albumin with 1 unit of T4 DNA ligase at 15oe for 12 hr. The ligation mixture was used to transform competent JM101. Single-stranded DNA from the resulting plaques was immobilized on nitrocellulose filters (61) and screened for the presence of inserts by first prehybridizing in lOx Denhardt's reagent (62), 0.2% SDS, 6x SSe (63), lmM EDTA for 2 hat 65oe, then hybridizing with 4x J05 eerenkov c.p.m./rnl of one of the internal oligonucleotides in 5x DEnhardt's reagent, 6x SSe at sooe for 12 hr. The filters were washed in 6x SSe at room temperature for 5 min, then at sooe for 5 min and autoradiographed. Single-stranded DNA was prepared from positive candidates and sequenced according to the method of Sanger eta!. (64). Plasmid Construction. To construct plasmid pBR322-~am, the Eco RI-Pst I fragment of pBR322 was subcloned into M13mp8. Oligonucleotide-directed mutagenesis (65) of the ~-lactamase gene was achieved using the 19-mer 5'TAAAAGTCTAeATCATTGG3'. After 37 verification of the mutant sequence, the Eco RI-Pst I fragment was transferred back into pBR322, yielding plasmid pBR322-pam. Pvu II fragments containing the synthetically constructed tRNALeuUAG or tRNALeu~serUAG gene together with the M13mp8-derived lac promoter and rrnC terminator were subcloned into the Pvu II site of pBR322-bam to yield pBR322-pam:tRNALeuUAG and pBR322-pam:tRNALeu~SerUAG. respectively. To construct plasmid pDA3-12, the 19-mer 5'ACGTAAAGCITAAACCGTC3' was used to create a Hind ill site in between the ribosome binding site and Pribnow box of the E. colifol gene (C. R. Morris, unpublished results) encoded by a 1 kilobase fragment contained in M13mp8 (66). This was done to facilitate subcloning of the gene into a tac expression vector. Subsequently, the 23-mer 5'GATAACGCGATTCfACGCfAACG3' was used for oligonucleotide-directed mutagenesis of the fol gene. The 247 -base pair (bp) Eco RI-Hind ill fragment of ptac12H (53) containing the tac promoter was ligated into the Eco RI-Hind III site ofpBR322 to create pTAClO. Subsequently, the DHFR amber, encoded by a 1,057-bp Hind ill fragment, was subcloned into the Hind ill site of pTAC10, yielding pDA3-12. The 392-bp Pvu II fragment from pBR322-pam:tRNALenuAa or pBR322-pam:tRNALeu~serUAG containing the lac promoter, synthetic tRNA gene and rrnC-derived terminator, was subcloned into the Pvu ll site ofpDA3-12 to yield pDA3-12:Leu or pDA3-12:Leu~Ser. DHFR Purification. Two liters ofMC1061/9100.1 carrying pTAClOl (the same construct as pDA3-12, except that thefol gene is wild type) in L broth were incubated at 37oc with aeration to an A 600 of 0.4-0.6, isopropyl thiogalactoside was added to 1 mM and the culture was incubated as before for an additional 4 hr. Cells were harvested and lysed (67) and the crude lysate was extracted with ammonium sulphate (68). The 90% ammonium sulphate-saturated insoluble fraction was brought to a protein concentration of 20 mg/ml and dialysed with 3 ml of methotrexate-Sepharose (69) overnight. The resin was batch-washed and DHFR eluted with folic acid (69). To remove nucleic acid and folate, peak fractions were dialized against 50mM potassium phosphate pH 8.0, lmM DTT and applied to a DEAE-Sephacel column (lml per mg protein) equilibrated in the same buffer. DHFR was eluted in a linear gradient of 0-0.4M KCl, dialized against water and lyophilized. Mutant DHFR was purified in a similar manner but on a larger scale (121 for the tRNALeu~SerUAG substituted mutant) and the peak fractions from the DEAE column were passed through methotrexate affinity and DEAE columns a second time. Protein 38 sequence analysis was carried out on an Applied Biosystems Mcxlel470A gas-phase protein sequenator (58) by the USC Microchemical Core Laboratory. RATIONALE To defme the specific interaction between E. coli serine-specific tRNAs and their cognate AAS, we set out to transform a leucine-specific amber suppressor tRNA into an amber suppressor tRNA recognized exclusively by the serine AAS. We assume that there exists a limited set of changes one could make to a leucine tRNA that would entirely block its recognition by the leucine AAS and allow recognition solely by the serine AAS. The choice of this particular transformation was in part dictated by the fact that serine and leucine amber suppressor alleles exist in E. coli, and because all serine and leucine tRNAs fall into the type II structure class. Type II tRNAs have large extra loops in a stem-loop configuration and therefore lack the triple-base tertiary interactions, which are found in type I tRNAs (28). In addition, it is possible that the extra loop is a distinct structural feature recognized by the AAS for type II tRNAs, although this notion has not been addressed experimentally. Six different tRNASer isoacceptors recognized by the E. coli serine AAS have been sequenced (five E. coli tRNASer species and a T4 tRNASer; refs 29-34). Comparison of these sequences provides the information necessary to design the transformation. Such an approach was used previously to delineate sequences important in the recognition of tRNAPhe by the yeast phenylalanine AAS (9,35). Inspection of the serine tRNA sequences reveals 35 positions in which a nucleotide is absolutely conserved (Figure 1). It is likely that the specific residues that are necessary and sufficient for recognition of the set of tRNA Ser molecules by serine AAS are limited to these 35 nucleotides. Of these, 14 occur in all tRNAs (the constant nucleotides) and an additional seven are present in tRNALeu5. The remaining 14, detailed in Figure 2a, are concentrated in the acceptor stem and D arm, together with one each in the T'rC loop, extra arm, and anticodon loop, and two comprising a base pair in the anticcxlon stem. Figure 2b shows the proposed changes for the leucine to serine transformation, the justifications for which are as follows. We chose to focus only on the acceptor stem and D stem and loop as these domains have been shown to be in contact with the synthetase in all cases studied. In this initial exercise, we have ignored the T'rC stem-loop, the extra arm and the anticodon stem and loop. Although the anticcxlon has been strongly implicated in 39 some tRNA-AAS interactions (13-15,26,36,37), the existence of at least six isoaccepting species of serine tRNA, recognizing six different codons, and the existence of a serine tRNA suppressor allele, argues against a crucial role for the anticodon region in the case of serine tRNA recognition. Accordingly, the conserved tRNASer_specific bases at positions 32 in the anticodon loop and the base pair 27-43 in the anticodon stem were not incorporated into the tRNALeu5 framework. We have thus superimposed the conserved serine nucleotides at positions 2, 71, 72 and 73 in the acceptor stem as well as positions 20-1, and 21 in the D loop, and base pair 11-24 in the D stem. Looking at probable tertiary interactions in the leucine and serine tRNAs, there are five base-pair interactions common to both type I and ll tRNAs that appear to be responsible for stabilization of the tertiary structure (5,6,28); four of these, G 18-U55 , G 19-C56, U 54-A58 and U 8-A14, involve constant nucleotides, and are therefore not directly affected by the alterations we propose. The fifth, a transtertiary base pair between residues 15 and 48, is G 15-C48 in all serine tRNAs. As this tertiary interaction is supplied in tRNALeu5 (albeit by an A-U pair), we elected not to alter bases 15 and 48. A difference in primary structure between tRNASer and tRNALeu which could account for a subtle difference in tertiary structure between the two sets of tRNAs, lies in the position of the constant nucleotides G 18 and G 19 in the D loop. Note in Figure 1 how the positions of G 18 and G 19 in tRNASer are offset by one nucleotide in comparison with G 18 and G 19 in tRNALeus. The question at this point becomes whether or not to maintain rigorously the leucine tRNA structure, and merely superimpose the predicted serine determinants on the leucine scaffold, or to alter the nature of the tertiary interactions to resemble those of a serine tRNA. As the position of 0 18 and 0 19 is a highly conserved feature of the serine tRNAs, we felt it may well be important for serine AAS recognition, especially in view of its location within the structure. Through an insertion and a deletion in the D loop of tRNALeus, these bases are kept in the same position relative to the D stem that is seen in all six tRNASer sequences. Finally, the 0-C to A-U change in the third base pair of the acceptor stem was not dictated by absolute conservation of that sequence among serine tRNAs but was suggested by a statistical study of E. coli tRNA sequences (W. H. McClain and H. B. Nicholas, personal communication) in which it was found that an A-U or U-A base pair at that position is significantly correlated with all tRNASer sequences and is anti-correlated with 40 other tRNA sequences. The rest of the proposed changes are compatible with the significant serine-specific residues detected in McClain's study. To effect this transformation of a leucine amber suppressor tRNA, we first synthesized a leucine amber suppressor tRNA gene and then altered it at the residues indicated in Figure 2b. SYNTHESIS OF tRNALEU SUPPRESSOR GENE The sequence of the starting leucine amber suppressor gene was based on the sequence of E. coli tRNALeus with substitution of the anticodon to CTA. tRNALeus recognizes the codon UUA and has a different sequence from the leucine amber suppressor supP, as became apparent from studies published after the present work was started (38,39). The synthetic gene, tRNALelluAo, is comprised of six overlapping oligonucleotides, which were annealed and ligated into the M13 vector mp8 containing a synthetic transcription terminator (see Figure 3). In this construction, transcription of the tRNA gene is directed from the lac promoter and is expected to terminate within the sequence derived from the ribosomal RNA cistron rrnC (40). No provision was made for the inclusion of natural 5' and 3' flanking sequences, as the nucleases that process tRNA precursor transcripts are not sequence-specific; instead, they recognize conserved features of secondary and tertiary structure common to all tRNAs (41). It was therefore anticipated that a tRNA could be correctly processed from any flanking sequence context. To ascertain that the synthetic gene produced a functional tRNA, we assayed for suppression of an amber mutation in an E. coli strain carrying the tRNALeuUAG gene. The amber mutation resides in the N-terminallacl portion of a lacl-Z fusion gene (42). Activity of the lacZ gene product, (>-galactosidase, is unaffected by the lacl portion of the fusion, hence quantitation of ~-galactosidase activity gives a measure of the extent of suppression of the amber mutation. The tRNALeuUAG gene, together with its terminator, was subcloned into plasmid pEMBL8+ (ref. 43) and transformed into the lacl-Z amber strain XAC-A16. Suppression efficiency was determined directly by assaying for (>-galactosidase activity (44). When tested in this way, the tRNALeuUAG suppressor is 60% efficient, producing substantial levels of (>-galactosidase. 41 SYNTHESIS OF tRNALEU~SERUAG GENE For construction of the tRNALeu~serUAG gene, four additional oligonucleotides were synthesized and combined with two of the tRNALeuUAG gene chains as shown in Figure 3. This gene was then transferred to the pEMBL8+ vector and its activity assayed using the lacl-Z amber strain. The tRNALeu~serUAG gene is active but produces a much weaker suppressor than its parent, exhibiting a 0.5-1% efficiency of suppression. By contrast, the supD suppressor (serine) is 35% efficient. To determine the specificity of this suppressor, two approaches were taken. The first approach uses an amber mutant in the ~lactamase gene that is functionally suppressed only by a serine suppressor. This approach allows the selection of serine amber suppressors. If our design of a suppressor with serine identity had been wrong by one or two nucleotides, the serine-specific amber could have been used to select an active serine suppressor by spontaneous or induced mutation. Alternatively, this mutant could be used to select stronger variants of a weak serine suppressor. In the second approach the tRNALeu~SerUAG suppressor was used to suppress an amber mutation near the beginning of the E. coli dihydrofolate reductase genefol, and subsequently the suppressed protein was sequenced. SERINE-REQUIRING AMBER MUTATION The active site of ~lactamase contains an essential serine residue (45,46), which participates in the cleavage of the ~lactam ring of ampicillin and thereby confers antibiotic resistance to cells harbouring the f3-lactamase gene amp. By oligonucleotide mutagenesis we converted the AGC serine codon at residue 68 (residue 70 according to Ambler's nomenclature, ref. 47) of pBR322 amp gene to TAG (Figure 4a). Figure 4b shows that this mutation can be suppressed in an E. coli supD strain (supD inserts serine) but not by supE (glutamine), supF (tyrosine) or supP (leucine). The only other amino acids, which might be expected to substitute for serine at this position based on some knowledge of the reaction mechanism (48), are cysteine and threonine. These substitutions have been made (46,49-51) and in the case of cysteine resulted in an enzyme with 1/30 of wild-type activity. In the case of threonine, the enzyme was inactive (51). The active-site amber mutation therefore meets our requirements as apparently it can be functionally suppressed 42 only by a serine-inserting suppressor. The synthetic suppressor genes tRNALeuUAG or tRNALeu~SerUAG, together with the lac promoter and rrnC terminator, were ligated into the Pvull site of pBR322-~Wu. Each plasmid was introduced into E. coli and the transformants were tested for growth on tetracycline and ampicillin. Only the tRNALeu~SerUAG suppressor allowed growth on ampicillin (Figure 4c). Futhermore, the cells containing tRNALeu~seruAo-suppressed !3-lactamase are fully viable in the presence of the drug at 80 J,Jg mi-L We conclude that tRNALeu~serUAG is a functional tRNA that inserts serine. This experiment cannot, however, exclude the possibility that the tRNA inserts other amino acids also. The only way to determine the specificity of