Identity of Saccharomyces cereuisiae tRNATrP Is Not Changed by an Anticodon Mutation That Creates an Amber Suppressor*

A C35+T mutation in an Escherichia coli tRNATm gene creates an amber suppressor which efficiently inserts glutamine in response to UAG codons in vivo (Soll, L., and Berg, P. Nature 223, 1340-1342). We have introduced the same change in a yeast tRNATrp gene and demonstrated that the tRNA acts as an efficient amber suppressor in vivo. Amino acid se- quence analyses were performed on chitinase produced by cells carrying the corresponding gene with a UAG codon at position 8 of the mature protein plus the mutant tRNATrp gene. In contrast to comparable ex- periments with E. coli, tryptophan is inserted at a frequency ~ 8 0 % by the yeast suppressor tRNATrp. Fur- thermore, in vitro charging experiments with the mutant tRNATm reveal no detectable increase in gluta- mine acceptor activity results from the C35+T transition. The identity elements in E. coli tRNAG’” are well characterized (Jahn, M., Rogers, J., and Soll, D. Nature 352, 258-260). Sequence comparisons of the tRNATm and tRNAG’” molecules from E. coli reveal that the amber suppressor tRNATm has four of five identity elements required for glutaminyl-tRNA synthetase recognition. A similar comparison in the yeast system shows only two of the five potential identity elements are present. We conclude that, in spite of substantial structural similarities between yeast and E. coli ami- noacyl-tRNA synthetases, fundamental differences can exist with regard

In recent years, an interest in elements of tRNA structure which control recognition by cognate and non-cognate aminoacyl-tRNA synthetases (RS)' has given rise to a considerable amount of research in this field (reviewed by Schimmel (1989) and Normanly (1989)). The anticodon has been shown t o be a strong determinant of identity in numerous tRNAs (Bare and Uhlenbeck, 1986;Pelka, 1988,1989;McClain et al., 1988McClain et al., , 1990McClain et al., , 1991Jahn et al., 1991;Putz et al., 1991;Himeno et al., 1991;Pallanck and Schulman, 1991;Hasegawa et al., 1992;Nazarenko et al., 1992;Pallanck et al., 1992;Rogers et al., 1992). Any tRNAs in which all members of an isoaccepting group have one or more common bases in the anticodon are candidates to utilize these positions as identity elements. This mode of recognition has been demonstrated for Escherichia coli tRNATrp both in uivo and i n * This work was supported by National Science Foundation Grant DMB-8802058. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact, $ T o whom correspondence should be addressed. Tel.:  ' The abbreviations used are: RS, tRNA synthetase(s); aa, aminoacyl; PTH, phenylthiohydantoin.
uitro (Yaniv et al, 1974;Himeno et al., 1991). Results from Soll and Berg (1969) as well as Celis et al. (1976) have shown that an amber suppressor form of E. coli tRNATw inserts predominantly Gln during translation of a corresponding nonsense codon position in vivo. Work by Knowlton et al. (1980) has indicated that E. coli tRNc:A has a dual specificity, charging equally well with either Trp or Gln in vitro and in vivo. Surprisingly, it appears that the translational apparatus somehow discriminates against the Trp-tRNAEEA.
Acylation of the with both Gln and Trp is perhaps not surprising in view of similarities in the structures of E. coli tRNATW and tRNAG1" at positions important for GlnRS recognition (Rogers et at., 1992). Both mutational analyses (Jahn et al., 1991) and the crystal structure determined for a tRNAG1"-GlnRS complex (Rould et al., 1989) unequivocally illustrate the importance of both the acceptor stem and anticodon for recognition. The relatively relaxed charging specificity of E. coli GlnRS also likely contributes to heterogeneous charging (Englisch-Peters et al., 1991). After losing a positive identity controlling element at the central position in the anticodon, E. coli t RNA: : * has a sufficiently lower affinity for TrpRS that the GlnRS becomes an effective competitor for this substrate (Knowlton et al., 1980). In support of this hypothesis, a G1 + A mutant in an amber suppressor form of E. coli tRNA5' also inserts Gln at UAG codons in vivo (Ghysen and Celis, 1974), suggesting that this base, along with U35 and G2:C71, are sufficient for positive recognition by GlnRS.