suppression is to sequence through the site of the amber mutation in a suppressed protein. tRNALEU-+SERuAG SPECIFICITY To determine the specificity of the tRNALeu~serUAG suppressor, we needed to use it for suppression of an amber mutation close to the beginning of a gene whose product is relatively small, easy to purify, and present at high concentration. The E. coli dihydrofolate reductase (DHFR) gene,fo/, proved an attractive candidate. We linked the fol gene, the sequence of which is known (52), to the tac promoter (53) in the expression vector pTAClO (Figure 5). The 159-amino acid DHFR protein represents approximately 10% of the total protein in fully induced strains of E. coli harbouring this plasmid, and is easily purified by methotrexate affinity chromatography. An amber mutation at a site near the N-terminus would allow sequence determination of the suppressed DHFR by Edman degradation (54) of the intact protein. Residues 2-8 constitute a strand of a !3-pleated sheet and are part of the NADP binding site. We chose position 10, a valine residue found in a loop on the surface of the protein, because this residue is not phylogenetically conserved and X-ray crystallographic data reveal that it is well removed from the active site (55-57). Substitution of any residue at this site should not affect the purification of DHFR on a methotrexate affinity column. Another point to consider is that the efficiency of amber suppression and therefore the yield of suppressed protein is dependent on the context of the amber codon. UAG codons followed by A residue are usually better suppressed than those followed by a G (42). Consequently, by oligonucleotide-directed mutagenesis, we altered the DHFR sequence GTA-GAT of codons 10 and 11 to TAG-AAT, changing the protein 43 sequence from Val-Asp to amber-Asn. This alteredfol gene, together with its tac promoter, was transferred to plasmids carrying either the tRNALeuUAG or tRNALeu--tSerUAG genes (Figure 5a). Figure 5b documents the purification of DHFR from a strain harlx>uring the wild-type gene under control of the tac promoter (pTAC101). Dihydrofolate reductase was purified from four strains: one carrying pTAC101, another a supD strain harbouring the DHFR amber mutant in the tac expression vector (pDA3-12), and the third and fourth carrying pDA3-12:Leu and pDA3-12:Leu~er, respectively. Each purified protein was sequenced by Edman degradation using an automatic gas-phase sequencer (58). In interpreting these results (Figure 6), it should be considered that the chromosomal copy of the fol gene is present in these strains, leading to fractional yield (usually small) of Val-Asp at positions 10 and 11. This sequence is very clearly seen when wild-type DHFR is sequenced but when the DHFR amber is suppressed by the tRNALeuuAG suppressor, the sequence is unambiguously Leu-Asn at positions 10 and 11. Overproduction of the plasmid-encoded DHFR is probably greater than 100-fold and the tRNAl..euuAG suppressor is efficient. When the DHFR amber is suppressed by supD, serine is clearly present in position 10, demonstrating that the well known problems in detecting serine in Edman degradation have been overcome. Here, valine is clearly not detected at position 10 but there is some aspartate at position 11; we believe this results from deamidation of asparagine. When the DHFR amber is suppressed by the tRNALeu--tSerUAG suppressor, there is a good yield of serine at position 10, but also a detectable (12-15%) amount of valine. The presence of valine is due to the fact that the tRNALeu--tSerUAG suppressor is weak, giving rise to a lower ratio of suppressed protein to wild-type protein. There is also a small amount ( 10%) of leucine at position 10; this is probably due to carry-over of the leucine from postion 8, but the experimental method used cannot exclude the possibility that the tRNAl...eu--tSerUAG suppressor inserts small amounts of leucine or valine. We could detect no other amino acid at position 10, and conclude that we have changed a strong leucine suppressor to a weak serine suppressor. 44 DISCUSSION Comparison of the sequences of six serine tRNAs, which are substrates for the E. coli serine AAS reveals several conserved nucleotides. Here we altered an amber suppressor allele of tRNALeus at 12 positions to yield a tRNA sequence containing a subset of these conserved serine residues-those located in the acceptor stem and in the D stem and loop. The resultant molecule is a weak suppressor, which inserts serine in response to amber nonsense codons. The specific acquisition of tRNASe:r identity was established by a highly discriminatory functional assay, while the loss of tRNALeu identity in exchange for unique acquisition of tRNASer identity was established by protein sequencing. Our study demonstrates that a limited number of nucleotides by virtue of their nature and position on a tRNA scaffold, are sufficient to dictate cognate synthetase recognition. It is noteworthy that this result was achieved with only primary and secondary structure information about the two tRNAs and no knowledge of the AAS structures. A number of questions remain of course. We have converted a strong suppressor into a weak suppressor. This reduction in efficiency may be the result of a weak interaction between the suppressor tRNA and serine AAS. Alternatively, it may reflect a perturbation in the general tRNA structure resulting in poorer competition with the release factors. THere are two possible routes to producing a better suppressor with serine specificity. First, we can further alter the sequence by including those changes in the anticodon stem and TPC loop that were conserved in all serine tRNAs, but which we chose to ignore. Our predictions about the importance of specific tertiary interactions in AAS recognition also warrant a more systematic investigation. Perhaps a better method, however, is to mutagenize the gene and select stronger serine suppressors by requiring suppression of the ~-lactamase active-site amber in the presence of increasingly higher levels of ampicillin. When a strong serine suppressor is generated, the sequence of the tRNA will be determined. At that point, we can begin a systematic analysis of the identity switch to determine which changes were necessary. Eventually, the minimum set of changes required to transform tRNALeu to tRNASer can be determined. This information will then be used to effect a different transformation, for example, tRNATyr to tRNASer. It will clearly be necessary to gather supporting in vitro kinetic and thermodynamic data for the various tRNAs and AAS involved. This iterative method can lead to a more detailed 45 understanding of the essence of serine tRNA identity. ACKNOWLEDGEMENTS We thank Joe Kraut for assistance in the design of the DHFR amber mutant, Lynn Williams of the USC Microchemical Core Laboratory for protein sequencing and Jeffrey Miller for discussion and bacterial strains. The manuscript was improved by comments from Olke Uhlenbeck, Paul Schimmel, LaDonne Schulman and Bill McClain. J.N. thanks Marge Strobel and Eric Phizicky for technical advice, and Rich Schwartz and Peter Johnson for advice and interest. This work was supported in part by an Nlll grant (GM32637) to J. A., an NSF grant (DMBA3-18244) toR. C. 0., an Nlll grant (GM06965) to S. J. H. and by a National Research Service Award (T32GM07616) from the National Institute of General Medical Sciences to J. N. REFERENCES 1. 2. 3. Chapeville, F. et a/.1962. Proc. Nat/. Acad. Sci. U.S A 48:1086-92 Crick, F. H. C. 1966. J. Mol. Bioi. 19:548-55 Sprinzl, M., Moll, J., Meissner, F., Hartmann, T.l985. Nucleic Acids Res. 4. 5. 6. 7. 8. 13:r1-49 Holley, R. W. et al. 1965. Science 147:1462-65 Robertus, J.D. et a/.1974. Nature 250:546-51 Kim, S. H. et a/.1974. Science 185:435-40 Schevitz, R. W. et a/. 1974. Nature 278:188-90 Westhof, E., Dumas, P., Moras, D.l985. J. Mol. Bio/. 184:119-45 9. 10. 11. 12. 13. 14. 15. Schimmel, P.R., Soli, D. 1979. Ann. Rev. Biochem. 48:601-48 Rich, A., Schimmel, P.R. 1977. Nucleic Acids Res. 4:1649-65 Schimmel, P.R. 1977. Acts.Chem.Res. 10:411-18 Schimmel, P. R.l979. in Transfer RNA: Structure, Properties and Recognition (eds. Schimmel, P.R., Soli, D., Abelson, J.) Cold Spring Harbor Laboratory, New York pp. 297-310 Schulman, L. H., Pelka, H., Susani, M.l983. Nucleic Acids Res. 11:1439-55 Schulman, L. H., Pelka, H. 1983. Proc. Nat/. Acad. Sci. U.S.A. 80:6755-9 Schulman, L. H., Pelka, H. 1985. Biochemistry 24:7309-14 46 22. Hooper, M. L., Russell, R. L., Smith, J.D. 1972. Febs. Lett. 22:149-55 Shimura, Y. eta/.1972. Febs. Lett. 22:144-48 Ghysen, A., Reyes, 0., Allende, C. C., Allende, J. E.l973. in Gene Expression and Its Regulation (eds Kenney eta/.) Plenum Press, New York pp. 209-16 Ozeki, H. eta/. 1980. in Transfer RNA: Biological Aspects (eds. Schimmel, P.R., Soll, D., Abelson, J.) Cold Spring Harbor Laboratory, New York pp. 341-62 Goodman, H. M., Abelson, J., Landy, A., Brenner, S., Smith, J. D.1968. Nature 217:1019-24 Steege, D. A., Soll, D. 1979. in Biological Regulation and Development Vol. I (ed. Goldberger, R. F.) Plenum Press, New York pp. 433-85 Smith, J.D., Celis, J. E. 1973. Nature new Bioi. 24~Q6-71 23. 24. 25. 26. Celis, J. E., Hooper, M. L., Smith, J. D.l973. Nature new Bioi. 244:261-64 lnokuchi, H., Celis, J. E., Smith, J. D.l974. J. Mol. Bioi. 85:187-92 Ghysen, A., Celis, J. E. 1974. J. Mol. Bioi. 83:333-51 Yaniv, M., Folk, W. R., Berg, P., Soll, L.l974. J. Mol. Bioi. 86:245-60 27. 28. 29. 30. 31. 32. 33. 34. Khorana, H. G. 1979. Science 203:614-25 Brennan, T., Sundaralingam, M.l976. Nucleic Acids Res. 3:3235-3251 Steege, D. A. 1983. Nucleic Acids Res. 11:3823-32 Yamada, Y., Ishikura, H.l973. Febs. Lett. 29:231-34 Ish-Horowicz, D., Clark, B. F. C.l973. J. Bioi. Chem. 248:6663-73 lshikura, H., Yamada, Y., Nishimura, S.l971. Febs. Lett.l6:68-70 McClain, W. H., Barrell, B. G., Seidman, J. G.l975. J. Mol. Bioi. 99:717-32 Grosjean, H., Nicoghosian, K., Haumont, E., Soll, D., Cedergren, R. 1985 16. 17. 18. 19. 20. 21. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. Nucleic Acids Res. 13:5697-706 Roe, B., Sirover, M., Dudock, B. 1973. Biochemistry 12:4146-54 Bruce, G. A., Uhlenbeck, 0. C. 1982. Biochemistry 21 :3921-26 Kisselev, L. L.l985. Prog. Nucleic Acids Res. Mol. Bioi. 32:237-66 Yoshimura, M., Inokuchi, H., Ozeki, H. l984. J . Mol. Bioi. 177:627-44 Thorbjarnardottir, S. eta/. 1985. J. Bact. 161:219-22 Young, R. A. l979. J . Biol.Chem. 254:12725-31 Abelson, J. 1979. Ann. Rev. Biochem. 48:1035-69 Miller, J. H., Albertini, A.M. 1983. J . Mol. Bioi. 164:59-71 Dente, L., Cesareni, G., Cortese, R.l983. Nucleic Acids Res. 11:1645-55 Miller, J. H. 1972. in Experiments in Molecular Genetics Cold Spring Harbor 47 Laboratory, New York pp. 352-55 45. Knott-Hunziker, V., Waley, S. G., Orlek, B.S., Sammes, P. G.l979. Febs. Lett. 99:59-61 46. Dalbadie-McFarland, G. et al. 1982. Proc. Nat/. Acad. Sci. U.S.A. 79:6409-13 47. Ambler, R. P. 1980. Phil. Trans. R. Soc. B289:321-31 48. Fisher, J., Belasco, J. G., Khosla, S., Knowles, J. R. 1980. Biochemistry 19:2895-901 49. Sigal, I. S., Harwood, B. G., Arentzen, R. 1982. Proc. Nat/. Acad. Sci. U.S.A . 79:7157-60 50. Sigal, I. S., DeGrado, W. F., Thomas, B. J., Petteway, S. R. 1984. J. Bioi. Chem. 259:5327-32 51. Dalbadie-McFarland, G., Neitzel, J. J., Richards, J. H.l986. Biochemistry 25:332-8 52. 53. Smith, D. R., Calvo, J. M.l980. Nucleic Acids Res. 8:2255-74 Amann, J. E., Brosius, J., Ptashne, M.l983. Gene 25:167-178 54. Edman, P., Begg, G.l967. Eur. J . Biochem. 1:80-91 55. Bolin, J. T., Filman, D. J., Matthews, D. A., Hamlin, R. C., Kraut, J. 1982. J. 56. 57. 58. 59. 60. 61. 62. 63. 64. Bioi. Chem. 257:13650-62 Matthews, D. A. eta/. 1977. Science 197:452-55 Filmann, D. J., Bolin, J. T ., Matthews, D. A., Kraut, J. 1982. J. Bioi. Chem. 257: 13663-72 Hewick, R. M., Hunkapiller, M. W., Hood, L. E., Dreyer, W. J. l981. J. Bioi. Chem. 256:7990-7 Yamaizumi, Z., Kuchino, Y., Harada, F., Nishimura, S., McCloskey, J. A. 1980. J. Bioi. Chem. 255:2220-5 Yarus, M., McMillan, C., Cline, S., Bradley, D., Snyder, M.l980. Proc. Nat!. Acad. Sci. U.S.A. 77:5082-96. Maniatis, T., Fritsch, E. F., Sambrook, J. 1982. in Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory, New York pp. 320-21 Denhardt, D. T. l966. Biochem. Biophys Res. Com. 23:641 -6 Maniatis, T ., Fritsch, E. F., Sambrook, J. 1982. in Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory Press, New York p. 447 Sanger, F., Nicklen, S., Coulson, A. R.l977. Proc. Nat/. Acad. Sci. U.S.A. 74:5463-7 48 68. Newman, A. J., Ogden, R. C., Abelson, J.l983. Cell35:117-25 Villafranca, J. E. et al., 1983. Science 222:782-8 Baccanari, D., Phillips, A., Smith, S., Sinski, D., Burchall, J. 1975. Biochemistry 14:5267-73 Baccanari, D.P., Averett, D., Briggs, C., Burchall, J.l977. Biochemistry 69. 16:3566-72 Baccanari, D.P., Stone, D., Kuyper, L., 1981. J. Bioi. Chern. 256:1738-47 65. 66. 67. 