In prior work, the suppressor activity of yeast tRNAETA has been documented in vivo (Kim and Johnson, 1988;Kim et al., 1990). To determine what amino acid this tRNA inserts in response to UAG, a chitinase gene altered to contain an amber codon near its amino terminus (Edwards et al., 1991) was expressed in Saccharomyces cerevisiae cells also containing the gene. Amino acid sequence analysis of chitinase isolated from this cell line identified 280% Trp at the position specified by UAG. Furthermore, in vitro charging assays using tRNAs produced by T7 RNA polymerase from synthetic genes corresponding to and t RNA: : * show no increase in acceptor activity for Gln with the amber suppressor tRNA.

EXPERIMENTAL PROCEDURES
The construction of YCpTrp and YCpTrpA, carrying the yeast tRNATv gene and a C 3 b T variant, respectively, has been reported (Kim et al., 1988). Yeast strain W303VctsI and plasmid constructs pCT28 and pCTam were generously provided by Drs. Paul Schimmel and Catherine Reynolds. The yeast strain W303VctsI is a derivative of W303-1B in which the chitinase gene has been disrupted by a directed integration (Kuranda and Robbins, 1991). The plasmid pCT28 carries the structural gene for chitinase, CTS1-I (Kuranda and Robbins, 1991), in the vector YEp352 (Hill et al., 1986). The plasmid pCTam contains a version of the CTS1-1 gene in which the codon for Asn' of the mature protein has been converted to an amber codon (Trezeguet etal., 1991). The W303Vctsl cells were transformed 217 by the procedure of It0 et al. (1983). Transformants were identified by selecting for nutritional markers and the presence of the expected plasmid(s) confirmed by Southern blot analysis. Chitinase was prepared essentially by the procedure of Kuranda and Robbins (1991). The protein isolated from each cell line was loaded onto a 7% polyacrylamide, 0.1% SDS gel, electrophoresed, and stained with Coomassie Blue for visualization. For amino acid sequence analysis, proteins in SDS-polyacrylamide gels were electroblotted to Immobilon polyvinylidene difluoride membrane (Bio-Rad), then stained with Coomassie Blue. The chitinase bands were cut from the membrane and subjected to sequence analysis using an Applied Biosystems 473A Protein Sequencer. Charging experiments were done using a commercial yeast aaRS preparation (Sigma). Bulk yeast tRNA was isolated as described by Monier et al. (1960) and deacylated following the procedure of Sarin and Zamecnik (1964). Transcripts representing tRNA:& and tRNAFg were made using T7 RNA polymerase with cloned oligodeoxyribonucleotide templates as described by Sampson and Uhlenbeck (1988). Nucleotide sequence analyses were done on the templates t o confirm their structure. Prior to use in charging reactions, the T7 transcripts were purified by both electrophoresis on 10% polyacrylamide, 8 M urea gels and reverse-phase chromatography, then renatured by slowly cooling from 65 "C in the presence of 20 mM MgC12. Charging reactions were done a t 37 "C in 30 mM Hepes (pH 7.5), 25 mM NH4Cl or KCl, 15 mM MgC12,2.5 mM ATP, 4 mM dithiothreitol, and either 12 p M [3H]Trp (16 Ci/mM) or 15 p M t3H]Gln (26 ci/mM). The T7 transcripts were present a t 1.0 p~ or the bulk tRNA a t 5.8 mg/ml. Aliquots were removed at the times indicated and spotted on GF/C disks (Whatman LabSales Inc., Hillsboro, OR), which were then immersed in cold 10% trichloroacetic acid. The disks were washed 5 times in cold 5% trichloroacetic acid, once in 65% ethanol, air-dried, and then radioactivity was measured in 5 ml of a liquid scintillation mixture (Beckman Instruments, Fullerton, CA).

A paucity
of naturally occurring nonsense suppressor tRNAs in yeast (Sherman, 1982), compared to E. coli (Raftery and Yarus, 1985), encouraged an attempt to create via in vitro genetics tRNAs which could function in this capacity. A C35+T mutant of the s. cerevisiue tRNA&, gene was created, tRNGTA, and its ability to suppress a number of amber mutations in vivo was demonstrated (Kim and Johnson, 1988). To determine whether this tRNA actually inserts Trp in response to UAG codons, we used a modified version of the yeast chitinase gene, CTS1, in which the eighth codon of the mature protein has been altered to UAG (Edwards et al., 1991;Trezueget et al., 1991). Chitinase is secreted by yeast with cleavage of a signal peptide. A high affinity chitin binding domain is localized at the C terminus (Kuranda and Robbins, 1991). The mutation at position 8 affects neither secretion nor chitin binding. Therefore, mutant chitinase with any amino acid substituted a t position 8 can be isolated in high purity directly from the media as a complex with insoluble chitin.
Chitinase was prepared in this way from a cell line in which the chromosomal copy of CTSl was inactivated by a directed integration. The cells produced normal amounts of the enzyme when transformed with a plasmid, pCT28, which carries the CTSl gene (Fig. 1). The same host cell, transformed with a plasmid carrying the C T S l gene with a UAG codon at position 8, pCTam, did not produce chitinase (data not shown). The inclusion of the tRNA& gene did not restore chitinase production, indicating neither resident nor cloned copies of the normal tRNATw gene can suppress the UAG mutation. However, in the presence of the tRNATw CUA gene carried on YCpTrpA, chitinase synthesis was restored (Fig.  1). This demonstrates that the chitinase produced by these cells is dependent on the presence of both the mutant CTSl gene and the suppressor tRNA. The quantity of chitinase recovered from cells containing pCTam+YCpTrpA was about 10-20-fold lower than from those containing pCT28 indicat- ing a 5-10% efficiency of suppression. Fig. 1 also demonstrates that, in the absence of t RNA: : A, no other protein of 110 kDa was recovered from the media by the purification scheme. Therefore, protein from this region of the SDS-polyacrylamide gel can be subjected to amino acid sequence analysis with minimal background from contaminating proteins.
The amino acid sequences determined for residues 7-16 of the chitinases produced in the presence of pCT28 and pCTam+YCpTrpA are reported in Fig. 2. Only the major amino acid is reported for each position except position 8. Because Trp is a labile amino acid, the molar yields from sequencing must be corrected. The presence of Trp at position 13 in the protein fortuitously provides an internal standard to correct for this instability. A correction factor was calculated using results from the analysis of protein from cells carrying pCTam+YCpTrpA. By averaging the yield of PTH amino acids a t positions 12 and 14, then dividing this value by the amount of PTH-Trp recovered a t position 13, a multiplier of 7.0 was derived, which was then applied to Trp recoveries a t position 8 of the same protein as well as position 13 of the pCT28 product (Fig. 2, A and B ) . As expected, Asn was the major amino acid recovered a t position 8 with protein from cells carrying pCT28. A minimum of 80% Trp was present in response to the UAG codon at the same residue in protein from the pCTam+tRNA::Asystem (Fig. 2B). Low levels (510% the amount of Trp) of Gln and Asp were recovered in this cycle. Similar amounts of background were also observed in the other sequencing cycles. We can therefore conclude that 280% of the amino acid inserted by t RNA: : A is indeed Trp.
Knowleton et al. (1980) demonstrated that, both in vivo and in vitro, approximately equal amounts of Trp and Gln were esterified to E. coli t RNA: : A.