49 FIGURE LEGENDS Figure 1. Serine tRNAs recognized by E. coli serine AAS (29-34) compared with E. coli tRNALeu5 (ref. 59 and K. Jones B. Reid, personal communication). Boxed nucleotides indicated absolute conservation among the five tRNASer sequences. The conserved serine nucleotides that differ from tRNALeu5 have been underlined in the tRNALeu5 sequence. A fifth E. coli serine tRNA has been sequenced (not shown) and is identical to that of E. coli 4, except for residue 20, which is a C (39). Ace. stem, acceptor stem. A. C., anticodon. Figure 2. a)Cloverleaf representation of conserved E. coli tRNASez- residues. The indicated nucleotides are those that are absolutely conserved in all six serine tRNAs and not found in tRNALeu5. b) tRNALeu5 with the changes proposed to confer tRNASerspecificity. The modified nucleotides are those present in tRNALeu5. Note that the changes in the D loop at positions 20-2 and 21 are equally well represented by insertion of an A between residues 21 and 22. Thus there are actually 11 independent changes. Figure 3. The tRNALeuUAG gene was constructed from six oligonucleotides containing the tRNALeuUAG sequence flanked by Eco RI and Pst I restriction endonuclease cohesive ends at the immediate 5' and 3' ends of the gene respectively. Junctions between oligonucleotides are indicated. The four oligonucleotides containing the appropriate sequence alterations for construction of tRNALeu-+Sez-UAG exactly replace four of the tRNALeuUAG oligonucleotides, and are shown above and below the tRNALe~AG sequence. The sequence of the terminator, synthesized as two chains, is indicated. Figure 4a. Sequence of the serine-requiring amber mutation in the amp gene of pBR322, generated by oligonucleotide-directed mutagenesis. Pvu II fragments containing the synthetically constructed tRNALeuUAG or tRNALeu-+SerUAG gene together with the M13mp8-derived lac promoter and rrnC terminator were subcloned into the Pvu ll site of pBR322-~am to yield pBR322-~am:tRNALeuUAG and pBR322-~am:tRNALeu-+SerUAG. respectively. Figure 4b. Suppression of the serine-requiring amber mutation. Isogenic E. coli strains carrying the supD, supE, supF and supP alleles were transformed with pBR322-~am and 50 plated on L plates containing either 20ug/ml tetracycline or 80ug/ml ampicillin. Figure 4c. Suppression of serine-requiring amber mutation by tRNALeu-+SerUAG· XAC-A16 was transformed with either pBR322-~am:tRNALeuUAG or pBR322-~am:tRNALeu-+SerUAG and plated as in (b), except that IPTG was added to lmM. ~-lactamase activity in the latter strain was 2% the level seen in a strain harbouring wild-type pBR322 and 5% the level seen when this mutation was suppressed by supD. Figure 5 a) Sequence of the fol amber. Asp 11 (GAT) was changed to Asn (AAT) to improve the context of the amber mutation. Pvu II fragments encoding either tRNALeuuAG or tRNALeu-+SerUAG along with the lac promoter and rrnC terminator were subcloned into the Pvu II site of plasmid pDA3-12 to yield plasmids pDA3-12:Leu and pDA3-12:Leu-+Ser, respectively. b) DHFR purification visualized by Coomassie blue staining of an SDS-polyacrylamide gel. Lane 1, crude extract; Lane 2, 90% ammonium sulphate-saturated insoluble fraction; Lane 3, flow through from methotrexate sepharose column; Lane 4, folic-acid eluate from methotrexate sepharose; Lane 5, peak protein fraction from DEAE sephacel column. Figure 6. Protein sequence analysis of wild-type DHFR (frrst column of graphs), DHFRam/Leu suppressor (second column), DHFRam/Ser suppressor (third column), and DHFRam/Leu-+Ser suppressor (fouth column). The picomolar yields of phenylthiohydantoin (PTII) derivatives of valine, aspartic acid, asparagine, leucine and serine are plotted against the residue number for wild-type DHFR, isolated from E. coli carrying pTAClOl, and mutant DHFR isolated from strains carrying pDA3-12:Leu, or pDA3-12:Leu-+Ser, as well as from a supD strain carrying pDA3-12. ~..... :!1 GCUGACCUG U C C G A G U - - Gm G 0 0 G A A G G ·A G C A C C C C U G C A A A G + G U G CGAGGCGUG G C A G A G U - - Gm G 0 0 U A A U G C A C C G G UCm U N G AxA A A C C GI G E. coli 4 Phage T4 Leu E. coli 5 tRNA T stem T loop T stem Ace. stem · C G G C G U U C G C G C U G - - - - - - QG C G G G T + C A A i U C C C G C U C C G G i !!. .!. C C A C A G U C GC U C C G GC G A C U - - - ~ A U A G GT + C A A A U C C U A U C GC C U C C G C C A U A U A C G G C A A C G U A U - - - - - C GG G G G T+ C G A A U C C C C C C C U C A C C G C C A U A U G C G G U C A A A A G C U G C AU C C G G G G T·+ C G A A U C C C C G C C U C A C C G C C A A GU A G GG GC A A C U C U A C - - - C GG GG G T + C A A A U C C C C C U C U C U C C GC C A C GA C C C G A A A G G G U U - - - - - C C A G A G T+ C G A A U C U C U G C G C U U C C G C C A 44 45 46 4747474747474747 474747 4747 47 47 4747 48149 50 51 52 53154 55 56 57 58 59 60161 62 63 64 65166 67 68 69 70 71 72173 74 7576 I 2 3 4 5 6 1 8 9 10 II 12 13 14 15 16 Extra arm G~ CCGGAUG G !!. G G A .!. 0 C - Gm G 0 .!. - ~ A C .!. C A A G G G A :!: U N A AAx A + C C C Q CGUGAGGUG G C C C A C A - - C C C 0 G A A C C C C C U C C C,c U C C UtA A C C G A G tRNA E. coli 3 A.C. stem GGAGACAU C C C C C A C C - - Gm C C 0 C A A C G G A C C C C U C U C G A A A A C C G C A.C . loop E. coli 2 A.C. stem CCAAGUGUG G C C C A G C - - Cm G 0 0 C A A C C C A C C C G UCm Uo U C AxA A A C C C C D stem E. coli I D loop 1234567 89 10 II 12 13 14 15 16 171718 19 20 20 20 21 22 23 24 25 26 27 28 29 30 31 32 3334 35 36 37 38 39 40 4042 43 I I 2 D stem Ser Ace. stem ..... VI :!1 16• II • • 21 D• A 20-1 • 19• e • G • • • I I I e • • • • •• 32C • - - e - e 27C - G43. e I • e ... ... • • •• e I •G• 24 • C46 • •• 59 •• • • • • A • I II I I • ••••• • • • I I I • C• N • - • - - • - 15 .. .. 2 G - C 11 3 AU -UA7o ~ ~ (JQ • - c 72 • • • G 73 Composite tRNASer \ ll AA \ l If IIIII A 1\\ u C UA ms216A A G G GCGGG C GUG U T If C Gm I II G CAC 1 UC G 0 A GA A CGI I GC G A-U GCI 1 G G -C G C U G A G -C UU G -c A -If 0 UAG c s4 U A-U CGCCC U G A G....._C-G-+C A....._C-G-U C -G G -c G -c c c A-G pG- u -c A UAG E. coli tRNALeu _. tRNALeu-+Ser lJl N Ul ~· 'Tl GTCCCGCTCCGTCCGCCACTGCA Plac Eco RI Pst I rrnC terminator GATCCTTAGCGAAAGCTAAGGATTTTTTTTA ACGTCTAGGAATCGCTTTCGATTCCTAAAAAAAATTCGA ' GCG GGC CTACCACCTT AGC CATCTGTG1iTTCCCTAAGATTTAGGGAGCCGCAAGC GCGACACGqCCAAGTTCAGGGCGAGGCCCATGGTG , 5 _ Pst I 3 GCCTG CCTACCGCCTTACC AACTTGCG CCAAGTTC AGGGCGAGGCAGGCGGT G AATTCGCCCGGATGGTGGAATCGGTAGACACAAGGGA~TCTAAATCCCTCGGCGTTCGCGCTGTGCGGGTTCA~GTCCCGCTCCGGGTACCACTGCA 3' ,AATTCGGACGGATGGCGGAATGGTTGAACGCAAGGGAT EcoRI 5 tJt Ul j:l) ~ ~ ~ .... {3 am: {3wt: "-,-J Act1ve site 10 {3 am sup tRNA gene pBR322-{3am rrnC terminator . . . . . . . • . . • . • . . . . . . . . . . . TAG . . . . .. CCC GAA GAA CGT TTT CCA ATG ATG AGC ACT TTT Pro Glu Glu Arg Phe Pro Met Met Ser Thr Phe GO pBR322 ,B-Iactamase amber ~ VI 55 -co <.0 ~ Q) en - a. ~ Q) E ..0 <( E 0 Q) (/) 0 E (.) -- -- -- -- 0 _J 0 +- I en. N N ::J Q) ~ ~ c ~ Q) -- -u..a. --wa. --a. ~ <..? a_ a. ::J (J) (f) 0 ::J (J) ::J (J) ::J (f) r<) 0::: CD CL - ~ 0 c: Q) 0 ~--,~-- .-. ! . ·. _; .~ (/) (/) Q) ~ CL CL :::::::J (f) Figure 4b ~ "T1 ..... (') .j:>.. ~ ()'q Tet UAG tRNA Leu-+Ser UAG t RNA Leu5 Amp Suppression of pBR322 ,8-loctomose amber with tRNA~~;ser .. VI 0\ 57 DHFR amber ~ 2 DHFR wt: lie Ser Leu lie Ala Ala Leu Ala Vol Asp Arg ATC AGT CTG ATT GCG GCG TTA GCG GTA GATCGC DHFR om : .•• om Asn •• • • •• TAG AAT •• • Ptoc pDo3- 12 Sup !RNA tet p terminator sup tRNA gene 1 2 3 Figure 5 4 5 a U Qf 111 u .. • • 10 •• lfll i.lill • OHFRan Leoci'le S\4>pressor m I! • • u •• I [ 180 U llJioLI lP llJ V. ill IlK a DH'Ran Seri'le Suppressor Ill II lOili Ill 11.1 • Ill DHFRanleo45e< Supp<essOf • • Ill 1U lh lf .tU ., 111 till Ml Uil • 15 15 300 5 5 10 10 15 15 I .0 IS ~ 240 so c. 160 5 I 5 10 15 ~ 1 I 10 IS ~~ 5 10 15 .............. I 10 , ~·. ~ 5 5 10 tlJJC__ t~~J t :~ 5 ~ m~ I~ i1$~ ] ~ 10 I1 15 5 0\ 10 5 15 ~ ~~~----------~A~S~N~~~~----------1 10 ~ 10 t ~! ]\/\ J 1~ 1\fi( , I ~~L&l 1 l~ ~~ .c:-; ..~:::::1\ J = ~ t i~!=xJ ~~ lb a Ut • ~· "'1'1 DH'R wl • 1 Ill ~I lit Ill IU WI Ill ._. I I I 00 (.JI 59 CHAPTER II Eight base changes are sufficient to convert a leucine-inserting tRNA into a serine-inserting tRNA 60 INTRODUCTION The identity of a tRNA is defined by the elements within the tRNA that are responsible for the correct aminoacylation of the tRNA by its cognate aminoacyl tRNA synthetase (AAS). The correct aminoacylation of a tRNA is the outcome of two types of interactions, 1) the positive interaction between a tRNA and its cognate AAS, and 2) the negative interaction between a tRNA and all 19 noncognate AASs. The determinants of these positive and negative interactions are in the nucleotides of any given tRNA. One method that has been used to determine which nucleotides comprise a tRNA's identity is referred to as an identity swap. In this approach, the identity of a tRNA is altered with the fewest changes, while retaining its complete biological function. Those changes responsible for redirection of AAS recognition constitute identity elements. In an effort to defme the identity elements of E. coli serine tRNA, we have developed a system to facilitate the identity swap approach. Amber suppressor tRNA genes containing the alterations predicted to redirect AAS recognition are assembled from oligonucleotides and inserted in cassette form into a plasmid, which supplies the inducible lac promoter and a transcription terminator (1). First, the biological function of the mutant tRNA is established by assaying the efficiency at which the tRNA can suppress a LacZ amber mutation. This mutation, in a non-essential portion of LacZ, can tolerate any amino acid. Second, we screen specifically for the insertion of serine using a mutant of the amp gene, which contains an amber mutation in the essential, active-site serine residue (2). Finally, the amino acid specificity of the supperssor is determined by sequencing the protein product of a suppressed amber mutant gene. For this purpose the E. coli dihydrofolate reductase (DHFR) gene,fol, with an amber mutation at codon ten is used. DHFR is easily purified by affmity chromatography on a methotrexate resin. X-ray crystallographic data (4, 5) reveals that residue ten of DHFR lies on the surface of the protein, well removed from the active site, and it appears that any amino acid is acceptable at this position. If more than one amino acid is inserted by the suppressor, they can be readily detected by this method. Protein sequence data acquired in this manner reveal the outcome of the competition among all 20 AAS in the cell for the suppressor. Previously, we reported the conversion of a leucine-inserting amber suppressor tRNA into a serine-inserting amber suppressor tRNA (1). Comparison of six E. coli serine-accepting tRNAs revealed 16 nucleotides, which were highly conserved among, or highly correlated with, serine tRNAs, yet not found in the leucine tRNA that we used as a 61 starting tRNA. Twelve of the 16 nucleotides were superimposed upon the amber suppressor allele oftRNALeu5, by constructing a tRNA gene from synthetic oligonucleotides that encoded the leucine tRNA with the twelve base changes that we wished to incorporate. The resultant tRNA, tRNALeu~serUAG was a weak suppressor (1 % efficient as determined by ~-galactosidase assays) yet it insetted predominantly serine (Table 1). Here, we describe our efforts to determine the minimum number of base changes necessary to effect the leucine to serine conversion, and to increase the suppression efficiency of the tRNA. We speculated that not all of the 12 nucleotide changes were necessary for the conversion and may have been responsible for the low suppression . efficiency of tRNALeu~serUAG· By reverting the mutations singly or in groups, we . determined that a minimum of eight base changes were required to effect the leucine to serine conversion. They were basepair C11-G24 in the dihydrouridine (D) stem, base pairs GrC71 , A 3 -U7o. and nucleotides Cn and G 73 , all in the acceptor stem. These eight nucleotides, when incorporated into a leucine tRNA, yield a tRNA that inserts exclusively serine with 40% suppression efficiency. METHODS Gene Construction. The tRNA genes were constructed from synthetic oligonucleotides (made by the Caltech Microchemical Facility), which were assembled as described previously (1), and cloned into the vector pJN10. pJN10 is derived from plasmids pBR322 (6) and pEMBL8+ (7). The Pvui-Pvuii fragment from pEMBL8+ containing a portion of the amp gene, the replication origin, and the polylinker, was ligated to the Pvui-Pvuii fragment of pBR322 that carries the tet gene. In pJN10, the amp gene lacks a Pst I site, and, additionally, contains an amber mutation at its active site serine. Prior to constructing pJNlO, the Eco RI site in pBR322 was removed by digesting pBR322 with Eco RI, filling in the "sticky ends" with Klenow and ligating. tRNA genes are ligated into the Eco RI-Pst I sites located in the polylinker region ofpJN10, which are flanked by the lac promoter and rrnC terminator, 5' and 3', respectively. Plasmid pJN10, into which a suppressor tRNA gene has been cloned, will be referred to as pJN10:sup. 62 Determination of suppression efficiency: Suppression efficiencies were detennined by assaying ~-galactosidase activity (7) in strain XAC-1 (F'laci373 LacZullBam proB+fil(lacproB)xlll nalA, rif, ara, argE~. which had been transformed with pJNlO:sup. Specificity assay: purification of DHFR: DHFR was isolated from XAC-1 carrying pJNlO:sup and plasmid pDAYQ. pDAYQ encodes the gene conferring resistance to chloramphenicol as well as the gene encoding DHFR,fol, with an amber mutation at residue ten (Normanly and Abelson, unpublished). The fol gene is under the control of the tac promoter (8). Cells were cultured in L broth (lOg/1 bactotryptone, 10g/1 NaCl, 5g!l yeast extract) supplemented with 5ug/ml tetracycline and 30ug/ml chloramphenicol to an A 600 of 0.4-0.8, at which time they were diluted fifty-fold into minimal M9 Glycerol medium (7), which had been supplemented with 5ug!ml tetracycline, 30ug/ml chloramphenicol, and lmM IPTG. The culture was incubated, shaking, at 370C for 12 to 18 hr., and the cells harvested. DHFR was purified from the cells as described in (1). N-terminal sequence analysis of DHFR was canied out at either the USC Microchemical Core Laboratory or the Caltech Microchemical Facility. RESULTS Figure 1a shows the nucleotides that are highly conserved in serine-accepting tRNAs (or as in the case of basepair 3-70, highly correlated) and which are not found in tRNAleu5UAG· our starting tRNA. Figure 1b shows the original leucine to serine conversion, tRNAleu~serUAG (1), with the 12 changes indicated. We first examined the contribution of the mutations in the D loop. In tRNAleu~serUAG· we had shifted the position of the conserved nucleotides G 18 and G 19 , based upon the observation that the position of these nucleotides (relative to the D stem), which is highly conserved in serine tRNAs, differs from the position ofG 18 and G 19 in leucine tRNAs. We predicted that this structural feature may play a role in serine AAS recognition, and therefore, altered the leucine D loop to introduce a serine-like configuration (Figure 1b). In mutant LSM6 (Figure 2) the D loop has the original tRNAleu5 sequence and configuration, while the mutations in 63 the D stem and acceptor stem are retained. This suppressor inserted exclusively serine at high efficiency (Table 1). We next examined the contribution of the mutations in the D stem (basepair 11-24) by generating a mutant that contained mutations in the acceptor stem only, and none in the D loop or stem. This variant, LSM4 (Figure 3), while an efficient suppressor, inserted a mixture of amino acids: serine (16% ), glutamine (39%) and leucine (38% ). Clearly, basepair C 11 -024 plays a role in serine acceptance. In all serine tRNAs base pair 11-24 is a C-0 and in tRNALeu5UAG is aU-A. We wondered if aU-G basepair would be acceptable at this position, thereby pinpointing 0 24 as the identity element Mutant LSMll (Figure 4) contained a U-0 at position 11-24, along with the original mutations in the acceptor stem This variant suppressed the ~-lactamase amber sufficiently to enable growth in the presence of ampicillin; however, the specificity assay revealed that LSM11 inserts predominantly leucine (62%), glutamine (33%) and only 1% serine. From these data we established that alterations to the D loop were unnecessary, and from the standpoint of suppression efficiency, most likely deleterious. Base pair C 11-024, however, plays a key role in serine AAS acceptance. We next set out to determine which of the original changes in the acceptor stem were actually necessary by reverting one base pair or nucleotide at a time, retaining the remainder of the acceptor stem changes as well as the essential C-0 pair at position 11-24. In mutant LSMlO, position 73 has been changed back to an A, while the other acceptor stem mutations, and the mutations at basepair 11-24 have been retained (Figure 5). This tRNA variant inserts exclusively leucine. In mutant LSM12 (Figure 6) basepair 2-71 has the original leucine sequence, C-0, and in mutant LSM13 (Figure 7) position 72 has the leucine sequence U. Both of these mutants insert predominantly leucine and glutamine (Table 1). In mutant LSM9 (Figure 8) the A-U at 3-70 has been changed to the C-0 found in tRNALeu5UAG· This tRNA inserts predominantly leucine (72% ), some serine (20%) and glutamine (6%), indicating that A 3-U70 is necessary for complete serine identity. The A-U pair at position 3-70 is not highly conserved in serine tRNAs, some have aU-A at this position. A mutant, LSM14, with aU-A at position 3-70, along with the other acceptor and D stem mutations (Figure 9) inserts predominantly leucine (65%) along with glutamine (19%) and serine (11 %). It would appear that aU-A pair at position 3-70 results in a loss of serine identity. This is odd considering that some serine tRNAs have a U3-A7o pair. It 64 is possible that an A 3 -U70 pair plays less a role in serine identity but rather acts as a deterrent to the leucine or glutamine AASs. From these reversion studies it is apparent that all of the original changes made in the acceptor stem were necessary to convert a leucine-accepting tRNA into a serine-accepting tRNA. Including basepair 11-24, the minimum number of changes to effect the leucine to serine conversion is eight, illustrated in Figure 2. This tRNA (LSM6) is also a very efficient suppressor, indicating biological function. It is possible that in this initial experiment, some of the serine identity elements would not have been discovered if, for example, they happened to be present in the tRNALeu5UAG that we used as a starting tRNA. To further define the elements of serine identity, we wished to convert another tRNA to serine acceptance. Therefore, we attempted to convert a tyrosine tRNA to a serine tRNA. We chose a tyrosine tRNA because in E. coli this species of tRNA, like the leucine and serine tRNAs, belongs to the type II class. Type II tRNAs are distinguished from type I tRNAs (in E. coli type I tRNAs constitute the remainder of tRNA species) primarily in the size of their extra loop, which is larger, 9 to 21 nucleotides, as opposed to 4 to 5 nucleotides for type I tRNAs (9). As the function of the extra arm is not understood, we elected to remain within the type II class for our identity swap experiments. Figure 10 shows the nucleotides highly conserved in serine tRNAs but not found in the su+3 allele of tRNATyr (10). The leucine to serine identity swap experiments revealed that alterations in the D loop were unnecessary for generating a serine accepting tRNA; furthermore, it was not necessary to make alterations in the anticodon stem of tRNALeu5UAG to obtain serine identity. We therefore chose to make only two changes to the tyrosine tRNA, those at postion 9 and 73. This mutant, Tyr~er1 (Figure 11), enabled growth in the presence of ampicillin, yet inserted predominantly glutamine (71% ), tyrosine (26%) and only 1% serine, as determined by the specificity assay (Table 1). While we may have introduced a serine determinant, we also either introduced a glutamine identity element or removed a deterrent. Also, Tyr-+Serl retained tyrosine identity elements. This result did not come as a complete suprise to us, as a tyrosine amber suppressor (su+3) with the same alteration at position 73 as in the Tyr~er1 mutant, inserts glutamine (11-13). We made two other attempts at a tyrosine to serine identity swap; Tyr~er2 (Figure 12) has alterations at all seven of the positions highly conserved in serine that are not found in tyrosine. Tyr~er3 (Figure 13) has changes at positions 9, 73, and basepair 3-70. Neither 65 mutant inserts more than 1% serine, both insert predominantly glutamine and some tyrosine (Table 1). DISCUSSION Previously, we reported the conversion of a leucine-inserting tRNA into a serine-inserting tRNA, brought about by making 12 base changes in the leucine tRNA (1). We have now determined which of those 12 changes were necessary, i.e., the minium number of changes required to effect the leucine to serine identity conversion. Only eight base changes are necessary, those at positions 11-24 in the D stem, 2-71, 3-70, 72, and 73 in the acceptor stem. In making the original tRNAl.eu-+SerUAG, we shifted the position of 0 18 and 0 19 in the leucine tRNA to conform to the configuration of a serine D loop. We found that not only was this alteration unnecessary for serine AAS recognition, it had a deleterious effect upon the suppression efficiency of the tRNA. We have also found that, while serine tRNAs may have either an A-U or U-A at position 3-70, in the leucine to serine identity conversion, an A 3 -U70 pair is essential, while a U 3 -A70 pair results in charging by the glutamine and leucine AASs. Possibly, the A3 -U70 acts less as an identity element for serine AAS, but in the context of the tRNAl.eu-+SerUAG sequence, acts as a deterrent to the leucine and glutamine AASs. It would be interesting to see what effect an A3 -U70 pair would have upon the identity of a serine tRNA which normally has aU-A at position 3-70. Identity swap experiments have been used to defme the identity elements for E. coli tRNAAla (14-16). A 0-U pair at position 3-70 is unique to alanine tRNAs, and all alanine tRNAs that have been sequenced to date have this 0 3 -U70 pair. Superimposing this 0 3 -U70 pair onto an amber alleles of E. coli tRNACys or tRNALYs, results in complete alanine identity (14, 16). When the 0 3 -U70 is superimposed upon a phenylalanine amber suppressor tRNA, the tRNA inserts predominantly alanine into amber codons, however it inserts some phenylalanine as well (14, 15). Yet when the 0 3 -U70 pair is superimposed upon a glycine amber suppressor tRNA, there is no conversion to alanine identity (15). These experiments highlight one of the problems encountered when attempting to define tRNA identity, and help to explain our failure to effect a complete conversion from tyrosine to serine identity. Merely superimposing predicted identity elements onto a tRNA is not 66 sufficient to effect a complete conversion of identity. One has to contend with the positive or negative recognition elements for all other AAS, which also reside in the tRNA being studied. While we were able to introduce serine identity elements into the tyrosine tRNA sequence (all the mutants; Tyr~Ser1 Tyr~Ser2 and Tyr~Ser3 were able to adequately suppress the serine requiring amber mutation in the amp gene), tyrosine identity elements remain in the tRNA. Additionally, we either introduced glutamine identity elements or removed glutamine AAS deterrents. As stated above, the introduction of a G at position 73 of a tyrosine amber suppressor (su+-3, the same allele used in our experiments) results in glutamine mischarging. Yet, in a leucine amber suppressor, an A to G change at position 73 does not result in glutamine mischarging. In fact, the phenomenon of glutamine mischargine is not new. A number of amber suppressor tRNAs are mischarged by the glutamine AAS. There must be a context effect; that is, there must be other glutamine determinants or a lack of glutamine deterrents present in those tRNAs that are mischarged by the glutamine AAS, while the opposite must be true for tRNAs that are not mischarged. Sequence comparison between tRNAs that are, or are not, mischarged by the glutamine AAS does not reveal any uniquely conserved nucleotides. In summary, identity swap experiments have proven instructive about the nature of serine tRNA identity, however, we will not be able to fully refme our model until we have a better understanding of the positive and negative identity elements of other tRNAs, glutamine tRNAs in particular. REFERENCES 1. Normanly, J., Ogden, R. C., Horvath, S. J., Abelson, J. 1986. Nature 321:213-19 2. Dalbadie-McFarland, G., et al. 1982. Proc. Nat/. Acad. Sci. U.SA. 79:6409-13 3. Bolin, J. T., Filman, D. J., Matthews, D. A., Hamlin, R. C, Kraut, J. 1982. J. Bioi. 4. Chern. 257:13650-62 Filman, F. J., Bolin, J. T., Matthews, D. A., Kraut, J. 1982. J. Bioi. Chern. 5. 257:13663-72 Bolivar, F., Rodriguez, R. L., Greene, P. J., Betlach M. C ., Heynecker, H. L., 6. Boyer, H. N.l977. Gene 2:95-113 Dente, L., Cesareni, G., Cortese, R.l983. Nucleic Acids Res. 11:1645-55 67 7. Miller, J. H.l972. in Experiments in Molecular Genetics. Cold Spring Harbor 8. Laboratory Press, Cold Spring Harbor, New York Amann, E., Brosius, J., Patashne, M . l983. Gene 25:167-78 9. 10. Brennan, T. , Sundaralingam, M. 1976. Nucleic Acids Res. 3:3235-51 Goodman, H. M., Abelson, J., Landy, A., Brenner, S., Smith, J. D. 1968. Nature 11. 217:1019-24 Shimura, Y., Aono, H., Ozeki, H., Sarabhai, A., Lamfrom, H., Abelson, J. 1972. Febs. Lett. 22:144-8 12. Hooper, M. L., Russel, R. L., Smith, J. 0.1972. Febs. Lett. 22:149-55 13. Ghysen, A., Celis, J. E. l974. J. Mol. Bioi. 83:333-51 14. Hou, Y. -M., Schimmel, P. R.l988. Nature 333:140-~ 15. McClain, W. H., Foss, K.l988. Science 240:793-6 16. McClain, W. H., Foss, K.l988. Science in press 68 Table 1. tRNAuAG Suppression Efficiency Growth on amp (given in % wild type ~-gal) Specificity (given in % amino acid inserted at DHFRam1o) Leu Ser 52-59 34-52 Leu~Ser 0.5-1 33-49 35-48 11-19 20-35 5-9 12 11-12 LSM6 LSM4 LSM11 LSMlO LSM12 LSM13 LSM9 LSM14 Tyr 14-16 18-25 + + + + + + + + Leu (99) Ser (92) Ser (90),Leu(10) Ser (92) Leu (38),Gln (39),Ser (16) Leu (62),Gln (33),Ser (<1) Leu (99) Leu (91),Gln (9) Leu (15), Gin (78),Ser (<1) Leu (72),Gln (6),Ser (20) Leu (65),Gln (19),Ser (11) Tyr (95) Tyr~Ser 1 6-8 + Gln (71),Tyr (26),Ser (1) Tyr~Ser 2 35-44 + Gln (71),Tyr (29),Ser(<1) Tyr~Ser 3 21-23 + Gln (82),Tyr (13), Ser (1) Values for percent amino acid inserted may not total100%, because background levels of less than 5% for any amino acid are not included with the exception of serine. While the values given for % serine inserted is below 5% in some cases, the ability to confer ampicillin resistance indicates that the mutant must be inserting some low level of serine. Estimates for% Gln inserted include Glu. amp; ampicillin 69 FIGURE LEGENDS Figure 1a. Composite of E. coli serine tRNAs. Nucleotides highly conserved in or correlated with serine tRNAs and not found in tRNAl.eu5UAG are indicated Adapted from (1). Figure 1b. tRNALeu5UAG with twelve base changes predicted to convert the tRNA to serine identity. Adapted from (1). Figure 2. Mutant I.SM6 Figure 3. Mutant I.SM4 Figure 4. Mutant LSM11 Figure 5. Mutant I.SM10 Figure 6. Mutant LSM12 Mutant LSM13 Figure 7. Figure 8. Figure 9. Figure 10. Mutant I.SM9 Mutant LSM14 Composite of E. coli serine tRNAs. Nucleotides highly conserved in serine tRNAs and not found in tRNATyr (su+3) are indicated. Figure 11. Mutant Tyr~Ser 1 Figure 12. Mutant Tyr~ser 2 Figure 13. Mutant Tyr~Ser 3 70 • • •G 73 • - c 72 G - C 11 AU -UA7o 2 3 .• .• .• .• - - • G • • II • •C• 15 I • D • A • 20-1 21 I I •G• 24 • 59 ••••• • A • I I I 1 I • ••••• • • • C48 •• I I •. 21C G43 • • • • • •- • • • 32C - • • • • • • Figure la • e I I e I • • • • •• 71 Figure lb 72 LSM6 Figure 2 73 LSM4 Figure 3 74 LSM11 A c c A-.G G-U -.c G.