T o determine whether S. cerevisiue t RNA: : A was a substrate for the yeast GlnRS, an in vitro system was established to allow direct measurement of charging. Following the methods developed by Sampson and Uhlenbeck (1988), synthetic oligodeoxyribonucleotides were synthesized and cloned such that T7 RNA polymerase could be used to produce transcripts representing maturesized tRNA:& and These transcripts were used as substrates for a crude aaRS preparation from yeast with either [3H]Trp or [3H]Gln present. Bulk, deacylated tRNA from yeast was used as a positive control for the activity of both RS enzymes (Fig. 3A). The results demonstrate that the C35+T transition which generates the amber suppressor does  9 1 0 1 1 1 2 1 3 1 4 1 5 1  not measurably affect the Gln acceptor activity of tRNATv ( Fig. 3, B and C). The levels of the Gln charging in Fig. 3 ( B and C) are equal and represent 1 2 times background.

DISCUSSION
An amber mutation in the CTSl gene of yeast can be suppressed at about 5-10% efficiency by tRNAEEA (Fig. l ) , and Trp is inserted at 280% of the suppression events (Fig.  2). The nearly quantitative insertion of Trp in this system is i n clear contrast to a comparable result in E. coli, where approximately -90% Gln is found at positions specified by a UAG codon when is present (Celis et al., 1976). This is quite surprising as other work (Knowlton et al., 1980) demonstrated that the kinetics of charging of tRNg:A by the E. coli GlnRS and TrpRS enzymes i n vitro are nearly identical. Also, i n vivo the accumulation of glutaminyl and tryptophanyl forms of the tRNA is nearly equivalent. This implies a very unusual selection must exist for the Gln in some aspect of E. coli translation. Again, the yeast system behaves rather differently as i n vitro charging experiments indicate no measurable Gln acceptor activity for the (Fig. 3). The anticodon mutation does impede charging with [3H]Trp by 10-20-fold. This may provide a partial explanation for the reduced amount of chitinase produced by yeast cells carrying pCTam+YCpTrpA (Fig. 1). The reduction in extent of [3H]Trp charging observed i n vitro with the amber suppressor mutant is likely due to a non-enzymatic deacylation reaction (Bonnett and Ebel, 1972 for Trp and Gln is presented in Fig. 4. The sequence elements that specify the charging identity for the E. coli system have been established (Himeno et al., 1991;Jahn et al., 1991) and are boxed in the figure. The C35-T transition in E. coli results in five of six identity elements for recognition by GlnRS being present in this molecule. Two of these similarities, U35 and G73, are known to be the predominant recognition sites (Ghysen and Celis, 1974;Jahn et al., 1991). Comparing the two yeast tRNAs in these putative recognition regions suggests a possible explanation for the different charging behavior. Only two of the five positions are equivalent between the yeast tRNAs corresponding to and tRNAG1" (Fig. 4). This suggests that the recognition elements used by the TrpRS and GlnRS enzymes have evolved differently in yeast and perhaps other eukaryotes (Nagel and Doolittle, 1991). This also appears to be the case for TyrRS, LeuRS, and MetRS (Edwards et al., 1991;Lee and Raj-Bhandary, 1991). Alternatively, the differences may reflect an intrinsic misacylation tendency of the E. coli GlnRS (Englisch-Peters et al., 1991). An important caveat emptor to these arguments is that the recognition sites for the yeast  1971;Hirsh, 1970) along with the tRNA& and tRNAgk sequences for both organisms (Rould et al., 1989;Tschumper and Carbon, 1982;Weiss and Friedberg, 1986).
Nucleotides known to be involved in recognition of the E. coli versions of these tRNAs by their cognate aaRS are boxed. The corresponding yeast sequences are also boxed. The U36 which converts tRNATm to an amber suppressor as well as bases which differ in tRNA&, and tRNA& are presented in brackets.
GlnRS and TrpRS enzymes have not been mapped. Experiments to identify these sites are currently in progress.