- C-G -.c A._ C-G -.u C-G G-C G-C A-UyyyyyUGA Gc GGG uAAG 11 Gu C GUG u UU G I I I G CAC / U G U A G A ~ AA_fic;~:>Cct G G-C GC~/ G G-C uuC G-C J--UA ucul' Figure 4 t 75 LSM10 A c c A 73 G-U ----.c G~ C-G ----.c A~ C-G ----. U C-G G-C G-C C A-UC<f<(<(<(UGA t uAAG Gu bCGGGuut C GUG U I I I G G G CAC ~ U u A GA AA-uGr/Sct; G G-C G G-C uuC G-C tf-UA ! UC'cf ucu~ Figure 5 76 LSM12 Figure 6 77 LSM13 Figure 7 78 LSM9 Figure 8 79 LSM14 Figure 9 80 • • • ·-· ·-· ·-· ·-· ·-· • ·-· G AU-AU •• •• G • ••• I I I • •u ••• • • • • • • • • •• •• I I I I I • • • • • •• •• •• C-G. • • •• •• •• •• •• •• • • • • •• • • ••• ·-· ·-· ·-· ·-· Figure 10 81 Tyr • Ser I Figure 11 82 Tyr • Ser2 Figure 12 83 Tyr ~ Ser3 Figure 13 84 CHAPTER Ill Construction of two Eschericia coli amber suppressor genes: tRNAPheCUA and tRNACysCUA Jennifer Normanlyl. Jean-Michel Masson2, Lynn G. Kleina2, John Abelsonl, and Jeffrey H. Miller2 I Division of Biology, California Institute of Technology, Pasadena, CA 91125; and 2Department of Biology, University of California, Los Angeles, CA 90024 [Proc. Natl. Acad. Sci. USA (1986) 83:6548-6552] 85 ABSTRACT Amber suppressor genes corresponding to Escherichia coli tRNAPhe and tRNACys have been constructed for use in amino acid substitution studies as well as protein engineering. The genes for either tRNAPheGAA or tRNACysGCA both with the anticodons 5' CfA 3' were assembled from four to six oligonucleotides, which were annealed and ligated into a vector. The suppressor genes are expressed constitutively from a synthetic promoter, derived from the promoter sequence of the E. coli lipoprotein gene. The tRNAPhe suppressor (tRNAPhecuA) is 54-100% efficient in vivo, while the tRNACys suppressor (tRNACyscuA) is 17-50% efficient. To verify that the suppressors insen the predicted amino acids, both genes were used to suppress an amber mutation in a protein coding sequence. N-terminal sequence analysis of the resultant proteins revealed that tRNAPhecuA and tRNACysCUA insert phenylalanine and cysteine, respectively. To demonstrate the potential of these suppressors, tRNAPheCUA and tRNACyscuA have been used to effect amino acid substitutions at specific sites in the E. coli lac repressor. INTRODUCTION Nonsense suppressor are alleles of tRNA genes altered in the anticodon, resulting in insertion of an amino acid in response to a termination codon. These suppressors have been important tools in bacterial genetics (1) as well as in the study of recognition of tRNA by aminoacyl tRNA synthetase (2). The use of nonsense suppressors also offers exciting possibilities for protein engineering, because numerous altered proteins with known amino acid changes can be generated easily. By this method, genes containing amber, ochre, or opal mutations (resulting in UAG, UAA, or UGA chain-terminating codons, respectively) are expressed in mutant strains that produce suppressor tRNAs capable of inserting an amino acid in response to the nonsense mutation. Many single amino acid substitutions at specified positions in proteins have already been made. For example, nonsense suppression has been used to create close to 400 mutant lac repressor proteins, each carrying a known amino acid replacement (3). Unfortunately, the range of amino acid exchanges made possible by nonsense suppression has been limited by the relatively small set of nonsense suppressors available. Until recently, only five different amino acids could 86 be inserted at an amber codon by suppressor tRNAs. The amber suppressor tRNAs that have been generated in vivo (supD, supE, supF, and supP), insert serine, glutamine, tyrosine, and leucine, respectively, in response to UAG codons. In addition, Murgola and coworkers have described a glycine-inserting amber suppressor derived from a series of in vivo genetic manipulations (4). [Ochre suppressors operating at reduced efficiencies can insert additional amino acids--for instance, supG directs the insertion of lysine at UAA and UAG codons (5), but the inefficient suppression greatly limits their utility for generating altered proteins.] With the exception of the glycine suppressor described by Murgola, the existing amber suppressor alleles arose from a single base change, resulting in the anticodon 5' VAG 3' (6-10). Creating new suppressor alleles by standard genetic means is unlikely, as virtually every other tRNA requires more than a single base change to yield a CUA anticodon. However, advances in oligonucleotide synthesis (11) have made possible the in vitro construction of new suppressor tRNAs, which permits us to greatly extend the utility of this approach. To expand the available collection of amber suppressors in E. coli, we have synthesized the genes for several tRNAs, altering the anticodon to enable recognition of UAG (amber) codons. Here we describe the construction of the genes for two such suppressor tRNAs, which insert phenylalanine and cysteine at high efficiency. and we demonstrate their utility in amino acid substitution studies of the E . coli lac repressor. METHODS Bacteria, Bacteriophage, Media, and Reagents. E. coli strain XAC-1 is F' laci373 lacZuJJBamProB+ I p- ~(lacproB)xJJJ nalA, rif, argEam, ara. The Fl phage, IR l was obtained from E. Meyerowitz. Indicator plates used for transformations were minimal M9 glucose (12) supplemented with ampicillin (lOOug/ml) and the indicator dye 5-bromo-4-chloro-3-indolyl ~-D-galactopyranoside (40ug/ml). Folic acid and methotrexate were purchased from Calbiochem and ICN, respectively. Oligonucleotide Synthesis and Purification. Oligonucleotides were synthesized on an Applied Biosystems 380A DNA synthesizer. Three to five A260 units of oligonucleotide were purified by electrophoresis through a 20% acrylamidenM urea gel, 87 visualized by UV shadowing, excised, and eluted in 0.3 M NaCl, 10mM Tris-HCl, pH 7.4, 1mM EDTA, 1% phenol, at 370C for 18 hr. The DNA was precipitated and washed with ethanol, dried, and resuspended in distilled H 20. Gene Synthesis. Each oligonucleotide (2ug) was phosphorylated in 70mM Tris-HCl, pH 7.6, lOmM MgC12, 5mM dithiothreitol (DTT), 100uM ATP, with 4 units of polynucleotide kinase for 1 hr at 370C. The phosphorylated oligonucleotides (80 pmol each) were mixed together in 100mM NaCl, heated to 800C for 5 min, then allowed to cool to room temperature over a period of 3 hr. The entire mixture was combined with vector pGFIB-1 (previously cleaved with Eco RI and Pst I) to give' a· mass ratio of 10: 1 (insert:vector). Subsequent ligation was carried out in 50mM Tris-HCl, pH 7.6, 10mM MgC12, 20 mM (DTT), 50mM NaCl, 1mM A TP, bovine serum albumin (50ug/ml), with 1 unit of T4 DNA ligase at 150C for 12 hr. The ligation mixture was used to transform competent E. coli XAC-1 cells. From purified colonies exhibiting the suppressor phenotype on indicator medium, single-stranded DNA was made by using the F1 phage IR1 as helper (13) and sequenced by the chain-termination method of Sanger et al. (14). Protein Purification. Purification of mutant dihydrofolate reductase (DHFR) protein was carried out by a combination of the methods ofBaccanari et al. (15-17) and is briefly described here. E . coli XAC-1 carrying plasmid pDA3-12:Phe was cultured in 61iters of minimal M9 glucose medium supplemented with ampicillin (50ug/ml) at 37oc with aeration for 18 hr. Cells were harvested and lysed (15) and the crude lysate was extracted with ammonium sulfate to 90% saturation (16). The precipitate was brought to a protein concentration of 20 mg/ml, 3 m1 of methotrexate Sepharose was added, and the mixture was dialyzed for 18 hr (16). The resin was batch-washed (17) and the DHFR was eluted with folic acid (17). At this point, the DHFR was 99% pure as judged by silver staining (18) of sodium dodecyl sulfate/polyacrylamide gels (19). To remove folate and nucleic acid, peak fractions were first dialyzed against 50 mM potassium phosphate, pH 8.0, 1 mM DDT, and then applied to a DEAE-Sephacel (Pharmacia) column (1 mg/ml protein) equilibrated with the same buffer. DHFR was eluted in a linear gradient of 0-0.4M KCl in 50mM potassium phosphate, pH 8.0, 1mM DTT. Peak fractions were pooled, dialyzed against water, and lyophilized. 88 Protein Sequencing. Mutant DHFR protein was sequenced on an Applied Biosystems Model470A gas-phase protein sequenator (20) by the USC Microchemical Core Laboratory. Mutant ~-galactosidase was purified (21) and sequenced by A. Fowler (University of Los Angeles). RESULTS Gene Construction. The sequences of E. coli tRNAPheaAA and tRNACysGCA with the alterations necessary to convert them to amber suppressor tRNAs are shown in Figure 1. The synthetic genes for tRNAPheCUA and tRNACysCUA are depicted in Figure 2. Each gene is constructed from a set of 4-6 complementary oligonucleotides ranging in size from 23 to 46 mucleotides, with 7 -base pair (bp) overlaps at the junctions. Each tRNA gene sequence is flanked at the 5' and 3' ends by Eco RI and Pst I sites, respectively. Since the E. coli cellular nucleases that process tRNA precursors recognize conserved structural elements within the tRNA and not surrounding sequences (25), we expected that these constructs would produce fully processed functional tRNAs. The oligonucleotides were annealed and ligated into the Eco RI-Pst I sites of plasmid pGFIB-1 as described in the Methods section. In this vector (Figure 3), the tRNA gene is expressed constitutively from a strong synthetic promoter, based on the promoter sequence of the E. coli lipoprotein gene lpp (26). Distal to the promoter and restriction site polylinker is a transcription terminator, which we have synthesized based on the termination sequence of the ribosomal RNA operon rrnC (27-28). Plasmid pGFIB-I also carries the F1 intergenic region, which in the presence of F1 helper phage allows the production of single-stranded DNA (13), thus facilitating dideoxy sequencing of the synthetic gene. To clone active suppressors, the ligation mixtures pGffi:Phe and pGFffi:Cys were transformed into E. coli stain XAC-1. This strain carries an F' lac proB episome with an amber mutation early in the lacZ gene. In an sup- strain, virtually no active ~-galactosidase (the lacZ gene product) is formed. Suppression of the amber mutation should yield functional ~-galactosidase, resulting in blue colonies on medium containing the indicator 5-bromo-4-chloro-3-indolyl ~-D-galactopyranoside. When XAC-1 was transformed with pGFIB:Phe or pGFIB:Cys, on average, 60% of the transformants were blue. To confmn 89 that this phenotype was plasmid encoded and not due to spontaneous reversion of the amber mutation, plasmid DNA was isolated from transformants that gave blue colonies and sequenced to ensure that there were no sequence anomalies in the synthetic construct. Determination of Suppression Efficiency. We have exploited the properties of a lac/-Z fusion system, to measure the efficiency of suppression. In this system, the lac/ and Z genes are fused, resulting in a hybrid protein with full ~-galactosidase activity (29). The N-terminal, /-encoded portion of the fusion is not required for ~-galactosidase activity, although continued transcription and translation through this portion of the fusion is necessary. Thus, the introduction of nonsense mutations that interrupt translation of the I portion of the message results in negligible ~-galactosidase synthesis, while suppression of these nonsense mutations restores ~-galactosidase synthesis. As fi-galactosidase activity is independent of the nature of the amino acid being insetted into the /-encoded portion of the hybrid protein and dependent only on the degree of transmission through the translation block, the level of ~-galactosidase activity is a direct measure of the efficiency of suppression. Previous studies of suppression at different nonsense sites have shown that the efficiency of suppression is dependent on the sequence immediately surrounding the nonsense site; the two bases 3' to the nonsense site are especially important (29, 30). Generally, amber codons followed by an A or G are suppressed more efficiently than those followed by a C or U, although codons followed by the sequence CU are well-suppressed. We have examined tRNAPhecuAand tRNACysCUA against a set of ambers in the lacl-Z fusion by assaying ~-galactosidase activity of the suppressed fusion protein (12). Table I shows the relative efficiency of the new suppressors as well as those for other suppressors. tRNAPheCUA has a suppression efficiency of 54-100%, while the efficiency of tRNACyscuA ranges from 17% to 50%, depending upon the context of the amber codon. Determination of Suppressor Specificity. A single change in the anticodon of E. coli tRNATrpresults in mischarging by the glutamine aminoacyl tRNA synthetase (31). Similarly, alteration of the anticodon of E. coli tRNAMet(f) results in mischarging by the glutamine synthetase (32-34). Given these precedents, we designed an assay to show directly that the new suppressors actually insen the predicted amino acid. tRNAPheCUA and 90 tRNACysCUA were each used to suppress an amber mutation in a protein-coding sequence. The resulting altered proteins were subsequently isolated and subjected toN-terminal sequence analysis to determine which amino acid(s) had been inserted in response to the amber codon. For the determination of the specificity of tRNAPhecuA, we constructed an amber mutation via oligonucleotide-directed mutagenesis, in the lOth residue (valine) of the E. coli fol gene encoding DHFR (35) (Figure 4). DHFR was chosen because it can be purified in virtually one step by methotrexate affmity chromatography. Furthermore, the structure of DHFR has been well-delineated by X-ray crystallography (39-41). The tenth residue of this protein is not highly conserved throughout phyla and is not in close proximity to the active site. Hence, we predicted that substitutions at this site would not significantly alter the protein's activity or its behavior during purification. We also converted the 11th residue from Asp to Asn. The purpose of this GAT to AAT conversion was to provide a better context for the amber mutation. Subsequently, we constructed a plasmid (pDA3-12:Phe) that carries both thefol amber gene under the control of the tac promoter (38) and tRNAPheCUA flanked by the lipoprotein promoter and rrnC terminator (Figure 4). DHFR was isolated from XAC-1 cells harboring this plasmid and subjected to Edman degradation on an automated protein sequenator (20). In Figure 5, the picomolar yields of phenylthiohydantoin-derivatized phenylalanine, valine, asparagine, and aspartic acid are plotted versus the residue number. It should be noted that the wild-type copy of the fol gene is also present in the host chromosome. Under control of the rae promoter, the plasmid-borne/of gene overproduces DHFR approximately 100-fold above the level produced from the chromosomal copy. This should result in minimal levels of Vas-Asp at positions 10 and 11 (compared with Phe-Asn) in the presence of a strong suppressor such as tRNAPheCUA· The major amino acid at position 10 is indeed phenylalanine, and no other amino acids are detected at a significant level, including valine or glutamine. In other studies, we have been able to detect 10% insertion of an amino acid at residue 10 (28). Therefore, to our limits of detection, phenylalanine is the sole amino acid inserted by the new suppressor. Residue 11 is predominantly asparagine, with a trace amount of aspartic acid. This may be the result of deamidation of asparagine or due to wild-type background. The specificity of tRNACysCUA was determined by A. Fowler in a similar fashion (personal communication). The synthetic tRNACysCUA gene was used to suppress an 91 amber mutation located at residue 17 in the lacZ gene. The resulting ~-galactosidase was purified and subjected toN-terminal sequence analysis. Cysteine is the major amino acid inserted at this site, and again glutamine is not detected at a significant level (data not shown). Substitutions in the lac Repressor. To demonstrate the potential of the new suppressors for structure function studies, we have used them to generate specific amino acid replacements in the E. coli lac repressor. The lac repressor is especially suited to amino acid substitution studies as its properties are well-understood, its activity is readily assayed, and a large number of amber mutations in the repressor gene, lacl, have been generated(42). We have previously described the use of suppressed nonsense mutations to generate altered lac repressor molecules (3). Close to 90 different nonsense sites in the lac! gene have been described, and the ability to add different amino acids by suppression offers the opportunity to study a heirarchy of amino acid substitutions at each site. Table IT shows several positions in the repressor that are particularly sensitive to certain amino acid substitutions. These data can yield useful information regarding the nature of the amino acid being inserted by the new suppressors. For instance, glutamine is the wild-type residue at positions 18 and 248. From Table 2 it is evident that these sites specifically require glutamine among the amino acids examined so far. Exchanges at position 18 affect operator binding, and substitutions at position 248 affect inducer binding (the 15 phenotype). These two sites can be used to determine whether a suppressor is inserting glutamine. Also, the phenylalanine at position 293 is not easily replaced without affecting the activity of the protein. Note that the phenylalanine-inserting suppressor restores the wild-type character, reinforcing the protein chemistry results for this suppressor. In addition, the tyrosines at position 17, 47, and 282, and the tryptophan at position 201 are sensitive to substitution. Only certain hydrophobic amino acids are allowed at positions 201 and 282, and the phenylalanine-inserting suppressor does restore activity in both cases. None of the previous substitutions for tyrosine at position 47 could restore even partial activity. However, as can be seen from Table 2, the phenylalanine-inserting suppressor can substitute reasonably well. Possibly, the ring moiety of the tyrosine is crucial at this position, and the phenylalanine fulfills a similar function. We note that phenylalanine cannot substitute for tyrosine at position 17, a residue that has been implicated in operator binding mediated by hydrogen bonding (43). This would be 92 expected, because the phenylalanine ring lacks the hydroxyl side chain of tyrosine that could participate in hydrogen bonding. DISCUSSION Nonsense suppressors can be generated by altering the anticodon of a tRNA to allow recognition of chain-termination codons. This approach has already met with success in several cases via oligonucleotide-directed mutagenesis (44-47). We have described the construction of two amber suppressor tRNA genes, tRNAPhecuA and tRNACysCUA• which insert the predicted amino acid at a high level of efficiency. By using the approach of total gene synthesis, we have been able to construct these suppressor genes very rapidly. Furthermore, the manner in which the genes are designed, i.e., without their natural 5' and 3' flanking sequences, does not preclude the synthesis of a functional tRNA. We have subsequently assembled the genes for several other suppressor tRNAs in an effort to create as large a "bank" of amber suppressors as possible (unpublished results). One potential obstacle to creating a complete collection of amber suppressors is the possibility that the anticodon of certain tRNAs is crucial for recognition by its cognate aminoacyl tRNA synthetase. We therefore developed an assay that would enable us to determine whether the anticodon had affected the specificity of the tRNA in question, by showing directly which amino acid it inserts into an amber site of a protein coding sequence. From these protein chemistry studies, it is evident that the new suppressors insert only the original amino acids, phenylalanine and cysteine, at least to the limits of detection of the protein sequencing systems. Data from amino acid replacement studies in the lac repressor, utilizing tRNAPheCUA and tRNACyscuA. support this conclusion. To extend the sensitivity of detection beyond the limits of protein sequencing, genetic systems similar to the one featured in Table 2 can be developed further. ACKNOWLEDGEMENTS We thank Lynn Williams of the University of Southern California Microchemical Core Laboratory and Audree Fowler of the University of California at Los Angeles for protein sequencing. The manuscript was improved by comments from V. Bankaitis, M. Clark, E. Grayhack, A. Lustig, E. Phizicky, and S. Ruby. This work was supported in part by a 93 National Science Foundation grant (DMB-8417353) to J. A. and J. H. M., a National Research Service Award (T32GM07616) from the National Institute of General Medical Sciences to J. N., and a North Atlantic Treaty Organization fellowship and Philippe Foundation fellowship to J.-M. M. REFERENCES 1. Steege, D. A., Soli, D. 1979. in Biological Regulation and Development, (ed. Goldberger, R. F.) Plenum Press, New York Vol. I, pp. 433-85 2. Ozeki, H., Inokuchi, H., Yamao, F., Kodaira, M., S~o, H., Ikemura, T., Shimura, Y.1980. in Transfer RNA: Biological Aspects (eds. Soli, D., Abelson, J. N, Schimmel, P.R.) Cold Spring Harbor Laboratory Press, New York pp. 341-2 3. Miller, J. H., Coulondre, C., Hofer, M., Schmeissner, U., Sommer, H., Schmitz, 4. A., Lu, P.1979. J. Mol. Bioi. 131:191-222 Murgola, E. J., Prather, N. E., Pagel, F. T., Mims, B. H., Hijazi, K. A. 1984. 10. Mol. Gen. Genet. 193:76-81 Gorini, L.1970. Ann. Rev. Genet. 4:107-34 Steege, D. A.1983. Nucleic Acids Res. 11:3823-32 Yoshimura, M., Inokuchi, H., Ozeki, H.1984. J. Mol. Bioi. 177:627-44 Thorbjarnardottir, S., Dingermann, T., Rafnar, T., Andreson, 0. S., Soli, D., Eggertsson, G.l985 J. Bacterial. 161:219-22 lnokuchi, H., Yamao, F., Sakano, H., Ozeki, H.1979. J. Mol. Bioi. 132:649-62 Goodman, H. M., Abelson, J., Landy, A., Brenner, S., Smith, J. D. 1968. Nature 11. 12. 217:1019-24 Khorana, H. G. 1979. Science 203:614-25 Miller, J. H. 1972. in Experiments in Molecular Genetics. Cold Spring Harbor 13. 14. Laboratory Press, New York Dente, L., Cesareni, G. Cortese, R.l983. Nucleic Acids Res. 11:1645-55 Sanger, F., Nicklen, S., Coulson, A. R. 1977. Proc. Nat/. Acad. Sci. U.S A . 15. 74:5463-7 Baccanari, D., Phillips, A., Smith, S., Sinski, D., Burchall, J. 1975. Biochemistry 5. 6. 7. 8. 9. 14:5267-73 94 16. Baccanari, D.P., Averett, D., Briggs, C., Burcha11, J. 1977. Biochemistry 16:3566-72 17. Baccanari, D.P., Stone, D., Kuyper, L.1981. J.Biol. Chern. 256:1738-47 18. Wray, W., Boulik:as, T., Wray, V. P., Hancock, R. l981. Anal. Biochem. 118:197-203 19. Laemmli, U. K.l970. Nature 227:680-5 20. Hewick, R. M., Hunkapiller, M. W., Hood, L. E., Dreyer, W. J. 1981. J. Bioi. Chern. 256:7990-7 21. Fowler, A. V., Zabin, 1.1983. J. Bioi. Chern. 258:14354-8 22. Barrell, B. G., Sanger, F. 1969. Febs. Lett. 3:275-8 23. 24. 25. 26. Mazzara, G. P., McClain, W. H.1977. J. Mol. Bioi. 117:1061-79 Deutscher, M. P.1984. CRC Crit. Rev. Biochem. 17:45-71 Abelson, J.l979. Ann. Rev. Biochem. 48:1035-69 Nakamura, K., Inouye, M.l979. Cell18:1109-17 27. Young, R. A.l979. J . Bioi. Chern. 254:12725-31 28. Normanly, J., Ogden, R. C., Horvath, S. J., Abelson, J. l986. Nature 321 :213-19 29. Miller, J. H., Albertini, 1983. J. Mol. Bioi. 164:59-71 30. 31. 32. Bossi, L. 1983. J. Mol. Bioi. 164:73-87 Yaniv, M., Folk, W . R., Berg, P., Soil, L.l974. J. Mol. Bioi. 86:245-68 Schulman, L. H., Pelka, H., Susani, M.l983. Nucleic Acids Res. 11:1439-55 33. Schulman, L. H., Pelka, H. 1983. Proc. Nat/. Acad. Sci. U.S.A. 80:6755-9 34. 35. Schulman, L. H. 1972. Proc. Nat/. Acad. Sci. U.S.A. 69:3594-7 Smith, D. R., Calvo, J. M. 1980. Nucleic Acids Res. 8:2255-74 36. 37. Newman, A. J., Ogden, R. C., Abelson, J. A.l983. Cel/35:117-25 Villafranca, J. E., Howell, E. E., Voet, D. H., Strobel, M.S., Ogden, R. C., 38. 39. Abelson, J.N., Kraut, J. 1983 Science 222:782-8 Amann, E., Brosius, J., Ptashne, M. 1983. Gene 25:167-78 Bolin, J. T. , Filman, D. J., Matthews, D. A., Hamlin, R. C., Kraut, J. l982. J. 40. Bioi. Chern. 257:13650-62 Matthews, D. A., Alden, R. A., Bolin, J. T., Freer, S. T., Hamlin, R., Xuong, N., Kraut, J., Poe, M., Williams, M., Hoogsteen, K. 1977. Science 197:452-5 95 41. 42. 43. Filman, D. J., Bolin, J. T., Matthews, D. A., Kraut, J. 1982. J . Bioi. Chern. 256: 13663;72 Miller, J. H . 1978. in The Operon, (eds. Miller, J. H., Reznikoff, W. S.) Cold Spring Harbor Laboratory, New York pp. 31-88 Steitz, T. A., Wever, I. T., Matthews, J. B.l982. Cold Spring Harbor Symp . Quant. Bioi. 47:419-26 44. Capone, J. P., Sharp, P. A., RajBhandary, U. L. 1985. EMBO J. 4:213-21 45. Laski, F. A., Belagaje, R., RajBhandary, U. L., Sharp, P. A. 1982.Proc. Nat/. 46. 47. Acad. Sci. U.SA . 79:5813-17 Laski, F. A., Belagaje, R., Hudziak, R. M., Capecchi, M. R., Palese, P., RajBhandary, U. L., Sharp, P. A. 1984. EMBO J. 3:2445-52 Hudziak, R. M., Laski, F. A ., RajBhandary, U. L., Sharp, P. A., Capecchi, M. R., 1982. Ce//31:137-46 96 Table 1. Suppression of amber mutations in a lac/-Z fusion. ~-Galactosidase activity,% of wild type Suppressor 021c A30 028c A26 A16 017c 013c tRNAPheCUA 79 100 78 100 78 54 84 tRNACysCUA 34 53 50 51 35 17 35 supD supE supF 27 54 26 67 70 32 10 62 80 25 21 26 11 10 6 0.8 100 100 43 62 11 46 30 58 supP 15 34 72 11 All assays were carried out at 37°C and were determined in duplicate (12). Values are given as the percentage of the wild-type fusion in the respective suppressor strains. Values· for supD, supE, supF, and supP are taken from Miller and Albertini (29). 97 Table 2. Amino acid substitutions in the lac repressor. Replacement by suppression Wild-type residue Ser Gin Tyr Leu Phe Cys Tyr-17 Try-47 Tyr-282 Trp-210 Phe-293 Gln-18 Gln-248 +I- -I+ + + + + s -1+ -I+ s -I+ +I+ + + s s s +Is +I+ + + s -I+ s +,Normal activity; s, DNA binding activity retained but not inducibility;-, no activity, presumably because of defects in folding, aggregation or DNA binding. Partial activities: +/-indicates a less severe defect that-/+. 98 FIGURE LEGENDS Figure 1. RNA sequences of E. coli tRNAPheGAA (22) and tRNACysGCA (23) with the base changes that would enable the tRNAs to recognize amber codons. Figure 2. Synthetic genes for tRNAPheCUA and tRNACysCUA· The first and last nucleotides encoding each tRNA (indicated by a dot) are immediately flanked by the cohesive ends of Eco RI and Pst I restriction endonuclease sites, respectively. Arrows indicate the oligonucleotide junctions, and the altered anticodon is underlined. Note that the tRNAPheCUA gene does not encode the 3'-tenninal C and A residues. These residues are added in vivo by tRNA nucleotidyltransferase (24). Figure 3. Plasmid pGFIB-1. This vector is a derivative ofpEMBL8+ (13). We have replaced the lac promoter with a synthetically constructed promoter (48) based upon the promoter sequence of the E. coli lipoprotein gene lpp (26). Two complementary oligonucleotides encoding a transcription termination sequence from the E. coli "nC operon (27) have been ligated into the Pst I and Hind ill site of the polylinker (28). The oligonucleotides encoding the tRNA genes are ligated into the Eco RI and Pst I sites of the polylinker. Figure 4. Construction of plasmid pDa3-12. Oligonucleotide-directed mutagenesis (36) was carried out on the E. colifol gene (35) encoded by a 1-k:ilobase fragment contained in the vector M13mp8 (37). Plasmid pDa3-12:sup tRNA, in this case pDa3-12:Phe, was constructed by a series of subcloning steps. The 247-bp Eco RI-Hind ill fragment of pTAC12H (38) containing the tac promoter was ligated into the Eco RI-Hind ill sites of pBR322 to create pTAClO. Subsequently, the DHFR amber encoded by a 1057-bp Hind ill fragment, was subcloned into the Hind ill site ofpTAClO, which yielded pDa3-12. The 350 bp Pvu il fragment from pGFIB:Phe containing the lipoprotein promoter, tRNAPhecuA, and the "nC terminator was ligated into pDa3-12 to yield pDa3-12:Phe. 99 Figure 5. The picomolar yields ofphenylthiohydantoin-derivatized phenylalanine, valine, asparagine, and aspartic acid have been plotted versus the residue number for DHFR purified from E. coli XAC-1 cells carrying pDa3-12:Phe. •• ...... j' "T1 A • • •• • G UCCGGT'I'C G tRNAGcA-. CUA Cys 66 .. • • • • G•C ex CUUGGriJIC AG•CG G7 G• C U G•C Phe I tRNAGAA-.CUA I c u A5 A•o/ UG AAA A UG cAA D GAGe CUCG t- • 'II C G G A G• C · U G AGC CU UAG U G•C G•C A•l/1 'I' A5 AACA • •• AUGUAG U A G DO C•GCU C G GA G•C G•C C•G G•C C•G G•C ~ • A A GGC CU CA D GA A u G•C C•G C•G C•G G•C A c c A c c 8 "Tl ..... N ~ (Jq • I I . EcoRI 5' - I I I GCCGCGCAATTGTTTCGCCAATACATCGCCTAAGATTTAGGCAGATCAGGCCAAGCTGAGGCCTTGCGCGGAGGTG I ~ . Pst I 5• AATTCGGCGCGTTAACAAAGCGGTTATGTAGCGGATTCTAAATCCGTCTAGTCCGGTTCGACTCCGGAACGCGCCTCCACTGCA • GCGGGCCTATCGAGTCAGCCATCTCGTCCCCTAAGATTTAGGGGCACAGGAACCAAGCTAAGGCTCAGGCCCGTG AATTCGCCCGGATAGCTCAGTCGGTAGAGCAGGGGATTCTAAATCCCCGTGTCCTTGGTTCGATTC CGAGTCCGGGCACTGCA CUA tRNACys EcoRI 5' CUA tRNAPhe Pst I 0"""" """" 102 Pstl pGFIB Figure 3 103 DHFR amber 2 DHFR wt: 12 II e Ser Leu lie Ala Ala Leu Ala Val Asp Arg ATC AGT CTG ATT GCG GCG TTA GCG GTA GAT CGC DHFR am: ... • • • • •• • •• ••• • • • am Asn ••• .. . . .. • • • • • • •• • • . . . • •• TAG AAT ••• ••• • • Ptac pDa3-12 SuptRNA / tet Pvu li terminator sup tRNA gene Figure 4 104 DHFRarn- Phenylalanine Suppressor MET ILE SER LEU ILE ALA ALA LEU ALA PHE ASN ARG VAL ILE PHE Q) 0 E ~ 10 5 180r-------------------------------------~--~ VAL ~ 0 E ~ 10 5 180~----------------------------------------~ ASN II Q) 0 E ~ 100 Q) 0 E ~ 10 5 ASP 75 50 II 25 10 5 Figure 5 105 CHAPTER IV Construction of Escherichia coli amber suppressor genes: Optimization of suppressor efficiency and determination of tRNA specificity 106 The work described herein was carried out in collaboration with Jeffrey H. Miller, Lynn G. Kleina, and Jean-Michel Masson; two other manuscripts detailing parts of this work are in preparation. ABSTRACT Using synthetic oligonucleotides, we have constructed a collection of E. coli amber suppressor tRNA genes (1, 2). These suppressors vary in efficiency, and we have been able to increase the suppression efficiency for some of the tRNAs via mutagenesis. In order to determine the specificity of these tRNAs, each was used to suppress an amber (UAG) nonsense mutation in the E. coli dihydrofolate reductase genefol. The mutant proteins were purified and subjected toN-terminal sequence analysis to determine which amino acid had been inserted by the suppressor tRNAs at the position of the amber codon. The suppressors can be classified into three groups, based upon the protein sequence information. Class I suppressors, tRNAA1a2cuA, tRNAGlyUCUA, tRNAHisACUA, tRNALysCUA· and tRNAProHCUA· inserted the predicted amino acid. The Class II suppressors, tRNAGluACUA, tRNAGlyTCUA· and tRNAllelCUA were either partially or predominantly mischarged by the glutamine aminoacyl tRNA synthetase (AAS). The Class III suppressors, tRNAArgCUA, tRNAAspMcuA, tRNAile2cuA, tRNAThr2cuA, tRNAMet(m)CUA and tRNAValCUA inserted predominantly lysine. INTRODUCTION We have set out to construct a complete collection of E. coli amber suppressor tRNA genes for use in studies of protein structure and function. We previously reported the construction of amber suppressor alleles of E. coli tRNAPhe and tRNACys from synthetic oligonucleotides (3). Subsequently, we have constructed amber suppressor alleles for an additionall5 tRNAs in a similar fashion (1, 2). These tRNAs, each with a CUA anticodon to enable recognition of amber codons, are constitutively expressed from a high copy number plasmid (4, 5). Table 1lists the suppression efficiencies of the newly generated tRNAs. For a number of the amber suppressor tRNAs the suppression efficiencies were so low that they would not be useful in amino acid substitution studies. We undertook to improve the suppression efficiencies of a number of the tRNAs by several means. 107 I. Optimization of suppressor efficiency Different lsoacceptors. We found that in some cases, the amber suppressor efficiencies varied among isoacceptors of a given tRNA species. This was the case for alanine, isoleucine, threonine and valine tRNAs. tRNAAlalcuA, tRNAilelcuA, tRNAThrlCUA• and tRNAVallcuA were all non-functional suppressors as evidenced by their very low suppression efficiencies. Simply generating an amber allele of a different isoacceptor for these tRNAs resulted in suppressor tRNAs which were significantly more efficient (Table 1). Subsequent to generating the Alal suppressor, we learned that the published sequence, which the tRNA gene was based on, was in question. The fact that the sequence is incorrect would help to explain why the Alal suppressor was inefficient. We do not know if the same holds for the other inefficient tRNAs. The Extended Anticodon Hypothesis. The extended anticodon hypothesis put forth by Y arus (6) states that "... the structures of the anticodon loop and the proximal anticodon stem are related to the sequence of the anticodon." Comparison of the anticodon stem and loop sequences from E. coli tRNAs reveals that they are related to the 3' nucleotide of the anticodon, referred to as the cardinal nucleotide. Put another way, for a given nucleotide at the 3' position of the anticodon, the sequence of the anticodon stem and loop (the extended anticodon) will have a predictable sequence. These fmdings are applicable to suppression efficiency in the following manner. To generate an amber suppressor allele, it is necessary to alter the anticodon to CUA, by one or more base changes. Y arus noted that for weak amber suppressors the mutations necessary to render an amber suppressor resided in the cardinal nucleotide, while strong amber suppressors did not have an alteration in the cardinal nucleotide. According to the extended anticodon hypothesis, when the cardinal nucleotide is altered as in the case of the weak suppressors, the proximal anticodon stem and loop sequences are no longer related to the cardinal nucleotide, resulting in decreased translational efficiency. To generate the aspartic acid suppressor, the cardinal nucleotide was altered from C to A, and the resultant suppressor was non-functional. Based upon the consensus sequences that were derived for each of the four possible cardinal nucleotides, the extended anticodon of tRNAAsPcuA suppressor deviated at five positions. Using oligonucleotide-directed mutagenesis, we altered those five nucleotides so that the extended anticodon would conform to the consenesus sequence for a cardinal nucleotide of "A." The resultant tRNA, tRNAAspMCUA· was tenfold more efficient than the original tRNAAsPcUA· 108 3' Double A. The extended anticodon hypothesis dictates that if the cardinal nucleotide is an A, the two 3' proximal nucleotides (37 and 38) should be A as well. Position 37 of the glutamic acid and histidine suppressors, both fairly weak, was changed to A, resulting in a two- to sixteen-fold increase in suppression efficiency (Table 1). It should be noted that in generating tRNAAla2cuA, the alternate alanine isoacceptor, position 37 was changed to an A as well. The Hybrid Molecule Approach. While the proline amber suppressor is non-functional (< 1% efficient), the phenylalanine amber suppressor is very strong (48-100% ). Speculating that the anticodon stem and loop do not play a role in proline AAS recognition, we generatd a tRNA variant (via gene synthesis) with the anticodon stem and loop region of tRNAPheCUA and the remaining regions of the tRNA from tRNAPro. This mutant, tRNAProHCUA had a suppression efficiency ranging from 8 to 60%. In summary, using the above approaches we were able to generate amber suppressor tRNAs for 14 species oftRNA that had reasonably high suppression efficiencies, from 6 to 100%, depending upon the context of the amber codon being suppressed (Table 1). We have not yet been able to generate an efficient suppressor allele for tRNAAsn. II. Determination of tRNA specificity The anticodon of a tRNA potentially plays a role in the recognition of the tRNA by its cognate aminoacyl tRNA synthetase (AAS). For example, altering the anticodon of the tryptophan or formyl-methionine tRNAs from E. coli, results in mischarging by the glutamine AAS (7 -1 0). Chemical modification of the E. coli tRNAArg or yeast tRNAPhe anticodons reduces cognate aminoacylation in vitro (11-14). It was possible, then, that the amber suppressor alleles that we had constructed would have altered specificity, i.e., the recognition of these tRNAs by their cognate AAS would be affected by the change introduced into the anticodon. Determining the amino acid specificity of the newly constructed suppressor tRNAs would reveal the extent to which the anticodon plays a role in AAS recognition for each of these tRNAs. Previously, we reported the development of a specificity assay in which a tRNA was used to suppress an amber mutation in the tenth residue of the E . coli dihydrofolate reductase (DHFR) genefol. The mutant protein was subsequently purified and partly sequenced to determine which amino acid the suppressor 109 tRNA had inserted at the amber codon (3, 5). In this paper we present the results of the specificity assays for the 14 new suppressor tRNAs. METHODS Bacterial Strains and Media. All strains were derived from E. coli XAC which is F!i(lacproB)xln• na/A, rif, argEam, ara. XACRF is XAC with uar and tnlO transduced into the chromosome. The episome (laci373 lacZuzl8amProB+) is designated -1 as in XAC-1 II II and XACRF-1. The episome (lacH'zyAtn5) is designated 9100.1 and is found in 11 11 XAC/9100.1 and XACRF/9100.1. The episome Zmet3 is (lacllacZam3) and is found in XAC/Zmet3. Minimal M9 media was prepared according to Miller (15) and supplemented with 100ug/ml ampicillin, and 30ug/ml chloramphenicol, both from Sigma. L broth was 10g/l Bactotryptone, 5g/l yeast extract, and 8g/l NaCl. Plasmid Construction. Plasmid pDA3-12 (3, 5) contains the coding region of the E . coli DHFR gene fol with an amber mutation at residue ten ifofamlO), under control of the tac promoter, cloned into the Eco RI-Hind III sites of pBR322. pDA3-12:Arg was constructed by subcloning the Pvu II fragment ofpGFIB:Arg (1), which carries the tRNAArgCUA gene expressed from the lpp promoter, and followed by the rrnC transcription terminator, into the Pvu II site of plasmid pDA3-12 (Figure 1a). pDA3-12:AspM, pDA3-12:lle2, pDA3-12:Thr2, and pDA3-12:Val were constructed in a similar fashion. To construct plasmid pDA5YC, a Hind ll-Eco RI adaptor was ligated to the Eco RI cohesive end of the Eco RI-Bam HI fragment of pDA3-12, and subcloned into the Hind III-Bam HI sites of plasmid pACYC184 (Figure 1b). Plasmid pDAYQ was constructed by attaching Sal I linkers to the Eco RI fragment from plasmid pJWlaci (16), which carries lacH, and ligating the entire fragment into the Sal I site ofpDA5YC (Figure 1b). All of the suppressor tRNA genes were subcloned into wild type lambda in the following manner. Xba I or Sal I linkers were ligated to the Pvu II fragment of each of the pGIB:suptRNAs (1, 2), which carries the tRNA gene, along with the lpp promoter and a transcription terminator. The linkered fragment was then ligated into the single Xba I site of wild type lambda (Figure 1c). Cloning techniques were as described by Davis et al. (17). Phage constructs containing only a single copy of the suppressor tRNA gene were used to lysogenize E. coli strains. 110 Purification ofDHFR. XAC-1 or XACRF-1 carrying both pGFIB:sup and pDAYQ; XAC/9100.1 or XACRF/9100.1 carrying pGFIB:sup and pDA5YC; XAC/9100.1 or XACRF/9100.1 carrying pDA3-12:suptRNA alone, were all cultured in minimal M9 media supplemented with ampicillin (chloramphenicol was added in the presence of plasmids pDA5YC orpDAYQ) to an A 600 of0.5 to 0.7. The cell cultures were diluted 100 fold into fresh minimal media, grown to an A 600 of0.4 to 0.6, at which time thefol gene was induced with isopropyl thiogalactoside (IPTG) at a final concentration of lmM, for 4 to 18 hr. Alternatively, XAC/9100.1 or XACRF/9100.1, which had been lysogenized with the lambda suppressor constructs and transformed with plasmid pDA3-12, were cultured and induced as above except that L broth supplemented with ampicillin was used instead of minimal medium. The cells were harvested, and DHFR was purified as described by Normanly et al. (3, 5) with only one notable change. A MonoQ column (Pharmacia) was used interchangeably with a DEAE-sephacel column (Pharmacia) in the final stage of the purification. Purified DHFR was subjected to N-terminal sequence analysis at the USC Microchemical Core Facility or the Caltech Microchemical Facility. Purification of ~-galactosidase. Srain XAC/Zmet3 carrying pGFIB:Val was cultured in minimal media as described for the purification of DHFR except that the IPTG induction was eliminated. Cells were harvested and frozen at -200C. Subsequent steps were carried out on ice with stirring. The cell pellet was thawed in 2.14 ml/g cells TEP (lOOmM Tris HCl, pH 7 .4, lOmM EDTA, lmM phenylmethylsulfonyl fluoride, PMSF). The following additions were made at 30 min. intervals: (1) 0.15 ml/g cells of lysozyme (17mglml in TEP); (2) 0.055 ml/g cells of 1M MgC12 and 0.015 ml/g cells of DNasei (13.2 mglml in TEP). The suspension was centrifuged at 12,000 x g for 15 min. at 4°C. The supernatant was saved and the pellet resuspended in 2.14 ml/g cells TEP. Steps 1 and 2 were repeated and the suspension centrifuged as above. The pellet was washed with 0.92ml!g cells TEP and centrifuged again. All three supernatants were pooled, brought to 40% saturation with ammonium sulfate and stirred on ice for 2 hr. The suspension was centrifuged at 19,000 x g for 30 min. at 4°C. The precipitate was resuspended in lml of TEP, and brought to a concentration of 4 mglml with TBSN ( lOmM Tris HCl, pH 8.0, 150mM NaCl, 0.2% NP-40). The suspension was equilibrated with lml ofPromega Proto Sorb Lac Z column matrix, batch washed with TBSN, and applied to a column. ~-galactosidase was eluted from the column with 100mM Na-Carbonate buffer, pH 111 10.8, according to the manufacturers instructions. Purified beta-galactosidase was dialyzed against distilled water, lyophilized and sequenced at the USC Microchemical Core Facility. RESULTS In order to determine the specificity of the amber suppressor tRNAs, we constructed strains which carried both the suppressor tRNA gene to be assayed andjo[amiO. We used three different methods to create these strains. In one case the suppressor tRNA gene and jo[amiO were carried on the same plasmid. In another case,jo[amlO resided on a plasmid, while the suppressor tRNA gene was integrated into the chromosome on a lambda prophage. Last, both the suppressor tRNA gene andjo[amiO were carried on separate, compatible plasmids. We used the E. coli strains, XAC and XACRF, which differ in two genetic markers. XACRF was derived from XAC by transduction of uarts (linked to tn10), into the chromosome. The uarts allele is a release factor mutant. Its effect is to enhance suppression efficiency, because the release factor competes poorly with the suppressor tRNA for the nonsense codon. This strain was employed for those suppressor tRNAs with lower efficiencies. Single Plasmid System. Previously, we described the plasmid pDA3-12, which carriesjo[amlO under control of the tac promoter (3, 5). The suppressor alleles corresponding to tRNAs Ile2, Val, AspM, Thr2, and Arg (1, 2) were subcloned into the pDA3-12 vector (Figure 1a). Lambda Prophage and Plasmid Encodedfol Amber. We have found the plasmids encoding suppressor tRNAs to be somewhat unstable unless suppression is required for growth. The instability is probably due to the high copy number of the vector (pGFIB-1) and the fact that the suppressor genes are expressed constitutively (1, 4). The E . coli strain XAC has an amber mutation in the ArgE operon, which will accept a variety of amino acids. Consequently, if the suppressor-carrying strains are maintained on minimal media, the plasmids are stable. Our long term goal is to compile a collection of suppressor tRNA genes, which can be used in any strain background. We speculated that if the suppressor genes were present in lower copy number they would be more stable and would not require the ArgE background. Additionally, we wished to determine the effect of lowered copy number on the suppression efficiency of the tRNAs. We cloned 13 of the suppressor tRNAs genes, expressed from a constitutive promoter, lpp, (4) and followed by 112 a transcription tenninator, rrnC (3, 5), into the single Xba I site of wild type lambda. These phage constructs were integrated into the chromosome of E. coli strain XAC and various derivatives, and the suppressor tRNAs were assayed for efficiency (data not shown) using the Laclam-Z system of Miller (18). It was clear that the copy number of these tRNA genes played an important role in their suppression efficiencies. Of the 13 suppressor-carrying prophage, only those encoding tRNAGlyTCUA· tRNAHisACUA and tRNAAla2CUA had suppression efficiencies high enough to render the tRNAs useful in amino acid substitution studies (data not shown). These strains were transformed with plasmid pDA3-12, which encodesjo[amlO, for the specificity assay (Figure 1b). Compatible Plasmid System. We also developed a system wherein the suppressor tRNA gene resided in pGFIB-1, andjo[amlO was carried on a compatible plasmid. We subcloned thejo[amJO gene into the Hind III-Bam HI site of pACYC184 to yield pDA5YC (Figure 1c). Because jo[amJO is under tac promoter control, it is necessary to maintain the plasmid in an lacJq strain (19). For added flexibility, we subcloned the lacH gene into plasmid pDA5YC to yield plasmid pDAYQ (Figure 1c). Suppressors for tRNAs Ilel, Ile2, Arg, AspM, Val, Thr2, ProH, Met(m), Lys, GluA and GlyU were transformed into either strains (XAC-1 or XACRF-1) carrying plasmid pDAYQ or strains (XAC/9100.1 or XACRF/9100.1) carrying plasmid pDA5YC. The cells were cultured in minimal media, and after induction of the fa/ gene with IPTG, DHFR was purified from these cells by affinity chromatography. N-tenninal sequence analysis of the purified proteins revealed which amino acid(s) had been inserted in response to the amber codon by each suppressor tRNA. Protein Sequence Data. Figure 2 shows the results from the protein sequencing. tRNAAla2CUA• tRNAHisACUA• tRNALysCUA• and tRNAProHCUA all inserted the predicted amino acid exclusively (Figure 2a). Interestingly, two different isoacceptors of tRNAGly behaved differently when their anticodons were altered to CUA. tRNAGlyUCUA inserted exclusively glycine, while tRNAGlyTCUA inserted a mixture of glycine (62%) and glutamine (37%) (Figure 2b). Similar results for the GlyT suppressor have been obtained elsewhere (20). The glutamic acid suppressor tRNAGluACUA inserted mostly glutamic acid (59%) but was mischarged by the glutamine AAS as well (17%). One isoacceptor of tRNAlle, tRNAllelCUA· was mischarged by the glutamine AAS (74%) and to some extent by the lysine AAS (16%), while another isoacceptor of lie, tRNAlle2cuA. was mischarged 113 exclusively by the lysine AAS (Figure 2b). Five other suppressor alleles, tRNAArgcuA, tRNAAspMCUA' tRNAMet(m)CUA, tRNAThr2cuA, and tRNAValCUA were mischarged by the lysine AAS (Figure 2c). In the cases of tRNAArgCUA· tRNAThr2cuA. and tRNAVa!CUA, some of the predicted amino acid was inserted as well, but not more than 10%. The average background level for any amino acid residue is approximately 3%. Amino acid levels of 5% and above were considered significant. The percent amino acid incorporated into protein may not add up to 100%, because background levels are not included. For example, while proline is the major amino acid inserted by tRNAProHcuA, and no more than 3% of any other amino acid was found, the value given for percent proline incorporated into protein is only 85%. As stated in the introduction, there is precedent for mischarging by the glutamine AAS when the anticodon is altered; however, mischarging of a tRNA by the lysine AAS was completely unexpected. Since we began this work others have observed mischarging by the lysine AAS (21). The existence of a second lysine AAS in E. coli, which is a heat shock protein, may explain the lysine mischarging (22). It occurred to us that the overproduction of DHFR might induce a heat shock response and cause mischarging by the lysine AAS. To rule out this possibility, we decided to assay the specificity of tRNAVaiCUA using another protein. We chose a mutant of the LacZ gene, with an amber mutation residing in the third residue. This gene is encoded episomally, and ~-galactosidase is not overproduced to the extent that DHFR is. pGFIB:Val was transformed into XAC{Zmet3, from which ~-galactosidase was purified. As for the fol amber, lysine was found to be the predominant amino acid inserted, with no more than 5% valine shown to be incorporated (data not shown). DISCUSSION We have generated amber functional suppressor alleles for 14 species of tRNA (Table 1), and determined their amino acid specificities. Tables 2 and 3 summarize the results. Table 2 gives the specificities of amber suppressor alleles generated in this work as well as from previous work by us and others. With the exception of tRNAAsn, functional supressor alleles exist for all tRNA species (and in some cases multiple isoacceptors). The specificities of these suppressors are known in vivo for all but tRNAMet(f)cuA, which has been examined in vitro (10). The specificity assays conducted upon these tRNAs reveal that they may be classified in one of three ways (Table 3). For Class I, altering the 114 anticodon to CUA does not affect cognate AAS recognition in vivo. For the other two classes of tRNA, such an alteration results in mischarging by either the glutamine (Class II) or lysine (Class ill) AAS. Interestingly, different isoacceptors of a given tRNA species may fall into separate classes, as in the case of tRNAllelcuA and tRNAile2cuA, as well as tRNAGlyUCUA and tRNAGlyTCUA· Our goal to generate a complete bank of amber suppressor tRNA genes has only been partially achieved. The suppressors that fall into class I will be useful in amino acid substitution studies, and to this end the existing collection of useful amber suppressors has been increased from 4 to 11. (A suppressor allele of tRNAGiyUCUA had been generated previously by genetic means but the exact specificity was unknown; 23). The tRNAs in class II and ill remain an obstacle. It is not clear what distinguishes one class of tRNA from another; sequence comparison among the three classes does not reveal any uniquely conserved nucleotides. A fortuitous mutation in the tRNAArgCUA suppressor has resulted in a variant that inserts exclusively arginine (24), although it is not immediately obvious why this is the case. Clearly, if we are to succeed in our goal of generating a complete collection of amber suppressor tRNA genes, each inserting the predicted amino acid, we must define those elements of class ll and ill tRNAs that dictate cognate AAS recognition. REFERENCES 1. Masson, J.-M., Kleina, L.G., Normanly, J.H., Miller, J.H., and Abelson, J. 1988 2. manuscript in preparation. Kleina, L.G., Normanly, J., Masson, J.-M., Miller, J.H., and Abelson, J. 1988. 3. manuscript in preparation. Normanly, J., Masson, J-M., Kleina, L.G., Abelson, J., and Miller, J.H. 1986. 4. 5. Proc. Nat. Acad. Sci. 83:6548-6552 Masson, J.-M., and Miller, J.H.l986. Gene,47:179-183 Normanly, J., Ogden, R.C., Horvath, S.J., and Abelson, J. (1986). Nature , 6. 7. 8. 9. 321:213-21 Yarns, M . 1982. Science, 218:646-652 Celis, J.E., Coulondre, C., and Miller, J.H. l976. J. Mol. Bioi. 104:729-734 Yaniv, M., Folk, W.R., Berg, P., and Soli, L.l974. J. Mol. Bioi. 86:245-268 Yarns, M., Knowlton, R., Soil, L. 1977. in Nucleic acid-protein recognition, pp. 10. 391-408 New York, Academic Press Schulman, L.H., and Pelka, H. 1985. Biochemistry, 24:7309-7314 115 11. 12. 13. 14. Chakraburrty, K. 1975. Nucleic Acids Res. 2:1793-1804 Bruce, A. G., and Uhlenbeck, O.C.l982. Biochemistry, 21:855-861 Bruce, A.G., and Uhlenbeck, O.C.l982. Biochemistry,21:3921-3927 Kisseliv, L.L. 1985. in Progress in Nucleic Acids Research and Molecular Biology 15. (Cohn, W.E. Moldave, K. eds.) Academic Press, Florida, v.32,237-266 Miller, J.H. 1972. In Experiments in Molecular Genetics, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. Zumstein, L., and Wang, J.C.l986. J. Mol. Bioi. 191:333-40 Davis, R.W., Botstein, D., and Roth, J.R.l982. InA Manualfor Genetic Engineering: Advanced Bacterial Genetics. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY Miller, J. H., and Albertini, A. M.l983. J. Mol. Bioi. 164:59-71 Amann, E., Brosius, J., and Ptashne, M. 1983. Gene, 25:167-178 McClain, W.H., and Foss, K.l988. Science 240:793-796 McClain, W.H., and Foss, K.l988. J. Mol. Bioi. 202:697-709 Hirshfield, LN., Bolch, P.L., BanBoge1en, R.A., and Neihardt, F.C.l981. J. Bacterio/.146:345-351 Murgola, E. J. 1985. Ann. Rev. Genetics.l9:57-80 McClain, W.H., Foss, K. 1988. Science in press. Normanly, J., Abelson, J. manuscript in preparation. Hewick, R.M., Hunkapiller, M.W., Hood, L.E., and Dreyer, W.J.l981. J. Bioi. Chem. 256:7990-7997 116 Table 1. Suppression efficiencies of amber suppressor tRNAs. Suppressor Suppression efficiency* Alal <1% <1% SET 1 Asp Asn Arg Cys Glu GlyT <1% 11-55% 17-54% 1-44% 44-67% 24-100% 1-38% 1-10% 9-29% GlyU His Ilel Lys Met(m) 5-45% Phe Pro Thr1 Vall 48-100% <1% <1% <1% Ala2 AspM 8-83% 10% GluA 8-100% 16-100% SET2 HisA Ile2 ProH Thr2 Va12 6-67% 8-60% 1-14% 18-83% *Suppression efficiencies are a direct measurement of 13-galactosidase activity(% of wild type) resulting from suppression of amber mutations in lacl-Z fusions (15). 117 Table 2. Compilation of amber suppressor tRNA specificities. Suppressor Specificity (given as percent amino acid Reference inserted at amber codon) Ala2 Ala 97% this work Lys Lys II ProH Pro His Gly,Gln 94% 85% Hi sA GlyT GlyU Tiel Met(m) Gly Gln,Lys Lys 94% 62%,37% 90% II 74%,16% 97% 93% 81% II II II lle2 AspM Arg Lys Lys Lys,Arg Thr2 Lys,Thr 91%,5% 86%,8% Val Lys,Val 84%,5% GluA Glu,Gln Tyr,Arg 59%,17% Phe Cys Phe Cys Leu5, and su+6 Ser(su+l) Gln(su+2) Leu Ser Gint Tyr(su+3) Trp(su+7) Tyr Gln,Trp Met(t) Gin* II 6%,6% 98% 99% 92% 93% 95% *Demonstrated in vitro only. tincludes Glu as well. II II II II II II 3 3 5 5 5 25 7 10 118 Table 3. tRNAs classified by the effect of a CUA anticodon upon AAS recognition. Class I Class II Class III No affect Lysine AAS upon charging Glutamine AAS mischarging Ala2 Ilel GlyU GlyT Met(f) lle2 Arg Met(m) GluA Trp (su+7) AspM Thr2 Cys Phe ProH HisA Lys Ser (su+I) Gln (su+2) Tyr (su+3) Leu (su+6;also synthetic Leu5) mischarging Val 119 FIGURE LEGENDS Figure 1a. The Pvu II fragment encoding the amber suppressor tRNA genes, flanked by the lpp promoter (4) and rrnC terminator (3), were subcloned into the Pvu II site of plasmid pDA3-12 (3). "suptRNA" in pDA3-12:suptRNA is generic, representing each of the suppressor tRNA genes that have been subcloned into the plasmid. For example pDA3-12 carrying the Arg suppressor is actually referred to as pDA3-12:Arg. Figure 1b. Xba I or Sac I linkers were ligated to the Pvu IT fragment shown in Figure 1a, and subcloned into the single Xba I site of wild type lambda DNA. The lambda constructs were used to lysogenize E . coli strains, which were then transformed with plasmid pDA3-12. Figure 1c. pGFIB:suptRNA is derived from plasmid pGFIB-1 (3). Again "suptRNA" is a generic term. Plasmid pDA5YC encodes the fol amber and a chloramphenicol resistance marker. Plasmid pDAYQ differs from pDA5YC in that the lacH gene is inserted into the Sal I site of pDA5YC. The lacH gene overproduces the lac repressor, which is necessary for the regulation of the fol gene's tac promoter. With the lacH gene encoded by the same plasmid that carries the fol amber, it is not necessary to use an E. coli strain that carries the lacH allele chromosomally. Figure 2. Protein sequence analysis of DHFR substituted at position ten by the newly generated suppressor tRNAs. Sequence analysis was carried out on an Applied Biosystems model 470A gas phase sequenator (26) by the USC Microchemical Core Laboratory. The picomolar yields ofphenylthiohydantoin (PTH) derivatives of various amino acids are plotted against the residue number. The N-terrninal sequence (up to 15 residues) of the DHFR mutant is indicated below each plot, and the percent yield of the amino acid at residue ten is indicated to the right of the plot. Only the most significant yields are presented; amino acids not shown can be presumed to be present at background levels. Protein sequence results from DHFR substituted at position ten with a) tRNAAla2CUA, tRNALysCUA, tRNAProHCUA, and tRNAHisACUA b) tRNAGlyTCUA, tRNAGlyUCUA, tRNAGluACUA, and tRNAllelCUA· *Data not shown. c) tRNAlle2 cuA, tRNAMet(m)CUA, tRNAArgCUA, tRNAAspMCUA, tRNAThr2cuA, and tRNAValCUA· 120 Ptac Eco R1 Hind ill pDA3-12:sup tRNA Bam H1 Ptpp sup tRNA gene Figure la TrrnC 121 Ptac pDA3-12 Bam H1 Xbal Xbal Ptpp sup tRNA gene Figure 1b TrrnC lambda 122 / Eco R1 I pDASYC pGFIB:sup tRNA Ptpp sup tRNA gene Figure lc .-. 5 10 • Pro 85% I 10 15 DHFRam1oftRNAHiaACUA • His Lya 94% 94% N ~ 5 10 15 ···'·'·'·'• 1 • 1 • 1 • ' · ' · ' · ' · ' ' Ml SLIAALAPNRVIG 0 1·. 1000 2000 ~~------------~------· 0 10000 20000 I I I I I 5 I =i " I • i I I 10 I II I I I I I I I .I 15 MISLIAA LAHNRVIG I ~ ~------------------, N DHFRM11of1RNAProHCUA 5 -: I .. MISLIAALAKNRVIG 1 o_._ • ...... G 15 ···'·'·'·'·'·'·'·'·'·'! 2000-. 4000 6000 DHFR81111oftRNALY•cuA Vl l I MISLIAALAANRV 0 I 2000 .Ala 87% 0 E Q. 0000 10000 ~· 'T1 6000 1 4000 D. DHFR81111 of1RNA~CUA oooo~------~~----------, 0" N ~· "TT Cl. e 0 i 0 10 15 10 15 M I S L I A A L AEJQN A V I G 5 o I' ,• •"•"•"•"•"•"•"•"•"•"•"•"•"I 1000 2000~--------------------------, DHFRam1 c{tRNAGluACUA M I S L I A A LAG/ON R V I G 5 0 1•,·,·,•,·,•,•,•,•,•,•,•,•,•,•1 1000 2000 3000 ~~----------------------, DHFRam1oftRNAGlyTCUA • 0 • 0 6% 6% Tyr" Arg' 59% 17% Gin Glu Gin 62% 37% Gly l eCl. 0 I ~ 5 10 ~ - __ - - ~ L l - I I I 0I 10000 5 ,· ,· ,·,·,· , · 10 , . , . , .. 15 ,•,•,•,•1 M I S L I A A L AQ/KN R VI G I 20000~----------------------· DHFRam1oftRNA1111 CUA MI ' SLIAALAGNRV 1 0 I• _ 1000 2000 3000~------------------------· DHFRam1 c{tRNAGtyUCUA 0 • 0 • 74% 16%' Lys 90"1. Gin Gtn Gly ~ ...... 125 DHFR..,1of!RNA-mlcuA 8000 6000 0 E ... 0000 • c .. ll7'llo • Lp - · I.. 2000 0 5 M I S L I A A l I Lp 10 15 A KN R V I G MISLIAALAKNRVIG IJ Alp • 1000 • Lp 1111. l 0 IJ Arg • 1000 0 5 10 5 15 M I S Ll A A L A K N R V 10 • I. "" Lp 8111. IJ Vol 11!1. 15 M IS L IAA L A K N RVI G G OHFA..,1 oJlRHATlw2CUA 0000 8000 5000 3000 I IJ Tlw • 2000 1000 0 .. ~ .. 5 10 I • Lp ... 4000 I • • 3000 • 2000 1000 0 15 5 M I S L IAA L A KN R V I G Ml Figure 2c 10 SLIAA LAKNRVI Lp-